- Email: [email protected]
Effects of temperature and pH on the biokinetic properties of thiocyanate biodegradation under autotrophic conditions
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effects of temperature and pH on the biokinetic properties of thiocyanate biodegradation under autotrophic conditions Jaai Kim a, Kyung-Jin Cho b, Gyuseong Han b, Changsoo Lee a,*, Seokhwan Hwang b,** a
School of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 689-798, Republic of Korea b School of Environmental Science and Engineering, POSTECH, Pohang, Gyungbuk 790-784, Republic of Korea
article info
abstract
Article history:
The simultaneous effects of temperature and pH on the biokinetic properties of thiocya-
Received 11 July 2012
nate biodegradation under mixed-culture, autotrophic conditions were investigated using
Received in revised form
response surface analysis (RSA) combined with biokinetic modeling. A partial cubic model,
28 September 2012
based on substrate inhibition biokinetics, was constructed for each kinetic coefficient in
Accepted 3 October 2012
Andrew model (i.e., maximum specific growth rate (mm), saturation coefficient (KS), and
Available online 22 October 2012
substrate inhibition coefficient (KSI)). Each model proved statistically reliable to approximate the responses of the kinetic coefficients to temperature and pH changes (r2 > 0.8,
Keywords:
p < 0.05). The response surface plots demonstrated that the biokinetic coefficients change
Biokinetic modeling
with respect to temperature and pH significantly and in different ways. The model
pH
response surfaces were substantially different to each other, indicating distinct correla-
Response surface analysis
tions between the independent (temperature and pH) and dependent (model response)
Temperature
variables in the models. Based on the estimated response surface models, temperature
Thiocyanate biodegradation
was shown to have significant effects on all biokinetic coefficients tested. A dominant influence of temperature on mm response was observed while the interdependence of temperature and pH was apparent in the KS and KSI models. Specific growth rate (m) versus substrate (i.e., thiocyanate) concentration plots simulating using the obtained response surface models confirmed the significant effects of temperature and pH on the microbial growth rate and therefore on the thiocyanate degradation rate. Overall, the response surface models able to describe the biokinetic effects of temperature and pH on thiocyanate biodegradation within the explored region (20e30 C and pH 6.0e9.0) were successfully constructed and validated, providing fundamental information for better process control in thiocyanate treatment. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Thiocyanate (SCN), a hazardous compound, is often found in industrial waste streams as it is used or generated in various industries including photofinishing, dyeing, coking, metal
separation, and electroplating (Ahn et al., 2004). Thiocyanate is toxic at low concentrations (1e2 mM) to higher animals including humans. It binds to proteins and inhibits enzymatic reactions (Wood et al., 1998), leading to damage to human central nervous system and thus severe clinical problems
* Corresponding author. Tel.: þ82 52 217 2822; fax: þ82 217 2819. ** Corresponding author. Tel.: þ82 54 279 2282; fax: þ82 54 279 8299. E-mail addresses: [email protected] (C. Lee), [email protected] (S. Hwang). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2012.10.003
252
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
such as nervousness, hallucination, psychosis, and convulsions (Lewis, 1992). Effective removal of thiocyanate is therefore of particular concern when treating wastewaters from the industries mentioned above. Biological treatment of thiocyanate has extensively been studied and several thiocyanate-degrading bacteria, e.g., Arthrobacter, Bacillus, Escherichia, Klebsiella, Methylobacterium, Pseudomonas, and Thiobacillus, were identified from various sources (Ebbs, 2004; Lee et al., 2003). These bacteria degrade and utilize thiocyanate as an energy and/or nutrient source through two distinct metabolic pathways: carbonyl and cyanate pathways (Bezsudnova et al., 2007; Ebbs, 2004). In the former, thiocyanate is first hydrolyzed to ammonia and carbonyl sulfide (COS) by thiocyanate hydrolase, and then COS breaks into H2S and CO2. In the latter, thiocyanate is initially degraded to cyanate (CNO) and H2S, and then CNO is hydrolyzed to ammonia and CO2 by cyanase. The sulfide and ammonia released can be utilized as electron donor, nitrogen source, or sulfur source for microbial growth. Like other biological processes, activities of the microorganisms involved in thiocyanate biodegradation are to be affected by environmental factors. Understanding such influence is helpful in devising methods for enhancing microbial activities and thus process performance. Among many environmental factors, temperature and pH are universally known as the major factors affecting microbial growth (Madigan et al., 2009; Rosso et al., 1995). The temperature and pH effects on microbial activities are well documented in literature (Laidler, 1984; Madigan et al., 2009; Nedwell, 1999; Tan et al., 1998). A few studies have investigated the effects of temperature and pH in biological thiocyanate treatment processes (Kim and Katayama, 2000; Lay-Son and Drakides, 2008; Vazquez et al., 2006). However, previous studies have examined only the responses of processes to such environmental factors but, to our knowledge, not their effects on the biokinetic properties of thiocyanate degrading communities. Biokinetics mathematizes the relationship between microbial growth and substrate consumption using a combination of coefficients, helping describe and predict process performance. From an engineering point of view, biokinetic modeling is a useful tool for understanding the basic mechanism resulting in microbial growth with pollutant removal under a certain condition. It is therefore plain that an understanding of how environmental factors affect the biokinetics of a biological process will provide fundamental information that helps better describe how and why the system changes at the process level. Accumulation of such information will form a basis for better design and operation of biological processes. In this study, we aimed to investigate the effects of temperature and pH on the biokinetic properties (i.e., biokinetic coefficients) of thiocyanate biodegradation under a mixed-culture, autotrophic condition. Taking into account the two factors that vary simultaneously, response surface analysis (RSA), an effective statistical technique for evaluating simultaneous effects of multiple variables (Montgomery, 1997), was employed. RSA was used to design the experimental runs to test and build the response models for biokinetic coefficients. This study provides a biokinetic insight into the biodegradation process of thiocyanate.
2.
Materials and methods
2.1.
Microbial source and culture medium
Activated sludge from a local sewage treatment plant (Pohang, Korea) was cultivated in a continuously stirred tank reactor (CSTR) with a working volume of 7 L to produce consistent inoculum for subsequent experiments. Thiocyanatedegrading microorganisms were enriched using a synthetic wastewater containing the following components in 1 L (Hung and Pavlostathis, 1997): 500 mg SCN, 838 mg KSCN, 500 mg NaHCO3, 385 mg K2HPO4, 129 mg KH2PO4, 50 mg MgSO4$7H2O, 7 mg KCl, 5 mg CaCl2$2H2O, 5 mg FeSO4$7H2O, and 5 mg MnSO4$H2O. The inoculum system was operated at 7.2-day hydraulic retention time (HRT), 30 C, and pH 8.5. After over 10 turnovers of steady-state operation (>99% SCN removal efficiency), the effluent from this system was used as seed inoculum for the microcultivation tests for biokinetic studies.
2.2.
Response surface analysis design
Response surface analysis (RSA) was employed to evaluate the simultaneous effects of temperature and pH on the biodegradation kinetics of thiocyanate with minimum number of experimental trials (Montgomery, 1997). A face-centered design (FCD) with extra trials added to augment the response surface model (Table 1) was applied to evaluate the biokinetic responses to changes in temperature and pH (i.e., independent variables). Based on literature values, the exploring ranges (center point variance) of temperature and pH were determined to be 30 10 C and 7.5 1.5, respectively (Ahn et al., 2004; Hung and Pavlostathis, 1997; Kim and Katayama, 2000; Lee et al., 2003). RSA was performed by a sequential procedure of collecting experimental data from each trial, estimating polynomials, and checking model adequacy (Lee et al., 2011),
Table 1 e Experimental design and observed results of the response surface analysis. Trials Conditions
T ( C)
pH
Observations 1
mm (d )
KSI KS (mg SCN/L) (mg SCN/L)
Original face-centered design 1 20 6.0 0.380 140.9 2 20 9.0 0.326 106.5 3 40 6.0 0.422 143.4 4 40 9.0 0.290 74.9 5 30 6.0 0.428 156.2 6 30 9.0 0.495 147.4 7 20 7.5 0.335 122.8 8 40 7.5 0.412 146.8 30 7.5 0.497 (0.036) 148.1 (18.9) 9a Augmented experimental 10 35 6.0 11 20 7.0 12 35 7.0 13 40 8.5
points 0.471 0.356 0.446 0.348
171.1 134.1 149.6 112.1
197.0 216.1 261.4 461.2 212.1 241.1 244.7 272.2 205.0 (26.3)
206.4 230.1 250.1 356.4
a Center point. Experiment was triplicated and the responses are presented as average values (standard deviation).
