Open Access

Hydrogen production and microbial kinetics of Clostridium termitidis in mono-culture and co-culture with Clostridium beijerinckii on cellulose

AMB Express20177:84

Received: 23 June 2016

Accepted: 24 September 2016

Published: 20 April 2017


Cellulose utilization by hydrogen producers remains an issue due to the low hydrogen yields reported and the pretreatment of cellulose prior to fermentation requires complex and expensive steps. Clostridium termitidis is able to breakdown cellulose into glucose and produce hydrogen. On the other hand, Clostridium beijerinckii is not able to degrade cellulose but is adept at hydrogen production from glucose; therefore, it was chosen to potentially enhance hydrogen production when co-cultured with C. termitidis on cellulose. In this study, batch fermentation tests were conducted to investigate the direct hydrogen production enhancement of mesophilic cellulolytic bacteria C. termitidis co-cultured with mesophilic hydrogen producer C. beijerinckii on cellulose at 2 g l−1 compared to C. termitidis mono-culture. Microbial kinetics parameters were determined by modeling in MATLAB. The achieved highest hydrogen yield was 1.92 mol hydrogen mol−1 hexose equivalentadded in the co-culture compared to 1.45 mol hydrogen mol−1 hexose equivalentadded in the mono-culture. The maximum hydrogen production rate of 26 ml d−1 was achieved in the co-culture. Co-culture exhibited an overall 32 % enhancement of hydrogen yield based on hexose equivalent added and 15 % more substrate utilization. The main metabolites were acetate, ethanol, lactate, and formate in the mono-culture, with also butyrate in the co-culture. Additionally, the hydrogen yield of C. beijerinckii only in glucose was 2.54 mol hydrogen mol−1 hexose equivalent. This study has proved the viability of co-culture of C. termitidis with C. beijerinckii for hydrogen production directly from a complex substrate like cellulose under mesophilic conditions.


Clostridium termitidis Clostridium beijerinckii Co-culture Hydrogen production Cellulose Microbial kinetics


Hydrogen (H2) is considered a clean and renewable energy resource that does not contribute to the greenhouse effect (Lee et al. 2014). The main source of H2 production from fermentation is carbohydrates, among which, cellulose is widely available in agricultural wastes and industrial effluents such as pulp/paper and food industries (Lee et al. 2014). In comparison to the use of natural mixed consortia, pure cultures have achieved higher H2 yields (Masset et al. 2012). Artificial microbial co-cultures and consortia can perform complex functions (Masset et al. 2012), such as, simultaneous hexose and pentose consumption (Eiteman et al. 2008), maintaining anaerobic conditions for obligate H2 producers, improving the hydrolysis of complex sugars, allowing fermentation over a wider pH range (Elsharnouby et al. 2013), and could be more robust to changes in environmental conditions (Brenner et al. 2008). Although, thermophiles have shown higher H2 production yields than mesophiles in the literature (Kumar and Das 2000; Lu et al. 2007; Munro et al. 2009; Ngo et al. 2012), mesophilic H2 production is more economical and reliable than thermophilic and hyperthermophilic production. Four co-culture experiments for biohydrogen production from pure cellulose, two at mesophilic and two at thermophilic conditions (Geng et al. 2010; Liu et al. 2008; Wang et al. 2008, 2009) have been reported. All of these studies have shown enhancement of H2 production compared to mono-cultures, with the highest H2 yield of 1.8 mol hydrogen mol−1 hexose achieved by the co-culture of Clostridium thermocellum JN4 and Thermoanaerobacterium thermosaccharolyticum GD17 at 60 °C (Liu et al. 2008), potentially due to synergism between the two cultures.

Clostridium termitidis ATCC 51846 is an anaerobic, mesophilic, cellulolytic bacterium isolated from the gut of a termite (Hethener et al. 1992), with reported H2 yields of 1.99 mol hydrogen mol−1 hexose from glucose, 1.11 mol hydrogen mol−1 hexose equivalent from cellobiose (Gomez-Flores et al. 2015), and 0.62 mol hydrogen mol−1 hexose equivalent from cellulose (Ramachandran et al. 2008). On the other hand, C. beijerinckii is a mesophilic H2 producer which is not able to degrade cellulose but is adept at H2 production from glucose (Masset et al. 2012). Clostridium beijerinckii H2 yields from glucose have been reported to be 1.9 and 2.8 mol hydrogen mol−1 hexoseadded or consumed (Lin et al. 2007; Masset et al. 2012), 2.5 mol hydrogen mol−1 hexoseconsumed (Pan et al. 2008), and 2 mol hydrogen mol−1 hexoseadded (Taguchi et al. 1992). These experiments differ from each other in the reactor size, medium and initial glucose concentration.

Additionally, reasonably accurate mathematical models able to predict biochemical phenomena as well as the determination of its parameters are essential since they provide the basis for design, control, optimization and scale-up of process systems (Huang and Wang 2010). Therefore, this study has two goals (1) evaluate the effect of co-culture of C. termitidis and C. beijerinckii on biohydrogen production and, (2) determine the microbial kinetics of C. termitidis in mono-culture and co-cultured with C. beijerinckii on cellulose.

Materials and methods

Microbial strain and media

The strains used were C. termitidis ATCC 51846 (American Type Culture Collection) and C. beijerinckii DSM 1820 (Deutsche Sammlung von Mikroorganismen und Zellkulturen). All chemicals for media and substrates were obtained from Sigma-Aldrich Canada Co. (Oakville, ON, Canada). Fresh cells of C. termitidis were maintained by successively transferring 10 % (v/v) of inoculum to ATCC 1191 medium containing 2 g l−1 of cellulose, whereas fresh cells of C. beijerinckii were maintained by successively transferring 10 % (v/v) of inoculum to ATCC 1191 medium containing 2 g l−1 of cellobiose. The ATCC 1191 medium was prepared according to Gomez-Flores et al. (2015).

