Abstract
Many mutualistic microbial relationships are based on nutrient cross-feeding. Traditionally, cross-feeding is viewed as being unidirectional from the producer to the recipient. This is likely true when a producer’s metabolic waste, such as fermentation products, provides carbon for a recipient. However, in some cases the cross-fed nutrient holds value for both the producer and the recipient. In such cases, there is potential for nutrient reacquisition by producer cells in a population, leading to competition against recipients. Here we investigate the consequences of inter-partner competition for cross-fed nutrients on mutualism dynamics using an anaerobic coculture pairing fermentative Escherichia coli and phototrophic Rhodopseudomonas palustris. In this coculture, E. coli excretes waste organic acids that provide carbon for R. palustris. In return, R. palustris cross-feeds E. coli ammonium (NH4+), a valuable nitrogen compound that both species prefer. To explore the potential for inter-partner competition, we first used a kinetic model to simulate cocultures with varied affinities for NH4+ in each species. The model predicted that inter-partner competition for cross-fed NH4+ could profoundly impact population dynamics. We then experimentally tested the predictions by culturing mutants lacking NH4+ transporters in both NH4+ competition assays and cooperative cocultures. Both theoretical and experimental results indicated that the recipient must have a competitive advantage in acquiring valuable cross-fed NH4+ to avoid collapse of the mutualism. Thus, the very metabolites that form the basis for cooperative cross-feeding can also be subject to competition between mutualistic partners.
Significance Mutualistic relationships, particularly those based on nutrient cross-feeding, promote stability of diverse ecosystems and drive global biogeochemical cycles. Cross-fed nutrients within these systems can be either waste products valued only by one partner or nutrients that both partners value. Here, we explore how inter-partner competition for a communally-valuable cross-fed nutrient impacts mutualism dynamics. We discovered that mutualism stability necessitates that the recipient have a competitive advantage against the producer in obtaining the cross-fed nutrient. We propose that the requirement for recipient-biased competition is a general rule for mutualistic coexistence based on the transfer of communally valuable resources, microbial or otherwise.
Introduction
Mutualistic cross-feeding of resources between microbes can have broad impacts ranging from influencing host health (1, 2) to driving global biogeochemical cycles (3–6). Cross-fed metabolites are often regarded as nutrients due to the value they provide to a dependent partner, the recipient. However, for the partner producing the nutrient, the producer, a cross-fed nutrient’s value can vary. On one extreme, the cross-fed metabolite is valued by the recipient but not the producer, as is the case for fermentative waste products (7–10). In other cases, a cross-fed metabolite holds value for both the recipient and the producer, as is the case for vitamin B12 (6, 11, 12) and ammonium (NH4+) (13, 14). Such communally-valuable cross-fed nutrients are subject to partial privatization (15), wherein the producer has mechanisms to retain a portion of the nutrient pool for itself. While most mutualism cross-feeding studies only consider unidirectional metabolite transfer from producer to recipient, we wondered whether these mechanisms for partial privatization could lead to competition between partner populations for communally-valuable cross-fed nutrients. It seems likely that such competition could influence mutualism stability, as is known to be the case for competition for exogenous limiting resources (8, 16–19). To the best of our knowledge inter-partner competition for cross-fed nutrients and its impact on mutualism dynamics have never been investigated.
One example of cross-feeding that could involve competition between mutualistic partners is NH4+ excretion by N2-fixing bacteria (Fig. 1A), called N2-fixers (13, 14). During N2 fixation, the enzyme nitrogenase converts N2 gas into two NH3 (20). In an aqueous environment, NH3 is in equilibrium with NH4+. At neutral pH, NH4+ is the predominant form but small amounts of NH3 can potentially leave the cell by diffusion across the membrane (21) (Fig. 1B). This inherent ‘leakiness’ for NH3 likely fosters NH4+ cross-feeding, as extracellular NH3 is available to neighboring microbes. Importantly, these neighbors can include clonal N2-fixers, as NH3/NH4+ is a preferred nitrogen source for most microbes. At concentrations above 20 μM, NH3 can be acquired by passive diffusion; below 20 μM, NH4+ is bound and transported as NH3 by AmtB transporters (Fig. 1B) (22). AmtB-like transporters are conserved throughout all domains of life (23). There is growing evidence that AmtB is used by N2-fixers to recapture NH3 lost by passive diffusion, as ΔAmtB mutants accumulate NH4+ in culture supernatants whereas wild-type strains do not (24–26). Thus, during NH4+ cross-feeding, AmtB likely facilitates both NH4+ acquisition by the mutualistic partner and recapture of NH4+ by the N2-fixer.
