CHEMICAL ENGINEERING TRANSACTIONS VOL. 57, 2017 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Sauro Pierucci, Jiří Jaromír Klemeš, Laura Piazza, Serafim Bakalis Copyright © 2017, AIDIC Servizi S.r.l. ISBN 978-88-95608- 48-8; ISSN 2283-9216 Modelling and Optimization of Poly-Aromatic-Hydrocarbons Biodegradation by Bulab 5738 Claudia Prasciolua, Valentina Perraa, Francesco Desogusa, Stefania Troncia, Nicoletta Currelib, Giuliano Saiua, Massimiliano Grossoa* aDipartimento di Ingegneria Meccanica, Chimica e dei Materiali, Università degli Studi di Cagliari, Via Marengo 3, Cagliari, Italy bDipartimento di Scienze Biomediche, Unità di Biochimica, Università degli Studi di Cagliari, Cittadella Universitaria, Monserrato, Cagliari, Italy massimiliano.grosso@dimcm.unica.it In this work, the bacterial consortium Bulab 5738 was used to simultaneously remove pyrene, phenanthrene and catechol from aqueous solutions. The bacterial population growth was estimated by means of optical density measurements while HPLC was used to quantify the pollutant concentration in the solution. The obtained data were used to model the systems, in term of biomass population and substrate concentration. The effects of pollutant concentration values were analysed using the outputs of a full factorial experimental design. 1. Introduction Polycyclic aromatic hydrocarbons (PAH) are highly toxic pollutants, with potential harmful effects on human health and environment, and difficult to remove by traditional treatments (Carta and Desogus, 2013). Bioremediation represents an environmentally sustainable method for treating polluted sites, although its application is still constrained by factors such as the unpredictable endpoint PAH concentrations and the lack of adequate monitoring tools to ensure active biodegradation processes (Vila et al., 2015). Bacteria and fungi showed to be able to efficiently remove PAHs, as reported in different studies on bioremediation (e.g., Palanisamy et al., 2014; Khan et al., 2013; Lu et al., 2011; Tronci and Spigno, 2015; Lee et al., 2015; Saiu et al., 2015, 2016), but their ability is negatively affected as the number of aromatic rings increases or when different organic compounds are present in the polluted soil. Recent studies showed that bacteria consortia may improve PAH degradation, because the presence of different microorganisms may result in enhanced PAH utilization, since metabolic intermediates produced by some organisms may work as substrates for the others (Mrozik, 2003; Zhong et al., 2011; Janbandhu and Fulekar, 2011). This study is focused on the evaluation of the bioremediation capability of the bacterial consortium provided by Buckman Laboratories International and named Bulab 5738. In more detail, the goal is to understand the consortium behaviour when the only source of carbon is represented by pyrene, phenanthrene and catechol. The presence of a surfactant (Tween 80) has been also considered for determining the process conditions leading to the best results in term of degradation rate and pollutant removal. The experimental conditions were selected using a full factorial experimental design, considering two levels for the compounds concentration values. The obtained experimental results were processed by statistical tools to rigorously reveal the impact of the pollutants on the bacterial population growth and on their removal efficiency. 