Experimental conditions

Batch fermentations were performed in media bottles (Wheaton, NJ, USA) with a working liquid volume of 500 and 210 ml of headspace. For the co-culture experiments, bottles containing 450 ml of ATCC 1191 medium and 1 g cellulose were tightly capped with screw caps with butyl septum, degassed by applying vacuum, sparged with high purity N2 gas, and autoclaved. Mono-culture bottles were inoculated with 10 % (v/v) of C. termitidis cultures, while co-culture bottles were inoculated with 10 % (v/v) of C. termitidis and C. beijerinckii cultures in a volumetric ratio of 1:1. All bottles were incubated at 37 °C in shakers (Max Q4000, Thermo Scientific, CA, USA). Three (3) ml liquid samples were taken at specific times for pH, metabolites, cellular protein content and cellulose analyses. Fermentations ran for 45 and 40 days for the mono-culture and co-culture, respectively. A total of 24 samples were taken for the mono-culture experiments whereas 21 samples were taken for the co-culture experiments. pH was initially set to 7.2 but was not controlled. Data shown are the averages of duplicate experiments. Additionally, fermentation on glucose 2 g l−1 by C. beijerinckii in the ATCC 1191 medium was performed in serum bottles (Wheaton, NJ, USA) with a working volume of 500 and 210 ml of headspace. Duplicate bottles were inoculated with 10 % (v/v) of fresh cultures. Bottles were incubated at 37 °C and 100 rpm for 48 h. Also, the initial pH was set to 7.2 but was not controlled.

Analytical methods

Cell growth was monitored by measuring cellular protein content, samples (1 ml) were placed in microcentrifuge tubes (VWR®, Polypropylene) and centrifuged (Corning® LSE™, NY, USA) at 10,000×g for 15 min. Supernatants were used for soluble product analysis by transferring to new microcentrifuge tubes. The pellets were re-suspended with 0.9 % (w/v) NaCl and centrifuged at the same aforementioned conditions. Supernatants were discarded, and 1 ml of 0.2 M NaOH was added to microcentrifuge tubes and vortexed to re-suspend the pellet. Microcentrifuge tubes were placed in a water bath at 100 °C for 10 min. After cooling, tubes were centrifuged and supernatants were collected for Bradford assay using bovine serum albumin (BSA) as standard, measured by a UV–visible spectrophotometer (Cary 50 Bio, Varian, Australia) at 595 nm. The cellulose pellet was quantified gravimetrically after being dried overnight at 100 °C (Liu et al. 2008). pH was measured using a B10P SympHony pH meter (VWR®). Ethanol, glucose, cellobiose, and lactic, formic, acetic, and butyric acids, were measured as follows: supernatants for metabolites analysis were filtered through 0.2 µm and measured using an HPLC (Dionex, Sunnyvale, CA, USA) consisting of a Dionex GP50 Gradient pump and a Dionex LC25 Chromatography oven equipped with an Aminex HPX-87H column (Bio-Rad) at 30 °C and 9 mM H2SO4 at 0.6 ml min−1 as mobile phase, connected to a Perkin Elmer 200 series refractive index detector (RID). Standard curves of metabolites, glucose and cellobiose were performed on ATCC 1191 medium. Cellular protein content was then converted to dry weight using the correlation dry weight (g l−1) = 0.0051 × protein (µg ml−1) (Gomez-Flores et al. 2015). For the estimation of the COD equivalents for the biomass dry weight, the empirical formula of the organic fraction of the biomass of C5H7O2N (Metcalf and Eddy 2003), and an organic fraction of 90 % of the cell dry weight (Pavlostathis et al. 1988), were assumed.

Gas measurements

Gas volume was measured by releasing the gas pressure in the bottles using appropriately sized glass syringes in the range of 5 to 100 ml to equilibrate with the ambient pressure (Owen et al. 1979). H2 analysis was conducted by employing a gas chromatograph (Model 310, SRI Instruments, Torrance, CA, USA) equipped with a thermal conductivity detector (TCD) and a molecular sieve column (Mole sieve 5A, mesh 80/100, 1.83 m × 0.32 cm). The temperatures of the column and the TCD detector were 90 and 105 °C, respectively. Argon was used as the carrier gas at a flow rate of 30 ml/min.

Modified Gompertz model

The following modified Gompertz model (Lay et al. 1999) was used to describe the H2 production.
$$ H\; = \;P\;exp\;\left\{ { - exp\left[ {\frac{{R_{max} \;e}}{P}\;\left( {\lambda \; - \;t} \right)\; + \;1} \right]} \right\} $$
where H is the cumulative H2 production (ml), P is the H2 production potential (ml), Rmax is the maximum H2 production rate (ml d−1) and λ is the lag time (d).