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
using Design Expert 7 software (Stat-Ease, Minneapolis, MN). In RSA computation, the independent variables are converted to coded values for computational convenience: the upper limit of a factor (i.e., temperature or pH in this study) is coded as þ1, the center level as 0, and the lower limit as 1. The bestfit response surface model was chosen by fitting increasingly complex polynomials from low to high orders (i.e., linear to partial cubic) to the experimental data. Model adequacy was checked based on the p-values of regression, lack-of-fit, and model coefficients.
2.3.
Evaluation of biokinetic coefficients
An automated growth-monitoring system equipped with a 200-well microcultivation unit (Bioscreen C, Labsystems) was used to produce batch data for the biokinetic characterization of thiocyanate degradation at different pH and temperature conditions (Table 1). Ten different initial thiocyanate concentrations, i.e., 10, 30, 50, 75, 100, 125, 150, 200, 300, and 400 mg/L, were tested, in five replicates each, for each trial condition, requiring 50 cultivations per trial condition. For statistical soundness, the center point (trial 9 in Table 1) was triplicated using three separately prepared sets of 50 cultivations. The microcultivations were performed using the synthetic wastewater fed to the inoculum system with appropriately modified thiocyanate concentrations. Each microcultivation mixture was set up with 390 mL of medium and 10 mL of inoculum (seeding rate, 2.5% v/v). The initial pH and the incubation temperature were set to the corresponding test conditions (Table 1) before the start of cultivation. Microbial growth was automatically measured and recorded in real time by optical density (600 nm, OD600). Specific growth rate (m) at a given thiocyanate concentration was first estimated using Eq. (1). The calculated m values were then plotted versus the initial thiocyanate concentrations. The Monod model (Eq. (1)) and the Andrew model (Eq. (2)) were tested in parallel by nonlinear least squares regression to describe the overall relationship between microbial growth and residual thiocyanate concentration (Shuler and Kargi, 2002): m¼
mm S KS þ S
mm m¼ KS S 1þ 1þ S KSI
(1)
(2)
where m is the specific growth rate, mm is the maximum specific growth rate in the absence of inhibition, S is the residual substrate concentration, KS is the saturation coefficient equivalent to the substrate concentration at which m is half of its maximum, KSI is the substrate inhibition coefficient which equals to the substrate concentration at which m is half of its maximum, KSI > KS.
3.
Results and discussion
3.1.
Evaluation of thiocyanate biodegradation kinetics
A specific growth rate versus substrate concentration plot (m-S plot), was constructed for each RSA trial condition (Table 1)
253
from the microcultivation results. In all trials, the m value increased, with increasing thiocyanate concentration, up to the peak value (0.137e0.194) and then gradually decreased with further increasing thiocyanate concentration (data not shown). This indicates that the m-S relationship follows a substrate inhibition pattern and thus would fit the Andrew model better than the Monod model (Lee et al., 2009). Correspondingly, all m-S plots showed a good fit to the Andrew model (r2 > 0.95) while the Monod model fit was relatively poor (r2 < 0.80). The Andrew model coefficients, i.e., mm, KS, and KSI (Eq. (3)) were estimated from the curve fitting results (Table 1).
3.2. Effects of temperature and pH on biokinetic properties Linear to partial cubic polynomials were sequentially examined with the observed results (Table 1) to describe the responses of dependent variables, i.e., mm, KS, and KSI. Original FCD points (trials 1e9 in Table 1) were run first to test the adequacy of linear models depicting the response surfaces of biokinetic coefficients. As the first-order regression was not statistically significant for all responses, four augmentation points (trials 10e13 in Table 1) were added to the original design to better describe the responses using a higher-order model (Montgomery, 1997). The augmented experimental runs were performed after the completion of the original FCD trials, implying that, given the high diversity of thiocyanatedegrading communities, the seed inocula for the original and augmented trials might be different in microbial community composition (Felfo¨ldi et al., 2010). Such possible microbial variations in inocula were thought to be minimal as a stable inoculum system was employed to provide consistent inoculum for all experiments in this study. Increasingly complex polynomials were applied to model the augmented data set with checking the statistical significances of the models and coefficients. As a result, for all kinetic coefficients, a partial cubic model was selected as the best model to approximate the response surface (Eq. (3), Table 2): Yi ¼ b0 þ b1 X1 þ b2 X2 þ b3 X1 X2 þ b4 X21 þ b5 X22 þ b6 X21 X2 þ b7 X1 X22
ð3Þ
where Yi is the predicted response of i (i ¼ mm, KS, or KSI), Xj is the independent variable j ( j ¼ temperature and pH in order), bk is the coefficient value estimated by the polynomial model. The mm response surface model showed a high r2 of 0.937 and the model regression was significant at the 0.1% a-level ( p ¼ 0.001). Lack-of-fit (LOF ) was not significant ( p > 0.05) and the model adequacy was validated by a high adequate precision (AP) value of 11.0. AP is a measure of the range of predicted responses relative to the average prediction error, in other words, a signal to noise ratio. A high AP value therefore indicates a statistically sound model and it is generally recommended to be greater than 4 for an adequate model (Design Expert 7). Consequently, the resulted mm model proved adequate to navigate the design space. The significance of individual terms in the model was also examined (Table 2). Two higher-order terms, X21 (temperature2) and X21 X2 (temperature2 pH) were significant at the 1% a-level. All other terms were, on the other hand, not significant at the 5% a-level (p > 0.05), with X1 (temperature) still showing a considerable
254
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Table 2 e Estimated response surface coefficients and model significance. mm model
Terms Coefficient Intercept X1 X2 X1X2 X21 X22 X21 X2 X1 X22
Model Regression Lack-of-fit
3.854 2.095 8.116 4.673 2.567 2.828 4.957 1.223
101 101 102 103 102 104 103
a
KS model
EMR
p-value
4.190 2.435 2.804 1.027 0.255 0.595 0.220
0.059 0.134 0.135 <104 0.230 0.009 0.204
KSI model
Coefficient
EMR
p-value
Coefficient
EMR
p-value
1495.708 66.455 332.087 15.593 0.608 16.074 0.113 0.625
1329.1 996.3 935.6 243.2 144.7 135.8 112.4
0.314 0.364 0.257 0.024 0.494 0.200 0.255
4096.521 191.076 1126.087 47.920 1.200 66.229 0.229 2.468
3821.5 3378.3 2875.2 479.8 596.1 275.3 444.3
0.199 0.239 0.004 0.005 0.205 0.101 0.014
p-value
R2
APb
p-value
R2
APb
p-value
R2
APb
0.001 0.869
0.937
11.009
0.029 0.798
0.825
6.791
<103 0.721
0.952
15.499
a Effect on model response. b Adequate precision.
p-value of 0.059. This indicates that temperature affects the model output more significantly than pH does within the experimental region. Such a strong effect of temperature is mirrored in the mm response surface (Fig. 1 and S1), where the response value changes sharply along the temperature axis. On the other hand, the model contours elongate along the pH axis to form a hill of concentric ellipses, indicating a low sensitivity of the model to pH variation. The influences of temperature and pH were examined by varying one variable with the other one fixed at its level for the maximum mm response. The maximum model output of 0.500 d1 was obtained at 30.13 C and pH 8.56 within the design boundary (Table 3). The mm response goes down from 0.500 to 0.450 d1 with decreasing pH from 8.65 to 6.15 (i.e., 1.667 decrease in coded unit (CU)) at the fixed temperature of 30.13 C. The same effect was achieved by increasing or decreasing temperature
Fig. 1 e Response surface plot of maximum specific growth rate (mm, dL1) with respect to temperature ( C) and pH.