Assessing the effects of inter-partner competition for a cross-fed nutrient would require a level of experimental control not possible in most natural settings. However, synthetic microbial communities, or cocultures, are well-suited to address such questions (27–29). We previously developed a bacterial coculture that features cross-feeding of waste products (organic acids) from Escherichia coli, and a communally-valuable nutrient (NH4+) from Rhodopseudomonas palustris Nx (Fig. 1A) (26). Here, using both a kinetic model and genetic manipulation to alter the affinity of each species in the coculture for NH4+, we demonstrate that inter-partner competition for cross-fed NH4+ plays a direct role in maintaining coexistence. Specifically, insufficient competition by E. coli for NH4+ resulted in a collapse of the mutualism. Mutualism collapse could be delayed or potentially avoided through higher net NH4+ excretion by R. palustris or increased E. coli population size. Our results suggest that, as a general rule, competition for a cross-fed nutrient in an obligate mutualism must be biased in favor of the recipient to avoid mutualism collapse and the potential extinction of both species.
Results
Competition for cross-fed NH4+ is predicted to shape mutualism population dynamics
Within our coculture (Fig. 1A), E. coli (Ec) ferments sugars into waste organic acids, providing essential carbon and electrons to R. palustris (Rp) Nx. R. palustris Nx is genetically engineered to excrete low micromolar amounts of NH4+, providing essential nitrogen for E. coli (26). NH4+ excretion by R. palustris Nx is due to mutation of NifA, the master transcriptional regulator of nitrogenase, which results in constitutive nitrogenase activity even in the presence of normally inhibitory NH4+ (30). In contrast to organic acids, which are only useful to R. palustris, NH4+ produced by R. palustris Nx is essential for the growth of both species; R. palustris uses some NH4+ for its own biosynthesis and excretes the rest, which serves as the nitrogen source for E. coli. However, R. palustris Nx can also take up NH4+ (30). Thus, we hypothesized that competition for cross-fed NH4+ between the R. palustris Nx producer population and the E. coli recipient population could influence mutualism dynamics.
We first explored whether competition for cross-fed NH4+ could affect the mutualism using SyFFoN, a mathematical model describing our coculture (26, 31). SyFFoN simulates population and metabolic dynamics in batch cocultures using Monod equations with experimentally-determined parameter values. As previous versions described NH4+ uptake kinetics only for E. coli (26, 31), we amended SyFFoN to include both an R. palustris NH4+ uptake affinity (Km) and higher R. palustris maximum growth rate (μMAX) when NH4+ is used (SI Appendix Table S1). We then simulated batch cocultures wherein the relative affinity for NH4+ varied between the two species (Fig. 2). The model predicted that coexistence is maintained when the R. palustris affinity for NH4+ is low relative to that of E. coli (Rp:Ec < 1); sufficient N2 is converted to NH4+ to support R. palustris growth and enough NH4+ is cross-fed to support E. coli growth. In contrast, when the R. palustris affinity for NH4+ is high relative to that of E. coli (Rp:Ec > 1), E. coli growth is no longer supported because E. coli cannot compete for excreted NH4+. However, high R. palustris cell densities were still predicted (Fig. 2) due to persistent, low-level organic acid cross-feeding stemming from E. coli maintenance metabolism, which can support R. palustris growth even when E. coli is not growing (31).
