Bioresource Technology 141 (2013) 233–239

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Metabolic flux network analysis of fermentative hydrogen production:Using Clostridium tyrobutyricum as an example

0960-8524/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.03.141

⇑ Corresponding author at: Department of Environmental Engineering, NationalCheng Kung University, No. 1, University Road, Tainan 701, Taiwan. Tel.: +886 62757575x65837; fax: +886 6 2752790.

E-mail address: whang@mail.ncku.edu.tw (L.-M. Whang).

Hai-Hsuan Cheng a, Liang-Ming Whang a,b,c,⇑, Che-An Lin a, I-Chun Liu a, Chao-Wei Wu a

a Department of Environmental Engineering, National Cheng Kung University, No. 1, University Road, Tainan 701, Taiwanb Sustainable Environment Research Center (SERC), National Cheng Kung University, No. 1, University Road, Tainan 701, Taiwanc Research Center for Energy Technology and Strategy (RCETS), National Cheng Kung University, No. 1, University Road, Tainan 701, Taiwan

h i g h l i g h t s

�Metabolic fluxes for C. tyrobutyricum in H2 production from glucose and lactate/acetate fermentation.� MFA results indicate that HRT presents a significant impact on the flux of hydrogen production from glucose.� At HRT between 4 and 18 h, increase of HRT increased hydrogen production but decreased lactate production.� At HRT <4 h decrease of HRT increased hydrogen production but decreased lactate production.� The flux for lactate, butyrate and acetate affects H2 production, due to their impacts on NADH, ferredoxin and ATP.

a r t i c l e i n f o

Article history:Available online 28 March 2013

Keywords:GlucoseLactate/acetateMetabolic fluxHydrogen productionHydraulic retention time

a b s t r a c t

This study applies metabolic flux network analysis (MFA) to evaluate the metabolic flux of fermentativehydrogen production (FHP) with the use of Clostridium tyrobutyricum fed with either glucose or lactate/acetate as substrates. The MFA results suggest that hydraulic retention time (HRT) presents significantimpact on hydrogen production from glucose. At HRT between 4 and 18 h, increase of HRT increasedhydrogen production but decreased lactate production, while at HRT below 4 h decrease of HRT increasedhydrogen production but decreased lactate production. The flux for lactate, butyrate and acetate seemedto affect H2 production, due presumably to their impacts on the balance of NADH, ferredoxin and ATP. It issuggested that the MFA can be a useful tool to provide valuable information for optimization and designof the fermentative hydrogen production process.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Development of a clean and renewable energy source has be-come an urgent need in modern times because of the depletionof fossil fuel, and also the possible correlation between global cli-mate change and its consumption (Jo et al., 2008). Hydrogen(H2), with the highest energy density and generating only waterafter combustion, is considered as one of the most promising alter-natives (Das and Veziroglu, 2001; Fang et al., 2006). Besides itscleanliness, in terms of no greenhouse gas (GHG) production dur-ing combustion, hydrogen has many benefits, such as being usedas the fuel for a conventional fuel cell, where electricity can be pro-duced without combustion. Among hydrogen production pro-cesses, fermentative hydrogen production (FHP) is an attractiveroute since a variety of carbon sources can be used as the

feedstock, such as starch (Liu and Shen, 2004), cellulose (Lay,2001), sucrose (Chen et al., 2001), glucose (Lee et al., 2008; Liet al., 2010; Whang et al., 2011), xylose (Zhu and Yang, 2004),and mixture of biomass and organic wastes (Das and Veziroglu,2001; Hawkes et al., 2008; Juang et al., 2011; Cheng et al., 2012).

It is widely known that various volatile fatty acids (VFAs) and/oralcohols are produced through dark fermentative hydrogen pro-duction. The theoretical H2 yield is 4 mol H2/mol glucose whenacetate is the sole product while a maximum yield of 2 mol H2/mol glucose was obtained when butyrate as the end product. Themaximum yield of 12 mol H2/mol glucose which should beachieved from the complete conversion of glucose to H2 and CO2

has never been attained in known biological system until now(Fang et al., 2006). The low conversion yield might presumablydue to the natural consequence that microorganism tend to pro-duce cell biomass evolutionarily instead of hydrogen, and to thecompeting reactions such as hydrogen consumption or divisionto other products (Hallenbeck and Benemann, 2002). In practice,H2 yield is related to the metabolic pathways and end-products,like when butyrate-to-acetate ratio (B/A ratio) and lactate yield

234 H.-H. Cheng et al. / Bioresource Technology 141 (2013) 233–239

have been proved to be related to H2 production. Therefore, it ap-pears that there are some critical metabolic activities in H2 fermen-tation which may be ignored in the currently proposed metabolicreactions and equations (Cai et al., 2011) and the potential ofFHP anaerobes has not totally explored and their metabolic net-work requires to be understood.

