Agricultural and Food Science in Finland 515 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. HPLC analysis of vitamin B 6 in foods Velimatti Ollilainen Department of Applied Chemistry and Microbiology, PO Box 27, FIN-00014 University of Helsinki, Finland, e-mail: velimatti.ollilainen@helsinki.fi ACADEMIC DISSERTATION to be presented, with the permission of the Faculty of Agriculture and Forestry of the University of Helsinki, for public criticism in lecture hall B2, Viikki on January 28th 2000, at 12 noon. 516 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Supervisor: Professor Vieno Piironen Department of Applied Chemistry and Microbiology University of Helsinki Reviewers: Dr. Rainer Huopalahti Department of Biochemistry and Food Chemistry University of Turku and Dr. Johan Lindeberg National Food Administration Uppsala Opponent: Dr. Rune Waagbø Institute of Nutrition Directorate of Fisheries Bergen 517 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. “Typically, the more useful a technology is, the more likely people are to adopt it before completely understanding it. Chromatography’s tremendous usefulness has ensured that its practice is far ahead of theory.” (Wirth 1994) Photo: pyridoxine hydrochloride crystals photographed with polarizing filters (9 x magn.) © Velimatti Ollilainen 518 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods 519 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Preface This study on vitamin B 6 was conducted at the Department of Applied Chemistry and Microbiology, University of Helsinki as a part of the project “The Nutrient Content of Finnish Foods - Water- Soluble Vitamins” during the years 1994–1999. The world of liquid chromatography combined with the analysis of vitamins is an attractive one. I’ve enjoyed this work which has been both challenging and educational. The development in liquid chromatography has provide sophisticated tools for vitamin analysts. However, there is still a lot of work is to be done. During these years, many people have been involved in this study. Without their support, this work would not have succeeded. I wish to thank them all. To my supervisor, professor Vieno Piironen, I like to express my deepest gratitude. All her advice and support in the course of time has been vital. I would like to thank her for reading my manuscript and her stimulating criticism. I wish to thank emeritus professor Pekka Koivistoinen who once made me “an offer I couldn’t refuse”. His attitude to food chemistry research was most encouraging. His enthusiasm pushed for- ward my interest in the field of liquid chromatography and vitamin analysis. Professor Pertti Varo is worthy of my sincere thanks for his expertise in the research group, and being an excellent teacher for me. Dr. Anu Hopia and Dr. Liisa Vahteristo are kindly acknowledged for our continual discussions about food chemistry, liquid chromatography or many other matters. Many new aspects arose during these innovative conversations and arguments. I am indebted to Dr. Anna-Maija Lampi and Dr. Päivi Ekholm for their patience with me with statistics. Antti Uusi-Rauva is acknowledged for all his ad- vice during the years, and especially when I was conducting isotope measurements. I wish to thank reviewers Dr. Rainer Huopalahti and Dr. Johan Lindeberg for their comments and justified criticism. It was so important to receive their opinion and outlook in this matter. My special thanks go to Paul Finglas for introducing me to the impressive world of intercompar- ison studies in the past years. I also want to thank Dr. Henk van den Berg and Dr. Antal Bognar for the most rewarding discussions concerning the vitamin measurements. It has been my pleasure to work with all three of you. Kirsti Risunen and Jorma Löytynoja are acknowledged for their technical skill and expertise in the course of sampling and maintaining the laboratory facilities. I also wish to thank the other staff and students of our department for the pleasant working atmosphere. The financial support provided by Ministry of Agriculture and Forestry, Ministry of Social Af- fairs and Health, and Centre for Metrology and Accreditation is gratefully acknowledged. Finally, I wish to thank my dear family and as well as my friends for their ever-continuing sup- port. My warmest thanks belong to Ulla who has gave me support during all these years. She had the strength to believe that this work would be finished some day. Helsinki, January 2000 Velimatti Ollilainen 520 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Contents Abstract ............................................................................................................... 526 1 Introduction ............................................................................................................ 527 2 Literature Review ................................................................................................... 528 2.1 Chemical and physical properties of vitamin B 6 compounds .................... 528 2.1.1 Nomenclature and chemical structures .............................................. 528 2.1.2 Spectral characteristics ....................................................................... 531 2.1.3 Chemical reactions .............................................................................. 532 2.1.4 Some reactions of pyridoxal and pyridoxal-5’-phosphate ............... 533 2.2 Nutrition and physiological functions .......................................................... 534 2.2.1 Utilization ............................................................................................. 534 2.2.1.1 Free and phosphorylated forms ............................................ 535 2.2.1.2 Glycosidically bound pyridoxines ........................................ 535 2.2.1.3 Chemically modified forms .................................................. 537 2.2.1.4 Interaction of other food components – food digestibility 538 2.2.2 Coenzyme function .............................................................................. 538 2.2.3 Proposed role(s) in hyperhomocyst(e)inaemia ................................. 539 2.2.4 Recommended dietary intake ............................................................. 540 2.3 Occurrence and distribution of vitamin B 6 in foods ................................... 541 2.3.1 Meat, offals and fish ............................................................................ 541 2.3.2 Plant foods ............................................................................................ 541 2.3.3 Milk, milk products and eggs ............................................................. 542 2.4 Analysis of vitamin B 6 compounds .............................................................. 543 2.4.1 Extraction and hydrolysis techniques ................................................ 544 2.4.1.1 Mineral acid extraction ......................................................... 544 2.4.1.2 Chemical deproteinating agents ........................................... 545 2.4.1.3 Enzymatic hydrolysis ............................................................ 547 2.4.2 Chromatographic techniques .............................................................. 549 2.4.2.1 Open column chromatography ............................................. 551 2.4.2.2 High-performance liquid chromatography .......................... 551 2.4.2.3 Gas chromatography and mass spectrometry ...................... 556 2.4.3 Microbiological methods .................................................................... 556 3 Objectives of the study .......................................................................................... 558 4 Materials and methods ........................................................................................... 558 4.1 Calibrants and enzymes ................................................................................. 558 4.2 Equipment ....................................................................................................... 558 4.3 Food samples, sample collection and pretreatment .................................... 559 4.4 Evaluation of liquid chromatographic method and extraction procedure 560 4.4.1 Analytical liquid chromatography ..................................................... 560 4.4.1.1 Chromatographic parameters and column testing protocol 560 4.4.1.2 Tested column packings and mobile phases ........................ 560 521 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 4.4.1.3 Post-column derivatization ................................................... 561 4.4.2 Preliminary experiments on extraction procedure ............................ 562 4.4.3 Solid-phase extraction ......................................................................... 562 4.4.4 Statistics ............................................................................................... 563 4.5 Chosen method for routine food analysis .................................................... 563 4.5.1 Extraction procedure ........................................................................... 563 4.5.2 Enzymatic digestion ............................................................................ 564 4.5.3 Liquid chromatographic separation ................................................... 564 4.5.4 Calculations and expressing the results ............................................. 565 4.5.5 Comparison of the results to national food composition tables ...... 565 4.5.6 Evaluation of the chosen method ....................................................... 565 4.5.6.1 Stability of pyridoxal-5’-phosphate ..................................... 566 4.5.6.2 In-house monitoring .............................................................. 566 4.5.6.3 Intercalibration and collaborative studies ........................... 566 4.5.6.4 Estimating the uncertainty of the results, and statistical calculations ............................................................................ 567 4.6 Characterization of bound pyridoxine ......................................................... 567 4.6.1 Isolation ................................................................................................ 567 4.6.2 β-glucosidase digestion ....................................................................... 568 4.6.3 Structural evaluation ........................................................................... 569 5 Results ..................................................................................................................... 569 5.1 Liquid chromatography ................................................................................. 569 5.1.1 Choice of the column and the mobile phase ..................................... 569 5.1.2 Post-column derivatization ................................................................. 570 5.2 Sample extraction ........................................................................................... 571 5.2.1 Preliminary experiments for the extraction procedure ..................... 571 5.2.2 Evaluation of the chosen perchloric acid extraction ........................ 573 5.2.3 Enzymatic digestion ............................................................................ 575 5.2.4 Solid-phase extraction ......................................................................... 576 5.3 Validity of the method chosen for routine food analysis ........................... 577 5.3.1 General parameters .............................................................................. 577 5.3.2 Uncertainty of results .......................................................................... 578 5.3.3 Interlaboratory and laboratory proficiency tests .............................. 578 5.4 Routine food analysis .................................................................................... 581 5.4.1 Flesh foods: meat, fish and poultry ................................................... 581 5.4.2 Dairy products and egg ....................................................................... 582 5.4.3 Plant-derived foods ............................................................................. 583 5.4.4 Comparison with national food composition tables ......................... 584 5.5 Characterization of bound pyridoxine ......................................................... 588 5.5.1 Enzymatic hydrolysis .......................................................................... 588 5.5.2 Proton NMR spectroscopy .................................................................. 588 5.5.3 FAB mass spectrometry ...................................................................... 592 6 Discussion ............................................................................................................... 592 6.1 Liquid chromatography ................................................................................. 592 6.1.1 Column packings ................................................................................. 592 522 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods 6.1.2 Mobile phase and column temperature .............................................. 595 6.1.3 Post-column derivatization ................................................................. 596 6.2 Evaluation of the chosen method for routine food analysis ....................... 596 6.2.1 Acid hydrolysis and the stability of PLP ........................................... 596 6.2.2 Enzymatic hydrolysis .......................................................................... 598 6.2.3 Solid-phase extraction ......................................................................... 599 6.2.4 Validity of the routine food analysis method .................................... 600 6.2.5 Laboratory proficiency ........................................................................ 601 6.3 Food analysis .................................................................................................. 602 6.3.1 Flesh foods; meat, poultry, and fish ................................................... 602 6.3.2 Dairy products and egg ....................................................................... 603 6.3.3 Plant-derived foods ............................................................................. 604 6.3.4 Comparison to national food composition tables ............................. 605 6.4 Characterization of isolated pyridoxine derivative ..................................... 605 6.4.1 β-glucosidase hydrolysis ..................................................................... 605 6.4.2 Proton NMR spectroscopy .................................................................. 605 6.4.3 FAB mass spectrometry ...................................................................... 607 7 Conclusions ............................................................................................................ 608 References .................................................................................................................... 609 Selostus ......................................................................................................................... 619 Appendices 1–2 523 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. List of abbreviations according to Riekkola 1995, and Huopalahti et al. 1994 α separation factor, α=k 2 /k 1 A s 2 peak symmetry, (peak symmetry)2, A s 2 =(b/a)2 calculated on the base line A 4.4 peak symmetry calculated on the basis of 4.4% peak height, A 4.4 =(b/a) calculated at 4.4% of peak height APCI atmospheric pressure chemical ionization C m concentration in the mobile phase CAS RN Chemical Abstract Service Number CEN European Commitee for Standardization CMC critical micelle concentration C s concentration in the stationary phase CV r coefficient of variation within laboratories CV R coefficient of variation between laboratories D minimum detecability DPN 4-deoxypyridoxine (4-desoxypyridoxine) DPN•HCl 4-deoxypyridoxine hydrochloride DPNP 4-deoxypyridoxine-5’-phosphate d p particle diameter (µm) EI electron impact ionisation ESD error standard deviation EU-MAT European Union Measuring and Testing Programme E% energy percent ε molar absorption coefficient (l mol-1 cm-1 or l mmol-1cm-1) F 5 width from peak start to total retention time at the 5% of peak height (min) FAB fast atom bombardment GC-MS gas chromatography – mass spectrometry Hcy homocyst(e)ine HP hydrophobicity index HPLC high-performance liquid chromatograph(y) HPN 6-hydroxypyridoxine (2-methyl-3,6-dihydroxy-4,5-bis(hydroxymethyl)-pyridine) i.d. inner diameter ISTD internal standard, internal standard method J coupling constant (Hz) k e retention factor (relative retention), k=(t R -t M )/t M l column length (cm, m) λ max wavelength value related to a maximum UV/VIS absorption (nm) LC-MS liquid chromatography – mass spectometry MALDI matrix associated laser desorption ionization M W molecular weight (g/mol) MCA monochloroacetic acid, ClCH 2 COOH MPA metaphosphoric acid, (HPO 3 ) n MPCSC 6-methyl-2-pyridone carboxaldehyde MS mass spectra, mass spectrometer 524 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods MSn multiple scan mass spectra m/z mass – charge -ratio N number of theoretical plate, N=16(t R / w b )2 NCI negative-ion chemical ionization N/l number of theoretical plates per meter N sc surface coverage (µmol/m2), N sc = [106 P C /1200n C -P C (M W -I)] [1/S] n C number of carbons in the bonded silane molecule NMR nuclear magnetic resonance PA 4-pyridoxic acid (2-methyl-3-hydroxy-4-carboxy-5-hydroximethylpyridine) P C percent carbon in the bonded phase PCA perchloric acid, HClO 4 PCI positive-ion chemical ionization pK equilibrium constant pK a dissociation constant PL pyridoxal (2-methyl-3-hydroxy-5-hydroxymethyl-4-carboxal-pyridine) PL•HCl pyridoxal hydrochloride PLP pyridoxal-5’-phosphate (2-methyl-3-hydroxy-5-[(phosphooxy)methyl]-hydroxime- thyl-4-carboxalpyridine) PM pyridoxamine (3-hydroxy-2-methyl-5-hydroxymethyl-4-aminomethylpyridine) PM•2HCl pyridoxamine dihydrochloride PMP pyridoxamine-5’-phosphate (3-hydroxy-2-methyl-5-[(phosphooxy)methyl]-4-ami- nomethylpyridine) PMP•2HCl pyridoxamine-5’-phosphate dihydrochloride PN pyridoxine (2-methyl-3-hydroxy-4,5-bis(hydroxymethyl)-pyridine) PN base pyridoxine cation (Mw 170 g mol-1) PN•HCl pyridoxine hydrochloride PNG pyridoxine glycoside(s), also 5’-O- β-D-glucopyranosylpyridoxine PNG% pyridoxine glycoside%, PNG%=(PNG/ΣΣ B 6 )×100 PNP pyridoxine-5’-phosphate(2-methyl-3-hydroxy-5-[(phosphooxy)methyl]-4-hy- droxymethylpyridine) PNX isolated derivative of pyridoxine PS-DVB poly(styrenedivylbenzene) R s peak resolution, Rs=2(t R2 -t R1 )/(w b1 +w b2 ) R r relative response factor RSD% relative standard deviation, percent, (mean/standard deviation)×100 RSD r repeatability relative standard deviation RSD R reproducibility relative standard deviation δ chemical shift (ppm) S specific surface area of the unbonded silica (m2/g) SAX strong anion-exchange SiOH silanol index value SI-MS secondary-ion mass spectra SCX strong cation-exchange(r) SPE solid-phase extraction SSA 5-sulfosalicylic acid, (2-hydroxy-5-sulfobenzoic acid) t M hold-up time, retention time of unretained compound t R total retention time 525 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. TCA trichloroacetic acid, Cl 3 CCOOH TEA triethylamine, (CH 3 CH 2 ) 3 N T USP tailing factor, T=w 5 /2F 5 V 0 interparticle volume of the column, void volume w b peak-width at base w h peak-width at half height w 5 width of the peak at the 5% height U unit ΣB6 sum of PLP, PMP, PNP, PL, PN, and PM ΣΣB6 sum of ΣB6 and PNG 526 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods HPLC analysis of vitamin B 6 in foods Velimatti Ollilainen Department of Applied Chemistry and Microbiology, PO Box 27, FIN-00014 University of Helsinki, Finland, e-mail: velimatti.ollilainen@helsinki.fi The objective of this work was to evaluate the methods for determination of vitamin B 6 in foods. To achieve this, the literature review focused on sample treatment and liquid chromatographic analysis of vitamin B 6 related compounds. In the experimental part, the chosen sample pretreatment and the high-performance liquid chromatographic (HPLC) method were validated, and used to produce vita- min B 6 data on various food items commonly consumed in Finland. The main emphasis of the sample treatment was on the extraction efficiency and the maintenance of the original concentration profile of the vitamers. Several acid extraction procedures were tested for this purpose. Perchloric acid was chosen as the extraction agent. Routine food analysis was then performed using dilute ice-cold per- chloric acid extraction followed by an internally standardized ion-paired reversed-phase liquid chro- matography. Food samples were hydrolyzed with β-glucosidase and alkaline phosphatase enzymes, phosphorylated and glycosylated vitamers were quantitated before and after the enzymatic digestion. This procedure enabled the extraction of vitamin B 6 compounds in their intact forms, and the meas- urement of free, phosphorylated and glycosylated forms. The maintenance of the concentration pro- file of the vitamers was verified by using 14C -labeled pyridoxal-5’-phosphate in the examination of the extraction procedure. The extraction efficiency and laboratory performance were confirmed by interlaboratory studies. Up-to-date data on vitamin B 6 content of about fifty common food items was produced. The data includes the results from meat and poultry, fish and fish product, dairy product, cereal and vegetable, and ready-to-eat food samples. Free and phosphorylated vitamin B 6 compounds were measured in all food groups, and the glycosylated vitamer fraction was analyzed in all plant- derived foods. The results obtained in this work showed that vitamin B 6 content of nearly all foods of plant origin was mainly comprised of glycosidically bound pyridoxine derivatives. These bound an- alytes are normally not taken into account in traditional analytical methods, and food composition tables lack the data of glycosylated pyridoxine. The role of the glycosylated pyridoxines need to be clarified in terms of their analytical and physiological nature. If, as it is currently assumed, the avail- ability of the bound forms is limited for humans, the role of vegetables, cereals and other foods of plant-origin as a source of vitamin B 6 , as well as the analytical methods should be reassessed. Key words: food, vitamin B complex, pyridoxine, pyridoxal, pyridoxamine, HPLC, liquid chroma- tography, analytical methods, vitamin analysis © Agricultural and Food Science in Finland mailto:velimatti.ollilainen@helsinki.fi 527 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. The pioneer work of Eijkman in 1906 is consid- ered to be the first recognition of the concept of vitamin as he suggested the presence of an “anti- polyneuritis factor” in rice which is “indispen- sable to health”. On the next decade, Funk’s the- ory of the four vitamins opened the new possi- bilities in nutrition research as the etiology of diseases could be linked to diet and was no long- er limited to the “germ theory” only. The subse- quent empirical phase combined with the devel- opment of the experimental tools yield the dis- covery of several vitamins in a relatively short time. The substances which are classified as vi- tamins according to present knowledge were identified within only five decades after Eijk- man’s work. The identification of vitamin B 6 compounds was one the most rapid one after the discovery that vitamin B complex includes several, chem- ically and physiologically dissimilar organic compounds. The vitamin “B 6 family” was sur- prisingly soon characterized after their discov- ery , within only one decade. At first the biolog- ical methods like microbiological assays enabled their measurement in various biological matri- ces, and these biological methods were rapidly followed by the intrumental chromatographic procedures. In the late 1970’s, Japanese research- ers showed that vitamin B 6 in plant-origin mate- rials is mainly constituted of carbohydrate de- rivatives of pyridoxine and their different vita- min activity from that of the free vitamers was proposed. In addition of traditional “vitamin ac- tions” for vitamin B 6 , the new roles for pyridox- ine related compounds in human health have also been suggested. These includes their proposed task for instance in hyperhomocyst(e)inaemia and in cancer. The growing interest in the interactions be- tween diet and health requires a more precise analytical methodology for vitamins. In the food labeling and maintaining the food data bases the appropriate methods are needed. However, the harmonization of analytical procedures is com- plicated by the fact that term vitamin is a phys- iological rather than chemical one, expressing a certain physiological activity which is related to the chemical substances responsible for this ac- tivity. In a traditional point of view, the vitamin B 6 activity is vested to the free and phosphor- ylated vitamers but the other derivatives of py- ridoxine may not be included to that considera- tion. In the beginning of the new century, Eijk- man’s and Funk’s question “what is a vitamin” is still of current interest as this must be con- cluded whenever the analytical procedures are being evaluated in the laboratory performing vitamin analysis. The first part of the present work includes the literature review mainly focusing on the methodology of vitamin B 6 for the biological materials. In the experimental section, the suit- ability of the high-performance liquid chroma- tographic method for food analysis was evaluat- ed, and the chosen method was used for the de- termination of vitamin B 6 compounds in the most common food items. 1 Introduction 528 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods 2.1 Chemical and physical prop- erties of vitamin B 6 compounds 2.1.1 Nomenclature and chemical structures Free and phosphorylated vitamers Vitamin B 6 compounds can be classified as de- rivatives of 2-methyl-3-hydroxypyridine (Fig. 1). Vitamin B 6 is a group name for compounds hav- ing vitamin B 6 activity: this group of compounds includes the free vitamers pyridoxine (2-methyl- 3-hydroxy-4,5-bis(hydroxymethyl)-pyridine, CAS RN 65-23-6), pyridoxal (2-methyl-3-hy- droxy-5-hydroxymethyl-4-carboxalpyridine), CAS RN 66-72-8 and pyridoxamine (3-hydroxy- 2-methyl-5-hydroxymethyl-4-aminomethylpyri- dine), CAS RN 85-87-0, and their phophorylat- ed forms; pyridoxal-5’-phosphate (2-methyl-3- hydroxy-5-[(phosphooxy)methyl]-hydroxime- thyl-4-carboxalpyridine, CAS RN 54-47-7), py- ridoxamine-5’-phosphate (3-hydroxy-2-methyl- 5-[(phosphooxy)methyl]-4-aminomethylpyrid- ine), CAS RN 529-96-4 and pyridoxine-5’-phos- phate (2-methyl-3-hydroxy-5-[(phosphooxy)- methyl]-4-hydroxymethylpyridine), CAS RN 447-05-2. Free vitamers are commercially avail- able as crystall ine hydrochlorides, l ike pyridoxine•HCl (CAS NR 58-56-0), pyridoxal•HCl (CAS No. 65-22-5), and pyridoxamine•2HCl (CAS RN 524-36-7). Pyridoxine•HCl (Mw 205.64 g mol-1) is the UPS reference standard. Members of ”the same vita- min family” are called vitamers (Combs 1992). The terms pyridoxol, PN or vitamin B 6 are also being used as a synonym for pyridoxine. In aqueous solutions vitamin B 6 compounds exist in various ionic forms depending on e.g. pH and temperature (Snell 1963); in a cationic form (I) in acidic environments, as a mixture of zwitterionic (II) and unionized (III) form in neu- tral solutions, and as an anion (IV) in alkaline solutions (Fig. 2). However, a dipolar ionic form of pyridoxine (II) predominates in neutral me- dia. Pyridoxamine is positively charged in neu- tral solutions due to its basic 4-aminomethyl group. The aldehyde group of pyridoxal enables also hemiacetal (V) and quinoidic (VI) structures in addition to free aldehyde (VII) and hydrate (VIII) forms (Fig. 3). Free and phosphorylated B 6 vitamers crys- tallize as white to off-white platelets or rods, the commercial preparations normally being hydro- chlorides. Their melting or decomposition points are generally over 200°C, the melting points of free bases being lower. In general, B 6 vitamers dissolve in water (ca. 0.5–1g/2ml) and in 95% ethanol (0.5–1g/100ml) but are practically insol- uble in most organic solvents. Pyridoxic acid is only slightly soluble in water and alcohol. Di- lute acidic vitamer solutions are rather stable and tolerate thermal processings (like autoclaving) but pyridoxal degrades in alkaline solutions. Pyridoxine, when autoclaved in a neutral solu- tion, forms a dimer in which the 4’-hydroxyme- thyl group of one pyridoxine molecule is linked with the nitrogen atom of the other pyridoxine 2 Literature Review Fig. 1. Structures of vitamin B 6 compounds. R1 R2 Common name -CH 2 OH -CH 2 OH Pyridoxine -CHO -CH 2 OH Pyridoxal -CH 2 NH 2 -CH 2 OH Pyridoxamine -COOH -CH 2 OH 4-pyridoxic acid -CH 3 -CH 2 OH 4-deoxypyridoxine -CH 2 OH -CH 2 OPO(OH) 2 Pyridoxine-5’-phosphate -CHO -CH 2 OPO(OH) 2 Pyridoxal-5’-phosphate -CH 2 NH 2 -CH 2 OPO(OH) 2 Pyridoxamine-5’-phosphate -CH 2 OH -CH 2 OC 6 O 5 H 11 Pyridoxine-5’-glucoside 529 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. molecule (Snell 1963). All vitamer forms are relatively light-sensitive and are destroyed by strong oxidazing agents. Glucosidically bound forms Nelson and his coworkers (1977) described the bound form of vitamin B 6 present in orange juice as a small nondialyzable (M w < 3500 daltons) molecule which binds both pyridoxine and pyri- doxal. The non-protein character of the vitamin B 6 conjugate present in orange juice was veri- fied by protease digestion as the results derived from the enzymatic hydrolysis indicated no in- crease in vitamin B 6 activity after protease treat- ment in contrast to a lyophilized yeast sample similarly treated. Thus, the isolate did not have a protein-binding nature. However, no further structure interpretation for this heat stable non- protein compound was given. Carbohydrate derivatives of pyridoxine are reported to be present only in plant foods. It is supposed that their utilization by humans as a source of vitamin B 6 is limited. The first eluci- ated compound, 5’-O-(β-D-glukopyranosyl) pyridoxine (C 14 H 21 NO 2 , Mw 331.1g mol-1)(Fig. 4), was isolated and characterized from rice bran by Yasumoto et al. (1977). It was proposed that Fig. 2. Ionic forms of pyridoxine. Fig. 3. Free aldehyde (VII), hydrate (VIII), quinoidic (VI) and hemiacetal (V) forms of pyridoxal. the transglycosylation of pyridoxine enables the formation of both 5’- and 4’-derivatives. How- ever, the 5’-substituted form was considered to be the dominate form. Its white crystals are sol- uble in water and in boiling 75% ethanol. The 530 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods β-glycosidic bond is reported to be stable in non- mineral acid solutions, such as in perchloric acid solution (Schramm and Bitsch 1993, Toukairin- Oda et al. 1989), in trichloroacetic acid solution (Gregory and Ink 1987, van Schoonhoven et al. 1994) and in metaphophoric acid solution (Sampson et al. 1995), or in neutral buffer solu- tions (Kabir et al. 1983a) but it is hydrolyzed in mineral acids especially when acid extraction is combined with a thermal process like autoclav- ing (Kawai et al. 1971a). Preparation of α-glyc- osylated pyridoxines (Suzuki et al.1997) and nonlabeled, deuterated as well as tritiated β-glu- cosidic forms (Gregory and Nakano 1997) has been recently reported. Other glycosylated vitamers reported in rice are 5’-O-(β-cellobiosyl)pyridoxine, 4’-O-(β-D- glucosyl)-5’-O-(β-cellobiosyl)pyridoxine and 5’-O-(β-glucotriosyl)pyridoxine (Tadera et al. 1988). A glycosylated pyridoxine, probably es- terified with malonic acid or 3-hydroxy-3-me- thyl-4-carboxy-butanoic acid, (named B6X) ex- isted in rice and wheat bran, peas and soybeans (Tadera et al. 1983, Tadera et al. 1985 and Tad- era et al. 1986a). This compound(s) was consid- ered to be an esterified pyridoxine glucoside as it gives a microbial response after alkaline treat- ment and hydrolysis of the glycosidic bond. This more precisely unidentified compound was only a minor of B 6 compounds present in those plant foods. A suggestion that PN glucosides exist in potato as a mixture of mono- and diglucosides was presented by Addo and Augustin (1988). Pyridoxine glucoside-forming activity seems to be distributed particularly to the microbe gen- era Sarcina and Micrococcus; pyridoxine 5’-α- glucoside and pyridoxine 4’-α-glucoside were synthesized from pyridoxine and glucosyl do- nors via transglucosidation (Ogata et al. 1968, Ogata et al. 1969a). Sucrose, maltose and phe- nyl-α-D-glucoside acted as glucosyl donors. The formation of β-galactosides of pyridoxine, 4’- O-galactopyranosyl-1–4-galactopyranosyl pyri- doxine, 4’-O-galactopyranosylpyridoxine, 5’-O- galactopyranosyl-pyridoxine using Sporobolo- myces singularis has been recently published by Suzuki and Uchida (1997). Referring to the data published by several research groups, it is obvious that a major por- tion of vitamin B 6 exists in the form of carbohy- drate derivatives of pyridoxine in the plant ma- terials. Their distribution, chemical forms and importance as a source of vitamin B 6 in humans is still not well understood. Traditional analyti- cal methods which are based on the release of free vitamers from the sample matrix do not rec- ognize these bound vitamin forms. For this rea- son the present food composition data bases do not contain this information. In addition, data concerning the changes in their amount during the maturation of vegetables, fruits, and related plant food materials, as well as their stability in the food processing systems are scarce. Other forms of vitamin B 6 The study of Bishop and Tryfiates (1989) indi- cated a novel vitamin B 6 metabolite identified as adenosine-N6-diethylthioether-N1-pyridox- imine-5’-phosphate (Fig. 5). Animal and human tumor cells incubated with pyridoxine formed this Schiff’s base conjugate of vitamin B 6 with Fig. 4. The structure of 5’-O-(β-D-glucopyranosyl) pyridoxine. Fig. 5. The structure of adenosine-N6-diethylthioether-N1- pyridoximine-5’-phosphate. 531 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Table 1. UV-Spectral characteristics of free B6 vitamers and pyridine (Metzler and Snell 1955). Compound Solvent λ max (nm) ε(1 mmol-1cm-1) Pyridoxine – cation 0.1M HCl 291 8.6 – neutral alcohol 286 5.7 – dipolar ion pH 6.8 324 7.2 – anion 0.1M NaOH 310 6.8 Pyridoxal – cation 0.1M HCl 288 9.0 – neutral 60% dioxane 280 4.1 – dipolar ion pH 6.9 317 8.9 – anion pH 10–11 302 5.7 Pyridoxamine 0.1 HCl 292 8.2 98% dioxane 287 3.4 pH 6.7 326 7.9 0.1M NaOH 310 7.2 4-Deoxypyridoxine – cation 0.1M HCl 282 8.3 – dipolar ion neutral 313 8.1 – anion 0.1M NaOH 301 7.1 Pyridine – cation 0.05M HCl 256 5.7 – uncharged 0.02M NH3 256 2.8 adenosine diethylthioether, and its formation was highest in the rapidly growing least differentiat- ed cells. The metabolite corresponded to 10–30% of the total vitamin B 6 metabolites and its for- mation was inversely related to tumor differen- tiation (Tryfiates et al. 1991). This novel vita- min B 6 compound was assumed to be a minor product of vitamin B 6 metabolism in tumors and cultured tumor cells (Gregory 1992). Based on the findings that cancer patients in the actice dis- ease phase had 3–4 fold higher plasma levels of this B 6 metabolite, a role for it as a circulating marker for human cancer detection was proposed (Tryfiates 1996, Tryfiates et al. 1996). 2.1.2 Spectral characteristics UV and fluorescence spectra Due to the ionic nature of B 6 vitamers, their UV absorbance maxima depend on pH and solvent used. Two absorption maxima are normally present in the UV-VIS absorption spectra of the vitamers (Fig. 6). These two maxima are con- sidered to be derived from those of pyridine. The Fig. 6. The UV absorption spectrum of pyridoxine at various pH values. (Reprinted with permission from [Metzler and Snell 1955, Journal of Americal Chemical Society 77: 2431– 2437].Copyright [1955] American Chemical Society.) 532 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods main absorbance maximum of pyridoxine in acid and neutral media lies at wavelengths of ca. 290 and 325nm, respectively, compared to the wave- length of 310nm in alkaline solutions. The high- est molar absorption coefficient is achieved in acid solutions; the value of coefficient is reduced when the maximum wavelength shifts to higher wavelengths (Metzler and Snell 1955). Table 1 shows the molar extinction coefficients for free B 6 vitamers and pyridine. Vitamin B 6 compounds possess a native state fluorescence; in acid media the maximum exci- tation and emission wavelengths are located at ca. 300 nm and 375 nm. In neutral solutions the maximum excitation and emission wavelengths lies at ca. 330–340 nm and 385–400 nm (Table 2). The maximum fluorescence intensity of py- ridoxine is found at pH 7 (Duggan et al. 1957). Natural fluorescence of B 6 vitamers is consid- ered to have the characteristics of the hydrox- ypyridine moiety (Peterson et al. 1955). Pyri- doxal-5’-phosphate exhibits a lower fluorescence response than the other vitamers. It is also low- er than that of free pyridoxal as the cyclic hemia- cetal structure is hindered in the phosphate ester form. The free aldehyde group present in PLP is considered to diminish the fluorecence of the aromatic system by withdrawing the electrons in the ring system. Raising the pH of the meas- uring solvent to neutral or to weakly alkaline enhanced the relative fluorescence of PLP (Bridges et al. 1966) especially when this pro- cedure is combined with the formation of a bi- sulphite adduct (Coburn and Mahuren 1983). A comprehensive fluorescence study on hydrox- ypyrines and vitamin B 6 compounds has been published by Bridges et al. (1966). UV-absorbance of the isolated 5’-O-(β-D- glucopyranosyl)pyridoxine fraction had a maxi- mum at the wavelength of 292 nm in 0.1M hy- drochloric acid as well as at 246 nm and 310 nm in 0.1M sodium hydroxide solution (Yasumoto et al. 1977). The molar fluorescence is assumed to be the same as that of pyridoxine (Gregory and Ink 1987). 2.1.3 Chemical reactions The free vitamers, pyridoxine, pyridoxal and pyridoxamine, exhibit the typical chemical re- actions of para unsubstituted aromatic phenols. Colorimetric measurements based on reactions Table 2. Fluorecence characteristics of B 6 vitamers (Bridges et al. 1966). Vitamer Excitation/emission Fluorescence pH range of the wavelenght (nm) intensity* maximum fluorescence Pyridoxine 332/400 238 6.5–7.5 320/380 168 12.0–14.00 Pyridoxal (hemiacetal) 330/382 207 6.0 310/365 283 12.0 Pyridoxal-5’-phosphate 330/410 11 6.0 315/370 17.5 12.0–14.0 Pyridoxamine 337/400 370 4.0–5.5 320/370 410 14.0 4-Pyridoxic acid 320/420 770 1.5–4.0 315/425 650 6.1–9.4 4-Pyridoxolactone 365/423 32 2.5–4.8 360/430 2150 8.7–13.0 * expressed as fluorescence intensity relative to the fluorescence intensity of 3-hydroxypyridine anion at pH11.0 considered as 100 533 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. with the diazo derivative of sulfanilic acid and with 2,6-dichloroquinone have been used for quantitative measurement of vitamin B 6 . In ad- dition, one or more alcoholic hydroxy groups in the molecule enable ester formation with acylat- ing reagents. Many of these reactions, however, have insufficient selectivity and/or sensitivity to be applied in the analysis of food and other bio- logical materials. 2.1.4 Some reactions of pyridoxal and pyridoxal-5’-phosphate Pyridoxal and its phosphate ester are considered to be the most reactive B 6 vitamers, and many of their reactions take place in sample treatment during the analytical procedure. Some of the re- actions related to the aldehyde group of the mol- ecule are utilized to modify the fluorescence detection in liquid chromatography while other reactions are involved with the formation of bounded forms of pyridoxal and its phosphate ester with proteins which may yield reduced vi- tamin B 6 activity. Transamination, the intercon- version of PL to PM and vice versa, leads to the changes of the vitamer distribution during sam- ple treatment compared to the original. Although they are rather stable in pure acid- ic and neutral solutions, many special reactions are involved with pyridoxal due to its reactive aldehyde group. Pyridoxal is quite stable in acid- ic solutions in dim environments and the pres- ence of light catalyzes its degradation. General- ly, pyridoxal-5’-phosphate reacts more rapidly and more completely than its free vitamer as the stabilizing hemiacetal structure of pyridoxal is disabled in phosphorylated vitamer; PLP has been reported to be 1.5–2 times more reactive than the free pyridoxal under similar conditions (Gregory and Hiner 1983). Phosphorylated py- ridoxal reacted with bisulphite forming a hydrox- ysulfonic derivative which is more fluorescent than native PLP in neutral or slightly alkaline media (Coburn and Mahuren 1983). The reac- tion with ammonia compounds after elimination of a water molecule yields a carbon-nitrogen double bond in the product molecule. Thus, PLP- semicarbazone is formed after the reaction of PLP with semicarbazide. This acid catalyzed nucleophilic attact reaction had an optimum in a slightly acidic pH, and the formed vitamer-sem- icarbazone conjugate exhibits strong fluores- cence at pH 12 (Gregory 1980a). Pyridoxal was converted to pyridoxine within two hours by re- duction by sodium borohydride in alkaline me- dium (10mM in 0.2M NaOH) (Chaikin and Brown 1949). To decrease the reaction time, a more concentrated borohydride solution (0.1M) was needed (Reitzer-Bergaentzle et al. 1993). This approach has been successfully utilized in an analytical method generally adopted in France. In aqueous solutions of amino acids the transamination reaction (Fig. 7) occurs slowly at room temperature. This reaction is catalyzed by heat and di- and trivalent metal ions, like cop- per, iron and aluminium salts (Metzler and Snell 1952). As the imine formation and its break- down reaction are rapid processes, the rate-lim- iting step in this non-enzymatic transamination was suggested to be the tautomeric rearrange- ment of the imine (Metzler 1957). The labiliza- tion of bonds in α-carbon (Fig. 7, III) allows for (i) racemization and elimation reactions as well as (ii) the decarboxylation and (iii) the aldoli- type reaction. The pH optimum for the reaction between pyridoxal and amino acids was estimat- ed to be 4.5 in a metal ion catalyzed reaction (Metzler and Snell 1952) and in a range of pH 5 to 7 in a non-catalyzed reaction (Cennamo 1964). In a model system described by Metzler and Snell (1952), an equimolar ratio of pyridoxal and pyridoxamine was formed within one hour when 10mM pyridoxal solution was heated (100°C) in the presence of 10mM glutamic acid at pH 5.0, and the same product mixture ratio was obtained from an equal mixture of pyridoxamine and ke- toglutaric acid. The use of an enzyme prepara- tion in the sample extraction procedure with a rather long incubation time produced a decreased amount of pyridoxal and the formation of pyri- 534 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods doxamine probably via the transamination route (van den Berg et al. 1996). Pyridoxal reacts with the amino group of pro- teins forming e.g. pyridoxyl-ε-lysine (Fig. 8a). When a pyridoxal solution containing cysteine is heated, a portion of pyridoxal was irreversi- bely converted to thiazolidine condensation product which still, after hydrolyzation, was re- ported to be available as a source of vitamin B 6 . The heating of condensed milk yielded the mer- capto derivative of pyridoxine (Fig. 8b) (Srnco- va and Davidek 1972). 