Abstract
Nutrient cross-feeding can stabilize microbial mutualisms, including those important for carbon cycling in nutrient-limited anaerobic environments. It remains poorly understood how nutrient limitation within natural environments impacts mutualist growth, cross-feeding levels, and ultimately mutualism dynamics. We examined the effects of nutrient limitation within a mutualism using theoretical and experimental approaches with a synthetic anaerobic coculture pairing fermentative Escherichia coli and phototrophic Rhodopseudomonas palustris. In this coculture, E. coli and R. palustris resemble an anaerobic food web by cross-feeding essential carbon (organic acids) and nitrogen (ammonium), respectively. Organic acid cross-feeding stemming from E. coli fermentation can continue in a growth-independent manner during nutrient limitation, while ammonium cross-feeding by R. palustris is growth-dependent. When ammonium cross-feeding was limited, coculture trends changed yet coexistence persisted under both homogenous and heterogenous conditions. Theoretical modeling indicated that growth-independent fermentation was crucial to sustain cooperative growth under conditions of low nutrient exchange. We also show that growth-independent fermentation sets the upper E. coli cell density at which this mutualism is supported. Thus, growth-independent fermentation can conditionally stabilize or destabilize a mutualism, indicating the potential importance of growth-independent metabolism for nutrient-limited mutualistic communities.
Conflict of interest The authors declare no conflict of interest.
Introduction
Mutualistic cross-feeding interactions between microbes crucially impact diverse processes ranging from human health (Hammer et al., 2014; Ramsey and Whiteley, 2009; Flint et al., 2007) to biogeochemical cycles (Morris et al., 2013; McInerney et al., 2010; Durham et al., 2015). Within most environments, microbial communities experience prolonged periods of nutrient limitation (Lever et al., 2015). In general, bacteria tolerate nutrient limitation by modulating their growth and metabolism (Ferenci, 2001; Russell and Cook, 1995; Wanner and Egli, 1990; Rittershaus et al., 2013; Lee et al., 1976). Sub-optimally growing and even non-growing cells can survive by retaining low metabolic activity to generate maintenance energy (Wanner and Egli, 1990; Russell and Cook, 1995; Rittershaus et al., 2013; Hoehler and Jørgensen, 2013). It is possible that growth-independent metabolic products could serve to cross-feed other microbes and thereby influence the initiation and/or endurance of microbial mutualisms under growth-limiting conditions. Nonetheless, most microbial cross-feeding studies view nutrient release as being tightly coupled to growth. While mutualism flux balance models tend to include growth-independent maintenance energy parameters (Harcombe et al., 2014; Chubiz et al., 2015), most other mutualism models do not, and few studies have examined the impact of growth-independent cross-feeding on mutualism dynamics (Shou et al., 2007; Megee III et al., 1972; Stolyar et al., 2007).
Growth-independent metabolism is fundamental to fermentative microbes during nutrient limitation, as they must continue to ferment and excrete products, albeit at a lower rate than during growth, to generate maintenance energy. For example, in the absence of electron acceptors, and starved for essential elements (i.e., nitrogen or sulfur), Escherichia coli generates energy by fermenting glucose in a growth-independent manner (Wanner and Egli, 1990; LaSarre et al., 2016). Fermentative microbes serve pivotal roles within natural anaerobic food webs, wherein their excreted products serve as nutrients for other microbes. As such, growth-independent fermentation could be an important cross-feeding mechanism within anaerobic communities.
Studying mutualistic cross-feeding in natural environments can be challenging due to environmental and genetic stochasticity. Synthetic microbial communities, or cocultures, offer an alternative approach that mimics key aspects of natural communities while providing a greater degree of experimental control (Widder et al., 2016; Momeni et al., 2011; Ponomarova and Patil, 2015; Lindemann et al., 2016). We previously developed a bacterial coculture to facilitate the study of mutualistic cross-feeding in anaerobic environments (LaSarre et al., 2016). Our coculture resembles other fermenter-photoheterotroph cocultures, which have primarily been studied for converting plant-derived sugars into H2 biofuel (Odom and Wall, 1983; Fang et al., 2006; Sun et al., 2010; Ding et al., 2009). However, unlike previous systems, our coculture enforces stable coexistence through bi-directional cross-feeding of essential nutrients. Specifically, E. coli ferments sugars to excreted organic acids, providing essential carbon and electrons for a genetically engineered Rhodopseudomonas palustris strain (Nx). R. palustris Nx has a NifA* mutation (McKinlay and Harwood, 2010a) that results in NH4+ excretion during N2 fixation, providing essential nitrogen for E. coli (Figure 1) (LaSarre et al., 2016). We previously used our coculture to examine the effects of increased NH4+ cross-feeding on coculture dynamics (LaSarre et al., 2016). In that study, theoretical modeling suggested that coexistence would persist even at very low NH4+ excretion levels (LaSarre et al., 2016). This prediction prompted us to ask herein: how does this mutualism contend with limitation of cross-fed nutrients?
