Upsala J Med Sci 102: 175-184, 1997 Image Analysis of the Duodenal Endocrine Cells in Mice with Particular Regard to Optical Densitometry M. El-Salhy and J. Mahdavi Section f o r Gastroenterology and Hepatology, Department of Medicine, University Hospital, Umed, Sweden ABSTRACT The endocrine cells in the murine proximal duodenum have been investigated by means of immunohistochemistry and computerized image analysis. Five endocrine cell types were identified, namely secretin-, gastric inhibitory polypeptide (GIP)-, gastrin-CCK-, somatostatin- and serotonin immunoreactive cells. The number of endocrine cells/mm3 epithelial cells was estimated and the cell secretory index (CSI) for different endocrine cell types was determined. Furthermore, the optical density of the cellular immunoreactivity and the immunoreactive area in the cell were determined and an index, cell immunoreactivity content was estimated as the optical density multiplied by the immunoreactive area. It has been suggested that the use of this index might better reflect the cellular peptide/amine content than does the CSI. Serotonin- immunoreactive cells were the predominant endocrine cell type, followed by gastrin/CCK-immunoreactive cells. The numbers of secretin-, GIP- and somatostatin immunoreactive cells were almost identical. All endocrine cell types were present both in crypts and in villi, but were, more numerous in the crypts, except for secretin which was more frequent in the villi. INTRODUCTION Image analysis first appeared as a readily available technique in 1963 with the introduction of the Quantitative Television Microscope (QTM). QTM was designed by Metals Research Ltd and was intended for use in metallographical 175 laboratories. A n early application in the biological field was the measuring, of the size of air spaces in the damaged lung and counting of silver grains in autometallograph (7, 9,11,12). The rapid developments in computer hardware and software made computer image analysis an easy, rapid, precise and objective tool for determining the number and volume of biological structures (3-5, 8, 10). In computerized image analysis the image from video-cameras is digitized into picture elements (pixels). The grey level (i.e. the brightness) of all the pixels in the image can be measured in terms of light transmitted or reflected, or as an optical density value. In coloured images, measurements are taken on the individual colour components within the image. Thus, computerized image analysis, in addition to traditional morphometry, makes it possible to measure the greyness or colour levels of a stained biological structure. The mouse is a n excellent experimental animal. The endocrine cells in the gut of this animal have not previously been subjected to detailed investigation. Apart from somatostatin ( 6 ) and peptide Y Y (PYY) (1) in the small intestine, no other endocrine cell types have been examined in the gut of normal mouse. The aim of the present study was to quantify duodenal endocrine cells in mice and to use the opportunity offered by computerized image analysis, using optical densitometry to determine an index reflecting the cellular peptidelamine content. Material and Methods Animals Ten mice ( 5 females and 5 males) aged 3 months and with an average body weight of 30 g (Bom: NMRI strain, B/S Bomholtgird Breeding and Research Centre, Denmark) were used. The mice were housed in cages, 5 mice in each cage, in a room with artificial light from 6 am to 6 pm. The animals were fed a standard pellet diet (Astra-Ewos AB, Sodertdje, Sweden) and allowed water ad 176 libitum. They were kept in the laboratory for 2 weeks prior to sacrifice in a C02 chamber, after an overnight fast. The investigation was approved by the local committee on animal ethics, UmeA University. Immunohistochemistry After sacrificing the animals, the proximal part of the duodenum was removed and fixed overnight in 4% buffered formaldehyde and embedded in paraffin wax. The deparaffinized sections were rehydrated, and immersed in 0.01 % hydrogen peroxide in Tris-HC1 buffer, pH 7.6 for 10 min in order to inhibit the endogenous peroxidase activity. To occupy the non-specific binding sites the sections were pre-treated with 1 % bovine serum albumin for 10 min. The sections were immunostained by the avidin-biotin complex (ABC) method (Dako A/S,Glostrup,Denmark)as described in detail previously (4). Briefly, The sections were incubated for 20 h with one of the primary antisera (for details concerning the antisera used, see Table 1). They were incubated with biotinylated swine anti-rabbit IgG, diluted 1:200 for l h . The sections were then incubated for 1 h with avidin-biotin-peroxidase complex, diluted 1:lOO. Staining of the sections was performed in 50 ml Tris buffer containing 10 pl of 30% H202 and 25 mg diaminobenzidine tetrahydrochloride (DAB), for 15 min. The same ABC-kit and biotinylated swine anti-rabbit IgG was used throughout the study. The incubation was performed at 4°C. Equal parts of avidin and biotinylated peroxidase