Possible fermentation pathways of H2 production by Clostridiumsp. utilizing glucose have been reported in several researches (Sri-dhar and Eiteman, 2001; Zhu and Yang, 2004; Cai et al., 2010), itseems that Clostridium sp. share similar pathways. Clostridium tyro-butyricum is one of saccharolytic clostridia which is wildly incu-bated in butyric acid production (Zhu et al., 2002; Wu and Yang,2003) and H2 fermentation (Jo et al., 2008; Whang et al., 2011)since it has many advantages over other species, including simplemedium for cell growth and relatively high product purity andyield (Michel-Savin et al., 1990). However, the impact of incuba-tion factors such as hydraulic retention time (HRT) or initial sub-strate concentration on metabolic flux distribution in thebiological H2 fermentative system applying C. tyrobutyricum is stillnot well-known. The quantification of the pathway fluxes becomesimportant for both bioreactor operation and metabolic engineeringin order to enhance H2 production. Metabolic flux analysis (MFA)has been considered as a useful methodology to calculate intracel-lular fluxes from extracellular fluxes, and would be helpful for thecomprehension of flux regulation in the H2 fermentation (Manishet al., 2007; Cai et al., 2011).

In this study, the metabolic network model for the fermentativehydrogen production of C. tyrobutyricum was developed. The MFAmethodology was employed to illustrate the possible mechanismand flux distribution during glucose and lactate/acetate metabo-lism of C. tyrobutyricum, and to study the effect of HRT and initialsubstrate concentration on the intracellular fluxes under a con-stant temperature and pH.

2. Methods

2.1. Bioreactor operation

Two continuous stirred tank reactor-type (CSTR) bioreactors,fed with glucose (GA) and lactate/acetate (LA) respectively, wereconducted in the present study. For the GA bioreactor, an isolatedpure culture of C. tyrobutyricum FYa102 from previous studies (Liet al., 2010; Whang et al., 2011) was applied as seeding sludge,and 12,000 mg/L of glucose, 1400 mg/L of ammonium chloride,and 360 mg/L of peptone were fed as a substrate. The seedingmicroorganism for LA bioreactor was obtained from a lab-scalecontinuous flow bioreactor fed with wasted residue from a bio-ethanol fermentation process (Juang et al., 2011) and C. tyrobutyr-icum was considered the major hydrogen-producing species in LAbioreactor (Wu et al., 2012). The predetermined concentration oflactate and acetate were served with a ratio of 10–3 as a substrate,and ammonium chloride and peptone were also added. The totalvolume of both bioreactors was 4 L and the working volume was2 L. The feedstock solutions were stored at 4 �C and continuouslyfed into the bioreactors using a peristaltic pump. Each liter of theinfluent feed was supplemented with the following chemicals (inmg): resazurin, 0.175; CaCl2�6H2O, 32.32; MgCl2�6H2O, 232.26;KCl, 167.81; MnCl2�4H2O, 63.87; CoCl2�6H2O, 3.87; H3BO3, 0.74;CuCl�2H2O, 0.35; Na2MoO4�2H2O, 0.33; ZnCl2, 0.27; FeCl2�4H2O,10.62; sodium thioglycolate, 217.35; KH2PO4, 119 (Liu et al.,2010; Whang et al., 2011). The pH in both bioreactors was con-trolled at 6.0 ± 0.1 with a pH controller by the addition of 5%H3PO4 and 10% NaOH throughout the experiments. The amountof biogases produced from both bioreactors was measured with awet-gas flow meter (Shinagawa W-NK-0.5B, Tokyo, Japan).

2.2. Sample preparation and analysis

Biogases collected from the bioreactors and batches were ana-lyzed by gas chromatograph (China GC 8900, Taipei, Taiwan)equipped with a thermal conductivity detector (TCD). A 2 m stain-less column packed with Hayesep Q (60/80 mesh) was installed inthe oven and the operational temperatures of the injection port,the oven, and the detector were set at 90, 60 and 60 �C, respec-tively. Nitrogen was used as the carrier gas at a flow rate of15 mL/min. Volatile fatty acids (VFAs), alcohols, and sugars weredetermined using high pressured liquid chromatograph (HitachiD2000 system, Tokyo, Japan) equipped with an ICSep COREGEL-87H3 column (7.8 mm � 300 mm) in a 42 �C oven and an RI Range16� detector. The eluent used was 0.008 N sulfuric acid at a flowrate of 0.6 mL/min. The pH, ammonium nitrogen (NH4

+-N), organicnitrogen, and volatile suspended solids (VSS) were measuredaccording to the standard methods (APHA, 1998).