Using theoretical and experimental approaches, we show that growth-independent fermentation is crucial for maintaining cooperative growth during limitation of cross-fed NH4+. Conversely, we also show that growth-independent fermentation can be detrimental to the mutualism under certain circumstances. Specifically, growth-independent fermentation can inhibit coculture growth when large E. coli populations magnify the otherwise low growth-independent fermentation rate; this leads to organic acid production that outpaces consumption and ultimately acidifies the environment and halts growth. Thus, growth-independent cross-feeding conditionally influences this mutualism in positive and negative manners and thereby sets both the upper and lower thresholds for cooperation.
Materials and Methods
Strains, plasmids, and growth conditions
Strains are listed in Supplementary Table 1. E. coli and R. palustris were cultivated on Luria-Burtani (LB) agar or defined mineral (PM) (Kim and Harwood, 1991) agar with 10 mM succinate, respectively. For determining colony forming units (CFU), LB agar or PM agar minus (NH4)2SO4 were used for E. coli and R. palustris, respectively. Cultures were grown in 10-mL of defined M9-derived coculture medium (MDC) (LaSarre et al., 2016) in 27-mL anaerobic test tubes. The medium was made anaerobic by bubbling with N2, sealed with rubber stoppers and aluminum crimps, and then autoclaved. After autoclaving, MDC was supplemented with cation solution (1 % v/v; 100 mM MgSO4 and 10 mM CaCl2) and glucose (25 mM). For defined N2 concentrations, the medium was bubbled with argon and after autoclaving defined volumes of N2 were injected through a 0.2 micron syringe filter. All cultures were grown at 30°C either laying horizontally under a 60 W incandescent bulb with shaking at 150 rpm (shaking conditions) or upright without agitation (static conditions). Static cultures were only mixed for sampling upon inoculation and at the termination of an experiment. Thus, growth rates were not measured under static conditions. Starter cultures were inoculated with 200 μL MDC containing a suspension of a single colonies of each species. Test cocultures were inoculated using a 1% inoculum from starter cocultures. For serial transfers, cocultures were incubated for either one week (shaking), two weeks (100% N2; static), or four weeks (18% N2; static) before transferring a 1% stationary phase inoculum to fresh medium.
Analytical procedures
Cell density was assayed by optical density at 660 nm (OD660) using a Genesys 20 visible spectrophotometer (Thermo-Fisher, Waltham, MA, USA). Growth curve readings were taken in culture tubes without sampling. Specific growth rates were determined using measurements between 0.1-1.0 OD660 where there is linear correlation between cell density and OD660. Final OD660 measurements were taken in cuvettes and samples were diluted into the linear range as necessary. To compare cell densities between growth conditions, CFUs were converted into growth yields by calculating CFUs per μmol glucose consumed, as N2 limitation prevented complete glucose consumption during the assay period. H2 and N2 were quantified using a Shimadzu (Kyoto, Japan) gas chromatograph (GC) with a thermal conductivity detector as described (Huang et al., 2010). Glucose, organic acids, and ethanol were quantified using a Shimadzu high-performance liquid chromatograph (HPLC) as described (McKinlay et al., 2005). NH4+ was quantified using an indophenol colorimetric assay as described (LaSarre et al., 2016).