Genetic disruption of AmtB NH4+ transporters affects relative affinities for NH4+
Bacterial cells generally acquire NH4+ through two mechanisms: passive diffusion of NH3, or active uptake by AmtB transporters (Fig. 1B). We hypothesized that deleting amtB genes in either species would result in a lower affinity for NH4+ in that species and thus could be used to test how relative NH4+ affinity impacts coculture dynamics. We generated ΔAmtB mutants of both E. coli and R. palustris and first characterized the effect of the mutations in monoculture. Deletion of amtB in E. coli had no effect on growth or fermentation profiles when NH4Cl was in excess (SI Appendix Fig. S1), consistent with previous observations where ΔAmtB growth defects were only apparent at NH4+ concentrations below 20 μM (22). In R. palustris ΔAmtB monocultures with N2 as the nitrogen source, growth trends were equivalent to those of the parent strain; however, R. palustris ΔAmtB excreted more NH4+ than the parent strain and about a third of that excreted by R. palustris Nx (SI Appendix Fig. S1C and D). In line with our hypothesis, NH4+ excretion by R. palustris ΔAmtB could be due to a decreased ability to reacquire NH4+ lost by diffusion, resulting in increased net NH4+ excretion. Alternatively, we considered that NH4+ excretion by R. palustris ΔAmtB could be due to improper nitrogenase regulation. In several other N2-fixers, proper nitrogenase regulation requires AmtB, for example to induce post-translational nitrogenase inhibition (switch-off) in response to NH4+ (25, 32). We tested whether R. palustris ΔAmtB exhibits NH4+-induced switch-off by adding NH4Cl to exponentially growing cultures and measuring H2 production, an obligate product of the nitrogenase reaction (33), as a proxy for nitrogenase activity. Upon adding NH4Cl, H2 production stopped in R. palustris ΔAmtB cultures. In contrast, H2 production only slowed slightly in R. palustris Nx cultures (SI Appendix Fig. S2), consistent with previous observations for NifA* strains (34, 35). Additionally, like the parent strain, R. palustris ΔAmtB did not produce H2 when grown with NH4+, unlike R. palustris Nx (SI Appendix Fig. S3). These observations demonstrate that R. palustris ΔAmtB is competent for NH4+-induced nitrogenase repression, and thus NH4+ excretion by R. palustris ΔAmtB is likely due to a poor ability to reacquire NH4+ lost by diffusion.
To test our hypothesis that deleting amtB would lower cellular affinity for NH4+, we directly competed all possible E. coli and R. palustris strain combinations in competition assays where ample carbon was available for each species but the NH4+ concentration was kept low; specifically, a small amount of NH4+ was added every hour to bring the final NH4+ concentration to 5 μM (Fig. 3). In this competition assay, the species that is more competitive for NH4+ should reach a higher cell density than the other species. In all cases, WT E. coli was more competitive for NH4+ than R. palustris. However, each R. palustris strain was able to outcompete E. coli ΔAmtB (Fig. 3), even though the R. palustris maximum growth rate is 4.6-times slower than that of E. coli (SI Appendix Fig. S1). Even R. palustris strains lacking AmtB outcompeted E. coli ΔAmtB (Fig. 3), indicating that R. palustris has a higher affinity for NH4+ than E. coli independent of AmtB. These data confirmed that deletion of amtB was an effective means by which to lower the relative affinity for NH4+ in each mutualistic partner.
Altering relative NH4+ affinities affects mutualistic partner frequencies
We then examined how relative affinities for NH4+ influenced mutualism dynamics by comparing the growth trends of cocultures containing either WT E. coli or E. coli ΔAmtB, paired with either R. palustris ΔAmtB, R. palustris Nx, or R. palustris NxΔAmtB, the latter of which we previously determined to exhibit 3-fold higher NH4+- excretion levels than the Nx strain in monoculture (26). For each R. palustris partner, cocultures with E. coli ΔAmtB grew slower than cocultures with WT E. coli (Fig. 4A,B). E. coli ΔAmtB also constituted a lower percentage of the population and achieved lower cell densities compared to WT E. coli when paired with the same R. palustris strain (Fig. 4C). These lower frequencies were consistent with the competitive disadvantage of E. coli ΔAmtB for excreted NH4+ (Fig. 3).