2.3. MFA and CellNetAnalyzer

MFA is a modeling tool based on stoichiometric reactions analysisand mass balances (Cai et al., 2011). It can provide information aboutthe intracellular metabolism conversion and metabolites exchangewith extracellular environment despite the overwhelming complex-ity of cellular responses. The incoming fluxes of each metabolitesnode will balance the outgoing fluxes under the assumption of con-structed metabolic network and pseudo-steady state (Cai et al.,2011). The CellNetAnalyzer, a software package of the MATLAB(Mathworks, Inc.) for metabolic networks as well as signaling andregulation networks analysis, was used to estimate and predict thestoichiometry of different products during H2 fermentation.

2.4. Metabolic network of C. tyrobutyricum

The glucose and lactate/acetate metabolic fluxes network in H2

production developed for MFA was evaluated and modified fromregular biochemical pathways of glucose metabolism reported forClostridium species (Sridhar and Eiteman, 2001; Cai et al., 2010).The network of C. tyrobutyricum is presented in Fig. 1, and theabbreviation of the metabolites and related reactions are summa-rized in Appendix A and B, respectively. The network was con-structed using 21 reactions (J1–J20, and biomass synthesis) and16 intracellular metabolites for the GA bioreactor, including glu-cose utilization (J1) and production of lactate, formate, acetate,butyrate, ethanol, H2 and CO2 (J6, J7, J8, J9, J10, J11, J13). For the LA bio-reactor, there were 18 reactions (J�2–J�6, J8, J�9, J10, J12–J20, and bio-mass synthesis) and 16 intracellular metabolites in the network,since glucose, ethanol and formate were not detected duringexperiments. The model construction and MFA analysis were per-formed using the CellNetAnalyzer, and the steady state specific up-take/production/growth rate (mmol/g VSS/h) at different HRTswere used as constraints in the model. Mass balance for carbonwas calculated to examine the reliability of output data and allthe flux distribution were normalized by the basis of glucose orlactate/acetate uptake rate and expressed as percentage.

3. Results and discussion

3.1. Node analysis of GA bioreactor

For the GA bioreactor, the metabolic flux distribution under theHRT between 2 and 18 h are shown in Fig. 2. The errors of carbonmass balance of each run were less than 5%. In general, HRT pre-sented a significant impact on H2 production. H2 production be-came unfavorable when HRT approached to 4 h which was thedoubling time of C. tyrobutyricum, as found in batch experiments.

Fig. 1. Metabolic network of C. tyrobutyricum in H2 production.

Fig. 2. Flux distribution of C. tyrobutyricum using glucose as substrate under different HRTs.

H.-H. Cheng et al. / Bioresource Technology 141 (2013) 233–239 235

In order to further understand the detailed flux distribution,node analysis was applied. Table 1 and Fig. 2 summarize the resultsof node analysis for the MFA of C. tyrobutyricum fed with glucose.

The first hinge of this network was the intermediate metaboliteG6P, the flux divided into two pathways at this node, recognizedas the Embden–Mayerhof–Paranas (EMP) pathway (J2) and the

Table 1Node analysis of C. tyrobutyricum fed with glucose.

HRT (h) Flux at G6P node Flux at PYR nodeJ1 = 100 J5 = 100

J2 J14 J2/J14 J6 J7 J8 J8/(J6 + J7) J8/J7

18 96.40 3.60 26.81 0.00 15.50 84.50 5.45 5.458 93.59 6.41 14.61 21.40 20.39 58.21 1.39 2.856 99.01 0.98 100.55 52.05 7.00 40.95 0.69 5.854 96.36 3.64 26.49 73.96 5.21 20.83 0.26 4.003 109.19 �9.21 70.93 20.59 8.49 0.09 0.412.5 86.46 13.53 6.39 58.74 31.59 9.67 0.11 0.312 80.63 19.35 4.17 47.05 37.17 16.73 0.20 0.45

HRT (h) Flux at PEP node Flux at acetyl-CoA nodeJ4 = 100 J7 + J8 = 100

J5 J12 J5/J12 J9 J10 J11 J9/J10

18 99.94 0.05 1894.14 7.82 45.28 1.60 0.178 99.94 0.05 1829.40 11.15 21.66 45.51 0.516 99.94 0.06 1728.75 15.68 26.49 31.31 0.594 99.93 0.06 1643.53 18.93 22.15 36.69 0.853 99.84 0.15 667.34 54.45 9.90 25.59 5.502.5 99.83 0.16 639.72 38.14 1.33 59.08 28.662 99.81 0.18 566.74 45.33 14.86 23.08 3.05