2.2 Nutrition and physiological functions 2.2.1 Utilization The bioavailability of a nutrient depends on its extent of intestinal absorption and metabolic uti- lization. Vitamin B 6 compounds can exist in forms that are readily absorbed but poorly me- tabolized to forms which act as active coen- zymes. On the other hand, intestinal absorption Fig. 7. Non-enzymatic transamination of pyridoxal. 535 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. may limit a vitamer’s utilization (Gregory 1988b). During food processing, like the spray- drying of milk, vitamers can chemically react with other food components to yied products that have decreased vitamin B 6 activity. In addition, the nature of the food matrix (like materials rich in fiber) is suggested to have an influence on vitamin bioavailability. In general, B 6 vitamers are freely absorbed as nonphosphorylated forms via passive diffu- sion in the jejunun and ileum. Phosphate esters are dephosphorylated during absorption by the membrane-bound alkaline phosphatase. Free vi- tamers are then rephosphorylated after crossing the cell membranes and phosphorylated pyridox- ine and pyridoxamine are oxidized to the bio- logically active coenzyme form, pyridoxal-5’- phosphate. 2.2.1.1 Free and phosphorylated forms As free pyridoxine, pyridoxal, pyridoxamine and their related phosphorylated forms are being converted to each other by numerous enzymes in tissues, it is thought that they possess equal biological vitamin activity in rats and humans. Trials performed on animals (Nguyen et al. 1983) and microbes (Polansky et al.1985), however, it has been shown that individual B 6 vitamers may have different biological activity or growth re- sponse depending on the indicator used. The vi- tamin level in diet supplementation or in the diet may also affect the results (Gregory and Lither- land 1986). For instance, the nonphosphorylat- ed vitamers gave a fairly uniform response on plasma pyridoxal-5’-phosphate concentration in rat but a slightly different growth response and feed efficiency was observed (Nguyen et al. 1983). It was shown that the response of the rat to free B 6 vitamers was dependent on the pro- tein level in the diet; the response decreased in low-protein diet compared to that of the normal dietary protein level. The interaction between dietary protein intake and the role of vitamin B 6 in protein metabolism (Nguyen et al. 1983) as well as the gross composition of a diet, and its processing history (Nguyen and Gregory 1983) was established. 2.2.1.2 Glycosidically bound pyridoxines The main vitamin B 6 fraction in plant-derived foods consists of glycosylated pyridoxine. It is reported that it forms 5–70% of the total vita- min content in plant foods but it is absent in an- imal products. Several different glycosylated py- ridoxine derivatives have been found in nature, however, the bioavailability studies have main- ly focused on 5’-O-(β-D-glukopyranosyl) pyri- doxine, so called pyridoxine-β-glycoside. An inverse relationship in human between the percentage of glycosylated vitamin B 6 present in food and the extent of availability of the vita- min B 6 was reported by Kabir et al. (1983b). In- complete utization of glycosidic pyridoxine as vitamin B 6 in the rat was showed by Ink et al. (1986) and the bioavailability of pyridoxine-β- glucoside was estimated to be ca. 40% or less Fig. 8. The structures of a) pyri- doxyl-ε-lysine and b) mercaptopy- ridoxine. 536 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods compared to that of pyridoxine. Trumbo et al. (1988) arrived at the same estimation; the utili- zation of pyridoxine glucoside relative to the molar response of pyridoxine was only 10–30% in the rat. Seven to nine percent of ingested py- ridoxine glycoside was excreted in urine in its intact form at all dosage levels investigated. Their results showed that pyridoxine glucoside is incompletely utilized and the extent of utili- zation in rat is not influenced by the amount of its intake. Further studies of Trumbo and Gre- gory (1988) revealed that pyridoxine glucoside and pyridoxine are well absorbed in the intes- tine but the glycosidic form is then poorly me- tabolized: ca. 20% of ingested bound form was converted to pyridoxine in rat. Pyridoxine glu- coside was absorbed without prior conversion to pyridoxine and the limiting factor seemed to be its hydrolysis to biologically active pyridox- ine. As vitamin B 6 deficiency did not enhance the utilization of pyridoxine glucoside, vitamin B 6 status has little or no effect on the utilization. Intact pyridoxine glucoside was not secreted in the milk of lactating rats although pyridoxine derived from the limited hydrolysis of pyridox- ine glucoside was delivered to the mammary gland (Trumbo and Gregory 1989). Pyridoxine glucoside comprised ca. 15% of the total vita- min B 6 content in Nepalese human milk (Rey- nolds 1988). However, this contrasts with other studies where little or no pyridoxine glucoside was found in human milk (Gregory and Ink 1987, Andon et al. 1989). The reason for the higher concentration of pyridoxine glucoside in the breast milk of Nepalese women compared with American women remain unclear. The utilization of glycosidic bound pyridox- ine was later found to be higher in humans than was earlier found in rats (Gregory et al. 1991, Nakano et al. 1997).When pyridoxine and its glucoside were admistrated simultaneously, a greater extent of utilization was measured; the bioavailability of orally administrated deuterat- ed pyridoxine glucoside ranged from 50% to 58% compared to that of pyridoxine. Utilization of glucoside considerably decreased when it was administrated intravenously. These findings sup- ported the suggestion that β-glucosidase activi- ty in the intestinal mucosa or microflora increas- es the availability of pyridoxine glucoside in human. Animal trials showed that the utilization of pyridoxine glucoside varies in different spe- cies. The glucosidic form was effectively ab- sorbed in rats but 80% of it was rapidly excreted in urine. Thus, its bioavailability was estimated to ca. 20% compared to that of pyridoxine (Gre- gory et al. 1991, Gilpert and Gregory 1992). The bioavailability of pyridoxine glucoside, based on urinary excretion of the administrated tritium labeled pyridoxine glucoside, was only 10–30% in rats, but 70% in mice and hamsters, and 90% in guinea pig. The use of mice or hamsters as model species in utilization trials instead of mice or rats was then suggested (Banks and Gregory 1994). However, the results of their study indi- cated that pyridoxine glucoside metabolism dif- fers between rat, mice and hamster as well as in human. Overall, the bioassay methods based on animals have been found, in many cases, to be unsuitable for evaluating the bioavailability of vitamin B 6 (Gregory and Litherland 1986). An effect of the physiological condition on utiliza- tion was reported by Cheng and Trumbo (1993). During pregnancy the utilization of pyridoxine glucoside in rats was increased and was quite similar to that of pyridoxine. Hormonal differ- ences like changes in the activity of uterine gly- cosidases may explain these findings. It is proposed that pyridoxine-β-glucoside nutritionally operates in several ways: it can be utilized as a source of partially available vita- min B 6 but it also acts as a weak antagonist hence preventing the utilization of pyridoxine (Gilpert and Gregory 1992). Pyridoxine glucoside inhib- ited the uptake of pyridoxine in rats. Utilization of pyridoxine was retarded in isolated rat liver cells incubated in a equimolar mixture of pyri- doxine and pyridoxine-β-glucoside. It was con- cluded that pyridoxine glucoside uses the same transportation system as pyridoxine. As the amount of transported glucoside was only 20% of transported pyridoxine, a permeability barri- er due to steric hindrance at the transporter was proposed by Kawai et al. (1972a) and Zhang et 537 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. al. (1993). Unlabeled pyridoxine glucoside ad- ministrated simultaneously with 14C-labeled py- ridoxine altered the metabolism and in vivo re- tention of pyridoxine in rats. Changes in meta- bolic patterns as well as possible inhibition of certain enzymes or modulation of the β-glucosi- dase activity were suggested (Nakano and Gre- gory 1995). It was proposed that the interaction between pyridoxine-β-glucoside and nonglyco- sylated B 6 vitamers occur also in human metab- olism (Hansen et al. 1996). Their study showed that women consuming a diet rich in PNG ex- hibited a decrease in vitamin B 6 status indica- tors. The total loss of the total vitamin B 6 intake was estimated to be 15–18%. Pyridoxine 5’-α-glucoside and pyridoxine 4’- α-glucoside are formed by certain microbe gen- era, like Sarcina and Micrococcus (Ogata et al. 1968, Ogata et al. 1969a) but these α-glucosylat- ed compounds are expected to occur less com- monly in nature. Their bioavailability as a source of pyridoxine differs remarkably from those of related β-glucosides. It has been demonstrated that α-glucosides are actively transferred into rabbit erythrocytes (Kawai et al. 1972b) and they are readily utilized by the rat (Joseph et al. 1996, Tsuge et al. 1996). The microbiological activity for Saccharomyces ovarum was 20% compared to that of pyridoxine for a 24h incubation. Micro- biological activity was increased after a pro- longed incubation time and reached 50% of the activity of equivalent of mole amount of pyri- doxine. (Kawai et al. 1971a ). Synthetized 4’-α- and 5’-α glycosylated derivatives of pyridoxine probably use different transport mechanism than pyridoxine thus having no inhibitory effect on pyridoxine transport. Both forms were readily converted to pyridoxine in rat liver. Pyridoxine- 5’-α-glucoside was more rapidly hydrolyzed than the related 4’-form in the experiment of Joseph et al. (1996) and it was concluded that 5’-α-glucoside of pyridoxine, if present in diet, may have some nutritional importance for the intake of vitamin B 6 . It is quite evident that more detailed infor- mation of the free, phosphorylated and glyco- sylated vitamers, and their distribution in food is essential. The vitamin intake and its relevance to vitamin B 6 status can not be completely eval- uated on the basis of current food composition data as proper information for different vitam- ers including glycosylated forms and their dis- tribution is not available. 2.2.1.3 Chemically modified forms The reaction products of B 6 vitamers are usually incorporated during food processing. More re- active vitamers like pyridoxal, pyridoxamine and their related phosphates, can be bound to an amino or sulfhydryl group of proteins forming Schiff’s bases or disulfides. Known products of these reactions are for instance pyridoxyl-ε- lysine and a thiazolidine condensation deriva- tive. The extent to which these modified forms are utilized vary. Generally, pyridoxyl-ε-lysine is poorly utilized (Tsuge et al. 1996) having only 50% activity relative to the free form (Gregory & Kirk 1977, Gregory 1980d). Even an antivita- min B 6 effect of intact ε-pyridoxyllysine has been put forward (Gregory and Kirk 1978a). Various vitamin activity values for sulfur- containing thiazolidine derivatives formed in evaporated heat-sterilized milk after reaction with cysteine have been reported; thiazolidine forms had no activity in rats, 20% activity in S. ovarum (former carlsbergensis), and 60–70% activity in Neurospora sitophila compared to that of pyridoxal (Bernhart et al. 1960). The reaction with free sulfhydryl groups of milk protein was proposed to yield the formation of bis-4-pyri- doxyl disulfide (Wendt and Bernhard 1960). However, the amount of the product formed via this reaction route was estimated to be so small that the decreased vitamin activity found in proc- essed foods could hardly be explained in this way. Thus, the formation of a compound active for microbes but inactive for mammals was pro- posed (Gregory and Kirk 1977). The formed pep- tide – pyridoxal phosphate complex gave evi- dence for an acid stable pyridoxylamino struc- ture. An inactive mebolite, 6-hydroxypyridoxine HPN (Fig. 9), was found after processing plant foods rich in ascorbic acid (vitamin C). It was 538 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods suggested that the formation of HPN was the result of the oxidative degradation of ascorbic acid (Tadera et al.1986b) and it was generated by hydroxyl radicals. The pH and temperature optimum for its formation were 4–7 and 30– 50°C, respectively. 6-Hydroxypyridoxine showed neither vitamin activity in S. uvarum (Tadera et al. 1986b) or in rat (Gregory and Leatham 1990) nor antivitamin activity in S. uvarum (Tadera et al. 1986b). It was concluded that the formation of this pyridoxine conjugate causes vitamin losses during food processing, cooking and storage in the presence of ascorbic acid. However, the data of its amount and rele- vance to the vitamin B 6 losses in processed foods is still scarce. 2.2.1.4 Interaction of other food components – food digestibility The fiber content of a diet has been proposed to lower vitamin B 6 availability by binding pyri- doxine to some extent. This conclusion was based on findings that vitamin B 6 was less ef- fectively utilized from whole-wheat bread than from white bread or from fortified white bread (Leklem et al. 1980). It is was suggested that increased fecal excretion contributed to the lower vitamin availability. However, the difference in utilization efficiencies was small. Lindberg et al. (1983) demonstrated that adding wheat bran to the diet slightly lowered the bioavailability of vitamin B 6 (ca. 17%) in young men but did not adversely affect vitamin B 6 status if the vitamin intake was adequote. The bioavailability of add- ed vitamin also was decreased in rat when vita- min B 6 was added to a rice breakfast cereal (Gre- gory 1980b). Later studies of Nguyen and asso- ciates could not confirm either the inhibitory effect of pectin, cellulose or wheat bran in rat and chicks (1981b) or in vitro physical interac- tion between vitamin B 6 and eight purified polysaccharide fractions (1981a). The same kind of results were obtained with wheat bran and cellulose in rats by Hudson et al. (1988) and in humans by Shultz and Leklem (1987). No effect of dietary fiber on the vitamin B 6 status among vegetarian and nonvegetarian elderly individu- als was found in the Netherlands (Löwik et al. 1990). The net bioavailability of vitamin B 6 in an average American diet ranged from 61% to 81% evaluated by plasma PLP data and from 73% to 92% according to urinary data compared to the availability of pure vitamin (Tarr et al. 1981). The availability of vitamin B 6 was report- ed to be higher in meat and fish foods than in plant-derived foods. It was proposed, however, that dietary fiber has little or no effect on utili- zation of vitamin B 6 in humans. On the other hand, the slow digestibility of fibrous food may need even more emphasis. In general, many results of the reported stud- ies can only be interpretated with difficulty as several factors, like interactions or reactions with other compounds present in foods, physiologi- cal status and nutrition of an individual as well as different chemical forms of vitamin B 6 com- pounds, will have an impact on the vitamin uti- lization. 2.2.2 Coenzyme function The B 6 vitamers are metabolically interconvert- ed via the dephosphorylation -phosphorylation, oxidation – reduction and deamination – ami- nation reactions by enzymes like pyridoxal ki- nase, phosphohydrolases, pyridoxal dehydroge- nase and transaminases. The metabolically ac- tive form of vitamin B 6 is pyridoxal-5’-phos- phate (PLP) which serves as a coenzyme in sev- eral enzyme systems. Flavin mononucleotide requiring pyridoxal phosphate oxidase (EC 1.4.3.5), regulating the conversion of pyridox- Fig. 9. The structure of 6-hydroxypyridoxine. 539 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. ine and pyridoxamine to the active coenzyme has been suggested to be the limiting enzyme in vitamin B 6 metabolism (Kazarinoff and McCormick 1975). Vitamin B 6 dependent enzymes are mainly involved with amino acid metabolism. The most evident group of enzymes is transaminases, most of which use α-ketoglutaric as an amino group acceptor. The activation coefficient of erythrocytic aspartate aminotransferase (EAST- AC) has been widely used as a parameter to indicate the availability of PLP and thus reflect the vitamin B 6 status of a body over a long pe- riod (van den Berg et al. 1978, Löwik et al. 1990, Costa de Carvalho et al. 1996). Other enzymes which are PLP coenzyme dependent include decarboxylases, and amino acid side- chain altering enzymes. Racemaces, involved in the utilization of amino acids, have been found in certain microorganisms. In addition, vitamin B 6 is also required for tryptophan-ni- acin conversion (kynereninase) and the synthe- sis of heme (δ-amino levulenic acid synthetase) and glycogen phosphorylase. The role of PLP in lipid metabolism converting linoleic acid to arachidonic acid and in steroid hormone recep- tors has also been suggested. Pyridoxal phosphate coenzyme is bound to various apoenzymes by the formation of a Schiff’s base between the keto-carbon of the coenzyme and the ε-amino group of a lysyl res- idue of the apoencyme. Reactions catalyzed by PLP seems to involve the binding of amino acid substrate to the internal aldimine group by the transaldimination reaction to form an external aldimine (Snell 1990). Delocalization of the elec- trons in the enzyme-substrate complex leads to the formation of a carbanion at the α-carbon of the substrate which weakens its bonds resulting in the heterolytic cleavage of one of the three bonds in the α-carbon. The bond located in α- carbon which is cleaved is determined by the particular PLP-dependent enzyme; in the decar- boxylation of amino acids, the formed enzyme- substrate complex weakens the bond of the car- boxyl group resulting in the loss of carbon diox- ide. 2.2.3 Proposed role(s) in hyperhomocyst(e)inaemia The increased homocysteine (Hcy) level in plas- ma seems to be a risk factor for vascular diseas- es. The concentrations of homocysteine rise in chronic renal failure (Wilcken and Gupta 1979) and elevated levels are associated with athero- sclerosis and coronary artery disease. Increased plasma levels are considered to be those over 15µmol/l while normal Hcy level varies from 5 to 15µmol/l depending on age and sex. Certain diseases and medical treatments as well smok- ing and the excessive consumption of coffee are known to raise Hcy plasma levels. Dietary fats, fruits, vegetables, vitamins and the amount of their consumption are expected to affect the ho- mocysteine levels in the plasma. The role of gen- otype is also under investigation. As metabolic evidence for low serum folate, vitamin B 12 and vitamin B 6 concentrations and elevation of se- rum homocysteine level in elderly people was found in studies carried out in Belgium, Germa- ny and the Netherlands, early detection of tissue deficiency of these vitamins was recommended (Joosten et al. 1993). In humans, dietary supple- mentation with vitamin B 6 enhances the glutath- ione activity and the effect is most apparent in elderly people (Grimble 1997). Cravo et al. (1996) suggested that the chronic alcoholism may increase the plasma Hcy level by interfer- ring with the disposal of homocysteine. Chronic alcoholism is known to interfere with one-car- bon metabolism which involves folate and vita- min B 6 . Significantly lower PLP level in serum and folate level in red blood cells were found among alcoholics than those of controls. Inter- action of PLP in the the desulfhydration route of cystein and homocystein to pyruvate and 2-ox- obutyrate has been proposed (Harper et al. 1979). Plasma folate, vitamin B 12 and pyridoxal phosphate were inversely associated with the plasma homocysteine concentration (Robinson et al. 1995, Selhub et al. 1995). Folate was the most effective agent for reducing total plasma homocysteine levels in cases of a vitamin B 12 540 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods deficiency in rats and in patients. Vitamin B 6 had no effect on fasting plasma total concen- tration but did reduce post-methionine load Hcy levels in plasma. Similar results were found in the Framingham Heart Study; homocysteine ex- hibited a strong inverse association with plas- ma folates but weaker associations with plas- ma vitamin B 12 or vitamin B 6 . Verhoef and as- sociates (1996) suggested that the impairment of remethylation of homocysteine to methionine which is dependent on folate and vitamin B 12 was the predominant cause for high homo- cysteine levels and not the vitamin B 6 related transsulfuration route. No inverse correlation between plasma levels of homocysteine and vitamin B 6 was found in their study performed in Boston. In general, plasma folate, cobalamine and pyridoxal phosphate levels are inversely related to homocysteine concentration. More detailed information concerning the factors and metabolic routes effecting homocysteine levels in plasma need further research. For instance, the clinical benefit of vitamin supplementation has not yet been demonstrated. 2.2.4 Recommended dietary intake The requirement for vitamin B 6 increases as the intake of protein increases. This relationship is connected to the role of pyridoxal-5’-phosphate in amino acid metabolism. The recommended dietary allowance for vitamin B 6 is 2.0mg/day for men and 1.6mg/day for women (Food and Nutrition Board 1989) which is adequate for an avarage protein intake of 100g/day for men and 60g/day for women. An additional allowance of 0.5mg of vitamin B 6 per day is recommended during lactation. The recommended daily intake for vitamin B 6 in the Nordic Countries is set at 0.015mg of vitamin B 6 per one gram protein when the pro- tein intake counts as 15 E% (NNR 1996). The same intake value is recommended for pregnant and lactating women. This recommended intake corresponds to a daily intake of 1.5mg and 1.2mg for men and women, respectively. The average requirement is set at 0.013mg/d per one gram protein and the highest recommended intake for an individual person is 50mg per day. The adequate protein intake from natural foods is considered to ensure the recommend- ed vitamin B 6 intake as vitamin B 6 compounds and proteins occur together in natural foods. Vi- tamin B 6 deficiency is rarely observed and it is assumed that the low vitamin intake is related to the people deficient in other B-complex vi- tamins too. The risk group includes elderly peo- ple whose energy and food intake is generally decreased and chronic alcoholics (Food and Nu- trition Board 1989). The amount of vitamin B 6 in food in Nordic countries is estimated to be 1.5–2.2mg of vitamin B 6 per 10MJ (NNR 1996). Nevertheless, there is still evidence for low vitamin B 6 intake among the population in the industrialized countries. PLP level in plasma samples from ca. 2500 elderly people partici- pating in the Euronut SENECA study (17 towns in 11 European countries) showed that the prev- alence of biochemical vitamin B 6 deficiency was widespread (Haller et al. 1991). Later on the marginal vitamin B 6 status of elderly peo- ple (546 persons, aged 74–76) in the SENECA study was reported by van der Wielen and cow- orkers (1996). Approximately 30% of the males and 40% of the females had dietary vitamin B 6 intake below the mean minimum requirements. A French nutrition survey study conducted in Burgundy (337 middle-aged and healthy sub- jects consisted of 157 males and 180 females) showed the vitamin B 6 intake below the French recommended dietary intake in 11% of the males and 28% of the females (Costa de Car- valho et al. 1996). Bailey and her associates (1997) reported that the risk for biochemical deficiency was greater for vitamin B 6 and B 2 than for folate, thiamine or ascorbic acid in the elderly British population studied (sixty men, eighty-five women). However, it was conclud- ed that the requirements of the elderly for vita- mins need review. 541 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 2.3 Occurrence and distribution of vitamin B 6 in foods Poultry, fish, offals, pork and egg are considered as rich sources of dietary vitamin B 6 . Good sources are also unmilled grain and rice, soy beans, oats and nuts whereas dairy products and red meat are assumed to be poor sources. The bioavailability of vitamin B 6 among different foods, however, varies widely. Food groups can be divided into two categories according to bound vitamers as well as the distribution of free and phosphorylated vitamers. The main component(s) in foods of the plant origin, espe- cially in vegetables and cereals, is pyridoxine which is bound to a carbohydrate moiety (Kabir et al. 1983a, Gregory and Ink 1987, van Schoon- hoven et al. 1994, Sampson et al. 1995, Samp- son et al. 1996). The content of either glycosylat- ed or free pyridoxine may be considered to be characteristic for the plant-derived foods while phosphorylated pyridoxal and pyridoxamine form the main vitamer fraction in flesh and dairy foods. Ca. 80% of the total vitamin B 6 activity in fresh animal tissues is derived mainly from phosphorylated pyridoxal and pyridoxamine and to a smaller extent from free pyridoxal and pyri- doxamine (Rabinowizt and Snell 1948, Polan- sky and Toepher 1969). Pyridoxine phosphate (PNP) is generally assumed to be a minor vita- min B 6 fraction in foods (Vanderslice et al. 1980, Tadera and Naka 1991). Variations in the sample treatment procedures (e.g. extraction media and conditions) and the later improvements in methodology, especially in modern liquid chromatography, make it diffi- cult and in some cases not even sensible to com- pare the vitamer distribution results derived from different era and research groups. 2.3.1 Meat, offals and fish Main B 6 vitamers present in flesh foods were either free or phosphorylated pyridoxal and py- ridoxamine, the proportion of free pyridoxine was smaller. Bowers and Craig (1978) found that the the sum of pyridoxamine and pyridoxal formed ca. 80% of the total vitamin B 6 content in chicken meat and Polansky and Toepfer (1969) obtained a figure of 92–95% the major vitamer being pyridoxal except in liver which was rich in pyridoxamine. A similar distribution was re- ported for beef, lamb, and pork. Phosphorylated forms of pyridoxal (63%) and of pyridoxamine (35%) predominated in raw chicken breasts (Olds et al. 1993), the total vitamin B 6 content being 0.4–0.6mg/100mg (Ang 1980, Olds et al. 1993). Free pyridoxamine counted only for 2% and no pyridoxine was reported by Olds et al. (1993). Almost an equal amount of total pyri- doxal and pyridoxamine (40%) was present in turkey meat while free pyridoxine accounted for as much as 20% of the total vitamin B 6 (Bowers and Craig 1978). The total vitamin B 6 content in fresh pork meat varied from 0.46 to 0.57 mg/ 100g, the major free vitamers being pyridoxam- ine and pyridoxal (Esteve et al. 1998). The vita- min content of flounder, oyster and shrimp was derived mainly from pyridoxal and pyridoxam- ine thus resembling meat. The main vitamer in the canned sockeye salmon and tuna was report- ed to be pyridoxamine (Polansky and Toepfer 1969). This is probably due to the transamina- tion during and after the canning process. 2.3.2 Plant foods As much as two thirds of the total vitamin activ- ity in some plant food is reported to be derived from bound pyridoxine; such typical materials are carrot, orange (Gregory and Ink 1987, Gre- gory 1988a) and cereals (Sampson et al. 1995, 1996). Banana was also rich in free pyridoxine. On the other hand, certain nuts and almonds con- tained only a low level of glycosylated pyridox- ine and the main form was free pyridoxine. The native glucosidase activity in almonds (Chiari et al.1997) may explain this phenomenon. The neutral pH optimum β-glucosidases of almonds 542 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods are reported to be capable of hydrolyzing a number of plant glucosides (Lai et al. 1992). A smaller amount of free and phosphorylated py- ridoxal was found in plant-derived foods. Pyri- doxamine content was reported to vary in dif- ferent materials; e.g. in tomato the content of total pyridoxamine is significant different to that of carrot. Kabir and his group (1983a) measured the vitamin B 6 content of several food items micro- biologically before and after β-glucosidase treatment. The amount of glucosidic bound py- ridoxine in plant foods varied from 5 to 82%; in general, high values were found in grains and legumes. Processed broccoli and cauliflower samples gave the higher values compared to those of the raw material. No explanation for the increased values was given. Sampson and coworkers (1995) showed that pyridoxine glu- coside and free pyridoxine consisted the main B 6 vitamers in wheat representing ca. 35–81%, and 21–64% of the total vitamin content in three wheat cultivars analyzed. In their study the amount of PLP and PMP accounted for 12% of the total vitamin content in one cultivar and only pyridoxine and its glycoside were present in two other wheat cultivars. A significant var- iation in the content of pyridoxine and pyridox- ine glucoside was found in 22 North American and Canadian wheats (Sampson et al. 1996). American hard wheat contained more vitamin B 6 , mostly in the form of pyridoxine glucoside, compared to Canadian hard wheat. The mean values for pyridoxine glucoside (67%), pyridox- ine (33%) were given while other vitamers were not reported. Glycosylated pyridoxine formed the major B 6 vitamer fraction also in legumes. Pyridoxine glu- coside accounted for 40%, 63% and 75% of the total vitamin B 6 in haricot beans, lentils and chick beans, respectively (Sierra and Vidal-Valverde 1997). The amount of free vitamers after acid phosphatase digestion varied 5–17% for pyridox- amine, 15–35% for pyridoxal, and 5–8% for py- ridoxine in their samples. 2.3.3 Milk, milk products and eggs The total vitamin B 6 content of fluid milk ranged from 0.03mg to 0.05mg/100g in the studies of Gregory and Mabbit (1959, 1961), Polansky and Toepher (1969), Dong et al. (1980), Coburn and Mahuren (1983), Laukkanen et al. (1988), Tou- kairin-Oda et al. (1989) and Sieber et al. (1996). Vitamin B 6 activity in fresh fluid milk consisted of both free and phosphorylated vitamers, pyri- doxal being the predominant form (Sieber et al. 1996, Argoudelis 1997) whereas pyridoxal phos- phate and pyridoxamine phosphate, and free pyridoxamine each formed less than 10% of the total content. A high proportion of PLP (36%) in milk has also been reported by Toukairin-Oda et al (1989). The results of Coburn and Mahuren (1983) showed that PLP was the main vitamer B 6 fraction (51% of the total amount) in fresh bovine milk while increased amount of PMP and pyridoxal were found in pasteurized milk. The amount of free pyridoxine in milk was low (Gre- gory and Mabbit 1961) or pyridoxine was not found (Rabinowitz and Snell 1948, Sieber et al. 1996, Argoudelis 1997). Lim et al. (1982) re- ported that pyridoxine formed almost 80% of the total vitamin B 6 content in non-fat dry milk, a result which is contradictory to the results pub- lished by other researchers. Neither the process- ing procedure nor possible enrichment of the material, however, was described in their paper. Practically no change in the vitamer distribution between fresh milk and skimmed milk powder was found in the results of Polansky and Toepher (1969). The change in the vitamer distribution was observed in “processed” milk products; the amount of pyridoxamine and PMP in the con- densed/evaporated or pastorized milk were in- creased compared to that of fresh milk. The in- creased proportion of pyridoxamine, probably derived from pyridoxal or its phosphorylated form, was present in cheeses (Polansky and Toepfer 1969, Coburn and Mahuren 1983, Bitsch and Möller 1989b). As only a small amount of PMP was present in fresh milk, it was suggested to be formed during the process (Gregory and 543 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Mabbit 1961). A short-term microwave heating of fresh milk (83°C for 4min) did not increase the amount of pyridoxamine according to Sie- ber et al. (1996). A lower total vitamin B 6 con- tent was observed in cultured butter milks than in milks whereas a higher level of vitamin B 6 was present in yoghurts (Polansky and Toepher 1969, Laukkanen et al. 1988). The vitamin B 6 content of human milk resem- bled that of cow’s milk; the total amount of 0.02mg/100g was reported by Morrison and Driskell (1985) and the proportion of pyridoxal, pyridoxamine, and pyridoxine was 81%, 13%, and 6%, respectively. The lower proportion of phosphorylated PL was present in human milk compared to that of bovine milk; free pyridoxal formed the main portion (65–83%) while PLP (6–28%), pyridoxine (0–4%), and PMP (1–2%) accounted for the rest of the vitamin B 6 activity (Coburn and Mahuren 1983, Hamaker et al. 1985, Bitsch and Möller 1989a). Only traces (less than one percent or not determined) of free pyridoxamine and no PNP was found. Raw chicken egg yolk contained 0.44mg vi- tamin B 6 per 100g and the major vitamer was pyridoxal-5’-phosphate (96%). Pyridoxamine (2%) and pyridoxamine-5’-phosphate (2%) were minor components. The amount of free pyridox- ine and pyridoxal were traces or not detected (Argoudelis 1996). The same kind of vitamer distribution was published by Toukairin-Oda et al. (1988). 2.4 Analysis of vitamin B 6 compounds Traditional sample extraction procedures and their modifications for vitamin B 6 analysis are based on mineralic acid hydrolysis combined with a heating step such as boiling in a water bath or an autoclaving process (AOAC 1995a, 1995b, COST91). An enzymatic digestion is commonly added after the heating treatment to disintegrate the sample matrix and to ensure the quantitative hydrolysis of phosphate esters (van den Berg et al. 1996). After these pretreatment steps only pyridoxine, pyridoxal, and pyridox- amine can be measured as phosphorylated forms are hydrolyzed to their corresponding free vita- mers. Mineralic acid sample disintegration and extraction combined with an autoclaving proc- ess hydrolyses, at least partially, the glycosidic bond present in glycosylated vitamin B 6 forms. The minor glucosidase activity present in mixed commercial enzyme preparations like Takadia- stase caused the hydrolysis of glycosidic bond as well (van Schoonhoven et al. 1994). Thus, the traditional extraction procedures with or with- out mixed enzyme preparations will most likely release the free B 6 vitamers from their corre- sponding phosphate and glycoside conjugates. Normally, the total vitamin B 6 content as the sum of free vitamers is reported after the microbio- logical assay. The content of individual free vi- tamers can be measured if cation-exchange chro- matographic separation on an open column is performed prior to the microbial assay. The in- formation on the original distribution of phos- phate esters and glycosidic vitamers in the sam- ple, however, is then lost. The use of the modern liquid chromatogra- phy allows the measurement of individual vita- mers, and the original vitamer distribution can be achieved. If the original vitamer distribution of the material is needed (free, phosphorylated and glycosylated vitamers), the procedures, con- ditions and reagents used in the sample pretreat- ment and chromatographic separation should be carefully chosen to avoid unnecessary hydroly- sis and interconversion of the vitamers but still maintaining the quantitativity of the extraction. The use of a heating step in the extraction pro- cedure helps the disintegration of the matrix but samples rich in protein inevitably produce chang- es in vitamer distribution due to transamination and dephosphorylation reactions. This should be taken into account when the individual vitamer data from studies using different methodology are compared. If free, phosphorylated as well as glycosidi- 544 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods cally bounded B 6 vitamers are to be measured, non-mineral acids or buffer solutions should be used in the sample extraction. All pretreatments and chromatographic separations should be per- form at room temperature or lower. Testing of the enzyme activity including also the probable minor side-activities of the preparations should be considered. 2.4.1 Extraction and hydrolysis techniques Extraction procedure for measurement of vita- min B 6 compounds usually includes hydrolysis in acidic media and is followed by a sample matrix disintegration with enzymatic hydrolysis. The aim of the acid treatment of the sample is to release protein-bound vitamin B 6 forms. Depend- ing on the extraction acid (mineral acids vs. non- mineral acids) and conditions (temperature and time) used, free vitamers are also released from their corresponding phosphate esters or glyco- sylated derivatives. The release of vitamers is normally completed by performing an enzymat- ic digestion after acid treatment. Proteases like pancreatin and pepsin have been used to verify the efficiency of the hydrolysis technique or in vitro digestion (Ekanayake and Nelson 1988). The hydrolysis of beef or soybean samples was verified by protease treatment; no release of B 6 vitamers was observed after the protease diges- tion compared to nonenzymatic treatment. Thus, the nonenzymatic sample treatment was consid- ered to be quatitative (Kabir et al. 1983). Ex- traction procedures for vitamin B 6 compounds are thoroughly reviewed in the papers of Grego- ry (1988a) and of Rizzolo and Polesello (1992). 2.4.1.1 Mineral acid extraction Thermal extraction in acidic solution is the most commonly used procedure to release protein- bound vitamers and to hydrolyze the phosphor- ylated forms. These methods are based on the work of Atkin et al. (1943) in which it was shown that maximal extraction for most biological sam- ples was achieved by autoclaving the samples in 0.055N sulfuric acid at 128°C for 60 minutes. A more concentrated acid solution, 0.44N, was needed for cereal matrix. Sulfuric acid was re- placed with hydrochloric acid, and the most suit- able condition for the extraction was suggested to range from pH 1.7 to 1.8 (Rubin et al. 1947). Hydrochloric acid extraction was further evalu- ated by Rabinowitz and Snell (1947); an auto- claving period of one hour with 0.055M hydro- chloric acid was adequate for hydrolyzing pyri- doxal-5’-phosphate whereas liberation of pyri- doxal from its phosphate was slower with more concentrated acid solutions. Several hours (4–5h) were needed to release pyridoxamine from its phosphate ester. Only 50% of synthetized PMP was hydrolyzed after 10h at 100°C in 6N sulfuric acid (Peterson et al. 1955). Rather vigorous extraction condition, 2M hydrochloric acid, was needed for bound vitam- ers present in rice bran matrix. Later on these vitamin B 6 compounds were identified as glyco- sylated derivatives of pyridoxine. On the basis of these studies, the sample pretreatment for the present microbiological assay was developed; extraction with 0.44M hydrochloric acid (two hours at 121°C ) for plant foods and with 0.055M hydrochloric acid (five hours at 121°C ) for an- imal products (Toepher and Polansky 1970). The modification of this method is still in use as a procedure recommended by AOAC (1995a, b). Mineral acid extraction with hydrochloric acid, the pH of the solution adjusted first to 1.7 and then increased to 4.6, without a thermal proc- ess was used for determination of vitamin B 6 in fortified infant formulas. High recovery values were obtained with this extraction procedure (Ayi et al. 1986). Autoclaving for 2 hours in 0.5M hydrochloric acid solution was presented as a sample disintegration step for pasta products. Recovery values ranged from 83% to 92%, low- est values achieved with pyridoxal (Tolomelli et al. 1991). A method applying nitric acid as an extractant for vitamin B 6 has also been published; pharmaceutical multi-vitamin preparations were dissolved in dilute nitric acid solution (1mM) and the mixture was heated on a boiling water bath 545 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. for 30 minutes. After filtration the sample was ready-to inject to liquid chromatograph (Callm- er and Davies 1974). Relatively low recovery (97%) for pharmaceutical preparation but rather good precision values (as a standard deviation for peak height values of repeated injections) were reported. At present, mineral acid extraction using ei- ther hydrochloric acid (Yasumoto et al. 1977, Addo and Augustin 1988, Ekanayake and Nel- son 1988, Steiner et al. 1993, Bognar and Olli- lainen 1997) or sulfuric acid (Wehling and Wet- zel 1984, Rees 1989, COST 1991, Olleta et al. 1993, Agostini and Godoy 1997, Esteve et al. 1998) are combined with a heating treatment (autoclaving or boiling water-bath) to ensure extraction efficiency. However, interferring compound(s) mainly preceding pyridoxal after mineral acid hydrolysis may complicate the in- terpretation of the reversed-phase chromatogram (Bognar 1985, COST1991). 