Prochlorococcus Exudate Stimulates Heterotrophic Bacterial Competition with Rival Phytoplankton for Available Nitrogen

ABSTRACT The marine cyanobacterium Prochlorococcus numerically dominates the phytoplankton community of the nutrient-limited open ocean, establishing itself as the most abundant photosynthetic organism on Earth. This ecological success has been attributed to lower cell quotas for limiting nutrients, superior resource acquisition, and other advantages associated with cell size reduction and genome streamlining. In this study, we tested the prediction that Prochlorococcus outcompetes its rivals for scarce nutrients and that this advantage leads to its numerical success in nutrient-limited waters. Strains of Prochlorococcus and its sister genus Synechococcus grew well in both mono- and cocultures when nutrients were replete. However, in nitrogen-limited medium, Prochlorococcus outgrew Synechococcus but only when heterotrophic bacteria were also present. In the nitrogen-limited medium, the heterotroph Alteromonas macleodii outcompeted Synechococcus for nitrogen but only if stimulated by the exudate released by Prochlorococcus or if a proxy organic carbon source was provided. Genetic analysis of Alteromonas suggested that it outcompetes Synechococcus for nitrate and/or nitrite, during which cocultured Prochlorococcus grows on ammonia or other available nitrogen species. We propose that Prochlorococcus can stimulate antagonism between heterotrophic bacteria and potential phytoplankton competitors through a metabolic cross-feeding interaction, and this stimulation could contribute to the numerical success of Prochlorococcus in nutrient-limited regions of the ocean.

Prochlorococcus sp. strain MIT9215 reached a higher maximum abundance in monoculture than in coculture with Synechococcus sp. strain WH7803, suggesting that competition in coculture caused a slight but significant reduction in the MIT9215 cell yield ( The addition of the marine heterotrophic bacterium Alteromonas macleodii strain EZ55 dramatically changed the outcome for the Synechococcus-Prochlorococcus cocultures (Fig. 1B). While the Prochlorococcus strain MIT9215 growth rate declined moderately, the addition of EZ55 to the coculture resulted in a nearly total loss of growth for Synechococcus strain WH7803 (P = 0.0018). In this AMP-MN medium, the EZ55 heterotroph grew rapidly to ;10 6 cells mL 21 , regardless of whether cyanobacteria were present (see below), indicating growth on trace contaminating organic carbon in the medium. The presence of the heterotroph in this nitrogen-limited medium thus shifted the phytoplankton community structure to one resembling open-ocean communities, with Prochlorococcus being numerically dominant over its rival Synechococcus.
The dynamics of resource competition were further investigated by challenging the cyanobacterial strains to invade established populations of their competitors when rare. At day 32 of growth in AMP-MN, a small inoculum (;3,000 cells mL 21 ) from Synechococcus strain WH7803 monocultures was added to cultures of Prochlorococcus strain MIT9215 with or without Alteromonas macleodii strain EZ55; reciprocally, MIT9215 monocultures were inoculated into cultures of WH7803 with or without EZ55. WH7803 cells were able to invade MIT9215 monocultures after a few days' lag and reach an almost equal abundance over the next 17 days (Fig. 2A). However, WH7803 failed to grow in MIT9215 cultures when EZ55 was present, dropping below the limit of detection shortly after inoculation (Fig. 2B).
In the reciprocal invasion assay, Prochlorococcus strain MIT9215 rapidly grew when inoculated into the Synechococcus strain WH7803 monoculture, with both organisms coexisting at equal abundances (Fig. 2C). In the presence of Alteromonas macleodii strain EZ55, MIT9215 was still able to invade a culture of WH7803 (Fig. 2D). Interestingly, with EZ55 present, the MIT9215 population displaced WH7803 as the majority phytoplankter in the culture: WH7803 exhibited a dramatic decline in abundance (Fig. 2D) that was not observed when EZ55 was absent (Fig. 2C). Thus, independent of the starting ratios or cell concentrations, the presence of the EZ55 heterotroph favored the growth of Prochlorococcus over Synechococcus when cultured in nitrogen-limited media.