Kinetic equations and modeling

As shown in Fig. 1, there are mainly 2 steps: hydrolysis of cellulose and fermentation of soluble sugars (glucose). In both cases, C. termitidis’ putative cellulosome (Munir et al. 2014) is responsible for the cellulose hydrolysis. Fermentation of soluble sugars is performed by C. termitidis in mono-culture, whereas in co-culture both, C. termitidis and C. beijerinckii ferment the soluble sugars. The soluble products in mono-culture are acetate, ethanol, lactate and formate. In the co-culture, the lactate present in the C. beijerinckii growth media acted as substrate, and butyrate was an additional soluble product.
Fig. 1

Schematic representation of the steps involved in cellulose fermentation in a mono-culture and b co-culture

Among the various reactions involving glucose, only acetate and butyrate pathways involve H2 production according to Eqs. 2 and 3, respectively, while ethanol and lactate are involved in a zero-H2 balance (Guo et al. 2010)
$$ C_{6} H_{12} O_{6} \; + \;2H_{2} O\; \to \;2CH_{3} COOH\; + \;2CO_{2} \; + \;4H_{2} $$
$$ C_{6} H_{12} O_{6} \; \to \;CH_{3} CH_{2} CH_{2} COOH\; + \;2CO_{2} \; + \;2H_{2} . $$
Lactate utilization is represented by Eq. 4 (Thauer et al. 1977).
$$ CH_{3} CHOHCOOH\; + \;H_{2} O\; \to \;CH_{3} COOH\; + \;CO_{2} \; + \;2H_{2} . $$
Because cellulose was not completely biodegraded, the use of a non-biodegradable factor So (g COD l−1) was needed as presented in Eq. 5.
$$ S\; = \;\mathop \int \limits_{0}^{t} \frac{dS}{dt}\; + \;S_{o} $$
where S is cellulose concentration (g COD l−1) and So is the non-biodegradable cellulose concentration remaining at the end of the fermentation. Soluble sugars from cellulose hydrolysis (cellobiose and glucose) were not detected in any of the fermentations, implying that cellulose hydrolysis was the rate-limiting step. Nevertheless, cellulose is an insoluble substrate and Monod model cannot be used. Therefore, a modified Monod approach, incorporating particulate organic matter (POM) (Metcalf and Eddy 2003) was used (Eq. 6).
$$ {{\upmu }}\; = \;\frac{{{{\upmu }}_{{\varvec{max}}} \;\left( {\frac{{\varvec{PO}}}{\varvec{X}}} \right)}}{{\varvec{K}_{\varvec{X}} \; + \;\left( {\frac{{\varvec{PO}}}{\varvec{X}}} \right)}} $$
where µmax (d−1) is the maximum specific growth rate, Kx is the half-velocity degradation coefficient (g COD PO g−1 COD biomass), PO is the particulate organic (cellulose) concentration (g COD l−1) and X is biomass concentration (g COD l−1) (Metcalf and Eddy 2003). The POM modeling approach considers the particulate substrate conversion rate as the rate-limiting process that is dependent on the particulate substrate and biomass concentrations. The particulate degradation concentration is expressed relative to the biomass because the particulate hydrolysis is related to the relative contact area between the non-soluble organic material and the biomass (Metcalf and Eddy 2003). All concentrations were expressed as g COD; for biomass the factor of 1.42 g COD g−1 biomass based on the empirical formula of C5H7O2N was used (Metcalf and Eddy 2003).
The two models are described as follows:
  1. a.
    Mono-culture (C. termitidis only). Biomass growth and PO consumption are described in Eqs. 7 and 8, respectively.
    $$ \frac{dX}{dt}\; = \;{{\upmu }}X\; = \;\frac{{{\upmu{ }}_{max} \;\left( {\frac{PO}{X}} \right)\;X}}{{\left[ {K_{X} \; + \;\left( {\frac{PO}{X}} \right)} \right]}} $$
    $$ \frac{dPO}{dt}\; = \; - \;\frac{{{{\upmu }}_{max} \;\left( {\frac{PO}{X}} \right)\;X}}{{Y_{{{X} / _{{PO}}}} \;\left[ {K_{X} \; + \;\left( {\frac{PO}{X}} \right)} \right]}} $$
    where YX/PO (g COD biomass g−1 COD PO) is the biomass yield (Shuler and Kargı 2002). Acetate, ethanol, lactate and formate production was modeled as described by Eq. 9.
    $$ \frac{dP}{dt}\; = \;\frac{{Y_{P} /_ {{PO}} }}{{Y_{{{{X}/_ {PO}}}} }}\; \frac{{{{\upmu }}_{max} \;\left( {\frac{PO}{X}} \right)\;X}}{{\left[ {K_{X} \; + \;\left( {\frac{PO}{X}} \right)} \right]}} $$
    where P and YP/PO are acetate, ethanol, lactate and formate concentrations (g COD l−1) and yields (g COD g−1 COD PO), respectively.
  2. b.
    Co-culture (C. termitidis and C. beijerinckii). No distinction in biomass measurement was done for each strain. Co-culture was modeled as a single strain with the addition of lactate as substrate and butyrate as product. Consequently, PO consumption is described in Eq. 8, biomass growth from cellulose and lactate is modeled by Eq. 10, and lactate consumption was considered a first order reaction (Eq. 11).
    $$ \frac{dX}{dt}\; = \;\frac{{{{\upmu }}_{max} \;\left( {\frac{PO}{X}} \right)\;X}}{{\left[ {K_{X} \; + \;\left( {\frac{PO}{X}} \right)} \right]}}\; + \;Y_{{X} {/ {L}}} K_{L} LX $$
    $$ \frac{dL}{dt} = - K_{L} LX $$
    where YX/L is the biomass yield from lactate (as g COD g−1 COD) and KL is the lactate consumption constant (l g−1 COD biomass d−1). Based on Eq. 4, acetate is also produced from lactate. Thus acetate kinetics are modeled by Eq. 12.
    $$ \frac{dA}{dt}\; = \;\frac{{Y_{{{{A} {/ {{ {PO}}}} }}} }}{{Y_{X}{{{{/ _{{ {PO}}}} }}} }}\;\frac{{{{\upmu }}_{max} \;\left( {\frac{PO}{X}} \right)X}}{{\left[ {K_{X} \; + \;\left( {\frac{PO}{X}} \right)} \right]}}\; + Y_{A}/_{L} K_{L}LX$$
    where YA/L is the acetate yield from lactate (g COD g−1 COD).