Mathematical modeling
A Monod model describing bi-directional cross-feeding in batch cultures, called SyFFoN_v2 (Syntrophy between Fermenter and Fixer of Nitrogen), was modified from our previous model (LaSarre et al., 2016) as follows: (i) a sigmoidal function, rather than a Monod function, was used to control the transition to growth-independent fermentation (10/(10+1.09(1000*uEc))); (ii) sigmoidal functions were used to transition from NH4+ excretion (1-(40/(40+1.29N))) to H2 production (40/(40+1.29N)) by R. palustris as N2 becomes limiting; (iii) a sigmoidal function was used to simulate the inhibiting effects of accumulated organic acids on both growth and metabolism for both species (bx/(bx+10(f+C))); (iv) a sigmoidal function was used to dampen growth-independent fermentation rates when consumable organic acids (lactate, succinate, and acetate) accumulate (rx*(100/(100+6C)) + rx_mono) and simulate the slow growth-independent fermentation observed in E. coli monocultures (LaSarre et al., 2016), compared to faster growth-independent fermentation in coculture; (v) R. palustris H2 production was coupled to consumable organic acid depletion, assuming that 0.5 CO2 are produced per H2 (McKinlay et al., 2014); (vi) the R. palustris Km for N2 (KN) was given a value of 6 mM, based on the change in growth rate at limiting N2 concentrations in coculture; (vii) product formation parameters (R and r) were increased to more accurately simulate observed growth rates in coculture; (viii) the E. coli acid resistance parameter (bEc) was increased relative to that for R. palustris (bRp) based on terminal pH values observed in E. coli monocultures versus cocultures. Equations are listed below with default values in Supplementary Table 2. SyFFoN_v2 runs in R studio and is available for download at: https://github.com/McKinlab/Coculture-Mutualism.
Equations 1 and 2 were used to describe E. coli and R. palustris growth rates:
Equations 3-12 were used to describe temporal changes in cell densities and extracellular compounds. Numerical constants in product excretion equations are used to account for molar stoichiometric conversions. Numerical constants used in sigmoidal functions are based on those values that resulted in simulations resembling empirical trends. All R and r parameters are expressed in terms of glucose consumed except for RA, which is the amount of NH4+ produced per R. palustris cell (Supplementary Table 2).
Where, μ is the specific growth rate of the indicated species (h-1). μMAX is the maximum specific growth rate of the indicated species (h-1). G, A, C, N, f, e, H and CO2 are the concentrations (mM) of glucose, NH4+, consumable organic acids, N2, formate, ethanol, H2, and CO2, respectively. All gasses are assumed to be fully dissolved. Consumable organic acids are those that R. palustris can consume, namely, lactate (3 carbons), acetate (2 carbons), and succinate (4 carbons). All consumable organic acids were simulated to have three carbons for convenience. Only net accumulation of formate, ethanol, CO2 and H2 are described in accordance with observed trends. K is the half saturation constant for the indicated substrate (mM). Ec and Rp are the cell densities (cells/ml) of E. coli and R. palustris, respectively. b is the ability of a species to resist the inhibiting effects of acid (mM). Y is the E. coli or R. palustris cell yield from the indicated substrate (cells / μmol glucose). Y values were determined in MDC with the indicated substrate as the limiting nutrient. R is the fraction of glucose converted into the indicated compound per E. coli cell during growth (μmol of glucose / E. coli cell), except for RA. Values were adjusted to accurately simulate product yields measured in cocultures and in MDC with and without added NH4Cl. RA is the ratio of NH4+ produced per R. palustris cell during growth (μmol / R. palustris cell). The default value was based on that which accurately simulated empirical trends. r is the growth-independent rate of glucose converted into the indicated compound (μmol / cell / h). Default values are based on those which accurately simulated empirical trends in coculture. r_mono is the growth-independent rate of glucose converted into the indicated compound by E. coli when consumable organic acids accumulate. Default values are based on linear regression of products accumulated over time in nitrogen-free cell suspensions of E. coli (LaSarre et al., 2016).
Results
Coexistence is maintained at reduced NH4+ cross-feeding levels
Previously, we found that stable coexistence and reproducible trends in our mutualistic coculture were dependent on the transfer of NH4+ from R. palustris Nx to E. coli (LaSarre et al., 2016). Adding NH4+ to the medium broke the dependency of E. coli on R. palustris and resulted in E. coli domination due to its higher intrinsic growth rate relative to that of R. palustris. Thus, the NH4+ cross-feeding level controls the E. coli growth rate within the mutualism. We were intrigued that theoretical modeling predicted mutualism coexistence would be maintained even at very low NH4+ cross-feeding levels, as such levels should severely limit or event halt E. coli growth (Supplementary Figure 1a) (LaSarre et al., 2016).