2.4.1.2 Chemical deproteinating agents Many extraction procedures using a non-miner- al acid for the release of the protein-bound B 6 vitamers from biological sample matrix have been reported in the literature; sample hydroly- sis with metaphosphoric acid, perchloric acid, sulfosalisylic acid, trichloroacetic acid or with certain buffers enables the extraction of phos- phoric acid esters and glycosidically bond con- jugates of vitamin B 6 in their intact forms. Sev- eral sample pretreatment procedures are pub- lished and their use is ideal when the “true” vi- tamer distribution of the sample is needed. How- ever, many of these methods still lack proper validation data on the extraction efficiency. Trichloroacetic acid (TCA) Trichloroacetic acid has been largely used for protein denaturation in the measurement of vi- tamin B 6 compounds. The main advantages for the use of trichloroacetic acid are its extraction efficiency, it has some ionic characteristics which may enhance the resolution of measured analytes in reversed-phase liquid chromatogra- phy, and it can be removed from the aqueous phase by organic solvent extraction with diethyl ether. The use of TCA or SSA has been recommend- ed for precipitation of protein and protein hy- drolysates. In addition certain enzymes, like α- chymotrypsin (Saidel and Modapallimattam 1970) and pepsin (Alfred and Narasinga Roo 1971), maintained their enzyme activity in di- lute TCA solutions. Denaturation and inactiva- tion occurred when 5% or more concentrated acid solution was used. At higher concentrations, TCA is suitable for removing the enzyme pro- tein by precipitation after the enzymatic hydrol- ysis. So the enzyme activy in the extraction so- lutions containing trichloroacetic acid should be taken into account. The compatibility of trichlo- roacetic acid in the microbiological assay should also be verified. Trichloroacetic acid extraction has been ap- plied to clinical studies (Coburn and Mahuren 1983, Hachey et al. 1985, Shephard et al.1987, Shephard et al. 1989, Mahuren and Coburn 1997), and to human milk (Hamaker et al. 1985) and food and feed analysis (van Schoonhoven et al. 1994) as well as measuring the sugar-deriva- tives of pyridoxine (Ink et al. 1986, Sierra and Vidal-Valverde 1997). As the phosphate ester and glucosidic bond were not hydrolyzed by TCA the extraction in lowered or ambient temperature, this extraction enabled the determination of free, phosphorylated and glycosylated compounds. In the above mentioned studies, the concentration needed for complete liberation of vitamers nor- mally ranged between 5–10% (w/v). Ubbink and coworkers (1986) showed that albumin bound PLP was quantitatively liberated by a trichloro- acetic acid concentration of 3% which is some- what more dilute than the solutions used by the other researchers. The advantage of using TCA for the separation efficiency of vitamers in the reversed-phase chromatography was stated by Hamaker et al. (1985); the resolution of pyri- doxal and its phosphate ester was dependent on the presence of TCA. The use of trichloroacetic acid as an extract- ing agent for vitamin B 6 compounds has been evaluated with both individual and collaborative 546 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods studies, thus the validity results for this extrac- tion agent are available (Hollman et al. 1993, Berg van den et al.1996). An acceptable repeat- ability as well as reproducibility value for the total vitamin B 6 content was achieved by using a 5% trichloroacetic acid solution followed by enzymatic hydrolysis for sample matrix disinte- gration and dephosphorylation. This data was derived from the work of an European collabo- rative study in which eleven laboratories partic- ipated during 1989–1997. Perchloric acid (PCA) Perchloric acid has been widely applied as a sam- ple deproteinizing agent. The acid concentration used for the extraction of vitamin B 6 compounds varied considerably; 0.1–0.5M solutions have been used for different foods depending on the protein content (Möller and Bitsch 1988, Bitsch and Möller 1989a, Chase et al. 1992), 0.4M for human plasma (Mascher 1993, 1997), and 1M for biological samples (Pierotti et al. 1984, Tsuge et al.1988, Kimura et al. 1996, Argoudelis 1997, Tsuge 1997). Reynolds and Brain (1992) used PCA only for deproteinizing the sample extract and not for the sample matrix disintegration. Perchloric acid has been used for measurement of both glycosylated derivatives and phosphate esters in their native form (Schramm and Bitsch 1993). Difficulties in the extraction were as- sumed to be derived from the protein-binding properties of pyridoxal phosphate still present in the dilute perchloric acid solutions (Edwards et al. 1989). An advantage of PCA is that it can be removed by precipitation with alkali, and thus diminishing the risk for the incompatibility of the sample extract/solvent with the chromato- graphic separation or with the enzymatic hydrol- ysis. Neutralizing the perchloric acid extraction solvent to pH of 7.5 with dilute potassium hy- droxide solution precipitates potassium perchlo- rate. Even alkali labile pyridoxal is assumed to be stable during this procedure. Sulfosalicylic acid (SSA) Some earlier studies applied sulfosalicylic acid as an acid extraction agent in amino acid meth- odology (Mondino et al. 1972). Later on Vanderslice and his colleagues (1980) published a routine liquid chromatographic method which included a sulfosalicylic acid extraction proce- dure for vitamin B 6 compounds. A 5% sulfosali- cylic acid solution was confirmed to be an ef- fective deproteinizing agent. The prevention of interconversion of B 6 vitamers was also ob- served. High vitamer recovery values for sever- al different food materials were reported. Since their work sulfosalicylic acid extraction has been used for several food items (Vanderslice et al. 1984, Gregory and Feldstein 1985), for ready- to-eat cereals (Vanderslice et al. 1981b) and for poultry meat (Olds et al. 1993). As sulfosalicyl- ic acid has a native fluorescence in the excita- tion/emission wavelength related to vitamin B 6 compounds it has to be removed before fluoro- metric measurement. Sulfosalicylic acid was re- moved with a purification column containing anion-exchange resin in the work of Vanderslice and coworkers (1981b). However, this purifica- tion step diluted the sample extract. This could be a disadvantage when the samples of low vita- min content are to be analyzed. Metaphosphoric acid (MPA) One of the first extraction methods for vitamin B 6 using metaphosphoric acid was published by Loo and Badger (1969). Rather low acid con- centration, 0.8% (0.17M), was needed for pro- tein denaturation in the assay of vitamin B 6 com- pounds in brain tissue. It was stated that the re- moval of lipids prior to metaphosphoric acid extraction effectively released B 6 vitamers. High- er acid concentrations were used in subsequent studies. A 0.5M metaphosphoric acid hydroly- sis was applied to chicken meat sample at tem- peratures below 15°C followed by a heating step in a boiling water bath for 4 minutes (Ang et al. 1988). High recovery values were reported when B 6 vitamers and pyridoxic acid of biological fluid and tissue samples were liberated with 5% to10% metaphosphoric acid solutions by Sharma and Dakshinamurti (1992). The method was suggest- ed to be suitable for analyzing biological liquids with a low B 6 vitamer content. Sampson and 547 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. coworkers (1995, 1996) described a deprotein- izing step with cold 5% acid solution for meas- uring the content of free and glycosylated vita- mers of different wheat cultivars. However, a low recovery for added pyridoxal and pyridoxal phosphate was reported Buffer solutios and water Dilute buffer solutions have been used for ex- traction of B 6 vitamers from various foods and other biological materials. Neutral phosphate buffer solution (0.1M, pH7.0) extracted pyri- doxal, pyridoxamine and their phosphate esters from rat tissue samples after which the protein precipitation was performed with 0.6M perchlo- ric acid solution (Gregory 1980a, Gregory et al.1981). High recoveries were obtained for all vitamers except for pyridoxal phosphate in cer- tain food matrix. Low recovery for PLP was as- sumed to be due to the pre-column derivatiza- tion with semicarbazide reagent or some impu- rities quenching the fluorescence intensity in the detection. Extraction with a neutral 0.1M phos- phate buffer was applied to the sample prepara- tion for determination of pyridoxine in multivi- tamin preparations (Razagui and Barlow 1989). This extraction solvent was also suitable for the direct spectrofluorometric measurement. Buffer solutions (pH 6.8) were used in sample extrac- tion for measurement of glycosylated vitamin B 6 in foods; vegetables, fruits, nuts, grains and leg- umes as well as animal product samples were stirred in 0.1M phosphate buffer solution for two hours at room temperature. The extraction meth- od described was suggested to be usefull for var- ious food matrix (Kabir et al. 1983a). In bioa- vailability studies of glycosylated pyridoxine Kabir et al. (1983b) used a 0.1M phosphate buff- er solution (pH5) for extraction of urine or faces samples. Extraction with an acetate buffer of pH 4.5 has been utilized for clinical samples (Hefferan et al. 1986, Bötticher and Bötticher 1987) , for fortified breakfast cereals (Gregory 1980c), vi- tamin preparations (Bühnert 1988) as well as for food samples (Reitzer-Bergaentzle et al. 1993). Acidic acetate solution was easily combined with the followed enzymatic hydrolysis procedure and the method was suitable for the simultaneous measurement of free B 1 , B 2 and B 6 vitamers in plasma and serum samples. The validity of the acetate extraction used by Reitzer-Bergaentzle et al. (1993) preceeding the pre-column deriva- tization of B 6 vitamers into pyridoxine was fur- ther evaluated by the collaborative study per- formed in France (Bergaentzle et al. 1995). A lower pH acetate buffer was used for multivita- min preparate; sample was sonicated with a di- lute solution of acetic acid (0.5M) and triethyl- amine (40mM) at the pH of 3.6 (Lam et al. 1984) or with liquid chromatography’s mobile phase (acetonitrile – 10mM phosphate buffer – triethyl- amine, 8:91.5:0.5, pH 3.8)(Maeda et al. 1989). Good accuracy and recovery values for pyridox- ine were obtained in both studies. Boiling water (Tadera et al. 1986a) or aque- ous ethanol (75%)(Tadera et al. 1988) was used for extraction in isolation of glycosylated pyridoxine(s) in cereal samples. Boiling 75% ethanol was considered to possess sufficient en- zyme inactivating capability since neither pyri- doxine glycoside or pyridoxine was formed from exogenous pyridoxine or glycosylated pyridox- ine, respectively. Also transglycosylation to form oligoglucosides caused by β-glucosidase was ruled out. However, the efficiency of ethanol – water -mixture for the extraction of cellobiocyl and glucotriocyl derivatives of pyridoxine in rice bran was not stated. Good recovery and repeata- bility values for pyridoxine content of multivi- tamin preparations and elemental diets were re- ported using a methanol – water -mixture (80:20) (Amin and Reusch 1987), water (Wang and Hou 1988a, Belal 1989) or 2M sodium chloride solu- tion (Iwase 1992) as a dissolving agent. In these studies the extraction procedure could be sim- plified as the vitamin B 6 content was related only to one vitamer, pyridoxine. 2.4.1.3 Enzymatic hydrolysis Liberation of B 6 vitamers from their correspond- ing phosphate esters or glycosides can be per- formed with acid hydrolysis combined with an autoclaving step or with an enzymatic hydroly- 548 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods sis. Rather long autoclaving time, ca. 4–5 hours, in 0.055M hydrochloric acid solution was need- ed for complete hydrolysis of PMP (Rabinowitz and Snell 1947). Glycosylated pyridoxine is ex- pected to be hydrolyzed more easily in autoclav- ing with mineralic acid solution than PMP. De- phosphorylation of PLP and PMP using an en- zymatic digestion is normally performed over- night and the most commonly used enzymes in vitamin B 6 analysis are phosphatases or the mixed enzyme preparations like Takadiastase and Claradiastase. The time required for releas- ing the bound pyridoxine with β-glucosidase treatment varied from 30min to several hours depending on digestion conditions. The enzyme treatment, especially with mixed enzyme prepa- rations, also enhance the extraction efficiency due to the sample matrix disintegration by sev- eral enzyme activities. Dephosphorylation Acid phosphatase preparation has been used for dephosphorylation of PLP and PMP present in food (Addo and Augustin 1988, Reitzer-Ber- gaentzle et al. 1993) and plasma samples (Gui- larte 1983, Hefferan et al. 1986, Mascher 1993, Mascher 1997). The amount of enzyme added to extraction/sample solution ranged from 16– 100U/g for food samples and 0.4–2U/g for plas- ma sample, the pH of the digestion solution be- ing 4.5 to 4.7. Earlier studies of Morrison and Driskell (1985) showed that hydrolysis of hu- man milk samples for 1 h at 37°C resulted in incomplete dephosphorylation. An incubation period of 16h was needed to liberate pyridoxal from its phosphate ester in plasma samples (Mascher 1993) which was a longer period than reported in many other papers. Thus prolonged incubation time, 12h to overnight at 37°C, to ensure the proper hydrolysis was recommend- ed. An acid phosphatase treatment combined with conversion of pyridoxal and pyridoxamine into pyridoxine is well documented including the collaborative study parameters (Reitzer-Ber- gaentzle et al. 1993, Bergaentzle et al. 1995). Hydrolysis based on alkaline phosphatase has been carried out after perchloric acid extraction. Next to precipitation of perchlorate with alkali, the pH of the sample solution match more easily with alkaline phosphatase hydrolysis omitting/ reducing the need for readjustment of pH. Rela- tively short incubation times were reported for alkaline phosphatase treatment; 30min at 25°C (Möller and Bitsch 1988, Bitsch and Möller 1989a). Mixed enzyme preparations, like Takadia- stase or Clarase, normally contain enough phos- phatase activity for dephosphorylation of pyri- doxal phosphate and pyridoxamine phosphate. The advantage of the use of a multi-enzyme pre- parate, Claradiastase, was reported by Bötticher and Bötticher (1987). The enzyme preparate used in their application contained several enzyme activities (diastase, proteinase, phosphatase) making it suitable for simultaneous extraction of B 1 , B 2 and B 6 vitamers in blood samples. Es- teve and cooperators (1998) applied the AOAC extraction procedure for thiamine and ribofla- vin to release vitamin B 6 compounds in fresh pork meat; this method included an enzymatic digestion with Takadiastase. A good accuracy estimated through recovery assays was report- ed. The use of Takadiastase has been proved to yield quantitative hydrolysis of PLP and PMP; the efficiency of dephosphorylation was estimat- ed by measuring the coefficient of variation and the average percentages of recovery/conversion (van Schoonhoven et al. 1994). Takadiastase hydrolysis was further evaluated by the intercal- ibration and collaborative studies carried out by BCR (van den Berg et al. 1996). It should be noted, however, that prolonged sample digestion with Takadiastase may induce transamination of pyridoxal to pyridoxamine changing the origi- nal vitamer distribution. Claradiastase hydroly- sis after mineral acid treatment did not yield higher vitamer levels in flour or vegetable sam- ples. On the contrary, a decreased amount of pyridoxal were observed with some samples (Bognar 1985). This was assumed to be a result of transamination or related reactions of pyri- doxal. Acid hydrolysis followed by the fungal amylase Clarase digestion was applied to forti- fied cereal products by Wehling and Wetzel 549 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. (1984). A high reproducibility and accuracy for the analytical method was reported. Neverthe- less, pyridoxine was the only B 6 vitamer found in samples analyzed. Mixed enzyme preparations used for vitamin B 6 analysis, like Takadiastase or Clarase, include several different enzyme activities. Commercial Takadiastase preparation (Serva 35740, Germa- ny) contained α- and β-amylase, adenosine deaminase, nuclease Si, and ribonuclease Ti ac- tivity. However, a negligble activity for alkaline phosphatase has been found. Cellulase, α-amy- lase, invertase, peptidase, phosphatase and sul- fatase activity were specified for Claradiastase (Fluka 27540, Switzerland). Enzyme mixture preparations Claradiastase and Takadiastase have resulted in a good extraction efficiency and their use also in multivitamin extraction was pro- posed (Steiner et al. 1993, van den Berg et al. 1996). A drawback of these mixed preparations can be that the enzyme activity varies from batch to batch and the amount of side-activities is sel- dom known by the user (Bognar and Ollilainen 1997). Thus, the miscellaneous results in litera- ture may reflect the variability of enzymes used in different studies. Hydrolysis of glycosylated vitamers Glucosidase treatment of sample extracts has been carried out in several studies in order to measure the amount of glycosylated B 6 vitam- ers in samples of plant origin (Kabir et al. 1983a, Kabir et al. 1983b), to test the proposed meth- odology (e.g. Tadera and Naka 1991) or to con- firm the isolated bounded vitamer form (Tadera et al. 1986a). Hydrolysis by β-glucosidase has been performed at pH 5 in phosphate or in ace- tate buffer for 2–5 hours at 37°C (Kabir et al. 1983a, Kabir et al. 1983b, Sampson et al. 1995, Sierra and Vidal-Valverde 1997). The increase in pyridoxine after this digestion step is mainly considered to be due to hydrolysis of 5’-O-(β- D-glucopyranosyl)pyridoxine, the bound B 6 vi- tamer first isolated from rice bran by Yasumoto et al. (1977). The structural evaluation of 5’-O- (β-D-glucopyranosyl)pyridoxine was confirmed by enzyme treatment as this structure was vul- nerable to the action of β-glucosidase and not to that of α-glucosidase. Further structural eluci- dation revealed that also three other pyridoxine glycosides present in rice could be hydrolyzed to free pyridoxine and sugar moieties with β-glu- cosidase (Tadera et al. 1988). In vitro studies of Kawai et al. (1973) clearly showed that both the formation and hydrolysis of pyridoxine-α-gly- coside are catalyzed by α-glucosidases. These α-glucosidase enzymes are particularly distrib- uted in the bacteria genera Sarcina and Micro- coccus (Kawai et al. 1972a, b, Kawai et al. 1973) and, the transglycosidation to vitamin B 6 by these microorganisms has been thoroughly studied (Ogata et al. 1969a, 1969b, Kawai et al. 1971b). The synthesis of pyridoxine glucoside by β-glu- cosidase of rice bran has also been reported (Iwa- mi and Yasumoto 1986). 2.4.2 Chromatographic techniques Ion-exchange chromatography or ion-pair re- versed-phase chromatography are the most com- monly used techniques for the separation of vi- tamin B 6 compounds due to their pH-depended ionic nature. Ionic interaction between the cati- onic pyridinium or methylamino moiety and the ion-pair reagent or cation-exchange support ena- bles their controlled retention. Partition chroma- tography of ionic B 6 vitamers using reversed- phase column packings (e.g. octadecyl or octyl phases) usually suffers from the poor retention for PLP and pyridoxamine. The anionic proper- ties of the analytes can only be utilized with the polymer based or related column packings as the dissociation constants of B 6 vitamers lies nor- mally outside the stability range of the normal silica based column packings. Most of the recently published reversed- phase chromatographic procedures for vitamin B 6 compounds have been performed using chem- ically bounded phases (CBP), the most preva- lent packing being the octadecyl phase (Table 3). The improvements in column packing mate- rials, that is the synthesis of the silica base and the chemically bonded phases, have enabled a 550 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Table 3. Conditions for liquid chromatographic determination of B 6 vitamers. Sample preparation Stationary phase Mobile phase Detection Samples/ Reference vitamers Extraction with 10%TCA, Vydac 401TP-B, 10µm, A. 0.02M HCl Fluorescence, Serum, various Cuburn and 2-amino-5-chlorobenzoic 300x4.6 mm id B. 0.1M sodium 330/400nm tissues, milks Mahuren acid as internal standard, phosphate buffer, Post-column 1983 remove of TCA with diethyl pH3.3 derivatization PLP, PMP, PL, ether C. 0.5M sodium with sodium PN, PM, phosphate buffer, phosphate buffer PNG pH 5.9 (1M, pH 7.5) gradient elution, containing 1.5ml/min 1mg/ml NaHSO3 (4.5ml/h) Homogenization with 5% SSA, Perkin-Elmer ODS, A. 0.033M phosphate Fluorescence, 330/ Plant foods, Gregory and DPN as internal standard, remove 3µm, 30x4.6mm id buffer, 8mM octanesul- 400nm Post-column human milk Ink 1987 of lipids with CH 2 Cl 2 , preparative fonic acic, pH2.2 derivatization with anion-exhange chromatography to B. 0.033M phosphate potassium phosphate PLP, PMP, PL, remove SSA buffer, pH2.2 – buffer (1M, pH 7.5) PN, PM, PA 2-propanol (93:7) containing 1mg/ml β-Glucosidase treatment to verify NaHSO3 (0.2ml/min) glycosylated vitamer gradient elution, 1.8ml/min Extraction with 1M PCA, DPN as TSK Gel ODS-80, 0.05M potassium Fluorescence, 330/ Various foods Tadera and internal standard, remove of PCA 250x4.6mm id phosphate, pH 3.5, 400nm Post-column Naka 1991 with KOH 0.12M sodium derivatization with PLP, PMP, perchlorate, potassium phosphate PNP, PL, PN, buffer (1M, pH 7.5) PM, PNG, PA – methanol (99:1) containing 1mg/ml NaHSO 3 isocratic elution, 0.5ml/min Extraction with 0.5-1M PCA, LiChrospher RP-18, A. methanol B. 0.03M Fluorescence, 330/ Liver, milk Bitch and DPN as internal standard, remove 5µm, 125x4.6mm id phosphate buffer, 400nm Post-column Möller 1989 of PCA with KOH, dephosphory- pH2.7 – 4mM derivatization with PLP, PMP, PL, lation with alkaline phosphatase octanesulfonic acid potassium phosphate PN, PM, PA buffer (0.5M, pH 7.5) gradient elution, containing 3.7mg/ml 1.5ml/min NaHSO 3 (0.07ml/min) Extraction with 0.05M sodium LiChrospher RP Select Acetonitrile – 0.05M Fluorescence, Yeast, wheat germ, Reitzer- acetate, deamination of PM to B, 5µm, 250x5mm id potassium phosphate, 290/395nm breakfast cereal Bergaentzle PL with glyoxylic acid, dephos- pH2,5 0.5mM et al. 1993 phorylation with phosphatase, heptanesulfonic acid reduction of PL to PN with NaHB4 isocratic elution, PN 1ml/min Extraction with 5%TCA, DPN as ODS Hypersil, 3µm, Methanol – 0.1M Fluorescence, 333/ Various plant foods, van internal standard, dephosphory- 125x4.6mm id phosphate buffer, 375nm Increase of meats, feeds, yeast Schoonhoven lation with Takadiastase pH2.15 – 1.25mM pH prior detection et al. 1994 octanesulfonic acid with 1M K2HPO4 isocratic elution, (0.3ml/min) 1.2ml/min PL, PN, PM Homogenization with 5% TCA, Spherisorb ODS2, Methanol- 0.033M dephospharylation with acid 10µm, 300x3.9mm id phosphate buffer phosphatase pH2.2 (2:98) β-Glucosidase digestion for isocratic elution, Fluorescence, 328/ Lentils, peas, beans Sierra and glycosylated vitamer(s) 1.2ml/min column 390nm Increase of Vival- temperature 17C pH prior detection PL, PN, PM, Valverde with 0.3M K 2 HPO 4 PNG 1997 (0.7ml/min) HCl; hydrochloric acid, DPN; 4-deoxypyridoxine, KOH; potassium hydroxide, PA; pyridoxic acid, PCA; perchloric acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PN; pyridoxine, PNG; pyridoxine glucoside, PM; pyridoxamine, PMP; pyri- doxamine-5’-phosphate, SSA; sulfosalicylic acid, TCA; trichloroacetic acid 551 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. large variety in specificity of the column mate- rials and yielded sophisticated separation pro- cedures. The choice of the composition of the mobile phase, that is organic solvent and buffer, depends on the desired pH and the sample ex- traction procedure and vice versa. Native state fluorescence, characteristic to vitamin B 6 related compounds, enables their fluorometric detection which is more selective and sensitive compared to that of UV detection. However, improvements in specificity and se- lectivity in the detection, e.g. use of a mass-se- lective and other MS applications, are still cer- tainly required to reduce the misinterpretation of the chromatographic data. 2.4.2.1 Open column chromatography In early studies of vitamin B 6 methodology (e.g. Toepher and Lehman 1961) the separation was based on cation-exchange chromatography in an open column as the cationic characters of pyrid- inium and aminomethyl group enables their ion- exchange chromatographic separation. Open column chromatography with Dowex AG 50–8X (Toepher and Lehman 1961), sulfonated Amber- lite CG120S or related resins (Loo and Badger 1969) for separation of free pyridoxal, pyridox- ine and pyridoxamine was described; changing the pH of the elution solvent, three vitamers can be separated according to their pK constants. This procedure combined with microbiological quantitation has been well documented and val- idation data from the collaborative studies is available (Toepher and Polansky 1970). This approach is still in use in the method recom- mended by AOAC (1995a, b). A disadvantage of this separation procedure is the limited effi- ciency and the unwanted dilution of the vitamer fractions. In addition, the method is useful only for free pyridoxine, pyridoxal and pyridoxam- ine as phosphorylated vitamers need to be hy- drolyzed prior to chromatography. Proper separation efficiency can still be achieved by using an ion-exchange resin with an open column technique despite the higher separation efficiency achieved in the HPLC with modern chemically bounded ion-exchange or partition column packings. Quite recently, a cat- ion-exhange chromatographic separation of vi- tamers using open-column chromatography was published by Srividya and Balasubramanian (1997). Chromatographic separation was per- formed prior to the spectrophotometric determi- nation of diazotized 2,4-dinitroaniline derivative of pyridoxine The proposed method was applied for pharmaceutical preparations, biscuit and rice samples. 2.4.2.2 High-performance liquid chromatography Ion-exchange chromatography High-performance liquid chromatographic cati- on-exchange separation has been applied to bio- logical samples (Coburn and Mahuren 1983, Hart and Hayler 1986 , Argoudelis 1988, Mahuren and Coburn 1997) and food materials (Wong 1978). Seven B 6 vitamers: PLP, PNP, PA, PMP, PL, PN and PM, and an internal standard (2-amino-5- chlorobenzoic acid) were separated within 36 minutes using a silica-based ion-exchange col- umn packing (Coburn and Mahuren 1983). The total analysis time varied from 50 to 60 minutes depending on the sample complexity. Baseline chromatographic separation of standards was achieved using a rather long gradient elution with phosphate buffers. Good precision and recovery was obtained with this method. However, to ver- ify the proper peak identity and purity, modifi- cations in solvent gradient were occasionally needed. Shephard and associates (1987) applied a cation-exchange chromatographic method for human plasma samples. Short column life was considered as a drawback of the method as the retention of early eluting compounds, pyridoxal phosphate and pyridoxic acid, decreased rapid- ly in the course of time. Cation-exhange chro- matography was further applied to human tissue samples (Shephard et al. 1989) and other bio- logical samples (Argoudelis 1988) in which good precision and recovery data for six vitamers was reported. Callmer and Davies (1974) published a cation-exhange chromatographic determination of pyridoxine in vitamin preparation using di- lute nitric acid as an eluent. 552 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods An anion-exhange chromatography of B 6 vi- tamers and pyridoxic acid was published by Vanderslice et al. (1979); vitamin B 6 compounds were separated as their anion forms at pH 10 with a BioRad A-25 anion exhance packing the elu- ent being sodium chloride-glycine buffer. This method was applied to food samples in the com- prehensive studies of Vanderslice and cowork- ers; for cereal, milk powder and flesh foods (Vanderslice and Maire 1980, Vanderslice et al. 1980), ready-to-eat cereals (Vanderslice et al. 1981b) and for chicken meat (Olds et al. 1993). If naturally fluorescent sulfosalicylic acid was used as a deproteinating agent in the extraction, it had to be removed by an anion-exchange pu- rification using a Dowex or Aminex A-50 resins prior to the chromatographic separation (Gregory and Ink 1987). The disadvantage of this clean- up procedure was the inevitable dilution of the vitamer fraction. Partition and ion-pair chromatography Dilute acidic solutions have been used as a mo- bile phase with the octadecyl phase; a dilute sul- furic acid (Bognar 1985, COST91 1995, Esteve et al. 1998) or an ammonium sulfate solution (Steiner et al. 1993) were used for the separa- tion of free vitamers in food samples. Despite the polar nature of B 6 vitamers, reversed-phase separation based on partition can yield proper relative retention; only pyridoxamine or pyri- doxal-5’-phosphate as such showed an inade- quate retention in the partition mode. More com- monly the separation of B 6 compounds is estab- lished with an ion-pair partition chromatography due to vitamers’ ionic nature. The control of re- tention and selectivity in the ion-pair chroma- tography have been thoroughly documented by Dong et al. (1988) and Schill (1989). The use of a complexing agent, the ion-pair reagent, enhanc- es the retention of an analyte to the stationary phase or alters the selectivity of the compounds. The retention of an organic ion is affected by the nature of an ion-pairing compound and its concentration, and the presence of other counter ions or components competing for the binding capacity of the stationary phase. By choosing the ion-pair reagent and the separation conditions, the relative retention of vitamin B 6 compounds, especially that of pyridoxamine, can be widely regulated. The chromatographic separation of free and phosphorylated vitamers as well as pyridoxine- β-glucoside was published by Gregory and Ink (1987). Reversed-phase separation on a 3µm ODS-column was performed with a gradient elu- tion of the mixture of 2-propanol, phosphate buffer and octanesulfonate at pH 2.2 (Gregory and Feldstein 1985) followed by Coburn’s and Mahuren’s (1983) enhanced fluorerence detec- tion with sodium bisulphite reagent. Good re- peatability was achieved and the procedure was applied to a number of plant-derived samples. The method and its modifications have been fur- ther applied to food and other biological materi- als later on by various research groups. The use of an alcyl derivative of sulfonic acid predomi- nates in ion-pair chromatography of B 6 com- pounds; a number of reports using hexane sul- fonic acid (Kawamoto et al. 1983, Wehling and Wetzel 1984, Bötticher and Bötticher 1987, Chase and Soliman 1990, Tolomelli et al. 1991, Reynolds and Brain 1992, Olleta et al. 1993, Blanco et al. 1994, Agostini and Godoy 1997), heptane sulfonic acid (Lam et al. 1984, Heffer- an et al. 1986, Bühnert 1988, Rees 1989, Iwase 1992, Reitzer-Bergaentzle et al. 1993) and oc- tane sulfonic acid (Bitsch and Möller 1989, Maeda et al. 1989, van Schoonhoven et al. 1994, Sharma and Daksinamurti 1992, Sampson et al. 1995) or a mixture of alcyl sulphonates (Jaumann and Engelhardt 1985, Addo and Augustin 1988) have been published. The phenolic character of pyridoxine was utilized by Belal (1989); tetrabutyl ammoniumhydroxide enabled the proper separation of pyridoxine and pyrithiox- ine in a selected pH of 8 in dosage forms. Enhanced selectivity for the separation of thiamin, riboflavin, pyridoxine and niacin using a reversed-phase column was achieved with a small amount of ammonium hydroxide in the mobile phase (Chase and Soliman 1990). The use of triethylamine did not give the same result in their application. A 16mM concentration of mon- 553 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. ochloroacetic acid (MCA) in the phosphate buff- er mobile phase insteadt of trichloroacetic acid improved significantly the resolution of pyridox- amine and its phosphate ester (Ekananayke and Nelson 1988). No explanation of the detailed mechanism for MCA for this particular resolu- tion was given. A similar affect for trichloroace- tic acid (TCA) on the separation efficiency has also been suggested (Bognar, personal commu- nication). Japanese researchers used sodium per- chlorate (100–120mM) in an isocratic mobile phase to separate six B 6 vitamers, pyridoxine β- glucoside, pyridoxic acid and deoxypyridoxine within a single run. Perchlorate anion was stat- ed to serve as an ion-pair reagent in this appli- cation (Tsuge et al. 1986, Tadera and Naka 1991). A small amount of PCA present in the mobile phase consisting of phosphate buffer enabled the separation of seven (Argoudelis 1997) and four- teen (Tsuge 1997) vitamin B 6 related compounds in a single isocratic run. A mobil phase (pH 2.5) containing sodium perchlorate (5mM), alkyl sul- phonate (10mM) and methanol separated four free vitamers well in a isocratic run within 14 minutes (Kawamoto et al. 1983). The silanol ef- fect between reversed-phase packings and basic analytes (Stadalius et al. 1988, Li 1992, Sykora et al. 1997) have been diminished by using or- ganic solvent modiers (McCalley 1995) in the mobile phase; free silanols in the packing sur- face were blocked by adding triethylamine (Mascher1993, Lam et al. 1984, Edwards et al. 1989, Agostini and Godoy 1997) or di-N- butylamine (Bötticher and Bötticher 1987) to the mobile phase used for vitamin B 6 analysis. The use of a volatile buffer in the mobile phase is favorable for LC-MS applications. Care- ri et al. (1996) used methanol and ammonium formate in the eluent to apply the mobile phase stream (150µl/min) into a quadrupole mass spec- trometer via a particle beam interface. Good re- peatability data (RSD% <6%) for pyridoxal, pyridoxine and pyridoxamine was presented. Detection Native state fluorescence of vitamin B 6 related compounds enables their selective and sensitive fluorometric detection which is considered to be superior to UV detection. Liquid chromatograph- ic separation of B 6 compounds is mainly done in acid or neutral medium due to the better separa- tion efficiency of ion-pair chromatography in an acidic enviromental (Dong et al. 1988) or to re- strictions of silica-based columns in alkaline conditions (Claessens et al. 1996). The maximum fluorescence intensity of pyridoxine is however at pH 7 (Duggan et al. 1957). The low fluorescence intensity of pyridoxal 5’-phosphate was enhanced with an innovative post-column derivatization by Coburn and Ma- huren (1983). After cation-exhange chromato- graphic separation of PLP, PNP, PA, PMP, PL, PN and PM on a Vydac 401TP-B column in acid- ic medium, the pH of the mobile phase was raised to neutral by mixing the eluent with a derivati- zation solution consisting of phosphate buffer (1M, pH7.5) and sodium bisulphite (1mg/ml) prior fluorometric detection. The elevated pH and the formed bisulphite adduct increased the intensity of pyridoxal 5’-phosphate in the FL detection allowing its measurement in the bio- logical samples with a low level of B 6 vitamers. This approach with modifications has been wide- ly accepted later on for the analysis of pyridoxal- 5’-phosphate either alone or among other vita- mers in food samples (Hamaker et al. 1985, Gre- gory and Ink 1987, Addo and Augustin 1988, Ekanayake and Nelson 1988, Bitsch and Möller 1989a, Tadera and Naka 1991, Olds et al. 1993) and in other biological materials (Reynolds and Brain 1992, Sharma and Daksinamurti 1992, Sampson et al. 1995, Mahuren and Coburn 1997). Various changes in Coburn’s and Ma- huren’s post-column derivatization have been presented; a pre-column modification for cell- free yeast culture media was published by Ar- goudelis (1988) whereas Kimura et al. (1996) added bisulphite reagent to the mobile phase. So no further equiment for derivatization in the lat- ter application was needed. In the method report- ed by Argoudelis (1997), the bisulphite adduct of PLP was formed in acidic pH without raising the pH of eluent with an alkaline solution. On the other hand the elevated pH itself has been 554 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods reported to produce a more sensitive and selec- tive detection by reducing the interference found in rice and wheat samples (van Schoonhoven et al. 1994) and legumes (Sierra and Vidal-Valverde 1997). Apparently there was no need for adding bisulphite to the post-column reagent in their applications. A post-column derivatization using the sem- icarbazide reaction for increasing the intensity of PLP was first reported by Gregory (1980a). Unfortunately, the fluorescence intensity of formed PLP semicarbazone was not at its maxi- mum in slightly acidic to neutral solutions. The advantage of this derivatization was that the re- action could also be performed in dilute trichlo- roacetic acid solutions (pH<2) compatible with the sample extraction procedure (Ubbink et al. 1986). The semicarbazide reaction was utilized in the determination of PLP in plasma samples by Vanderslice and Maire (1980) and Mascher (1993, 1997). Gregory and his associates (1981) published a procedure to deaminate pyridoxam- ine and its 5’-phosphate ester to related aldehyde vitamers by a glyoxylate reagent. The fluores- cence of PLP was enhanced by a further semi- carbazide reaction. The glyoxylate reaction pro- vided quantitative deamination of PM and PMP, and the formed PLP semicarbazone was stable enough for multiple analysis. The enzymatic hydrolysis of phosphorylated vitamers with an acid phosphatase digestion combined with a gly- oxylic acid deamination of pyridoxamine to py- ridoxal followed by reduction with sodium boro- hydride was presented by Reitzer-Bergaentzle and coworkers (1993). The advantage of this excellent pre-column derivatization was that only one vitamer, pyridoxine, needed to be fine- ly measured. The validity of the method was fur- ther tested by a collaborative study in which good repeatability and reproducibility was obtained (Bergaentzle et al. 1995). This method has been put forward as an official method in France for vitamin B 6 determination in foods. Potassium cyanide acted as a catalyst in the oxidation of PL and PLP to 4-pyridoxic acid lac- tone and 4-pyridoxic acid 5’-phosphate, respec- tively (Ohishi and Fukui 1968). The formed de- rivatives possessed enhanced fluorescence inten- sity compared to the intact aldehydic vitamers. No effect on the fluorescence characteristics of other vitamers was found. This method was suc- cesfully applied to food (Tsuge et al.1988, Tou- kairin-Oda et al. 1989) and other biological sam- ples (Millart and Lamiable 1989, Tsuge 1997). UV detection usually lacks the sufficient sen- sitivity and selectivity for measuring vitamin B 6 compounds in foods and other complex biologi- cal materials. The few published papers are mainly limited to the use of UV detection after an ion-exchange purification of extracts (Wong 1978), in the fraction collection step prior to mass spectrometry (Hachey et al. 1985) or to the determination of vitamin B 6 in the vitamin prep- arations or enriched foods (Agostini and Godoy 1997). The wavelength of a conventional UV detector was set at 254 nm for multivitamin cap- sule preparations (Amin and Reusch 1987, Mae- da et al. 1989) and 280–290 nm for the commer- cial multivitamin preparations (Callmer and Davies 1974, Lam et al. 1984, Bühnert 1988, Hauenstein 1990, Iwase 1992). An on-line pre- column step was needed as an enrichment and clean-up procedure for measuring vitamin B 6 with UV-detection in the vitamin enriched grape sugar (Jaumann and Engelhardt 1985). The ca- pability of a photodiode array detector for vita- min analysis was evaluated in the study of Arai and Hanai (1988). This system was capable of detecting six vitamins within one chromato- graphic run. Difficulties in the quantitative and qualitative analysis of a vitamin mixture was supposed to be solved by combining the diode array detection on UV region with fluorometry. The reaction of pyridoxine with 2,6-dibro- moquinone-4-chlorimide in the presence of am- monia yieded a colored compound which exhib- its VIS absorption maximum at 650nm (Ka- wamoto et al. 1983). The proposed method was reported to be approximately seven times more sensitive for measuring pyridoxine in vitamin preparation than the conventional UV detection at 254 nm, and a notable improvement in the selectivity was reported. No further applications of this method have been presented. 555 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Electrochemical detection for the determina- tion of vitamin B 6 in multivitamin preparations has been proposed by Wang and Hou (1988a, b) and Hou and Wang (1990). The advantage of their application was the improved selectivity and wider linear dynamic range compared to that of UV detection. A subnanogram detection limit (0.5–1ng for PL depending on detection mode, 2ng for PN, and 1ng for PM) was achieved. A mechanism for oxidation of pyridoxal to a dim- erisation product was proposed by Hart and Hay- ler (1986); the approach was suggested for meas- uring the circulating vitamin B 6 as pyridoxal in plasma. Internal standard The accuracy and precision of HPLC analysis can be remarkably improved by using a thor- oughly validated internal standard procedure. The choice of an internal standard depends on for instance the chromatographic separation and the commercial availability of the reagent. 4-Deoxypyridoxine has been adopted in rou- tine use for vitamin B 6 analysis for food sam- ples (Gregory and Feldstein 1985, Gregory and Ink 1987 van Schoonhoven 1994, Esteve et al. 1998) and other biological materials (Pierotti et al. 1984, Morrison and Driskell 1985, Shar- ma and Dakshinamurti 1992). Its chemical char- acteristics resemble those of other vitamers and it is usually eluted in the middle of the reversed- phase or ion-paired reversed-phase chromato- gram. However, the relatively low fluorescence quantum yield of 4-deoxypyridoxine compared to other vitamers may limit its usefulness (Tou- kairin-Oda et al. 1989). In addition, its strong- er retention would increase the total analysis time in the cation-exhange chromatography pre- sented by e.g. Coburn and Mahuren (1983). In this case, deoxypyridoxine as an internal stand- ard was substituted by 2-amino-5-chlorobenzo- ic acid. Shephard and associates (1987) modi- fied the cation-exchange chromatographic method by using 4’-deoxypyridoxine-5’-phos- phate (DPNP) as an internal standard as its chemical structure and fluorescence properties were closer to B 6 vitamers than those of 2-ami- no-5-chlorobenzoic acid used by Coburn and Mahuren (1983). 3-Hydroxypyridine has also been put forward an internal standard for food samples (Vanderslice et al. 1979, Vanderslice et al. 1980, Olds et al. 1993, Vanderslice et al. 1984). It was stated that this compound was stable, did not coeluate with other vitamers and behaved in the same way as the vitamers throughout the ana- lytical process; hydroxypyridine had fluores- cence characteristics similar to other vitamers and it was well separated from the other vitamin B 6 compounds in their ion-exhange chromato- gram. However, in the reversed-phase chroma- tographic method of Ang et al. (1988) neither 3- hydroxypyridine nor 4-pyridoxic acid were suit- able for use as internal standards as those com- pounds were not separated from all the other peaks derived from chicken samples. Isopyridoxal was used for the analysis of B 6 vitamers in cell-free culture media (Argoudelis 1988) and in milk and egg yolk (Argoudelis 1997). Both deoxypyridoxine and isopyridoxal fullfilled the requirements for an internal stand- ard to quantitate six or seven vitamers in the sam- ples. However, the use of isopyridoxal as an in- ternal standard was recommended due to the better fluorescence response of iso-PL and its more suitable retention characteristics compared to those of DPN. The easy synthesis of isopyri- doxal was also put forward as an advantage as this compound is not commercially available at present. An internally standardized HPLC assay for vitamin B 6 in human serum published by Rey- nolds and Brain (1992) employed pyridoxamine 5’-phosphate (PMP) as an internal standard for the measurement of pyridoxal 5’-phosphate (PLP). PLP was the only vitamer found in sig- nificant quantities in their serum samples. PMP exhibited a more suitable retention than free vi- tamers in their chromatographic system. Ubbink et al. (1986) applied 6-methyl-2-pyridone car- boxaldehyde (MPCSC) as an internal standard for PLP in plasma analysis. A good separation of MPCSC and the semicarbazone derivatives of PL and PLP was achieved using a reversed- 556 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods phase chromatography system. Simultaneous determination of pyridoxine, thiamine, and ri- boflavin in infant formulas was internally stand- ardized by using m-hydroxybenzoic acid as it could be measured both with fluorometric de- tection (used for vitamin B 6 ) and UV spectro- metric (used for vitamin B 1 and B 2 ) (Chase et al. 1992). Acetanilide was used in the study of Mae- da et al. (1989) for simultaneous determination of nicotinamide, thiamine, riboflavin and its phosphate, pyridoxine, cyanocobalamine, caf- feine and sodium benzoate in oral liquid tonics. 