For R. palustris strains lacking AmtB, the effects on population trends varied. Consistent with our previous work, R. palustris NxΔAmtB supported higher WT E. coli percentages and cell densities (Fig. 4C) (26). With high NH4+ excretion levels from R. palustris NxΔAmtB, faster E. coli growth leads to rapid organic acid accumulation, which acidifies the environment, inhibits R. palustris growth, and leaves organic acids unconsumed (Fig. 4D) (26). Surprisingly, although R. palustris ΔAmtB excreted less NH4+ than R. palustris Nx in monoculture, R. palustris ΔAmtB supported a higher WT E. coli population in coculture and consumable organic acids accumulated (Fig. 4C, D). These trends resemble those from cocultures with R. palustris NxΔAmtB (Fig. 4C, D), which has a high level of NH4+ excretion (SI Appendix Fig. S1D). Unlike Nx strains, which have constitutive nitrogenase activity due to a mutation in the transcriptional activator NifA (30), R. palustris ΔAmtB has WT NifA. Thus, R. palustris ΔAmtB can likely still regulate nitrogenase expression, and thereby its activity, in response to nitrogen starvation. We hypothesized that in coculture with WT E. coli, R. palustris ΔAmtB might experience heightened nitrogen starvation, as NH4+ consumption by WT E. coli would limit NH4+ reacquisition by R. palustris ΔAmtB (in an R. palustris ΔAmtB monoculture any lost NH4+ would remain available to R. palustris). We therefore tested whether coculture conditions stimulated higher nitrogenase activity using an acetylene reduction assay. In agreement with our hypothesis, R. palustris ΔAmtB had increased nitrogenase activity in coculture conditions compared to monocultures, whereas R. palustris Nx, which exhibits constitutive nitrogenase activity, showed similar levels in both conditions (SI Appendix Fig. S4). Thus, the relatively high WT E. coli population in coculture with R. palustris ΔAmtB is likely due to both the competitive advantage for acquiring NH4+ over R. palustris ΔAmtB (Fig. 3) and higher NH4+ cross-feeding levels due to increased nitrogenase activity.
E. coli must have a competitive advantage for NH4+ acquisition to avoid mutualism collapse
We were surprised to observe that cocultures of R. palustris ΔAmtB paired with E. coli ΔAmtB showed little growth when started from a single colony of each species (Fig. 4A), a method that we routinely use to initiate cocultures (26, 31). We reasoned that the higher R. palustris ΔAmtB affinity for NH4+ relative to E. coli ΔAmtB (Fig. 3) likely led to community collapse as predicted by SyFFoN (Fig. 2). Even though SyFFoN had predicted R. palustris growth when outcompeting E. coli for NH4+ (Fig. 2), SyFFoN likely underestimates the time required to achieve these densities, if they would be achieved at all, as SyFFoN does not take into account cell death, which is known to occur when E. coli growth is prevented (31). Consistent with the hypothesis that poor coculture growth was due to a competitive disadvantage of E. coli ΔAmtB for NH4+, SyFFoN simulations indicated that starting with a larger E. coli ΔAmtB population would increase the probability that any given E. coli cell would acquire NH4+ versus R. palustris and thereby overcome the competitive disadvantage of E. coli ΔAmtB for NH4+ (SI Appendix Fig. S5). Indeed, we observed greater growth of both species when cocultures were inoculated with equal or higher relative densities of E. coli ΔAmtB versus R. palustris ΔAmtB (SI Appendix Fig. S5).