2. Methods 2.1 Chemicals and culture medium Phenanthrene, pyrene and catechol (all >99% purity) used in degradation experiments were purchased from Sigma Aldrich. Tween 80, a non ionic surfactant, was supplied by Fluka. The mineral medium contains the DOI: 10.3303/CET1757057 Please cite this article as: Prasciolu C., Perra V., Desogus F., Tronci S., Currelli N., Saiu G., Grosso M., 2017, Modelling and optimization of poly-aromatic-hydrocarbons biodegradation by bulab 5738, Chemical Engineering Transactions, 57, 337-342 DOI: 10.3303/CET1757057 337 following salts: NaCl, MgSO4, (NH4)2SO4, FeCl3, CaSO4, Na2HPO4, KH2PO4. All the salts were added in large excess to supply the needs of nutrients for the microorganisms growth and to be a not limiting factor for it. 2.2 Microorganisms The microorganisms used in this study were the mixed bacterial culture Bulab 5738 containing many different species of bacterial strains including Rhodococcus, Aeromonas, Streptomyces, Bacillus, Pseudomonas. The mixed bacterial culture appears as a light brown granular powder containing lyophilized microorganisms on a substrate of cereal bran and freeze dried enzymes. 2.3 Analytical methods The bacterial growth was determined by UV–visible spectrophotometer (JENWAY 7300) at 600 nm by measuring the absorbance of the cell suspensions, as already done by Carta and Desogus (2010) and Carta et al. (2006). Samples of the solution containing the pollutant were taken at the beginning and at the end of the bioremediation experiments in order to assess whether the chemicals had actually been degraded or not. Aqueous pollutants were quantified using high-performance liquid cromatography. The samples were only filtered using 0.20 µm filter before the analysis. The EPA 8310 method was used with the following equipment: Agilent 1200 Infinity LC system, column Zorbax Eclipse plus C18 (4.6·100mm, 3.5 µm particle size). Elution was performed using a two solvents gradient (A: water – B: acetonitrile): the elution started with 40% of B followed by a linear gradient to 95% of B. The injection volume was 5 µl, and detection was performed through DAD at 230, 240, 254, 270, 350 nm with 400 nm reference wavelength and spectrum acquisition. 3. Experimental design The capability of the bacterial consortium to degrade the pollutants was monitored by jointly varying the concentrations of pyrene (PYR), phenanthrene (PHE) and catechol (CAT), which range from 5 to 15 mg/l. Each experimental run was replicated twice, leading to 16 experiments carried on in 250 ml Erlenmeyer flasks. The factorial design 23 is reported in Table 1. Table 1: Factorial design Run PYR (mg/l) PHE (mg/l) CAT (mg/l) 1 5 5 5 2 15 5 5 3 5 15 5 4 15 15 5 5 5 5 15 6 15 5 15 7 5 15 15 8 15 15 15 Two different models were considered to describe the growth of the bacterial population in terms of absorbance (A) measured at different sampling time: i) a Gompertz growth model reported in Eq. 1 and ii) a zero-order kinetic model reported in Eq. 2 𝑑𝐴 𝑑𝑡 = 𝜇1𝐴(𝑙𝑜𝑔𝐴∞ − 𝑙𝑜𝑔𝐴) 𝐴(𝑡0) = 𝐴0 𝑡0 ≤ 𝜆 (1) 𝑑𝐴 𝑑𝑡 = 𝜇2 (2) In Eq. 1, the parameter  is the lag time, which is the interval of time where the growth rate is almost null. The statistical analysis was then performed by using the estimated parameters 1 and 2 as outputs of the experiments. 338 4. Results The absorbance values registered at the different sampling time are reported in the left panel of Figure 1, in the understanding that they represent an indirect measurement of the microbial population. For sake of clarity, only one replicate is shown. Figure 1: Absorbance values of the samples collected during the experimental runs in Table 1 (left panel) and comparison between model estimation (continuous line) and experimental data B05-I (blach diamond) along with the confidence interval (right panel). It is r evident a diauxic growth behaviour that can be explained by considering the kind of molecules present in the solution. First, bacteria are reasonably able to degrade the simplest compound, which is catechol. Then, they can metabolize phenanthrene and pyrene, which are the most refractory compounds in the solution. The Gompertz model (Eq. 1) showed to quantitatively describe the first growth step, with an average value for the Mean Square Error (MSE) scalar equal to 1.71·10-5, with values ranging from MSEmin=2.64·10 -6 to MSEmax=3.91·10 -5, whereas the zero-order