Ethanol, formate and butyrate were described by Eq. 9, where P and YP/PO are also butyrate concentration (g COD l−1) and yield (g COD g−1 COD PO).

Microbial kinetics were estimated from the growth phase only, ignoring the lag phase. Kinetic parameters were estimated using MATLAB® R2014a. The solver function used for numerical integration of the ordinary differential equations i.e. Ode45, implemented fourth/fifth order Runge–Kutta methods. Initial guesses were manually adjusted to obtain a good fit to the data, and average percentage errors (APE) and root mean square errors (RMSE) were calculated. The complete nomenclature is shown in Table 1.
Table 1



Meaning and units


Lactate consumption constant (l g−1 COD biomass d−1)


Substrate utilization rate (g COD PO g−1 COD biomass d−1)


Half-velocity degradation coefficient (g COD PO g−1 COD biomass)


Maximum specific growth rate (d−1)


Non-biodegradable factor (g COD l−1)


Acetate yield from lactate (g COD g−1 COD lactate)


Acetate yield from particulate organic (g COD g−1 COD PO)


Butyrate yield from particulate organic (g COD g−1 COD PO)


Ethanol yield from particulate organic (g COD g−1 COD PO)


Formate from particulate organic (g COD g−1 COD PO)


Lactate yield from particulate organic (g COD g−1 COD PO)


Biomass yield from lactate (g COD g−1 COD lactate)


Biomass yield from particulate organic (g COD biomass g−1 COD PO)


C. beijerinckii on glucose experiment

Clostridium beijerinckii degraded glucose in 46 h with an initial lag phase of 22 h and had a yield of 2.54 mol hydrogen mol−1 glucose (Additional file 1: Figure S1a). pH dropped from 7.1 to 6.2. With a 28 % higher H2 yield over C. termitidis for the same substrate (Gomez-Flores et al. 2015) and under the same operating conditions, with the exception of using 500 ml of working volume instead of 400 ml, C. beijerinckii was chosen to potentially enhance H2 production when co-cultured with C. termitidis on cellulose by serving as a high H2 producer from glucose formed from cellulose hydrolysis by C. termitidis.

At the same time, a correlation between dry weight and cellular protein content was developed for C. beijerinckii in a similar way to the correlation for C. termitidis (Gomez-Flores et al. 2015). A 20 % cellular protein content was obtained, in close agreement with the 19 % obtained for C. termitidis in the aforementioned study (Additional file 1: Figure S1b).

Hydrogen production from cellulose

The H2 production profiles in Fig. 2a clearly depict the enhancement in H2 production from co-culture over mono-culture. H2 production showed long lag phases of up to 17 days. The results of the modified Gompertz model are shown in Table 2. The overall H2 production for the co-culture compared with the mono-culture increased by 30 % to 326 ml. Moreover, the H2 production rate in the co-culture of 26 ml d−1 was double the 12 ml d−1 observed in the mono-culture.
Fig. 2

C. termitidis mono-cultured in 2 g l−1 cellulose and co-cultured with C. beijerinckii 2 g l−1 cellulose. a Cumulative H2 production profiles. b pH profiles. Data points are the averages of duplicates, lines above and below represent the actual duplicates

Table 2

H2 yields and Gompertz parameters of C. termitidis mono-cultured and co-cultured with C. beijerinckii on 2 g l−1 cellulose


Cellulose consumed (%)

H2 yields

Gompertz parameters

mol H2 mol−1 hexose eq.added

mol H2 mol−1 hexose eqconsumed

P max a (ml)

R m b (ml d−1)

λc (d)


















aH2 production potential

bMaximum H2 production rate

cLag phase

Figure 2b shows the pH profiles. During the lag phases, all cultures exhibited a marginal decrease in pH from 7.2 to around 7. Concurrent with the H2 production, the pH dropped to around 6.1. As the optimum pH range for C. termitidis growth has been reported to be >5 to <8.2 (Hethener et al. 1992), the pH changes observed in mono-culture fermentations were assumed not to impact the microbial kinetics. For C. beijerinckii DSM 1820 growth, the pH range reported is from 5.2 to 7.3, with the former reported as inhibitory (Masset et al. 2012). As the observed pH changes in the co-culture fermentation were within the growth range reported for both strains, pH changes were assumed not to affect the microbial kinetics.

Cellulose was not completely consumed in neither case but co-culture enhanced the extent of cellulose utilization by 15 % to about 93 % (Table 2).

Table 2 also shows the H2 yields based on hexose equivalent added and consumed. The H2 yield of 1.92 mol hydrogen mol−1 hexose equivalentadded obtained in the co-culture was 32 % greater than the H2 yield obtained by the mono-culture of 1.45 mol hydrogen mol−1 hexose equivalentadded. Also, the H2 yield of 2.05 mol hydrogen mol−1 hexose equivalentconsumed in the co-culture was 14 % greater than the H2 yield obtained by the mono-culture of 1.8 mol hydrogen mol−1 hexose equivalentconsumed.