2.4.2.3 Gas chromatography and mass spectrometry Gas chromatography is generally poorly suited for the analysis of vitamin B 6 compounds due to their ionic nature and lack of volatility unless they are dephosphorylated and their polar groups are derivatized. Rigorous pretreatment is also needed to remove nonvolatile compounds and water from the extract. Derivatization procedures include formation of acetyl (Hachey et al. 1985) or trifluoroacetyl (Lim et al. 1982) derivatives. The use of an electron-capture detector enabled a low detection limit (10pg) for all three free vitamer (Lim et al. 1982). Hachey and coworkers (1985) developed an analysis method for vitamin B 6 forms in biolog- ical samples by isotope dilution mass spectro- metry. Deuterated vitamers, served as internal standards, were added at the time of trichloro- acetic acid extraction and the aliquot of extract was fractionated using cation-exhange HPLC. The fractions were further purified with ion-ex- hange chromatography, PL and hydrolyzed PLP were reduced to pyridoxine, and the vitamer frac- tions were acetylated. Mass spectrometry record- ed electron impact mass spectra and the quanti- fication was based on the isotopic ratios of deu- terated and unlabeled ions. The tediousness and complexity of the method limits its use for rou- tine food analysis. This approach, however, may serve as a accurate reference method for vali- dating other analytical procedures. Liquid chromatography – mass spectrometry (LC-MS) is more easily applied to vitamin B 6 compounds than GC-MS. Using a semi-micro- bore column (Ultracarb ODS, Phenomenex, USA) and a binary gradient elution consisting of methanol and ammonium formate (20mM, pH3.57) seven water-soluble vitamins in a stand- ard solution were separated and detected using particle-beam mass spectometry (Careri et al. 1996). For electron impact spectra of the B 6 vi- tamers, the intensity of the molecule ion was high for all vitamers, and especially for pyridoxal (m/ z 167, 67%). The relative abundance of the ions observed was higher in chemical negative (NCI) and positive (PCI) ionization mode compared to that of electron impact ionization. The detection limit values for pyridoxal, pyridoxine and py- doxamine reported were 6ng, 225ng and 400ng, respectively. 2.4.3 Microbiological methods Microbiological assays for the determination of vitamins are based on “the nutritional require- ment of a microorganism for a certain vitamin” (Fawell 1990). Bacteria, molds, protozoa and yeasts can be utilized for the measurement of vitamin B 6 . Protozoa, like Tetrahymena or Ochramonas, have more developed ingestive and digestive system than yeasts or bacteria. Thus, their response to bound vitamers should resem- ble more closely that of higher animals (Voigt et al. 1979b). On the other hand, protozoa are more sensitive to the growth inhibition caused by salts, and only a few papers concerning their use in vitamin analysis have been published. So yeasts like Saccharomyces and Kloeckera are most fre- quently used at present especially as no bacteria utilizes equally all three B 6 vitamers. A compre- hensive summary of microbiological methods for vitamin analysis was presented by Bell (1974). The requirement of vitamin B 6 for the growth of a microorgamism and its use for vitamin de- termination was published by Atkin et al. (1943). The problem related to a microbiological assay is the variable growth response of an organism to different vitamers. The growth response of Saccharomyces uvarum (formerly S.carlsberg- 557 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. ensis) for pyridoxamine was 60–80% compared to that of pyridoxal or pyridoxine (Rabinowitz and Snell 1948, Gregory and Mabbit 1959, Po- lansky and Toepher 1969, Gregory 1982, Gre- gory 1983). The physiological fundamentals, the influence of pH, ions and temperature on the transport system and metabolism in yeast cells has been examined by Shane and Snell (1976). The extent of utilization of phosphorylated vita- mers also varies between microorganisms; the bioavailability of PMP was higher for Strepto- coccus faecalis than for S. uvarum or Str. fae- cium when mild acid extraction was performed for milk samples (Gregory and Mabbit 1961). Str. faecalis utilized pyridoxamine phosphate as readily as the free vitamer when the mild acid hydrolysis was insufficient to completely hydro- lyze pyridoxamine phosphate. Different respons- es of a microorganism for individual B 6 vitamers inevitability lead to uncertainties in the results. The response differences were later corrected by using an open-column ion-exhange column to separate B 6 vitamers prior microbiologic incu- bation and an individual calibration was per- formed for each vitamer (Toepher and Lehmann 1961, Toepher and Polansky 1970, Polansky 1981). This approach is used in the method rec- ommended by the AOAC (1995a, b). However, the chromatographic separation step limits the method’s usefulness for routine food analysis (Favell 1990). The most succesfully used yeasts in the as- says are S. uvarum (S. carlsbergensis) and Kloeckera brevis (K. apiculata). The former microorganism has been widely recognized for de- terminations of vitamin B 6 (Toepher and Polan- sky 1970, Bell 1974, Favell 1990, AOAC 1995a, b). However, an interference factor called “neg- ative drift”, especially in cereal sample was re- ported by Gregory (1980c); an inverse relation between the result and the amount of extract taken into the analysis caused variations in the results. The inhibition effect of neutralization salts (sodium and potassium chlorides) or food preservatives (sodium sorbate, propionate and benzoate) was suggested to explain this annoy- ing phenomenon (Voigt et al. 1979a, b, Guilarte 1984). It was concluded that the standards should be treated as samples to diminish the possible drift. The choice of buffer should also be con- sidered. Kloeckera brevis has been reported to show equal responses to three free vitamers (Gui- larte 1983, Guilarte and Tsan 1983) which is in contrast to the results of Gregory (1983). Whether this discrepancy was due to the sub- culturing or some other factors was not resolved. It seems likely that S. uvarum and also K. brevis only grow on the free vitamers or at least utilize phosphorylated vitamers to a lesser extent (Po- lansky 1981). If so, the dephosphorylation of related esters either by autoclaving or enzymat- ically hydrolysing seems to be essential during the course of analysis. A smart application for the measurement of vitamin B 6 was the use of a microbial biosensor system (Endo et al. 1995). This system used an immobilized Saccharomyces uvarum combined to a Clark -type oxygen electrode, and the deter- mination was based on the respiratory activity of the microorganism in the presence of vitamin B 6 . Oxygen consumption measured as a current decrease in electrode system was related to the vitamin B 6 amount in the media. A good corre- lation between proposed method and convention- al biological assay was achieved. A high sample input of the biosensor system was reported and one assay could be completed within 15 min. 558 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods This study was a part of the research “Nutri- ent Content of Finnish Foods; Water-Soluble Vitamins” at the Department of Applied Chem- istry and Microbiology/Food Chemistry. The aim of this work was to evaluate and improve the methodology used for the determination of vitamin B 6 , to assess the suitability of the de- veloped HPLC method to routine food analy- sis and to determine the vitamin B 6 content of the most common food items consumed in Fin- land. To achieve this – the extraction procedures compatible for liq- uid chromatographic determination of free, phosphorylated and glycosylated vitamin B 6 compounds in foods were assessed – the validity of the method and laboratory’s performance was evaluated by intercalibra- tion and collaborative parameters – up-to-date data on vitamin B 6 content of common food items was produced – the main glycosylated vitamer fraction of certain plant-derived foods was measured 3 Objectives of the study 4 Materials and methods 4.1 Calibrants and enzymes Vitamin calibrants were purchased from Sigma (MO, USA): 4-deoxypyridoxine hydrochloride (D-0501), 4-pyridoxic acid (P-9630), pyridoxal- 5-phosphate (P-9255), pyridoxal hydrochloride (P-9130), pyridoxamine-5-phosphate hydrochlo- ride (P-9505), pyridoxamine dihydrochloride (P- 9380) and pyridoxine hydrochloride (P-9755). Labeled pyridoxal phosphate, ([4,5- 14C]pyridoxal-5’-phosphate (Fig. 10), measured activity 314kBq (8.5µCi), 71mBq/mmol (1.93mCi/mmol)) was delivered by Amersham International plc, UK. Concentrations of the standard stock solutions were checked spectro- photometrically using molar absorption coeffi- cients (Table 4). Coefficient values obtained from the manufacturer (Claypool 1994) were in accordance to those published by Metzler and Snell (1955). The performance of spectropho- tometry was checked against a certified refer- ence holmium oxide kyvette (Milton Roy, USA). Alkaline phosphatase from calf intestine (EC 3.1.3.1., Boehringen-Mannheim, Germany) and β-glucosidase from almonds (EC 3.2.1.21., Flu- ka, Switzerland) were used in the enzymatic hydrolysis. The hydrolysis efficiency of phos- phatase and of β-glucosidase was tested with sample extracts. 4.2 Equipment An analytical liquid chromatography equiment controlled by a Millennium Chromatography Fig. 10. 14C-labeled pyridoxal-5’-phosphate (14C-carbons marked as *). 559 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Manager software package (v.2.12.x – 2.15.x, Waters, USA) consisted of three chromatogra- phy pumps (Waters M510, USA), an autosam- pler (700 Satellite or 712, Waters, USA), an air- bath column temperature module (Waters, USA) and a fluorescence detector (M470, Waters, USA). A binary gradient elution of the mobile phase was performed with two HPLC pumps while the third pump was used for the post-col- umn derivatization. Data signal was collected via an A/D converter (SAT/IN interface, Waters, USA) and processed with the Millennium Chro- matography Manager software package. Cal- culations in internal standard method were veri- fied using a HP 42S calculator (Hewlett-Pack- ard, USA). Preparative liquid chromatography was per- formed using a Delta Prep liquid chromato- graph (Waters, USA) with a UV-VIS (Waters 484, USA) and a fluorescence detector (Waters 470, USA). A column packing device manufac- tured by Shandon (UK) used for packing the in-house made analytical HPLC columns in- cluded a single piston pressure intensifier unit and a stainless steel slurry chamber (50ml, Shandon, UK). A Varian Unity-500 NMR-spektrometer (Var- ian, USA) was used for measuring the proton NMR spectra. Mass spectra was performed with a Finnigan MAT 90 mass spectrometry (Finni- gan, USA). 14C-activity was measured using a Wallac 1411 Liquid scintillation counter (LKB, Sweden). 4.3 Food samples, sample collection and pretreatment Food samples analyzed were chosen on the ba- sis of their consumption in Finnish diet and their estimated role in vitamin B 6 intake. Sales vol- umes of individual food items purchased from main wholesale food chains were used for se- lecting items to be analyzed. The food sample list was prepared in cooperation with the Finn- ish Food Data Bank (National Institute of Health, Helsinki). Sampling was performed during June 1994 and January 1995. Collected food samples represented common food items in six main food groups; meat and poultry (16 items), dairy prod- ucts and egg (9 items), vegetables and nuts (8 items), cereals (7 items), fish (6 items), others (5 items), the total number of analyzed food items being 51 (Appendix 1). Food collection was generally based on the procedure applied in previous vitamin studies performed at our department (Piironen 1986, Heinonen 1990, Mattila 1995, Vahteristo 1998); ten retail shops, located in the Helsinki area and Table 4. Molar absorption coefficients (e) of vitamin B 6 compounds (Claypool 1994). Compound ε λ max solution, M W (l mmol-1cm-1) (nm) pH (g mol-1) PLP 5.02 388 0.1M phosp. buffer, pH7 247.1 PMP•HCL 8.37 326 0.1M phosp. buffer, pH7 284.6 PL•HCl 8.96 288 0.1M HCl 203.6 PN•HCL 7.30 323.8 0.1M phosp. buffer, pH7 205.6 DPN•HCl 8.10 314 0.1M phosp. buffer, pH7 189.6 PM•2HCl 4.6 253 0.1M phosp. buffer, pH7 241.1 DPN•HCl; 4-deoxypyridoxine hydrochloride, PL•HCl; pyridoxal hydrochloride, PLP; pyridoxal-5’ -phosphate, PN•HCl; pyridoxine hydrochloride, PM•2HCl; pyridoxamine dihydrochloride, PMP•HCl; pyridoxamine-5’-phosphate hydrochloride 560 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods representing the main wholesale food chains in Finland, were selected based on their market share. If the commodities were not available in these shops, other shops, related to the same wholesale chain if possible, were used. Pork liv- er, pork kidney, reindeer, lamb and elk samples were purchased from a slaugher house or from a wholesale house. Pike and pikepearch were caught from Baltic Sea, whitefish from Baltic Sea and from Lake Inari. Rainbow trout was cultivated in the sea area. Baker’s yeast was from Alko Oy, the Finnish State Alcohol Company. The size of collected subsamples varied from 300g to 1kg depending on the food item and its package size; a minimum subsam- ple size was kept at approx. 300g. Ten subsam- ples per each food item were collected and the subsamples were pooled to form a composite sample. Pretreatment was carried out as soon as pos- sible in practice after sample collection. Meat samples were deboned and cut into cubes (ca. 1cm3) with a stainless steel knife. Any excess of adipose and connective tissue as well as artifi- cal coverings were removed. Frozen chicken (broiler) and hen, were allowed to thawn, deboned, deskinned, and the meat and fat were cut into cubes with scissors. Frozen peas were mixed as such. Fish samples were obtained as fillets. Egg yolk was separated from raw eggs using normal household methods. Liquids (e.g. milks) and powders (e.g. flours) were mixed and pooled. Subsamples of each food item were mixed in a plastic container and divided into ca. 200g portions. These portions were vacuum- packed into polyethylene-nylon laminate bags, frozen and stored at –20°C until analysed. Pre- treament and the sample preparation were done under dimmed light and in cool conditions when- ever practically possible. A commercial infant formula, fortified with pyridoxine, was used as an in-house reference material. The content of three packages (of 400g) were mixed, divided into portions of 10 to 20g, vacuum-packed into aluminium coated polyeth- ylene-nylon laminate bags, frozen and stored at -70°C until analysed. 4.4 Evaluation of liquid chromatographic method and extraction procedure 4.4.1 Analytical liquid chromatography 4.4.1.1 Chromatographic parameters and column testing protocol Interparticle volume (V 0 ), retention factor (k e ), separation factor (α), peak resolution (Rs), sym- metry (A s 2) and the plate number of the column (N) calculated according to Snyder and Kirkland (1979) were used to characterize the chromato- graphic performance. The interparticle volume of the column was measured using either uracil or sodium nitrate detected at a wavelength of 254nm (Wells and Clark 1981). The limit of de- tection (LOD) was evaluated as three times sig- nal-to-noise ratio. Octadecyl columns were char- acterized by measuring the column performance (as theoretical plate number, N), the silanol in- dex (SiOH) and the hydrophobicity (HP) values according to Walters (1987). The aminopropyl packing was tested using manufacturer’s intruc- tions (Spherical Materials and Columns for HPLC 1976). 4.4.1.2 Tested column packings and mobile phases Reversed-phase and reversed-phase ion-pair chromatographic techniques were applied to sep- arate vitamin B 6 compounds. Chromatographic separation was performed using octadecyl or aminopropyl phases (Table 5). Novapak C18 (Waters, USA), µBondapak C18 (Waters, USA), Vydac 201HSB (Separations Group, USA), Spherisorb S5ODS2 (formerly PhaseSepara- tions, UK) and Spherisorb S5NH 2 (formerly PhaseSeparations, UK) columns consisted of chemically modified silica. The silica-based col- umn packings chosen were mainly those of low- acidity and/or end-capped materials and thus expected to be suitable for the separation of ba- sic compounds (Walters 1987, Stadalius and 561 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Snyder 1988). Two polymer packings tested, based on the poly(styrene divylbenzene) (PS- DVB) structure, were Polyspher RP-18 (Merck, Germany) and Hamilton PRP-1 (Hamilton, USA). The columns tested were commercial prod- ucts except for S5ODS2 and S5NH 2 columns which were home-made. Octadecyl or aminopro- pyl phase (Spherisorb, PhaseSeparations , UK), ca. 3g of packing material per 50ml, was sus- pended in acetone or in a mixture of methanol: water (90:10), respectively. The slurry was packed downward with a constant pressure (450atm, 46MPa), the packing solvent being methanol (Spherical Materials and Columns for HPLC 1976). All analytical columns apart from the aminopropyl column were connected with a guard column (Novapak C18 cartridge 10x3.9 mm i.d., Waters, USA or Spherisorb S5ODS2 25x2.1 mm i.d., PhaseSeparations, UK). The mobile phase constituted of buffer (pre- pared from orthophosphoric acid, 85% Baker 6024 or its potassium salts), dilute acetic acid (Merck 1.00063, >99.8%, Germany) or dilute sulfuric acid. 2-Propanol, methanol, ace- tonitrile, all of HPLC grade (Baker, USA or Rathburn, UK), were tested as organic solvents in the mobile phase. Peak tailing, especially that of pyridoxamine, was diminished using triethyl- amine (Fluka 09340, pa., Switzerland) in the mobile phase (Stadalius and Snyder 1988). In order to increase the retention of pyridoxam- ine, an ion-pair was formed using alkyl sul- fonates (1-octanesulfonic acid, sodium salt, Sigma O-8380, appr. 98%, USA). The mobile phase was then a mixture of the organic sol- vent and the buffer. 4.4.1.3 Post-column derivatization The detectability of pyridoxal-5’-phosphate was improved by post-column derivatization. The fluorescence response of PLP was enhanced by forming a sulphite adduct in an elevated pH (Coburn and Mahuren 1983); a sodium phos- Table 5. Characteristics of the tested HPLC column packings. Column Particle Pore Surface Pore Carbon Type of phase End-capped packing diameter diameter area volume loading ligand, (µm) (nm) (m2/g) (cm3/g) (%, w/w) functionality S5ODS2 5 8 220 12 tri yes S5NH 2 5 8 220 2 tri yes 201HSB5 5 8 500 13.5 poly yes µC18 10 12.5 330 10 mono yes C18 4 6 120 0.3 7 mono yes PRP-1 5 7.5 415 0.79 * * * RP-18 9 305µm 3505µm * * * Packing Brand name, particle size, column dimensions, manufacturer S5ODS2 Spherisorb S5ODS2, 5µm, 250x4,6mm i.d., PhaseSep, UK S5NH 2 Spherisorb S5NH 2 , 5µm, 250x4,6mm i.d., PhaseSep, UK 201HSB5 Vydac 201HSB5, 5µm, 250x4,6mm i.d., Separation Group, USA µC18 µBondapak C18, 10µm, 300x3,9mm i.d., Waters, USA C18 Novapak C18, 4µm, 150x3,9mm i.d., Waters, USA PRP-1 Hamilton PRS-1, 5µm, 250x4,1mm i.d., Hamilton, USA RP-18 Polyspher RP-18, 9µm, 150x4.6mm id, Merck, Germany * poly(styrene-divinylbenzene) based 562 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods phate buffer solution (0.5M, pH7.5) containing sodium bisulphite (36mM) was pumped into the mobile phase through a union tee prior detec- tion. The fluorescence excitation and emission wavelengths were then set at 330nm and 400nm, respectively. The effect of post-column deriva- tization on the fluorescence intensity of B 6 vita- mers was evaluated by comparing the detector signal to that of a non-derivatized detection (emission wavelength 300nm, exitation wave- length 375nm). 4.4.2 Preliminary experiments on extraction procedure One mineral acid and four non-mineral acids were tested for sample extraction using plant- derived samples. The procedures used were based on literature review. Sulfuric acid extrac- tion (0.1M solution prepared from Merck 731, 95–97%, Germany) was used with or without an autoclaving process (Bognar 1985, COST91 1985). Non-mineral acid extractions were 0.5M perchloric acid (made from Baker 6063, 70– 72%, The Netherlands) (Bitsch and Möller 1989a), 5% (w/v) sulfosalicylic acid (Gregory and Ink 1987), 0.5M metaphosphoric acid (pre- pared from Fluka 79615, ~65% HPO 3 , puriss, Switzerland) (Ang et al. 1988) and 10% (w/v) trichloroacetic acid (made from Merck 807, z.A., ACS, Germany) (Coburn and Mahuren 1983). Autoclaving was omitted in non-miner- al acid extractions. All analysis were done in duplicate. The criteria for the chosen extraction proce- dure were the extraction efficiency, maintaining if possible the B 6 vitamer distribution in its in- tact form during extraction, and the compability to the reversed-phase liquid chromatographic separation procedure. After this preliminary ex- periment, the main emphasis was focused on the investigation of the perchloric acid extraction procedure (see 4.5.1). 4.4.3 Solid-phase extraction Solid-phase extraction (SPE) was tested to find out whether interferring compounds present in the sample extracts could be removed, and vita- min B 6 compounds concentrated as well. Parti- tion chromatography, using octadecyl, octyl, phenyl and cyanopropyl phases (Analytichem International, USA or Baker, the Netherlands), and cation-exchange chromatography, using car- boxymethyl, propylsulfonic acid and benzenesul- fonic acid phases (Analytichem International, USA or Baker, the Netherlands) were evaluated. The amount of sorbent material in SPE cartridg- es ranged from 300–500mg. A strong anion-ex- change phase, trimethylaminopropyl from Ana- lytichem International (USA), was used to re- move sulfosalicylic acid when samples were extracted according to the method of Gregory and Ink (1987). SPE packings were conditioned according to the manufacturers’ guide lines; first with meth- anol followed by a buffer solution compatible to sample extract. Adequately diluted sample extract solution was applied to the SPE cartridge at a flow rate not exceeding 2ml/min. Total ion- ic capacity of the phases (meq/g packing mate- rial), informed in manufacturer’s instructions were used. The flow rate was regulated using a vacuum manifold (SPE21, Baker, the Nether- lands) with additional valves, made of stainless steel and teflon, for each SPE cartridge. The SPE cartridge was washed with the solution in which the sample was applied or with the solvent of equal or less solvent/ionic strenght. Analyte(s) was eluted from the SPE cartridge by enhancing the solvent strength of the eluent; increasing the portion of the organic solvent, the ionic strength or ion selectivity, or changing the pH. Washing and elution volumes were maintained ca. two times the interparticle volume of the cartridge (void volume, bed volume) which was estimat- ed to be 1.2µl/mg cartridge packing. Elution was repeated until the analyte was eluted from the phase. All fractions (sample applying solvent, washings as well as elution fractions) were col- lected and analyzed by HPLC. 563 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 4.4.4 Statistics Comparison of means was performed by using an ANOVA or a t-test. When the tests’ basic as- umptions of equal variances were not matched, the Kruskal-Wallis or the Mann-Whitney test to compare the medians of the samples, or the Kol- mogorov-Smirnov test to compare the distribu- tions of the samples was performed. A multiple range test was performed to determine which means were different from the others. Box and whisker plot was used to detect outside and far outside data points (outlier values). All tests were performed at a 95.0% confidence level. If the amount of repetititon for individual measurements was inadequate for a proper sta- tistical evaluation, the variation was assumed to the same (approximately 12%) for each mean values, and the statistical evaluation was per- formed using these “recalculated” values. This approach is mentioned in the context of the re- sults when applied. All statistical calculations were performed using Statgraphic Plus for Win- dows v.3.0 software package (Manugistics, USA). 4.5 Chosen method for routine food analysis 4.5.1 Extraction procedure A modified perchloric acid extraction of Bitsch and Möller (1989a) combined with an alkaline phosphatase and β-glucosidase digestion was chosen for routine food analysis from the tested extraction procedures (see. 4.4.2). Food samples were extracted with ice-cold dilute perchloric acid followed by enzymatic hydrolysis. All sam- ple extracts were digested with alkaline phos- phatase while extracts from plant-derived sam- ples were also treated with β-glucosidase (Fig. 11). All determinations were done in triplicate; Fig. 11. Schematic presentation of the routine food analysis. (PA = pyridoxic acid, DPN = 4-deoxypyridox- ine, PL = pyridoxal, PLP = pyridoxal-5’-phosphate, PN = pyridoxine, PNG = pyridoxine glucoside, PM = pyridoxamine, PMP = pyridoxamine-5’-phosphate). 564 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods three separate vacuum-packed sachets of each pooled food item were analyzed. Homogenized sample (3–10g) was mixed with 50 ml 0.5M perchloric acid, 4-deoxypyri- doxine was added as an internal standard, and the mixture was homogenized in an ice-water bath for 60 sec using an Ultra-Turrax blender. The homogenate was made up to a known vol- ume (normally 100ml) and filtrated. A portion was taken (8ml) and the pH was adjusted to 7.5 with 5M and 0.1M potassium hydroxide solu- tion to precipitate perchlorate. After 30 minutes the mixture was filtered or centrifuged and made up to a known volume (10ml). 4.5.2 Enzymatic digestion In order to analyse free, phosphorylated and glycosylated vitamin B 6 derivatives, the sam- ple extract was divided into three parts. Free forms (PA, PL, PN, and PM) were determined using non-enzymatically treated extract, and two other portions of sample extract were digested with alkaline phosphatase and with β-glucosidase. Phosphorylated forms (PLP, PMP, and in case of baker’s yeast also PNP) were quantitated af- ter alkaline phosphatase treatment (Bitsch and Möller 1989a). An alkaline phosphatase solution (40µl containing 60U) was added to sample ex- tract solution (2ml), and the mixture was incu- bated for 30 min at room temperature. Sodium acetate solution (0.25M) and dilute hydrochlo- ric acid (0.2M) were used to adjust the pH to 3.8, and the mixture was filtered prior to liquid chromatography. Every sample extract was in- jected in duplicate. Sample extracts from plant- derived foods containing glycosylated vitamin B 6 derivatives (PNG) were digested also with β- glucosidase (Gregory and Ink 1987). In that case, a 2ml portion of the sample extract solution from which perchlorate has been precipitated was ad- justed to the pH 5.0 with sodium acetate solu- tion (0.25M) and dilute hydrochloric acid solu- tion (0.2M). β-Glucosidase solution (20U in 0.2ml) was added, and the incubation at 37°C was stopped after 4h by adding 50µl of trichlo- roacetic acid (100% w/v). After filtration the mixture was ready to inject into the liquid chro- matograph. Enzyme blank samples were treated like food samples. Two HPLC injections were done for each enzyme hydrolyse extract. The amount of alkaline phosphatase needed to convert phosphate esters into their free forms was tested by monitoring the phosphorylated and their free vitamers using beef steak as a test material. The efficiency of β-glucosidase hydrol- ysis was evaluated using the extract from raw carrot. The disappearance of the tentatively iden- tified glycosylated pyridoxine, and the simulta- neously increased amount of free pyridoxine were assumed to be markers for adequate glu- cosidase activity during the glucosidase diges- tion. 4.5.3 Liquid chromatographic separation An ion-paired reversed-phase chromatography was chosen to separate vitamin B 6 compounds in routine food analyses. The separation of vita- min B 6 forms was performed using a Waters Novapak C18 cartridge column (4µm, Table 6. Factors for converting vitamin B 6 compounds to pyridoxine base (PN, Mw 170 g mol-1). Compound PLP* PA PMP•HCl* PL•HCl PN•HCl PM•2HCl PNG* Factor 0.836 0.929 0.7059 0.836 0.8274 0.7059 0.8274 * PLP determined as PL, PMP as PM, and PNG as PN PA; pyridoxic acid, PL•HCl; pyridoxal hydrochloride, PLP; pyridoxal-5’-phosphate, PN•HCl; pyridoxine hydrochloride, PM•2HCl; pyridoxamine dihydrochloride, PMP•HCl; pyridoxamine-5’-phosphate hydrochloride, PNG; pyridoxine glucoside 565 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 150x3.9mm i.d., Waters, USA,) connected to a guard column (Novapak C18 cartridge 10x3.9 mm i.d., Waters, USA). The mobile phase, a modification of Gregory and Ink (1978), con- sisted of 33 mM phosphate buffer and 8 mM 1- octanesulfonate (pH 2.2), and 2-propanol. After a 2 min lag phase (2% 2-propanol), a linear gra- dient from 2% to 20% of 2-propanol was per- formed in 10 minutes followed by a 10 minute lag phase (20% 2-propanol). The total analysis time was 37 minutes. Post column reagent, so- dium bisulphite (36mM) in phosphate buffer (0.5M, pH 7.5), at a flow rate of 0.1ml/min was used to enhance the fluorescence of pyridoxal- 5’-phosphate, the total flow rate of mobile phase being 1.2 ml/min. Column temperature was set at 30°C with an air-bath module (Waters, USA). The injection volume ranged normally 50 to 75µl. Detection was based on the measurement of fluorescence signal (excitation and emission wavelengths set at 330nm and 400nm, respec- tively). 4.5.4 Calculations and expressing the results The results were calculated using an internal standard method in which peak area ratios were plotted against concentration ratios at nine con- centration levels. 4-Deoxy-pyridoxine was used as the internal standard compound. The fluores- cence response against concentration was fitted using a linear equation. The results were enzyme blank corrected. Calculations were performed with a Millennium 2010 Data Manager -software package (v.2.12.x – 2.15.x, Waters, USA) or by using a HP 42S calculator (Hewlett-Packard, USA). Phosphorylated vitamers, PLP and PMP, were determined as their free vitamers before and after alkaline phosphatase hydrolysis: PLP = PL amount after phosphatase treatment – PL amount before phosphatase treatment PMP = PM amount after phosphatase treatment – PM amount before phosphatase treatment In the case of baker’s yeast, the amount of pyridoxine phosphate was also determined: PNP = PN amount after phosphatase treatment – PN amount before phosphatase treatment Amount of glucosylated pyridoxine was de- termined as total PNG calculated as PN before and after β-glucosidase digestion: Total PNG = PN amount after glucosidase treatment – PN amount before glucosidase treatment Vitamin B 6 content (ΣB6), expressed as PN, was calculated as the sum of free and phosphor- ylated forms: ΣB6 = PLP + PMP + PNP + PL + PN + PM Total vitamin B 6 content (ΣΣB6) included glucosidic pyridoxine (calculated as PN) as well: ΣΣB6 = ΣB6 + total PNG Results of all vitamin B 6 compounds were expressed as pyridoxine, M w 170 g mol-1 (Tab- le 6). Values were given with two or three sig- nificant digits based on uncertainty calculations (see 4.5.6.4) and repeatability relative standard deviation (RSD r ). RSD r of the replicate analysis (n=3) for food analysis normally varied between 1–15%. 4.5.5 Comparison of the results with national food composition tables In order to compare the results of this study to national food composition tables, some foreign data was recalculated to standardize the vitamin B 6 expression form. A conversion factor of 0.8274 was used to convert the amount of pyri- doxine hydrochloride to that of pyridoxine. 4.5.6 Evaluation of the chosen method Validity of the chosen method for routine food analysis was investigated by evaluating the sta- bility of PLP during the analytical procedure, measuring the chromatographic parameters for 566 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods the internal standard, and determining recovery values for added standards, monitoring repeata- bility by using an in-house reference material, and by participating in both international and national intercalibration studies. Laboratory pro- ficiency for performing vitamin B 6 analysis was tested by taking part in the collaborative and certification studies organized by BCR’s EU- MAT -programme. 4.5.6.1 Stability of pyridoxal-5’-phosphate The stability of phosphorylated pyridoxal in the chosen perchloric acid extraction procedure was further tested using 14C-labeled pyridoxal-5’- phosphate as the aldehydic form is considered to be the most reactive vitamer in aqueous solu- tion containing amino acids and related com- pounds. The effect of extraction media and the sample matrix on interconversion and hydroly- sis of PLP were evaluated by adding labeled pyridoxal phosphate into the extraction proce- dure with the sample matrix. Tested sample ma- trices were beef, raw carrot and whole wheat flour. All analyses were done in duplicate. [4,5-14C]pyridoxal-5’-phosphate in aqueous solution was diluted with water and the activity of 39.3kBq (1.1µCi, corresponding approx. 140µg of labeled PLP) was added to the food sample prior to homogenization. The sample was extracted using the chosen perchloric acid pro- cedure excluding enzyme hydrolysis, and an extract portion of 75 µl was injected into ana- lytical liquid chromatography. Five fractions per minute (ca. 240µl) were collected during the whole chromatographic run. A liquid scintilla- tion cocktail (3ml, Hionic-Fluor scintillation cocktail, Packard Instrument, USA) was added to the collected fraction, and the 14C-activity was measured using a liquid scintillation counter. The quench correction was performed for each ma- terial, and the results were calculated as disinte- grations per minute (dpm). The purity of the la- beled pyridoxal phosphate was measured by in- jecting diluted 14C-PLP solution into analytical liquid chromatography, collecting fractions over the chromatogram and measuring the activity of the collected fractions. 4.5.6.2 In-house monitoring Peak identification was based on the retention time in the chromatogram, spiking the sample extract with standard solutions, and on phos- phatase and glucosidase treatments. Recovery of an added standard was measured using beef steak and a commercial infant formula as sample ma- trixes. A commercial product, fortified infant formula, was used as an in-house control sam- ple; the control sample was analyzed every three week. Retention factor (k e ), relative response (peak area /concentration), peak width at half height (w h ), peak symmetry (A 4.4 ) and peak tail- ing (T, according to USP) for the internal stand- ard ( 4-deoxypyridoxine) were monitored using the Millennium software’s system suitability option. 4.5.6.3 Intercalibration and collaborative studies The validity of the procedure for routine food analysis was verified with intercalibration and collaborative proficiency tests. Laboratory per- formance was evaluated by taking part in BRC EU-MAT Measurement and Testing Programme and later on by participating the BCR’s certifi- cation study. Twelve European laboratories par- ticipated in the 3rd EU-MAT intercomparison in which the vitamin B 6 results derived both from the in-house methods of individual laboratories and from a common “optimal” extraction proto- col were given (van den Berg et al. 1996). Certi- fication data consisted of the results from four- teen laboratories. Sample materials were lyophi- lized pork liver, mixed vegetable, whole meal flour and milk powder. Both high performance liquid chromatographic and microbiological methods were used in the above mentioned stud- ies. Chromatographic performance was further evaluated by an interlaboratory study between two German laboratories and our laboratory (Bognar and Ollilainen 1997); standard solutions and sample extracts, prepared by the Federal Research Centre for Nutrition (Stuttgart, Germa- ny) were analyzed in all three laboratories. All analyses were performed with two liquid chro- 567 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. matographic methods (an isocratic reversed- phase chromatography with FL detection and a gradient reversed-phase chromatography with FL detection) and with one microbiolocical assay using Saccharomyces cereviase ATCC 9080 as the test organism. Also a two sample intercomparison (with whole wheat flour and beef steak samples) was performed with VTT Technical Research Centre (Finland) in which rerversed-phase liquid chro- matographic results were compared to those de- rived from a microbiological assay in which Sac- charomyces cereviase (uvarum) ATCC 9080 was used as the test organism. 4.5.6.4 Estimating the uncertainty of the results, and statistical calculations The uncertainty of the results was estimated by calculating the expanded uncertainty for pyri- doxine results of the in-house control sample. Based on the internal standard standardization of pyridoxine, the 95% confidence interval (CI) of the result was estimated according to Caulcutt and Boddy (1983). Error standard deviation (ESD) values used for the evaluation of uncer- tainty were produced by the Millennium 2010 Data Manager software package (v. 2.12.X– 2.15.X., Waters, USA). Estimating the expanded uncertainty of the result included balance calibration and perform- ance, repeatability in weighing, error in volu- metric glassware and pipettes, and temperature dependense of volumetric glassware, purity and formula weight of weighed reference (standard) materials, and the performance of spectrophoto- metric measurement. If the limits given in cer- tificates or in equipment specifications were expressed without the confidence level or the shape of the distribution was not known, a rec- tangular distribution was used (Eurachem 1995). The following assumptions were also made; a random error of the test method was not related to the concentration and the residuals were nor- mally distributed. Outlying variance, homogeneity of varianc- es, and normality of distribution of mean values were tested in intercalibration and collaborative studies using the Cochran test, the Bartlett test and the Kolmogorov-Smirnov-Lilliefors -test, respectively. A Nalimov or Dixon’s test was per- formed to identify outlying mean values (Caul- cutt and Boddy 1983, Sokal and Rohlf 1995). When the homogeneity of variances did not match the Bartlett test, either a non-parametric Kruskal-Wallis or a two-sample comparison, Mann-Wilcoxon test, was applied. The tests were performed at a 95.0% confidence level. The sta- tistical calculations were performed by Stat- graphic Plus for Windows 3.0 software package (Manugistics, USA). 4.6 Characterization of bound pyridoxine Pyridoxine glycoside (PNG) from raw carrot and whole wheat flour was extracted with ice-cold perchloric acid solution and purified using a pre- parative liquid chromatography. PNG fraction was further purified using reversed-phase and cation-exhange chromatography. The glycosylat- ed pyridoxine fraction was characterized with NMR and mass spectrometry, and with enzymat- ic hydrolysis (Fig. 12). 4.6.1 Isolation 1 kg of raw carrot was homogenized in 2000ml perchloric acid (0.5M) for three minutes in an ice-water bath. The homogenate was filtered through a nylon cloth to remove excess solid material and the filtrate was then centrifuged. The filtrate was further concentrated prior to preparative chromatography in a vacuum-ro- tavapor, the temperature of the water bath not exceeding 20°C. The same procedure was also applied to the wheat sample. A Delta Prep liquid chromatograph (Waters, USA) was connected to a UV-VIS -detector (Waters 484, USA) and a fluorescence detector (Waters 470, USA). The preparative liquid chro- 568 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods magraphic method, except for the flow rate of the mobile phase, was scaled up on the basis of the analytical liquid chromatographic method. Separation of the glucosidic fraction of PN was performed using a µBondapak C18 column (15– 20µm, 100x25mm i.d., radial compression col- umn, Waters, USA ) connected to a guard col- umn (µBondapak C18, 10µm, 10x25mm i.d., Waters, USA). The flow rate of the gradient elu- tion, a mixture of 2-propanol, phosphate buffer, and ion-pair reagent, was set at 10ml/min. Post- column derivatization was omitted in the pre- parative chromatography. Fractions were manu- ally collected, and the individual fractions con- taining bound pyridoxine were pooled. Pooled fractions were concentrated in a vac- uum-rotavapor and the residues were redissolved into 9 ml of water. The fractions were injected into the preparative chromatograph in order to remove buffer salts and ion-pair reagent. The mobile phase was then a mixture of methanol and water (85:15) the flow rate being 10ml/min. The fractions were manually recollected, and checked by analytical liquid chromatography. Desalted, bound pyridoxine containing fractions were pooled and lyophilized. Proposed bound pyridoxine fractions were further purified with ion-exchange open-column chromatography based on the method of Grego- ry and Ink (1978). An open-column (32x20mm) was packed with a cation-exhange resin (Dowex AG 50W-X8, 200–400 mesh), regenerated with 1M ammonium hydroxide solution and equili- brated with 33mM ammonium hydrogenphos- phate solution (pH 2.2). The lyophilized extract, dissolved into a 30ml portion of water, was ap- plied to column, and the column was washed with 50mM ammonium acetate solution (pH4.0, 25ml) and with 250mM ammonium acetate so- lution (pH4.0, 65ml). The analyte was eluted from the resin with 250mM ammonium acetate solution (pH7.0, 65ml). Fractions of 10ml were collected and checked using analytical liquid chromatography. The ammonium acetate buffer used in the ion- exhange chromatography was removed by ana- lytical liquid chromatography using 10% (v/v) methanol as a mobile phase, the flow rate being 1.2ml/min. Lyophilized analyte extract was dis- solved in 10% methanol (1ml) , injected and the fractions containing bound pyridoxine were col- lected, pooled, lyophilized and dissolved in deu- terium oxide (99.8%, Merck, Germany). The sample was relyophilized and redissolved in deu- terium oxide (99.95%, Merck, Germany) prior NMR measurement. 4.6.2 β-glucosidase digestion As the reversed-phase chromatogram normally contains three peaks, named as X 1 (k e ~8.3), X 2 (k e ~10.0), and X 3 (k e ~11.8) which disappear after the digestion with β-glucosidase, the frac- tion that gives an increased fluorescence re- sponse in analytical liquid chromatography at the k-value of pyridoxine (k e ~13.3–13.6) being tak- en for further testing. The amount of this frac- tion was then measured as pyridoxine using analytical HPLC. Quantitation with an external standard method was based on the assumption Fig. 12. Isolation and characterization of glycosylated py- ridoxine. (NMR = nuclear magnetic resonance spectrosco- py, FAB-MS = fast atom bombardment mass spectrosco- py). 569 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. that the molar fluorescence responses of pyri- doxine and its glycosylated derivative are equal (Gregory and Ink 1987). 4.6.3 Structural evaluation Proton-NMR of the isolate was measured in deu- terium oxide using a Varian Unity-500 NMR- spectrometer at the Institute of Biotechnology (Helsinki). Proton-NMR-spectra and COSY and TOCSY correlation spectrum from wheat were recorded. The carrot fraction contained impuri- ties in a such amount that the evaluation of its NMR spectrum was rejected. Proton spectra of D(+)-glucose (Merck 8337, Germany) and pyri- doxine hydrochloride (Sigma P-9755, USA) dis- solved in D 2 O were measured in a same manner as for bound pyridoxine fractions. The FAB-ionization technique was used to perform mass spectrum of the analyte. The sam- ple was applied to the mass spectrometry via a directly coupled sample probe. Glycerol and ni- trobenzyl amine (for positive ion measurement) and triethanol amine (for negative ion measure- ment) were used as matrix modificators. Meas- urements were done in low resolution mode with Finnigan MAT 90 mass spectrometry equipment at the Environment laboratory of Helsinki. 