236 H.-H. Cheng et al. / Bioresource Technology 141 (2013) 233–239

Pentose Phosphate (PP) pathway (J14). The flux ratio of J2/J14 rangedfrom 15 to 100 when HRT >4 h to below 10 when HRT <4 h, sug-gesting that lower HRT condition was suitable for PP pathway.However, EMP pathway was the main route for C. tyrobutyricumduring glucose fermentation. At the PEP node, the fraction of fluxmostly was toward to pyruvate (J5) instead of OAA (J12) underevery HRT conditions, with only a slightly increase of OAA distribu-tion to 0.17% when HRT decreased to 2 h.

In glucose metabolism network, pyruvate was another impor-tant intermediate. Pyruvate could be converted either to lactate(J6) or acetyl-CoA, but the production of acetyl-CoA could eitherco-generate formate (J7) or reduced ferredoxin and CO2 (J8). Amongthe three routes, J8 was a critical route responsible for productionof hydrogen from reduced ferredoxin through hydrogenase.Accordingly, it would be unfavorable to H2 production if the fluxdistributed to the production of lactate (J6) or formate (J7). The ra-tio of J8/(J6 + J7) decreased from 5.5 to approximately 0.1 with a de-creased HRT, indicating considerable amount of flux distributedinto the routes which were unfavorable to H2 production at lowerHRT conditions especially lower than 4 h. The highest flux of lac-tate distribution was attained at HRT of 4 h, which was closed tothe doubling time of C. tyrobutyricum (Whang et al., 2011). The for-mation of lactate was regulated by two enzymes, lactate dehydro-genase (LDH) and inducible NAD-independent lactatedehydrogenase (iLDH). LDH converts pyruvate to lactate whileiLDH reverses the reaction, and increase or decrease of their activ-ity is attributed to the ratio of NADH/NAD+ and pH (Wu and Yang,2003). In addition, microbial growth rate may also affect the activ-ities of LDH and iLDH.

Formate production was at a minimum level of 5.2% at HRT of4 h, while the maximum flux for formate of 37.2% appeared atHRT of 2 h. In general, a lower HRT promoted a higher formate fluxdistribution which generated acetyl-CoA without generating re-duced ferredoxin, resulting in the ratio of J8/J7 lower than at HRT<4 h. At the node of acetyl-CoA, three major external metabolitesof glucose fermentation were produced, including acetate, butyrateand ethanol. Ethanol distribution varied between 23% and 59%, ex-cept for a minimum value of 1.6% occurred at HRT of 18 h, indicat-ing a lower NADH2 consumption. The ratio of butyrate to acetate(B/A, J10/J9) has been considered as an important indicator for H2

production (Liu et al., 2010). In this study, the ratio of J10/J9 washigher than 1 at HRT >4 h, and this ratio decreased to lower than1 when HRT was lower than 4 h.

3.2. MFA of LA bioreactor

Lactate utilization by Clostridium sp. has been studied when car-bohydrates were present (Jo et al., 2008; Juang et al., 2011; Chenget al., 2012), but limited information was available for using lactateas a sole energy source during batch experiments. Matsumoto andNishimura (2007) isolated three Clostridium species from slurrysolution, and concluded that these species can produce hydrogenfrom acetate and lactate. The presence of acetate was found to benecessary during lactate fermentation, but the role of acetate, how-ever, was still unclear. Diez-Gonzalez et al. (1995) suggested thatacetate may serve as an electron acceptor during lactate utilizationand acetate was not the limiting substrate (Woolford, 1984).

The metabolic flux distribution of C. tyrobutyricum in LA biore-actor under different HRT and lactate/acetate concentration issummarized in Fig. 3. The flux was presented in the order of volu-metric loading rate (VLR), ranging from 21.8 to 83.5 kg-COD/m3/day. The maximum distribution of H2 production (J13) was 47% atVLR of 27.8 kg-COD/m3/day, which was under the HRT of 18 hand lactate/acetate concentration of 15,000/4500 mg/L. Under thesame condition, the highest ratio for lactate/acetate utilizationwas found to be 2.78, indicating the lowest acetate consumption.A lower acetate consumption indicated a lower ATP requirementfor activation energy from acetate to acetyl-CoA, and thus, resultedin a higher NADH production during conversion of lactate to pyru-vate. This can be confirmed by a lowest flux distribution of 4.72%from FdH2 to NADH and a maximum flux distribution for H2 pro-duction (47%). The flux distribution for butyrate (J10) at differentHRT and substrate concentration was similar ranging from 38.4%to 43.3%, since butyrate was the major end product of lactate andacetate fermentation (Diez-Gonzalez et al., 1995). In the pathwayof lactate utilization, butyrate production was the only ATP gener-ation step and the generated ATP was further utilized for activationof acetate.

