Improved microbial conversion of de-oiled Jatropha waste into biohydrogen via inoculum pretreatment: process optimization by experimental design approach

1 Center for Materials Cycles and Waste Management Research, National Institute for Environmental Studies, Tsukuba, Japan. 2 Research Institute on Bioengineering, Membrane Technology and Energetics, University of Pannonia, Egyetem u. 10, 8200 Veszprém, Hungary. 3 Department of Environmental Engineering, Daegu University, Republic of Korea. 4 Department of Environmental Engineering and Science, Feng Chia University, 40724 Taichung, Taiwan.


Introduction
The facts of ever increasing energy consumption, diminishing reservoirs of fossil fuels and threatening environmental problems awaked the scientists to find acceptable fuel alternatives both from the ecological and energetic points of views.As a result, various green energy carriers have been proposed, among which hydrogen is a promising candidate due to its unique characteristics (Akil and Jayanthi, 2014).However, the majority of hydrogen is currently derived from non-renewable sources such as methane conversion and oil/naphtha (Ewan and Allen, 2005).Consequently, clean technologies should be developed to make hydrogen a more attractive energy carrier.For this purpose, biological approaches are among the emerging opportunities (Hallenbeck, 2009).In fact, the dark fermentative way to produce hydrogen is one of the most extensively studied fields and is currently the most promising one when practicality is considered (Han et al., 2012;Wang et al., 2012;Diamantis et al., 2013;Sarma et al., 2013).However, utilization of abundant and inexpensive lignocellulosic wastes, such as de-oiled Jatropha waste (DJW) needs adequate microbial consortia (Fan et al., 2006).To obtain such bacterial populations, the seed pretreatment is often a key step.Nevertheless, the most appropriate technique to enrich the hydrogen-producing strains has to be found specifically for each case due to the diversities in the microbial population structure of seed sources of different origins (Mohammadi et al., 2011;Wang et al., 2011).In addition to the importance of seed pretreatment, it is also known that biotechnological hydrogen formation is influenced by the environmental conditions applied.Among them, temperature, pH and substrate concentration are most crucial (Hawkes et al., 2007;Wang and Wan, 2009a).Therefore, these factors must carefully be optimized to improve hydrogen generation e.g. by the experimental design approach (Wang and Wan, 2009b).
In our previous study, it was shown that the DJW could be successfully converted into hydrogen and methane by untreated, mixed microbial flora (Kumar and Lin, 2013).In this work, firstly, the selection of feasible seed pretreatment was attempted.Subsequently, the enhancement of hydrogen production was aimed by optimizing the substrate (i.e.DJW) concentration, pH and temperature using the experimental design approach.

Inoculum and DJW
The seed inoculum was collected from a municipal wastewater treatment plant located in central Taiwan.The main features of the inoculum were previously described elsewhere (Kumar and Lin, 2013).The cellulose (14.1%), hemicellulose (24.2%) and lignin (30.4) contents of the DJW were measured by the FIBERTEC TM analyzer as indicated in our previous study (Kumar et al., 2012).

Inoculum pretreatment methods
Several methods were tested for inoculum pretreatment.The heat shock treatment was conducted by heating the sludge in a water bath at 95 o C for 30 min.The acid pretreatment was performed by adjusting the pH of the sludge to 3.0 for 24 h.During base pretreatment, the pH of the sludge was increased to 11.0 and maintained for 1 d.The chemical treatment was carried out by adding KNO3 (1000 mg/L) to the sludge.The combination of acid and heat pretreatment involved acidification and subsequent heat shock, meanwhile the base and heat strategy was also established by exposing the sludge firstly to base and then to heat, according to the procedure mentioned above.

Experiment I: selection of efficient inoculum pretreatment method
Batch vials (holding capacity of 225 mL) with a working volume of 150 mL were used for the fermentation.The bottles contained 30 mL of seed sludge (pretreated by various ways), dried DJW substrate to a final concentration of 50 g/L, 5 mL of the nutrient solution and drops of either 1N HCl or NaOH to adjust the pH of the solution to neutral.The final working volume was made up with tap water.The composition of the nutrient solution used can be found elsewhere (Kumar and Lin, 2013).The vials were purged with argon for 5 min to provide a fully anaerobic environment.Afterwards, the batch bottles were placed in a reciprocal air-bath shaker at 150 rpm agitation rate and temperature of 55 o C. The volume and composition of the biogas formed were measured periodically.

Experiment II: statistical parameter optimization
The batch tests were performed by following the procedure given in Section 2.3, except that the batch vials used had a total and working volume of 125 mL and 60 mL, respectively.This change was made in order to reduce the amount of chemicals and substrate needed.The initial measurement conditionssubstrate concentration, temperature and pHwere set according to the experimental design matrix (Table 1).A fivelevel central composite design (CCD) and response surface methodology (RSM) were used to optimize substrate concentration, pH and temperature.After optimization, verification experiments were conducted in parallel.

