Modeling mixed culture fermentations - Water Science & Technology

Q IWA Publishing 2008 Water Science & Technology—WST | 57.4 | 2008

493

Modeling mixed culture fermentations; the role of different electron carriers R. Kleerebezem, J. Rodrı´guez, M. F. Temudo and M. C. M. van Loosdrecht

ABSTRACT A recently established mixed culture fermentation (MCF) model has been modified to account for the role of different electron carriers in the process. The MCF-model predicts the product spectrum as a function of the actual environmental conditions using a thermodynamic optimization criterion while satisfying a number of constraints. Other improvements made to the original model are the inclusion of formate as fermentation end-product, and gas-liquid mass transfer. The model is adequately capable of reproducing experimental results in terms of butyrate and formate versus hydrogen/carbon dioxide production. The model is not capable of predicting the production of an ethanol/acetate mixture as measured at higher pH-values, suggesting specific biochemical control. Catabolic acetate production can potentially be explained by anabolic requirements for a specific electron donor like NADH. Key words

R. Kleerebezem M. F. Temudo M. C. M. van Loosdrecht Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands E-mail: [email protected]; [email protected]; [email protected] J. Rodrı´guez Sustainable Environment Research Centre, University of Glamorgan. Pontypridd, CF37 1DL Wales, UK E-mail: [email protected]

| acidogenesis, electron carrier, fermentation, mathematical model

INTRODUCTION Anaerobic fermentation of carbohydrates by mixed micro-

of a paper factory, or for the production of specific building

bial cultures of bacteria is an important process in natural

blocks for further processing as is required in PHA

and man-made environments. The fatty acids produced in a

production.

mixed culture fermentation (MCF) process can be used for

We have been investigating the MCF process using a

methane containing biogas production, or for the production

dual approach: experimentally we have studied glucose

of polyhydroxyalkanoates (PHA, bioplastics). MCF plays

fermentation as a function of the pH in chemostat type

furthermore an important role in processes like storage of

bioreactors (Temudo et al. 2007), and based on theoretical

grass, paper production and the storage of domestic waste in

considerations we have developed a mathematical model

landfills.

that predicts the product spectrum in mixed culture

The dominant pathways for product formation in the

fermentation process as a function of the operational

MCF process generate volatile fatty acids (acetate, propio-

variables (Rodriguez et al. 2006). Even though the model

nate, butyrate, lactate) or solvents (i.e. ethanol), but despite

predictions were partially confirmed by the experimental

extensive research efforts it remains unclear which path-

observations, major differences were found as well. We have

ways dominate in which conditions (Zoetemeyer et al.

suggested that one of the reasons for the differences found is

1982a,b; Mosey 1983). The work described here aims for

based on the use of different electron carriers in the different

prediction of the product spectrum of the MCF process as a

catabolic steps of the fermentation process. In this paper we

function of the operational variables applied. This may

introduce this concept of different electron carriers in the

allow for prevention of the formation of certain unwanted

MCF process model in order to reduce the gap between the

products like the production of butyrate in the water cycle

experimental and theoretical results. A number of additional

doi: 10.2166/wst.2008.094

494

R. Kleerebezem et al. | Role of electron carriers in mixed culture fermentations

Water Science & Technology—WST | 57.4 | 2008

improvements to the model structure are furthermore

block. NAD/NADH is assumed as electron carrier in the

suggested, and a drastically different numerical solution

process, and considered to be in thermodynamic equili-

method is described.

brium with the actual hydrogen partial pressure in the medium. The model includes furthermore a number of

MODEL DESCRIPTION

constraints as defined by the process thermodynamics and the intracellular concentrations organic acids. The

The anaerobic fermentation model described in this paper is

prediction of the final catabolic product spectrum is

based on the model described in the paper by Rodriguez

established in a growth maximization routine based on

et al. (Rodriguez et al. 2006). The model considers the

the assumption of ecological selection of the dominant

dominant catabolic reaction pathways for glucose fermen-

pathway by maximization of the biomass yield. A full

tation in a simplified metabolic network scheme as shown

description of this model as well as the predicted product

in Figure 1. Metabolic energy production in the form of ATP

spectrum as a function of a number of operational variables

is considered to occur due to substrate level phosphoryl-

can be found elsewhere (Rodriguez et al. 2006). Below the

ation and due to generation of transmembrane proton

changes made to the model will be described.

gradient in thermodynamically strongly favorable oxidation reactions in the butyrate and propionate pathways. The excretion of volatile fatty acids is furthermore associated with ATP consumption or production as a function of the

Modification 1: electron carriers different from NAD/NADH

actual environmental conditions. Biomass growth is the

Temudo et al. (2007) suggested that besides NADH2/NAD,

ATP-consuming reaction using glucose as biomass building

ferredoxin (Fd/FdH2) has to play an important role in

Figure 1

|

Scheme of catabolic reactions considered in the model with the free energy change per electron of the electron transfer reactions in the different pathways, as well as the free energy change for the reduction of the three electron carriers considered. Based on theoretical considerations and literature information the most plausible electron carrier for the different electron transfer steps are identified. Standard free energy change values of the different reactions are corrected for a pH of 7, and concentrations dissolved species of 1 mM (Hanselmann 1991; Thauer et al. 1977).