Table 2 summarizes the results of node analysis for the MFA ofC. tyrobutyricum fed with lactate/acetate. At the node of pyruvate,more than 60% of the fraction flux was toward to Acetyl-CoA withco-generation of reduced ferredoxin and CO2 (J8). About 20% to 35%of flux distributed back to PEP (J�5) and further produced materialsfor biomass synthesis via gluconeogenesis pathway. Two fluxes,including pyruvate (J8) and acetate (J�9), flowed into the acetyl-CoA node and butyrate (J10) was formed as the end-product. Thehighest ratio of J8–J�9 was found to be 1.95, at which lowest acetate

Fig. 3. Flux distribution of C. tyrobutyricum using lactate/acetate as substrate under different VLRs.

Table 2Node analysis of C. tyrobutyricum fed with lactate/acetate.

VLR (kg-COD/m3/day) Flux at PYR node Flux at acetyl-CoA nodeJ�6 = 100 J10 = 100

J�5 J8 J8/J�5 J8 J�9 J8/J�9

21.8 39.62 60.38 1.52 92.05 107.97 0.8527.8 29.69 70.31 2.37 132.32 67.69 1.9541.7 34.49 65.51 1.90 109.39 72.93 1.2155.6 19.65 80.35 4.09 127.08 90.62 1.7483.5 26.62 73.38 2.76 130.58 69.43 1.88

VLR (kg-COD/m3/day) Flux at PEP nodeJ�5 = 100

J�4 J12 J�4/J12

21.8 99.86 0.13 780.4427.8 99.89 0.10 1018.6541.7 99.92 0.07 771.3355.6 99.86 0.13 1345.1283.5 99.88 0.11 917.64

H.-H. Cheng et al. / Bioresource Technology 141 (2013) 233–239 237

was consumed with a higher production of reduced ferredoxin,resulting in a highest hydrogen production flux.

3.3. Effect of metabolites on H2 production under different HRTs

In Fig. 2, the maximum flux distribution of C. tyrobutyricum fedwith glucose for H2 production was found to be 161.4% at HRT of18 h. The H2 production flux decreased with a decreased HRT from18 to 4 h, and the production flux increased to 94.2% at HRT of 2 h.The lowest flux for ethanol production was also found at HRT of18 h with a value of 3.2%. It is possible that a lower ethanol produc-tion reserves more NADH for other flux since NADH is required forethanol formation. These reserved NADH can be used for butyrateproduction (J10) or ferredoxin reduction (J20), and thus promotes H2

production. Additionally, an increased butyrate production may in-

crease the ATP formation which can be used for activation of ace-tate, and thus, promoted the lactate utilization (Diez-Gonzalezet al., 1995). This was also confirmed by the 0% lactate flux ob-served at HRT of 18 h. Furthermore, previous studies have indi-cated that lactate accumulation in lactate/acetate-fed bioreactorscan be attributed to a relatively slow growth of Clostridium sp. onlactate and acetate compared with on glucose (Diez-Gonzalezet al., 1995; Jo et al., 2008; Whang et al., 2011). Nevertheless, anegative correlation between lactate flux and H2 production in glu-cose metabolism under different HRTs is illustrated in Fig. 4A andthis relationship agreed with the statement that lactate productionreduces H2 production due presumably to the competition ofNADH (Heyndrickx et al., 1990; Lee et al., 2009; Wang et al., 2009).

The B/A ratio has been considered as an indicator in evaluatingH2 production, with a higher B/A ratio corresponding to a higher H2

A

B

Fig. 4. The relationship between H2 flux and lactate flux (A) or B/A ratio (B) underdifferent HRTs using C. tyrobutyricum fed with glucose.