by 10.92 C (0.546 CU) from 30.13 C at the fixed pH of 8.56. These, together with the high effect on model response (EMR) values of temperature terms than of pH terms (Table 2), confirm that temperature has significantly greater influence on the mm model response (Fig. 1 and S1). EMR measures the magnitude and direction (i.e., positive or negative) of each term’s influence on the model prediction (i.e., designed range of a term corresponding coefficient value). The temperature term (X1) showed a 1.5e19-fold greater EMR value than the other terms and the EMR of the squared temperature term ðX21 Þ was 4.0-fold higher than of the squared pH term ðX22 Þ. These results agree well with the conventional wisdom that temperature is one of the most critical factors in controlling microbial growth in various environments (Madigan et al., 2009; Westermann et al., 1989). The KS response surface model exhibited a good approximation (r2 ¼ 0.825) significant at the 5% a-level ( p ¼ 0.029) and LOF was not significant ( p > 0.05). The model AP value was 6.791 which is high enough to assure the model adequacy (AP > 4). The resulted partial cubic model was therefore suggested to be suitable to approximate the response surface of KS. The squared temperature term ðX21 Þ was the only model term significant at the 5% a-level (Table 2). The maximum model response of 158.0 mg SCN/L was shown at 33.29 C and pH 6.53 (Table 3), with yielding a wide plateau around the peak point on the response surface (Fig. 2 and S2). This indicates that the model is not very sensitive to temperature or pH variations around the maximum response condition. The response contours run diagonally down the plateau toward the upper and lower right corners, suggesting a potential interdependence of temperature and pH in the corner areas. The KS response decreases from 158.0 to 150.0 mg SCN/L with either increasing pH from 6.53 to 7.83 (i.e., 0.867-CU rise) along the temperature axis at 33.29 C or decreasing temperature from 33.29 to 25.47 C (i.e., 0.782-CU drop) at the fixed pH of 6.53. This suggests that temperature does not have dominant effect over pH, although the contours are somewhat elongated along the pH axis, on the KS response surface plot. This corresponds with the comparable EMR values of temperatureand pH-related terms (Table 2), indicating a relatively even influence of two variables on the model response. As KS is the
255
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Table 3 e Predicted biokinetic coefficient values at their maximum and minimum conditions. Numbera
Conditions
Responses
1
T ( C)
pH
mm (d )
KS (mg SCN /L)
KSI (mg SCN/L)
aA (mg SCN/L d)
Maximum 1 2 3
mm KS KSI
30.13 33.29 40.00
8.56 6.53 9.00
0.500 0.468 0.294
144.7 158.0 83.4
228.4 221.3 447.0
3.46 103 2.96 103 3.53 103
Minimum 3 3 4
mm KS KSI
40.00 40.00 21.42
9.00 9.00 6.00
0.294 0.294 0.392
83.4 83.4 148.1
447.0 447.0 194.5
3.53 103 3.53 103 2.65 103
a Condition number, refer to Fig. 5.
concentration of substrate at which m is mm/2, a lower KS value means a higher tendency to keep growth rate from falling under low substrate conditions, i.e., higher substrate affinity (Shuler and Kargi, 2002). Substrate affinity relies on the functioning of cell membrane system that consists of colloidal complexes of phospholipids and proteins. Too high temperature denatures membrane-bound transport proteins while membrane layers become viscous by freezing effect at too low temperatures, hindering nutrient transportation and diffusion (Nedwell, 1999). The activity of an enzyme active site, the functioning part of an enzyme, is controlled by its ionization state directly affected by pH changes (Tan et al., 1998). Therefore, an enzyme will lose its function outside a suitable pH range, causing inhibition of ‘active uptake’ of substrate mediated by active transport proteins. These support our observation of dynamic changes in KS response associated with changes temperature and pH. The KSI response surface model showed an excellent fit (r2 ¼ 0.952) with a low p-value of <103. LOF was not significant ( p > 0.05) and the model AP value was sufficiently high for an
Fig. 2 e Response surface plot of saturation coefficient (KS, mg SCNL/L) with respect to temperature ( C) and pH.
adequate model (AP ¼ 15.50). These suggest that the model can make statistically reliable predictions of KSI response. Concerning individual model terms, X1X2 (temperature pH) and X21 (temperature2) were significant at the 1% a-level and X1 X22 (temperature ⅹ pH2) at the 5% a-level (Table 2). The KSI response surface plot generated a saddle-shaped surface with no pinnacle within the design boundary (Fig. 3 and S3). This saddle-shaped response surface is a signal of a significant effect of the interaction between the independent variables on the KSI response (Murthy et al., 2000), corresponding to the high significance of the interaction term X1X2 ( p ¼ 0.004; Table 2). The lowest and highest KSI outputs were on the lower left and upper right corners of the explored boundary, respectively. The contour level increases diagonally toward the upper right corner, especially rapidly in the high temperature and high pH region (ca. >30 C and >pH 7.5). This suggests that temperature and pH are interdependent for the KSI response, with the KSI model response being more sensitive to environmental changes in the high temperature and pH region.
Fig. 3 e Response surface plot of substrate inhibition coefficient (KSI, mg SCNL/L) with respect to temperature ( C) and pH.
256
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Substrate inhibition causes a drop in growth rate at high substrate concentrations where S > KSI (see Eq. (2)), indicating that a lower KSI means a higher potential for substrate inhibition (Shuler and Kargi, 2002). Here, all possible inhibitory effects of thiocyanate, such as competitive inhibition of anion transport, cell acidification, and binding to cellular components (Olson et al., 2003), are biokinetically simplified into KSI to have two affinity coefficients for thiocyanate in the biokinetic expression, KS and KSI (Eq. (2)). Although little information is available on the relationships between KSI and environmental factors, the biokinetic equation tells that substrate inhibition is attributed to the undesirable binding and functioning of substrate. This suggests that the affinity of inhibitory binding, directly related to KSI, varies with chemical or physical alterations of the non-target components, particularly proteins, possibly by changes in temperature and pH (Nedwell, 1999; Tan et al., 1998). Additionally, each response surface model was examined for any weakness by generating a residual plot with all experimental points (Fig. 4). All residual plots show good predictions of biokinetic responses across the observed values with scattered residuals with no structure or pattern, indicating random and homogenous variances. This, together with the relatively low coefficient of variation values of the obtained response surface models (6.16e10.76%; data not shown), again supports the adequacy and reliability of the models.
3.3. Simulation of meS relationships under different conditions As can be inferred from the Andrew model (Eq. (3)), a combination of a higher mm, a lower KS, and a higher KSI leads to a higher microbial growth rate and thus more efficient utilization of substrate (or treatment of contaminant). The apparent maximum specific growth rate ðm0m Þ was observed at a substrate concentration ðS0m Þ between KS and KSI (Shuler and Kargi, 2002). Table 3 summarizes the temperature and pH conditions for the maximum and minimum responses of each kinetic coefficient within the design boundary. The minimum mm and KS outputs and the maximum KSI output were commonly observed at the corner point of the highest temperature and pH (40 C, pH 9.0) (Figs. 1e3). Correspondingly, trial 4 among the 13 experimental trials showed the lowest mm and KS observations and the highest KSI observation (Table 1). Specific growth rate versus thiocyanate concentration plots corresponding to the conditions 1 to 4 described in Table 3 were generated and denoted as curves 1 to 4 (Fig. 5). Despite mm for curve 3 is significantly lower than that for any other curve, curve 3 is always above curves 2 and 4, indicating faster microbial growth in condition 3 than in conditions 2 and 4. This is owing to the negative effects of higher KS and lower KSI values for curves 2 and 4. Although conditions 2 and 4 were of comparable KS and KSI values, curve 2 illustrates markedly superior growth rate due to the significant difference in mm. Curve 1 which has the highest m0m shows the fastest growth over the thiocyanate concentration range of 7.6e663.0 mg/L and curve 3 dominates over curve 1 outside the range. As can be inferred from Eq. (2), the substrate inhibition effect (i.e., S/ KSI) is insignificant at low thiocyanate concentrations (where S << KSI) and the growth rate is largely affected by substrate
Fig. 4 e Residual plots of the response surface models for (A) maximum specific growth rate (mm), (B) saturation coefficient (KS), and (C) substrate inhibition coefficient (KSI). affinity (i.e., KS/S ). This suggests that the growth following curve 3 is favored under substrate limited conditions (S < 7.6 mg/L) due to its significantly lower KS value (1.74 fold). When thiocyanate concentration becomes over 7.6 mg SCN/ L, the growth following curve 1 is favored likely due to the
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
257
the descending order of the aA value. It is therefore suggested that a biological thiocyanate treatment process may be kept in the best condition by closely controlling temperature and pH. This could be a possible practical use of the response surface models constructed in this study (refer to Eq. (3) and Table 2) for effective thiocyanate treatment removal.
4.
Conclusions
The effects of temperature and pH on the biokinetic properties of thiocyanate degradation under mixed-culture, autotrophic conditions was successfully examined by a combination of RSA and biokinetic modeling.