The explanation that mutualism collapse was due to a competitive advantage of R. palustris ΔAmtB over E. coli ΔAmtB for NH4+ called into question why cocultures pairing E. coli ΔAmtB with either R. palustris Nx or R. palustris NxΔAmtB did not collapse as well (Fig. 4), given that in all of these pairings E. coli ΔAmtB is at competitive disadvantage (Fig. 3). We hypothesized that a relatively high NH4+ excretion level by these latter R. palustris strains (SI Appendix Fig. S1D) could compensate for a low E. coli NH4+ affinity. To explore this hypothesis we simulated cocultures with the R. palustris affinity for NH4+ set high relative to that of E. coli (Rp:Ec = 1000) and varied the R. palustris NH4+ excretion level (Fig. 5). Indeed, increasing R. palustris NH4+ excretion was predicted to overcome a low E. coli affinity for NH4+ and support growth of both species (Fig. 5). The only exception was at the highest levels of NH4+ excretion, where R. palustris growth was predicted to be inhibited due to rapid E. coli growth and subsequent accumulation of organic acids that acidify the environment (Fig. 5) (26). These simulations suggested that R. palustris Nx and NxΔAmtB supported coculture growth with E. coli ΔAmtB due to higher NH4+ excretion levels (SI Appendix Fig. S1D), whereas a combination of low NH4+ excretion by R. palustris ΔAmtB (SI Appendix Fig. S1D) and a low affinity for NH4+ by E. coli ΔAmtB led to collapse of the mutualism in this pairing.
So far, we had only considered the effect of severe discrepancies in NH4+ affinities between the two species (e.g., 1000-fold difference in Km values in our simulations) as a mechanism leading to coculture collapse within the time period of a single culturing. However, we wondered if a subtle discrepancy in NH4+ affinities could lead to coculture collapse if given more time. We therefore simulated serial transfers of cocultures with partners having different relative NH4+ affinities (Fig. 6A, B). At equivalent NH4+ affinities (Fig. 6A), both species were predicted to be maintained over serial transfers. However, when the relative affinities approached a threshold (relative Rp:Ec = 2.75), cell densities of both species were predicted to decrease over serial transfers (Fig. 6B). This decline in coculture growth is due to E. coli being slowly but progressively outcompeted for NH4+ by R. palustris. As the difference between the R. palustris and E. coli populations expands, R. palustris cells have a greater chance of acquiring NH4+ than the smaller E. coli population, further starving E. coli and simultaneously cutting off R. palustris from its supply of organic acids from E. coli.
The above prediction prompted us to investigate if cocultures pairing R. palustris Nx with E. coli ΔAmtB were stable through serial transfers. We focused on cocultures with R. palustris Nx rather than R. palustris NxΔAmtB because R. palustris Nx has AmtB and would therefore be most likely to outcompete E. coli ΔAmtB. Strikingly, after eight serial transfers of cocultures pairing R. palustris Nx with E. coli ΔAmtB we observed coculture collapse (Fig. 6C). This observation is in stark contrast to cocultures of R. palustris Nx paired with WT E. coli, which we have serially transferred for over 100 times with no extinction events (unpublished data). These results indicate that the recipient population must have a competitive advantage for a cross-fed nutrient versus the producer population to avoid mutualism collapse.
Discussion
Here we demonstrate that mutualistic partners can compete for a cross-fed nutrient upon which the mutualistic interaction is based, in this case NH4+. This competition can impact partner frequencies and mutualism stability. Efficient nutrient reacquisition by the producer can render nutrient excretion levels insufficient for cooperative growth, starving the recipient and leading to tragedy of the commons (36). Conversely, recipient-biased competition for a cross-fed nutrient drives cooperative directionality in nutrient exchange and thereby promotes mutualism stability. One implication of these results is that inter-partner competition can influence the level of resource privatization. Within microbial interdependencies, partial privatization has primarily been thought to depend on mechanisms used by the producer to retain a portion of a communally-valuable resource (15). Our data indicate that for excreted resources having a transient availability to both mutualists, recipient acquisition mechanisms can also influence the level of producer privatization, as the competition impacts how much of a cross-fed resource will be shared versus re-acquired. In effect, recipient-biased competition avoids tragedy of the commons by enforcing partial privatization of a communally-valuable resource. The importance of the recipient having the upper hand in inter-partner competition likely applies to other synthetic cocultures and natural microbial mutualisms that are based on the cross-feeding of communally-valuable nutrients, including amino acids (37, 38) and vitamin B12 (6, 11). The same rule could also apply to inter-kingdom and non-microbial examples of cross-feeding (e.g., plants and pollinators, nutrient transfer between plants and bacteria or fungi (39)) and cooperative feeding (e.g., honeyguide bird and human harvesting of bee hives (40), cooperative hunting between grouper fish and moray eels (41)). In such cases, increased privatization of a cross-fed or shared resource, for example through producer-biased competition, could threaten the mutualism upon which both species depend (15, 39, 42).