were mixed with the Tris-buffer and allowed to stand for 1 h at 4°C. Specificity controls were the same as those described earlier (4). Briefly, the primary antibodies were replaced by non-immune rabbit serum, and the primary antisera pre-incubated for 24 h with the corresponding or structurally related peptide (50-75 pg/ml diluted antibody) at 4°C. Computerized image analysis The Quantimet 500MC image processing and analysis system (Leica, Cambridge, England) linked to an Olympus microscope type BX50 was used. The software used in this system was the Leica Windows-based image analysis tool kit "QWIN" (version 1.02) and an interactive programming system Quips (version 1.02). Table 1. Details of the antisera used. All antisera were raised in rabbit. *Specific for CCWgastrin C-terminus. **Specific for glucagon N-terminus and cross-reacts with both pancreatic and enteroglucagon. Antibody against Source Working Code no Porcine secretin Eurodiagnostica, Malmo, 1: 1000 R-787502 dilution Sweden Bovine gastric Eurodiagnostica 1:1600 R-786403- inhibitory peptide B2 ( G I R Synthetic human Eurodiagnos tica 1 : 10,000 R-783511 gastrin-17 * Synthetic human Dakopatts, Glostrup, 1: 1600 A566 somatostatin Denmark Porcine pancreatic Eurodiagnostica 1 : 2500 B-3 1 glucagon** Porcine motilin Eurodiagnostica 1 : 5000 R-842206 Serotonin Eurodiagnostica 1: 600 B 45-1 178 The number of endocrine cells and their secretory index were determined as described in detail previously (2)using an automated standard sequence analysis operation. Measurements were made on 40 randomly chosen fields (20 from the crypts and 20 from the villi) for each peptide and animal. These fields were chosen from three to five sections, at least 50 pm apart. The sections were examined with X20 objective and each field observed in the monitor represented 0.034 mm2 area of the tissue. In Quantimet 500MC the optical density is sampled and coded as a grey level value in the range 0 to 255. Grey level measurements can be calibrated to known standards and Quantimet 500MC utilizes a "look-up table" in which each grey level from 0 to 255 is related to its calibrated value. Using QUIPS, an automated standard sequence operation was created to measure the optical density in different endocrine cell types. The microscope illumination intensity was constant. The operating parameters of the CCD-video camera were kept constant during all the measurements performed in this study. Ten nucleated cells were randomly chosen for each endocrine cell type and animal. Quantification was undertaken with a X40 objective and each pixel corresponded to 0.206 pm. In this automated analysis process, the image was converted to a monochrome image (black and white) and the operator calibrated the grey level so that the section background staining was regarded zero. This was done in every section. The optical density was measured in the cell cytoplasm as mean grey, being equal to the sum of the optical density divided by the number of pixels measured, where sum of optical density is the sum of the calibrated optical density values for each pixel within the measured object. In addition to the mean grey, the immunoreactive area was measured in each cell. The cell immunoreactivity content (CIC) was calculated as the mean grey for the cell, multiplied by the immunoreactive area in pm2. The data from each field were tabulated, computed and statistically analysed automatically. 179 Fig. 1. Secretin-immunoreactive cells in the villi (A), and gastridCCK- immunoreactive cells (B) in the crypts of proximal duodenum of a mouse. x350. 180 Immunohistochernistry Secretin- (Fig. l a ) , gastric inhibitory polypeptide (G1P)-, gastridCCK- (Fig. I b), somatostatin- and serotonin-immunoreactive cells were found both in the crypts and in villi, though they were more numerous in the crypts-except for secretin-immunoreactive cells, which occurred mostly in the villi. No glucagon- or motilin-immunoreactive cells couId be detected in the examined proximal part of the mouse duodenum. The endocrine cells varied in shape, from flask- to basket-shaped and somatostatin-immunoreactive cells often showed a basal process. There was no immunostaining when the primary antibody was replaced by non-immune rabbit serum or when the primary antisera were pre-incubated with the corresponding antigen. Pre-incubation of the primary antibody with a structurally related antigen had no effect on the immunostaining. Computerized image analysis The results of image analysis of different endocrine cell types are summarized in Table 2. It seems that the most abundant cell is serotonin-immunoreactive cell type followed by gastridCCK-immunoreactive cells. The numbers of secretin- .GIP- and somatostatin-immunoreactive cells are almost the same. DISCUSSION In a previous report (2) the cell secretory index (CSI) was suggested a s a parameter reflecting the endocrine cell peptide/amine content, which is a summation of cellular peptide/ amine synthesis and release. CSI represents volume of the cellular immunoreactive secretory granules. T h e cellular peptidelamine content is not only reflected by the immunoreactive secretory granule volume, but also by the intensity of immunostaining. In the present 181 Table 2. The cells, the cell content (CIC) number of various endocrine cell types per mm3 of epithelial secretory index (CSI) mm3 per cell and cell immunoreactivity in the duodenum of mice (mean+SE) Endocrine cell type Number of CSI CIC cells Crypts 3060~498 81.Lk_11.5 Secretin Villi 5078~3 14 16 1.693 1.8 Total 4036k284 84.827.7 5334.8k3002.6 Crypts 691 12443 77.8k4.6 GIP Villi 53 1 e 4 2 8 71.324.2 Total 61 112343 75. e 3 . 