Microbial products and kinetics

The experimental and modeled biomass and cellulose profiles are illustrated in Fig. 3, which emphatically demonstrates that the co-culture was able to utilize more cellulose than mono-culture and the ultimate biomass growth was similar in all cases.
Fig. 3

Experimental and modeled growth kinetics. a Mono-culture. b Co-culture

Figure 4 shows the experimental and modeled metabolites profiles. Neither glucose nor cellobiose from cellulose hydrolysis were detected in any of the fermentations, implying that cellulose hydrolysis was the rate limiting factor. Clostridium termitidis metabolites on cellulose were acetate, ethanol, lactate, and formate, in agreement with Ramachandran et al. (2008). In mono-culture experiments, acetate and ethanol were produced during biomass growth, while, formate and lactate exhibited lag phases and were not detected until day 38. H2 production peaked around day 44 for the mono-culture experiment, concurrent with all metabolites peak.
Fig. 4

Experimental and modeled profile of metabolites in a mono-culture and b co-culture

Clostridium beijerinckii DSM 1820 soluble products from glucose have been reported by Masset et al. (2012) to be butyrate, acetate, formate, lactate, in addition to butanol, acetone and isopropanol by Chen and Hiu (1986), although, other strains of C. beijerinckii (i.e. L9 and Fanp3) have been demonstrated to produce ethanol from glucose (Lin et al. 2007; Pan et al. 2008). In the co-culture experiment acetate and butyrate were produced as lactate was consumed. It is noteworthy that only butyrate production peaked on day 40, concurrent with the H2 peak in the co-culture.

Mathematical models that accurately predict biochemical phenomena provide the basis for design, control, optimization and scale-up of process systems (Huang and Wang 2010). Kinetic parameters of the mathematical model are shown in Table 3. The co-culture exhibited the highest µmax (0.2 d−1), thus rationalizing the end of the fermentation test before the mono-culture. In this regard, the impact of the synergy in microbial kinetics was notorious, with µmax in co-culture of 0.2 d−1 double the 0.1 d−1 observed in mono-culture. It is noteworthy that the maximum specific growth rates achieved on glucose and cellobiose by C. termitidis of 0.22 and 0.24 h−1, respectively (Gomez-Flores et al. 2015), are more than 50 times greater than those achieved by the same strain on cellulose. The half-saturation constant, Kx, varied between 0.42 and 1.1 g COD cellulose g−1 COD biomass. PO/X values (Additional file 1: Figure S2) are significantly greater than the Kx values, i.e. the growth rate throughout the experiments equals µmax. The recommended value for the hydrolysis rate of carbohydrates in the anaerobic digestion model (ADM1) (Batstone et al. 2002) is 0.25 d−1 at mesophilic conditions which is comparable to the growth rates obtained in the present study, clearly emphasizing that the biodegradation of cellulose is hydrolysis-limited.
Table 3

Kinetic parameters obtained in MATLAB of C. termitidis mono-cultured and co-cultured with C. beijerinckii on 2 g l−1 cellulose




\( S_{o}\) a \( ({\text{g COD}} \,{\text{l}}^{ - 1} ) \)



\( \upmu_{max} ({\text{d}}^{ - 1} ) \)



\( Y_{{x}{/ { L}}}\) b



\( Y_{{x/PO}}\) c



K m d



\( Y_{L/PO} \) e



\(Y_{F/PO}\) f



\(Y_{A/PO}\) g



\(Y_{E/PO}\) h



\(Y_{B/PO}\) i



\(Y_{A/L}\) j






K x k



NA Not applicable

aNon-biodegradable factor

bBiomass yield from lactate (g COD g−1 COD lactate)

cBiomass yield (g COD g−1 COD PO)

dg COD PO g−1 COD biomass d−1

eLactate yield (g COD g−1 COD PO)

fFormate yield (g COD g−1 COD PO)

gAcetate yield (g COD g−1 COD PO)

hEthanol yield (g COD g−1 COD PO)

iButyrate yield (g COD g−1 COD PO)

jAcetate yield from lactate (g COD g−1 COD lactate)

kg COD PO g−1 COD biomass

Co-culture experiment reflected a slightly lower biomass yield than monoculture (0.25 vs 0.3 g COD g−1 COD cellulose). YX/L (biomass yield from lactate) was assumed to be the same as YX/PO (biomass yield from cellulose) and YA/L (acetate yield from lactate) was calculated as follows:
where fA/L is the stoichiometric relationship based on Eq. 4 of 1 mol acetate per mol lactate, calculated in g COD as 0.66. YA/L was calculated to be 0.49 g COD acetate g−1 COD lactate and the theoretical H2 production from lactate was also calculated based on Eq. 4 and subtracted from the measured H2 produced. The modified H2 yields from cellulose in the co-culture experiment were 1.72 mol hydrogen mol−1 hexose equivalentadded and 1.84 mol hydrogen mol−1 hexose equivalentconsumed, approximately 19 % higher than the mono-culture based on hexose added. Nevertheless, the calculated H2 from lactate may be overestimated since it is theoretical.

The average percentage errors (APE) and RMSE calculated for the modeled biomass, substrate and metabolites are the in Additional file 1: Table S1. Biomass and cellulose exhibited the lowest average percentage errors, within the range of 4–8 %, followed by PO/X with the highest value of 11 % in co-culture. For both lactate and formate in mono-culture, the model significantly under estimated the lag phase, as evident from Fig. 4a. Accordingly, the APE excluding the lag phase for lactate and formate were 12 and 11 % and including lag phases was 81 % in both cases.


