**DOI:**10.1128/AEM.00333-06

## ABSTRACT

A simple model (termed the syntrophy model) for simulating the contribution of coaggregation to interspecies hydrogen fluxes between syntrophic bacteria and methanogenic archaea is described. We applied it to analyzing partially aggregated syntrophic cocultures with various substrates, revealing that large fractions of hydrogen molecules were fluxed in aggregates.

The syntrophic interaction between fermentative bacteria and methanogenic archaea is an essential part of methanogenesis (18) and has been found in such ecosystems as rice paddy fields (5), freshwater sediments (4), mammalian digestive tracts (29), petroleum-contaminated soil (12), and anaerobic digesters for organic waste treatment (6, 9, 18, 21, 24). In this process, reducing equivalents (i.e., H_{2} and/or formate) produced by fermentative bacteria should be efficiently consumed by methanogenic archaea in order for the bacteria and archaea to grow actively (2, 3, 24, 27). This is particularly important for syntrophic volatile fatty acid (such as butyrate, propionate, and acetate) oxidation, since this reaction is endergonic under standard conditions and is thermodynamically feasible only when the H_{2} partial pressure (or formate concentration) is kept very low (1, 6, 20, 24, 27).

It has been theoretically suggested that close physical contact between volatile fatty acid-fermenting syntrophic bacteria (syntrophs) and methanogenic archaea (methanogens) is important for efficient interspecies electron transfer (2, 19, 24, 28). Our previous study has also indicated that coaggregation facilitated interspecies hydrogen transfer between a propionate-oxidizing syntroph, *Pelotomaculum thermopropionicum* SI, and a hydrogen-consuming methanogen, *Methanothermobacter thermautotrophicus* ΔH (11). However, since we were unable to separately determine hydrogen flux in aggregates and that between dispersed cells, the contribution of aggregates to interspecies hydrogen transfer has not yet been quantitatively evaluated. In the present study, we developed and applied a model (named the syntrophy model) for simulating the contribution of coaggregation to interspecies hydrogen flux between syntrophs and methanogens.

## Model development.

Hydrogen flux was estimated based on Fick's diffusion law (equation 1).
$$mathtex$$\[J{=}D_{\mathrm{H}2}\ \frac{C_{\mathrm{H}2-\mathrm{syntroph}}{-}C_{\mathrm{H}2-{\Delta}\mathrm{H}}}{d}\]$$mathtex$$(1)
In this equation, *J* is the interspecies hydrogen flux, *D*_{H2} is the H_{2} diffusion constant in water (at 55°C), *C*_{H2-syntroph} is the H_{2} concentration immediately outside a syntroph cell, *C*_{H2-}_{Δ}_{H} is the H_{2} concentration immediately outside a ΔH cell, and *d* is the average distance between the syntroph and ΔH cells (for units of parameters, refer to the tables). Total interspecies hydrogen flux (*Q*_{H2}) is stoichiometrically correlated with the methane production rate (four times the methane production rate) and calculated by multiplying the *J* value by the total surface area of hydrogen-releasing syntroph cells.
$$mathtex$$\[Q_{\mathrm{H}2}{=}X_{\mathrm{syntroph}}\ {\cdot}\ V\ {\cdot}\ A_{\mathrm{syntroph}}\ {\cdot}\ J\]$$mathtex$$(2)
In this equation, *X*_{syntroph} is the cell concentration of syntroph, *V* is the culture volume, and *A*_{syntroph} is the surface area of a syntroph cell.