5 Results 5.1 Liquid chromatography 5.1.1 Choice of the column and the mobile phase Peak symmetry of pyridoxal, pyridoxine, deox- ypyridoxine, and pyridoxamine in reversed- phase chromatography was measured using dif- ferent column packings and mobile phases (Ta- ble 7). Excellent peak shape for free B 6 vitamers was achieved with the µBondapak packing in spite of a relative high silanol index value (SiOH) for that particular packing material measured by Walters’ test (Table 8). The characteristics of the column packings were needed because the sepa- ration efficiency is greatly influenced by the cat- ionic nature of vitamin B 6 compounds. Peak shape of pyridoxamine, e.g. using 201HSB5 packing, could be improved by adding triethyl- amine as a modifier to the mobile phase but this reduced peak shape for the other free vitamers. Vitamin B 6 compounds possess both hydro- phobic and ionic properties which were utilized in liquid chromatographic separation. Chemical- ly bonded reversed-phases were chosen to be test- ed for separation of B 6 vitamers since adequate retention for free vitamers is generally achieved with these packings. However, the use of an ion- pair reagent was needed to ensure the adequate retention in liquid chromatography for phospho- rylated pyridoxamine (k e ~ 20 in the chosen method for routine food analysis). Rather mod- est column efficiency (as N/m) in Walters’ test was achieved with poly(styrene- divinylbenzene)(PS-DVB) based “reversed- phase like” polymer packing compared to silica based column packings (Table 8). The amino- propyl phase gave an insufficient retention of vitamin B 6 compounds, and its further testing was discontinued. A good peak separation was achieved using 2-propanol as a organic modifier in the gradient elution (Fig. 13). Selectivity of isopropanol was found to be superior to acetonitrile or methanol (data not shown). By lowering the pH of the mobile phase to approx. 2–3, the baseline dis- turbances could also be diminished which result- ed in increased selectivity of fluorescense de- tection. The low pH of the mobile phase also 570 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods On this basis, the selection of the chroma- tography mode for separation of vitamin B 6 com- pounds was later focused on ion-paired reversed- phase chromatography. Chromatographic param- eters (Tables 7 and 8) favoured the use of Wa- ters Novapak C18 cartridge column (4µm, 150x3.9mm i.d., Waters, USA) which was later on connected with a guard column (Waters No- vapak C18 cartridge) in routine food analysis. Retention of pyridoxamine was adjusted to a value of ca. k e 20 by choosing 1-octanesulfonate to serve as an ion-pair reagent. Gradient elution then consisted of 33 mM phosphate buffer and 8 mM 1-octanesulfonic acid (pH 2.2), and 2-pro- panol column temperature being set at 30°C. Retention factor for PLP, PA, PMP, PL, PN, DPN, and PM were then 1.4, 2.1, 5.6, 12.2, 13.6, 14.6 and 21.3, respectively. Tentatively identi- fied 5’-O-β-D-glucopyranosyl pyridoxine was eluted preceeding PL at k e value of 11.6–11.8 in this chromatographic system. 5.1.2 Post-column derivatization The relative fluorescence responses of B 6 vita- mers differs noticeable, pyridoxal-5’-phophate and 4-deoxypyridoxine (internal standard) hav- Table 8. Characterization of octadecyl columns tested according to Walters method (1987). Packing SiOH HP Efficiency (N/m) S5ODS2 0.51 5.14 52 000 201HSB5 0.64 3.96 11 000 µC18 1.71 3.18 14 000 C18 0.59 4.56 84 000 RP-18 0.20 nd* 4 400 SiOH; silanol index, HP; hydrophobicity index nd* antracene not eluted in a proper retention time Packings: S5ODS2 Spherisorb S5ODS2, 5µm, 250x4,6mm i.d., PhaseSep, UK 201HSB5 Vydac 201HSB5, 5µm, 250x4,6mm i.d., Separation Group, USA µC18 µBondapak C18, 10µm, 300x3,9mm i.d., Waters, USA C18 Novapak C18, 4µm, 150x3,9mm i.d., Waters, USA RP-18 Polyspher RP-18, 9µm, 150x4.6mm id, Merck, Germany Table 7. Peak symmetry values for B6 vitamers in reversed-phase columns tested. Column Mobile phase Symmetry (A s 2 ) PL PN DPN PM S5ODS2 40mM H 2 SO 4 ,isoc. 1.8 5.4 1.0 33mM phosphate buffer, pH2.2, isoc. 3.1 5.2 1.0 170mM HOAc, MeOH, IP, grad. 11.1 7.1 16.0 3.2 201HSB5 170mM HAc, MeOH, IP, grad. 2.1 2.0 1.3 2.5 170mM HAc, MeOH, IP, grad., 1%TEA 3.2 3.2 3.1 1.0 µC18 170mM HAc, MeOH, IP, grad. 1.0 1.0 1.0 1.0 C18 33mM phosphate buffer, pH2.2, IPA, grad. IP 2.1 2.1 2.0 Columns: S5ODS2 Spherisorb S5ODS2, 5µm, 250x4.6mm i.d., PhaseSeparation, UK 201HSB5 Vydac 201HSB5, 5µm, 250x4.6mm i.d., Separation Group, USA µC18 µBondapak C18, 10µm, 300x3.9mm i.d., Waters, USA C18 Novapak C18, 4µm, 150x3.9mm i.d., Waters, USA HAc; acetic acid, IP; ion-pair, IPA; 2-propanol, MeOH; methanol, TEA; triethylamine favoured the conditions required for the forma- tion of an ion-pair between alkyl sulfonates and cationic analytes like pyridoxamine. 571 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. ing the lowest responses (Fig. 14). The response of PLP was increased five-fold after the forma- tion of a sulphite adduct at elevated pH. The rel- ative effect of derivatization on the responses of other vitamers was smaller. Post-column deri- vatization reagent, 35mM sodium hydrogensul- phite in 0.5M phosphate buffer (pH7.5) mixed with the mobile phase yielded the highest re- sponse for PLP. Optimum condition for PLP was thus obtained by pumping a derivatization rea- gent through a union tee into the mobile phase with a flow rate of 0.1ml/min which correspond- ed approximately 3mM hydrogen sulphite con- centration in the mobile phase prior to detection (Fig. 15). In post-column derivatized LC system, the fluorescence excitation and emission wave- lengths were set at 330nm and 400nm, respec- tively. 5.2 Sample extraction 5.2.1 Preliminary experiments for the extraction procedure The result of the total B 6 vitamin content of a broccoli sample seemed to differ using four dif- ferent non-mineral acids as the extracting solu- tion. Tested non-mineral acids, sulfosalicylic acid, metaphosphoric acid, trichloroacetic acid, Fig. 13. Ion-paired reversed-phase chromatogram of a) a carrot sample and b) standards. (PA = pyridoxic acid, DPN = 4- deoxypyridoxine, PL = pyridoxal, PLP = pyridoxal-5’-phosphate, PN = pyridoxine, PM = pyridoxamine, PMP = pyridox- amine-5’-phosphate). 572 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods and perchloric acid showed also varieties in the vitamer distribution (Fig. 16a). The statistical difference in the results could not be directly evaluated due to an insufficient number of re- peated measurements. If the variation in the in- dividual mean values of the total results for each acid treatment was assumed to be the same (and of magnitude 12%), a statistical difference within the results was found; the results derived from the sulfosalicylic acid and perchloric acid ex- tractions were consistent, but they were differ- ent compared to those of metaphosphoric acid and trichloroacetic acid extractions, and vice versa. However, this was not the case with the banana matrix; in the total vitamin results of banana sample, the sulfosalicylic acid and trichloroacetic acid hydrolysis gave the same result (Fig. 16b). Sulfuric acid digestion without an autoclav- ing process extracted bounded pyridoxine but it hydrolyzed partially the glycosidic bond (Fig. 17). When sulfuric acid extraction was Fig. 14. Relative fluorescence responce of vitamin B 6 com- pounds with and without post-column derivatization. The relative fluorescence responce of PN without post-column derivatization set at 100. (DPN = 4-deoxypyridoxine, FL = relative fluorescence, PL = pyridoxal, PLP = pyridoxal-5’- phosphate, PN = pyridoxine, PM = pyridoxamine, PMP = pyridoxamine-5’-phosphate). Fig. 15. The effect of the amount hydrogensulfite in the mobile phase on the fluorescence responce of PLP. (FL = relative fluorescence, PLP = pyridoxal-5’-phosphate). Fig. 16. Effect of sample extraction on vitamin B 6 content of a) a raw broccoli sample extracted with non-mineral acids combined with a β-glucosidase treatment (replicate analyses) and, b) a banana sample (replicate analyses), no enzyme treatment, PLP not determined. (SSA; sulfosalicylic acid hydrolysis, MPA; metaphosphoric acid hydrolysis, TCA; trichlo- roasetic acid hydrolysis, PCA; perchloric acid hydrolysis, PL = pyridoxal, PLP = pyridoxal-5’-phosphate, PN = pyridoxine, PNG = pyridoxine glucoside, PM = pyridoxamine, PMP = pyridoxamine-5’-phosphate). 573 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Fig. 17. Effect of sample extraction on vitamin B 6 content of a carrot sample extracted with sulfuric acid and with perchloric acid (replicate analysis). (H 2 SO 4 ; sulfuric acid hydrolysis with and without β-glucosidase treatment, PCA; perchloric acid hydrolysis with and without β-glucosidase treatment, PA = pyridoxic acid, PL = pyridoxal, PLP = py- ridoxal-5’-phosphate, PN = pyridoxine, PNG = pyridoxine glucoside, PM = pyridoxamine, PMP = pyridoxamine-5’- phosphate). combined with an autoclaving treament (121°C, 30min), glycosidic pyridoxine in carrot extract was almost completely hydrolysed to free pyri- doxine (data not shown). Homogenizing the sam- ple with an ice-cold perchloric acid and omit- ting the autoclaving process extracted glyco- sylated pyridoxine in its intact form; practically no pyridoxine was measured after plain acid treatment while this was not the case with sulfu- ric acid. The explanation for an increased amount of PM after perchloric acid/glucosidase diges- tion remained unclear as the phosphatase activ- ity of the used enzyme preparate was not meas- ured. The sample treatment procedure in which the samples were extracted with dilute ice-cold perchloric acid (and excluding the autoclaving step) was chosen for a more detailed examina- tion. 5.2.2 Evaluation of the chosen perchloric acid extraction Perchloric acid extraction was further tested us- ing 14C-labeled pyridoxal-5’-phosphate. The measured total 14C-activity of purchased stand- ard compound was 314 kBq and the purity was approximately 76% calculated on the basis of distribution of percentage 14C-activity of untreat- ed standard in the reversed-phase chromatogram (Fig. 18a). 12% and 8% of the total 14C-activity was located in the chromatogram at retention factor values of k e =0.5 and k e =5.6. The former 14C-active fraction eluted at near interparticle volume (V 0 ) was considered as an unknown com- pound (named unknown 1) whereas the latter active fraction eluted at the same retention fac- tor value as PMP. A small change in 14C-activity distribution was found after the labeled pyridoxal phosphate standard was extracted with the routine perchlo- ric acid extraction procedure (Fig. 18b). The activity in PLP fraction decreased by approxi- mately 3% after perchloric acid extraction, and increment of activity was found in the PMP frac- tion, in unknown 3 fraction and in PL fraction representing the change of 0.8, 1.6 and 0.8% in the total activity, respectively. The sample matrix effect on the stability of PLP during the acid extraction was evaluated by adding labeled PLP standard to beef steak, car- rot and whole wheat flour samples and measur- ing the activity distribution in the reversed-phase chromatograms (Fig. 18c-e). Approximately, 10% of 14C-activity of labeled PLP added to beef steak and carrot samples was lost during the ex- traction procedure and the main increased ac- tivity was relocated in fractions of PL and of unknown 2 (Fig. 18c, d). The chemical structure of the supposed break-down product of PLP, re- ferred to as unknown compound 2, is not known. The percentage activity related to the released free PL due to hydrolysis of phosphate ester lin- gage of PLP was 7% and 14% in beef and carrot sample matrices, respectively (Fig. 19). For the whole wheat flour matrix, the 14C-activity in pyridoxal fraction was increased by approxi- mately 26%. When labeled PLP standard was extracted with any of three tested sample matrices, part of the 14C-activity was relocated in a unknown com- pound 2 fraction (k e ~3.9) instead of unknown compound 3 fraction (k e ~11). Formation of the 574 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Fig. 18. Distribution of 14C-activity in a reversed-phase chromatogram of a) a labeled PLP standard sample directly injected into HPLC omitting perchloric acid extraction procedure, b) a labeled PLP standard sample measured by HPLC af- ter the perchloric acid procedure, c) a labeled PLP standard added to beef steak sample, extracted with perchloric acid solution and measured by HPLC d) a labeled PLP standard added to carrot sample, extracted with perchloric acid solution and measured by HPLC and e) a labeled PLP standard added to whole wheat sample, extracted with perchloric acid solution and measured by HPLC. Five fractions (each of ca. 240µl) per minute collected dur- ing the whole chromatographic run (20min). Dpm = disin- tegrations per minute, PLP = pyridoxal-5’-phosphate. a b c d e 575 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. latter fraction was found during the extraction procedure without any food matrix. 5.2.3 Enzymatic digestion Phosphorylated pyridoxal and pyridoxamine were converted into their free forms using an alkaline phosphatase preparate from calf intes- tine. The minimum amount of alkaline phos- phatase needed to ensure adequate dephosphor- ylation was estimated to be 0.4–0.8U enzyme per mg sample in this sample extraction procedure (Fig. 20).The amount of alkaline phosphatase needed to release free vitamers from their phos- phate esters was tested using beef steak sample which is rich both in pyridoxamine phosphate and pyridoxal phosphate; the amount of phos- phorylated pyridoxamine and pyridoxal in beef covered approximately 44% and 46% of the vi- tamin B 6 activity, respectively. The carrot matrix where pyridoxine gluco- side accounts for the major portion of vitamin B 6 compounds was used for testing glucosidase Fig. 19. Changes in the vitamer distribution of 14C-labeled pyri- doxal-5’-phosphate added into three food matrices (beef, carrot and wheat), and compared to per- chloric acid extracted 14C-PLP standard calculated on the basis of 14C -activities. (PL = pyridoxal, PLP = pyridoxal-5’-phosphate, PN = pyridoxine, PM = pyridoxamine, PMP = pyridoxamine-5’-phos- phate, Unk 1,2,3,4 = Unknown compounds). Fig. 20. Effect of the amount of alkaline phosphatase on the hy- drolysis of phosphorylated pyri- doxal and pyridoxamine in beef steak matrix. (PL = pyridoxal, PLP = pyridoxal-5’-phosphate, PN = pyridoxine, PM = pyridoxamine, PMP = pyridoxamine-5’-phos- phate, Total = sum of PLP, PL, PM, PMP and PN). 576 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods hydrolysis. A completely disappeared PNG peak (k e =11.6, t R = 9.8 min) in the reversed-phase chro- matogram was considered to show an adequate enzymatic hydrolysis. No fluorescence signal at a k e value of 11.6 in the chromatogram was measured after hydrolysis procedure using ca. 0.13U of β-glucosidase per mg sample. 5.2.4 Solid-phase extraction Purification of the sample extract and concen- tration of vitamin B 6 compounds using a solid- phase extraction (SPE) technique was tested. As the adequate retention on reversed-phase SPE materials was not achieved due to a large varia- tion in polarity of B 6 compounds (data not shown), the separation of vitamin B 6 compounds based on the cationic nature of either pyridin- ium ion or aminomethyl group was evaluated. Weak cation-exchange functional group such as carboxymethyl (pK~4.8) retained vitamers from standard solutions but it was not effective enough with sample extracts (data not shown). Free B 6 derivatives, PL, PN, DPN and PM, could be iso- lated with a stronger cation-exhange phase by using the benzene sulfonic acid phase (Fig. 21a). Propyl sulfonic acid phase did not retain free vitamers sufficiently (Fig. 21b.). Also phospho- rylated vitamers (PLP and PMP) were lost using both phases. The most cationic compound, PM, retained strongly in both phases (Fig. 22). When the extraction method of Gregory and Ink (1987) was evaluated, sulfosalicylic acid (SSA) has to be removed by an anion-exhange SPE because of its natural fluorescence proper- ties. Removal of SSA by solid-phase extraction led to an unwanted dilution of the analyte ex- tract. As the same extraction efficiency was achieved with the more easily removable per- chloric acid (PCA), extraction with sulfosalicylic acid was excluded. Fig. 21. Retention of vitamin B 6 compounds on a) a benzene sulfonic acid SPE phase (replicate analysis) and b) a propyl sulfonic acid SPE phase (replicate analysis). (0 = fraction through the cartridge,W1-2 = washings, E1-E4 = elutions, DPN = 4-deoxypyridoxine, PL = pyridoxal, PLP = pyridoxal-5’-phosphate, PN = pyridoxine, PM = pyridoxamine, PMP = pyri- doxamine-5’-phosphate). Fig. 22. Retention of pyridoxamine on a propyl sulfonic acid (PRS) and benzene sulfonic acid (SCX) SPE-phases (replicate analysis). (0 = fraction through the cartridge, W = washing, E1-E4 = elutions). 577 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 5.3 Validity of the method chosen for routine food analysis 5.3.1 General parameters Retention factor (k e ) values for PLP, PA, PMP, PL, PN, DPN, and PM in the chosen ion-paired reversed-phase chromatographic separation were 1.4, 2.1, 5.6, 12.2, 13.6, 14.6 and 21.3, respec- tively. The retention factor, chromatographic peak shape (peak width at half height, asymme- try, USP tailing) and relative response (peak area/ concentration) for 4-deoxypyridoxine (DPN) in calibration and quantitation runs were monitored during the routine food analysis (Table 9). The retention of 4-deoxypyridoxine on the chroma- tographic system was reproducible, the relative standard deviation of retention factor being ca. 2%. A higher variation was found in the peak width at half height and in the relative responce values. However, the relative standard deviation of peak symmetry or USP tailing were both un- der 10%. The limit of detection (LOD) was estimated to be 50–150pg pyridoxine per injection which corresponded approximately to 1.4–4.2µg pyri- doxine per 100g food sample using the present- ed routine food analysis procedure. The recov- ery of an added analyte to beef steak and in- house reference material (commercial infant for- mula, fortified with pyridoxine) ranged normal- ly from 72 to 107% (Table 10). The highest var- iation (150%) was measured for PLP. The repeat- ability relative standard deviation (RSD r ) for the vitamin B 6 content of an in-house reference ma- terial (a commercial fortified infant formula powder) was 5.2% (n=6, triplicate analysis), mean and standard deviation being 0.44 and 0.023 mg PN/100g fresh weight (Fig. 23). Table 9. Repeatability parameters for 4-deoxypyridoxine (internal standard). Sample group Repeatability relative standard deviation (RSDr ) for n k’ w _ R r A 4.4 T Standards 178 3.0 9.9 14.0 9.4 5.9 Meat and meat products 174 2.3 10.0 14.3 7.7 5.0 Milk and milk products 90 1.7 6.6 11.4 9.5 6.1 Cereals and vegetables 359 1.8 21.4 14.5 7.7 3.7 Fish and fish products 68 0.8 5.8 7.9 7.1 4.8 Miscellanous 13 0.1 3.6 10.6 n.d n.d Sum 882 Weighed mean 2.0 13.9 12.9 8.1 4.7 n; number of chromatographic runs, ke; retention factor, wh; peak-width of at half height, Rr; relative responce (peak area/concentration injected), A 4.4 ; peak asymmetry calculated as width at the 4.4% peak height, T; tailing factor according to USP, n.d.; not determined Table 10. Recovery (%) of added B6 vitamers (mean±SD, n=4). Vitamer Beef steak Infant formula PLP 77±16.5 150±50 PA 77±14.2 83±5.2 PMP 90±4.8 82±16.8 PL 83±0.7 72±9.2 PN 98±2.3 95±7.3 DPN 94±3.8 107±9.4 PM 95±2.2 107±10.4 DPN; 4-deoxypyridoxine, PA; pyridoxic acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PN; pyridoxine, PM; pyridoxamine, PMP; pyridoxamine-5’-phosphate 578 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods 5.3.2 Uncertainty of results The total uncertainty of results was divided be- tween errors caused by calibration and those due to preparation of analysis sample (Table 11). Uncertainly related to internal standard stand- ardization, calculated as RSD%, was double that for preparation of the analysis sample. The total uncertainy for pyridoxine was estimated to be ca. 12%. Based on this calculation, the results derived from the routine food analysis were ex- pressed using two or three significant numbers. Uncertainty related to the spectrophotometric measurement of the concentration of the stock standard solutions was approx. 1%. The 95% confidental interval (CI 95% ) for concentration, slope, and intercept for in-house reference ma- terial were calculated on the basis of four ran- domly chosen calibration data sets of pyridox- ine, pyridoxal, and pyridoxamine (Table 12). The origin value was located in the calculated CI 95% of the intercept in all twelve calibration sets ex- cept in two cases (pyridoxine; calibration no 2, pyridoxal; calibration no 3). 5.3.3 Interlaboratory and laboratory proficiency tests Validity of the analytical method for routine food analysis and the performance of the laboratory were evaluated by taking part in EU Measure- ment and Testing (MAT) Programme’s 3rd inter- comparison. Each laboratory reported its results for a tested (“optimal”) method and for the lab- oratory’s own (“in-house”) method. Results from twelve laboratories were submitted (Fig. 24 a- c). Repeatability relative standard deviation (RSD r ) for a particular laboratory (variation within a laboratory) varied from 5% to 13% be- ing much lower than the reproducibility relative standard deviation RSD R (variation between lab- oratories) (Table 13). Our laboratory’s results (laboratory no. 2 in Figures 24a-c) were located within the 95% confidental interval calculated for either all data or outlier excluded data. Fur- thermore each laboratory’s performance could Table 11. Total uncertainty of pyridoxine result derived from an in-house reference material (infant formu- la) (mean value 0.324mg PN/100g fresh weight). Procedure mean±SD uncertainty as % range Calibration 0.324 ± 0.03708 11.4 0.287 – 0.361 Sample preparation 0.324 ± 0.01591 4.9 0.308 – 0.340 Total uncertainty (u Tot ) 0.324 ± 0.04035 12.4 0.284 – 0.364 Result 0.32 ± 0.040 Fig. 23. Repeatability of determination of the in-house ref- erence material (a commercial fortified infant formula pow- der); mean value of means, standard deviation, and repeat- ability relative standard deviation 0.44 mg PN/100g FW, 0.023 mg PN/100g FW and 5.2%, respectively. 579 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Ta bl e 12 . U nc er ta in ty a ss oc ia te d w ith th e ca lib ra tio n of a ) p yr id ox in e, b ) p yr id ox al a nd c ) p yr id ox am in e; 9 5% c on fi de nt in te rv al (C I 95 % ) f or c on ce nt ra tio n of a n in -h ou se re fe re nc e m at er ia l ( in fa nt f or m ul a) , f or s lo pe , a nd f or in te rc ep t ( 4- de ox yp yr id ox in e as a n in te rn al s ta nd ar d, n in e co nc en tr at io n le ve ls , t w o in je ct io ns a t e ac h le ve l) . C al ib ra tio n C on ce nt ra tio n Sl op e In te rc ep t lo w er m ea n up pe r ra ng e lo w er m ea n up pe r ra ng e lo w er m ea n up pe r ra ng e a) py ri do xi ne 1 0. 28 69 26 0. 32 40 08 0. 36 10 89 0. 07 4 5. 24 53 90 5. 36 32 90 5. 48 11 90 0. 24 -2 .6 63 73 7 1. 24 19 10 5. 14 75 57 7. 8 2 0. 32 28 39 0. 33 03 29 0. 33 78 19 0. 01 5 4. 50 48 73 4. 52 32 83 4. 54 16 93 0. 03 7 0. 13 63 60 0. 80 14 12 1. 46 64 64 1. 3 3 0. 27 61 47 0. 29 61 15 0. 31 60 84 0. 04 0 5. 01 55 26 5. 03 63 34 5. 13 47 14 0. 12 -1 .4 35 90 8 0. 55 01 52 2. 53 62 12 4. 0 4 0. 31 59 17 0. 34 75 36 0. 37 91 55 0. 06 3 5. 16 37 61 5. 07 51 20 5. 36 10 81 0. 20 -0 .3 20 16 4 2. 43 78 70 5. 19 59 04 5. 5 M ea n 0. 05 0. 2 5 SD 0. 02 6 0. 09 2. 7 b) py ri do xa l 1 0. 05 90 81 0. 08 27 63 0. 10 64 44 0. 04 7 4. 14 63 67 4. 20 35 42 4. 26 07 17 0. 11 -3 .4 23 12 0 -1 .4 90 64 0 0. 44 18 45 3. 86 2 0. 12 02 35 0. 12 77 33 0. 13 52 31 0. 01 5 3. 66 81 08 3. 68 94 92 3. 71 08 76 0. 04 3 -0 .9 66 25 0 -0 .4 25 84 0 0. 11 45 83 1. 08 3 0. 08 98 94 0. 10 90 92 0. 12 82 90 0. 03 8 4. 12 23 00 4. 16 53 00 4. 20 83 00 0. 08 6 -3 .1 17 33 0 -1 .5 62 72 0 -0 .0 08 12 0 3. 11 4 0. 02 90 69 0. 05 67 34 0. 08 43 98 0. 05 5 3. 69 05 27 3. 74 99 76 3. 80 94 25 0. 12 -3 .3 54 47 0 -1 .6 58 68 0 0. 03 71 22 3. 39 M ea n 0. 04 0. 09 3 SD 0. 01 7 0. 03 4 1. 2 c) py ri do xa m in e 1 0. 03 78 05 0. 05 35 77 0. 06 93 48 0. 03 2 11 .8 90 76 0 12 .1 65 78 0 12 .4 40 79 0 0. 55 -3 .8 61 20 0 -0 .1 21 58 0 3. 61 80 34 7. 5 2 0. 05 22 38 0. 05 61 51 0. 06 00 64 0. 00 78 8. 91 08 10 8. 94 92 23 8. 98 76 36 0. 07 7 -0 .4 96 50 0 0. 25 00 33 0. 99 65 64 1. 5 3 0. 03 03 68 0. 03 84 84 0. 04 66 00 0. 01 6 14 .5 38 65 0 14 .3 12 89 0 14 .5 38 65 0 0. 45 -4 .3 92 56 0 -1 .6 56 35 0 1. 07 98 70 5. 5 4 0. 04 94 24 0. 05 67 79 0. 06 41 34 0. 01 5 11 .8 31 12 0 11 .7 07 61 0 11 .5 84 10 0 0. 25 -1 .8 52 45 0 -0 .4 35 11 0 0. 98 22 24 2. 8 M ea n 0. 02 0. 3 4 SD 0. 01 0 0. 21 2. 7 580 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Table 13. Statistics of EU-MAT interlaboratory study on vitamin B 6 (Berg van den et al. 1996). Pig’s liver Mixed vegetables Wholemeal flour In-house Optimal In-house Optimal In-house Optimal Number of laboratories 12 10 12 10 12 11 Mean of means* 2.01 1.57 0.41 0.38 0.27 0.26 RSDr (%) 6 13 12 5 11 7 RSD R (%) 63 36 38 40 51 43 Number of outliers** – 1 – – – – * as PN (mg/100g) ** according to Nalimov test RSD r ; repeatability relative standard deviation, RSD R ; reproducibility relative standard deviation Fig. 24. Intercomparison results of EU-MAT study on vita- min B 6 x = mean value, laboratory 2 = University of Hel- sinki. Samples: a) pig’s liver, b) mixed vegetable and c) wholemeal flour. (Redrawn from the results of van den Berg et al. (1996) with permission from Dr. van den Berg). a b c be assessed by its results for the BCR’s certifi- cation study for reference materials; our labora- tory’s vitamin B 6 results were included within the 95% confidental interval calculated from the mean values derived from eleven European lab- oratories. 581 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Moreover, the sample extracts of pig’s liver, mixed vegetables, and whole wheat flour pre- pared by the Federal Research Centre for Nutri- tion (Stuttgart, Germany) were analyzed in three laboratories. Two liquid chromatographic pro- cedures (HPLC1; isocratic method, HPLC2; gra- dient method) and one microbiological assay were compared. The results derived from two chromatographic methods were correlative to each other (Fig. 25). There was a statistically significant difference in the results of pigs’ liver sample extract between liquid chromatographic and microbiological assay; the medians at the 95% confidence level indicated that a lower re- sult was obtained using a microbiological assay. This was not the case with the two other sam- ples. Whole wheat flour and beef steak samples were also analyzed in two laboratories in Fin- land (Fig. 26). In this comparison, the microbio- logical assay seemed to give the same (for beef sample) or a little higher value (for whole wheat four) compared to those of the liquid chromato- graphic method when the variation in the mean values for both methods is assumed to be the same (and of the magnitude of 12%). 5.4 Routine food analysis 5.4.1 Flesh foods: meat, fish and poultry The most abundant B 6 vitamers in beef, pork, lamb, reindeer, elk, and poultry meat were pyri- doxal-5’-phosphate and pyridoxamine-5’-phos- phate representing ca. 88.3–94.4 % of the vita- min B 6 content (Table 14). A significant amount of free pyridoxine was present only in beef shoul- der and kidney sample. Offals like pork liver, beef liver, broiler liver, beef kindey, and pork kidney also contained free vitamers, especially pyridoxamine. Poultry meat was rich in pyri- doxal phosphate. Inactive metabolite, 4-pyridox- ic acid, was the main vitamin B 6 compound in pork and beef liver but not in kidney or in poul- try liver. A low percentage of PLP was found in processed food items; sausages and meat balls Fig. 25. Interlaboratory results of the circulated sample extracts. Redrawn from the results of Bognar and Ollilainen (1997). Sample treatment: D1 = 5% trichloroacetic acid, 20°C, 30min, b-glucosidase, D3 = 0,1 M hydrochloric acid, 120°C, 30min, β-glucosidase, F1 = 5% trichloroacetic acid, 20°C, 30min, Takadiastase, F2 = 0,1 M hydrochloric acid, 120°C, 30min, Takadiastase. Methods: HPLC1 = isocratic elution (BFE, Germany), HPLC2 = gradient elution (University of Helsinki), MA = microbiological assay (LUF, Germany). (Redrawn from the results of Bognar and Ollilainen (1997) with permission from Dr. Bognar). 582 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods which contained mainly PMP. The highest vita- min contents were found in liver; 0.85, 0.81, and 0.66mg/100g for beef liver, broiler liver, and pork liver, respectively. Distribution of B 6 vitamer in fish resembled that of meat; the amount of vitamin B 6 consisted mainly of phosphorylated pyridoxal and pyridox- amine. However, free vitamers and especially pyridoxic acid were also present (Table 15). 5.4.2 Dairy products and egg Free and phosphorylated pyridoxal, and pyridox- amine phosphate but also to some extent free pyridoxamine, were typical for liquid milk prod- ucts whereas free and phosphorylated pyridox- amine formed the major vitamin fraction in cheese (Table 16). Pyridoxal phosphate was the dominating vitamer in egg yolk. Infant formula based on cow milk was fortified with pyridox- ine. Only a small amount of pyridoxine was found in baker’s yeast and egg yolk. In addition baker’s yeast contained a minor fraction of py- ridoxine phosphate. This vitamer was not found in any other food item in this study. Fig. 26. A two-sample (beef steak and wheat flour) com- parison, HPLC (Univ of Helsinki), microbiological assay (VTT Technical Research Centre, Finland). Table 14. Vitamer distribution and vitamin B 6 content of meat and meat products (mean±SD, n=3, as PN mg/100g fresh weight). Food item Vitamer Sum PLP PA PMP PL PN PM Minced meat, beef 0.17±0.004 0.006±0.0033 0.16±0.003 0.024±0.0014 0.0047±0.0006 tr 0.36 Minced meat, beef and pork 0.21±0.019 0.018±0.0026 0.075±0.0022 0.017±0.0001 0.014±0.0001 0.0053±0.00002 0.32 Beef shoulder 0.14±0.014 0.0043±0.00097 0.09±0.014 0.0061±0.00055 0.06±0.082 0.006±0.0011 0.30 Pork shoulder 0.22±0.014 0.020±0.0043 0.087±0.0094 0.010±0.0018 0.0159±0.00067 0.013±0.0015 0.35 Elk, steak 0.352±0.0089 n.d. 0.129±0.0038 0.022±0.0014 0.0062±0.00020 0.0061±0.00012 0.52 Lamb, steak 0.112±0.0044 0.0065±0.00097 0.060±0.0019 0.0106±0.00059 0.0071±0.00039 0.0047±0 0.19 Reindeer, steak 0.20±0.036 n.d. 0.187±0.0091 0.0093±0.00087 0.0057±0.00064 0.0086±0.00072 0.41 Beef liver 0.17±0.013 0.55±0.069 0.23±0.005 0.026±0.001 0.069±0.003 0.36±0.005 0.85 Beef kidney 0.019±0.0009 0.016±0.0011 0.070±0.0046 0.041±0.0029 0.073±0.0028 0.171±0.0042 0.37 Pork liver 0.135±0.0013 1.07±0.078 0.30±0.011 0.037±0.0056 0.0694±0.00053 0.116±0.0019 0.66 Pork kidney 0.108±0.0065 0.035±0.0083 0.269±0.0047 0.065±0.0049 0.0212±0.00027 0.073±0.0017 0.54 Broiler liver 0.093±0.0071 0.0076±0.00070 0.179±0.0043 0.11±0.011 0.021±0.0015 0.40±0.019 0.81 Broiler without skin 0.33±0.022 0.019±0.0011 0.054±0.0013 0.022±0.0030 0.0072±0.00093 0.0062±0.00014 0.42 Hen without skin 0.313±0.0079 0.023±0.0012 0.081±0.0015 0.015±0.0018 0.0100±0.00047 0.0061±0.00001 0.43 Dry sausage, salami type 0.0016±0.00025 0.0066±0.00058 0.095±0.0063 0.0110±0.00063 0.0147±0.00067 0.134±0.0062 0.26 Sausage, “lenkki” 0.0033±0.00093 0.0013±0.00080 0.042±0.0024 0.0187±0.00094 0.0073±0.00064 0.0058±0.00020 0.077 Meatballs 0.0159±0.00096 0.006±0.0013 0.0414±0.00043 0.0088±0.00070 0.017±0.0022 0.0090±0.00070 0.092 Broiler meatballs 0.029±0.0051 0.0124±0.00082 0.048±0.0025 0.0097±0.00056 0.0174±0.00020 0.0110±0.00036 0.11 PA; 4-pyridoxic acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PM; pyridoxamine, PMP; pyridoxamine-5’-phosphate, PN; pyridoxine, Sum; PLP+PMP+PL+PN+PM 583 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Table 15. Vitamer distribution and vitamin B 6 content of fish (mean±SD, n=3, as PN mg/100g fresh weight). Food item Vitamer Sum PLP PA PMP PL PN PM Baltic herring, fillets 0.061±0.0077 0.006±0.0010 0.08±0.014 0.0099±0.00084 0.00626±0.000096 0.021±0.0033 0.18 Pike, sea, fillets 0.13±0.012 0.0360±0.00076 0.039±0.0023 0.0069±0.00014 n.d. 0.00043±0.000056 0.17 Pikepearch, sea, fillets 0.11±0.024 0.036±0.0087 0.044±0.0072 0.0081±0.00033 0.0031±0.00020 0.005±0.0015 0.17 Rainbow trout, cultivated, fillets 0.301±0.0097 0.013±0.0010 0.058±0.0036 0.0093±0.00095 0.0147±0.00067 0.0055±0.00053 0.39 Whitefish, sea, fillets 0.29±0.010 1.9±0.33 0.046±0.0021 0.0219±0.00077 0.0056±0.00062 0.0049±0.00017 0.37 Roe paste, smoked, salted 0.0008±0.00069 n.d. 0.0005±0.0004 0.0014±0.00048 0.00289±0.00047 0.038±0.0010 0.0432 PA; 4-pyridoxic acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PM; pyridoxamine, PMP; pyridoxamine-5’-phosphate, PN; pyridoxine, Sum; PLP+PMP+PL+PN+PM Table 16. Vitamer distribution and vitamin B 6 content of dairy products, egg, and bakers yeast (mean±SD, n=3, as PN mg/ 100g fresh weight). Food item Vitamer Sum PLP PA PMP PL PN PM Milk, 1.9% fat 0.010±0.0011 0.022±0.0026 0.009±0.0013 0.020±0.0026 n.d. 0.0045±0.00056 0.044 Milk, 3.8% fat 0.0069±0.00048 0.0196±0.00068 0.006±0.0017 0.0200±0.00035 n.d. 0.0047±0.00014 0.038 Whipping cream 0.0035±0.00028 0.024±0.0011 0.005±0.0016 0.0054±0.00024 n.d. 0.0027±0.00014 0.017 Cheese, Edam 40 0.0004±0.00035 n.d. 0.030±0.0035 0.0016±0.00100 n.d. 0.015±0.0016 0.047 Cheese, Edam 20 n.d. 0.0048±0.00081 0.049±0.0028 0.0015±0.00047 n.d. 0.018±0.0027 0.068 Cream cheese n.d. 0.0019±0.00069 0.0229±0.00064 n.d. n.d. 0.007±0.0013 0.030 Skim milk powder 0.090±0.0013 0.101±0.0021 0.085±0.0086 0.168±0.0037 n.d. 0.042±0.0025 0.39 Infant formula* 0.006±0.0013 0.040±0.0080 0.050±0.0014 0.0499±0.00074 0.35±0.0126 0.050±0.0012 0.52 Egg yolk 0.303±0.0031 0.025±0.0014 0.0039±0.00064 0.0055±0.00029 0.0066±0.00064 0.0105±0.00044 0.33 Baker’s yeast** 0.04±0.015 n.d. 0.19±0.037 0.099±0.0049 0.017±0.0028 0.0471±0.00035 0.40 * fortified with pyridoxine **PNP: 0.019±0.0049 PA; 4-pyridoxic acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PM; pyridoxamine, PMP; pyridoxamine-5’-phosphate, PN; pyridoxine, Sum; PLP+PMP+PL+PN+PM 5.4.3 Plant-derived foods The presence of bound pyridoxine and several vitamers was characteristic for plant derived foods: six vitamer and glycosylated pyridoxine were present in almost every food item (Table 17a and 18a). Generally, pyridoxine and its bound form (pyridoxine glycoside) as well as free and phos- phorylated pyridoxamine form a significant por- tion of B 6 vitamer content in cereals. Typically, pyridoxal phosphate and pyridoxic acid were only minor components except in rolled oats in which a relatively large portion of pyridoxic acid 584 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods ished rice) to 51% (wheat germ), an average be- ing ca. 40% of the total vitamin B 6 content (Ta- ble 17b). Several vitamers including pyridoxine glycoside(s) were present in vegetables, roots and nuts and the ratio of glycosylated pyridox- ine to free and phosphorylated vitamers had a large variation; PNG % was ca. 8, 40 and 77 % for frozen peas, broccoli and carrot, respective- ly (Tables 18a-b). Nearly half of the total amount of vitamin B 6 in peanut was in glycosylated form(s) whereas only a low concentration of bound pyridoxine, 0.0021mg/100g representing less than 1% of total vitamer content, was found in hazelnut. The PNG portion in baby-foods was higher in foods based on vegetable ingredients than those based on mixed vegetable and meat ingredients: ready-to-eat potato-carrot puree baby-food contained almost two third of its to- tal vitamer as glycosidic pyridoxine. 5.4.4 Comparison with national food composition tables The present data was generally in accordance to the results published in national food composi- Table 17a. B 6 vitamer distribution in cereal and rice (mean±SD, n=3, as PN mg/100g fresh weight). Food item Vitamer PLP PA PMP PL PN PM PNG Barley flour 0.007±0.0020 0.0254±0.00089 0.0081±0.00046 0.009±0.0011 0.028±0.0010 0.033±0.0012 0.077±0.0076 Oats, rolled n.d. 0.108±0.0051 0.032±0.0020 0.013±0.0014 0.033±0.0030 0.0272±0.00078 0.03±0.016 Rye flour 0.0109±0.00032 n.d. 0.024±0.0013 0.0144±0.00072 0.062±0.0016 0.02866±0.000064 0.119±0.0044 Wheat flour, graham 0.0074±0.00074 n.d. 0.016±0.0016 0.012±0.0012 0.040±0.0040 0.016±0.0016 0.036±0.0036 Mixed wheat and rye flour 0.0079±0.00080 0.0072±0.00021 0.0099±0.00086 0.0100±0.00036 0.031±0.0014 0.0138±0.00020 0.071±0.0019 Wheat germ 0.083±0.0014 n.d. 0.152±0.0032 0.037±0.0037 0.156±0.0059 0.108±0.0036 0.56±0.012 Rice. polished 0.0038±0.00031 0.0088±0.00072 0.0087±0.00061 0.0129±0.00095 0.021±0.0013 0.0175±0.00082 0.020±0.0028 PA; 4-pyridoxic acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PM; pyridoxamine, PMP; pyridoxamine-5’-phosphate, PN; pyridoxine, PNG; pyridoxine glucoside (calculated as PN) Table 17b. Vitamin B 6 and total vitamin B 6 contents and PNG percentage of cereal and rice (mean±SD, n=3, as PN mg/100g fresh weight). Food item ΣB6 ΣΣB6 PNG% Barley flour 0.084 0.16 47.7 Oats, rolled 0.11 0.14 23.4 Rye flour 0.14 0.26 45.8 Wheat flour, graham 0.091 0.13 28.5 Mixed wheat and rye flour 0.073 0.14 49.1 Wheat germ 0.54 1.1 51.0 Rice. polished 0.064 0.084 23.5 ΣB6; PLP + PMP + PL + PN + PM, ΣΣB6; ΣB6 + PNG, PNG%; PNG/ΣΣB6 × 100 was found. The highest vitamer content in flour samples was found in rye, 0.26mg/100g, al- though the difference between analyzed flour samples was rather small. Barley flour, mixed wheat and rye flour, and polished rice contained all six vitamers, even though pyridoxic acid was not measured or found in whole wheat or rye flour. Glycosidic derivative(s) of pyridoxine was found in all cereal samples including rice, and its portion varied from 23% (rolled oats and pol- 585 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Table 18a. B 6 vitamer distribution in plant-derived foods (mean±SD, n=3, as PN mg/100g fresh weight). Food item Vitamer PLP PA PMP PL PN PM PNG Broccoli 0.032±0.0024 0.106±0.0073 0.0149±0.000810 0.0214±0.00009 0.021±0.0016 0.0021±0.0002 0.060±0.0050 Carrot 0.009±0.0026 0.0044±0.00089 0.0019±0.00016 0.011±0.0012 n.d. 0.00057±0.00053 0.073±0.0047 Potato 0.028±0.0055 0.030±0.0033 0.007±0.0042 0.018±0.0026 0.020±0.0017 0.024±0.0013 0.17±0.032 Tomato 0.0093±0.00060 0.020±0.0036 0.0030±0.00070 0.012±0.0015 0.023±0.0050 0.023±0.0015 0.017±0.0038 Peas, frozen 0.023±0.0026 n.d. 0.0395±0.00083 0.0109±0.00026 0.0080±0.00049 0.0136±0.00054 0.0087±0.00071 Hazelnuts 0.014±0.0066 0.067±0.0027 0.0042±0.00079 0.30±0.038 0.0032±0.00048 0.00396±0.000056 0.0021±0.00056 Peanuts n.d. 0.061±0.0042 0.0107±0.00043 0.050±0.0099 0.036±0.0032 0.0127±0.00030 0.082±0.0051 Baby food, veal and vegetables n.d. 0.016±0.0023 0.0151±0.00054 0.022±0.0017 0.0013±0.00018 0.0120±0.00026 0.0210±0.00064 Baby food, potato and carrot puree n.d. n.d. 0.0025±0.00013 0.0143±0.00045 n.d. 0.0037±0.00020 0.0381±0.00088 PA; 4-pyridoxic acid, PL; pyridoxal, PLP; pyridoxal-5’-phosphate, PM; pyridoxamine, PMP; pyridoxamine-5’-phosphate, PN; pyridoxine, PNG; pyridoxine glucoside (calculated as PN), ΣB6; PLP + PMP + PL + PN + PM, ΣΣB6; ΣB6 + PNG, PNG% ; PNG/ΣΣB6 × 100 Table 18b. Vitamin B 6 and total vitamin B 6 contents and PNG percentage of plant-derived foods (as PN mg/100g fresh weight). Food item ΣB6 ΣΣB6 PNG% Broccoli 0.091 0.15 39.9 Carrot 0.022 0.096 76.7 Potato 0.098 0.27 63.6 Tomato 0.070 0.087 19.7 Peas, frozen 0.095 0.10 8.4 Hazelnuts 0.33 0.33 0.6 Peanuts 0.11 0.19 42.7 Baby food, veal and vegetables 0.051 0.072 29.3 Baby food, potato and carrot puree 0.021 0.059 64.9 ΣB6; PLP + PMP + PL + PN + PM, ΣΣB6; ΣB6 + PNG, PNG% ; PNG/ΣΣB6 × 100 tion tables (Tables 19–23). However, meat sort- ing or labeling was found to hamper this com- parison. One of the main differences was found in the values for baker’s yeast; relatively high value was reported in Danish food composition table compared to other food composition tables. This difference can not only be explained by the variation in dry matter content. 586 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Table 19. Comparison of national food data bases; vitamin B 6 content of meat and meat products (as PN mg/100g fresh weight). Den1) Fin Ger Swe UK1) US1) This study Minced meat, beef – 0.27 .. 0.44 0.22 0.22 0.36 Minced meat, beef and pork – – – – – – 0.32 Beef shoulder 0.360 0.32 .. 0.50 0.19 – 0.30 Pork shoulder 0.26 0.31 .. 0.36 0.37 0.29 0.35 Elk, steak – .. – 0.44 – .. 0.52 Lamb, steak 0.17 0.20 0.13 0.20 0.17 0.14 0.19 Reindeer, steak – .. – 0.44 – .. 0.41 Beef liver 0.83 0.83 0.71 1.00 0.69 0.78 0.85 Beef kidney 0.36 0.32 0.39 0.43 0.26 0.42 0.37 Pork liver 0.70 0.68 0.59 0.85 0.56 0.57 0.66 Pork kidney 0.370 0.25 0.55 0.44 0.21 0.36 0.54 Broiler liver 0.66 0.80 0.80 0.80 0.33 0.63 0.81 – Broiler without skin 0.36 0.43 0.50 0.43 0.35 0.36 0.42 Hen without skin 0.50 0.6 .. – – 0.23 0.43 Dry sausage, salami type 0.12 0.15 .. 0.15 0.12 0.09 0.26 Sausage, lenkki – 0.15 0.14 0.15 – 0.11 0.077 Meatballs – – – 0.20 – – 0.092 Broiler meatballs – – – 0.20 – – 0.11 1) values recalculated to pyridoxine base .. value missing – food item not comparable or missing Den; Møller 1985, Fin; Rastas et al. 1993, Ger; Souci et al. 1994, Swe; SLV 1993, UK; Holland et al 1991, US; USDA 1999 Table 20. Comparison of national food data bases; vitamin B6 content of fish (as PN mg/100g fresh weight). Den1) Fin Ger Swe UK1) US1) This study Baltic herring, fillets – 0.37 .. 0.37 – – 0.18 Pike, sea, fillets 0.12 0.12 0.15 0.12 – 0.097 0.17 Pikepearch, sea, fillets – 0.3 .. 0.12 – – 0.17 Rainbow trout, cultivated, fillets – 0.98 .. 0.69 – 0.512 0.39 Whitefish, sea, fillets – 0.3 – 0.12 – 0.248 0.37 Roe paste, smoked, salted 0.17 .. – 0.2 – – 0.043 1) values recalculated to pyridoxine base .. value missing – food item not comparable or missing Den; Møller 1985, Fin; Rastas et al. 1993, Ger; Souci et al. 1994, Swe; SLV 1993, UK; Holland et al 1991, US; USDA 1999 587 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Table 21. Comparison of national food data bases; vitamin B 6 content of dairy products and egg (as PN mg/100g fresh weight). Den1) Fin Ger Swe UK1) US1) This study Milk, 1.9% fat 0.040 0.04 0.046 0.05 0.05 0.036 0.044 Milk, 3.9% fat 0.039 0.04 0.036 0.04 0.05 0.035 0.038 Whipping cream 0.016 0.02 0.036 0.03 0.03 0.022 0.017 Cheese, 20 Edam type 0.066 0.07 – – – – 0.068 Cheese, 40 Edam type 0.07 0.06 0.073 0.09 0.07 0.063 0.047 Cream cheese 0.046 0.04 0.056 0.06 0.03 0.039 0.030 Infant formula, fortified – – – – 0.52 Skim milk powder 0.40 0.6 0.280 0.49 0.50 0.286 0.39 Egg yolk 0.25 0.3 0.300 0.3 0.25 0.324 0.33 Baker’s yeast 0.91 – 0.684 0.6 0.50 0.356 0.40 1) values recalculated to pyridoxine base – food item not comparable or missing Den; Møller 1985, Fin; Rastas et al. 1993, Ger; Souci et al. 1994, Swe; SLV 1993, UK; Holland et al 1991, US; USDA 1999 Table 22. Comparison of national food data bases; vitamin B 6 content of cereal (as PN mg/100g fresh weight). Den1) Fin Ger Swe UK1) US1) This study2) Barley flour 0.23) 0.33 0.5603) 0.33 – 0.263 0.16 Wheat flour, graham 0.41 0.35 0.46 0.35 0.41 0.282 0.13 Mixed wheat and rye 0.09 – – 0.30 – – 0.14 Oats, rolled 0.17 0.18 0.160 0.18 0.27 .. 0.14 Rye flour 0.29 0.35 0.233 0.35 0.29 0.222 0.26 Wheat germ 0.79 0.492 1.82 2.73 1.08 1.1 Rice, polished 0.12 0.30 0.150 0.15 0.26 0.290 0.084 1) values recalculated to pyridoxine base 2) glycosylated forms included 3) whole grain .. value missing – food item not comparable or missing Den; Møller 1985, Fin; Rastas et al. 1993, Ger; Souci et al. 1994, Swe; SLV 1993, UK; Holland et al 1991, US; USDA 1999 588 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Table 23. Comparison of national food data bases; vitamin B 6 content of plant-derived foods (as PN mg/100g fresh weight). Den1) Fin Ger Swe UK1) US1) This study2) Broccoli 0.17 0.4 0.280 0.16 0.11 0.132 0.15 Carrot 0.054 0.12 0.270 0.07 0.11 0.122 0.096 Potato 0.12 0.44 0.307 0.26 0.36 0.215 0.27 Tomato 0.083 0.10 0.100 0.10 0.11 0.066 0.087 Peas, frozen 0.108 0.12 0.0463) 0.12 0.07 0.101 0.10 Hazelnuts – 0.61 0.313 0.61 0.49 0.506 0.33 Peanuts 0.4 0.59 0.440 0.30 0.49 0.288 0.19 Baby food, veal and vegetables – – – – 0.098 0.072 Baby food, potato and carrot puree – – – – 0.067 0.059 1) values recalculated to pyridoxine base 2) glycosylated forms included 3) canned – food item not comparable or missing Den; Møller 1985, Fin; Rastas et al. 1993, Ger; Souci et al. 1994, Swe; SLV 1993, UK; Holland et al 1991, US; USDA 1999 5.5 Characterization of bound pyridoxine The reversed-phase chromatogram of whole wheat extract contained three main unknown peaks, named X 1 (k e ~8.3), X 2 (k e ~10.0), and X 3 (k e ~11.8) which disappeared during the sample extract digestion with β-glucosidase (Fig. 27). Further emphasis was focused on the X 3 frac- tion which eluted before pyridoxal (k e ~12.2). The chemical structure of isolated X 1 or X 2 fractions remains unclear. The X 3 fraction was collected and repurified by open-column ion-exchange chromatography. The amount of isolated and purified X 3 analyte (later called PNX), quanti- tated as pyridoxine using an external standard method, was approximately 0.075µmol (13µg) and 0.1µmol (17µg) starting from 1000g of raw carrot and 750g of wholemeal flour. The recov- ery after ion-exchange purification was estimat- ed to be ca. 65%. 