To test this prediction, we sought to experimentally manipulate R. palustris NH4+ excretion. In our previous study, cocultures were grown under a 100% N2 headspace (LaSarre et al., 2016). We reasoned that limiting the N2 supply could lower the amount of NH4+ excreted as R. palustris Nx would potentially retain more NH4+ for itself and excrete less for E. coli. To limit the N2 concentration, we injected known volumes of N2 into coculture tubes with an argon-filled headspace. In agreement with our expectation, supernatants from R. palustris monocultures with 18% N2 contained half as much NH4+ compared to 100% N2 monocultures (Figure 2). Thus, we concluded that N2 limitation was a suitable approach to manipulate NH4+ excretion levels.
To examine the degree of N2 limitation that would support coexistence, we grew cocultures with a range of N2 concentrations and monitored H2 yields and growth rates. We used H2 yield as a proxy for N2 limitation because H2 is an obligate product of nitrogenase, even under 100% N2 (Eq 1). As N2 becomes limiting, nitrogenase produces more H2, eventually producing pure H2 in the absence of N2 (Eq 2) (McKinlay and Harwood, 2010b; Gest and Kamen, 1949).
Thus, progressively more nitrogen limitation should result in progressively more H2 produced. N2-limited cocultures were incubated horizontally with shaking to promote gas mixing. As expected, the coculture H2 yield increased as N2 concentration decreased (Figure 3a). Nitrogen limitation was also evident from the coculture growth rate, which decreased as the N2 concentration decreased (Figure 3b; Supplementary Figure 2a). Notably, cocultures still grew at the lowest concentration of N2 that we tested, 6% N2 in the headspace, indicating that sufficient NH4+ was released to permit mutualistic growth.
Moving forward, we focused on 18% N2 to characterize how lower NH4+ cross-feeding affected coculture dynamics (Figure 4). Model simulations predicted that less NH4 cross-feeding would result in a decrease in the E. coli population (Supplementary Figure 1a), In agreement with this, we observed that E. coli made up 5% of the population in the cocultures with 18% N2, which was significantly lower than the 9% E. coli frequency observed in cocultures with 100% N2. To assess coculture reproducibility, we also performed serial transfers of cocultures with 18% N2. Growth yield, H2 yield, and growth rates were all reproducible across serial transfers (Figure 4). This reproducibility indicated that coexistence was stable despite the lower level of NH4+ sharing.
Coexistence is maintained in heterogeneous environments that decrease NH4+ cross-feeding
Spatial structuring can profoundly impact microbial mutualistic interactions. In some cases, defined spatial structuring is important or even required for coexistence (Summers et al., 2010; Hom and Murray, 2014; Harcombe, 2010; Kim et al., 2008). In other cases, well-mixed environments are sufficient to promote cooperative relationships (Pande et al., 2014; Hillesland and Stahl, 2010; Mee et al., 2014). We hypothesized that the homogeneous environment in our shaking cocultures might dampen the impact of low NH4+ cross-feeding levels. Thus, next we examined if a heterogeneous environment would affect coexistence within nitrogen-limited cocultures.
One way to induce a heterogeneous environment is by incubating in static conditions, wherein cocultures are not agitated. Static incubation was expected to result in a gradient of N2 availability throughout the depth of the coculture, as N2 has to diffuse from the headspace into the liquid. Static incubation also resulted in substantial cell settling within the coculture. Thus, we hypothesized that a heterogeneous environment would develop wherein some cells would experience a higher degree of nitrogen limitation than others, and R. palustris NH4+ excretion levels would thus vary. Confirming this hypothesis, static R. palustris Nx monocultures with 100% N2 showed less NH4+ excretion than in shaken cultures (Figure 2). Furthermore, in static R. palustris Nx monocultures with only 18% N2, NH4+ was only detectable after a longer 4-week incubation time (Figure 2). Thus, under static conditions, consumption of N2 by R. palustris likely exceeds the rate at which dissolved N2 is replenished from the headspace, resulting in N2 limitation and thereby low NH4+ excretion levels.