495


electron transfer in the MCF process. NADH2 is an unlikely

Modification 2: gas-liquid mass transfer and formate

electron donor for the generation of molecular hydrogen at

versus hydrogen/carbon dioxide as electron sink

the elevated hydrogen partial pressures in the MCF process ( pH2 , 0.01– 0.50 atm). This range of pH2 values suggests the actual NADH2/NAD ratio pressure to be least 1 £ 104 which is physiologically not feasible. We have therefore introduced the generation of FdH2 in the model associated with the oxidation of pyruvate to acetyl-CoA as shown in Figure 1. FdH2 allows for generation of molecular hydrogen at FdH2:Fd-ratios of at least approximately 0.1. Since the pyruvate to acetyl-CoA conversion reaction is not of importance in all product formation pathways, the overall stoichiometry of the catabolic pathways considered changes

At pH-values higher than 6, formate (C1) was found as dominant electron sink in the anaerobic fermentation of glucose (Temudo et al. 2007). At lower pH-values molecular hydrogen/carbon dioxide (H2/CO2) dominates as product from what we assume to be reoxidation of FdH2. The split between H2/CO2 and C1 was modeled as a thermodynamic equilibrium reaction as suggested by Temudo et al. (2007) and shown in Figure 2. To include the gas-liquid mass transfer from CO2 and H2 mass transfer coefficients were defined for both compounds.

as shown in Table 1. The oxidation state of NADH2/NAD and FdH2/Fd is assumed to be fully uncoupled, and the NADH2/NAD is treated as a conserved moiety, whereas FdH2 generation is directly linked to the production of H2/CO2 (or formate, see below)

Modification 3: maximum intracellular concentration volatile fatty acids The intracellular concentrations of organic acids are

The role of FADH2 is unchanged from the original

included in the model structure to describe the energetic

model: the reoxidation of FADH2 is assumed to be coupled

requirements for product extrusion due to pH-gradients

to reduction of NAD and the free energy available in this

across the cytoplasmic membrane. At acid extracellular

electron transfer reaction is assumed to be coupled to the

pH-values, metabolic energy needs to be invested to

extrusion of a proton across the cytoplasmic membrane.

maintain a near neutral intracellular pH, whereas at higher

The resulting proton motive force is assumed to drive ATP-

extracellular pH-values product excretion is suggested to be

production using a membrane bound ATP-ase resulting in

associated with the generation of metabolic energy in the

the stoichiometries shown in Table 1. The stoichiometry of

form of a proton motive force.

the anabolism was initially considered to include the

In the previous version of the model, a maximum value

production of inorganic carbon due to decarboxylation

was assumed for the intracellular concentration of the

reactions (Gommers et al. 1988). In the model described

individual organic acids formed. This is not realistic since

here we have excluded the anabolic decarboxylation

the intracellular organic acid concentrations will normally

resulting in a standard stoichiometry of the anabolism as

be defined by the intracellular pH-buffer capacity and

shown in Table 1. Table 1

|

Catabolic product formation stoichiometries considered in the MCF model. All yields are expressed in mol per mol glucose (S)

Product

YP/S

YATP/S

YNADH2/S

YFdH2/S

Acetate

2.0

4.0

2.0

2.0

Propionate

2.0

2.7

2 2.0

0.0

Butyrate

1.0

3.3

0.0

2.0

Ethanol

2.0

2.0

2 2.0

2.0

Lactate

2.0

2.0

0.0

0.0

Biomass (CH1.8O0.5N0.2)p

6.0

2 18.0

2 0.6

0.0

p The stoichiometry of the biomass production pathway has been changed to enable the inclusion of different electron carriers in the anabolic pathways.

Figure 2

|

Schematic drawing of the formate and hydrogen/inorganic carbon mass transfer and thermodynamic equilibria considered in the model. Values used for the constants have been presented elsewhere, except for the kLaL values for H2 and CO2 which were both set at 1.8 h21.

496



consequently the sum of the individual organic acids

entire pH-range of 4 to 9 the model predicts butyrate as

formed. We have added the constraint of a maximum total

dominant catabolic product. Minor amounts of acetate are

concentration of organic acids to the model structure.

produced due to the anabolic requirement for NADH2 (see Table 1). At pH-values between 7 and 9 a minor fraction of

Modification 4: numerical solution method applied

the glucose fermented is converted to a propionate acetate mixture. Compared to the production of butyrate the ATP-

As opposed to the first version of the model we have now

based yield as shown in Table 1 are equal, but at higher

implemented the model evaluation procedure as a global

pH-values the transmembrane proton gradient enables the

optimization routine based on a combined randomized

generation of small amounts of ATP favoring the pro-

initial parameter evaluation and a Newton-Raphson based

duction of two acid equivalents (acetate and propionate)

search routine. Optimization is based on maximization of

over one acid equivalent (butyrate).