In order to apply this estimation method to partially aggregated cocultures, *Q*_{H2} between aggregated cells and that between dispersed cells were separately estimated as follows.
$$mathtex$$\[Q_{\mathrm{H2-agg}}{=}X_{\mathrm{agg-syntroph}}\ {\cdot}\ V\ {\cdot}\ A_{\mathrm{syntroph}}\ {\cdot}\ D_{\mathrm{H}2}\ \frac{C_{\mathrm{H}2-\mathrm{syntroph}}{-}C_{\mathrm{H}2-{\Delta}\mathrm{H}}}{d_{\mathrm{agg}}}\]$$mathtex$$(3)$$mathtex$$\[Q_{\mathrm{H2-dis}}{=}X_{\mathrm{dis-syntroph}}\ {\cdot}\ V\ {\cdot}\ A_{\mathrm{syntroph}}\ {\cdot}\ D_{\mathrm{H}2}\ \frac{C_{\mathrm{H}2-\mathrm{syntroph}}{-}C_{\mathrm{H}2-{\Delta}\mathrm{H}}}{d_{\mathrm{dis}}}\]$$mathtex$$(4)$$mathtex$$\[Q_{\mathrm{H}2}{=}Q_{\mathrm{H2-agg}}{+}Q_{\mathrm{H2-dis}}\]$$mathtex$$(5)
In these equations, *Q*_{H2-agg} and *Q*_{H2-dis} are the total hydrogen flux between aggregated cells and that between dispersed cells, respectively, *X*_{agg-syntroph} and *X*_{dis-syntroph} are the concentration of aggregated syntroph cells and that of dispersed cells, respectively, and *d*_{agg} and *d*_{dis} are the mean interspecies distance between aggregated cells and that between dispersed cells, respectively. This scheme for estimating a *Q*_{H2} value was named “the syntrophy model.”

## Growth and coaggregation.

We have reported that cells in coculture of *P. thermopropionicum* SI (8) and *M. thermautotrophicus* ΔH were partially aggregated (11). The present study also analyzed cocultures of strain ΔH with butyrate-oxidizing *Syntrophothermus lipocalidus* TGB-C1 (22) and acetate-oxidizing *Thermacetogenium phaeum* PB (7). They were grown in 100-ml serum vials containing 50 ml of a culture medium as described elsewhere previously (22). The culture medium was supplemented with 0.1% Bacto yeast extract (Difco) and a growth substrate at 17 to 20 mM. Cultivation was conducted at 55°C under an atmosphere of N_{2} plus CO_{2} (80/20 [vol/vol]) without shaking. Cultivation was initiated by inoculation with 5 ml of a preculture in the same medium. Hydrogen releases from acetate by strain PB and from butyrate by strain TGB-C1 were examined in monocultures inoculated with 5 ml of precultures grown on methanol (PB) and crotonate (TGB-C1), in which methanol or crotonate was completely lost. Microscopic analyses and gas chromatography were conducted as described previously (11).

We found that although cells in monocultures of PB and TGB-C1 (grown on methanol and crotonate, respectively) were fully dispersed, their cells in cocultures with ΔH were partially aggregated (see Fig. S1 and S2 in the supplemental material for growth curves and phase-contrast micrographs, respectively). Cells in their cocultures were also observed by field emission-scanning electron microscopy (10), revealing that aggregates were comprised of both syntroph and methanogen cells (see Fig. S3 in the supplemental material). In addition, extracellular filamentous appendages were found in these cocultures, which connected syntroph cells to methanogen cells (see Fig. S3 in the supplemental material).

Our previous study also found that strain SI utilized flagellum-like filaments for making contact with strain ΔH (11). Together with the results of the present study, we consider that the connection of syntroph and methanogen cells with filaments is a widespread phenomenon. In order to know more specifically how these filaments contribute to coaggregation, molecular analyses of filaments should be done. It has been known that extracellular filaments function as adhesins in many different types of bacteria (10, 14, 15, 17, 23, 26); information in those studies will be useful for examining the role of extracellular filaments of syntrophs.

*X*_{agg-syntroph} and *X*_{dis-syntroph}.