kinetics (Eq. 2) demonstrated to be more adequate for fitting the final experimental data (average value for the MSE scalar equal to 7.87·10-3, with values ranging from MSEmin=1.62·10 -4 to MSEmax=1.61·10 -2). The parameter estimation was carried out by means of the Least Squares Method (Levenberg-Marquardt algorithm) and the obtained values are reported in Table 2. As illustrative example, the comparison between the model estimation and the experimental data for the run B05-I is shown in the right panel of Figure 1. Table 2: Parameters estimated for the Gompertz model ( and 2) and for the zero-order kinetics (2) Run λ [h] μ1 [h -1] μ2 [h -1] B01-I 3.93775 0.07294 0.00000 B02-I 4.01294 0.06252 0.00639 B03-I 4.16928 0.05638 0.00758 B04-I 5.34434 0.05069 0.00839 B05-I 4.11486 0.07925 0.00000 B06-I 4.37582 0.06498 0.00675 B07-I 4.39583 0.05902 0.00754 B08-I 5.28789 0.06260 0.00927 B01-II 3.88736 0.07385 0.00000 B02-II 3.97566 0.06245 0.00652 B03-II 4.30350 0.05752 0.00692 B04-II 5.30830 0.05245 0.00795 B05-II 4.03412 0.07863 0.00345 B06-II 4.49460 0.08017 0.01167 B07-II 4.40283 0.06106 0.00815 B08-II 5.22947 0.06108 0.00825 A statistical analysis on  and  estimated for the different experimental runs can give information on the effects of the process conditions on the biological system. An ANOVA test was performed (Montgomery, 2013) and the related results are reported on Table 3 and Table 4 for the parameters 1 and 2, respectively. 35302520151050 1.0 0.8 0.6 0.4 0.2 0.0 Time (h) O p ti c a l d e n s it y ( 6 0 0 n m ) B01-I B02-I B03-I B04-I B05-I B06-I B07-I B08-I 35302520151050 1.0 0.8 0.6 0.4 0.2 0.0 Tempo O p ti c a l d e n s ti ty ( 6 0 0 n m ) - B 0 5 -I Regression 95% C I Time (h) 339 In more detail, for each source of variation (reported on the column 1), the table shows the corresponding degrees of freedom (DoF), the sum of squares (SS) of the source of variation, the adjusted Mean Square (MS) error, the F-ratio statistics and the p-value associated to that source of variation. The effect contribution is considered significant when p< 0.05. The significant factors (i.e. with a p-value <0.05) are put in bold type. The main outcomes of the analysis are listed below: i. all the compounds are statistically significant for 1 whereas the higher order interactions are not; ii. pyrene and phenanthrene have an impact on 2 along with their interaction. The impact of catechol is negligible, thus confirming that it has mainly consumed during the first phase of the experiments. Table 3. Analysis of variance for the parameter 1 Source DoF SS MS F-ratio p-value PYR 1 0.0001087 0.0001087 7.16 0.028 PHE 1 0.0008121 0.0008121 53.51 0.000 CAT 1 0.0002101 0.0002101 13.85 0.006 PYR-PHE 1 0.0000468 0.0000468 3.09 0.117 PYR-CAT 1 0.0000343 0.0000343 2.26 0.171 PHE-CAT 1 0.0000013 0.0000013 0.09 0.778 PYR-PHE-CAT 1 0.0000017 0.0000017 0.11 0.744 Residual error 8 0.0001214 0.0000152 Pure error 8 0.0001214 0.0000152 Total 15 0.0013367 Table 4. Analysis of variance for the parameter 2 Source DoF SS MS F-ratio p-value PYR 1 0.0000622 0.0000622 26.09 0.001 PHE 1 0.0000535 0.0000535 22.46 0.001 CAT 1 0.0000080 0.0000080 3.36 0.104 PYR-PHE 1 0.0000367 0.0000367 15.38 0.004 PYR-CAT 1 0.0000003 0.0000003 0.11 0.749 PHE-CAT 1 0.0000027 0.0000027 1.14 0.318 PYR-PHE-CAT 1 0.0000003 0.0000003 0.11 0.748 Residual error 8 0.0000191 0.0000024 Pure error 8 0.0000191 0.0000024 Total 15 0.0001826 The dependence of the parameters 1 and 2 on the process conditions can be described by the relationships reported in Eq. (3) and (4), respectively, where only the significant effects selected by the ANOVA are taken into account. 