the catabolic ATP-production while maintaining the con-

The thermodynamic equilibrium assumption for C1

straint of (i) thermodynamically favorability of all reactions,

versus H2/CO2 furthermore induces a shift from H2/CO2 to

(ii) thermodynamic equilibrium between C1 and H2/CO2,

C1 at increasing pH. This trend is comparable to our

(iii) the ATP-balance, and (iv) the maximum intracellular

experimental observations and suggests that C1 dehydro-

concentration organic acids. Variables to be optimized are

genation is a thermodynamically controlled process.

the intra- and extracellular concentrations acetate (C2),

With regard to the predicted liquid composition of

propionate (C3), butyrate (C4), and lactate (Lac), the

organic acids the measured trend of ethanol and acetate

extracellular concentration ethanol (EOH) and biomass (X),

production at higher pH-values is not reproduced by the

and the split factor for C1 versus H2/CO2.

model. This suggests that the transition observed at increasing pH-values from C4 to C2/EOH is not controlled by the thermodynamic optimization criterion proposed here.

MODEL RESULTS AND DISCUSSION Glucose fermentation pattern as a function of the pH

In this respect it should be noted that the actual difference in ATP-yield between C4 and C2/EOH production is very small (3.3 versus 3.0 molATP molC621). An unknown

Figure 3 shows the glucose fermentation pattern as a

specific biochemical or kinetic reason is suggested to explain

function of the pH as predicted by the model. Across the

the transition in product formation observed.

Figure 3

|

Product spectrum from anaerobic glucose fermentation (0.06 M) as a function of the pH. The left graph shows the concentrations acetate (C2), propionate (C3), butyrate (C4) and biomass (X) as a function of the pH. The right graph shows the concentrations gaseous carbon dioxide and molecular hydrogen expressed as mol per litre liquid, as well as the liquid concentrations formate (C1) and bicarbonate (HCO3).

497



less C4) is predicted to be produced because C2 production is associated with NADH2 production (Table 1). This prediction corresponds to our experimental observation that in all experimental conditions C2 is one of the dominant catabolic end products (Temudo et al. 2007). In future research we will try to elucidate the role of different electron carriers in the anabolic pathway and the impact on the catabolic product spectrum.

ACKNOWLEDGEMENTS This work was supported by the Dutch Technology Foundation (STW), project no. DPC.5904 Figure 4

|

Fermentation product spectrum as a function of the anabolic NADH2 requirements during anaerobic glucose fermentation. The standard stoichiometry as defined in Table 1 is YNADH2 /X ¼ 20.1 mol mol21. Calculations were conducted at a pH of 6.0 and for 0.06 M glucose conversion.

Electron carriers required in the anabolism One of the uncertainties in the model is the type of electron carriers used in the anabolism. The net consumption of reduced electron carriers corresponds only to 0.2 mol electron per mol biomass as shown in Table 1, but various anabolic steps are reported to be associated with NADH2 or even NADPH2 oxidation. This may imply that the anabolic stoichiometry proceeds according to the following generalized reaction stoichiometry: 20:17·C6 H12 O6 2 ð0:1 þ xÞ·NADH2 2 0:2·NH3 þ CH1:8 O0:5 N0:2 þ x·FdH2 This equation suggests a net transfer of electrons from NADH2 to FdH2 and subsequently C1 and H2/CO2 during anabolism. Varying the variable x between 0 and 1 and solving the model results in the dependency of the product composition shown in Figure 4. Upon the requirement of more NADH2 in the anabolism, more C2 (and consequently

REFERENCES Gommers, P. J. F., Vanschie, B. J., Vandijken, J. P. & Kuenen, J. G. 1988 Biochemical limits to microbial-growth yields –an analysis of mixed substrate utilization. Biotechnol. Bioeng. 32(1), 86 –94. Hanselmann, K. W. 1991 Microbial energetics applied to waste repositories. Experientia 47(7), 645–687. Mosey, F. E. 1983 Mathematical-modeling of the anaerobicdigestion process –regulatory mechanisms for the formation of short-chain volatile acids from glucose. Water Sci. Technol. 15(8 –9), 209 –232. Rodriguez, J., Kleerebezem, R., Lema, J. M. & van Loosdrecht, M. C. M. 2006 Modeling product formation in anaerobic mixed culture fermentations. Biotechnol. Bioeng. 93(3), 592–606. Temudo, M. F., Kleerebezem, R. & van Loosdrecht, M. C. M. 2007 Influence of the pH on open (mixed) culture fermentation of glucose, a chemostat study. Biotechnol. Bioeng. 98(1), 69 –79. Thauer, R. K., Jungermann, K. & Decker, K. 1977 Energyconservation in chemotropic anaerobic bacteria. Bacteriol. Rev. 41(1), 100– 180. Zoetemeyer, R. J., Arnoldy, P., Cohen, A. & Boelhouwer, C. 1982a Influence of temperature on the anaerobic acidification of glucose in a mixed culture forming part of a 2-stage digestion process. Water Res. 16(3), 313– 321. Zoetemeyer, R. J., Vandenheuvel, J. C. & Cohen, A. 1982b pH influence on acidogenic dissimilation of glucose in an anaerobic digester. Water Res. 16(3), 303– 311.