In order to estimate *Q*_{H2} values, we first determined *X*_{agg-syntroph} and *X*_{dis-syntroph} values for strains SI, PB, and TGB-C1 in cocultures with strain ΔH. For this, we measured optical densities at 600 nm (OD_{600}) of a coculture before and after gentle homogenization with a tissue grinder, as described previously (11), until the OD_{600} no longer increased. Phase-contrast images of several homogenized cultures revealed that cell aggregates were fully dispersed (data not shown). The OD_{600} values measured were converted to a total cell concentration (a sum of syntroph and methanogen cells, *X*_{total}) using equations 6 to 8; these equations were produced from standard curves obtained by measuring OD_{600} and cell concentrations (DAPI [4′,6′-diamidino-2-phenylindole] counts) (11) of several fully dispersed cultures.
$$mathtex$$\[X_{\mathrm{total}\ (\mathrm{SI}/{\Delta}\mathrm{H})}({\times}\ 10^{7}\ \mathrm{cells\ ml}^{{-}1}){=}122\ {\cdot}\ \mathrm{OD}_{600}{-}0.92\]$$mathtex$$(6)$$mathtex$$\[X_{\mathrm{total}\ (\mathrm{TGB}-\mathrm{C}1/{\Delta}\mathrm{H})}({\times}\ 10^{7}\ \mathrm{cells\ ml}^{{-}1}){=}318\ {\cdot}\ \mathrm{OD}_{600}{-}3.40\]$$mathtex$$(7)$$mathtex$$\[X_{\mathrm{total}\ (\mathrm{PB}/{\Delta}\mathrm{H})}({\times}\ 10^{7}\ \mathrm{cells\ ml}^{{-}1}){=}158\ {\cdot}\ \mathrm{OD}_{600}{-}3.44\]$$mathtex$$(8)
The concentration of total aggregated cells (*X*_{agg}) was determined using equation 9, while the concentration of total dispersed cells (*X*_{dis}) was calculated from *X*_{agg} using equation 10.
$$mathtex$$\[X_{\mathrm{agg}}\ (\%){=}\ \frac{X_{\mathrm{total}}\ \mathrm{after\ dispersion}{-}X_{\mathrm{total}}\ \mathrm{before\ dispersion}}{X_{\mathrm{total}}\ \mathrm{after\ dispersion}}{\times}100\]$$mathtex$$(9)$$mathtex$$\[X_{\mathrm{dis}}{=}X_{\mathrm{total}}{-}X_{\mathrm{agg}}\]$$mathtex$$(10)*X*_{agg} and *X*_{dis} values can also be estimated from DAPI microscopic counts of cells before and after homogenization, while the values determined by the OD measurement agreed well with those determined by the DAPI counts (data not shown).

In order to estimate a ratio of the number of syntroph cells to that of total dispersed cells, dispersed cells were observed using phase-contrast micrography (see Fig. 1A, for example) and fluorescence microscopy. Syntroph cells and methanogen cells could be discriminated according to the differences in cell shape and the *F*_{420} autofluorescence (9). An *X*_{dis-syntroph} value was estimated from *X*_{dis} and the ratio of the number of syntroph cells. On the other hand, the ratio of the number of syntroph cells to that of the total aggregated cells was determined based on data from thin-section images of coaggregates obtained by transmission electron microscopy (TEM) (Fig. 1B). TEM images were obtained by a standard procedure (13) using an H7000 transmission electron microscope (Hitachi). As shown in Fig. 1B, syntroph cells could be distinguished from ΔH cells by shape and darkness. An *X*_{agg-syntroph} value was estimated from *X*_{agg} and the ratio of the number of syntroph cells (*n* > 90 for all cases).

Table 1 summarizes *X*_{agg}, *X*_{agg-syntroph}, *X*_{dis}, *X*_{dis-syntroph}, *X*_{total}, and *X*_{total-syntroph} values for four types of cocultures at mid-log growth phases, namely, SI plus ΔH grown on ethanol, SI plus ΔH grown on propionate, TGB-C1 plus ΔH grown on butyrate, and PB plus ΔH grown on acetate. Table 1 shows that SI/ΔH (propionate) and PB/ΔH (acetate) coaggregated at relatively high ratios, although the *X*_{agg} values were not high compared to *X*_{dis} values in all cases. In addition, we found that methanogen cells occupied aggregates more abundantly than did syntroph cells; this trend was prominent in cocultures of SI/ΔH (ethanol) and TGB-C1/ΔH (butyrate). We deduce that syntroph cells should have more vigorously aggregated when they were grown on energetically unfavorable substrates, i.e., propionate and acetate.

*d*_{agg} and *d*_{dis}.