3 2294.52 1429.4 Crypts 10503~782 11 1 . e 7 . 3 GastrirdCCK Villi 9481~638 130.828. 5 Total 9979~427 12125.8 1924.221633.5 Crypts 64602264 83.3k6.1 Somatostatin Villi 5000k353 80.8k4.7 Total 5737~178 83.3~5.1 6441.3k3336.3 Crypts 44626k3075 142.522.3 Serotonin Villi 232462769 151.6k7.6 Total 3480 1 ~ 1 9 0 1 142.223.5 1447.2~108.6 study the CIC was suggested to be a more accurate parameter, reflecting the cellular peptide/amine content. In this parameter both the area and intensity of immunostaining are taken into account. It is worth noting, however, that the optical density measured here is a relative value, being related to the background staining of the section as zero. There are several possible sources of error in measuring the optical density of immunostaining: those caused by 182 immunostaining and those caused by the equipment used. Use of different dilutions of the first, second and third layers, or with DAB, as well as using different incubation times or temperatures, would result in different staining intensities. Furthermore, amplification, when ABC method is used, can vary if different kits, different biotinylated IgGs, or different incubation times with avidin and biotinylated peroxidase are used. Moreover, the optical density measurements are affected by the image brightness caused by the illumination of the microscope and by the operating parameters of the CCD-video camera. Being aware of all these pitfalls and by standardizing all these factors, optical density measurements and CIC can be useful in studies on endocrine cells, where comparison between the cell secretory contents in different groups is required. The present study revealed five endocrine cell types in the proximal duodenum of the mouse. These cell types are similar in appearance and distribution to those described in the human duodenum (4,13). The finding that somatostatin- immunoreactive cells were more frequent in the crypts disagrees with an earlier report that they occurred mostly in the villi (6). ACKNOWLEDGEMENT This study was supported by grants from the Medical Faculty, Umeg University and the Familial Amyloidosis Association (FAMY). REFERENCES 1. El-Salhy, M.& Lundqvist, M.: Immunocytochemical investigation of polypeptide Y Y (PYY) and pancreatic polypeptide (PP) in the gastrointestinal tract of rodents. Biomed Res 5: 401-404, 1984. 2. El-Salhy, M., Sandstrom, O., Nasstrom, E., Mustajbasic, M.& Zachrisson, S.: Application of computer image analysis in endocnne cell quantification. Histochem J 29: 2 49-256, 1977 3 . El-Salhy, M.& Suhr, 0.: Endocrine cells in rectal biopsy specimens from patients with familial amyloidotic plyneuropathy. S a n d J Gastroenterol 3 1:68-73, 1996. 4. El-Salhy, M., Suhr, O., Stenling, R., Wilander, E., Grimelius, L.: Impact of familial associated polyneuropathy on duodenal endocrine cells. Gut 35: 1413- 1418, 1994. 5. El-Salhy, M., Danielsson, A . , Stenling, R.& Grimelius, L.: Colonic endocrine cells in inflammatory bowel disease. J Intern Med 342: 413-419, 1977. 183 6. Evans, G.S.: The distribution of endocrine cells along mouse small intestine. Bombesin v d somatostatin producing cells. Virchows Arch B Cell Pathol Incl Mol Pathol 58: 165-72, 1989. 7. Gonzalez, R.C. & Wintz, P.: (1987). Digital image processing. 2nd edn. Addison-Wesley, Reading. 1987 8. Kubota, Y . , Peters, R.E., Ottaway, C.A., Tubbs R.R., Farmer, R.G. & Fiocchi, C.: Colonic vasoactive intestinal peptide nerves in inflammatory bowel disease. Gastroenterology 102: 1242-1251, 1992. 9. Rosenfeld, A. & Kak, A.: Digital picture processing. 2nd edn.Academic Press,New York, 1982. 10. Ronnblom, A., Danielsson, A . , El-Salhy, M.: Intestinal endocrine cells in myotonic dystrophy: An immunocytochemical and computed image analytical study. (Submitted for publication) 11. Russ, J.C.: Computer assisted microscopy: the measurement and analysis of images. Plenum Pub1 Co, New York, 1990. 12. Serra ,J. (1984) Image analysis and mathematical morphology,vol. 1. Academic Press, New York, 1984. 13. Sjolund, K., Sandtn, G., H s a n s o n , R. & Sundler, F.: Endocrine cells in human intestine. An immunocytochemical study. Gastroenterology 85: 1120- 1130, 1983. Correspondence to: Magdy El-Salhy, MD, PhD, Section for Gastroenterology and Hepatology, Department of Medicine, University Hospital, S-901 85 Urn&, Sweden. Tel: +46-90-7853867 Fax: +46-90-120923. E-mail: magdy.el-salhy @ medicin.umu.se 184