238 H.-H. Cheng et al. / Bioresource Technology 141 (2013) 233–239

yield (Lee and Rittmann, 2009; Liu et al., 2010). In our study, the H2

production flux also showed a positive correlation with the ratio ofbutyrate flux to acetate flux (J10/J9) as indicated in Fig. 4B. Follow-ing the metabolism of glucose for fermentative hydrogen produc-tion, theoretically, 4 or 2 mol of hydrogen are produced byconverting 1 mol of glucose to 2 mol of acetate or 1 mol of buty-rate, respectively. In the case, a decrease in the B/A ratio tendedto decrease hydrogen production, and therefore, the stoichiometryapparently disagreed with the correlation found in this study. Pos-sible explanations for the above disagreement have been discussedin previous study (Cheng et al., 2012) including the metabolism oflactate and acetate for butyrate and hydrogen production. Thismechanism often took place in the systems fed with organicwastes (Cavinato et al., 2011; Juang et al., 2011; Cheng et al.,2012), but in this study, it seemed that this mechanism may alsooccur in the system fed with glucose as the sole substrate.

3.4. Metabolic flux analysis on biological H2 production

In order to achieve a higher H2 production yield, optimization ofcellular metabolic pathways is necessary and this requires anextensive analysis and understanding of metabolic pathways andtheir regulation in the H2-producing organisms (Oh et al., 2008).Manish et al. (2007) were the pioneers applying MFA to biologicalhydrogen production with Escherichia coli and indicated that themutant strain without ldh may increase hydrogen production.Based on a metabolic-pathway model for Citrobacter amalonaticusY19, Oh et al. (2008) suggested that a hydrogen yield of 8.7 mol/mol glucose can be possible if glucose metabolism was directedto the PP pathway and NAD(P)-linked hydrogenase was used toproduce hydrogen. Cai et al. (2010) developed a metabolic networkmodel of Clostridium butyricum W5 and reported that pH conditionwas a significant factor on hydrogen yield than initial glucose con-centration according to MFA. In the following studies, by blockingthe butyrate formation pathway, Cai et al. (2011) demonstrated adecreased hydrogen yield accompanied with an increased yieldof ethanol using the hbd-deficient strain of W5. Furthermore, the

aad-deficient mutant of W5, by inactivating the ethanol formationpathway, attained an enhanced lactate production but a decreasedhydrogen generation (Cai et al., 2013). By applying MFA on resultsobtained from fed-batch experiments, Jiang et al. (2013) indicatedthat pH condition was critical to H2 production yield, with a max-imum production of 3.9 mol-H2/mol-glucose by a butyrate over-producing C. tyrobutyricum mutant. To date, MFA has been appliedfor biological hydrogen production mostly using results obtainedfrom batch or fed-batch experiments. The current study, for thefirst time, applied MFA for biological hydrogen production usingexperimental results obtained from continuous flow bioreactors.Furthermore, this study evaluated the impacts of HRTs or dilutionrates on biological hydrogen production from two different sub-strates, glucose or lactate/acetate. This work provided helpfulinformation on optimization of operational parameters and devel-opment of molecular and metabolic engineering to improve yieldand productivity of biological H2 production.

4. Conclusion

The metabolic fluxes network of C. tyrobutyricum in H2 produc-tion from glucose and lactate/acetate fermentation was evaluatedin this study. MFA results indicate that HRT presents a significantimpact on the metabolic flux of hydrogen production from glucose.At HRT between 4 and 18 h, increase of HRT increased hydrogenproduction but decreased lactate production, while at HRT below4 h decrease of HRT increased hydrogen production but decreasedlactate production. The flux for lactate, butyrate and acetateseemed to affect H2 production, due presumably to their impactson the balance of NADH, ferredoxin and ATP.

Acknowledgements

The authors would like to acknowledge the financial supportfrom the National Science Council of Taiwan under Grant NSC98-3114-E-006-013, NSC 100-3113-E-006-017, and NSC 101-3113-E-006-016.

Appendix A. Metabolites used in metabolic flux model

Abbreviation

Compound

AcCoA

Acetyl CoA CO2 Carbon dioxide E4P D-Erythrose 4-phosphate Eth Ethanol F6P D-Fructose 6-phosphate FdH2 Reduced ferredoxin G3P Glyceraldehyde 3-phosphate G6P Glucose 6-phosphate Glu Glucose HAc Acetate HBu Butyrate HFo Formate HLa Lactate NADH Reduced-nicotinamide adenine dinucleotide OAA Oxaloacetate PEP Phosphoenolpyruvate PYR Pyruvate R5P D-Ribose 5-phosphate Ru5P D-Ribulose 5-phosphate S7P D-Sedoheptulose 7-phosphate X5P Xylulose 5-phosphate

H.-H. Cheng et al. / Bioresource Technology 141 (2013) 233–239 239

Appendix B. Reactions used in metabolic flux model

Reaction no.