𝜇1 = 0.06472 − 0.002607 ∙ 𝑐 𝑃𝑌𝑅 − 0.007124 ∙ 𝑐𝑃𝐻𝐸 + 0.003624 ∙ 𝑐𝐶𝐴𝑇 (3) 𝜇2 = 0.0062 + 0.0020 ∙ 𝑐 𝑃𝑌𝑅 + 0.0018 ∙ 𝑐𝑃𝐻𝐸 − 0.0015 ∙ 𝑐𝑃𝐻𝐸 ∙ 𝑐𝑃𝑌𝑅 (4) In Eqs. (3) and (4), ci represents the initial concentration of the compound. It can be easily seen that pyrene and phenanthrene strongly reduce the population growth rate 1, whereas catechol has a positive effect. On the other hand, during the second phase represented by the parameter 2, the two PAHs increases the population growth rate, with a synergic effect shown by the positive sign of the coefficient related to the interaction term. This behavior indicates that bacteria are able to adapt to pyrene and phenanthrene, which initially had an inhibitory effect on the population growth, and they become able to use PAH as carbon source. The results are also confirmed by the Pareto chart calculated using the Yates algorithm (Morgan, 1991) and reported in Figure 2.a and 3.a for 1 and 2, respectively. For sake of clarity, also the significance threshold value (with a significance level =0.05) is reported with the solid line. It was confirmed that all the pollutants affect the first phase of bacteria population growth, whereas the second phase depends on the concentration of PAHs and on their interaction. The main effects are also shown in Figure 2.b and 3.b. 340 Figure 2: a) Pareto chart of standardized effects for the  parameter; b) Main effect behaviour. Figure 3: a) Pareto chart of standardized effects for the  parameter; b) Main effect behaviour. The solutions were also analysed by HPLC to measure the pollutant removal efficiency (REp) in Eq. (5) 𝑅𝐸𝑝 = 𝐶𝑝 𝑖 −𝐶𝑝 𝑓 𝐶𝑝 𝑖 (5) where the superscripts i and f indicate respectively initial and final concentration for the pollutant p. The results are reported in Table 5, where the data regarding catechol confirm that it is easily degraded by the consortium for each investigated condition. The higher removal efficiency of phenanthrene (83.1%) is observed for the second run (02-II), which corresponds to the lowest PAHs initial concentration (5 g/l) and the highest catechol initial concetration (15 g/l). On the other hand, the minimum value for REPHE (38.8%) is obtained when the initial concentration of all the compounds is equal to 5 g/l (Runs 01-I,II). Also pyrene showed a similar behaviour, as it presented the higher removal efficiency (75.7%) when the initial concentration of the two PAHs was at the lowest level (5 mg/l) and that of catechol was maximum (15 mg/l) (Run 02-I), whereas the lowest removal (efficiency of 18.3%), like for phenanthrene, occurred during Run 01-I, when all the components started from the lowest concentration (5 mg/l). In all the experiments the average value of the removal efficiency for both pyrene and phenanthrene was about 55%. Table 5. Removal efficiency (percentage) for the components in all the experimental runs. Run RECAT REPHE REPYR Run RECAT REPHE REPYR 01-I 100.0 38.6 27.6 01-II 100.0 40.1 18.3 02-I 100.0 76.6 75.7 02-II 100.0 83.1 73.8 03-I 100.0 49.2 53.2 03-II 100.0 44.6 55.3 04-I 100.0 52.5 55.3 04-II 100.0 44.8 62.6 05-I 100.0 70.9 47.5 05-II 100.0 69.5 39.9 06-I 100.0 40.4 74.8 06-II 100.0 48.3 75.4 07-I 100.0 60.0 55.6 07-II 100.0 52.8 56.7 08-I 100.0 55.1 61.4 08-II 100.0 49.0 61.2 BC ABC AC AB A C B 876543210 T e r m Standardized Effect 2.306 A P y rene B P henanthrene C C atechol F actor N ame 155 0.072 0.068 0.064 0.060 155 155 0.072 0.068 0.064 0.060 Pyrene M e a n Phenanthrene Catechol AC ABC BC C AB B A 543210 T e r m Standardized Effect 2.306 A P y rene B P henanthrene C C atechol F actor N ame 155 0.008 0.007 0.006 0.005 0.004 155 155 0.008 0.007 0.006 0.005 0.004 Pyrene M e a n Phenanthrene Catechol 341 5. Conclusions In this paper it is presented and discussed the degradation ability of the bacterial consortium Bulab 5378 towards phenanthrene and pyrene, aided by the presence of catechol, in terms of the biomass growth rate and of the removal efficiency for each compound. The diauxic behaviour of the growing bacterial populations probably indicates two distinct phases: at first, they degradate the less stable molecular structure (catechol); after