In our system, AmtB transporters were crucial determinants of inter-partner competition for NH4+. We were intrigued to find that when both species lacked AmtB, R. palustris out-competed E. coli for NH4+ (Fig. 5), enough so to collapse the mutualism within a single culturing (Fig. 3). Whether by maximizing NH4+ retention or re-acquisition, R. palustris, and perhaps other N2-fixers, might have additional mechanisms aside from AmtB to minimize loss of NH4+ as NH3. These mechanisms could include a relatively low internal pH to favor NH4+ over NH3, negatively-charged surface features, or relatively high affinities by NH4+-assimilating enzymes such as glutamine synthetase. There are several reasons why it would be beneficial for N2-fixers to minimize NH4+ loss. First, N2 fixation is expensive, both in terms of the enzymes involved (43) and the reaction itself, costing 16 ATP to convert one N2 into two NH3 (33). Passive loss of NH3 would only add to this cost, as more N2 would have be fixed to compensate. Second, loss of NH4+ could benefit nearby microbes competing against an N2-fixer for separate limiting nutrients (14, 44). The possibility that N2-fixers could have a superior ability to retain or acquire NH4+ independently of AmtB is not farfetched. Bacteria are known to exhibit differential abilities to compete for nutrients. For example, iron acquisition commonly involves iron-binding siderophores, but siderophores can be chemically distinct and thereby differ in their affinity for iron (45). Strategies to utilize siderophores as a shared resource are also numerous, leading to different cooperative or competitive outcomes in microbial communities (45, 46). One must consider that additional mechanisms for acquiring NH4+ beyond AmtB might likewise exist. As our results have raised the potential for inter-partner competition for cross-fed resources themselves, understanding the physiological mechanisms that confer competitive advantages for nutrient acquisition between species will undoubtedly aid in describing the interplay between competition and cooperation within mutualisms.
Materials and Methods
Strains and growth conditions
Strains, plasmids, and primers are listed in SI Appendix Table S2. All R. palustris strains contained ΔuppE and ΔhupS mutations to facilitate accurate colony forming unit (CFU) measurements by preventing cell aggregation (47) and to prevent H2 uptake, respectively. E. coli was cultivated on Luria-Burtani (LB) agar and R. palustris on defined mineral (PM) (48) agar with 10 mM succinate. (NH4)2SO4 was omitted from PM agar for determining R. palustris CFUs. Monocultures and cocultures were grown in 10-mL of defined M9-derived coculture medium (MDC) (26) in 27-mL anaerobic test tubes. To make the medium anaerobic, MDC was bubbled with N2, then tubes were 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), unless indicated otherwise. E. coli monocultures were also supplemented with 15mM NH4Cl. All cultures were grown at 30°C laying horizontally under a 60 W incandescent bulb with shaking at 150 rpm. Starter cocultures were inoculated with 200 μL MDC containing a suspension of a single colony of each species. Test cocultures were inoculated using a 1% inoculum from starter cocultures. Serial transfers were also inoculated with a 1% inoculum. Kanamycin and gentamycin were added to a final concentration of 100 μg/ml for R. palustris and 15 μg/ml for E. coli where appropriate.
