Full Text - Optimization of fermentation conditions for pristinamycin production by immobilized Streptomyces pristinaespiralis using response surface methodology

Optimization of fermentation conditions for pristinamycin production by immobilized Streptomyces pristinaespiralis using response surface methodology

Li-Jing Zhang¹ · Xi Zheng¹ · Zhi-Hua Jin*¹ · Sheng Hu¹ · Mei-Ru He¹

Financial support: The National Natural Science Foundation of China (No. 20976161) and the Natural Science Foundation of Zhejiang Province, China (No. Y4100123).

Keywords: fermentation conditions, immobilization, pristinamycin, response surface methodology, Streptomyces pristinaespiralis.

Response surface methodology was used to optimize the fermentation conditions for the production of pristinamycin by immobilization of Streptomyces pristinaespiralis F213 in shaking ﬂask cultivation. Seed medium volume, fermentation medium volume and shaking speed of seed culture were found to have signiﬁcant effects on pristinamycin production by the Plackett-Burman design. The steepest ascent method was adopted to approach the vicinity of optimum space, followed by central composite design for further optimization. A quadratic model was built to ﬁt the pristinamycin production. The optimum conditions were found to be seed medium volume of 29.5 ml, fermentation medium volume of 28.8 ml, and shaking speed of seed culture at 204 rpm. At the optimum conditions, a production of 213 mg/l was obtained, which was in agreement with the maximum predicted pristinamycin yield of 209 mg/l. This is the first report on pristinamycins production by immobilized S. pristinaespiralis using response surface methodology.

Pristinamycin, which was first discovered in France in 1962, is produced by Streptomyces pristinaespiralis (Corvini et al. 2000; Voelker and Altaba, 2001; Corvini et al. 2004). Pristinamycin consists of approximately 30% pristinamycin I (PI_A, PI_B, PI_C) and 70% pristinamycin II (PII_A, PII_B). Because these compounds bind to 23S rRNA of the 50S ribosomal subunit of bacteria and thereby inhibit protein synthesis (Ng and Gosbell, 2005), pristinamycin exhibits a prolonged post-antibiotic effect, and has been considered as an alternative for infections due to penicillin- and macrolide-resistant S. pneumonia (Qadri et al. 1997). So there is a growing interest of pristinamycin production in terms of both the strain improvement and the fermentation process optimization. In the last decades, attention has been paid mainly to cloning and analysis of genes involved in the pristinamycin synthesis (Sezonov et al. 1997; Bamas-Jacques et al. 1999; Jin et al. 2010). Whereas so far, few reports have been found on conditions optimization of pristinamycin fermentation. Under conventional batch conditions, the pristinamycins yield with the buffered synthetic medium was about 100 mg/l by a spontaneous mutant (Corvini et al. 2004).

Fermentation of immobilized microbial cells has recently gained much attention among many biotechnological approaches, because of its advantage over conventional free cell systems in respect to retention of high cell density, operational stability, higher efficiency of catalysis, higher volumetric productivity and lower shear stress (Adinarayana et al. 2005; Givry et al. 2008). The adsorption in porous material such as polyurethane foam (PUF) is a very simple immobilization used in a liquid fermentation (de Ory et al. 2004; Quezada et al. 2009). Many examples of microbial cells immobilization on PUFs have been reported in literatures, such as bacteria (de Ory et al. 2004), microalgae (Yamaguchi et al. 1999), basidiomycetes (Guimaraẽs et al. 2005), ascomycetes (Hama et al. 2006) and Cyanobacteria (Chetsumon et al. 1993). It is well known that designing proper culture conditions is a prerequisite in the production of metabolites (Maia et al. 2001). Nevertheless, the optimization of fermentation conditions for pristinamycin production by immobilized cell has not been made so far.

The conventional optimization method, one-factor-at-a-time approach, is time consuming and incapable of detecting the true optimum, especially due to the complex interactions among various physicochemical parameters (Wang et al. 2008). Response surface methodology (RSM) has overcome these drawbacks, therefore it can evaluate the relative signiﬁcance of several variables simultaneously, especially in the presence of complex interactions. As a result, RSM is used popularly to solve multivariate problems and has proved to be powerful and useful for the optimization of the target metabolites production (Rahman et al. 2004; Katapodis et al. 2006; Li et al. 2006; Sayyad et al. 2007).