We calculated a *d*_{dis} value by supposing that dispersed cells were randomly distributed in the liquid phase (2, 11, 16, 24). In contrast, a *d*_{agg} value was determined by analyzing thin-section TEM pictures of aggregates (see Fig. 1B, for example). In this analysis, we supposed that cells were cylindrical and determined a mean radius of syntroph cells (*R*_{s}), a mean radius of ΔH cells (*R*_{d})b and a mean minimal interspecies distance (*d*_{agg-min}) by analyzing over 80 syntroph/ΔH pairs for each coculture. From these values, an interspecies distance at angle θ (Fig. 1B) was calculated using equation 11.
$$mathtex$$\[d_{\mathrm{agg}}({\theta}){=}[(R_{\mathrm{s}}\ {\cdot}\ \mathrm{sin}{\theta}\ {-}\ R_{\mathrm{d}}\ {\cdot}\ \mathrm{sin}{\theta})^{2}{+}(R_{\mathrm{s}}{+}d_{\mathrm{agg-min}}{+}R_{\mathrm{d}}{-}R_{\mathrm{s}}\ {\cdot}\ \mathrm{cos}{\theta}{-}R_{\mathrm{d}}\ {\cdot}\ \mathrm{cos}{\theta})^{2}]^{0.5}\]$$mathtex$$11The θ value ranged from 0 to π/2. For estimating *d*_{agg}, π/2 was divided into *n* parts, and *d*_{agg}(θ) at each θ point was calculated. Equation 12 was used for estimating *d*_{agg} from the *d*_{agg}(θ) values.
$$mathtex$$\[d_{\mathrm{agg}}{=}\ \frac{{{\sum}_{{\theta}\ {=}\ 0}^{{\pi}/2}}\ d_{\mathrm{agg}}({\theta})}{n}\]$$mathtex$$(12)
In our analyses, *n* was arbitrarily set at 90. Table 2 presents parameters in equation 11 and *d*_{agg} and *d*_{dis} values determined for the cocultures used in the present study. It is shown that the *d*_{agg} values for the different cocultures were similar to each other and that the *d*_{dis} values were approximately 100-fold higher than the *d*_{agg} values.

## Estimation of hydrogen fluxes.

In order to estimate *Q*_{H2-agg} and *Q*_{H2-dis}, parameters in equations 3 and 4, other than *X*_{agg-syntroph}, *X*_{dis-syntroph}, *d*_{agg}, and *d*_{dis}, were determined as follows. *D*_{H2} was cited from data reported previously by Wise and Houghton (30). *C*_{H2-}_{Δ}_{H} was defined as the minimum H_{2} concentration, above which an H_{2}-consuming methanogen can gain energy by carbonate respiration (6). *A*_{syntroph} was estimated by assuming that cells were cylindrical and measuring diameters and lengths of syntroph cells in the field emission-scanning electron microscopy photos (see Fig. S3 in the supplemental material) (11). *C*_{H2-syntroph} was determined to be the H_{2} concentration at a time point of microscopic analysis in the exponential growth phase, which was obtained from an H_{2} release curve (see Fig S1 in the supplemental material for strains PB and TGB-C1) (see reference 11 for strain SI). These values are summarized in Table 2.

Table 3 summarizes *Q*_{H2-agg} and *Q*_{H2-dis} values estimated as described above, in which we also present *Q*_{H2} values and total H_{2} fluxes experimentally determined from methane production rates (see Fig. S1 in the supplemental material) (11). The estimation revealed that although numbers of syntroph cells in aggregates (*X*_{agg-syntroph}) were relatively small in all cocultures, they contributed largely to total hydrogen fluxes (49 to 92%). Notably, coaggregation was found to contribute largely to syntrophic propionate and acetate oxidation. It is also shown that the estimated total hydrogen fluxes (*Q*_{H2} values) agreed well with those experimentally determined, supporting the adequacy of the syntrophy model.

We show here that the simulation model was applicable to several different syntrophic methanogenic cocultures irrespective of coculture members and substrates, suggesting that the model is widely applicable for partially aggregated heterogeneous systems. Although previous studies have theoretically discussed the importance of close physical contact for propionate oxidation (2, 11, 25, 28), the model could quantitatively show that coaggregation was important not only for oxidation of propionate, butyrate, and acetate (representing thermodynamically unfavorable substrates) but also for ethanol oxidation. Based on the results, we suggest that aggregation is the key factor for engineering syntrophic methanogenesis, for which the simulation model described herein will be useful.

## ACKNOWLEDGMENTS

We thank Yoichi Kamagata for providing *T. phaeum* PB and *S. lipocalidus* TGB-C1 and Hiroyuki Imachi for providing *P. thermopropionicum* SI. We also thank Masanori Arita for valuable advice for model development, Kohei Nakamura and Miho Enoki for valuable suggestions, and Yasuo Igarashi and Hiroshi Ikenaga for continuous encouragement. We are grateful to Katsutoshi Hori and Mika Atsumi for help in electron microscopy and Reiko Hirano for technical assistance.

This work was supported by the New Energy and Industrial Technology Development Organization (NEDO).

## FOOTNOTES

- Received 10 February 2006.
- Accepted 28 April 2006.

- Copyright © 2006 American Society for Microbiology