Hydrogen production

COD balances calculated by summation of metabolites, H2, cellulose and cells as g COD l−1 at the beginning and end of fermentations are presented in Table 4. The COD balances closed within 3–8 % of the initial, thus confirming the reliability of the data. Theoretical H2 production from acetate and butyrate shown in Table 4 was calculated based on 848 ml hydrogen g−1 acetate and 578 ml hydrogen g−1 butyrate (Eqs. 2, 3). The theoretical values were consistent with the H2 measured during the experiment with an average percent difference of 1 % of the theoretical H2. C. beijerinckii DSM 1820 produced a H2 yield of 2.54 mol hydrogen mol−1 glucose, added or consumed, in line with the 1.9 and 2.8 mol hydrogen mol−1 hexoseadded or consumed (Lin et al. 2007; Masset et al. 2012), 2.5 mol hydrogen mol−1 hexoseconsumed (Pan et al. 2008), and 2 mol hydrogen mol−1 hexoseadded (Taguchi et al. 1992). On the other hand, while the highest reported mesophilic H2 yield by co-culture on cellulose is 1.31 mol H2 mol−1 hexose with Clostridium acetobutylicum X9 and Ethanoigenens harbinense B49 (Wang et al. 2008), and the highest thermophilic H2 yield is 1.8 mol H2 mol−1 hexose with C. thermocellum JN4 and T. thermosaccharolyticum GD17 (Liu et al. 2008), the results from this study (Table 2) reveal a significantly improved H2 yield in the co-culture of C. termitidis and C. beijerinckii compared to the literature. The achievement of a yield of 1.92 mol hydrogen mol−1 hexose using two mesophilic cultures represents about 50 % improvement of the literature at similar conditions. Although the aforementioned yield is only 7 % higher than the maximum thermophilic yield, the balance of thermal energy input and output based on hydrogen in this study is still more favorable than reported elsewhere in the literature.
Table 4

COD balance and theoretical H2 production of C. termitidis mono-cultured and co-cultured with C. beijerinckii on 2 g l−1 cellulose


Metabolitesa (g COD l−1)

H 2 b (g COD l−1)

Cellulose (g COD l−1)

Biomassc (g COD l−1)

Total COD (g COD l−1)

COD balanced (%)

Theoretical H2 (ml)

Experimental H2 (ml)

Difference (%)

From acetic acid

From butyric acid








































aMetabolites COD accounts for the sum of acetate, butyrate, lactate, formate and ethanol as g COD l−1

bCalculated based on 8 g COD g−1 H2

cBiomass COD was calculated by multiplying dry weight (g l−1) × 0.9 × 1.42 (g COD g−1 biomass)

dCOD mass balance = (Final TCOD/Initial TCOD) × 100 %

Based on the modeled acetate and butyrate profiles, modeled H2 profiles shown in Fig. 5 were calculated in a similar manner as the theoretical H2 shown in Table 4, with 848 ml H2 g−1 acetate and 578 ml hydrogen g−1 butyrate (from stoichiometry of Eqs. 2 and 3), and 1.067 g COD g−1 acetate and 1.82 g COD g−1 butyrate. The modeled H2 profiles closely match the experimental H2, as verified with the low APE values ranging from 10 to 15 % and RMSE values (9–13 ml).
Fig. 5

Experimental and modeled H2 profiles for a mono-culture and b co-culture

Microbial products and kinetics

Anaerobic lactate consumption has been reported by different inoculums, such as soil, kitchen waste compost, Clostridium diolis JPCC H-3, Clostridium butyricum JPCC H-1, C. acetobutylicum P262, and also C. beijerinckii JPCC H-4 (Diez-Gonzalez et al. 1995; Grause et al. 2012; Lee et al. 2010; Matsumoto and Nishimura 2007). Nevertheless, in some cases, acetate has been simultaneously consumed. The metabolic pathways reported in the literature are shown in Eqs. 4, 15 and 16 (Costello et al. 1991; Diez-Gonzalez et al. 1995; Grause et al. 2012; Matsumoto and Nishimura 2007; Thauer et al. 1977):
$$ CH_{3} CH\;\left( {OH} \right)COOH\; + \;0.5CH_{3} COOH\; \to \; 0.75CH_{3} CH_{2} CH_{2} COOH\; + \;0.5H_{2} \; + \;CO_{2} \; + \;0.5H_{2} O $$
$$ CH_{3} CH\left( {OH} \right)COOH\; + \;0.43CH_{3} COOH\; \to \; 0.7CH_{3} CH_{2} CH_{2} COOH\; + \;0.57H_{2} \; + \;CO_{2} \; + \;0.7H_{2} O $$

The evident lactate consumption in co-culture fermentations shown in Fig. 4b, could be assumed to follow Eq. 4 since acetate was produced simultaneously.

Apparently, co-culture fermentation exhibited acetate consumption after day 27 (Fig. 4b), which could be explained by Eqs. 15 and 16, although lactate was below the detection limit during this period of time. In contrast, mono-culture fermentation did not exhibit this phenomenon because C. termitidis does not produce butyrate; thus acetate consumption in co-culture fermentations could be attributed to the presence of C. beijerinckii. Interestingly, the co-culture experiment of C. thermocellum JN4 and T. thermosaccharolyticum GD17 on cellulose reported by Liu et al. (2008) also consumed lactate with acetate production whereas C. thermocellum JN4 in mono-culture did not; no explanation of this phenomenon was attempted by the authors.

Desvaux et al. (2000) found a µmax of 0.056 h−1 with C. cellulolyticum grown on 2.4 g cellulose l−1 with a biomass yield of 36.5 g of cells mol−1 hexose equivalent (or 0.2 g cells g−1 hexose). Kinetics on cellulose have been also explained by alternative models to Monod. For example, Holwerda and Lynd (2013) found that the best fit to their results on C. thermocellum was with a substrate utilization rate that is both first order with respect to substrate and first order in cells. Recently, Gupta et al. (2015) found a µmax of 0.05 d−1 on cellulose using mesophilic anaerobic digested sludge (ADS) and a Ks of 2.1 g l−1, which is four times lower than that achieved by C. termitidis in the present study.