In the present study, a surface response methodology was applied to optimize fermentation conditions for pristinamycins production using S. pristinaespiralis immobilized on PUFs. The Plackett-Burman (PB) design was used at ﬁrst to screen critical factors from a number of process variables, then the steepest ascent method to approach the experimental design space, and finally central composite design (CCD) was applied to further estimate the relationship between the variables and response as well as optimize the levels. The optimal conditions of seed medium volume, fermentation medium volume and shaking speed of seed culture for maximum pristinamycin yield have been quantiﬁed with response surface methodology.

The Streptomyces pristinaespiralis strain F213 derived from the chemical mutagenesis of wild-type S. pristinaespiralis ATCC25486 was used in the study and the spore solution of the strain was cryopreserved in 20% (v/v) glycerol at -80ºC, which was 5 x 10⁶ spores/ml. The seed medium (pH 7.0) was composed of (g/l): soluble starch 15, glucose 10, soybean ﬂour 15, peptone 5, yeast extract 5, KNO₃ 2.5, NaCl 2, and CaCO₃4. For fermentation of immobilized mycelia, the production medium (pH 6.5) was composed of (g/l): soluble starch 40, glucose 10, soybean ﬂour 25, peptone 5, yeast extract 3, ﬁsh extract 10, (NH₄)₂SO₄1.5, MgSO₄∙7H₂O 3.5, KH₂PO₄0.2, CaCO₃4. Both media were sterilized for 30 min at 121ºC.

Commercial polyurethane foam (PUF, Qitai Foam Co. Ltd, Shanghai, P.R. China) was used as the carrier and cut in the same length (1 cm), thickness (0.5 cm) and different width, the size of the PUFs were determinated by volume. The PUFs were soaked and swollen in 95% alcohol for 24 hrs to remove impurities and washed several times with sufficient distilled water to remove the alcohol. The PUFs were dried in vacuum oven at 80ºC for 8-10 hrs and added to the seed medium before sterilization.

The fermentation condition before optimization was as follows: for seed culture, 1 ml of spore solution was inoculated into a 250 ml shaking ﬂask containing 30 ml seed medium, in which the size and amount of PUFs was 0.35 cm³ and 17.5 cm³/100 ml. After incubation for 42 hrs with shaking at 240 rpm at 28ºC, then 1.5 cm³PUFs on which mycelia were immobilized were transferred with sterilized tweezers to 30 ml fermentation medium in 250 ml ﬂasks. The fermentation was carried out at 25ºC with shaking at 210 rpm for 72 hrs.

To determinate yield of pristinamycin, one volume of the whole fermentation broth containing mycelia and PUFs was directly mixed with two volumes of methanol for 1 hr. After centrifugation (4000 x g, 10 min), the supernates were analyzed by high-performance liquid chromatography using a Hypersil C18 column (4.6 by 250 mm), 0.1 M potassium phosphate buffer (pH 2.9) and acetonitrile (55:45, v/v) as the mobile phase, 1 ml/min ﬂow rate with detection at 206 nm. Commercial pristinamycin from Rhone-Poulenc Rorer Co. (Montrouge, France) was used as a reference standard.

The preliminary single-factor experiments revealed that the major variables affecting the pristinamycin production were seed medium volume, carrier amount, seed age, carrier size, inoculum volume of fermentation, fermentation medium volume, shaking speed of both seed culture and fermentation culture. These variables were chosen for further optimization.

PBD was used to screen the most important factors influencing pristinamycin production. The experimental design with the name, symbol code, and level of the variables is shown in Table 1. Each independent variable is represented in two levels, high and low, which are denoted by (+) and (-), respectively. Three dummy variables were studied in 12 experiments to calculate the standard error. Pristinamycin fermentation was carried out in duplication and the average value was taken as the response. Usually, the variable with P-value of < 0.05 was considered to have a signiﬁcant effect on the response and was selected for further optimization.

To find the neighbourhood of the optimum condition quickly, we used the method of the steepest ascent. The experiments were applied to determine a suitable direction by increasing or decreasing the variables according to the results obtained from the Plackett-Burman design (Gheshlaghi et al. 2005).

To describe the optimum culture conditions to enhance the pristinamycin production, the response surface methodology was performed with central-composite design. The levels of each variable and the design matrix are given in Table 2. The low, middle, and high levels of each variable were designated as -1.682, -1, 0, and 1, 1.682, respectively.

The Design Expert software (Version 7.0.0, Stat-Ease, Minneapolis, USA) was used for the experimental design and the analysis of variance (ANOVA) for the data. The quality of the polynomial model equation was judged statistically by the coefficient of determination R², and its statistical signiﬁcance was tested by an F-test. The signiﬁcance of the regression coefficients was determined by a t-test.