Prochlorococcus exudate drives heterotroph N competition with Synechococcus. Critically, the inhibitory effect of Alteromonas macleodii strain EZ55 on Synechococcus strain WH7803 growth was absent if the Prochlorococcus MIT9215 strain was not included. WH7803 showed no significant difference in growth between mono-and cocultures with EZ55 in AMP-MN during exponential growth (Fig. 3A) (P = 0.91). This outcome suggested that Prochlorococcus may be secreting a factor(s) that stimulates the competition of EZ55 for a resource(s) shared by WH7803. To test this, EZ55 and WH7803 were placed in coculture competition in medium preconditioned by MIT9215. Whether MIT9215 cells were removed (via filtration) prior to competition (Fig. 4A) or remained in the medium ( Fig. 4B and Fig. S2), the outcome was the same, and the WH7803 maximal abundance was reduced by an order of magnitude when cocultured with EZ55 compared to its inoculation alone in MIT9215-conditioned medium. As shown in Fig. 3A, this growth differential was not observed in the same growth medium when MIT9215 was absent and did not precondition the medium. and Synechococcus strain WH7803 (C and D) in AMP-MN artificial seawater medium with and without Alteromonas macleodii strain EZ55 was determined. On day 32, cultures of the cyanobacteria without Alteromonas were inoculated as a minority into the cultures of the rival cyanobacterium with and without Alteromonas to assess the ability to invade. Error bars represent 1 standard deviation of the geometric mean (n = 3). We next considered two hypotheses for the Prochlorococcus-driven loss of Synechococcus strain WH7803 growth in the presence of Alteromonas macleodii strain EZ55: Prochlorococcus is driving EZ55 to either compete for limited resources or produce a factor that is toxic to WH7803. Carbon and nitrogen amendment studies favored the former over the latter hypothesis.
Prochlorococcus releases a large fraction of fixed carbon as dissolved organic carbon during nitrogen-limited growth (45), so we reasoned that this excess source of carbon and energy could be energizing Alteromonas macleodii strain EZ55 to compete with Synechococcus for nitrogen in this nitrogen-limited medium. Pyruvate was examined as a proxy for the Prochlorococcus exudate and, like the exudate, allowed EZ55 to prevent the growth of Synechococcus strain WH7803 (Fig. 3A). Notably, in tripartite cultures, the In AMP-MN medium, which is identical to artificial medium for Prochlorococcus autoclaved (AMP-A) except for the omission of nitrogen addition (see Materials and Methods), nitrogen is the limiting resource for both Prochlorococcus and Synechococcus ( Fig. S1A and B); other nutrients were provided in excess. As such, we reasoned that if Alteromonas macleodii strain EZ55 was restricting the growth of Synechococcus strain WH7803, it was likely via competition for nitrogen. Consistently, the addition of excess nitrogen to the medium as either NH 4 1 or NO 3 2 restored the ability of WH7803 to grow in the presence of pyruvate or exudate-stimulated EZ55, whether at the onset of cocultivation ( Fig. 3C and Fig. 4A and B) or after WH7803 had ceased growth for several days (Fig. 3E). Notably, in these coculture studies, pyruvate additions enabled EZ55 to grow to levels several orders of magnitude higher when nitrogen was in excess (Fig. 3D) but not when nitrogen was limiting (Fig. 3B), suggesting that inhibition by EZ55 requires excess carbon relative to nitrogen.
Nitrogen competition in three-member cocultures. While the concentration of total bioavailable N in AMP-MN has been established ( Fig. S1), the constituent N species are not known. We hypothesized that while the Prochlorococcus strain consumes NH 4 1 , the Synechococcus and heterotroph strains compete for a residual N resource that Prochlorococcus cannot utilize but that the other two can, namely, NO 3 2 or NO 2 2 (46). To test this hypothesis, we generated a transposon insertion mutant of Alteromonas macleodii strain EZ55 with a loss-of-function mutation in the nirB gene (nitrite reductase large subunit). The nirB mutant cannot utilize nitrate or nitrite as a nitrogen source and, unlike the wild type (WT) (Fig. 5A), cannot prevent the growth of Synechococcus strain WH7803 in tripartite cultures with Prochlorococcus strain MIT9215 (Fig. 5B). The nirB mutation did not impact the growth of the Alteromonas strain ( Fig. 5C and D), suggesting that this was present in AMP-MN and that wild-type EZ55 is able to outcompete WH7803 for this resource (when activated by the Prochlorococcus exudate).
Competition outcomes are robust with regard to genotype. To determine the extent to which strain genotype impacts the outcomes of cocultivation, we modified the mixed-culture experiments by replacing Prochlorococcus strain MIT9215, Synechococcus strain WH7803, or Alteromonas macleodii strain EZ55 with different strains of Prochlorococcus, Synechococcus, or heterotrophic bacteria, respectively. Like MIT9215, high-lightadapted Prochlorococcus sp. strain MIT9312 or MED4 outcompeted WH7803 in the presence of EZ55 (Fig. 6A), and like WH7803, Synechococcus sp. strains CC9605 and WH8102 were outcompeted by MIT9215 in the presence of EZ55 (Fig. 6B).