Fig. 5 e Specific growth rate (m) versus thiocyanate concentration (S ) plots in conditions 1e4 in Table 3 where the biokinetic coefficients (mm, KS, and KSI) are at their maximum or minimum values. The low substrate concentration region (S < 15 mg/L) of each plot is shown enlarged in the inset. The different lines represent the following: solid black line (curve 1); solid grey line (curve 2); dashed black line (curve 3); and dashed grey line (curve 4).
effect of significantly greater mm value (1.70 fold). Interestingly, the superiority is reversed again as thiocyanate concentration becomes over 663.0 mg/L. This can be ascribed to the effect of the significantly higher KSI value for curve 3 than for curve 1 (1.96 fold). The negative impact of substrate inhibition on microbial growth becomes prominent at high substrate concentrations, particularly where S > KSI (Hung and Pavlostathis, 1999). Specific microbial growth rate, determining growth rate and thus substrate utilization rate, is a function of substrate concentration involving kinetic coefficients (Eqs. (1) and (2)). Changes in temperature and pH can cause changes in the kinetic coefficient values, potentially leading to significant variations in the biokinetic properties of a biological process as shown in Fig. 5. Full-scale biological treatment systems are generally run in continuous mode at low residual substrate concentrations for stable treatment of pollutants, suggesting that biokinetic behavior in low-substrate environments is likely of greater interest than that in nutrient-rich conditions in practical applications. The most decisive factor in determining growth rate under substrate-limited conditions is substrate affinity which is often measured by KS (Nedwell, 1999). However, KS alone cannot be a robust indicator of substrate affinity. This can be easily seen from two meS curves with the same KS but different mm (Nedwell, 1999). A more reliable indicator is specific substrate affinity (aA ¼ mm/KS) that reflects the initial slope of a m-S curve (Button, 1993). An aA value can thus describe how a combination of kinetic coefficients behaves, in terms of growth rate, in low-substrate environments (Nedwell, 1999). The aA calculations for the different conditions listed in Table 3 accord with the simulation results in Fig. 5. At low thiocyanate concentrations below 7.6 mg/L, the predicted growth is superior in order of curves 3, 1, 2, and 4, according to
(1) The microcultivation study of thiocyanate biodegradation presented substrate inhibition pattern and Andrew model successfully describe the thiocyanate degradation biokinetics (r2 > 0.95 for all trials). (2) A partial cubic model was constructed to describe the response surface of each biokinetic coefficient (mm, KS, and KSI) and the model reliability was verified by statistical tests and residual plotting. All models showed a good fit (r2 ¼ 0.825e0.952) significant at the 5% a-level. (3) The biokinetic coefficients were significantly affected by changes in temperature and pH in distinct manners, leading to the markedly different response surface models and plots. (4) The squared temperature term ðX21 Þ was the only term significant at the 5% a-level in all response surface models, denoting the significant effect of temperature on the biokinetic coefficients studied. A dominant influence of temperature was observed in the mm model, whereas the interdependence of temperature and pH was pronounced in the KS and KSI models. Overall, the response surface approximations demonstrated that temperature and pH have significant and unique effects on the biokinetic properties of thiocyanate biodegradation. The constructed response surface models provide basic biokinetic information that, together with further studies, may lead to a better control of thiocyanate treatment process.
Acknowledgments This work was supported by the Korea Research Foundation grant funded by the Korean Government (MOEHRD) and the New & Renewable Energy Technology R&D program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean Ministry of Knowledge Economy (Grant no. 20103020090050). The authors are also grateful for the support of the 2012 Research Fund of Ulsan National Institute of Science and Technology (UNIST).
Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2012.10.003.
258
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
references
Ahn, J.-H., Kim, J., Lim, J., Hwang, S., 2004. Biokinetic evaluation and modeling of continuous thiocyanate biodegradation by Klebsiella sp. Biotechnology Progress 20, 1069e1075. Bezsudnova, E.Y., Sorokin, D.Y., Tikhonova, T.V., Popov, V.O., 2007. Thiocyanate hydrolase, the primary enzyme initiating thiocyanate degradation in the novel obligately chemolithoautotrophic halophilic sulfur-oxidizing bacterium Thiohalophilus thiocyanoxidans. Biochimica et Biophysica Acta (BBA) e Proteins & Proteomics 1774, 1563e1570. Button, D.K., 1993. Nutrient-limited microbial growth kinetics: overview and recent advances. Antonie van Leeuwenhoek 63, 225e235. Ebbs, S., 2004. Biological degradation of cyanide compounds. Current Opinion in Biotechnology 15, 231e236. Felfo¨ldi, T., Sze´kely, A.J., Gora´l, R., Barka´cs, K., Scheirich, G., Andra´s, J., Ra´cz, A., Ma´rialigeti, K., 2010. Polyphasic bacterial community analysis of an aerobic activated sludge removing phenols and thiocyanate from coke plant effluent. Bioresource Technology 101, 3406e3414. Hung, C.-H., Pavlostathis, S.G., 1997. Aerobic biodegradation of thiocyanate. Water Research 31, 2761e2770. Hung, C.-H., Pavlostathis, S.G., 1999. Kinetics and modeling of autotrophic thiocyanate biodegradation. Biotechnology and Bioengineering 62, 1e11. Kim, S.-J., Katayama, Y., 2000. Effect of growth conditions on thiocyanate degradation and emission of carbonyl sulfide by Thiobacillus thioparus THI115. Water Research 34, 2887e2894. Laidler, K.J., 1984. The development of the Arrhenius equation. Journal of Chemical Education 61, 494e498. Lay-Son, M., Drakides, C., 2008. New approach to optimize operational conditions for the biological treatment of a highstrength thiocyanate and ammonium waste: pH as key factor. Water Research 42, 774e780. Lee, C., Kim, J., Chang, J., Hwang, S., 2003. Isolation and identification of thiocyanate utilizing chemolithotrophs from gold mine soils. Biodegradation 14, 183e188. Lee, C., Kim, J., Hwang, K., Hwang, S., 2009. Fermentation and growth kinetic study of Aeromonas caviae under anaerobic conditions. Applied Microbiology and Biotechnology 83, 767e773.
Lee, C., Lee, S., Cho, K.-J., Hwang, S., 2011. Mycelial cultivation of Phellinus linteus using cheese-processing waste and optimization of bioconversion conditions. Biodegradation 22, 103e110. Lewis, R.J., 1992. Sax’s Dangerous Properties of Industrial Materials. Van Nostrand Reinhold Co., New York. Madigan, M.T., Martinko, J.M., Dunlap, P.V., Clack, D.P., 2009. Brock Biology of Microorganisms. Pearson Education, Inc., San Francisco, CA. Montgomery, D.C., 1997. Design and Analysis of Experiments, fourth ed. John Wiley and Sons, Inc, New York, NY. Murthy, M.S.R.C., Swaminathan, T., Rakshit, S.K., Kosugi, Y., 2000. Statistical optimization of lipase catalyzed hydrolysis of methyloleate by response surface methodology. Bioprocess Engineering 22, 35e39. Nedwell, D.B., 1999. Effect of low temperature on microbial growth: lowered affinity for substrates limits growth at low temperature. FEMS Microbiology Ecology 30, 101e111. Olson, G.J., Brierley, J.A., Brierley, C.L., 2003. Bioleaching review part B. Applied Microbiology and Biotechnology 63, 249e257. Rosso, L., Lobry, J., Bajard, S., Flandrois, J., 1995. Convenient model to describe the combined effects of temperature and pH on microbial growth. Applied and Environmental Microbiology 61, 610e616. Shuler, M.L., Kargi, F., 2002. Bioprocess Engineering: Basic Concepts. Prentice Hall PTR, Upper Saddle River. Tan, Y., Wang, Z.-X., Marshall, K.C., 1998. Modeling pH effects on microbial growth: a statistical thermodynamic approach. Biotechnology and Bioengineering 59, 724e731. Vazquez, I., Rodriguez, J., Maranon, E., Castrill, L., Fernandez, Y., 2006. Simultaneous removal of phenol, ammonium and thiocyanate from coke wastewater by aerobic biodegradation. Journal of Hazardous Materials 137, 1773e1780. Westermann, P., Ahring, B.K., Mah, R.A., 1989. Temperature compensation in Methanosarcina barkeri by modulation of hydrogen and acetate affinity. Applied and Environmental Microbiology 55, 1262e1266. Wood, P.A., Kelly, P.D., McDonald, R.I., Jordan, L.S., Morgan, D.T., Khan, S., Murrell, J.C., Borodina, E., 1998. A novel pinkpigmented facultative methylotroph, Methylobacterium thiocyanatum sp. nov., capable of growth of thiocyanate or cyanate as sole nitrogen sources. Archives of Microbiology 169, 148e158.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effects of temperature and pH on the biokinetic properties of thiocyanate biodegradation under autotrophic conditions Jaai Kim a, Kyung-Jin Cho b, Gyuseong Han b, Changsoo Lee a,*, Seokhwan Hwang b,** a