Based on our previous single-factor experiments, the importance of the eight culture conditions, namely, seed medium volume, carrier amount, inoculum amount, seed age, fermentation medium volume, shaking speed of both seed culture and fermentation culture for the pristinamycin production was analyzed by PBD. The experimental design and corresponding pristinamycin yields were shown in Table 1, whereas Table 3 shows the effects of these factors on the response and signiﬁcant levels.

Based on the statistical analysis, fermentation medium volume, with a probability value of 0.022, was determined to be the most signiﬁcant factor, followed by shaking speed of seed culture (0.047), and seed medium volume (0.053), so these three factors were considered in the further optimization. In the results, R² was found to be 0.9602, which means the model could explain 96.02% of the total variations in the system.

PBD results indicated that the effect of seed medium volume was positive, whereas that of fermentation medium volume and shaking speed of seed culture was negative. Thus it is predicted that increasing seed medium volume (X₁), while decreasing fermentation medium volume (X₂) and shaking speed of seed culture (X₃) should result in a higher production of pristinamycin. The centre point of the PBD has been considered as the origin of the path. The response for this point was determined as the average of responses for all the runs. From the results of the path of the steepest ascent, it is clearly seen that the yield proﬁle shows a maximum 202 mg/l at run 2 (Table 4). It suggested that this point might be near the region of the maximum pristinamycin yield response. Consequently, this point was chosen as the central point of CCD.

Based on the identiﬁcation of variables by the PBD and the steepest ascent method, the experiments were performed according to a CCD experimental plan together with experimental results (Table 5). In order to predict the maximum pristinamycin production corresponding to the optimum levels of the three variables, a second-order polynomial model was proposed to calculate the optimum levels of these variables. By applying the multiple regression analysis on experimental data, a second-order polynomial model in coded unit explains the role of each variable and their second-order interactions in producing pristinamycin. All terms, regardless of their significance, were included in the following second-order polynomial equation:

Where Y is the predicted pristinamycin production, X₁ is the seed medium volume, X₂ is the fermentation medium volume and X₃ is the shaking speed of seed culture. The predicted level of pristinamycin production at each experimental point calculated by the regression equation was shown in Table 5 along with the observed data.

Furthermore, the results of the second-order response surface model in the form of analysis of variance (ANOVA) were shown in Table 6. The P-value was used as a tool for checking the signiﬁcance of each coefficient. The smaller the P-value, the more signiﬁcant is the corresponding coefficient (Rahman et al. 2004). The Fisher’s F-test with a very low probability value [(P_model > F) = 0.0002] demonstrated that the model was highly signiﬁcant. Among the model terms, X₂,

and

had significant effects on pristinamycin production with a probability of not less than 95%, however, X₁, X₃ and the interaction terms of X₁, X₂, and X₃ seemed to be insigniﬁcant.

The fitness of the model can be checked by the determination coefficient (R²) and the adjusted determination coefficient (Adj R²). Here the value of R²for (Equation 1) was 0.9232, implying that 92% of the variability in the response could be explained by the model. The value of Adj R²was 0.8542 and it was also high enough to advocate for the significance of the model. The model also indicated statistically insigniﬁcant lack of ﬁt [(P_model> F) = 0.9403], so the model was supposed to be adequate for prediction within the range of variables employed. The coefficient of variation (CV) indicates the degree of precision with which the experiments are compared (Tanyildizi et al. 2006; Zhu et al. 2007) and the lower reliability of the experiment is usually indicated by high value of CV (Rahulan et al. 2009). In the present case, the lower value of CV (19.28%) indicates a better precision and reliability of the experiments performed. The normal plot of residuals was shown in Figure 1, as the most important diagnostic for the model, the normal probability plot of the residuals, came up by default. A linear pattern verified normality in the error term, i.e., there were no signs of any problems in the data (Wang and Liu, 2008).

The 3D response surface curve and 2D contour plot are generally the graphical representation of the regression equation. The three dimensional response surface and their corresponding 2D contour plots for the pristinamycin production against any two independent variables while the other independent variable maintained at zero levels were presented in Figure 2 and Figure 3. The graphical representation provides a method to visualize the relation between the response and experimental levels of each variable, and the type of interactions between test variables (Rahulan et al. 2009). The optimum value of each variable was located based on the hump in the three-dimensional plot, or from the central point of the corresponding contour plot.