As a final constraint on the Synechococcus-heterotroph coculture outcomes, different marine heterotrophic bacteria were substituted for Alteromonas macleodii strain EZ55: Phaeobacter sp. strain Y3F and Vibrio fischeri strain ES114. When grown in Nreplete AMP-A with or without pyruvate or N-limited AMP-MN without pyruvate, coculturing with any of the three heterotrophs did not cause any significant deviation of the Synechococcus strain WH7803 maximal abundance compared to the monoculture control ( Fig. S4A to C). However, as with EZ55, the addition of pyruvate to AMP-MN caused a reduction in the WH7803 maximal abundance when in coculture with YF3 or ES114 compared to either the monoculture control (

DISCUSSION
In this study, we describe conditions under which the dominance of Prochlorococcus over rival phytoplankton is reproduced in culture. Importantly, we observed that Prochlorococcus outgrows Synechococcus under low-nitrogen conditions, simulating the North Pacific Subtropical Gyre, and only in the presence of heterotrophic bacteria, simulating the multitrophic mixed community of the ocean.
In the NPSG, where nitrogen is thought to limit growth (3,4,13,47), Prochlorococcus can outnumber Synechococcus (and other members of the phytoplankton community) by several orders of magnitude (6,8,44). In these nitrogen-limited waters, heterotrophic bacteria can grow to between 300,000 and 500,000 cells mL 21 and outnumber phytoplankton (48)(49)(50). Our low-nitrogen culture medium recapitulated these trends: heterotrophs grew to an only slightly higher abundance of 10 6 cells mL 21 , and in tripartite cultures, the dynamics of the picocyanobacteria favored Prochlorococcus over Synechococcus, regardless of the relative starting abundances.
Our results suggest that Prochlorococcus acts indirectly, through a heterotroph intermediate, to dictate the growth outcome of its rival Synechococcus in low-nitrogen environments. In low-nitrogen, low-organic-carbon medium, Prochlorococcus scavenges a residual source(s) of nitrogen, apparently with a superior capability relative to Alteromonas and Synechococcus. Alteromonas can grow on residual organic carbon until it becomes growth arrested by a lack of carbon and energy. In this state, it is poised to compete for nitrogen but lacks the carbon and energy resources to do so unless fed by Prochlorococcus. Once fed, Alteromonas can begin to compete with Synechococcus for an alternative nitrogen source(s). The inability of a mutant Alteromonas strain lacking the capacity for NO 3 2 /NO 2 2 utilization to arrest the growth of Synechococcus suggests that the competition involves one or both of these nitrogen species, resources that both Synechococcus and wild-type Alteromonas can utilize but that the strains of Prochlorococcus examined in this study cannot. Nitrate-utilizing strains of Prochlorococcus were recently isolated (51), and future studies in tripartite cultures with these strains could prove informative. In the paragraphs that follow, we unpack this model to discuss the key supporting evidence and identify unanswered questions.
Our study implicates the release of organic carbon by Prochlorococcus for the stimulation of Alteromonas to outcompete Synechococcus for nitrogen. Neither Prochlorococcus nor Alteromonas acting alone was sufficient to diminish the growth of Synechococcus, but when together in a tripartite community, they diminished Synechococcus growth.
Importantly, this effect was observed only when nitrogen was limiting in the medium; the addition of excess nitrogen was all that was needed to restore Synechococcus growth. The latter result also argues against the production of a growth-limiting substance by Alteromonas as the explanation for the growth arrest of Synechococcus.
The Prochlorococcus exudate was sufficient to stimulate N competition by Alteromonas, as was a proxy form of the Prochlorococcus exudate, pyruvate. Prochlorococcus exudes a large fraction of fixed carbon as dissolved organic matter (52)(53)(54), much of which is bioavailable to heterotrophic bacteria (55,56). Recently, it was observed that Prochlorococcus can also release membrane vesicles (57), which may serve as complex nutrients for cooccurring heterotrophs. Critically, under nitrogen limitation, the release of dissolved organic matter by Prochlorococcus is exacerbated (45,58). The specific form(s) of released organic carbon that stimulated Alteromonas competition for nitrogen in this study is not known, but it is rather curious that the Synechococcus exudate was not sufficient for this effect: bipartite cultures of Alteromonas and Synechococcus stably coexisted in low-N medium. Synechococcus is known to release organic carbon, and this release increases under nutrient limitation (59), so this distinction between Prochlorococcus and Synechococcus exudates warrants further investigation.