School of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 689-798, Republic of Korea b School of Environmental Science and Engineering, POSTECH, Pohang, Gyungbuk 790-784, Republic of Korea
article info
abstract
Article history:
The simultaneous effects of temperature and pH on the biokinetic properties of thiocya-
Received 11 July 2012
nate biodegradation under mixed-culture, autotrophic conditions were investigated using
Received in revised form
response surface analysis (RSA) combined with biokinetic modeling. A partial cubic model,
28 September 2012
based on substrate inhibition biokinetics, was constructed for each kinetic coefficient in
Accepted 3 October 2012
Andrew model (i.e., maximum specific growth rate (mm), saturation coefficient (KS), and
Available online 22 October 2012
substrate inhibition coefficient (KSI)). Each model proved statistically reliable to approximate the responses of the kinetic coefficients to temperature and pH changes (r2 > 0.8,
Keywords:
p < 0.05). The response surface plots demonstrated that the biokinetic coefficients change
Biokinetic modeling
with respect to temperature and pH significantly and in different ways. The model
pH
response surfaces were substantially different to each other, indicating distinct correla-
Response surface analysis
tions between the independent (temperature and pH) and dependent (model response)
Temperature
variables in the models. Based on the estimated response surface models, temperature
Thiocyanate biodegradation
was shown to have significant effects on all biokinetic coefficients tested. A dominant influence of temperature on mm response was observed while the interdependence of temperature and pH was apparent in the KS and KSI models. Specific growth rate (m) versus substrate (i.e., thiocyanate) concentration plots simulating using the obtained response surface models confirmed the significant effects of temperature and pH on the microbial growth rate and therefore on the thiocyanate degradation rate. Overall, the response surface models able to describe the biokinetic effects of temperature and pH on thiocyanate biodegradation within the explored region (20e30 C and pH 6.0e9.0) were successfully constructed and validated, providing fundamental information for better process control in thiocyanate treatment. ª 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Thiocyanate (SCN), a hazardous compound, is often found in industrial waste streams as it is used or generated in various industries including photofinishing, dyeing, coking, metal
separation, and electroplating (Ahn et al., 2004). Thiocyanate is toxic at low concentrations (1e2 mM) to higher animals including humans. It binds to proteins and inhibits enzymatic reactions (Wood et al., 1998), leading to damage to human central nervous system and thus severe clinical problems
* Corresponding author. Tel.: þ82 52 217 2822; fax: þ82 217 2819. ** Corresponding author. Tel.: þ82 54 279 2282; fax: þ82 54 279 8299. E-mail addresses: [email protected] (C. Lee), [email protected] (S. Hwang). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2012.10.003
252
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
such as nervousness, hallucination, psychosis, and convulsions (Lewis, 1992). Effective removal of thiocyanate is therefore of particular concern when treating wastewaters from the industries mentioned above. Biological treatment of thiocyanate has extensively been studied and several thiocyanate-degrading bacteria, e.g., Arthrobacter, Bacillus, Escherichia, Klebsiella, Methylobacterium, Pseudomonas, and Thiobacillus, were identified from various sources (Ebbs, 2004; Lee et al., 2003). These bacteria degrade and utilize thiocyanate as an energy and/or nutrient source through two distinct metabolic pathways: carbonyl and cyanate pathways (Bezsudnova et al., 2007; Ebbs, 2004). In the former, thiocyanate is first hydrolyzed to ammonia and carbonyl sulfide (COS) by thiocyanate hydrolase, and then COS breaks into H2S and CO2. In the latter, thiocyanate is initially degraded to cyanate (CNO) and H2S, and then CNO is hydrolyzed to ammonia and CO2 by cyanase. The sulfide and ammonia released can be utilized as electron donor, nitrogen source, or sulfur source for microbial growth. Like other biological processes, activities of the microorganisms involved in thiocyanate biodegradation are to be affected by environmental factors. Understanding such influence is helpful in devising methods for enhancing microbial activities and thus process performance. Among many environmental factors, temperature and pH are universally known as the major factors affecting microbial growth (Madigan et al., 2009; Rosso et al., 1995). The temperature and pH effects on microbial activities are well documented in literature (Laidler, 1984; Madigan et al., 2009; Nedwell, 1999; Tan et al., 1998). A few studies have investigated the effects of temperature and pH in biological thiocyanate treatment processes (Kim and Katayama, 2000; Lay-Son and Drakides, 2008; Vazquez et al., 2006). However, previous studies have examined only the responses of processes to such environmental factors but, to our knowledge, not their effects on the biokinetic properties of thiocyanate degrading communities. Biokinetics mathematizes the relationship between microbial growth and substrate consumption using a combination of coefficients, helping describe and predict process performance. From an engineering point of view, biokinetic modeling is a useful tool for understanding the basic mechanism resulting in microbial growth with pollutant removal under a certain condition. It is therefore plain that an understanding of how environmental factors affect the biokinetics of a biological process will provide fundamental information that helps better describe how and why the system changes at the process level. Accumulation of such information will form a basis for better design and operation of biological processes. In this study, we aimed to investigate the effects of temperature and pH on the biokinetic properties (i.e., biokinetic coefficients) of thiocyanate biodegradation under a mixed-culture, autotrophic condition. Taking into account the two factors that vary simultaneously, response surface analysis (RSA), an effective statistical technique for evaluating simultaneous effects of multiple variables (Montgomery, 1997), was employed. RSA was used to design the experimental runs to test and build the response models for biokinetic coefficients. This study provides a biokinetic insight into the biodegradation process of thiocyanate.
2.
Materials and methods
2.1.
Microbial source and culture medium
Activated sludge from a local sewage treatment plant (Pohang, Korea) was cultivated in a continuously stirred tank reactor (CSTR) with a working volume of 7 L to produce consistent inoculum for subsequent experiments. Thiocyanatedegrading microorganisms were enriched using a synthetic wastewater containing the following components in 1 L (Hung and Pavlostathis, 1997): 500 mg SCN, 838 mg KSCN, 500 mg NaHCO3, 385 mg K2HPO4, 129 mg KH2PO4, 50 mg MgSO4$7H2O, 7 mg KCl, 5 mg CaCl2$2H2O, 5 mg FeSO4$7H2O, and 5 mg MnSO4$H2O. The inoculum system was operated at 7.2-day hydraulic retention time (HRT), 30 C, and pH 8.5. After over 10 turnovers of steady-state operation (>99% SCN removal efficiency), the effluent from this system was used as seed inoculum for the microcultivation tests for biokinetic studies.
2.2.
Response surface analysis design
Response surface analysis (RSA) was employed to evaluate the simultaneous effects of temperature and pH on the biodegradation kinetics of thiocyanate with minimum number of experimental trials (Montgomery, 1997). A face-centered design (FCD) with extra trials added to augment the response surface model (Table 1) was applied to evaluate the biokinetic responses to changes in temperature and pH (i.e., independent variables). Based on literature values, the exploring ranges (center point variance) of temperature and pH were determined to be 30 10 C and 7.5 1.5, respectively (Ahn et al., 2004; Hung and Pavlostathis, 1997; Kim and Katayama, 2000; Lee et al., 2003). RSA was performed by a sequential procedure of collecting experimental data from each trial, estimating polynomials, and checking model adequacy (Lee et al., 2011),
Table 1 e Experimental design and observed results of the response surface analysis. Trials Conditions
T ( C)
pH
Observations 1
mm (d )
KSI KS (mg SCN/L) (mg SCN/L)
Original face-centered design 1 20 6.0 0.380 140.9 2 20 9.0 0.326 106.5 3 40 6.0 0.422 143.4 4 40 9.0 0.290 74.9 5 30 6.0 0.428 156.2 6 30 9.0 0.495 147.4 7 20 7.5 0.335 122.8 8 40 7.5 0.412 146.8 30 7.5 0.497 (0.036) 148.1 (18.9) 9a Augmented experimental 10 35 6.0 11 20 7.0 12 35 7.0 13 40 8.5
points 0.471 0.356 0.446 0.348
171.1 134.1 149.6 112.1
197.0 216.1 261.4 461.2 212.1 241.1 244.7 272.2 205.0 (26.3)
206.4 230.1 250.1 356.4
a Center point. Experiment was triplicated and the responses are presented as average values (standard deviation).