Figure 2 depicted the three dimensional plot and its respective contour plot showing the response surface from the interaction between seed medium volume (X₁) and fermentation medium volume (X₂) while keeping other variable at its zero level. It can be seen from Figure 2, when fermentation medium volume was at a fixed level, the pristinamycin production increased with the increasing volume of seed medium but decreased beyond the range. On the contrary, when seed medium volume was at a fixed level, the effect of fermentation medium volume on the production response was similar to that of seed medium volume. These indicated that maximum pristinamycin could be obtained in the middle volume of both seed medium and fermentation medium. The elliptical nature of the contour plots stated clearly that the interaction between seed medium volume and fermentation medium volume were signiﬁcant.

Figure 4 showed the effect of fermentation medium volume (X₂) and shaking speed of seed culture (X₃) on pristinamycin production while the third variable was ﬁxed at its middle level. It was obvious that higher level of fermentation medium volume and lower shaking speed of seed culture would not increase the yield of pristinamycin. This may be that low dissolved oxygen results in the inhibitory effects to pristinamycin synthesis. When fermentation medium volume was at higher level (exceed 33 ml), pristinamycin production varied little with the change of shaking speed levels in seed culture. It was also noticed that pristinamycin production increased rapidly first and then decreased with gradually increasing value in fermentation medium volume, while shaking speed of seed culture was fixed at a certain level. This indicated that the requirements of dissolved oxygen and shearing strength to mycelia were quite high in pristinamycin fermentation by immobilized Streptomyces pristinaespiralis F213.

Figure 3 depicted the effect of seed medium volume (X₁) and shaking speed of seed culture (X₃) on pristinamycin production while fermentation medium volume was fixed at its zero level. As shown in Figure 3, when seed medium volume maintained at moderate level (27.5 ~ 32.5 ml), pristinamycin production varied little with the change of shaking speed of seed culture. However, pristinamycin production was sensitive to the change of shaking speed of seed culture, it tended to decrease rapidly with the increasing or decreasing shaking speed of seed culture, when the volume of seed medium was oversize or undersize.

Based on the quadratic model, the optimal values of each test variables in coded levels were as follows: X₁ = - 0.11, X₂ = - 0.24 , and X₃= - 0.19, whose actual values were seed medium volume 29.5 ml, fermentation medium 28.8 ml, and shaking speed of seed culture 204 rpm. The model predicted that the production of pristinamycin could reach 209 mg/l under the optimal condition. To verify the predicted result, validation experiment was carried out in triplicate test and the observed experimental production of pristinamycin was 213 ± 9 mg/l, which was closer to the predicted response. However the pristinamycin yield was 91 mg/l before optimization, a 1.34-fold increase had been obtained.

The RSM design applied in the present investigation have been successfully used in many metabolite production for optimization of immobilization conditions (Sankpal and Kulkarni, 2002; Aybastier and Demir, 2010; Liu et al. 2010). However, to the best our knowledge, there are no reports of optimization of pristinamycin production by immobilized S. pristinaespiralis using statistical experimental design.

Response surface methodology used in this investigation suggested the importance of dissolved oxygen supply for pristinamycin production in immobilization fermentation. A highly signiﬁcant quadratic polynomial obtained by the CCD was very useful for determining the optimal conditions with signiﬁcant effects on pristinamycin production. Validation experiments were also performed to verify the accuracy of the model, and the results indicated that the predicted value agreed with the experimental values well. A maximum pristinamycin yield of 213 mg/l was achieved, which was 2.34-fold higher than that before optimization, but it was lower than the reported yield of 412 mg/l (Xu et al. 2009). The less yield may be due to the difference of the strain used: the strain used in this study is a mutant strain, while that in the literature is a recombinant created from genome shuffling. Thus, for the fermentation of immobilized S. pristinaespiralis on PUFs, the response surface methodology was found to be a favourable strategy to optimize fermentation conditions for pristinamycin production. Being convenient and effective, this method might be useful in optimization of the immobilized fermentation for the overproduction of other metabolites.

ADINARAYANA, K.; JYOTHI, B. and ELLAIAH, P. (2005). Production of alkaline protease with immobilized cells of Bacillus subtilis PE-11 in various matrices by entrapment technique. AAPS Pharmscitech, vol. 6, no. 3, p. E391-E397. [CrossRef]

BAMAS-JACQUES, N.; LORENZON, S.; LACROIX, P.; DE SWETSCHIN, C. and CROUZET, J. (1999). Cluster organization of the genes of Streptomyces pristinaespiralis involved in pristinamycin biosynthesis and resistance elucidated by pulsed-field gel electrophoresis. Journal of Applied Microbiology, vol. 87, no. 6, p. 939-948. [CrossRef]

CORVINI, P.F.; GAUTIER, H.; RONDAGS, E.; VIVIER, H.; GOERGEN, J.L. and GERMAIN, P. (2000). Intracellular pH determination of pristinamycin-producing Streptomyces pristinaespiralis by image analysis. Microbiology, vol. 146, no. 10, p. 2671-2678.