Reaction Reverse

J1

Glu + ATP ? G6P J2 G6P ? F6P J�2

J3

F6P + ATP ? 2 G3P J�3

J4

G3P ? ATP + NADH + PEP + H2O J�4

J5

PEP ? ATP + PYR J�5

J6

NADH + Pyr ? HLa J�6

J7

PYR + CoA ? AcCoA + HFo J8 PYR + CoA ? AcCoA + CO2 + FdH2

J9

AcCoA ? ATP + CoA + HAc J�9

J10

2 AcCoA + 2 NADH ? ATP + HBu J11 AcCoA + 2 NADH ? CoA + Eth J12 PEP + CO2 ? OAA J13 FdH2 ? H2

J14

G6P + H2O ? 2 NADH + CO2 + Ru5P J15 Ru5P ? X5P J16 Ru5P ? R5P J17 R5P + X5P ? G3P + S7P J18 G3P + S7P ? F6P + E4P J19 E4P + X5P ? F6P + G3P J20 NADH ? FdH2

Biomass formation0.0016 G6P + 0.0055 F6P + 0.0547 G3P + 0.7957 ATP + 0.0028

PEP + 0.0115 AcCoA + 0.0371 OAA + 0.073 R5P

References

American Public Health Association, 1998. Standard Methods for the Examinationof Water and Wastewater. American Public Health Association, Washington,DC, USA.

Cai, G., Jin, B., Saint, C., Monis, P., 2010. Metabolic flux analysis of hydrogenproduction network by Clostridium butyricum W5: effect of pH and glucoseconcentrations. Int. J. Hydrogen Energy 35, 6687–6690.

Cai, G., Jin, B., Monis, P., Saint, C., 2011. Metabolic flux network and analysis offermentative hydrogen production. Biotechnol. Adv. 29 (4), 375–387.

Cai, G., Jin, B., Monis, P., Saint, C., 2013. A genetic and metabolic approach toredirection of biochemical pathways of Clostridium butyricum for enhancinghydrogen production. Biotechnol. Bioeng. 110 (1), 338–342.

Cavinato, C., Bolzonella, D., Fatone, F., Cecchi, F., Pavan, P., 2011. Optimization oftwo-phase thermophilic anaerobic digestion of biowaste for hydrogen andmethane production through reject water recirculation. Bioresour. Technol. 102(18), 8605–8611.

Chen, C.C., Lin, C.Y., Chang, J.S., 2001. Kinetics of hydrogen production withcontinuous anaerobic cultures utilizing sucrose as the limiting substrate. Appl.Microbiol. Biotechnol. 57 (1), 56–64.

Cheng, H.H., Whang, L.M., Wu, C.W., Chung, M.C., 2012. A two-stage bioprocess forhydrogen and methane production from rice straw bioethanol residues.Bioresour. Technol. 113, 23–29.

Das, D., Veziroglu, T.N., 2001. Hydrogen production by biological processes: asurvey of literature. Int. J. Hydrogen Energy 26 (1), 13–28.

Diez-Gonzalez, F., Russell, J., Hunter, J., 1995. The role of an NAD-independentlactate dehydrogenase and acetate in the utilization of lactate by Clostridiumacetobutylicum strain P262. Arch. Microbiol. 164 (1), 36–42.

Fang, H.H.P., Li, C., Zhang, T., 2006. Acidophilic biohydrogen production from riceslurry. Int. J. Hydrogen Energy 31 (6), 683–692.

Hallenbeck, P.C., Benemann, J.R., 2002. Biological hydrogen production;fundamentals and limiting processes. Int. J. Hydrogen Energy 27 (11–12),1185–1193.

Hawkes, F.R., Forsey, H., Premier, G.C., Dinsdale, R.M., Hawkes, D.L., Guwy, A.J.,Maddy, J., Cherryman, S., Shine, J., Auty, D., 2008. Fermentative production ofhydrogen from a wheat flour industry co-product. Bioresour. Technol. 99 (11),5020–5029.

Heyndrickx, M., De Vos, P., De Ley, J., 1990. H2 production from chemostatfermentation of glucose by Clostridium butyricum and Clostridiumpasteurianum in ammonium-and phosphate limitation. Biotechnol. Lett. 12(10), 731–736.

Jiang, L., Song, P., Zhu, L., Li, S., Hu, Y., Fu, N., Huang, H., 2013. Comparison ofmetabolic pathway for hydrogen production in wild-type and mutantClostridium tyrobutyricum strain based on metabolic flux analysis. Int. J.Hydrogen Energy 38 (5), 2176–2184.

Jo, J.H., Lee, D.S., Park, D., Park, J.M., 2008. Biological hydrogen production byimmobilized cells of Clostridium tyrobutyricum JM1 isolated from a food wastetreatment process. Bioresour. Technol. 99 (14), 6666–6672.