Analytical procedures
The soluble metabolic products (SMPs) including volatile fatty acids (VFAs) and alcohols (e.g.ethanol an butanol) were determined by GC-FID (Shimadzu GC-14, Japan).The volume of biogas was measured by a glass syringe at room temperature and 1 atm.The gas composition was analyzed by GC-TCD (China Chromatograph 8700T).All the measurement conditions are available in our previous work (Kumar and Lin, 2013).Kinetic analysis using the modified Gompertz equation was performed as previously reported (Kumar and Lin, 2013).Microbial community analysis (PCR-DGGE) was conducted by following the protocol described by Sivagurunathan et al., (2014).

Influence of various pretreatment methods on hydrogen yield and formation rate
In Figure 1A, it can be seen that both biogas and hydrogen generation were highly dependent on the applied seed sludge pretreatment, demonstrating that selection of appropriate pretreatment is a key step for hydrogen generation.The obtained production performances are listed in Table 2, where it is shown that heat treatment was the most efficient method in comparison with the other methods employed.On the other hand, chemical, base and acid pretreatments as well as the control produced not only hydrogen but also methane.Thus, these methods were not suitable for efficient hydrogen production due to the co-generation of methane (Wang et al., 2012).The reason assumed beyond the superiority of heat-associated treatment is that it preferentially enhances the growth of the endosporeforming Clostridium species, which are considered as good hydrogenproducers (Lin et al., 2006;Li and Fang, 2007).The kinetic analyses of the various pretreatment methods are tabulated in Table 2, where it could be observed that treatment conditions significantly influenced the bacterial lag phase time.This can be explained by the distinct adaptation capabilities of the microbes to the changes in the environmental conditions caused by the pretreatments applied.

SMP distribution
Monitoring the SMPs during anaerobic fermentation is suggested to be used to judge the bioreactor's performance and consequently the appropriateness of the available hydrogen producing cultures (Dabrock et al., 1992;Khanal et al., 2004).
The concentrations of VFAs and solvents formed in the hydrogen evolution step for different sludge pretreatment methods are summarized in Figure 1B, where it is indicated that the SMPs were dominated by butyric-and acetic acids accounting for 76.6-90.7% of the total SMPs.
The other side products such as propionate, valerate, and alcohols showed less significant contributions (9.3-23.4%) to the total SMPs.The high HBu/HAc ratio indicated an efficient biohydrogen generation system, since efficient hydrogen production is usually associated with high HBu production.Similar findings were reported in other studies as well (Lin and Chang, 1999;Chen et al., 2002;Lin et al., 2006;Demirbas, 2007;Hawkes et al., 2007).
According to Figure 1B, it can be pointed out that the highest HBu/HAc ratio (3.1) could be attained via applying heat pretreatment.These research findings demonstrate that metabolic activity (e.g.SMPs release) of the microbes is related to the pretreatment method used to promote the growth of the reliable hydrogen-evolving organisms.

Optimization of process parameters affecting biohydrogen production
The main operational parameters in biohydrogen production are substrate concentration, temperature and pH.In order to evaluate their effects on hydrogen fermentation of DJW, central composite design and response surface methodology were employed.
During these experimental runs, the heat-pretreated sludge was used.The experimental design matrix along with the corresponding response values of the dependent variable HPR are shown in Table 1.
The levels of the independent variables were chosen based on the preliminary experimental results.Analysis of Variance (ANOVA) was carried out by using Statistica 8 software to get the significance of each process variable.The results of ANOVA are listed in Table 3.The impacts of the parameters scoped were ranked based on the obtained Pvalues.
Basically, a smaller P-value stands for higher influence and only factors having P-values <0.05 can be considered as statistically important ones (Guo et al., 2009).

(Eq. 1)
In the present study, according to the ANOVA results, the R 2 value of 0.986 suggested that there was a reliable agreement between the experimental data and the values predicted by the model (Chong et al., 2009).Using the results obtained the contour plots could be constructed with the two-dimensional projections of the fitted three-dimensional surfaces (Fig. 2).These graphs were intended to illustrate the effects of two independent variables on HPR while keeping the level of the third factor at its center value (Fig. 2A-C).As it appears in Figure 2A-C, all the contour plots demonstrated a round ridge running around the center point implying that the interactions between the factors had only low importance (Kim et al., 2004).This is confirmed by the data presented in Table 3 as well, where it can be observed that the interactive effects (X1X2, X1X3, X2X3) are insignificant (P-value>0.05).