This study is the first to model C. termitidis microbial kinetics on cellulose and in co-culture with C. beijerinckii. High H2 yields at mesophilic temperature directly from cellulose of 1.8 and 2.05 mol hydrogen mol−1 hexose equivalentconsumed in mono-culture and co-culture, respectively, were achieved as compared to the literature. Cellulose degradation by the co-culture was 15 % higher than the mono-culture of C. termitidis. The viability of C. termitidis and C. beijerinckii producing H2 together was evidenced.


Authors’ contributions

MGF did the experimental design, laboratory work, data analysis, development of the code in Matlab, modeling, and paper writing. GN contribution was supervision, critical and data interpretation, paper review, and corrections. HH did paper review. All authors read and approved the final manuscript.



Competing interests

The authors declare that they have no competing interests.

Ethics approval and consent to participate

Not applicable since this article does not contain any studies with human participants or animals performed by any of the authors.


This work was supported by the Eastern platform of the Biofuel Network. The authors acknowledge the support by Consejo Nacional de Ciencia y Tecnologia de Mexico (CONACYT) and Alianza para la Formacion e Investigacion en Infraestructura para el Desarrollo de Mexico, awarded to Maritza Gomez-Flores.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

Department of Chemical and Biochemical Engineering, Faculty of Engineering, University of Western Ontario
Department of Civil and Environmental Engineering, Faculty of Engineering, University of Western Ontario