Juang, C.P., Whang, L.M., Cheng, H.H., 2011. Evaluation of bioenergy recoveryprocesses treating organic residues from ethanol fermentation process.Bioresour. Technol. 102 (9), 5394–5399.

Lay, J.J., 2001. Biohydrogen generation by mesophilic anaerobic fermentation ofmicrocrystalline cellulose. Biotechnol. Bioeng. 74 (4), 280–287.

Lee, H.S., Rittmann, B.E., 2009. Evaluation of metabolism using stoichiometry infermentative biohydrogen. Biotechnol. Bioeng. 102 (3), 749–758.

Lee, H., Salerno, M., Rittmann, B., 2008. Thermodynamic evaluation on H2

production in glucose fermentation. Environ. Sci. Technol. 42 (7), 2401–2407.

Lee, H.S., Krajmalinik-Brown, R., Zhang, H., Rittmann, B.E., 2009. An electron-flowmodel can predict complex redox reactions in mixed-culture fermentativeBioH2: microbial ecology evidence. Biotechnol. Bioeng. 104 (4), 687–697.

Li, S.L., Whang, L.M., Chao, Y.C., Wang, Y.H., Wang, Y.F., Hsiao, C.J., Tseng, I.C., Bai,M.D., Cheng, S.S., 2010. Effects of hydraulic retention time on anaerobichydrogenation performance and microbial ecology of bioreactors fed withglucose-peptone and starch-peptone. Int. J. Hydrogen Energy 35 (1), 61–70.

Liu, G.Z., Shen, J.Q., 2004. Effects of culture and medium conditions on hydrogenproduction from starch using anaerobic bacteria. J. Biosci. Bioeng. 98 (4), 251–256.

Liu, I.C., Whang, L.M., Ren, W.J., Lin, P.Y., 2010. The effect of pH on the production ofbiohydrogen by Clostridia: thermodynamic and metabolic considerations. Int. J.Hydrogen Energy 36 (1), 439–449.

Manish, S., Venkatesh, K., Banerjee, R., 2007. Metabolic flux analysis of biologicalhydrogen production by Escherichia coli. Int. J. Hydrogen Energy 32 (16), 3820–3830.

Matsumoto, M., Nishimura, Y., 2007. Hydrogen production by fermentation usingacetic acid and lactic acid. J. Biosci. Bioeng. 103 (3), 236–241.

Michel-Savin, D., Marchal, R., Vandecasteele, J.P., 1990. Butyric fermentation:metabolic behaviour and production performance of Clostridium tyrobutyricumin a continuous culture with cell recycle. Appl. Microbiol. Biotechnol. 34 (2),172–177.

Oh, Y.K., Kim, H.J., Park, S., Kim, M.S., Ryu, D.D.Y., 2008. Metabolic-flux analysis ofhydrogen production pathway in Citrobacter amalonaticus Y19. Int. J. HydrogenEnergy 33 (5), 1471–1482.

Sridhar, J., Eiteman, M.A., 2001. Metabolic flux analysis of Clostridiumthermosuccinogenes. Appl. Biochem. Biotechnol. 94 (1), 51–69.

Wang, X., Monis, P.T., Saint, C., Jin, B., 2009. Biochemical kinetics of fermentativehydrogen production by Clostridium butyricum W5. Int. J. Hydrogen Energy 34(2), 791–798.

Whang, L.M., Lin, C.A., Liu, I.C., Wu, C.W., Cheng, H.H., 2011. Metabolic and energeticaspects of biohydrogen production of Clostridium tyrobutyricum: the effects ofhydraulic retention time and peptone addition. Bioresour. Technol. 102, 8378–8383.

Woolford, M.K., 1984. The Silage Fermentation. Marcel Deccer, Inc., New York, USA.Wu, Z., Yang, S.T., 2003. Extractive fermentation for butyric acid production from

glucose by Clostridium tyrobutyricum. Biotechnol. Bioeng. 82 (1), 93–102.Wu, C.W., Whang, L.M., Cheng, H.H., Chan, K.C., 2012. Fermentative biohydrogen

production from lactate and acetate. Bioresour. Technol. 113, 30–36.Zhu, Y., Yang, S.T., 2004. Effect of pH on metabolic pathway shift in fermentation of

xylose by Clostridium tyrobutyricum. J. Biotechnol. 110 (2), 143–157.Zhu, Y., Wu, Z., Yang, S.T., 2002. Butyric acid production from acid hydrolysate of

corn fibre by Clostridium tyrobutyricum in a fibrous-bed bioreactor. ProcessBiochem. 38 (5), 657–666.