To determine how heterogeneous environments affected coculture trends under NH4+ cross-feeding limitation, we performed serial transfers of cocultures under static conditions with either 100% or 18% N2 in the headspace, every 2 or 4 weeks, respectively. These longer incubation times were necessary to achieve similar final cell densities between shaking and static environments. Static cocultures with 100% N2 had higher H2 yields than shaken cocultures (Figure 5a). This was expected given that a subset of the static coculture was experiencing N2 limitation (see Equations 1 and 2). Supplying only 18% N2 amplified this trend further (Figure 5a). In agreement with previously modeled effects of decreased NH4+ cross-feeding (Supplementary Figure 1a), R. palustris growth yields remained similar under all conditions whereas E. coli growth yields decreased in static cocultures relative to shaken cocultures with 100% N2 (Figure 5b). The decrease in E. coli growth yield was exacerbated in static cocultures with 18%, wherein NH4+ cross-feeding levels should be even lower (Figure 2). Coexistence was maintained over serial transfers regardless of N2 availability (Figure 5b). Collectively, these data demonstrate the robustness of our coculture to low NH4+ cross-feeding levels in both homogenous and heterogeneous environments.
Growth-independent fermentation is crucial for coexistence at low cross-feeding levels
Limiting NH4+ cross-feeding resulted in lower E. coli growth yields but had no effect on R. palustris growth yields (Figure 5). R. palustris growth yields depend on carbon acquisition from E. coli. Thus, despite lower E. coli cell densities during low NH4+ cross-feeding, similar levels of organic acids were still provided to R. palustris. We hypothesized that this disparity in growth yields could be due to growth-independent fermentation by E. coli. By this hypothesis, NH4+-limited E. coli would grow at a slower rate but would continue to use fermentation for maintenance energy; consequently, R. palustris would receive a slower but continuous supply of organic acids for growth and N2 fixation. Ultimately, E. coli would assimilate less glucose as a larger proportion would be used for maintenance whereas R. palustris would receive a similar or even greater amount of carbon from E. coli. This hypothesis in turn implies that growth-independent fermentation by E. coli is important for sustaining R. palustris metabolism and thereby coculture viability under low cross-feeding levels.
Growth-independent fermentation is essential for E. coli maintenance energy and is intimately tied to central metabolism. Thus, rather than try to genetically eliminate growth-independent fermentation we instead used a modeling approach to gauge its importance during N2 limitation. To simulate the impact of N2 limitation on coculture dynamics we first modified our previous model (LaSarre et al., 2016) to account for the effects of N2 limitation on the shift from NH4+ to H2 production by R. palustris (SyFFoN_v2; Supplementary Table 2). We then adjusted SyFFoN_v2 parameters to simulate growth rate and metabolite yield data observed at various N2 concentrations (Figure 3 and Supplementary Figure 2, lines). In doing so we found that parameters based on E. coli monoculture data (LaSarre et al., 2016) could not accurately simulate coculture growth rates observed at low N2 concentrations. Rather, E. coli growth-independent fermentation levels had to be increased by up to two-orders of magnitude to accurately simulate empirical growth rates (Figure 3b) (Supplementary Table 2). The need for these changes suggests that R. palustris consumption of fermentation products pulls E. coli fermentation by minimizing end-product inhibition, analogous to what has been observed in other fermentative cross-feeding systems (Iannotti et al., 1973; Hillesland and Stahl, 2010). SyFFoN_v2 accurately predicted H2 yields (Figure 3a), normalized growth rates (Figure 3b), and product yields (Supplementary Figure 2) between 15% and 100 % N2. Below 15% N2, R. palustris likely transitions into a physiological starvation response that our model does not predict, such as the diversion of resources to storage products (McKinlay et al., 2014). We also verified that SyFFoN_v2 could reproduce trends from our previous study, (LaSarre et al., 2016), namely the effects of added NH4+ (Supplementary Figure 3) and varying the R. palustris NH4+ excretion levels (Supplementary Figure 1a).