this, the microorganisms have become adapted to the more recalcitrant species phenanthrene and pyrene, which are finally (but not completely) degraded. The statistical analysis of the experimental data shows that all the components influence the bacterial growth rate: to be precise, catechol has a positive effect, whereas phenanthrene and pyrene exercise a negative one, decreasing the growth rate, when catechol is still present in the suspension; after this, also the two PAHs (singularly and together in a synergistic way) give a positive contribute to the development of the bacterial populations. The chemical analyses on the solutions always showed an about complete degradation of catechol, whereas phenanthrene and pyrene were consumed with an average removal efficiency of 55%, ranging from 38.6 to 83.1% for phenanthrene and from 18.3 to 75.7% for pyrene, depending on their initial concentration and on the relative abundance of catechol, which increases the degradation rate for the two PAHs in the second growth phase. Reference Carta R., Desogus F., 2010, The effect of low-power microwaves on the growth of bacterial populations in a plug flow reactor, AIChE J. 56, 1270-1278. Carta R., Desogus F., 2013, The enhancing effect of low power microwaves on phenol oxidation by the Fenton process, J. Environ. Chem. Eng. 1, 1292-1300. Carta R., Desogus F., Errico M., 2006, Effect of microwave radiation on the growth rate of Bacillus clausii at 37°C, CHISA 2006 – Proceedings of the 17th International Congress of Chemical and Process Engineering, August 27-31 2006, Prague, Czech Republic, Paper ID: 70283. Janbandhu A., Fulekar M.H., 2011, Biodegradation of phenanthrene using adaptedmicrobial consortium isolated from petrochemical contaminated environment, J. Hazard. Mater. 187, 333–340. Khan S., Afzal M., Iqbal S., Khan Q.M., 2013, Plant-bacteria partnerships for the remediation of hydrocarbon contaminated soils, Chemosphere 90, 1317–1332. Lee H., Yun S.Y., Jang S., Kim G.H., Kim J.J., 2015, Bioremediation of polycyclic aromatic hydrocarbons in creosote-contaminated soil by Peniophora incarnata KUC8836, Bioremediat. J. 19(1), 1-8. Lu X.Y., Zhang T., Fang H.H.P., 2011, Bacteria-mediated PAH degradation in soil and sediment, Appl. Microbiol. Biotechnol. 89, 1357–1371. Palanisamy N., Ramya J., Kumar S., Vasanthi N., Chandran P., Khan S., 2014, Diesel biodegradation capacities of indigenous bacterial species isolated from diesel contaminated soil, J. Environ. Health Sci. Eng. 12, 142-147. Montgomery D.C., 2013, Design and Analysis of Experiments, John Wiley & Sons Inc., Hoboken, NJ. Morgan E., 1991, Chemometrics, Open Learning in Experimental Design, Wiley, New York, USA. Mrozic A., 2003, Bacterial degradation and bioremediation of polycyclic aromatic hydrocarbons. Polish J. Environ. Studies 12, 15–25. Saiu G., Poggi F., Tronci S., Grosso, M., Lallai A., Cadoni E., Curreli N., 2015, Detection of parameters enhancing the performance of white-rot fungi for degradation of Poly-Aromatic Hydrocarbons through design-of-experiment methodologies, Chemical Engineering Transactions 43, 271-276, DOI: 10.3303/CET1543046 Saiu G., Tronci S., Grosso M., Cadoni E., Curreli N., 2016, Biodegradation of polycyclic aromatic hydrocarbons by pleurotus sajor-caju, Chemical Engineering Transactions 49, 487-492, DOI: 10.3303/CET1649082 Tronci S., Spigno G., 2015, Development of hybrid models for a vapor-phase fungi bioreactor, Math. Probl. Eng., Article ID 801213, 2015,11, DOI: 10.1155/2015/801213 Vila J., Tauler M., Grifoll M., 2015, Bacterial PAH degradation in marine and terrestrial habitats, Curr. Opin. Biotechnol. 33, 95-102. Zhong Y., Luan T., Lin L., Liu H., Tam N.F.Y., 2011, Production of metabolites in the biodegradation of phenanthrene fluoranthene and pyrene by the mixed culture of Mycobacterium sp. and Sphingomonas sp., Bioresour. Technol. 102, 2965–2972. 342