  1. Batstone DJ, Keller J, Angelidaki I, Kalyuzhnyi S, Pavlostathis SG, Rozzi A, Sanders W, Siegrist H, Vavilin V (2002) Anaerobic digestion model no. 1 (ADM1). IWA Publishing, LondonGoogle Scholar
  2. Brenner K, You L, Arnold FH (2008) Engineering microbial consortia: a new frontier in synthetic biology. Trends Biotechnol 26:483–489. doi: View ArticlePubMedGoogle Scholar
  3. Chen JS, Hiu SF (1986) Acetone-butanol-isopropanol production by Clostridium beijerinckii (synonym, Clostridium butylicum). Biotechnol Lett 8:371–376. doi: View ArticleGoogle Scholar
  4. Costello DJ, Greenfield PF, Lee PL (1991) Dynamic modelling of a single-stage high-rate anaerobic reactor—I. Model derivation. Water Res 25:847–858. doi: View ArticleGoogle Scholar
  5. Desvaux M, Guedon E, Petitdemange H (2000) Cellulose catabolism by Clostridium cellulolyticum growing in batch culture on defined medium. Appl Environ Microbiol 66:2461–2470. doi: View ArticlePubMedPubMed CentralGoogle Scholar
  6. Diez-Gonzalez F, Russell JB, Hunter JB (1995) The role of an NAD-independent lactate dehydrogenase and acetate in the utilization of lactate by Clostridium acetobutylicum strain P262. Arch Microbiol 164:36–42. doi: View ArticleGoogle Scholar
  7. Eiteman MA, Lee SA, Altman E (2008) A co-fermentation strategy to consume sugar mixtures effectively. J Biol Eng 2:3. doi: View ArticlePubMedPubMed CentralGoogle Scholar
  8. Elsharnouby O, Hafez H, Nakhla G, El Naggar MH (2013) A critical literature review on biohydrogen production by pure cultures. Int J Hydrogen Energy 38:4945–4966. doi: View ArticleGoogle Scholar
  9. Geng A, He Y, Qian C, Yan X, Zhou Z (2010) Effect of key factors on hydrogen production from cellulose in a co-culture of Clostridium thermocellum and Clostridium thermopalmarium. Bioresour Technol 101:4029–4033. doi: View ArticlePubMedGoogle Scholar
  10. Gomez-Flores M, Nakhla G, Hafez H (2015) Microbial kinetics of Clostridium termitidis on cellobiose and glucose for biohydrogen production. Biotechnol Lett 37:1965–1971. doi: View ArticlePubMedGoogle Scholar
  11. Grause G, Igarashi M, Kameda T, Yoshioka T (2012) Lactic acid as a substrate for fermentative hydrogen production. Int J Hydrogen Energy 37:16967–16973. doi: View ArticleGoogle Scholar
  12. Guo XM, Trably E, Latrille E, Carrère H, Steyer JP (2010) Hydrogen production from agricultural waste by dark fermentation: a review. Int J Hydrogen Energy 35:10660–10673. doi: View ArticleGoogle Scholar
  13. Gupta M, Gomez-Flores M, Nasr N, Elbeshbishy E, Hafez H, Hesham El Naggar M, Nakhla G (2015) Performance of mesophilic biohydrogen-producing cultures at thermophilic conditions. Bioresour Technol 192:741–747. doi: View ArticlePubMedGoogle Scholar
  14. Hethener P, Brauman A, Garcia JL (1992) Clostridium termitidis sp. nov., a cellulolytic bacterium from the gut of the wood-feeding termite, Nasutitermes lujae. Syst Appl Microbiol 15:52–58. doi: View ArticleGoogle Scholar
  15. Holwerda EK, Lynd LR (2013) Testing alternative kinetic models for utilization of crystalline cellulose (Avicel) by batch cultures of Clostridium thermocellum. Biotechnol Bioeng 110:2389–2394. doi: View ArticlePubMedGoogle Scholar
  16. Huang WH, Wang FS (2010) Kinetic modeling of batch fermentation for mixed-sugar to ethanol production. J Taiwan Inst Chem Eng 41:434–439. doi: View ArticleGoogle Scholar
  17. Kumar N, Das D (2000) Enhancement of hydrogen production by Enterobacter cloacae IIT-BT 08. Process Biochem 35:589–593. doi: View ArticleGoogle Scholar
  18. Lay JJ, Lee YJ, Noike T (1999) Feasibility of biological hydrogen production from organic fraction of municipal solid waste. Water Res 33:2579–2586. doi: View ArticleGoogle Scholar
  19. Lee ZK, Li SL, Kuo PC, Chen IC, Tien YM, Huang YJ, Chuang C, Wong SC, Cheng SS (2010) Thermophilic bio-energy process study on hydrogen fermentation with vegetable kitchen waste. Int J Hydrogen Energy 35:13458–13466. doi: View ArticleGoogle Scholar
  20. Lee KS, Whang LM, Saratale GD, Chen SD, Chang JS, Hafez H, Nakhla G, El Naggar MH (2014) Biological hydrogen production: dark fermentation. In: Sherif SA, Goswami DY, Stefanakos EK, Steinfeld A (eds) Handbook of hydrogen energy. The CRC Press series in mechanical and aerospace engineering. CRC Press, Boca RatonGoogle Scholar
  21. Lin PY, Whang LM, Wu YR, Ren WJ, Hsiao CJ, Chang SLLS (2007) Biological hydrogen production of the genus Clostridium: metabolic study and mathematical model simulation. Int J Hydrogen Energy 32:1728–1735. doi: View ArticleGoogle Scholar
  22. Liu Y, Yu P, Song X, Qu Y (2008) Hydrogen production from cellulose by co-culture of Clostridium thermocellum JN4 and Thermoanaerobacterium thermosaccharolyticum GD17. Int J Hydrogen Energy 33:2927–2933. doi: View ArticleGoogle Scholar
  23. Lu W, Wen J, Chen Y, Sun B, Jia X, Liu M, Caiyin Q (2007) Synergistic effect of Candida maltosa HY-35 and Enterobacter aerogenes W-23 on hydrogen production. Int J Hydrogen Energy 32:1059–1066. doi: View ArticleGoogle Scholar
  24. Masset J, Calusinska M, Hamilton C, Hiligsmann S, Joris B, Wilmotte A, Thonart P (2012) Fermentative hydrogen production from glucose and starch using pure strains and artificial co-cultures of Clostridium spp. Biotechnol Biofuels 5:1–15. doi: View ArticleGoogle Scholar
  25. Matsumoto M, Nishimura Y (2007) Hydrogen production by fermentation using acetic acid and lactic acid. J Biosci Bioeng 103:236–241. doi: View ArticlePubMedGoogle Scholar
  26. Metcalf L, Eddy HP (2003) Wastewater engineering: treatment and reuse, 4th edn. McGraw-Hill, New YorkGoogle Scholar
  27. Munir RI, Schellenberg J, Henrissat B, Verbeke TJ, Sparling R, Levin DB (2014) Comparative analysis of carbohydrate active enzymes in Clostridium termitidis CT1112 reveals complex carbohydrate degradation ability. PLoS ONE 9:e104260. doi: View ArticlePubMedPubMed CentralGoogle Scholar
  28. Munro SA, Zinder SH, Walker LP (2009) The fermentation stoichiometry of Thermotoga neapolitana and influence of temperature, oxygen, and pH on hydrogen production. Biotechnol Prog 25:1035–1042. doi: View ArticlePubMedGoogle Scholar
  29. Ngo TA, Nguyen TH, Bui HTV (2012) Thermophilic fermentative hydrogen production from xylose by Thermotoga neapolitana DSM 4359. Renewable Energy 37:174–179. doi: View ArticleGoogle Scholar
  30. Owen WF, Stuckey DC, Healy JB, Young LY, McCarty PL (1979) Bioassay for monitoring biochemical methane potential and anaerobic toxicity. Water Res 13:485–492. doi: View ArticleGoogle Scholar
  31. Pan CM, Fan YT, Zhao P, Hou HW (2008) Fermentative hydrogen production by the newly isolated Clostridium beijerinckii Fanp3. Int J Hydrogen Energy 33:5383–5391. doi: View ArticleGoogle Scholar
  32. Pavlostathis SG, Miller TL, Wolin MJ (1988) Kinetics of insoluble cellulose fermentation by continuous cultures of Ruminococcus albus. Appl Environ Microbiol 54:2660–2663PubMedPubMed CentralGoogle Scholar
  33. Ramachandran U, Wrana N, Cicek N, Sparling R, Levin DB (2008) Hydrogen production and end-product synthesis patterns by Clostridium termitidis strain CT1112 in batch fermentation cultures with cellobiose or α-cellulose. Int J Hydrogen Energy 33:7006–7012. doi: View ArticleGoogle Scholar
  34. Shuler ML, Kargı F (2002) Bioprocess engineering: Basic concepts. Prentice Hall PTR, Upper Saddle RiverGoogle Scholar
  35. Taguchi F, Hang JD, Takiguchi S, Morimoto M (1992) Efficient hydrogen production from starch by a bacterium isolated from termites. J Ferment Bioeng 73:244–245. doi: View ArticleGoogle Scholar
  36. Thauer RK, Jungermann K, Decker K (1977) Energy conservation in chemotrophic anaerobic bacteria. Bacteriol Rev 41:100–180PubMedPubMed CentralGoogle Scholar
  37. Wang AJ, Ren NQ, Shi YG, Lee DJ (2008) Bioaugmented hydrogen production from microcrystalline cellulose using co-culture - Clostridium acetobutylicum X-9 and Etilanoigenens harbinense B-49. Int J Hydrogen Energy 33:912–917. doi: View ArticleGoogle Scholar
  38. Wang A, Gao L, Ren N, Xu J, Liu C (2009) Bio-hydrogen production from cellulose by sequential co-culture of cellulosic hydrogen bacteria of Enterococcus gallinarum G1 and Ethanoigenens harbinense B49. Biotechnol Lett 31:1321–1326. doi: View ArticlePubMedGoogle Scholar


© The Author(s) 2016