A B
Moreover, it is to be concluded that the maximum HPR value could be achieved within the design boundaries.Hence, the optimal conditions yielding the highest HPR could be derived by RSM as follows: substrate concentration of 205 g/L, initial pH of 6.53 and temperature of 55.1 o C, where peak HPR value of 1.36 L H2/L-d was estimated.
Although experimental design methods have definite benefits, the statistically estimated optimum conditions need to be validated (Mu et al., 2006).Under the predicted optimal conditions, the HPR value was found as 1.42±0.03LH2/L-d.This proved that the actual optimal HPR satisfactorily matched its statistically-forecasted value (1.36 L H2/L-d) with only 4% difference between them.Therefore, the results obtained confirmed the feasibility of using the experimental design and RSM for biohydrogen process development.

C B
A comparison with other works focusing on substrate concentration, temperature and pH for enhanced biohydrogen production is given in Table 4.As a conclusion, it can be pointed out that most of these investigations were performed using the classic, so-called "one-factor-at-a-time" method.On the other hand, only a few studies were dedicated to the experimental design.The optimum value obtained in the current research is quite comparable with the results of other reports employing complex (lignocellulosic) substrate materials.

Microbial community pattern
The DGGE profile obtained using the bacterial primer set EUB968gc-UNIV1392r revealed the structural composition of the microbial communities based on the V6 region of the 16s rRNA.The DGGE profile of the hydrogen producing microbial community using DJW under optimal condition are depicted in Figure 3A.As shown, four individual distinct bands patterns representing four different strains were observed.The phylogenetic tree distribution was established using the bootstrap neighbor joining method (Saitou and Nei, 1987) as shown in Figure 3B.
The microbial load in the seed inocula enriched with de-oiled Jatropha waste was made up by Clostridium sp., and Lactobacillus sp.According to the major bands shown in Figure 3A, Clostridium sp. was the most abundant followed by Bacillus sp.The bacterial species C. thermopalmarium, C. thermobutyricum, and B. ginsengihumi were previously reported as potential hydrogen producing bacteria ( Wiegel et al., 1989;Geng et al., 2010;Walton et al., 2010).Clostridium Sp., a low G+ C content bacterium, is known to generate hydrogen along with butyrate and acetate as major SMPs (Chen et al., 2002;Levin et al., 2004).
In addition, C. thermopalamarium was reported to efficiently utilize cellulose (a major component of the DJW) to produce hydrogen.Moreover, Bacillus sp. was also reported to produce lactic acid from hemicellulose extracts (Walton et al., 2010).The inoculum source i.e. the sewage sludge selected in the present study was a rich source of hemicelluloytic and cellulolytic bacteria (Sleat et al., 1984).
Under the optimum conditions employed in the present study, Clostridium sp. and Bacillus sp. were the predominant microorganisms.while the former was more abundant.Clostridium species are known hydrogen producers with butyrate type fermentation in which acetic acid and alcohols are the minor metabolites (Levin et al., 2004).This was in agreement with the results of the SMPs analysis, suggesting that butyrate was the major metabolite observed during the fermentation, followed by acetic acid and ethanol.The PCR-DGGE based sequence analysis also revealed the presence of dominant butyrate-mediated hydrogen-producing bacteria present in the mixed cultures.

Conclusion
Various sludge pretreatment methods were tested for biohydrogen production using de-oiled Jatropha waste as feedstock.It turned out that among the methods studied, heat pretreatment was most effective in order to enrich efficient hydrogen-producing microorganisms.Moreover, central composite design was successfully used for the optimization of the most crucial operational parameters.The distribution pattern of the SMPs showed that the fermentation mainly followed the acetate-butyrate pathway.Moreover, PCR-DGGE results revealed that Clostridium sp. were majorly present under the optimal conditions.

Fig. 2 .
Fig.2.(A) Contour plots showing the effects of initial pH and substrate concentration on hydrogen production rate (HPR), (temperature: 55 o C), (B) Contour plots showing the effects of initial pH and temperature on hydrogen production rate (HPR), (substrate concentration: 200 g/L), (C) Contour plots showing the effects of temperature and substrate concentration on hydrogen production rate (HPR), (pH: 6.5).

Fig. 3 .
Fig.3.A:The DGGE profile of the hydrogen-producing microbial community under optimum conditions using DJW, and B: Phylogenetic tree showing the relatedness of the sequences identified in the mixed cultures.The tree was constructed based on maximum composite likelihood method using the neighbor-joining algorithm with 1,000 bootstrapping.E.coli was selected as the outgroup species.The scale bar represents 0.1 substitutions per nucleotide position.Numbers at the nodes are the bootstrap values.

Table 1 .
Central composite design for optimizing the hydrogen production rate.
SS : sum of squares df : degree of freedom MS : mean square

Table 2 .
Biogas production performance under various pretreatment conditions.

Table 4 .
Comparison of the results with relevant literature data.