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
using Design Expert 7 software (Stat-Ease, Minneapolis, MN). In RSA computation, the independent variables are converted to coded values for computational convenience: the upper limit of a factor (i.e., temperature or pH in this study) is coded as þ1, the center level as 0, and the lower limit as 1. The bestfit response surface model was chosen by fitting increasingly complex polynomials from low to high orders (i.e., linear to partial cubic) to the experimental data. Model adequacy was checked based on the p-values of regression, lack-of-fit, and model coefficients.
2.3.
Evaluation of biokinetic coefficients
An automated growth-monitoring system equipped with a 200-well microcultivation unit (Bioscreen C, Labsystems) was used to produce batch data for the biokinetic characterization of thiocyanate degradation at different pH and temperature conditions (Table 1). Ten different initial thiocyanate concentrations, i.e., 10, 30, 50, 75, 100, 125, 150, 200, 300, and 400 mg/L, were tested, in five replicates each, for each trial condition, requiring 50 cultivations per trial condition. For statistical soundness, the center point (trial 9 in Table 1) was triplicated using three separately prepared sets of 50 cultivations. The microcultivations were performed using the synthetic wastewater fed to the inoculum system with appropriately modified thiocyanate concentrations. Each microcultivation mixture was set up with 390 mL of medium and 10 mL of inoculum (seeding rate, 2.5% v/v). The initial pH and the incubation temperature were set to the corresponding test conditions (Table 1) before the start of cultivation. Microbial growth was automatically measured and recorded in real time by optical density (600 nm, OD600). Specific growth rate (m) at a given thiocyanate concentration was first estimated using Eq. (1). The calculated m values were then plotted versus the initial thiocyanate concentrations. The Monod model (Eq. (1)) and the Andrew model (Eq. (2)) were tested in parallel by nonlinear least squares regression to describe the overall relationship between microbial growth and residual thiocyanate concentration (Shuler and Kargi, 2002): m¼
mm S KS þ S
mm m¼ KS S 1þ 1þ S KSI
(1)
(2)
where m is the specific growth rate, mm is the maximum specific growth rate in the absence of inhibition, S is the residual substrate concentration, KS is the saturation coefficient equivalent to the substrate concentration at which m is half of its maximum, KSI is the substrate inhibition coefficient which equals to the substrate concentration at which m is half of its maximum, KSI > KS.
3.
Results and discussion
3.1.
Evaluation of thiocyanate biodegradation kinetics
A specific growth rate versus substrate concentration plot (m-S plot), was constructed for each RSA trial condition (Table 1)
253
from the microcultivation results. In all trials, the m value increased, with increasing thiocyanate concentration, up to the peak value (0.137e0.194) and then gradually decreased with further increasing thiocyanate concentration (data not shown). This indicates that the m-S relationship follows a substrate inhibition pattern and thus would fit the Andrew model better than the Monod model (Lee et al., 2009). Correspondingly, all m-S plots showed a good fit to the Andrew model (r2 > 0.95) while the Monod model fit was relatively poor (r2 < 0.80). The Andrew model coefficients, i.e., mm, KS, and KSI (Eq. (3)) were estimated from the curve fitting results (Table 1).
3.2. Effects of temperature and pH on biokinetic properties Linear to partial cubic polynomials were sequentially examined with the observed results (Table 1) to describe the responses of dependent variables, i.e., mm, KS, and KSI. Original FCD points (trials 1e9 in Table 1) were run first to test the adequacy of linear models depicting the response surfaces of biokinetic coefficients. As the first-order regression was not statistically significant for all responses, four augmentation points (trials 10e13 in Table 1) were added to the original design to better describe the responses using a higher-order model (Montgomery, 1997). The augmented experimental runs were performed after the completion of the original FCD trials, implying that, given the high diversity of thiocyanatedegrading communities, the seed inocula for the original and augmented trials might be different in microbial community composition (Felfo¨ldi et al., 2010). Such possible microbial variations in inocula were thought to be minimal as a stable inoculum system was employed to provide consistent inoculum for all experiments in this study. Increasingly complex polynomials were applied to model the augmented data set with checking the statistical significances of the models and coefficients. As a result, for all kinetic coefficients, a partial cubic model was selected as the best model to approximate the response surface (Eq. (3), Table 2): Yi ¼ b0 þ b1 X1 þ b2 X2 þ b3 X1 X2 þ b4 X21 þ b5 X22 þ b6 X21 X2 þ b7 X1 X22
ð3Þ
where Yi is the predicted response of i (i ¼ mm, KS, or KSI), Xj is the independent variable j ( j ¼ temperature and pH in order), bk is the coefficient value estimated by the polynomial model. The mm response surface model showed a high r2 of 0.937 and the model regression was significant at the 0.1% a-level ( p ¼ 0.001). Lack-of-fit (LOF ) was not significant ( p > 0.05) and the model adequacy was validated by a high adequate precision (AP) value of 11.0. AP is a measure of the range of predicted responses relative to the average prediction error, in other words, a signal to noise ratio. A high AP value therefore indicates a statistically sound model and it is generally recommended to be greater than 4 for an adequate model (Design Expert 7). Consequently, the resulted mm model proved adequate to navigate the design space. The significance of individual terms in the model was also examined (Table 2). Two higher-order terms, X21 (temperature2) and X21 X2 (temperature2 pH) were significant at the 1% a-level. All other terms were, on the other hand, not significant at the 5% a-level (p > 0.05), with X1 (temperature) still showing a considerable
254
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Table 2 e Estimated response surface coefficients and model significance. mm model
Terms Coefficient Intercept X1 X2 X1X2 X21 X22 X21 X2 X1 X22
Model Regression Lack-of-fit
3.854 2.095 8.116 4.673 2.567 2.828 4.957 1.223
101 101 102 103 102 104 103
a
KS model
EMR
p-value
4.190 2.435 2.804 1.027 0.255 0.595 0.220
0.059 0.134 0.135 <104 0.230 0.009 0.204
KSI model
Coefficient
EMR
p-value
Coefficient
EMR
p-value
1495.708 66.455 332.087 15.593 0.608 16.074 0.113 0.625
1329.1 996.3 935.6 243.2 144.7 135.8 112.4
0.314 0.364 0.257 0.024 0.494 0.200 0.255
4096.521 191.076 1126.087 47.920 1.200 66.229 0.229 2.468
3821.5 3378.3 2875.2 479.8 596.1 275.3 444.3
0.199 0.239 0.004 0.005 0.205 0.101 0.014
p-value
R2
APb
p-value
R2
APb
p-value
R2
APb
0.001 0.869
0.937
11.009
0.029 0.798
0.825
6.791
<103 0.721
0.952
15.499
a Effect on model response. b Adequate precision.
p-value of 0.059. This indicates that temperature affects the model output more significantly than pH does within the experimental region. Such a strong effect of temperature is mirrored in the mm response surface (Fig. 1 and S1), where the response value changes sharply along the temperature axis. On the other hand, the model contours elongate along the pH axis to form a hill of concentric ellipses, indicating a low sensitivity of the model to pH variation. The influences of temperature and pH were examined by varying one variable with the other one fixed at its level for the maximum mm response. The maximum model output of 0.500 d1 was obtained at 30.13 C and pH 8.56 within the design boundary (Table 3). The mm response goes down from 0.500 to 0.450 d1 with decreasing pH from 8.65 to 6.15 (i.e., 1.667 decrease in coded unit (CU)) at the fixed temperature of 30.13 C. The same effect was achieved by increasing or decreasing temperature
Fig. 1 e Response surface plot of maximum specific growth rate (mm, dL1) with respect to temperature ( C) and pH.