Generation of R. palustris mutants
R. palustris mutants were derived from wild-type CGA009 (49). Generation of strains CGA4004, CGA4005, and CGA4021 was described previously (26). For generation of strain CGA4026 (R. palustris ΔAmtB) the WT nifA gene was amplified using primers JBM1 and JBM2, digested with XbaI and BamHI, and ligated into plasmid pJQ200SK to make pJQnifA16. This suicide vector was then introduced into CGA4021 by conjugation, and sequential selection and screening was performed as described (50) to replace nifA* with WT nifA. Reintroduction of the WT nifA gene was confirmed by PCR and sequencing.
Generation of the E. coli ΔAmtB mutant
P1 transduction (51) was used to introduce ΔamtB::Km from the Keio collection strain JW0441-1 (52) into MG1655. The ΔamtB::Km genotype of kanamycin-resistant colonies was confirmed by PCR and sequencing.
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 (i.e., Tube OD660). Specific growth rates were determined using readings 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. H2 was quantified using a Shimadzu (Kyoto, Japan) gas chromatograph (GC) with a thermal conductivity detector as described (53). Glucose, organic acids, formate and ethanol were quantified using a Shimadzu high-performance liquid chromatograph (HPLC) as described (54). NH4+ was quantified using an indophenol colorimetric assay as described (26).
Nitrogenase activity
Nitrogenase activity was measured using an acetylene reduction assay (43). Cells from 10-mL cultures were harvested and resuspended in 10-mL fresh MDC medium in 27-mL sealed tubes pre-flushed with argon gas. Suspensions were incubated in light for 1 h at 30°C to recover. Then, 250 μl of 100% acetylene gas was injected into the headspace to initiate the assay, and ethylene production was measured over time by gas chromatography as described (43). Ethylene levels were normalized to total R. palustris CFUs in the 10-ml volume.
NH4+ competition assay
Fed-batch cultures were performed in custom anaerobic 75-ml serum vials with side sampling ports. Each vial contained a stir bar and 30-mL of MDC, and was sealed at both ends with rubber stoppers and aluminum crimps. Each vial was supplemented with 25 mM glucose, 1 % v/v cation solution and 20 mM sodium acetate. Starter monocultures of each species were grown to equivalent CFUs/mL in MDC tubes containing limiting nutrients (3 mM sodium acetate for R. palustris and 1.5 mM NH4Cl for E. coli), and 1 mL of each species was inoculated into the serum vials. These competition cocultures were incubated at 30°C under a 60 W incandescent bulb with stirring at 200 rpm (Thermo Scientific) for 96 h. Each serum vial was constantly flushed with Ar to maintain anaerobic conditions. NH4Cl was fed from a 500 μM NH4Cl stock using a peristaltic pump (Watson-Marlow) on an automatic timer (Intermatic DT620) at a rate of 0.33 mL/min once an hour for a final concentration of ~ 5 μM upon each addition. Samples were taken at 0 and 96 h for quantification of CFUs.
Mathematical modeling
A Monod model describing bi-directional cross-feeding in batch cultures, called SyFFoN_v3 (Syntrophy between Fermenter and Fixer of Nitrogen), was modified from our previous model (31) to allow for competition between E. coli and R. palustris for NH4+ as follows: (i) an equation for R. palustris growth rate on NH4+ was added to boost the R. palustris growth rate when acquiring NH4+ and (ii) the ability for R. palustris to consume NH4+ was added along with a Km of R. palustris for NH4+ (KAR). Equations and default parameter values are in the SI Appendix and Table S1. SyFFoN_v3 runs in R studio and is available for download at: https://github.com/McKinlab/Coculture-Mutualism.
Acknowledgments
We thank Richard Phillips (Indiana University) for providing equipment for the NH4+ competition assay. We also thank Jay Lennon (Indiana University) for helpful discussions on the manuscript. 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.
Footnotes
Conflict of interest. The authors declare no conflict of interest.