To examine how growth-independent fermentation influenced this mutualism, we used SyFFoN_v2 to simulate the effect of N2 limitation on population dynamics in the presence or absence of growth-independent fermentation (Figure 6). With growth-independent fermentation included, the model predicted that mutualistic growth would be sustained at low N2 concentrations. E. coli final cell densities were predicted to decline as N2 levels fall below ~30% while R. palustris final cell densities would decline as N2 levels fall below ~20% (Figure 6a). Without growth-independent fermentation, simulations predicted a truncated range of N2 concentrations that would support coculture growth (Figure 6b). In fact, the simulations suggested that growth-independent fermentation is necessary at N2 concentrations where we observed reproducible coculture growth trends (Figure 6). The model predicted similar trends when NH4+ excretion levels were varied in place of N2 availability (Supplementary Figure 1).
A closer inspection of simulated cross-feeding levels revealed why coculture growth would not be supported at low N2 concentrations (or low NH4+ cross-feeding levels) without growth-independent fermentation. At high N2 levels (100% N2), growth-coupled fermentation alone is sufficient to support coculture growth, as any increase in populations results in progressively more metabolites exchanged over time (Supplementary Figure 4a). However, near the transitional N2 concentration where coculture growth is predicted to fail in the absence of growth-independent fermentation (28% N2), metabolite excretion levels decrease as populations grow, resulting in continuously less essential resources for subsequent generations despite available glucose; in other words, cross-feeding spirals into a cycle of diminishing returns (Supplementary Figure 4b). Our data indicate that growth-independent fermentation can circumvent diminishing returns. Fermentation products will always be produced, and thus R. palustris will eventually grow to a density that collectively excretes sufficient NH4+ to allow E. coli growth. Indeed, when growth-independent fermentation is included at 28% N2, growth-independent cross-feeding by E. coli stimulates sufficient reciprocal NH4+ excretion to sustain coculture growth (Supplementary Figure 4c). These simulations strongly suggest that growth-independent fermentation permits cooperative growth at low NH4+ excretion levels that would otherwise be insufficient.
Growth-independent metabolism prevents cooperative growth at high E. coli cell densities
On a per cell basis, growth-independent fermentation is considerably slower than fermentation during growth (Russell and Cook, 1995). However, we reasoned that a high E. coli cell density could amplify this low rate such that organic acid production would be substantial at a population level. We previously demonstrated that dose-dependent toxicity governs mutualism dynamics in our coculture; specifically, organic acids play a beneficial role as a carbon source for R. palustris, but a detrimental role when they accumulate enough to acidify the medium (LaSarre et al., 2016). Thus, we hypothesized that if E. coli cell densities were sufficiently high, the collective growth-independent fermentation rate might destabilize the mutualism by producing organic acids faster than the smaller R. palustris population could consume them, resulting in coculture acidification and growth inhibition.
To test this hypothesis, we first simulated coculture growth from different initial species densities. The model correctly predicted that a common equilibrium would be reached from a wide range of initial E. coli densities (Figure 7a) (LaSarre et al., 2016). However, in agreement with our hypothesis the model also predicted a maximum initial E. coli density that would allow cooperative growth (Figure 7a). An upper limit was experimentally verified, albeit at a lower E. coli density than what was predicted (Figure 7a. At an initial E. coli density of ~2 x 109 CFU / ml, the pH reached acidic levels known to prevent R. palustris growth and metabolism (LaSarre et al., 2016) (Figure 7a). As a result, neither species' population increased (Figure 7a). These results contradicted predictions when growth-independent fermentation was omitted from the model, as there was no predicted initial E. coli density that would prevent cooperative growth (Figure 7b). Simulations indicated that it is the initial E. coli cell density rather than the initial species ratio that determines if coculture growth will be prevented through dose-dependent toxicity (Figure 7c). The inhibitory effect of high initial E. coli cell densities could be offset by a high initial R. palustris cell density enabling organic acid consumption at a rate sufficient to hamper accumulation (Figure 7c). However, simulations suggest that an initial R. palustris concentration of 1010 cells / ml would be required to fully offset the acidification from an initial E. coli cell density of 109 cells / ml, mainly because NH4+cross-feeding by R. palustris would stimulate E. coli growth and thereby accelerate fermentation and organic acid accumulation. Thus, while growth-independent fermentation is a stabilizing factor at low NH4+ exchange levels, it can also serve to destabilize the mutualism at high E. coli densities.