GIVRY, S.; PREVOT, V. and DUCHIRON, F. (2008). Lactic acid production from hemicellulosic hydrolyzate by cells of Lactobacillus bifermentans immobilized in Ca-alginate using response surface methodology. World Journal of Microbiology & Biotechnology, vol. 24, no. 6, p. 745-752. [CrossRef]

GUIMARAẼS, C.; PORTO, P.; OLIVEIRA, R. and MOTA, M. (2005). Continuous decolourization of a sugar reﬁnery wastewater in a modiﬁed rotating biological contactor with Phanerochaete chrysosporium immobilized on polyurethane foam disks. Process Biochemistry, vol. 40, no. 2, p. 535-540. [CrossRef]

JIN, Z.; JIN, X. and JIN, Q. (2010). Conjugal transferring of resistance gene ptr for improvement of pristinamycin-producing Streptomyces pristinaespiralis. Applied Biochemistry and Biotechnology, vol. 160, no. 6, p. 1853-1864. [CrossRef]

KATAPODIS, P.; CHRISTAKOPOULOU, V. and CHRISTAKOPOULOS, P. (2006). Optimization of xylanase production by Thermomyces lanuginosus in tomato seed meal using response surface methodology. World Journal of Microbiology & Biotechnology, vol. 22, no. 5, p. 501-506. [CrossRef]

LI, H.; LIANG, W.Q.; WANG, Z.Y.; LUO, N.; WU, X.Y.; HU, J.M.; LU, J.Q.; ZHANG, X.Y.; WU, P.C. and LIU, Y.H. (2006). Enhanced production and partial characterization of thermostable α-galactosidase by thermotolerant Absidia sp. WL511 in solid-state fermentation using response surface methodology. World Journal of Microbiology & Biotechnology, vol. 22, no. 1, p. 1-7. [CrossRef]

LIU, J.H.; ZHANG, Y.Y.; XIA, Y.M. and SU, F. (2010). Optimization of immobilization conditions of Candida antarctica lipase based on response surface methodology. Chemical and Biochemical Engineering Quarterly, vol. 24, no. 2, p. 203-209.

MAIA, M.M.D.; HEASLEY, A.; DE MORAIS, M.M.C.; MELO, E.H.M.; MORAIS JR., M.A.; LEDINGHAM, W.M. and LIMA FILHO, J.L. (2001). Effect of culture conditions on lipase production by Fusarium solani in batch fermentation. Bioresource Technology, vol. 76, no. 1, p. 23-27. [CrossRef]

QUEZADA, M.A.; CARBALLEIRA, J.D. and SINISTERRA, J.V. (2009). Monascus kaoliang CBS 302.78 immobilized in polyurethane foam using iso-propanol as co-substrate: Optimized immobilization conditions of a fungus as biocatalyst for the reduction of ketones. Bioresource Technology, vol. 100, no. 6, p. 2018-2025. [CrossRef]

SAYYAD, S.A.; PANDA, B.P.; JAVED, S. and ALI, M. (2007). Optimization of nutrient parameters for lovastatin production by Monascus purpureus MTCC 369 under submerged fermentation using response surface methodology. Applied Microbiology and Biotechnology, vol. 73, no. 5, p. 1054-1058. [CrossRef]

VOELKER, F. and ALTABA, S. (2001). Nitrogen source governs the patterns of growth and pristinamycin production in 'Streptomyces pristinaespiralis'. Microbiology, vol. 147, no. 9, p. 2447-2459.

XU, B.; JIN, Z.H.; JIN, Q.C.; LI, N.H. and CEN, P.L. (2009). Improvement of pristinamycin production by genome shuffling and medium optimization for Streptomyces pristinaespiralis. Biotechnology and Bioprocess Engineering, vol. 14, no. 2, p. 175-179. [CrossRef]

YAMAGUCHI, T.; ISHIDA, M. and SUZUKI, T. (1999). An immobilized cell system in polyurethane foam for the lipophilic micro-alga Prototheca zopfii. Process Biochemistry, vol. 34, no. 2, p. 167-172. [CrossRef]

Note: Electronic Journal of Biotechnology is not responsible if on-line references cited on manuscripts are not available any more after the date of publication.