5.5.1 Enzymatic hydrolysis The collected PNX fractions were combined and a part of the pooled fraction was enzymatically hydrolyzed using a β-glucosidase preparate; iso- lated PNX yielded the increased fluorescence signal at a k e value of 13.6 in reversed-phase chromatogram. Thus, the enzymatic hydrolysis of the PNG fraction increased the size of the chromatographic peak at the region of PN. Spik- ing of the enzymatically treated PNX extract with pyridoxine standard solution confirmed this. The amount of pyridoxine derived from bound form was almost equimolar to PNX calculated as free pyridoxine. 5.5.2 Proton NMR spectroscopy The isolated and purified PNX fraction was dis- solved in deuterium oxide (99.8%), lyophilized and redissolved in deuterium oxide (99.95%). The proton spectrum of the isolated PNX frac- 589 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Fig. 27. A reversed-phase chroma- togram of carrot sample: a) before b-glucosidase hydrolysis, b) after β-glucosidase hydrolysis and, c) enzyme blank. Fig. 28. Proton spectrum of PNX 3 fraction isolated from wholemeal flour. σ (ppm) 590 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Table 24. Signals in 1HNMR spectrum of PNX isolated from wholemeal flour measured in deuterium oxide (t=35°C). Chemical shift Signal form Coupling constant Tentative Analyte δ H (ppm) (intensities) J (Hz) characterization 0.9 Unknown 1.3 Unknown 1.35 multiple methyl protons in unknown (impurity)** CH3CH2OH, impurity 1.6 Unknown 1.8 Unknown 2.8 Unknown 3.2 Unknown 3.85 douple douplet Ja,b ~ 3.4 carbohydrate moiety* (1:3) 3.95 douple douplet Ja,b ~ 3.8 carbohydrate moiety* (3:1) 4.15 quartet Ja,b ~ 7,2 methylene protons in unknown (impurity)** (1:3:3:1) Ja,c ~ 13.9 CH 3 CH 2 OH, impurity Ja,d ~ 20.7 Jb,c ~ 6.8 Jb,d ~ 13.5 Jc,d ~ 6.8 4.35 quartet (?) Ja,b ~ 3.4 carbohydrate moiety* (1:3:3:1) Ja,c ~ 3.0 Ja,d ~ 9.4 Jb,c ~ 3.0 Jb,d ~ 6.0 Jc,d ~ 3.0 4.45 douple douplet Ja,b ~ 3.0 anomeric H in C1-glu carbohydrate moiety* (1:1:1:1) Ja,c ~ 4.9 Ja,d ~ 8.3 Jb,c ~ 1.9 Jb,d ~ 5.3 Jc,d ~ 3.4 4.6–4.8 Water 4.85 douplet carbohydrate moiety*, signal disappeared at 23°C 6.15 douplet Ja,b ~ 6.4 carbohydrate moiety* 8.30 two singlets (J ~ 42.5) aromatic H in C6-pyr pyridoxine moiety 8.40 tion of wholemeal flour showed that the analyte was pure enough for proper NMR measurement (Fig. 28) whereas the fraction isolated from car- rot still contained impurities. Therefore, the pro- ton spectrum of PNX fraction (Table 24) and COSY and TOCSY spectra (Fig. 29) of only whole wheat flour were measured. The NMR spectra of D-glucose and pyridoxine hydrochlo- ride in deuterated water were also recorded (Ta- ble 25). 591 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. Fig. 29. COSY- (a) and TOCSY-spectra (b) of PNX fraction of wholemeal flour. a b 592 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods 5.5.3 FAB mass spectrometry FAB mass spectra of PNX fraction isolated from whole meal flour and carrot samples were meas- ured in glycerol, nitrobenzylalcohol and trieth- anolamine matrices (Appendix 2). Table 25. Signals in 1HNMR spectrum of D-glucose and pyridoxine hydrochloride measured separately in deuterium oxide (t=35°C). Chemical shift Signal form Coupling constant Tentative Analyte δ H (ppm) (intensities) J (Hz) characterization 4.57 douplet Ja,b ~ 8.1 axial anomeric proton D-glucose (1:1) in β-anomer 4.7 singlet impurity water 5.16 douplet Ja,b ~ 3.9 equatorial anomeric proton in D-glucose (1:1) α-anomer Intensities of δ H 4.57 and δ H 5.16 signals: 1:2 α- and β-anomer in D-glucose equilibrium stage in deuterium oxide 2.5 singlet methyl protons pyridoxine; 2C-CH 3 4.7 singlet impurity water 4.9 singlet (?) 4’- methylene protons pyridoxine; (-CH 2 -O-) 5C-5’CH 2 -O- 8.0 singlet aromatic proton pyridoxine; 6C-H Intensities of δ H 2.5, δ H 4.9 and δ H 8.0 signals: 1:4.5:4.9 6 Discussion 6.1 Liquid chromatography 6.1.1 Column packings The chromatographic retention of B 6 vitamers based on partition chromatography is mainly affected by the substitution of the pyridinium ring while the peak symmetry is dependent on the cationic nature of the analyte. The cationic nature of B 6 vitamers, due to the pK-values of pyridinium nitrogen group and functional groups (4-carboxal, 4-hydroxymethyl, and 4-aminome- thyl) is characteristic for these compounds. This ionic nature of vitamers enables the use of ion- exchange- or ion-pair-chromatography in their separation. It was therefore expected that the proper resolution and separation efficiency achieved with octadecyl packings is a result of both ionic and partition characteristics of the analytes. An aminopropyl phase possessed only a limited retention for vitamin B 6 compounds, as can be also seen in the report of Belal (1989), and its use was thus rejected. Adsorption activity Peak tailing (poor peak symmetry), caused by the coulombic interaction between cationic vi- tamers (pK-value for pyridinium ring 7.9–9.0 according to Snell 1963) and especially that of in PM (aminomethyl group; pK 10.5) and either residual silanols and/or metal activity of the base 593 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. silica was found at some extent with all column packings. This characteristic should be especially true for non-end capped bounded phases (Sta- dalius and Snyder 1988). The effect of this un- favorable phenomenon on separation efficiency can be deminished by choosing a fully end- capped, low-acidic, and low-metal activity sili- ca material. The concentration of silanol groups in silica packing was presumed to be ca 8µmol/ m2 (Unger et al. 1976, Köhler and Kirkland 1987). Lower silanol surface concentration val- ues expressed as cation-exchange capacity, 0.7– 0.8 meq/g for plain silica , were obtained when the surface silanol was measured by direct titra- tion with alkali in the presence of aqueous salts (Khurana and Ho 1988). It should be noted that in the latter study only one commercial silica gel packing was tested. It has been estimated that due to steric hindrance in the silica surface ge- ometry only half of the silanol surface can be bound and the rest of the surface, ca. 4.2 µmol/g (Nasuto and Rózylo 1995, Bereznitski et al. 1996, Buszewski et al. 1997), must be deriva- tized by using an additional process. The desired silica packing for basic compounds should only contain a few unbonded groups for adsorption interaction. Various methods, such as secondary silanization (Nasuto and Rozylo 1995), douple endcapping (Fiorianti et al. 1995), the use of a high density coverage (Buszewski 1991) or mo- bile phase additives (Musch and Massert 1988, Li 1992), and various silica treatment procedures (Köhler and Kirkland 1987) have been applied to reduce the unwanted residual silanol activity. An excellent peak shape for the B 6 vitamers in the present study was achieved with a µBon- dapak packing (Table 7) even thought a relative high silanol index value was measured for that particular packing material by Walters’ test (Ta- ble 8). Similar results are reported by Wehling and Wetzel (1984) and Bötticher and Bötticher (1987). The efficiency of this column material (expressed as plate numbers) was lower than those of other octadecyl phases (except for pol- ymer phases) due to larger particle size (10µm), and contrastingly the silanol index value was clearly higher than in other ODS phases. Thus, silanol activity alone did not explain this result. Asymmetry values for B 6 vitamers were gener- ally higher using Spherisorb ODS2 and Vydac HS packings. The difference of a µBondapak or Novapak packing compared to other octadecyl phases is propably due to silica synthesis chem- istry but its details were not available. With oth- er column packings, peak shapes, especially for pyridoxamine, could be improved by adding tri- ethylamine to the mobile phase as a modifier (Table 7). However, the asymmetry values should be considered only as a guideline since the measurement of column efficiency is affect- ed by many factors. Even the slightest use of amine modifiers, ion-pair reagents or buffer salts in the mobile phase can change the efficiency, and once the column has been exposed to these compounds, the native selectivity/efficiency may be permanently changed. The use of basic mod- ifiers in the mobile phase may also shorten the column’s life. In addition, the variation in pH of the mobile phase after adding triethylamine or ammonia into the mobile phase was considered to be harmful to the succesful use of amine mod- ifiers. Traces of various metals at some extent are always incorporated in silica matrix (van den Driest et al. 1988); concentration levels in com- mercial parent silica can vary remarkably: e.g. 10–350 µg/kg of aluminium, 40–420µg/kg iron, and 10–5600µg/kg sodium. Their concentration should be under 50 ppm level which is suggest- ed to be an accepted upper limit for ideal sup- port (Buszewski et al. 1997). The interaction between metal impurities in the silica support and ionic analyte may be seen as tailing peaks and as reduced column efficiency. The deterio- ration of column efficiency for ionic compounds is reported to be quite similar to residual silanol effect. Accurate data, however, on metal impu- rity concentrations in commercial silica supports is not readily available. This makes it difficult to evaluate what is the primary reason for the poor peak shape of ionic analytes in certain pack- ing materials. Unwanted adsorption interactions were ex- pected to be diminished by choosing a high hy- 594 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods drophobicity, and a low silanol activity column packing material such as Waters Novapak C18 support (Waters, USA). Phase type and pore size The octadecyl column packings in this compar- ison were of the monofunctional (both Waters packings) and tri- or polyfunctional type (Spherisorb and Vydac packings). Monomeric chemically bounded phases in liquid chromatog- raphy are probably more preferred to polymeric phases; the advantages of a monomeric phase are faster mass transfer, increased stability in aque- ous solvents, and more homogenous organic coating with decreased amounts of residual si- lanol groups. Thus, many of the current com- mercial packings are synthesized using the mo- nosilane process. The monomeric chemically bounded silica phase combined with a high sur- face coverage was the most apolar stationary phase and possessed the maximal selectivity for aromatic test substances (Buszewski and Galush- ko 1995). It was also proposed that the chemi- cally bound ligands of the phases with dense CBP coverage existed in a more ordered configura- tion owing to stronger chain-to-chain interac- tions. However, certain compounds with some de- gree of planarity in their molecule shape are more effectively separated using polymeric phases; better resolution of polycyclic aromatic hydro- carbons (Sander and Wise 1984) and carotenoids (Jinno and Lin 1995) have been achieved using polymeric coatings instead of monomeric phas- es. By decreasing the temperature, an enhanced selectivity for separation of α- and β-carotenes was obtained, the optimum temperature being at 20°C. Phase transitions in chemically bonded phase were then assumed to cause this selectiv- ity – temperature -dependence. In addition, col- umn selectivity was assumed to be more direct- ly related to bonded phase surface coverage val- ues (N as µmol/m2) whereas the absolute reten- tion was more dependent on carbon loading. The morphology of the polymeric phases and its rel- evance to specificity of separation, however, is still poorly understood. Column separation efficiency using two poly(styrene-divinylbenzene)(PS-DVB) based polymeric reversed-phase packings, Polyspher RP-18 and Hamilton PRP-1, was remarkably lower than that of silica based materials. How- ever, better column efficiency values for an ani- on-exchange polydivinylbenzene resin have been reported (Nair et al. 1996). The decreased effi- ciency could not be explained by the particle size as d p of the polymeric supports was almost equal to those of silica based particles. Generally speaking, retention factor values were much higher than those of silica based supports show- ing an increased hydrophobicity of the polymeric packing. Hydrophobicity value for the polymer- ic packing could not be determined by using the test procedure developed for chemically modi- fied octadecyl silica packings. Thus, the column chemistry related to polymeric support and its functionality differs notably from ordinary, sili- ca based chemically bound octadecyl phases. This was also shown e.g. by Pastores et al. (1995) in their work with a non-porous PS-DVB pack- ing which was developed to eliminate all pore diffusion effects. The pore size of the parent silica material is recommended to have a sufficient large diame- ter and pore system to ensure the fast mass trans- fer of solutes and adsorption-desorption kinet- ics. For low molecular weight analytes (M W <3kD), pore sizes of 6–15nm is recommended (Regnier 1987). The columns tested in the present work belong to this category. Larger pore size was needed for efficient separation of com- pounds with higher molecular weight (Pearson et al. 1982). The effect of pore structure chang- es during the surface modification on mass trans- fer of the solutes is still somewhat unclear (Un- ger et al. 1976). Several chemically bounded octadecyl phas- es were tested to separate B 6 vitamers in various food matrices. Reversed-phase columns tested were based on the literature review; the separa- tion efficiency achieved by small mesopore re- versed-phase particles(d p ~3–5µm) is normally superior to e.g. ion-exchange chromatography. The sufficient retention factor of even phospho- 595 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. rylated compounds (namely PLP and PMP) can be achieved by using octadecyl phases. The re- tention factor values (k e ) for PLP, PA, PMP, PL, PL, PN, DPN, and PM were 1.4, 2.1, 5.6, 12.2, 13.6, 14.6 and 21.3, respectively. The relatively high retention factor for pyridoxamine (k e > 20) was needed to avoid coelution of pyridoxamine with two baseline disturbance “peaks” in reten- tion factor values between 15 and 17. System peaks were assumed to be the source of these baseline disturbances. Tentatively identified 5’- O-β-D-glucopyranosyl pyridoxine was eluted at the k e -value of 11.8, just before PL. Even ade- quate relative retention was obtained for the first eluted B 6 vitamer, pyridoxal-5’-phosphate. How- ever, early eluting substances may time to time, depending on the sample matrix, disturb PLP’s proper baseline measurement. This column packing comparison showed the importance of column testing or their evaluation by other means to make sure that column pack- ing is suitable for a particular analytical purpose, and this is especially true for basic analytes. The development of column packings and their chem- istry may be considered to be the main driving force in liquid chromatrography today, and an ever-increasing number of different products are launched every year. This means that any col- umn test, such as ours, should be considered as a guideline as it represents the situation at the time it was done. 6.1.2 Mobile phase and column temperature The mobile phase used for routine food analysis was a modification of Gregory and Ink’s (1987) method consisting of a gradient elution with a mixture of 2-propanol and phosphate buffer. A better column efficiency for separating B 6 vita- mers was achieved using 2-propanol as the or- ganic modifier than using methanol or ace- tonitrile. Methanol has been reported to yield a better peak shape of basic compounds compared to acetonitrile when the influence of the organic solvent modifier in reversed-phase chromatog- raphy was studied by McCalley (1995). In our study a more stable retention for B 6 vitamers was achieved by using a rather low pH (2.2) of 50mM phosphate buffer in the mobile phase. The re- tention of pyridoxamine was adjusted with oc- tanesulfonic acid as ion-pair reagent in the mo- bile phase. The retention of pyridoxamine can be readjusted whenever needed by changing the length of alkyl chain of the ion-pair reagent (hy- drophobicity/size of the ion-pair moiety). The use of octanesulfonate (sodium salt) produced an adequate retention as well as improved peak resolution and peak shape for pyridoxamine in our chromatographic system. Concentration of the ion-pair reagent in the mobile phase (C m ) was adjusted to 8mM in the present work in line with the findings of Dong et al. (1988) and Knox and Hartwick (1981). The latter group measured the maximum concentration of alkyl sulfates lo- cated in the octadecyl bonded silica surface. This reflected the maximum surface concentration (C s ) of ca. 2 µmol/m2 for octyl sulphonate and pointing to a nearly monomolecular layer of ad- sorbed counter ion. The critical micelle concen- tration (CMC), the concentration after which the relative retention would sharply decrease, was still significantly higher for the alkyl sulphonates studied. The same kind of findings were also earlier reported by Giles and coworkers (1974). In the work of Knox and Hartwick it was as- sumed that the degree of retention of a positive- ly charged analyte was directly based on the charge formed on the surface of stationary phase, proving that the formation of the ion-paired com- plex (hetaeron – cationic compound) in the mo- bile phase plays a minor role, if any. Column stability is affected by the type and the concentration of the organic mobile phase modifiers. A decreased stability in basic eluents containing methanol was reported using phos- phate or carbonate buffer compared to borate or glycine buffers (Claessens et al. 1996). Howev- er, the usefulness of borate or glycine buffer in vitamin B 6 analysis is somewhat limited due to its restricted buffer capacity in the mobile phas- es at low pH ranges. The acidic pH of the mo- 596 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods bile phase in the reversed-phase chromatogra- phy was needed to achieve the proper selectivi- ty and retention for the vitamers. Phosphate buff- ers were also considered to be free from some interacting or unwanted reactions; borate buff- ers form complexes with B 6 vitamers while gly- cine buffers may interact with pyridoxal (Met- zler and Snell 1955). As the column temperature was set at 30°C without thermostating the mobile phase, the for- mation of a temperature gradient in the column was evident. It was estimated that the ambient room temperature in the laboratory normally varied between 20 and 25°C. Thus, the radial temperature profile caused by the viscous heat dissipation of the eluent was opposed to that formed by the unthermostatted mobile phase. The temperature difference in our chromato- graphic system is in accordance with the results of Welsch et al. (1996) which showed that the optimum precooling of the eluent was achieved when the temperature of the eluent was ca. 8°C below the column temperature. This temperature difference was estimated to be enough to com- pensate for the viscous heat dissipation, and its advantages for the separation efficiency and re- producibility were stated in their report. 6.1.3 Post-column derivatization The molar absorptivities as well as relative flu- orescence responses of B 6 vitamers differ notice- able from each other and are influenced by the selected pH value (Metzler and Snell 1955). The lowest response (expressed as a ratio of fluores- cence intensity and concentration) was measured with pyridoxal and pyridoxal-5’-phosphate, and with 4-deoxypyridoxine (internal standard). The fluorescence response for PLP using bisulphite adduct reagent as a post-column reagent was ca. five-fold compared to that of an underivatized compound (Fig. 14). The effect on other vitam- ers was weaker. Only the response of pyridoxal decreased with post-column derivatization, and the difference was considered to be insignificant in practical laboratory work. The use of pre-col- umn derivatization with sodium hydrogensul- phite in an ion-exchange chromatography of B 6 vitamers reduced the fluorescence intensity of other vitamers (Argoudelis 1988). This contrasts with our results but may be explained by the dif- ferent pH-values of the mobile phase used in their ion-exchange and in our present partition chromatography. Optimum conditions in our study were achieved by pumping the derivatiza- tion reagent, 35mM sodium hydrogensulphite in 0.5M phosphate buffer (pH 7.5) into the mobile phase (Fig.15). Increasing either the amount of hydrogen sulphite in the derivatization reagent or the flow rate of the derivatization solvent did not yield a higher response. 6.2 Evaluation of the chosen method for routine food analysis 6.2.1 Acid hydrolysis and the stability of PLP Perchloric acid extraction procedure was tested using 14C-labeled pyridoxine-5’-phosphate. The measured total 14C-activity of purchased stand- ard was 314 kBq the purity being 76.0% (Fig. 18a). 12% and 8% of the total 14C-activity was eluated in the chromatogram at the same reten- tion as unknown 1 (k e 0.5) and PMP (k e 4.5), re- spectively. Unknown 1 represented a breakdown product derived from PLP; as 14C-carbon atoms were located in a rather stable pyridinium ring in positions 4 and 5, this breakdown product probably still consisted of a pyridinium ring. The second minor activity fraction was eluted at a retention factor of 4.5, which is the same as that of PMP. When different sample extraction pro- cedures for lyophilized pork liver samples were tested, a portion of pyridoxal was assumed to have been converted into pyridoxamine (Bog- nar and Ollilainen 1997). This transamination phenomenon was then considered to be the re- sult of a long incubation time in takadiastase 597 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. hydrolysis, and was caused by the sample ma- trix as well. Whether this conversion occurs sim- ilarly in different buffer solutions remains un- clear. Only a minor change in the activity distribu- tion occured when labeled pyridoxal phosphate standard was extracted with perchloric acid. The activity of PLP decreased during perchloric acid extraction, and an increase in activity was found at the retention factor values of pyridoxamine phosphate, unknown 3 and pyridoxal. The total change approx. 3% was measured. Thus, the most labile B 6 vitamer in this study, pyridoxal 5’-phosphate, can be quantitatively extracted using an ice-cold perchloric acid precedure and then separated with ion-paired reversed-phase chromatography in its nearly intact form when the sample matrix effect is excluded. The sample matrix affected on the stability of PLP during the extraction. When labeled phos- phorylated pyridoxal was exposed to beef, car- rot and whole wheat flour matrices, and then extracted and measured via HPLC, the distribu- tion of 14C activity was changed. 14C-activity of labeled PLP added to beef steak and carrot sam- ples was partially lost during the extraction pro- cedure. The main activity increase was located in the fractions of pyridoxal and unknown 2 (Fig. 19). The increased amount of free pyridoxal due to hydrolysis of phosphate ester lingage of PLP was, however, only 7% and 14% in beef and carrot sample matrices, respectively. Added 14C- labeled PLP consisted ca. 8% of the total activi- ty as an “impurity” of which the main part was located after extraction procedure either in PL or in unknown 2 fraction. The results showed that the conversion of one B 6 vitamer to another vitamer form in these conditions includes both interconvertion of a functional group and ester linkage hydrolysis. This was also found in the work of Bognar and Ollilainen (1997). The main B 6 vitamers naturally present in beef are PLP, PMP and PL representing 65%, 23% and 5% of the vitamin content making it a suitable matrix for stability testing. In the case of whole wheat flour, 26% of originally added 14C-activity was found either in pyridoxal or in PNG fraction since these derivatives were eluted very near each other and could not be exactly separated using this fraction collection. In carrot, ca. 70% of total vitamin fraction is made up of gluco- sylated pyridoxine, while PLP (9%) and pyri- doxal (11%) formed the minor vitamer fractions. The chemical structure of the suggested break- down product of PLP, referred as unknown com- pound 2, was not solved. When labeled PLP was extracted without any sample matrix, the main fraction containing 14C-activity was unknown compound 3 instead of unknown 2 in the food matrices tested. Changes in these 14C-activity patterns were considered to be reflected in the matrix effect on the vitamin B 6 compound distribution. In wheat matrix the main change, probably the de- phosphorylation of PLP into free pyridoxal, took place to a greater extent than in other two matri- ces. It is noteworthy that the interconversion of pyridoxal to pyridoxamine was quite minimal, and a small increase of pyridoxamine was ob- served showing that vitamin B 6 compounds in beef and carrot matrices can be measured nearly in their unaltered forms using perchloric acid as an extracting agent. In spite of the rather gentle extraction proce- dure, that is perchloric acid hydrolysis performed on an ice-water bath protected from light, the degradation, interconversion and dephosphoryla- tion of pyridoxal-5’-phosphate could not be com- pletely avoided in extraction and the preceding sample preparation. This was especially true for the wheat matrix. Dephosphorylation of B 6 vita- mers has been reported to be associated with the unit operations used such as acid hydrolysis (Pe- terson et al. 1955) or exposure to salt and alco- hol solutions (Shimada et al. 1993), and the rap- id dephosphorylation step was followed by a slow transamination phase (Shimada et al. 1993). The expected reactivity of pyridoxal’s aldehyde group with amino compounds during applied extraction with ice-cold perchloric acid could not be observed. So the transamination of pyridoxal or pyridoxal-5’-phosphate to related amino forms reported by many reseachers (Metzler and Snell 1952, Gregory and Kirk 1977) was not 598 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods found when the extraction procedure with ice- cold perchloric acid was used. In practice no dif- ference in the distribution of 14C-activity in beef sample compared to that of carrot or wheat sam- ple was recorded in the reversed-phase chroma- togram. This was probably due to some sort of inactivation, and also a less suitable pH-value of acid extractant for that particular reaction(s). In a citrate-phosphate buffer solution model, the maximum degradation rate for pyridoxal was reported at pH 5, pyridoxamine loss increased in higher pH values while pyridoxine was con- sidered to be the most stable vitamer (Saidi and Warthesen 1983). A factor that possibly also di- minished the transamination reaction rate in the present study was the lower extraction tempera- ture compared to traditional sample treatments. Thermal destruction of B 6 vitamers at elevated temperatures has been reported by many research groups (Srncova and Davidek 1972, Evans et al. 1981, Navankasattusas and Lund 1982, Grego- ry and Hiner 1983). The catalytic nature of tran- sition metal and aluminium ions in transamina- tion (Snell 1954, Cennamo 1964) could also be reduced by the suggested ion-pairing character of perchlorate anion. Formation of ε-pyridoxyl- lysine via a non-enzymatic browning reaction as demonstrated by Gregory and Kirk (1978a) was not considered to occur to any great degree in our procedure as this reaction is generally re- lated to long-term storage of dehydrated materi- als. 6.2.2 Enzymatic hydrolysis Alkaline phosphatase treatment Pyridoxal and pyridoxamine were released from their phosphorylated forms by an alkaline phos- phatase enzyme prepared from calf intestine. Beef steak matrix which is rich in both pyridox- amine phosphate and pyridoxal phosphate was used as a test matrix since phosphate ester link- age in pyridoxamine phosphate has been report- ed to be more stable e.g. during acidic thermal extraction procedures than that of pyridoxal phosphate (Gregory and Mabbit 1961). The adequate amount of alkaline phosphatase was ca. 0.4–0.8U enzyme /mg sample in this sample extraction procedure (Fig. 20). Increas- ing the amount of enzyme did not yield a higher vitamin B 6 content. The enzyme preparation used was free of B 6 vitamers; no measurable amount of any vitamer was found in the enzyme’s blank analysis, thus higher enzyme concentrations can be used if needed. Removing enzyme protein by precipitation can be performed after hydrolysis with trichloroacetic acid, but this was not found to be necessary; no deterioration of column per- formance was observed when this step was omit- ted from the routine procedure. The limited availability of pyridoxamine-5’- phosphate reference standard may complicate method development work in the future. Many reagent suppliers have informed us that this ref- erence material is no longer being commercial- ly produced (personal communications), and it may not be obtainable in future. As the phos- phate ester linkage in PMP is considered to be more permanent than that of PLP during the de- phosphorylation process, the hydrolysis efficien- cy of pyridoxamine phosphate should somehow be confirmed. β-Glucosidase treatment Carrot matrix was chosen for testing glucosi- dase hydrolysis. Glucosically bounded pyridox- ine forms ca. 50–70% of the total B 6 vitamer content in carrot (Gregory and Ink 1987). Its fraction can be clearly separated from other peaks using e.g. the ion-paired reversed-phase chromatography. However, several peaks (nor- mally two to three) in our chromatogram dis- appeared after β-glucosidase incubation. The fraction (k e =11.6, t R = 9.8 min) eluting right be- fore pyridoxal (k e =12.2, t R =10.2 min) in the reversed-phase chromatogram was assumed to be 5’-O-β-D-glucopyranosyl pyridoxine (PNG). No measurable signal at k e of 11.6 in the chro- matogram was observed after hydrolysing the sample extract (0.13U of β-glucosidase per mg sample). The chemical nature of those two oth- er peaks remained unknown. The tentative peak identification based on the glucosidase treat- 599 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. ment only can be complicated by the fact that a minor amount of side-activities, such as phos- phatase activity, were unfortunately present in this enzyme preparation. Thus, the disappear- ance of some of the unknown compounds in the chromatogram after the enzymatic hydrolysis with glucosidase can partly be as a result of hy- drolysis of the possibly occuring phosphate es- ter linkages. As the measurement of glycosylat- ed pyridoxine was based on the difference in the pyridoxine amount before and after the β- glucosidase treatment, the presence of minor but unwanted phosphatase activity in enzyme preparate may lead to misinterpretations for samples which contain phosphorylated pyridox- ine as well. Pyridoxine phosphate, however, was only found in baker’s yeast and not at all in plant-derived samples. 6.2.3 Solid-phase extraction Partition chromatography based on chemically bonded reversed-phases, namely octadecyl, oc- tyl, phenyl, and cyanopropyl sorbents, was first evaluated. The results of those tests revealed that the polarity characteristics of B 6 vitamers dif- fers to such an extent that proper retention and elution conditions with one sorbent material suit- able for all the vitamers was quite unlikely to be found. The most polar vitamers, phosphate es- ters of pyridoxal and pyridoxamine, and pyri- doxic acid, did not retain enough in reversed- phase packings from aqueous sample extract matrices. Free vitamers being less polar analytes retained in the octadecyl and the octyl phases but not in a more polar cyanopropyl phase. Nor did phenyl sorbent which based on π-electron interactions work sufficiently well. When the analytical procedure was considered as a whole, it was thought to be useless to apply the same separation phenomenon, partition chromatogra- phy both in the purification step and in the ana- lytical separation. Cation-exchange chromatographic purifica- tion/concentration proved to be more effective than partition in sample extract purification as all B 6 vitamers show more or less cationic fea- tures due to their cationic pyridinium or ami- nomethyl chromophore. A weak cation-exchange sorbent containing the carboxymethyl group (-CH 2 COOH) retained free PL, PN, DPN, and PM in the standard solutions. The pK value for carboxymethyl group is ca. 4.8 being suitable for strong cations also as the negative charge of the sorbent can be neutralized by adjusting pH. However, this ionic interaction was greatly de- creased when the added standards were coelut- ed with the sample matrix. Because the weak cation-exchange sorbent failed to retain free vi- tamers in the sample extracts, two strong cati- on-exchange phases were then tested. Most com- mercial sorbents have either the propylsulfonate or propylbenzenesulfonate group chemically bonded to the parent silica. Due to the low pK value of the sulfonate ion, cationic analytes had be eluted using high ionic strengths and/or by neutralizing the positive charge of the analyte. Strong cation-exchange phases are thus general- ly suitable for weak cations, such as the pyrid- inium ion, only. Modifying the elution solvent system, pyridoxamine was succesfully retained and eluted using aromatic sulfonic acid as cati- on exhange material. The elution efficiency, that is the release of PM, was then improved by us- ing an elution solvent of high pH value combined with a high solvent strength because of pyridox- amine’s basic nature (pK ~10.5). However, this yielded to alkaline elution solvents which, in general, are incompatible with the traditional HPLC columns due to the lability of silica phase in alkaline media. This l imit- ing factor has been resolved by using poly(styrenedivylbenzene) or alumina based col- umn packings in liquid chromatography which are more stable in large pH areas like pH 1–13 (Mao and Fung 1997). These chemically resist- ant polymer phases enable the use alkaline mo- bile phases and/or injection solvents. The sepa- ration efficiency of polymer phase columns test- ed in our work was, however, insufficient for adequate separation of vitamin B 6 compounds and unknown substances in food samples. 600 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods The removal of sulfosalicylic acid prior to liquid chromatography was necessary when the extraction method of Gregory and Ink (1987) was applied as sulfosalicylic acid possesses a natu- ral fluorescence very similar to B 6 compounds. Anion-exchange solid-phase extraction with tri- methylaminopropyl as a functional group trapped sulfosalicylic acid effectively from the sample extract. The selecticity of the proposed clean-up procedure is complicated by the dissociation constant (pK <2.5) ) value of the phosphate res- idue in pyridoxal-5’-phosphate and in pyridox- amine-5’-phosphate. For the quantitative elution of phosphorylated vitamers from the anion-ex- change sorbent, the elution solvent volume of several cartridge void volumes was required. Thus, an unwanted dillution of the analyte solu- tion during this clean-up procedure was inevita- ble. In practical term, this would have caused an additional concentration step. In summary, one solid-phase extraction ma- terial suitable for quantitative analysis of both more polar (PLP, PA, and PMP) and less polar (PL, PN and PM) vitamin B 6 compounds was not found as B 6 vitamers represent a wide range of polarities due to both cationic and anionic func- tional groups being present. Secondly, the fluo- rescence characteristics of vitamin B 6 com- pounds are related to the 3-hydroxypyridinium structure which lacks the selectivity for vitamin B 6 compounds if the selective retention is based on the pyridinium ion. At the pH range of 1–2, 3-hydroxypyridine occurs as a cation and a di- polar ion, and their derivatives are fluorescent (Bridges et al. 1966). Any unknown compound present in the chromatogram derived from the fluorescence detector’s signal may have a simi- lar chemical hydroxypyridine structure. Thus, purification based on the ion-exhange properti- ties of pyridinium ion may not yield proper se- lectivity in solid-phase extraction followed by fluorescence detection. Evidence of this phenom- enon can be found in the results of Wong (1978). When cation-exchange liquid chromatography was applied to the separation of B 6 vitamers, a strong unknown peak preceding the pyridoxal peak interferred with the baseline separation and there was still a need for cleaning procedure for the sample extracts. For the above mentioned reasons, the ion- exchange solid-phase extraction in purification and in routine food analyses was omitted. 6.2.4 Validity of the routine food analysis method The retention factor values (k e ) for B 6 vitamers ranged from 1.4 to 21.3 which was considered to be adequate for quantitative work. As a gen- eral rule, retention factors ranging from 2 to 6 are recommended (Snyder and Kirkland 1979), the upper limit is, however, a guideline. Thus vitamin B 6 compounds except for PLP fulfill the minimum conditions in this work. In the routine food analysis method chosen, the amount of PLP was calculated on the basis of pyridoxal as the difference between the enzyme treated and non- treated sample extract. Using this method, the early elution of PLP in ion-paired reversed-phase chromatography was circumvented. Chromato- graphic peak performance (peak symmetry or peak tailing and relative response) as well as the repeatability of retention for the internal stand- ard, 4-deoxypyridoxine, were good (Table 9). The variations in the above mentioned values are derived from both sample matrix effect and the decreased column performance during routine analysis period. Thus the continuous documen- tation of these parameters was necessary. The accepted performance limits should ultimately match the quality requirements of an individual laboratory, and only general recommendations are given in international directions like ISO/IEC Guide 25 (1990). The sensitivity of the method was generally sufficient, and the main difficulties were related to the proper identification of the analyte’s sig- nal. This task occasionally required laborious standard spiking procedures for the sample ex- tracts. The recovery results of an added stand- ard were satisfactory for individual vitamers ranging from 72% to 107%. A similar total vita- 601 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. min B 6 recovery values was reported by Grego- ry (1980c) for their liquid chromatographic method. A recovery value of 150% for PLP was achieved in our work for fortified infant formu- la sample thus showing there were problems with this most labile vitamin B 6 compond in that par- ticular sample matrix. This phenomenon was not found in beef sample matrix. In addition, the early elution of PLP in our ion-paired reversed- phase chromatographic system made the relia- ble measurement of PLP more difficult. Phos- phorylated pyridoxal was then measured in the routine food analyses as the difference between the PL amount of phosphatase hydrolysed ex- tract and of nontreated sample extract. The greatest source of uncertainty inherent to the results was the standardization even though nine concentration levels were used in the stand- ardization procedure. It was estimated that the un- certainty related the to internal standard stand- ardization was two times that for the preparation of analysis sample. The total uncertainty for py- ridoxine was approximately 12%. Thus, the main efforts to reduce the uncertainty in the results should be further apparently focused on improve- ment of the standardization procedure. Intercalibration studies showed good agree- ment between 12 European laboratories for pig’s liver, mixed vegetable, and wholemeal flour samples (van den Berg et al. 1996). Our labora- tory’s results in that intercalibration study were in accordance with the mean values calculated from the results derived from the participating laboratories. Generally, the results derived from each laboratories’ own in-house methods re- vealed some inconsistencies between liquid chromatographic and microbiological assays. The coefficient of variation within laboratory (CV r ) was considered to be acceptable (ranging from 5% to 13%) whereas the variation between laboratories (CV R ) was higher than expected. Similar results for vegetable and wheat sam- ple extracts were achieved with two different HPLC methods and one microbiological assay when standard solutions and sample extracts were circulated between three laboratories. The results of pig’s liver sample derived from a microbiological assay were lower than those of liquid chromatography. The disparity between these results may partly be explanied by the dif- ferent growth responses of the microorganism Saccharomyces uvarum for individual B 6 vitam- ers. It is generally recommended that microbio- logical assays should be standardized with each individual vitamer if possible to reduce possible growth response errors. However, vitamer B 6 distribution of pig’s liver did not markedly dif- fer from that of vegetable sample. Both liver and vegetable matrices were rich in pyridoxamine (and/or phosphate), and their portion covers ap- proximately 70% of the total vitamin B 6 content while in flour sample more than two thirds are derived from pyridoxine and its glucosylated form(s). If the result disagreement is only caused by the lower growth response of pyridoxamine for the microorganism used, this same phenom- enon should have been found in vegetable sam- ple extracts as well which was not the case. Moreover, an intercalibration between two lab- oratories in Finland showed that the results of the microbiological assay were the same or slightly higher than results derived from a liq- uid chromatographic method. The factor(s) which caused the disagreement in the results between the liquid chromatographic method and the microbiological assay remained unresolved. 