by 10.92 C (0.546 CU) from 30.13 C at the fixed pH of 8.56. These, together with the high effect on model response (EMR) values of temperature terms than of pH terms (Table 2), confirm that temperature has significantly greater influence on the mm model response (Fig. 1 and S1). EMR measures the magnitude and direction (i.e., positive or negative) of each term’s influence on the model prediction (i.e., designed range of a term corresponding coefficient value). The temperature term (X1) showed a 1.5e19-fold greater EMR value than the other terms and the EMR of the squared temperature term ðX21 Þ was 4.0-fold higher than of the squared pH term ðX22 Þ. These results agree well with the conventional wisdom that temperature is one of the most critical factors in controlling microbial growth in various environments (Madigan et al., 2009; Westermann et al., 1989). The KS response surface model exhibited a good approximation (r2 ¼ 0.825) significant at the 5% a-level ( p ¼ 0.029) and LOF was not significant ( p > 0.05). The model AP value was 6.791 which is high enough to assure the model adequacy (AP > 4). The resulted partial cubic model was therefore suggested to be suitable to approximate the response surface of KS. The squared temperature term ðX21 Þ was the only model term significant at the 5% a-level (Table 2). The maximum model response of 158.0 mg SCN/L was shown at 33.29 C and pH 6.53 (Table 3), with yielding a wide plateau around the peak point on the response surface (Fig. 2 and S2). This indicates that the model is not very sensitive to temperature or pH variations around the maximum response condition. The response contours run diagonally down the plateau toward the upper and lower right corners, suggesting a potential interdependence of temperature and pH in the corner areas. The KS response decreases from 158.0 to 150.0 mg SCN/L with either increasing pH from 6.53 to 7.83 (i.e., 0.867-CU rise) along the temperature axis at 33.29 C or decreasing temperature from 33.29 to 25.47 C (i.e., 0.782-CU drop) at the fixed pH of 6.53. This suggests that temperature does not have dominant effect over pH, although the contours are somewhat elongated along the pH axis, on the KS response surface plot. This corresponds with the comparable EMR values of temperatureand pH-related terms (Table 2), indicating a relatively even influence of two variables on the model response. As KS is the
255
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Table 3 e Predicted biokinetic coefficient values at their maximum and minimum conditions. Numbera
Conditions
Responses
1
T ( C)
pH
mm (d )
KS (mg SCN /L)
KSI (mg SCN/L)
aA (mg SCN/L d)
Maximum 1 2 3
mm KS KSI
30.13 33.29 40.00
8.56 6.53 9.00
0.500 0.468 0.294
144.7 158.0 83.4
228.4 221.3 447.0
3.46 103 2.96 103 3.53 103
Minimum 3 3 4
mm KS KSI
40.00 40.00 21.42
9.00 9.00 6.00
0.294 0.294 0.392
83.4 83.4 148.1
447.0 447.0 194.5
3.53 103 3.53 103 2.65 103
a Condition number, refer to Fig. 5.
concentration of substrate at which m is mm/2, a lower KS value means a higher tendency to keep growth rate from falling under low substrate conditions, i.e., higher substrate affinity (Shuler and Kargi, 2002). Substrate affinity relies on the functioning of cell membrane system that consists of colloidal complexes of phospholipids and proteins. Too high temperature denatures membrane-bound transport proteins while membrane layers become viscous by freezing effect at too low temperatures, hindering nutrient transportation and diffusion (Nedwell, 1999). The activity of an enzyme active site, the functioning part of an enzyme, is controlled by its ionization state directly affected by pH changes (Tan et al., 1998). Therefore, an enzyme will lose its function outside a suitable pH range, causing inhibition of ‘active uptake’ of substrate mediated by active transport proteins. These support our observation of dynamic changes in KS response associated with changes temperature and pH. The KSI response surface model showed an excellent fit (r2 ¼ 0.952) with a low p-value of <103. LOF was not significant ( p > 0.05) and the model AP value was sufficiently high for an
Fig. 2 e Response surface plot of saturation coefficient (KS, mg SCNL/L) with respect to temperature ( C) and pH.
adequate model (AP ¼ 15.50). These suggest that the model can make statistically reliable predictions of KSI response. Concerning individual model terms, X1X2 (temperature pH) and X21 (temperature2) were significant at the 1% a-level and X1 X22 (temperature ⅹ pH2) at the 5% a-level (Table 2). The KSI response surface plot generated a saddle-shaped surface with no pinnacle within the design boundary (Fig. 3 and S3). This saddle-shaped response surface is a signal of a significant effect of the interaction between the independent variables on the KSI response (Murthy et al., 2000), corresponding to the high significance of the interaction term X1X2 ( p ¼ 0.004; Table 2). The lowest and highest KSI outputs were on the lower left and upper right corners of the explored boundary, respectively. The contour level increases diagonally toward the upper right corner, especially rapidly in the high temperature and high pH region (ca. >30 C and >pH 7.5). This suggests that temperature and pH are interdependent for the KSI response, with the KSI model response being more sensitive to environmental changes in the high temperature and pH region.
Fig. 3 e Response surface plot of substrate inhibition coefficient (KSI, mg SCNL/L) with respect to temperature ( C) and pH.
256
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
Substrate inhibition causes a drop in growth rate at high substrate concentrations where S > KSI (see Eq. (2)), indicating that a lower KSI means a higher potential for substrate inhibition (Shuler and Kargi, 2002). Here, all possible inhibitory effects of thiocyanate, such as competitive inhibition of anion transport, cell acidification, and binding to cellular components (Olson et al., 2003), are biokinetically simplified into KSI to have two affinity coefficients for thiocyanate in the biokinetic expression, KS and KSI (Eq. (2)). Although little information is available on the relationships between KSI and environmental factors, the biokinetic equation tells that substrate inhibition is attributed to the undesirable binding and functioning of substrate. This suggests that the affinity of inhibitory binding, directly related to KSI, varies with chemical or physical alterations of the non-target components, particularly proteins, possibly by changes in temperature and pH (Nedwell, 1999; Tan et al., 1998). Additionally, each response surface model was examined for any weakness by generating a residual plot with all experimental points (Fig. 4). All residual plots show good predictions of biokinetic responses across the observed values with scattered residuals with no structure or pattern, indicating random and homogenous variances. This, together with the relatively low coefficient of variation values of the obtained response surface models (6.16e10.76%; data not shown), again supports the adequacy and reliability of the models.
3.3. Simulation of meS relationships under different conditions As can be inferred from the Andrew model (Eq. (3)), a combination of a higher mm, a lower KS, and a higher KSI leads to a higher microbial growth rate and thus more efficient utilization of substrate (or treatment of contaminant). The apparent maximum specific growth rate ðm0m Þ was observed at a substrate concentration ðS0m Þ between KS and KSI (Shuler and Kargi, 2002). Table 3 summarizes the temperature and pH conditions for the maximum and minimum responses of each kinetic coefficient within the design boundary. The minimum mm and KS outputs and the maximum KSI output were commonly observed at the corner point of the highest temperature and pH (40 C, pH 9.0) (Figs. 1e3). Correspondingly, trial 4 among the 13 experimental trials showed the lowest mm and KS observations and the highest KSI observation (Table 1). Specific growth rate versus thiocyanate concentration plots corresponding to the conditions 1 to 4 described in Table 3 were generated and denoted as curves 1 to 4 (Fig. 5). Despite mm for curve 3 is significantly lower than that for any other curve, curve 3 is always above curves 2 and 4, indicating faster microbial growth in condition 3 than in conditions 2 and 4. This is owing to the negative effects of higher KS and lower KSI values for curves 2 and 4. Although conditions 2 and 4 were of comparable KS and KSI values, curve 2 illustrates markedly superior growth rate due to the significant difference in mm. Curve 1 which has the highest m0m shows the fastest growth over the thiocyanate concentration range of 7.6e663.0 mg/L and curve 3 dominates over curve 1 outside the range. As can be inferred from Eq. (2), the substrate inhibition effect (i.e., S/ KSI) is insignificant at low thiocyanate concentrations (where S << KSI) and the growth rate is largely affected by substrate
Fig. 4 e Residual plots of the response surface models for (A) maximum specific growth rate (mm), (B) saturation coefficient (KS), and (C) substrate inhibition coefficient (KSI). affinity (i.e., KS/S ). This suggests that the growth following curve 3 is favored under substrate limited conditions (S < 7.6 mg/L) due to its significantly lower KS value (1.74 fold). When thiocyanate concentration becomes over 7.6 mg SCN/ L, the growth following curve 1 is favored likely due to the
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
257
the descending order of the aA value. It is therefore suggested that a biological thiocyanate treatment process may be kept in the best condition by closely controlling temperature and pH. This could be a possible practical use of the response surface models constructed in this study (refer to Eq. (3) and Table 2) for effective thiocyanate treatment removal.
4.
Conclusions
The effects of temperature and pH on the biokinetic properties of thiocyanate degradation under mixed-culture, autotrophic conditions was successfully examined by a combination of RSA and biokinetic modeling.