Discussion
In this study, we demonstrated that growth-independent cross-feeding can circumstantially impede or promote mutualism. Mutualism destabilization by growth-independent metabolism depends on dose-dependent nutrient toxicity. Specifically, destabilization occurs when a cross-fed nutrient produced in a growth-independent manner accumulates sufficiently to inhibit growth of the partner species. Destabilization of a natural mutualism by growth-independent metabolism would require specific conditions. In our system, organic acid toxicity is relatively low, in part due to the buffered medium. Thus, an extremely high initial E. coli cell density was required before growth-independent fermentation could inhibit cooperative growth via culture acidification. However, inhibitory effects linked to growth-independent metabolism could occur at cell densities relevant to natural systems in a less well-buffered system or if the cross-fed metabolite toxicity was intrinsically high. For example, notoriously toxic compounds like cyanide (Harris and Knowles, 1983) and antibiotics (Barnhill et al., 2010; Dantas et al., 2008) can serve as nutrients for some bacteria as long as concentrations are maintained at low concentrations.
While the likelihood of mutualism destabilization by growth-independent metabolism is difficult to gauge, promotion of cross-feeding relationships by growth-independent metabolism is likely to be widespread. Vast areas of the Earth's biosphere are limited for key nutrients (Lever et al., 2015), and it is well appreciated that nutrient limitation can promote cross-feeding in natural environments (Seth and Taga, 2014; Hom and Murray, 2014). However, it is poorly understood how established mutualisms respond to perturbations that limit cross-feeding itself. It is thought that exchange rates within obligate mutualisms must be sufficient to support sustained growth of both species in order to avoid eventual extinction (Shou et al., 2007). Our results demonstrate that growth-independent cross-feeding can ease this requirement. In our system, growth-independent fermentation by E. coli can preserve the mutualism amid unfavorable NH4+ exchange levels by continually cross-feeding organic acids. This persistent cross-feeding stimulates R. palustris growth and NH4+ excretion, thereby lifting both species out of starvation. In other words, persistent growth-independent cross-feeding facilitates cooperative success over an extended range of excretion levels compared to metabolites whose excretion is dependent on growth. Given that the majority of microbes in natural environments are in a state of dormancy or low metabolic activity (Hoehler and Jørgensen, 2013; Jørgensen and Marshall, 2016; Lever et al., 2015), we postulate that metabolite release is more likely to be growth-independent. As such, growth-independent metabolism could better serve to initiate and maintain mutualisms in natural environments. Separately, although growth-independent fermentation promoted growth of both partners under our study conditions, it is imaginable that mutualistic cross-feeding could purely support maintenance energy requirements in some cases, thereby promoting survival until nutrient availability improves.
Organic acids, and other fermentation products, are important metabolic intermediates in anaerobic food webs (McInerney et al., 2008; Jørgensen and Marshall, 2016). Growth-independent fermentation could therefore play an important role under nutrient-limiting conditions by sustaining mutualistic relationships with acetogens, methanogens, photoheterotrophs, and anaerobically respiring microbes that rely on fermentation products for carbon and electrons. However, contributions of growth-independent metabolism to mutualisms need not be restricted to fermentation nor to natural environments. Generation of maintenance energy is likely a universally essential process. Thus, mutualistic relationships encompassing diverse lifestyles could conceivably be preserved at low metabolic rates, provided that the limiting nutrient(s) still permits the excretion of factors required to sustain partner viability. Additionally, growth-independent cross-feeding could also benefit industrial bioprocesses, which commonly use growth-limiting conditions to boost product yields. Indeed, growth-independent cross-feeding likely sustained our coculture during N2-limiting conditions under which the highest H2 yields were observed (Figures 4 and 5). Applications of microbial consortia for industrial processes is gaining interest (Sabra et al., 2010) but the effects of nutrient limitation have yet to be investigated. Clearly, the role of growth-independent metabolic activities in fostering microbial cooperation deserves closer appraisal in both natural and applied systems.
Supplementary Figures and Tables
Acknowledgements
We thank David Kysela and Amelia Randich for discussions on the model name, SyFFoN. This work was supported in part by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, under Award Number DE-SC0008131, by the U.S. Army Research Office, grant W911NF-14-1-0411, and by the Indiana University College of Arts and Sciences.