6.2.5 Laboratory proficiency The performance of the laboratory was tested by participating in the intercomparison and certifi- cation studies. These results showed that both the analytical procedure and laboratory work corresponded to the general status between Eu- ropean laboratories involved in these studies. The results calculated on the basis of both ex- ternal standard and internal standard standardi- zation were in agreement showing that the cho- sen internal standard, 4-deoxypyridoxine, and its use fulfill the requirements for quantitative work. Our success in the interlaboratory studies and in a certification study (Ollilainen et al., manu- script) we participated in was considered to show 602 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods our laboratory’s proficiency to carry out vitamin B 6 analyses. The sets of interlaboratory studies clearly showed that further harmonizing work in the field of vitamin B 6 analysis is still needed. This task should include evaluation of the chromatograph- ic separation and its parameters, the sample pret- eatments (enzymatic and/or acid hydrolysis), and the standardization procedures for chromato- graphic and microbiological assays as well. 6.3 Food analysis In order to produce up-to date food composition data, a considerable amount of food items need to be analysed, and it can be considered that the value of the data obtained is due to the fact that the same validated analytical procedure is ap- plied to various samples. Sample pooling pro- cedure, as performed in this study, limits the obtainable information as the variation between subsamples is lost (Stewart 1995). In practice, in order to handle this number of food items us- ing the described, quite laborious analytical methodology, this compromise had to be made. If information of the variation between subsam- ples is needed, it will inevitable multiply the number of samples resulting in a need for in- creased economical resources. Further studies related to food composition work in future should be focused on producing reliable data on the uncertainty of sampling procedures as well. This data would be of considerable help in the accreditation of laboratories. 6.3.1 Flesh foods; meat, poultry, and fish Phosphorylated vitamers, pyridoxal-5’-phos- phate and pyridoxamine-5’-phosphate formed the major fraction of vitamin B 6 content in meat samples (Table 14) and pyridoxine, more char- acteristic for plant-derived foods, was present only in small amounts. In this respect, beef kid- ney and shoulder samples were exceptional; free pyridoxine covered one fifth of the total vita- min content in both samples. No detectable free pyridoxine in pork meat or pork meat products has been reported by Esteve et al. (1998); how- ever, small amounts of free PN could be detect- ed in all our meat samples. Free vitamers in our work, however, were generally present in offals. Livers were rich in vitamin B 6 containing ca. 0.85, 0.81, and 0.66mg/100g in beef liver, broil- er liver, and pork liver, respectively. Poultry meat contained almost solely pyridoxal phosphate; PLP covered more than 70% of the total vitamin content. Vitamin B 6 levels in kidney samples were much lower. 4-pyridoxic acid, the metabol- ic end-product of pyridoxine compounds was the main vitamin B 6 compound in livers. It was note- worthy that no PNP could be detected in the liv- er samples of our study. In some rat liver sam- ples but not in all samples, minor amounts of pyridoxine phosphate were found according to Vanderslice and coworkers (1981a). The major- ity of the vitamin B 6 activity in fish was com- posed of phosphorylated pyridoxamine and py- ridoxal as the total sum of PLP and PMP ranged between 80–90% of the total vitamin B 6 amount. Free vitamers were only minor constituents ex- cept in Baltic herring. Processing seems somewhat to alter the vi- tamer distribution in meat samples; a lower per- centage of PLP was present in processed foods (sausages and meatball food items) while this was not the case with PMP. This might reflect the lability of phosphate ester linkages of pyri- doxal phosphate present in beef and pork. Pyridoxal and pyridoxamine are reported to be the predominant vitamers in flesh foods (Po- lansky and Toepher 1969). Due to heating proc- esses like cooking or canning, at least 70% of the total vitamin content was in the form of py- ridoxamine whereas the original, non-processed material contained more pyridoxal and less py- ridoxamine. These different levels of pyridox- amine in raw and processed products were caused by the transamination of pyridoxal. The formation of pyridoxamine as a result of heat- ing pyridoxal together with glutamic acid 603 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. reached a maximum value at pH 6 at 100°C (Sigg 1985). Conversion of pyridoxamine in part to pyridoxine was observed when the temperature was further raised at 140°C (Sigg 1985). Tem- perature dependence of transamination toward the formation of pyridoxamine has also been re- ported in steamed sea urchin gonads (Shimada et al. 1993), in cooked meats (Bowers and Craig 1978) and in cereal based food model systems (Gregory and Kirk 1978b). It is quite obvious that the extent of transam- ination varies markedly depending on unit oper- ations to which the food material has been ex- posed. Furthermore, the interconversion of B 6 vitamers inevitably also occurs during the labo- ratory operations including sample storage and pretreatment, extraction, digestion and purifica- tion steps prior to analytical measurement. This complicates the evaluation of the vitamer distri- bution data of individual studies, and the earlier published data is not always comparable to present results. In the analytical sense, meat samples were considered as “easy” samples in terms of sam- ple pretreatment and enzymatic hydrolysis step even thought a more vigorous hydrolysis is need- ed due to more stable phosphorylated pyridox- amine (Peterson et al. 1955, Toepher and Polan- sky 1970). The strong hydrolysis of muscle and related tissues, especially if acid treatment is combined with a heating procedure, may yield breakdown products of pyrimidine-like and sim- ilar chemical structures. As the fluorescence characteristics of B 6 vitamers is based on the structure of hydroxypyridine moiety (Peterson et al. 1955), the formation of unknown or inter- ferring components which are seen in the chro- matogram will be hardly avoided. This was clear- ly shown in the analytical procedure applied by COST91 (1985) when sulfuric acid extraction followed by an autoclaving step was performed. An unknown, and in many cases also interfer- ence, fraction was eluted near to the pyridoxal in the reversed-phase chromatogram. Even if B 6 vitamers can be separated from these com- pounds, difficulties in the accurate baseline measuring of vitamin B 6 analytes should be ex- pected. In addition, as many of these interfer- ring compounds shows the same polarity nature in partition chromatography and the same ionic characters in ion-exchange chromatography, their specific removal during the sample treat- ment prior to analytical liquid chromatography will be a challenging task in vitamin B 6 analy- sis. Thus, a selective sample clean-up step with solid-phase extraction or related techniques would be of great help in the analytical proce- dure. 6.3.2 Dairy products and egg A different distribution of B 6 vitamers was found in liquid milks and egg yolk compared to other food groups. The predominant vitamer in light- processed items such as liquid milks was free and phosphorylated pyridoxal covering togeth- er approximately 50 to 70% of the total vitamin B 6 activity. The highest portion of PLP, over 90%, was found in egg yolk, a result which is in line with Argoudelis (1997) and Toukairin-Oda et al. (1989). Changes in the vitamer distribution between fresh and condensed or dried milk presented by Gregory and Mabbit (1961) were not seen in our skim milk powder sample (Table 16) as its vita- mer distribution resembled that of milk or cream. In their early studies it was suggested that part of phosphorylated pyridoxamine found in freeze- dried milk was derived from pyridoxal during the processing steps. However, the presence of pyridoxal phosphate in fresh milk could not be measured due to the limitations in their analyti- cal procedure. Thus, a possible substrate for transamination, free or phosphorylated pyri- doxal, remained unclear in their results. The total amount of vitamin B 6 in cheese sam- ples, as expected, correlated in reverse order to the fat content; a lower vitamin amount was found in cheese with a higher fat content. The vitamer distribution in cheese was also remark- ably different than that in pasteurized milk; both free and phosphorylated pyridoxal almost com- pletely disappeared while the relative amount of 604 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods pyridoxamine and its phosphate in cheese were 2–3 fold compared to that of milk. Whether this was caused by processing or microbiological growth in the cheese maturation process (Gre- gory and Mabbit 1961) was beyond the scope of this study. Pyridoxine phosphate, most likely pyridox- ine-5’-phosphate, was found in baker’s yeast and it was the only food sample in our study in which the pyridoxine content increased after phos- phatase digestion. This increase in free pyridox- ine was considered to indicate the occurrence of phosphorylated pyridoxine (PNP) in the sample. However, the effect of other unknown enzyme side activities or vitamer interconversion can- not be totally excluded. Results reported by Tou- kairin-Oda and coworkers (1989) showed that approx. 13% of the total vitamin B 6 content of their baker’s yeast sample was derived from phospho- rylated pyridoxine. Taking this into consideration, the PNP results of our study for yeast showed low- er proportion of PNP than that of the Japanese group. The portion of PNP in other food groups was lower according to above mentioned study. 6.3.3 Plant-derived foods The primary vitamin B 6 fraction of plant derived foods in our study was glucosylated pyridoxine. In this aspect foods of plant origin differ remark- ably from other food groups. Only some plant foods like almonds and hazel nuts lack bound derivative(s) of pyridoxine; it has been proposed that this was due to the natural β-glucosidase activity present in these materials (Chiari et al. 1997, He and Withers 1997, Lai et al. 1992). Total vitamin B 6 activity in cereals is derived from pyridoxine or its bound form (pyridoxine glycoside). Free and phosphorylated pyridoxam- ine and pyridoxal and free pyridoxic acid form only a minor part of vitamin B 6 compounds. All six vitamers were present in barley flour, mixed wheat and rye flour samples, and also polished rice contained all six vitamers. This is partially contradicts to the results of Sampson et al. (1995) as they did not detect any free pyridoxamine in three wheat cultivars studied. Our results for flour samples were somewhat higher than those reported by Michela and Lorenz (1976); the to- tal vitamin B 6 content for wheat and rye flour were 0.04mg/100g FM and 0.04mg/100g where- as our results for wheat and rye were 0.13mg/ 100g FM and 0.26mg/100g FM. No vitamer dis- tribution was given in their results. Differencies in the milling process make it difficult to com- pare these results as the ash content was not giv- en. The effect of milling on the vitamin B 6 con- tent in wheat, triticale and rye was clearly seen in their study; bran was rich in total vitamin B 6 (1.1mg/100g FM) and the milling reduced their vitamin content to one tenth part that of grain. Glycosidic derivative(s) of pyridoxine was de- termined in all cereal samples; the average por- tion of total PNG was ca. 40% of the total vita- min B 6 content. These results are in agreement with earlier studies (Kabir et al. 1983a, Gregory and Ink 1987, Sampson et al. 1995). In baby- foods based on vegetable ingredients this bound vitamer fraction constituted about two thirds of the total vitamin content. Most commonly used analytical procedures include either sample extraction with mineral acids or enzymatic digestion of food matrix pri- or to quantitative measurement of B 6 vitamers. Many enzyme preparations, like takadiastase, contain normally enough glucosidase side-activ- ity to hydrolyze the glucosidic linkage present in bound vitamin forms. In an analytical sense this simplifies the determination of the total vi- tamin B 6 content and it is often a desired char- acteristics in an enzyme preparation. It has also been considered as a marker of the usefulness of an enzyme preparation. Hydrolysis of bound vi- tamers and phosphorylated forms enables the direct measurement of the total vitamin B 6 con- tent without the need for determination of each vitamin form present. However, after these sam- ple pretreatment steps the information of origi- nal vitamer distribution in food is lost. Plant foods are considered to be the major dietary contributor of vitamin B 6 ; the portion of the intake derived from cereals and grain, from fruits and vegetables, and from legumes and nuts 605 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. were 21%, 43% and 8%, respectively. Their add- ed up vitamin B 6 content represented almost 70% of daily intake for elderly persons (Manore et al. 1990). The average PNG content of plant- derived foods was estimated to be 40% in our study. In some food items glucosylated pyridox- ine may form as much as 60–70 percent of the total vitamin B 6 activity. Thus, the role of gly- cosidically bond pyridoxine in plant-derived foods cannot be overlooked. More detailed data on the vitamer distribution and occurrence in biological samples like food, and the role of gly- cosidic forms of pyridoxine in human nutrition is clearly needed. This requires new physiologi- cal and analytical/ methodological studies. The data for availability in different animal species and in humans is inadequate but also the effect of processing (Kabir et al. 1983b) and food ma- trix on utilization should be evaluated. It seems that the availability of B 6 vitamers from differ- ent food matrices varies (Gregory 1980c). 6.3.4 Comparison to national food composition tables The total vitamin B 6 content of dairy products in the present work were in accordance to those values in national food composition tables when the amount of bound pyridoxine was summed together with other B 6 vitamers (Table 21). Some differences were still found, for example in the values for baker’s yeast; a relatively high value was reported in the Danish food composi- tion table compared to other food composition tables. 6.4 Characterization of isolated pyridoxine derivative The amount of isolated PNX fraction (k e ~11.8), quantitated as pyridoxine, was approximately 1.3µg/100g and 2.03µg/100g in carrot and wholemeal flour, respectively. This calculation is based on the assumption that the molar fluo- rescence response of glycosylated pyridoxine is equal to that of pyridoxine (Gregory and Ink 1987). The recovery after ion-exchange purifi- cation was estimated to be ca. 65%. The content of glycosylated pyridoxine varied from 100 to 900 µg/100g in carrot and wheat flour accord- ing to Gregory and Ink (1987, Schramm and Bitsch (1991), Bitsch and Schramm (1992), Schramm and Bitsch (1993), and Sampson et al. (1995). Considering those literature values, a high percentage of analyte was lost during the isolation and fractionating process. 6.4.1 β-glucosidase hydrolysis The reversed-phase chromatogram of plant-ori- gin sample extracts contained several (normally three) peaks, (k e ~8.3, ~10.0, ~11.8) which dis- appeared during the β-glucosidase digestion. When the proposed main bound vitamin fraction (PNX at k e ~11.6–11.8) was enzymatically hy- drolyzed with β-glucosidase, it yielded an in- creased pyridoxine peak in the chromatogram. The retention of two other fractions in cation- exchange resin was also different than that of PNX, thus showing unequal cationic features for these two unknown analytes compared to that of PNX. No data on their structures or their chem- ical natures have been presented in the litera- ture. The most retained unknown fraction (PNX) was considered to include the main glycosylat- ed pyridoxine fraction, and this analyte fraction was taken into NMR spectroscopy and FAB mass spectrometry for further characterization. 6.4.2 Proton NMR spectroscopy The 1H spectrum of the PNX fraction from wholemeal flour (k e ~11.8) was measured as the isolated analyte was considered to be pure enough for NMR measurement (Fig. 28) where- as the carrot fraction still contained impurity/ 606 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods unknown compound(s). The anomeric proton of glucose unit was tentatively characterized ac- cording to the signal at the δ-value of 4.45 (J a,d ~ 8Hz)(Table 24). The signal of an anomeric pro- ton is expected to be located in a low field due to the attachment of two oxygens atoms to C-1, and should therefore be distinguishable from the other signals (Hall 1964). The measured chemi- cal shift as well as coupling constant values for anomeric proton in the sugar moitety correspond- ed to those found in literature for β-D-gluco- pyranose (Lemieux et al. 1958, Rudrum and Shaw 1965). Pure D-glucose, dissolved in D 2 O, at equilibrium showed two douplet signals at δ-values 4.57 (J a,b ~ 8.1) and 5.16 (J a,b ~ 3.9) apart from the other carbohydrate moiety signals (Table 25). The intensities of these douplets (1:2) corresponded to tautomeric equilibria of glucose in water; 36% of α-D-glucose and 64% β-D-glu- cose (Rudrum and Shaw 1965, Lemieux and Stewens 1966). These findings were also in accordance with the results for the axial H-1 proton in β-D-glu- copyranose (δ~4.3–4.4 ppm, J~8Hz). Conforma- tion of C1 structure has been confirmed by free energy calculation of Angyal (1968). NMR measurements of an anomeric proton in D-glu- copyranose (Lemieux and Stevens 1966) or D- glucopyranose pentaacetates (Lemieux and Ste- vens 1965), revealed that equatorial protons give their signal to a lower field than chemically sim- ilar but axially orientated protons. The anomer- ic proton in β-glucopyranosyl moiety of a phe- nyl propanoid glycoside in lemon was found at δ4.47 (J~8) (Matsubara et al. 1991). The chem- ical shift of the douplet signal for anomeric pro- ton in 2-phenyl ethyl-β-glucoside located at δ- value of 4.82 ppm ( J~7Hz) was reported by Umehara and coworkers (1988). The correspond- ing signal of the equatorially orientated proton in α-glucoside was found at a δ-value of 4.80 ppm ( J~3.8Hz)(Shu and Lawrence 1994). The anomeric proton of the glucose moiety in glycosylated pyridoxine located in the high field area (δ~4.44 ppm, J~6Hz) was reported by Gregory and Ink (1987); these values were con- sidered to show the β-glycosidic linkage in glu- cosylated pyridoxine. A douplet at δ 4.22 (J~7), measured in CD 3 OD, was ascribed to the β-ano- meric proton in bound glucose of 5’-O-(β-D-glu- copyranosyl) pyridoxine isolated from rice bran (Yasumoto et al. 1977). A singlet (at δ~7.78) was suggested as a signal derived from the aromatic proton of pyridoxine. Tadera et al. (1988) report- ed NMR data for 5’-O-(β-cellobiosyl) pyridox- ine; two singlets at δ 4.25 and 4.27 indicated β- glycosidic linkage in the analyte. The anomeric proton of 5’-O-[6–O-(3-hydroxy-3-methyl-4- carboxy-butanoyl)-β-D-glucopyranosyl] pyri- doxine was characterized according to signal at δ 4.21( J=7.0) by Tadera and associates (1983). An anomeric proton resonance of pyridoxine 4’- and 5’-α-D-glucosides measured in deuterated dimethyl sulfoxide has been found at δ 4.89 (1H, d, J=3.5) and δ 4.68(1H, d, J=3.0), respectively (Suzuki et al. 1997). A low field signal (δ8.3–8.4, J ~ 42.5) as a douplet or two individual singlets in our results was/were assumed to be derived from the aro- matic C-6 proton in pyridoxine. The value for the coupling constant was extraordinarily wide. On the other hand, two singlet interpretation did not correlate with the proposed pyridoxine moitety in 5’-O- β-D-glucopyranosylpyridoxine. No low field area of the NMR spectrum was pre- sented by Gregory and Ink (1987). Evaluation of the aromatic proton region in the NMR spec- tra of synthetized and isolated pyridoxine-5’-α- D-glucopyranoside were also missing in the re- sults of Ogata et al. (1969a, b). The low field area of spectrum in our study representing the area for aromatic protons was also different than that of 5’-O-(β-cellobiosyl) pyridoxine in rice bran reported by Tadera et al. (1988) as only one singlet (δ~7.92) was present in their spectrum. Suzuki and coworkers (1997) reported that aro- matic protons in 4’- and 5’-O- (α-D-glucopyran- osyl) pyridoxines gave resonance signals at δ 7.91 (1H, s, PN, 6-H) and at δ 7.86 (1H, s, PN, 6-H). Thus, aromatic proton in pyridoxine was seen in both above mentioned studies as a sin- glet which was in accordance with expectations. Pyridoxine hydrochloride gives a four sig- nal NMR spectrum in deuterium oxide: δ~2.68, 607 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. δ~4.84, δ~5.04 and δ~8.21 for methyl protons (2’C), methylene protons in 4’CH 2 -, methylene protons in 5’CH 2 -, and aromatic proton in 6C, respectively (Handbook of Proton-NMR Spec- tra and Data 1985). Our results for the pyridox- ine moiety of the analyte is generally in accord- ance with that data, except for the signal of meth- ylene protons in the 4’-carbon as it was over- lapped by the water signal. Overall, the NMR signal for hydroxyl protons could not be record- ed since the measurements were performed in deuterium oxide. A high field signal in our work at δ~2.8 was characterized as an impurity. Spin-spin coupling and total spin correlation of the molecule were measured using COSY- and TOCSY-spectra (Fig. 29), respectively. Signals at δ-values of 3.85, 3.95, 4.35, 4.45, 4.85 and 6.15 in whole wheat flour spectrum were as- sumed to be sign of a spin-spin coupling of the adjacent atoms. Signal intensities of the former coupling system also shoved that these signals were assigned to one coupling system. This re- sult agreed with the proposed structure of glyc- osylated pyridoxine. Signals at 1.35 and at 4.15 belong to another coupling system, representing an impurity present in the NMR sample. The in- tensity of the proposed system (δ 3.85–3.95– 4.35–4.45–4.85– 6.15) did not correlate with the calculated amount of the isolated fraction when the concentration of the glycosylated fraction was measured using HPLC data. Overall NMR- signal level of that coupling system was estimat- ed to be higher than the result derived from the chromatographic data. Event though the resolution in the proton NMR spectrum was good, the exact chemical structure of the isolated pyridoxine derivate could not be elucidated using this proton NMR data. Most of the signals can be explained by the proposed molecule structure (5’-O- β-D-glu- copyranosyl pyridoxine) but too many uncertain- ties were still present in the interpretation of the proton spectra. Furthermore, the absolute amount of the isolate did not match the intensity of the proton NMR signal; the amount of isolated com- pound was considered to be higher than expect- ed when the intensity of the NMR signal was taken into account. Thus, it should be critically evaluated whether this discrepancy was due to different molecule stucture of the isolate than expected and/or to the presence of impurity (or impurities) in the isolated PNG fraction. 6.4.3 FAB mass spectrometry FAB induced mass spectra of the PNX fraction (k e ~11.8) isolated from wholemeal flour and car- rot samples were measured using either glycer- ol, nitrobenzylamine or triethanolamine as a matrix (Appendix 2). The highest m/z-value (621.4, 60%) as a molecule ion using a nitroben- zylamine matrix (in the positive mode) was con- sidered to be too high to show the proposed gly- cosylated pyridoxine structure when m/z 171 formed the base peak (100%). In addition, if molecular weights are normally given by abun- dant [M+H]+ ions in positive-ion FAB spectra, the above mentioned m/z-value match with dif- ficulty. Odd-numbered m/z-value for molecule ion, if a truly nonprotonated analyte molecule ion is present in the mass spectra, requires nor- mally an odd number of nitrogen atoms in the proposed structure, which is the case in proposed structure of glycosidically bound pyridoxine. Two triglucosides of pyridoxine identified from rice bran, 4’-O-(β-D-glucosyl)-5’-O-(β- cellobiosyl)pyridoxine and 5’-O-(β- glucotriosyl)pyridoxine, gave both [M+H]+ ions at an m/z-value of 656 in the secondary ion mass spectrometry (SI-MS). The [M+H]+ ion for a diglucoside of PN, 4’-O-(β-cellobiosyl) pyridox- ine, was detected at an m/z value of 494 (Tadera et al. 1988). Unexpectedly high m/z-values were also found using a glycerol matrix as the observed m/z-values neither correlated with the suggest- ed protonated oligomers of glycerol adducts ([glycerol]nH+, m/z (92n+1)) nor with other pro- posed adducts ([M+NH4]+, [M+Na]+). The mass fraction m/z 171 present in positive-mode CI with glycerol and nitrobenzylalcohol matrices in our work was found in the EI mass spectra of 5’-O- β-D-glucopyranosyl pyridoxine by Bitsch 608 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods and Schramm (1992). No mass spectrometric interpretation, however, was given in their re- port. Careri and her coworkers (1996) have report- ed electron impact (EI) and chemical ionization (CI) mass spetrometric data for water-soluble vitamin standards produced by a particle beam (PB) mass spectrometry. For the electron impact spectra of free pyridoxine, the intensity of the molecule ion was relatively high 169 (50%), oth- er major fragments being m/z 94 (M-75, 100%), 106 (M-63, 77%) and 151 (M-18, 74%). Un- bound pyridoxine using a chemical ionization in negative-ion mode (NCI) formed the major ions as m/z=167 (100%), 149 ( 74%), 168 (55%), and 151 (35%), and in positive-ion mode (PCI) as m/z= 152 (100%), 170 (76%), 136 (19%). PCI proved to be the most suitable ionization mode for pyridoxine and pyridoxamine. None of above mentioned molecule ions or fragmentation ions could be clearly recognized in our mass spectra. FAB-MS gave some evidence that both iso- lates (from carrot and wheat flour) contained the same analyte but the exact interpretation of the mass spectra was interfered from the impurities, probably traces derived from the mobile phase and ion-pair reagents. This being the case, the structure for the proposed 5’-O-(β-D-glucop- yranosyl) pyridoxine could not be confirmed on the basis of our mass spectra results. Whether this discrepancy is mainly caused by impure an- alyte or a less suitable ionization technique in mass spectrometric measurements remains un- clear. In order to continue this work, further pu- rification of the analyte as well as the use of other MS techniques, like MALDI or APCI-MSn, is probably needed. 7 Conclusions In summary, the liquid chromatographic meas- urement of vitamin B 6 compounds in a complex matrix, like foods and other biological materi- als, has proved to be a challenge for an analyti- cal laboratory. Depending on the data needed, different analytical approaches can be chosen. Sample treatment, including extraction and de- tection procedures, chromatographic separation and the quantitation of the analytes will have an effect on the data and its accuracy which are obtained using a particular method. The present data shows that vitamin B 6 com- pounds can be determined in their intact forms, and the native distribution of vitamers is main- tained during the analytical procedure. Cold per- chloric acid extraction followed by enzymatic digestions prior to liquid chromatographic anal- ysis produces data on free, phosphorylated and glycosylated vitamin B 6 compounds. In this ap- proach plant-derived samples were treated with two enzymatic digestions in addition to a non- treated sample which need, however, resources in a laboratory. At present, the vitamin B 6 method chosen in a laboratory depends on how detailed data is re- quired and for what purpose. The European Com- mittee for Standardization (CEN) has begun to harmonize the methodology for vitamin B 6 . In that work a method will be found which will serve as an European standard for quantitative measurement of vitamin B 6 in foods. Further work will be then focused on the sample prepa- ration technique and on the method proficiency testing data available. In addition, further progress in the improvements in identification in the present techniques would be of help. The general selectivity limitations in the present liq- uid chromatographic methods may be solved to some extent by the development of LC-MS or CE-MS methodology. Reassessment of the bio- logical methods, like microbiological assays, is also needed since these methods are still in gen- 609 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. eral utilization, and their use is expected to con- tinue in the near future in spite of the liquid chro- matography’s suitability. Secondly, the latest Finnish Food Composi- tion Table completely lacks data for vitamin B 6 , and the data users are involved in consulting the foreign food data bases. In general, the analyti- cal results of the amount of glycosylated pyri- doxine in foods are scarce too. The strength of the present data is that it is produced by one method including the interlaboratory data, and the results are thus comparable with each other within the food samples or food groups analyzed. It is therefore to be hoped that the results of the present study will meet those needs. Thirdly, in the present work it was clearly seen that glycosidically bound pyridoxine derivative(s) accounted for the main portion of vitamin B 6 in nearly all foods of plant origin, and this bound analyte(s) is not taken into ac- count in the traditional methods. The role of the glycosylated pyridoxine(s), needs to be clarified both in analytical and physiological aspects. The utilization of the glycosidically bound forms has remained somewhat unclear despite of the inten- sive study performed in this field. If the availa- bility of the bound forms is limited for human as it is more or less assumed at present, it means that the role of vegetables, cereals and other plant-derived foods as a source of vitamin B 6 should be reassessed. References Addo, C. & Augustin, J. 1988. Changes in the vitamin B-6 content in potatoes during storage. Journal of Food Science 53: 749–752. Agostini, T.S. & Godoy, H.T. 1997. Simultaneous deter- mination of nicotinamide, nicotinic acid, riboflavin, thiamin, and pyridoxine in enriched Brazilian foods by HPLC. Journal of High Resolution Chromatogra- phy 20: 245–248. Alfred, G. & Narasinga Roo, M.S. 1971. A study of the inactivation of pepsin by trichloroacetic acid. Indian Journal of Biochemistry & Biophysics 8: 28–31. Amin, M. & Reusch, J. 1987. High-performance liquid chromatography of water-soluble vitamins. Part 3. Si- multaneous determination of vitamins B1, B2, B6, B12 and C, nicotinamide and folic acid in capsule preparations by ion-pair reversed-phase high-per- formance liquid chromatography. Analyst 112: 989– 991. Andon, M.B., Reynolds, R.D., Moser-Veillon, P.B. & Howard, M.P. 1989. Dietary intake of total and glyc- osylated vitamin B6 and the vitamin B6 nutritional status of unsupplemented lactating woman and their infants. American Journal of Clinical Nutrition 50: 1050–1058. Ang, C.Y.W. 1980. Vitamin B 6 and niacin contents of broil- er meat of different strains, sexes, and production regions. Journal of Food Science 45: 898–900. – , Centiarelli, M. & Eitenmiller, R.R. 1988. A simple liq- uid chromatographic method for determination of B6 vitamers in raw and cooked chicken. Journal of Food Science 53: 371–375. Angyal, S.J. 1968. Conformational analysis in carbohy- drate chemistry. I. Conformational free energies. The conformations and α:β ratios of aldopyranoses in aqueous solutions. Australian Journal of Chemistry 21: 2737–2746. AOAC 1995a. AOAC Official method 961.15 vitamin B6 (pyridoxine, pyridoxal, pyridoxamine) in food extracts. Microbiological method. In: Cunniff, P. (ed.). Official methods of analysis of AOAC International. Vol II. 16th ed. AOAC International, Arlington, USA. ISBN 0-935584-54-4. – 1995b. AOAC Official method 985.32 vitamin B6 (py- ridoxine, pyridoxal, pyridoxamine) in ready-to-feed milk-based infant formula. Microbiological method. In: Cunniff, P. (ed.). Official methods of analysis of AOAC International. Vol II. 16th ed. AOAC Interna- tional, Arlington, USA. ISBN 0-935584-54-4. Arai, Y. & Hanai, T. 1988. Capability of a photodiode ar- ray detector for vitamin analysis. Journal of Liquid Chromatography 11: 2409–2418. Argoudelis, C.J. 1988. Simple isocratic high-performance liquid chromatographic method for the separation of the six vitamers of vitamin B6. Journal of Chroma- tography Biomedical Applications 424: 315–323. – 1997. Simple high-performance liquid chromato- graphic method for the separation of all seven vita- min B 6 -related compounds. Journal of Chromatogra- phy A 790: 83–91. Atkin, L., Schulzt, A.S., Williams, W.L. & Frey, C.N. 1943. Yeast microbiological methods for determination of vitamins. Pyridoxine. Industrial & Engineering Chem- istry Analyses Ed. 15: 141–144. Ayi, B.K., Yuhas, D.A. & Deangelis, N.J. 1986. Simulta- 610 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods neous determination of vitamins B 2 (riboflavin) and B6 (pyridoxine) in infant formula products by reversed phase liquid chromatography. Journal of the Associ- ation of Official Analytical Chemists 69: 56–59. Bailey, A.L., Maisey, S., Southon, S., Wright, A.J., Fin- glas, P.M. & Fulcher, R.A. 1997. Relationships be- tween micronutrient intake and biochemical indica- tors of nutrient adequacy in a “free-living’ elderly UK population. British Journal of Nutrition 77: 225–242. Banks, M.A. & Gregory, J.F. 1994. Mice, hamsters and guinea pigs differ in efficiency of pyridoxine-5’-β-D- glucose utilization. Journal of Nutrition 124: 406–414. Belal, F. 1989. Ion-pairing high-performance liquid chro- matographic determination of pyridoxine and pyrithi- oxine in dosage form. Analytical Letters 22: 1897– 1907. Bell, J.G. 1974. Microbiological assay of vitamins of the B group in foodstuffs. Laboratory Practice 23: 235– 242. Bereznitski, Y., Jaroniec, M., Kruk, M. & Buszewski, B. 1996. Adsorption characterization of octyl bonded phases for high performance liquid chromatography. Journal of Liquid Chromatography & Related Tech- nologies 19: 2767–2784. Berg, H. van den, Schaik, F. van, Finglas, P.M. & de Froid- mont, I. 1996. Third EU MAT intercomparison on method for the determination of vitamins B-1, B-2 and B-6 in Food. Food Chemistry 57: 101–108. – , Schreurs, W.H.P. & Joosten, G.P.A. 1978. Evalua- tion of the vitamin status in pregnancy. International Journal for Vitamin and Nutrition Research 48: 12– 21. Bergaentzle, M., Arella, F., Bourguignon, J.B. & Hassel- mann, C. 1995. Determination of vitamin B6 in foods by HPLC – A collaborative study. Food Chemistry 52: 81–86. Bernhart, F.W., D’Amato, E. & Tomarelli, R.M. 1960. The vitamin B6 activity of heat-sterized milk. Archives of Biochemistry and Biophysics 88: 267–269. Bishop, R.E. & Tryfiates, G.P. 1989. Biosynthesis of a novel form of vitamin B6 by tumor cells. Journal of Nutrition 119: 189–195. Bitsch, R. & Möller, J. 1989a. Analysis of B6 vitamers in foods using a modified high-performance liquid chro- matographic method. Journal of Chromatography 463: 207–211. – & Möller, J. 1989b. Evaluation of the vitamin B6 con- tent in foods by HPLC analysis. In: Nutrient availa- bility: chemical and biological aspects. Special Pub- lication 72. Royal Society of Chemistry. p. 77–79. – & Schramm, W. 1992. Free and bound vitamin B6 - derivatives in plant foods. Chemical Reactions in foods II. Working Party Proceedings. Prague, Sep. 23–25, 1992. Blanco, D., Sanchez, L.A. & Gutierrez, M.D. 1994. De- termination of water soluble vitamins by liquid chro- matography with ordinary and narrow-bore columns. Journal of Liquid Chromatography 17: 1525–1539. Bognar, A. 1985. Bestimmung von vitamin B6 in Lebens- mitteln mit Hilfe der Hochdruckflüssig-Chromatogra- phie (HPLC). Zeitschrift für Lebensmittel-Untersuc- hung und Forschung 181: 200–205. – & Ollilainen, V. 1997. Influence of extraction on the determination of vitamin B6 in food by HPLC. Zeitschrift für Lebensmittel-Untersuchung und For- schung 204: 327–335. Bötticher, B. & Bötticher, D. 1987. A new HPLC-method for the simultaneous determination of B 1 -, B 2 - and B 6 -vitamers in serum and whole blood. International Journal for Vitamin and Nutrition Research 57: 273– 278. Bowers, J.A. & Craig, J. 1978. Components of vitamin B6 in turkey breasts muscle. Journal of Food Sci- ence 43: 1619–1921. Bridges, J.W., Davies, D.S. & Williams, R.T. 1966. Fluo- rescence studies on some hydroxypyridines includ- ing compounds of the vitamin B6 group. Biochemi- cal Journal 98: 451–468. Bühnert, H. 1988. Bestimmung von pyridoxin, Thiamin und Riboflavin in einem Nährstoffkonzentrat (Pale- um). Lebensmittel und Biotechnologie 2: 84–85. Buszewski, B. 1991. Properties and biomedical applica- tions of packings with high-density coverage of C18 chemically bonded phase for high-performance liq- uid chromatography and solid-phase extraction. Jour- nal of Chromatography 538: 293–301. – , Gadzata-Kopciuch, R.M., Markuszewski, M. & Kalis- zan, R. 1997. Chemically bonded silica stationary phases: synthesis, physiochemical characterization, and molecular mechanism of reversed-phase HPLC retention. Analytical Chemistry 69: 3277–3284. – & Galushko, S.V. 1995. Relation between structure and physicochemical properties of chemically bond- ed phase, and retention in RP HPLC. Acta Chroma- tographica 4: 33–49. Callmer, K. & Davies, L. 1974. Separation and determi- nation of vitamin B1, B2, B6 and nicotinamide in com- mercial vitamin preparations using high performance cation-exchange chromatography. Chromatographia 7: 644–650. Careri, M., Cilloni, R., Lugani, M.T. & Manini, P. 1996. Analysis of water-soluble vitamins by high-perform- ance liquid chromatography-particle beam-mass spectrometry. Analytical Communications 33: 159– 162. Caulcutt, R. & Boddy, R. 1983. Statistics for analytical chemists. Chapman and Hall. London. ISBN 0-412- 23730-X. 253 p. Cennamo, C. 1964. Metal ion independent non-enzymatic transamination between pyridoxal and amino acid esters. Biochimica Biophysica Acta 93: 323–332. Chase, G.W., Jr., Landen, W.O., Jr., Eitenmiller, R.R. & Soliman, A.-G.M. 1992. Liquid chromatographic de- termination of thiamine, riboflavin, and pyridoxine in infant formula. Journal of AOAC International 75: 561–565. – & Soliman, A.-G.M. 1990. Analysis of thiamin, ribo- flavin, pyridoxine and niacin in multivitamin premix- es and supplements by high-performance liquid chro- matography. Journal of Micronutrient Analysis 7: 15– 26. Chaikin, S.W. & Brown, W.G. 1949. Reduction of alde- hydes, ketones and acid chlorides by sodium boro- hydride. Journal of the American Chemical Society 611 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 71: 122–125. Cheng, S. & Trumbo, P.R. 1993. Pyridoxine-5’-β-D-glu- coside: metabolic utilization in rats during pregnan- cy and availability to the fetus. Journal of Nutrition 123: 1875–1879. Chiari, M., Gelain, A., Riva, S. & Tura, D. 1997. Analysis of mandelonitrile lyase and beta-glucosidase from sweet almonds by combined electrophoretic tech- niques. Electrophoresis 18: 2050–4. Claessens, H.A., van Straten, M.A. & Kirkland, J.J. 1996. Effect of buffers on silica-based column stability in reversed-phase high-performance liquid chromatog- raphy. Journal of Chromatography A. 728: 259–270. Claypool, J. 1994. Sigma research technical service note. Sigma Chemical Company, MO, USA. Coburn, S.P. & Mahuren, J.D. 1983. A versatile cation- exchange procedure for measuring the seven major forms of vitamin B 6 in biological samples. Analytical Biochemistry 129: 310–317. Combs, G.F., Jr. 1992. What is a vitamin? In: The vita- mins. Fundamental aspects in nutrition and health. Academic Press. Inc. San Diego, California, ISBN 0- 12-183490-5. p. 3–8. COST91. 1985. Methods for the determination of vita- mins in food. Recommended by COST91. Brubach- er, G., Müller-Mulot, W. & Southgate, D.A.T. (eds.). Elsevier, Essex, UK. ISBN 0-85334-339-X. 166 p. Costa de Carvalho, M.J., Guilland, J.C., Moreau, D., Boggio, V. & Fuchs, F. 1996. Vitamin status of healthy subjects in Burgundy (France). Annals of Nutrition and Metabolism 40: 24–51. Cravo, M.L., Gloria, L.M., Selhub, J., Nadeau, M.R., Camilo, M.E., Resende, M.P., Cardoso, J.N., Leitao, C.N. & Mira, F.C. 1996. Hyperhomocysteinamia in chronic alcoholism: correlation with folate, vitamin B- 6, and vitamin B-12 status. American Journal of Clin- ical Nutrition 63: 220–224. Dong, M.H., McGown, E.L., Schwenneker, B.W. & Sau- berlich, H.E. 1980. Thiamine, riboflavin, and vitamin B6 contents of selected foods as served. Journal of the American Dietetic Association 76: 156–160. Dong, M.W., Lepore, J. & Tarumoto, T. 1988. Factors af- fecting the ion-pair chromatography of water-soluble vitamins. Journal of Chromatography 442: 84–95. Driest P.J., van den, Ritchie, H.J. & Rose, S. 1988. Octa- decyl bonded phases on silica gel supports. LC-GC 6: 124–127. Duggan, D.E., Bowman, R.L., Brodie, B.B. & Udenfriend, S. 1957. A spectrofluorometric study of compounds of biological interest. Archives of Biochemistry and Biophysics 68: 1–14. Edwards, P., Liu, P.K.S & Rose, G.A. 1989. A simple liq- uid chromatographic method for measuring vitamin B 6 compounds in plasma. Clinical Chemistry 35: 241– 245. Ekanayake, A. & Nelson, P.E. 1988. Applicability of an in vitro method for the estimation of vitamin B6 biolog- ical availability. Journal of Micronutrient Analysis 4: 1–15. Endo, H., Kamata, A., Hoshi, M., Hayashi, T. & Watan- abe, E. 1995. Microbial biocensor system for rapid determination of vitamin B6. Journal of Food Science 60: 554–557. Esteve, M.J., Farre, R., Frigola, A. & Carcia-Cantabella, J.M. 1998. Determination of vitamin B6 (pyridoxam- ine, pyridoxal and pyridoxine) in pork meat and pork meat products by liquid chromatography. Journal of Chromatography A 795: 383–387. Eurachem 1995. Quantifying uncertainty in analytical measurements. EURACHEM Working group on un- certainty in chemical measurement. 1st ed. Crown Copywright. ISBN 0-948926-08-2. 87 p. Evans, S.R., Gregory III, J.F. & Kirk, J. 1981. Thermal degradation kinetics of pyridoxine hydrochloride in dehyd rated model food system. Journal of Food Science 46: 555–558, 563. Fawell, D.J. 1990. Vitamin analysis – microbiological methods. Journal of Micronutrient Analysis 7: 237– 246. Fiorianti, J., Debellis, M., Arraiano, J. & Salotto, A.W. 1995. A comparison of douple endcapped vs. end- capped cyano columns utilizing high performance liq- uid chromatography. [Conference Meeting Abstract]. 210th ACS National Meeting. Chicago, IL, August 20– 24, Iss. Pt.1, CHED-146. Food and Nutrition Board 1989. Recommended dietary allowances. National Research Council. 