Fig. 5 e Specific growth rate (m) versus thiocyanate concentration (S ) plots in conditions 1e4 in Table 3 where the biokinetic coefficients (mm, KS, and KSI) are at their maximum or minimum values. The low substrate concentration region (S < 15 mg/L) of each plot is shown enlarged in the inset. The different lines represent the following: solid black line (curve 1); solid grey line (curve 2); dashed black line (curve 3); and dashed grey line (curve 4).
effect of significantly greater mm value (1.70 fold). Interestingly, the superiority is reversed again as thiocyanate concentration becomes over 663.0 mg/L. This can be ascribed to the effect of the significantly higher KSI value for curve 3 than for curve 1 (1.96 fold). The negative impact of substrate inhibition on microbial growth becomes prominent at high substrate concentrations, particularly where S > KSI (Hung and Pavlostathis, 1999). Specific microbial growth rate, determining growth rate and thus substrate utilization rate, is a function of substrate concentration involving kinetic coefficients (Eqs. (1) and (2)). Changes in temperature and pH can cause changes in the kinetic coefficient values, potentially leading to significant variations in the biokinetic properties of a biological process as shown in Fig. 5. Full-scale biological treatment systems are generally run in continuous mode at low residual substrate concentrations for stable treatment of pollutants, suggesting that biokinetic behavior in low-substrate environments is likely of greater interest than that in nutrient-rich conditions in practical applications. The most decisive factor in determining growth rate under substrate-limited conditions is substrate affinity which is often measured by KS (Nedwell, 1999). However, KS alone cannot be a robust indicator of substrate affinity. This can be easily seen from two meS curves with the same KS but different mm (Nedwell, 1999). A more reliable indicator is specific substrate affinity (aA ¼ mm/KS) that reflects the initial slope of a m-S curve (Button, 1993). An aA value can thus describe how a combination of kinetic coefficients behaves, in terms of growth rate, in low-substrate environments (Nedwell, 1999). The aA calculations for the different conditions listed in Table 3 accord with the simulation results in Fig. 5. At low thiocyanate concentrations below 7.6 mg/L, the predicted growth is superior in order of curves 3, 1, 2, and 4, according to
(1) The microcultivation study of thiocyanate biodegradation presented substrate inhibition pattern and Andrew model successfully describe the thiocyanate degradation biokinetics (r2 > 0.95 for all trials). (2) A partial cubic model was constructed to describe the response surface of each biokinetic coefficient (mm, KS, and KSI) and the model reliability was verified by statistical tests and residual plotting. All models showed a good fit (r2 ¼ 0.825e0.952) significant at the 5% a-level. (3) The biokinetic coefficients were significantly affected by changes in temperature and pH in distinct manners, leading to the markedly different response surface models and plots. (4) The squared temperature term ðX21 Þ was the only term significant at the 5% a-level in all response surface models, denoting the significant effect of temperature on the biokinetic coefficients studied. A dominant influence of temperature was observed in the mm model, whereas the interdependence of temperature and pH was pronounced in the KS and KSI models. Overall, the response surface approximations demonstrated that temperature and pH have significant and unique effects on the biokinetic properties of thiocyanate biodegradation. The constructed response surface models provide basic biokinetic information that, together with further studies, may lead to a better control of thiocyanate treatment process.
Acknowledgments This work was supported by the Korea Research Foundation grant funded by the Korean Government (MOEHRD) and the New & Renewable Energy Technology R&D program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean Ministry of Knowledge Economy (Grant no. 20103020090050). The authors are also grateful for the support of the 2012 Research Fund of Ulsan National Institute of Science and Technology (UNIST).
Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2012.10.003.
258
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 5 1 e2 5 8
references
Ahn, J.-H., Kim, J., Lim, J., Hwang, S., 2004. Biokinetic evaluation and modeling of continuous thiocyanate biodegradation by Klebsiella sp. Biotechnology Progress 20, 1069e1075. Bezsudnova, E.Y., Sorokin, D.Y., Tikhonova, T.V., Popov, V.O., 2007. Thiocyanate hydrolase, the primary enzyme initiating thiocyanate degradation in the novel obligately chemolithoautotrophic halophilic sulfur-oxidizing bacterium Thiohalophilus thiocyanoxidans. Biochimica et Biophysica Acta (BBA) e Proteins & Proteomics 1774, 1563e1570. Button, D.K., 1993. Nutrient-limited microbial growth kinetics: overview and recent advances. Antonie van Leeuwenhoek 63, 225e235. Ebbs, S., 2004. Biological degradation of cyanide compounds. Current Opinion in Biotechnology 15, 231e236. Felfo¨ldi, T., Sze´kely, A.J., Gora´l, R., Barka´cs, K., Scheirich, G., Andra´s, J., Ra´cz, A., Ma´rialigeti, K., 2010. Polyphasic bacterial community analysis of an aerobic activated sludge removing phenols and thiocyanate from coke plant effluent. Bioresource Technology 101, 3406e3414. Hung, C.-H., Pavlostathis, S.G., 1997. Aerobic biodegradation of thiocyanate. Water Research 31, 2761e2770. Hung, C.-H., Pavlostathis, S.G., 1999. Kinetics and modeling of autotrophic thiocyanate biodegradation. Biotechnology and Bioengineering 62, 1e11. Kim, S.-J., Katayama, Y., 2000. Effect of growth conditions on thiocyanate degradation and emission of carbonyl sulfide by Thiobacillus thioparus THI115. Water Research 34, 2887e2894. Laidler, K.J., 1984. The development of the Arrhenius equation. Journal of Chemical Education 61, 494e498. Lay-Son, M., Drakides, C., 2008. New approach to optimize operational conditions for the biological treatment of a highstrength thiocyanate and ammonium waste: pH as key factor. Water Research 42, 774e780. Lee, C., Kim, J., Chang, J., Hwang, S., 2003. Isolation and identification of thiocyanate utilizing chemolithotrophs from gold mine soils. Biodegradation 14, 183e188. Lee, C., Kim, J., Hwang, K., Hwang, S., 2009. Fermentation and growth kinetic study of Aeromonas caviae under anaerobic conditions. Applied Microbiology and Biotechnology 83, 767e773.
Lee, C., Lee, S., Cho, K.-J., Hwang, S., 2011. Mycelial cultivation of Phellinus linteus using cheese-processing waste and optimization of bioconversion conditions. Biodegradation 22, 103e110. Lewis, R.J., 1992. Sax’s Dangerous Properties of Industrial Materials. Van Nostrand Reinhold Co., New York. Madigan, M.T., Martinko, J.M., Dunlap, P.V., Clack, D.P., 2009. Brock Biology of Microorganisms. Pearson Education, Inc., San Francisco, CA. Montgomery, D.C., 1997. Design and Analysis of Experiments, fourth ed. John Wiley and Sons, Inc, New York, NY. Murthy, M.S.R.C., Swaminathan, T., Rakshit, S.K., Kosugi, Y., 2000. Statistical optimization of lipase catalyzed hydrolysis of methyloleate by response surface methodology. Bioprocess Engineering 22, 35e39. Nedwell, D.B., 1999. Effect of low temperature on microbial growth: lowered affinity for substrates limits growth at low temperature. FEMS Microbiology Ecology 30, 101e111. Olson, G.J., Brierley, J.A., Brierley, C.L., 2003. Bioleaching review part B. Applied Microbiology and Biotechnology 63, 249e257. Rosso, L., Lobry, J., Bajard, S., Flandrois, J., 1995. Convenient model to describe the combined effects of temperature and pH on microbial growth. Applied and Environmental Microbiology 61, 610e616. Shuler, M.L., Kargi, F., 2002. Bioprocess Engineering: Basic Concepts. Prentice Hall PTR, Upper Saddle River. Tan, Y., Wang, Z.-X., Marshall, K.C., 1998. Modeling pH effects on microbial growth: a statistical thermodynamic approach. Biotechnology and Bioengineering 59, 724e731. Vazquez, I., Rodriguez, J., Maranon, E., Castrill, L., Fernandez, Y., 2006. Simultaneous removal of phenol, ammonium and thiocyanate from coke wastewater by aerobic biodegradation. Journal of Hazardous Materials 137, 1773e1780. Westermann, P., Ahring, B.K., Mah, R.A., 1989. Temperature compensation in Methanosarcina barkeri by modulation of hydrogen and acetate affinity. Applied and Environmental Microbiology 55, 1262e1266. Wood, P.A., Kelly, P.D., McDonald, R.I., Jordan, L.S., Morgan, D.T., Khan, S., Murrell, J.C., Borodina, E., 1998. A novel pinkpigmented facultative methylotroph, Methylobacterium thiocyanatum sp. nov., capable of growth of thiocyanate or cyanate as sole nitrogen sources. Archives of Microbiology 169, 148e158.