10th ed. National Academic Press, Washington DC, USA. ISBN 0-309-04041-8. Giles, C.H., Smith, D. & Huitson, A. 1974. A General treat- ment and classification of the solute adsorption iso- therm. I Theoretical. Journal of Colloid Interface Sci- ence 47: 755–765. Gilpert, J.A. & Gregory III, J.F. 1992. Pyridoxine-5’-β-D- glucoside affects the metabolic utilization of pyridox- ine in rats. Journal of Nutrition 122: 1029–1035. Gregory III, J.F. 1980a. Determination of pyridoxal 5’- phosphate as the semicarbazone derivative using high-performance liquid chromatography. Analytical Biochemistry 102: 374–379. – 1980b. Bioavailability of vitamin B-6 in nonfat dry milk and a fortified rice breakfast cereal product. Journal of Food Science 45: 84–86, 114. – 1980c. Comparison of high-performance liquid chro- matographic and Saccharomyces uvarum Methods for the determination of vitamin B6 in fortified break- fast cereals. Journal of Agricultural and Food Chem- istry 28: 486–489. – 1980d. Effect of e-pyridoxyllysine bound to dietary protein on the vitamin B6 status of rats. Journal of Nutrition 110: 995–1005. – 1982. Relative activity of the nonphosphorylated B6 vitamers for Saccharomyces uvarum and Kloeckera brevis in vitamin B6 microbiological assay. Journal of Nutrition 112: 1643–1647. – 1983. Microbiological assays of total vitamin B-6 us- ing the Yeasts Saccharomyces uvarum and Kloeck- era brevis. (Responce to Drs. Guilarte and Tsan Let- ter). Journal of Nutrition 113: 722–724. – 1988a. Methods for determination of vitamin B6 in foods and other biological materials: A critical review. Journal of Food Composition and Analysis 1: 105– 123. – 1988b. Recent developments in methods for the as- 612 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods sessment of vitamin bioavailability. Food Technolo- gy October 1988: 230, 233, 235, 237–238. – 1992. A novel vitamin B6 metabolite may be a circu- lating marker of cancer. Nutrition Reviews 50: 295– 297. – & Feldstein, D. 1985. Determination of vitamin B-6 in foods and Other Biological Materials by paired-ion high-performance liquid chromatography. Journal of Agricultural and Food Chemistry 33: 359–363. – & Hiner, M.E. 1983. Thermal stability of vitamin B6 compounds in liquid model food systems. Journal of Food Science 48: 1323–1327. – & Ink, S.L. 1987. Identification and quantification of pyridoxine-β-glucoside as a major form of vitamin B6 in plant-derived foods. Journal of Agricultural and Food Chemistry 35: 76–82. – & Kirk, J.R. 1977. Interaction pf pyridoxal and pyri- doxal phosphate with peptides in a model food sys- tem during thermal processing. Journal of Food Sci- ence 42: 1554–1557, 1561. – & Kirk, J.R. 1978a. Vitamin B-6 Activity for rats of e- pyridoxyllysine bound to dietary protein. Journal of Nutrition 108: 1192–1199. – & Kirk, J.R. 1978b. Assesment of roasting effect on the vitamin B6 stability and bioavailability in dehy- drated food systems. Journal of Food Science 43: 1585–1589. – & Leatham, K. 1990. Lack of vitamin B6 activity of 6- hydroxypyridoxine. J Food Sci Off Publ Inst Food Technol 55: 1143–1146. – & Litherland, S.A. 1986. Efficacy of the rat bioassay for the determination of biologically available vitamin B-6. Journal of Nutrition 116: 87–97. – , Manley, D.B. & Kirk, J.R. 1981. Determination of vi- tamin B-6 in animal tissues by reversed-phase high- performance liquid chromatography. Journal of Agri- cultural and Food Chemistry 29: 921–927. – & Nakano, H. 1997. Preparation of nonlabeled, triti- ated, and deuterated pyridoxine 5’-β-D-glucoside and assay of pyridoxine-5’-β-D-glucoside hydrolase. Methods in Enzymology 280:58–65. – , Trumbo, P.R., Bailey, L.B., Toth, J.P., Baumgartner, T.G. & Cerda, J.J. 1991. Bioavailability of pyridox- ine-5’-β-D-glucoside Determined in humans by Sta- ble-Isotope Methods. Journal of Nutrition 121: 177– 186. Gregory, M.E. & Mabbit, L.A. 1959. The effect of heat on the vitamin B6 of milk. I. Microbiological tests. Jour- nal of Dairy Research 26: 203–214. – & Mabbit, L.A. 1961.The occurance of pyridoxamine phosphate in milk. Journal of Dairy Research 28: 293–301. Grimble, R.F. 1997. Effect of antioxidative vitamins on immune function with clinical applications. Interna- tional Journal for Vitamin and Nutrition Research 67: 312–320. Guilarte, T.R. 1983. Radiometric microbiological assay of vitamin B6: Assay simplification and sensitivity study. Journal of the Association of Official Analyti- cal Chemists 66: 58–61. – 1984. Effect of sodium chloride and potassium chlo- ride on growth responce of yeasts Saccharomyces uvarum and Kloeckera brevis to free vitamin B6. Jour- nal of the Association of Official Analytical Chemists 67: 617–620. – & Tsan, M.-F. 1983. Microbiological assays of total vitamin B-6 using the yeasts Saccharomyces uvar- um and Kloeckera brevis (Letters to the editor). Jour- nal of Nutrition 113: 721–722. Hachey, D.L., Coburn, S.P., Brown, L.T., Erbelding, W.F., DeMark, B. & Klein, P.D. 1985. Quantitation of vita- min B6 in biological samples by isotope dilution mass spectrometry. Analytical Biochemistry 151: 159–168. Hall, L.D. 1964. Nuclear magnetic resonance. In: Wolfrom, M.L. & Tipson, R.S. (eds.). Advances in carbohydrate chemistry. vol. 19. Academic Press, New York. p. 51–93. Haller, J., Löwik, M.R.H., Ferry, M. & Ferro-Luzzi A. 1991. Nutritional status: blood vitamins A, E, B6, B12, folic acid and carotene. Euronut SENECA investigators. European Journal of Clinical Nutrition 45: 63–82. Hamaker, B., Kirksey, A., Ekanayake, A. & Borschel, M. 1985. Analysis of B-6 Vitamers in human milk by re- verse-phase liquid chromatography. American Jour- nal of Clinical Nutrition 42: 650–655. Handbook of proton-NMR spectra and data. 1985. Sasa- ki, S.-I. (ed.). Vol. 3. Academic Press, Tokyo. ISBN 0-12-064503.188 p. Hansen, C.M., Leklem, J.E. & Miller, L.T. 1996. Vitamin B-6 status indicators decrease in women consuming a diet high in pyridoxine glucoside. Journal of Nutri- tion 126: 2512–2518. Harper, H.A., Rodwell, V.W. & Mayes, P.A. 1979. Review of physiological chemistry. 17th ed. Lange medical Publishers, Los Altos, CA, USA. 245 p. Hart, J.P. & Hayler, P.J. 1986. Preliminary studies towards an assay for circulating vitamin B6 levels in plasma using high-performance liquid chromatography with Electrochemical Detection. Analytical Proceedings 23: 439–441. Hauenstein, B. 1990. Vitamine. Teil I. Gehaltsbestimmung der vitamine B1, B2B6, C, Nicotinamid und Folsäure in Multivitaminpräparaten. Schweizerische Labora- toriums Zeitung 47: 224–228. He, S. & Withers, S.G. 1997. Assignment of sweet al- mond beta-glucosidase as a family 1 glycosidase and identification of its active site nucleophile. Journal of Biological Chemistry 272: 24864–7. Hefferan, T.E., Chrisley, B.M. & Driskell. J.A. 1986. Quan- titationof B 6 vitamers in rat plasma by high-perform- ance liquid chromatography. Journal of Chromatog- raphy Biomedical Applications 374: 155–161. Heinonen, M. 1990. Carotenoids and retinoids in Finnish foods and the average diet. EKT Series 811. Univer- sity of Helsinki. Academic Dissertation. 75 p. Holland, B., Welsh, A.A., Unwin, I.D., Buss, D.H., Paul, A.A. & Southgate, D.A.T. 1991. McCance and Wid- dowsson’s the composition of foods. 5th ed. The Royal Society of Chemistry and Ministry of Agricul- ture, Fisheries and Food. Richard Clay Ltd, Bungay, Suffolk. ISBN 0-85186-391-4 Hollman, P.C.H., Slangen, J.H., Wagstaffe, P.J., Faure, U., Southgate, D.A.T. & Finglas, P.M. 1993. Intercom- parison of methods for the determination of vitamins 613 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. in foods; Part 2: Water-soluble vitamins. Analyst 118: 481–488. Hou, W. & Wang, E. 1990. Liquid chromatography with series dual-electrode electrochemical detection for riboflavin. Analyst 115: 139–141. Hudson, C.A., Betschart, A.A. & Oace, S.M. 1988. Bioa- vailability of vitamin B-6 from rat diets containing wheat bran or cellulose. Journal of Nutrition 118: 65– 71. Huopalahti, R., Mäkelä, E., Siren, H. & Vuorela, H. 1994. Dictionary of chromatography and related methods. Association of Finnish Chemical Society, Helsinki, Finland. ISBN 952-90-5998-1. 254 p. Ink, S.L., Gregory, J.F. III & Sartain, D.B. 1986. Determi- nation of pyridoxine β-glucosidase bioavailability us- ing intrinsic and extrinsic labeling in the rat. Journal of Agricultural and Food Chemistry 34: 857–862. ISO/IEC 25 Guide 1986. General requirements for the competence of calibration and testing laboratories. Iwami, K. & Yasumoto, K. 1986. synthesis of pyridoxine- β-glucoside by rice bran β-glucosidase and its in situ absorption in rat small intestine. Nutrition Research 6: 407–414. Iwase, H. 1992. Determination of nicotinamide and pyri- doxine in an elemental diet by column-switching high- performance liquid chromatography with UV-detec- tion. Journal of Chromatography 625: 377–381. Jaumann, G. & Engelhardt, H. 1985. On line trace en- richment for isolation of vitamin B2 in food samples. Chromatographia 20: 615–617. Jinno, K. & Lin, Y. 1995. Separation of carotenoids by high-performance liquid chromatography with poly- meric and monomeric octadecylsilica stationary phas- es. Chromatographia 41: 311–317. Joosten, E., van den Berg, A., Riezler, R., Naurath, H.J., Lindenbaum, J., Stabler, S.P. & Allen, R.H. 1993. Metabolic evidence that deficiencies of vitamin B-12 (cobalamine), folate, and vitamin B-6 occur commonly in elderly people. American Journal of Clinical Nutri- tion 58: 468–476. Joseph, T., Tsuge, H., Suzuki, Y. & McCormick, D.B. 1996. Pyridoxine 4’-α- and 5’-α-D-glucosides are taken up and metabolized by isolated rat liver cells. Journal of Nutrition 126: 2899–2903. Kabir, H., Leklem, J. & Miller, L.T. 1983a. Measurement of glycosylated vitamin B6 in foods. Journal of Food Science 48: 1422–1425. – , Leklem, J. & Miller, L.T. 1983b. Relationship of the glycosylated vitamin B6 content of foods to vitamin B6 bioavailability in humans. Nutr. Rept. Int. 28: 709– 716. Kazarinoff, M.N. & McCormick, D.B. 1975. Rabbit Liver pyridoxamine (pyridoxine) 5’-phosphatase oxidase. Purification and properties. Journal of Biological Chemistry 250: 3436–3442. Kawai, F., Horii, T., Yamada, H. & Ogata, K. 1972a. For- mation of pyridoxine glucoside-synthesizing enzyme (α-glucosidase) of Micrococcus sp. No. 431. Agricul- tural and Biological Chemistry 36: 2607–2609. – , Yamada, H. & Ogata, K. 1971a. Properties of pyri- doxine glucoside. Journal of Vitaminology 17: 121– 124. – , Yamada, H. & Ogata, K. 1971b. Studies on transgly- cosidation to vitamin B6 by microorganisms. IV Puri- fication of a bacterial enzyme catalyzing pyridoxine glycoside synthesis. Agricultural and Biological Chemistry 35: 184–190. – , Yamada, H. & Ogata, K. 1972b. Studies on transgly- cosylation to vitamin B6 by microorganism. VII. Trans- port of pyridoxine glucoside into rabbit erythrocytes. Journal of Vitaminology 18: 183–188. – , Yamada, H. & Ogata, K. 1973. The formation and hydrolysis pyridoxine glucoside by various α-glucosi- dases. Agricultural and Biological Chemistry 37: 1963–1965. Kawamoto, T., Okada, E. & Fujita, T. 1983. Post-column derivatization of vitamin B6 using 2,6-dibromoqui- none-4-chlorimide. Journal of Chromatography 267: 414–419. Khurana, A.L. & Ho, C.-T. 1988. Determination of sur- face silanols of silica gels and HPLC bonded phas- es. Journal of Liquid Chromatography 11: 3205–3211. Kimura, M., Kanehira, K. & Yokoi, K. 1996. Highly sensi- tive and simple liquid chromatographic determination in plasma of B6 vitamers, especially pyridoxal 5’- phosphate. Journal of Chromatography A 722: 296– 301. Knox, J.H. & Hartwick, R.A. 1981. Mechanism of ion-pair liquid chromatography of amines, neutrals, zwitteri- ons and acids using anionic hetareons. Journal of Chromatography 204: 3–21. Köhler, J. & Kirkland, J.J. 1987. Improved silica-based column packings for high-performance liquid chro- matography. Journal of Chromatography 385: 125– 150. Lai, L.B., Gopalan, V. & Glew, R.H. 1992. Continuous spectrophotometric assays for beta-glucosidases acting on the plant glucosides L-picein and pruna- sin. Analytical Biochemistry 200: 365–9. Lam, F.-L., Holcomb, I.J. & Fusari, S.A. 1984. Liquid chro- matographic assay of ascorbic acid, niacinamide, pyridoxine, thiamine, and riboflavin in multivitamin- mineral preparations. Journal of the Association of Official Analytical Chemists 67: 1007–1011. Laukkanen, M., Antila, P., Antila, V. & Salminen, K. 1988. The water-soluble vitamin contents of Finnish liquid milk products. Meijeritieteellinen Aikakauskirja XLVI: 7–24. Leklem, J.E., Miller, L.T., Perera, A.D. & Peffers, D.E. 1980. Bioavailability of vitamin B-6 from wheat bread in humans. Journal of Nutrition 110: 1819–1828. Lemieux, R.U., Kullnig, R.K., Bernstein, H.J. & Schnei- der, W.G. 1958. Configurational effects on the pro- ton magnetic resonance spectra of six-membered ring compounds. Journal of the American Chemical Society 80: 6098–6105. – & Stevens, J.D. 1965. Substitutional and configural effect on chemical shift in pyranoid carbohydrate de- rivatives. Canadian Journal of Chemistry 43: 2059– 2070. – & Stevens, J.D. 1966. The proton magnetic reso- nance spectra and tautomeric equilibria of aldoses in deuterium oxide. Canadian Journal of Chemistry 44: 249–262. 614 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods Li, C. 1992. Silanol group Effect in C 18 bonded phases in HPLC. Chromatographia 34: 182–184. Lim, K.L., Young, R.W., Palmer, J.K. & Driskell, J.A. 1982. Quantitative separation of B6 vitamers in selected foods by a gas-liquid chromatographic system equipped with an electron-capture detector. Journal of Chromatography 250: 86–89. Lindberg, A.S., Leklem, J.E. & Miller, L.T. 1983. The ef- fect of wheat bran on the bioavailability of vitamin B- 6 in young men. Journal of Nutrition 113: 2578–2586. Loo, Y.H. & Badger, L. 1969. Spectrofluorometric assay of vitamin B 6 analogues in brain tissue. Journal of Neurochemistry 16: 801–804. Löwik, M.R., Schrijver, J., van den Berg, H., Hulshof, K.F., Wedel, M. & Ockhuizen, T. 1990. Effect of dietary fiber on the vitamin B6 status among vegetarian and nonvegetarian elderly (Dutch nutrition surveillance system). Journal of the American College of Nutri- tion 9: 241–249. Maeda, Y., Yamamoto, M., Owada, K., Sato, S. & Naka- zawa, H. 1989. Simultaneous liquid chromatograph- ic determination of water-soluble vitamins, caffeine, and preservative in oral liquid tonics. Journal of the Association of Official Analytical Chemists 72: 244– 247. Mahuren , J.D. & Coburn, S.P. 1997. Determination of 5- pyridoxic acid, 5-pyridoxic acid lactone, and other vitamin B6 compounds by cation-exchange high-per- formance liquid chromatography. Methods in Enzy- mology 280: 22–29. Manore, M.M., Vaughan, L.A. & Lehman, W.R.1990. Con- tribution of various food groups to dietary vitamin B- 6 intake in free-living low-income elderly persons. Journal of the American Dietetic Association 90: 830– 834. Mao, Y. & Fung, B.M. 1997. Use of alumina with anchored polymer coating as packing material for reversed- phase high-performance liquid chromatography. Jour- nal of Chromatography, A. 790: 9–15. Mascher, H. 1993. Determination of total pyridoxal in human plasma following oral administration of vita- min B 6 by high-performance liquid chromatography with post-column derivatization. Journal of Pharma- ceutical Sciences 82: 972–974. – 1997. High-performance liquid chromatographic de- termination of total pyridoxal in human plasma. Meth- ods in Enzymology 280: 12–21. Matsubara, Y., Yusa, T., Sawabe, A., Iizuka, Y. & Okamo- to, K. 1991. Structure and physiological activity of phenyl propanoid glycosides in lemon (Citrus limon BURM. f.) Peel. Agricultural and Biological Chemis- try 55: 647–650. Mattila, P. 1995. Analysis of cholecalsiferol, ergocalcif- erol and their 25-hydroxylated metabolites in foods by HPLC. EKT Series 995. University of Helsinki. Academic Dissertation. 81 p. McCalley, D.V. 1995. Influence of organic solvent modi- fier and solvent strength on peak shape of some ba- sic compounds in high-performance liquid chroma- tography using reversed-phase column. Journal of Chromatography (A) 708: 185–194. Metzler, D.E. 1957. Equilibria between pyridoxal and amino acids and their imines. Journal of the Ameri- can Chemical Society 79: 485–490. – & Snell, E.E. 1952. Some transamination reactions involving vitamin B6. Journal of the American Chem- ical Society 74: 979–983. – & Snell, E.E. 1955. Spectra and ionization constants of the vitamin B6 group and related 3-hydroxypyrid- ine derivatives. Journal of the American Chemical So- ciety 77: 2431–2437. Michela, P. & Lorenz, K. 1976. The vitamins of triticale, wheat, and rye. Cereal Chemistry 53: 853–861. Millart, H. & Lamiable, D. 1989. Determination of pyri- doxal 5’-phosphate in human serum by reversed- phase liquid chromatography combined with spec- trofluorimetric detection of 4-pyridoxic acid 5’-phos- phate as a aerivative. Analyst 114: 1225–1228. Møller, A. 1985. Levnedsmiddeltabeller [Food Composi- tion Tables], Frederikshavn. ISBN 87-7320-279-7. Möller, J. & Bitsch, R. 1988. Einsatz der Hochleistungs- Flüssigchromatographie (HPLC) zur simultanen Be- stimmung der pyridoxin-Derivative aus biologischen Materialen. Ernährungs-Umschau 35: 324–328. Mondino, A., Bongiovanni, G., Fumero, S. & Rossi, L. 1972. An improved method of plasma deproteination with sulphosalicylic acid for determining amino acids and related compounds. Journal of Chromatography 74: 255–263. Morrison, L.A. & Driskell, J.A. 1985. Quantities of B6 vi- tamers in human milk by high performance liquid chromatography. Journal of Chromatography 337: 249–258. Musch, G. & Massert, D.L. 1988. Isolation of basic drugs from plasma using solid-phase extraction with a cy- anopropyl-bonded phase. Journal of Chromatogra- phy Biomedical Applications 432: 209–222. Nair, L.M., Kildew, B.R. & Saari-Nordhaus, R. 1996. En- hanging the anion separations on a polydivinylben- zene-based anion stationary phase. Journal of Chro- matography A, 739: 99–110. Nakano, H. & Gregory, J.F. 1995. Pyridoxine and pyri- doxine-5’-β-D-glucoside exert different effects on tis- sue B-6 vitamers but similar effects on β-glucosidase activity in rats. Journal of Nutrition 125: 2751–2762. – , McMahon, L.G. & Gregory, J.F. 1997. Pyridoxine-5’- β-D-glucoside exhibits incomplete bioavailability as a a source vitamin B-6 and partially inhibits the utili- zation of co-ingested pyridoxine in humans. Journal of Nutrition 127: 1508–1513. Nasuto, R. & Rózylo, J.K. 1995. Column packing with chemically bonded methyl-vinyl stationary phase for RP HPLC. Acta Chromatographica 4: 50–63. Navankasattusas, S. & Lund, D.B. 1982. Thermal destruc- tion of vitamin B6 vitamers in buffer solutions and in cauliflower puree. Journal of Food Science 47: 1512– 1518. Nelson, E.W., Burgin, C.W. & Cerda, J.J. 1977. Charac- terization of food binding of vitamin B-6 in orange juice. Journal of Nutrition 107: 2128–2134. Nguyen, L.B. & Gregory, J.F. 1983. Effects of food com- position on the bioavailbility of vitamin B-6 in the rat. Journal of Nutrition 113: 1550–1560. – , Gregory III, J.F., Burgin, C.W. & Cerda, J.J. 1981a. 615 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. In vitro binding of vitamin B6 by selected polysac- charides, lignin and wheat bran. Journal of Food Sci- ence 46: 1860–1862. – , Gregory III, J.F. & Damron. B.L. 1981b. Effect of se- lected polysaccharides on the bioavailability of pyri- doxine in rats and chicks. Journal of Nutrition 111: 1403–1410. – , Hiner, M.E., Litherland, S.A. & Gregory III, J.F. 1983. Relative biological activity of nonphosphorylated vi- tamin B-6 compounds in the rat. Journal of Agricul- tural and Food Chemistry 31: 1282–1287. NNR 1996. Nordiska näringsrekommendationer [Recom- mended dietary allowances in Nordic Countries, in Swedish]. Nord 1999: 28. Nordiska Ministerrådet, Copenhagen. ISBN 92-9120-930-9. Ogata, K., Tani, Y., Uchida, Y. & Tochikura, T. 1968. Mi- crobial formation of pyridoxine glucoside, a new de- rivative of vitamin B 6 . Preliminary notes. Biochimica Biophysica Acta 165: 578–579. – , Tani, Y., Uchida, Y. & Tochikura, T. 1969a. Studies on transglycosidation to vitamin B6 by microorgan- isms. I. Formation of a new vitamin B6 derivative, pyridoxine glucoside, by Sarcina lutea. Journal of Vi- taminology 15: 142–150. – , Uchida, Y., Kurihara, N., Tani, Y. & Tochikura, T. 1969b. Studies on transglycosidation to vitamin B6 by microorganisms. II Chemical structure of pyridox- ine glycoside. Journal of Vitaminology 15: 160–166. Ohishi, N. & Fukui, S. 1968. Further study on the reac- tion products of pyridoxal and pyridoxal 5’-phosphate with cyanide. Archives of Biochemistry and Biophys- ics 128: 606–610. Olds, S.J., Vanderslice, J.T. & Brochetti, D. 1993. Vita- min B6 in raw and fried chicken by HPLC. Journal of Food Science 58: 505–507, 561. Olleta, F.J., Ljanos, E., Barcos, R., Ancín, M.C. & Martin, O. 1993. Determinación de vitaminas B1, B2, B6 y C en champion y lenteja por HPLC. Efecto de la trans- formación en conserva y el tiempo de almacenami- ento. Química Analítica 12: 155–158. Ollilainen, V., Finglas, P.M. van den Berg, H., & de Froid- mont-Görtz, I. Certification of B-group vitamins (B 1 , B2, B6 and B12) in four food reference materials. Jour- nal of Agricultural and Food Chemistry: 00: 000–000. Pastores, G., Lonardo, C., Variano, B. & Anderson, M.L. 1995. The use of a non-porous reversed phase col- umn for resolution of porcine insulin from low molec- ular weight amides in the matrix. Journal of Liquid Chromatography 18: 3049–3059. Pearson, J.D., Lin, N.T. & Regnier, F.E. 1982. The im- portance of silica type for reverse-phase protein sep- aration. Analytical Biochemistry 124: 217–230. Peterson, E.A., Sober, H.A. & Meister, A. 1955. Pyridox- amine phosphate (2-methyl-3-hydroxy-4-aminome- thyl-5-pyridylmethyl-phosphoric acid). Biochem. Preparations 3: 29–34. Pierotti, J.A., Dickinson, A.G., Palmer, J.K. & Driskell, J.A. 1984. Liquid chromatographic separation and quantitation of B6 vitamers in selected rat tissues. Journal of Chromatography Biomedical Applications 306: 377–382. Piironen, V. 1986. Tokoferolit ja tokotrienolit elintarvikkeis- sa ja keskimääräisessä suomalaisessa ruokavalios- sa [Tocopherols and Tocotrienols in foods and in av- erage Finnish diets, in Finnish]. EKT Series 811. University of Helsinki. Academic Dissertation. 173 p. Polansky, M.M. 1981. Microbiological assay of vitamin B-6 in foods. In: Leklem, J.E. & Reynolds, R.D. (eds.). Methods in vitamin B6 nutrition. Analysis and status assessments. Plenum Press, New York. p. 21–44. – , Reynolds, R.D. & Vanderslice, J.T. 1985. Vitamin B6. In: Augustin, J. et al. (eds.). Methods in vitamin as- say. John Wiley and Sons, New York. p. 417. – & Toepher, E.W. 1969. Vitamin B6 components in some meats, fish, dairy products, and commercial infant formulas. Journal of Agricultural and Food Chemistry 17: 1394–1397. Rabinowitz, J.C. & Snell, E.E. 1947. The vitamion B6 group. Extraction procedures for the microbiological determination of vitamin B 6 . Ind. Eng. Chem. Anal. Ed. 19: 277–280. – & Snell, E.E. 1948. The vitamin B6 Group. XIV. Dis- tributions of pyridoxal, pyridoxamine, and pyridoxine in some natural products. Journal of Biological Chem- istry 176: 1157–1167. Rastas, M., Seppänen, R., Knuts, L.-R., Karvetti, R.-L. & Varo, P. (eds.). 1993. Ruoka-aineiden ravintoaine- sisältö, Livsmedlens näringsinnehåll, Nutrient com- position of foods. Kansaneläkelaitos, Helsinki. ISBN 951-669-352-0. 461 p. Razagui, I.B. & Barlow, P.J. 1989. A direct spectrofluori- metric procedure for the determination of pyridoxine in multivitamin preparations. Journal of Micronutri- ent Analysis 6: 119–126. Rees, D.I. 1989. Determination of nicotinamide and pyri- doxine in fortified food products by HPLC. Journal of Micronutrient Analysis 5: 53–61. Regnier, F.E. 1987. HPLC of biochemical macromole- cules: The first decade. Chromatographia 24: 241– 251. Reitzer-Bergaentzle, M., Marchioni, E. & Hasselmann, C. 1993. HPLC determination of vitamin B6 in foods after pre-column derivatization of free and phospho- rylated vitamers into pyridoxine. Food Chemistry 48: 321–324. Reynolds, R.D. 1988. Bioavailability of vitamin B6 from plant foods. American Journal of Clinical Chemistry 48: 863–867. Reynolds, T.M. & Brain, A. 1992. A simple internally-stand- ardized isocratic HPLC assay for vitamin B 6 in hu- man serum. Journal of Liquid Chromatography 15: 897–914. Riekkola, M.-L. (ed.). 1995. Nomenclature for chroma- tography. IUPAC Recommendations 1993. IUPAC & Association of Finnish Chemical Societies, Helsinki, Finland. Rizzolo, A. & Polesello, S. 1992.Chromatographic deter- mination of vitamins in foods. Review. Journal of Chromatography 624: 103–152. Robinson, K., Mayer, E.L., Miller, D.P., Green, R., van Lente, F., Gupta, A., Kottke-Marchant, K., Savon, S.R., Selhub, J., Nissen, S.E., Kutner, M., Topol, E.J. & Jacobsen, D.W. 1995. Hyperhomocysteinemia and low pyridoxal phosphate. Common and independent 616 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods reversible risk factors for coronary artery disease. Circulation 92: 2825–2830. Rubin, S.H., Schener, J. & Hirschberg, E. 1947. The avail- ability of vitamin B6 in yeast and liver for growth of Saccharomyces carlsbergensis. Journal of Biologi- cal Chemistry 167: 599–611. Rudrum, M. & Shaw, D.F. 1965. The structure and con- formation of some monosaccharides in solution. Jour- nal of Chemical Society I: 52–57. Saidel, L.S. & Modapallimattam, G. 1970. The effect of protein precipitants on α-chymotrypsin and other enzymes. Indian Journal of Biochemistry and Bio- physics 7: 157–162. Saidi, B. & Warthesen, J.J. 1983. Influence of pH and light on the kinetics of vitamin B6 degradation. Jour- nal of Agricultural and Food Chemistry 31: 876–880. Sander, L.C. & Wise, S.A. 1984. Synthesis and chacter- ization of polymeric C 18 stationary phases for liquid chromatography. Analytical Chemistry 56: 504–510. Sampson, D.A., Eoff, L.A., Yan, X.L. & Lorenz, K. 1995. Analysis of free and glycosylated vitamin B6 in wheat by high-performance liquid chromatography. Cereal Chemistry 72: 217–221. – , Wen, Q.B. & Lorenz, K. 1996. Vitamin B6 and pyri- doxine glucoside content of wheat and wheat flours. Cereal Chemistry 73: 770–774. Schill, G. 1989. High performance ion-pair chromatogra- phy. Journal of Biochemical and Biophysical Meth- ods 18: 249–270. Schramm, W. & Bitsch, R. 1991. Distribution of pyridox- ine in plant foods as affected by storing and process- ing. Proc. Euro Food Chemistry VI: 885–890. Sep. 22–26, 1991, Hamburg, Germany. – & Bitsch, R. 1993. Selective measurement of vita- min B-6 derivatives in plant foods. Bioavailability ‘93 Proc. 417–421. May. 9–12, 1993, Ettlingen, Germa- ny. Schoonhoven, J. van, Schrijver, J., Berg, H. van den, & Haenen, G.R.M.M. 1994. Reliable and sensitive high- performance liquid chromatographic method with fluorometrix detection for the analysis of vitamin B-6 in foods and feeds. Journal of Agricultural and Food Chemistry 42: 1475–1480. Selhub, J., Jacques, P.F., Bostom, A.G., D’Agostini, R.B., Wilson, P.W.F., Belanger, A.J., O’Leary, D.H., Wolf, P.A., Schaeffer, E.J. & Rosenberg, I.H. 1995. Asso- ciation between plasma homocysteine concentrations ans extracranial carotid-artery stenosis. New Eng- land Journal of Medicine 332: 286–291. Shane, B. & Snell, E.E. 1976. Transport and metabolism of vitamin B6 in the yeast Saccharomyces carlsber- gensis 4228. Journal of Biological Chemistry 1042– 1051. Sharma, S.K. & Dakshinamurti, K. 1992. Determination of vitamin B6 vitamers and pyridoxic acid in biologi- cal samples. Journal of Chromatography Biomedical Applications 578: 45–51. Shephard, G.S., Louw, M.E.J. & Labadarious, D. 1987. Analysis of vitamin B6 vitamers in in plasma by cati- on-exchange high-performance liquid chromatogra- phy. Journal of Chromatography Biomedical Appli- cations 416: 138–143. – , Westhuisen, L. van der, & Labadarious, D. 1989. Analysis of vitamin B6 vitamers in human tissue by cation-exchange high-performance liquid chromatog- raphy. Journal of Chromatography Biomedical Appli- cations 491: 226–234. Shimada, K., Nakagawa, S., Hotta, K., Shibata, K. & Yagi, T. 1993. Effect of processing and storage on the fate of vitamin B1, B2 and B6 and nicotinamide of sea urchin gonads. Journal of Agricultural and Food Chemistry 41: 1021–1025. Shu, C.-K. & Lawrence, B.M. 1994. Synthesis of 2-phe- nylethyl α-glucoside. Journal of Agricultural and Food Chemistry 42: 1732–1733. Shultz, T.D. & Leklem, J.E. 1987. Vitamin B-6 status and bioavailability in vegetarian women. American Jour- nal of Clinical Nutrition 46: 647–651. Sieber, R., Eberhard, P., Fuchs, D., Gallmann, P.U. & Strahm, W. 1996. Effect of microwave heating on vi- tamins A, E, B1, B2 and B6 in milk. Journal of Dairy Research 63: 169–72. Sierra, I. & Vidal-Valverde, C. 1997. A simple method to determine free and glycosylated vitamin B6 in leg- umes. Journal of Liquid Chromatography & Related Technologies 20: 957–969. Sigg, S. 1985. Thermische Abbaukinetik von vitamin B6 in Modellösungen and in Rindfleisch. Diplomarbeit. Institute für Biologische and Ernährungswissen- schaft, Pharmakologie und Toxikologie. Universität Hohenheim, Stuttgart-Hohenheim. 104 p. SLV 1993. Livsmedelstatabell [Food composition tables, in Swedish]. Statens livsmedelsverk, Sweden. Nord- stedts tryckeri. ISBN 91-7714-026-5. Snell, E.E. 1954. The vitamin B6 group. The reversible interconversion of PL and PM by transamination re- action. Journal of the American Chemical Society 67: 194–197. – 1963. Vitamin B6. In: Florkin, M. & Stotz, E.H. (eds.). Water-Soluble vitamins, Hormones, Antibiotics. Com- prehensive Biochemistry. Vol 11. Elsevier, Amster- dam. – 1990. Vitamin B6 and decarboxylation of histidine. In: Dakshinamurti, K. (ed.). Vitamin B6. Annals of the New York Academy of Sciences 585: 1–12. Snyder, L.R. & Kirkland, J.J. 1979. Introduction to mod- ern liquid chromatography. 2nd ed. John Wiley & Sons. New York. ISBN 0-471-03822-9. 863 p. Sokal, R.R. & Rohlf, F.J. 1995. Biometry. The principles and practice of statistics in biological research. 3rd ed. W.H. Freeman and Company, New York. ISBN 0- 7167-2411-1. 887 p. Souci, S.W., Fachmann, W. & Kraut, H. 1994. Food com- position and nutrition tables. 5th ed. Medpharm Sci- entific Publishers, Stuttgart. Spherical materials and columns for HPLC. 1976. Phase Separations, UK. 8 p. Srividya, K. & Balasubramanian, N. 1997. Spectrophoto- metric determination of pyridoxine hydrochlorine in pharmaceutical preparations and foods. Journal of AOAC International 80: 1368–1373. Srncova, V. & Davidek, J. 1972. Reaction of pyridoxal-5- phosphate with proteins. Reaction of pyridoxal with milk serum protein. Journal of Food Science 37: 310– 617 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. 312. Stadalius, M.A., Berus, J.S. & Snyder, L.R. 1988. Reversed-phase HPLC of basic samples. LC-GC 6: 494–500. Steiner, I., Hruschka, G., Washüttl, J. & Kroyer, G. 1993. Ausgewählte vitamine des B-Komplexes in ver- schiedenen Lebensmitteln nach konventioneller Zu- berei tung and nach Mikrowel lenbehandlung. Ehrährung 17: 221–225. Stewart, K.K. 1995. Total compositing, Editorial. Journal of Food Composition and Analysis 8: 89. Suzuki, Y., Doi Y., Uchida, K. & Tsuge, H. 1997. Enzy- matic preparation of pyridoxine 4’- and 5-α-D-gluco- sides. Methods in Enzymology 280: 66–71. – & Uchida, K. 1997. Formation of β-galactosides of pyridoxine using Sporobolomyces singularis. Meth- ods in Enzymology 280: 71–77. Sykora, D., Tesarova, E. & Popl, M. 1997. Interactionsof basic compounds in reversed-phase high-perform- ance liquid chromatography. Influence of sorbent character, mobile phase composition, and pH on the retention of basic compounds. Journal of Chroma- tography A 758: 37–51. Tadera, K., Arima, M., Yoshino, S., Yagi, F. & Kobayashi, A. 1986b. Conversion of pyridoxine into 6-hydroxypy- ridoxine by food components, especially ascorbic acid. Journal of Nutritional Science and Vitaminolo- gy (Tokyo) 32: 267–277. – , Kaneko, T. & Yagi, F. 1986a. Evidence for the occur- rence and distribution of a new type of vitamin B6 conjugate on plant foods. Note. Agricultural and Bio- logical Chemistry 50: 2933–2934. – , Kaneko, T. & Yagi, F. 1988. Isolation and structural elucidation of three new pyridoxine-glycosides in rice bran. Journal of Nutritional Science and Vitaminolo- gy 34: 167–177. – , Mori, E., Yagi, F. Kobayashi, A. & Imabeppu, M. 1985. Isolation and structure of a minor metabolite of pyri- doxine in seedlings of Pisum sativum L. Agricultural and Biological Chemistry 47: 403–408. – , Nagano, K., Yagi, F., Kobayashi, A., Imada, A. & Tan- aka, S. 1983. Isolation and structure of a new me- tabolite of pyridoxine in seedlings of Pisum sativum L. Agricultural and Biological Chemistry 37: 1357– 1359. – & Naka, Y. 1991. Isocratic paired-ion high-perform- ance liquid chromatographic method to determine B6 vitamers and pyridoxine glucoside in foods. Note. Ag- ricultural and Biological Chemistry 55: 563–564. Tarr, J.B., Tamura, T. & Stokstad, E.L.R. 1981. Availabil- ity of vitamin B6 and pantothenate in an avarage american diet in man. American Journal of Clinical Nutrition 34: 1328–1337. Toepher, E.W. & Lehmann, J. 1961. Procedure for chro- matographic separation and microbiological assay of pyridoxine, pyridoxal, and pyridoxamine in food ex- tracts. Journal of the Association of Official Analyti- cal Chemists 44: 426–430. – & Polansky, M.M. 1970. Microbiological assay of vi- tamin B 6 and its components. Journal of the Associ- ation of Official Analytical Chemists 53: 546–550. Tolomelli, B., Laffi, R. & Marchetti, S. 1991. Chromatogra- fia liquida per il dosaggio della vitamina B 6 negli ali- menti. Industrie Alimentari 30: 1055–1058. Toukairin-Oda, T., Sakamoto, E., Hirose, N., Mori, M., Itoh, T. & Tsuge, H. 1989. Determination of vitamin B6 derivatives in foods and microbiological materials by reversed-phase HPLC. Journal of Nutritional Science and Vitaminology 35: 171–180. Trumbo, P.R. & Gregory III, J.F. 1988. Metabolic utiliza- tion of pyridoxine-β-glucoside in rats: Influence of vi- tamin B-6 status and route of administration. Journal of Nutrition 118: 1336–1342. – & Gregory III, J.F. 1989. The fate of dietary pyridox- ine-β-glucoside in the lactating rat. Journal of Nutri- tion 119: 36–39. – , Gregory III, J.F. & Sartain, D.B. 1988. Incomplete uti- lization of pyridoxine-β-glucoside as vitamin B-6 in the rat. Journal of Nutrition 118: 170–175. Tryfiates, G.P. 1996. Adenosine-N6-diethylthioether-N1- pyridoximine-5’-phosphate, a novel marker for human cancer detection. Anticancer Research 16: 2201– 2204. – , Bishop, R.E. & Murgo, A.J. 1991. Vitamin B6 and can- cer: A novel pyridoxal 5-phosphate conjugate in tu- mor cells and blood of cancer patients. Anticancer Research 11: 1281–1284. – , Gannet, P.M., Bishop, R.E., Shastri, P.K., Ammons, J.R. & Arbogast, J.G. 1996. Vitamin B6 and cancer: Synthesis and occurrance of adenosine-N6-diethyl- thioether-N-pyridoximine-5’-phosphate, a circulating human tumor marker. Cancer Research 56: 3670– 3677. Tsuge, H. 1997. Determination of vitamin B6 vitamers and metabolites in a biological sample. Methods in Enzymology 280: 3–12. – , Maeno, M., Hayakawa, T. & Suzuki, Y. 1996. Com- parative study of pyridoxine-α,β-glucosides, and phosphopyridoxyl-lysine as a vitamin B6 nutrient. Journal of Nutritional Science and Vitaminology 42: 377–386. – , Oda, T. & Miyama, H. 1986. Separation and determi- nation of vitamin B6 derivatives by reversed-phase HPLC. Agricultural and Biological Chemistry 50: 195– 197. – , Toukairin-Oda, T., Shoji, T., Sakamoto, E., Mori, M. & Suda, H. 1988. Fluorescence enhancement of PLP for application to HPLC. Agricultural and Biological Chemistry 52: 1083–1086. Ubbink, J.B., Serfontein, W.J. & de Villiers, L.S. 1986. Analytical recovery of protein-bound pyridoxal-5’- phosphate in plasma analysis. Journal of Chroma- tography Biomedical Applications 375: 399–404. Umehara, K., Hattori, I., Miyase, T., Ueno, A., Hara, S. & Kageyama, C. 1988. Studies on the constituents of leaves of Citrus unshiu Marcov. Chemical & Phar- maceutical Bulletin 36: 5004–5008. Unger, K.K., Becker, N. & Roumeliotis, P. 1976. Recent developments in the evaluation of chemically bond- ed silica packings for liquid chromatography. Jour- nal of Chromatography 125: 115–127. USDA 1999. U.S. Department of Agriculture, Agricultural Research Service. USDA Nutrient Database for Standard Reference, Release 12. Nutrient Data Lab- 618 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Ollilainen, V. HPLC analysis of vitamin B 6 in foods oratory Home Page. Available at Internet: http:// www.nal.usda.gov/fnic/foodcomp/Data/SR10/. Vahteristo, L. 1998. Food folates and their analysis: De- termination of folate derivatives and their stability by high-performance liquid chromatography. EKT Series 1150. University of Helsinki. Academic Dissertation. 75 p. Vanderslice, J.T., Brownlee, S.R.,. & Cortissoz, M.E. 1984. Liquid chromatographic determination of vita- min B6 in foods. Journal of the Association of Official Analytical Chemists 67: 999–1006. – & Maire, C.E. 1980. Liquid chromatographic separa- tion and quantification of B6 vitamers at plasma con- centration levels. Journal of Chromatography 196: 176–179. – , Maire, C.E. & Beecher, G.B. 1981a. Extraction and quantitation of B-6 vitamers from animal tissues and human plasma: A preliminary study. In: Leklem, J.E. & Reynolds, R.D. (eds.). Methods in vitamin B6 Nu- trition. Analysis and Status Assessments. Plenum Press, New York. p. 123–148. – , Maire, C.E., Doherty, R.F. & Beecher, G.R. 1980. Sul- fosalicylic acid as an extraction agent for vitamin B6 in food. Journal of Agricultural and Food Chemistry 28: 1145–1149. – , Maire, C.E. & Yakupkovic, J. E. 1981b. Vitamin B6 in ready-to-eat cereals: Analysis by high performance liquid chromatography. Journal of Food Science 46: 943–946. – , Stewart, K.K. & Yarmas, M.M. 1979. Liquid chroma- tographic separation and quantification of B6 vitam- ers and their metabolite, pyridoxic acid. Journal of Chromatography 176: 280–285. Verhoef, P., Stampfer, M.J., Buring, J.E., Gaziano, J.M., Allen, R.H., Stabler, S.P., Reynolds, R.D., Kok, F.J., Hennekens, C.H. & Willett, W.C. 1996. Homocysteine metabolism and risk of myocardinal infarction: Rela- tion with vitamin B6, B12, and folate. American Jour- nal of Epidemology 143: 845–859. Voigt, M.N., Eitenmiller, R.R. & Ware, G.O. 1979a. Vita- min analysis by microbial and protozoan organisms: Responce to food preservatives and neutralization salts. Journal of Food Science 44: 723–728, 737. – , Eitenmiller, R.R. & Ware, G.O. 1979b. Comparison of protozoan and conventional methods of vitamin analysis. Journal of Food Science 44: 729–734. Walters, M.J. 1987. Classification of octadecyl-bonded liquid chromatography columns. Analytical Chemis- try 53: 465–469. Wang, E. & Hou, W. 1988a. Determination of water-solu- ble vitamins using high-performance liquid chroma- tography and electrochemical or absorbance detec- tion. Journal of Chromatography 447: 256–262. – & Hou, W. 1988b. Determination of vitamin B6 using high-performance liquid chromatography with elec- trochemical detection. Microchemical Journal 37: 338–344. Wehling, R.L. & Wetzel, D.L. 1984. Simultaneous deter- mination of pyridoxine, riboflavin, and thiamin in for- tified cereal products by high-performance liquid chromatography. Journal of Agricultural and Food Chemistry 32: 1326–1331. Wells, M.J.M & Clark, C.R. 1981. Liquid chromatograph- ic elution characteristics of some solutes used to measure column void volume on C18 bonded phas- es. Analytical Chemistry 53: 1341–1345. Welsch, T., Schmidt, M., Kutter, J. & Kalman, A. 1996. Temperature of the eluent: A neglated tool in high- performance liquid chromatography? Journal of Chro- matography, A 728: 299–306. Wendt, G. & Bernhard, F.W. 1960. The structure of a sul- fur-containing compound with vitamin B6 Activity. Archives of Biochemistry and Biophysics 88: 270– 272. Wielen, R.P.J. van der, Löwik, M.R.H., Haller, J., Berg, H. van den, Ferry, M. & Staveren, W.A. van. 1996. Vitamin B-6 malnutrition among elderly Europeans – the SENECA Study. Journal of Gerontology – Series A – Biol. Sci. & Med. Sci. 51, 6: B417–424. Wilcken, D.E. & Gupta, V.J. 1979. Sulphur containing amino acids in chronic renal failure with particular reference to homocystine and cysteine-homocysteine mixed disulphide. European Journal of Clinical In- vestigation 9: 301–307. Wirth, M.J. 1994. Spectroscopic probing of reversed- phase chromatographic retention. LC-GC 7: 626, 628–630. Wong, F. 1978. Analysis of vitamin B6 in extractives of food materials by high-performance liquid chroma- tography. Journal of Agricultural and Food Chemis- try 26: 1444–1446. Yasumoto, K., Tsuji, H., Iwami, K. & Mitsuda, H. 1977. Isolation from rice bran of a bound form of vitamin B6 and its identification as 5’-O-(β-D-glucopyrano- syl) pyridoxine. Agricultural and Biological Chemis- try 41: 1061–1067. Zhang, F., Gregory, J.F. & McCormick, D.B. 1993. Pyri- doxine-5’-β-D-glucoside competitively inhibits uptake of vitamin B-6 into isolated rat liver cells. Journal of Nutrition 123: 95–89. 619 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Vol. 8 (1999): 515–619. SELOSTUS B 6 -vitamiinin määrittäminen elintarvikkeista suuren erotuskyvyn nestekromatografisella (HPLC) menetelmällä Velimatti Ollilainen Helsingin yliopisto B 6 -vitamiini (pyridoksiini) on vesiliukoinen vitamii- ni, joka esiintyy erilaisissa kemiallisissa muodoissa biologisissa materiaaleissa kuten eläinkudoksissa, kasvisolukoissa ja niitä sisältävissä elintarvikkeissa. Elimistössä B 6 -vitamiini edesauttaa monia biokemi- allisia reaktioita mm. valkuaisaineiden ja glykogee- nin aineenvaihdunnassa. Perinteisen vitamiinivaiku- tuksen lisäksi pyridoksiini saattaa liittyä mm. altis- tumiseen sydän- ja verisuonitaudeille. Tieteellistä näyttöä näistä uusista yhteyksistä ei kuitenkaan vie- lä ole saatu. Perinteisesti B 6 -vitamiini on määritetty elintarvik- keista mikrobiologisesti, jolloin pyridoksiini on va- pautettu ns. sidotuista muodoista. Tässä työssä sel- vitettiin nestekromatografian (HPLC:n) soveltuvuutta B 6 -vitamiinin vapaiden ja sidottujen muotojen mää- rittämiseen elintarvikkeista. Kehitetyn menetelmän luotettavuus varmistettiin mm. Euroopan unionin jär- jestämissä laboratorioiden välisissä vertailututkimuk- sissa. Menetelmällä määritettiin 50 elintarvikenimik- keen eri B 6 -vitamiinimuotojen pitoisuudet. Runsaas- ti pyridoksiinia tai sen johdannaisia sisälsivät mm. liha ja sisäelimet (maksa ja munuaiset), kananmunan keltuainen ja hasselpähkinä. Kasviperäisten elintar- vikkeiden B 6 -vitamiini koostui suurelta osin sidotuis- ta glykosyloiduista vitamiinimuodoista. Niiden hy- väksikäytettävyydestä elimistössä on varsin ristirii- taisia tietoja. Työssä saadut elintarvikkeiden B 6 -vita- miinitulokset liitetään Kansanterveyslaitoksen elin- tarvikkeiden koostumustietopankkiin ja niitä hyödyn- netään mm. ravitsemussuunnittelussa. Työ osoitti ole- van tarpeellista yhdenmukaistaa käytettäviä määritys- menetelmiä, jotta eri laboratorioissa saadut tulokset olisivat paremmin vertailukelpoisia. Tätä menetelmi- en yhtenäistämistä jatketaan mm. Euroopan Standar- doimisliiton (CEN) työryhmissä. 620 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Appendix 1 (omille sivuilleen Ei numeroida) MEAT AND POULTRY PY-1 Minced beef (steak) PY-2 Minced meat, beef and pork PY-5 Pork shoulder, deboned PY-6 Beef, shoulder chuck, deboned PY-3 Beef liver PY-10 Pork liver PY-11 Pork, kidney PY-4 Beef, kidney PY-7 Broiler, liver (PY-28 -”-) PY-12 Dry sausage, salami type PY-13 Sausage, ”lenkki” PY-15 Lamb, leg PY-16 Elk, steak PY-14 Reindeer, steak PY-9 Broiler, boneless, skinless PY-8 Hen, boneless, skinless CEREALS PY-74 Whole wheat flour (graham) PY-41 Wheat germ PY-40 Rye flour PY-42 Barley flour PY-44 Oats, rolled PY-43 Mixed wheat and rye flour, ”sämpyläjauho” PY-45 Rice FISH AND FISH PRODUCTS PY-75 Rainbow trout, fillets PY-96 Pikeperch, sea, fillets PY-70 Baltic herring, fillets PY-97 Whitefish, sea, fillets PY-95 Pike, sea, fillets PY-76 Roe paste, salted, smoked READY-TO-EAT FOODS PY-30 Meatballs PY-31 Broiler meat balls PY-32 Veal-vegetable stew (baby food) PY-33 Potato-carrot puree (baby food) DAIRY PRODUCTS AND EGGS PY-20 Whole milk, standardized 3.9 % fat PY-21 Milk, standardized 1.9 % fat PY-22 Whipping cream PY-25 Cream cheese, ”Turunmaa” PY-23 Cheese, Edam type, 40% PY-24 Cheese, Edam type, 20% PY-26 Skimmed milk powder PY-27 Infant formula powder, based on cow’s milk PY-71 Egg yolk VEGETABLES, POTATO AND NUTS PY-80 Tomato PY-81 Broccoli PY-82 Carrot PY-83 Potato PY-84 Peas, frozen PY-91 Peanuts PY-92 Hazelnuts PY-93 Almonds (PY-94 Almonds) MISCELLANEOUS PY-90 Yeast, bakers FOOD SAMPLES FOR VITAMIN B 6 ANALYSIS 621 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Appendix 2 (omille sivuilleen Ei numeroida) FAB-MS spectra of PNX from carrot using a) glyserol, b) nitrobenzylamine and c) triethanolamine as a matrix. a) b) c) 622 A G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D Appendix 2 (omille sivuilleen Ei numeroida) FAB-MS spectra of PNX from wholemeal flour using a) glyserol, b) nitrobenzylamine and c) tri- ethanolamine as a matrix. a) b) c) Title Preface Contents List of abbreviations Abstract Introduction Literature Review Objectives of the study Materials and methods Results Discussion Conclusions References SELOSTUS