As with carbon, the nitrogen species involved in the tripartite interactions are not yet completely identified and could include both inorganic and organic sources for growth. Our artificial seawater medium lacked nitrogen amendment, but trace amounts of nitrogen from unknown sources could support microbial growth to 10 6 cells mL 21 . Due to the volatility of ammonia and reported cases of ammonia contamination in other systems (60), we suspect that it serves as a major component of the unamended medium. As the preferred nitrogen source for Prochlorococcus and most microbes, we suspect that ammonia is the primary nitrogen source consumed by Prochlorococcus, whether in mono-or mixed cultures. However, strain MIT9215 has the genetic potential to utilize urea as well (37,46), so this species cannot be ruled out. Nitrate and/or nitrite is likely a component of the medium, as Synechococcus strain WH7803 can utilize nitrate or nitrite as a sole nitrogen source (46), and Alteromonas became unable to prevent Synechococcus growth when the nitrite/nitrate utilization pathway of the heterotroph was knocked out. While some strains of Prochlorococcus can utilize nitrite and nitrate (51), the ones assayed in this study could not. Whether or not the nitrate/nitrite-utilizing Prochlorococcus strains can also compete with Synechococcus for this resource could be resolved in future studies.
In the ocean, Prochlorococcus and Synechococcus compete for a variety of nitrogen sources, including organic forms such as amino acids (29,(61)(62)(63)(64)(65). In a 2019 study, Berthelot et al. observed that cooccurring populations of Prochlorococcus, Synechococcus, and photosynthetic picoeukaryotes in the N-limited North Pacific Subtropical Gyre all utilize ammonia, urea, and nitrate although to different extents (62).
While capable of sourcing their nitrogen from organic carbon molecules like amino acids, marine heterotrophs have been shown to also compete with phytoplankton for inorganic nitrogen in the form of ammonia or nitrate (66)(67)(68)(69). Heterotrophs can account for 30% or more of inorganic nitrogen uptake at some locations (70,71), and in some studies, inorganic nitrogen accounted for half or more of the total nitrogen acquired by heterotrophs (72,73).
Importantly, the ability of heterotrophs to compete for inorganic nitrogen appears to be stimulated by organic carbon. Several studies by the Kirchman group and others noted the necessity for sufficient carbon for inorganic N uptake by bacteria (67,68,(73)(74)(75)(76). These results reflect the importance of C/N balance for heterotrophic growth, which has been recognized in studies of Escherichia coli and other heterotrophs. For Escherichia coli, carbon limitation depletes the tricarboxylic acid (TCA) cycle intermediate and key substrate for inorganic nitrogen assimilation, a-ketoglutarate (2-oxoglutarate) (77). Consequently, C-starved cells have diminished rates of ammonium assimilation and potentially other N utilization pathways (77). Notably, a recent study found that Alteromonas significantly reduced the expression of genes involved in nitrogen metabolic pathways under carbon and iron colimitation (78).
The stimulation of inorganic nitrogen uptake in these studies is entirely consistent with our observations of Alteromonas and other marine heterotrophs in N-limited medium. Like E. coli, carbon-limited Alteromonas may be deprived of the necessary a-ketoglutarate for the assimilation of ammonia or nitrate. Alternatively, or in addition, carbon limitation may deprive the cells of the energy needed to drive the transport of these substrates. In either case, the provision of organic carbon by Prochlorococcus appears to satisfy the requirements for enhanced inorganic nitrogen uptake and assimilation by these heterotrophs, outcompeting Synechococcus in the process.
Previous studies have highlighted the beneficial effects of heterotroph interactions with picocyanobacteria (40-42, 59, 79-82). Previously, we described how heterotrophic bacteria protect Prochlorococcus from oxidative stress (12,38). Coe et al. (83) and Roth-Rosenberg et al. (84) have shown that heterotrophs promote the survival of Prochlorococcus during long-term light and nutrient (N or P) deprivation, respectively. Christie-Oleza et al. (59) found a similar relationship between Synechococcus and a marine roseobacter. In that study, long-term coexistence under nutrient limitation was facilitated by an exchange of resources between the phototroph and heterotroph.
Interactions between picocyanobacteria have been less well characterized, but a recent study by Knight and Morris (85) showed that Synechococcus could aid the growth of Prochlorococcus under conditions simulating ocean acidification. The mechanism of this help was not identified, but because these cocultures were grown in the presence of Alteromonas sp. EZ55, the authors speculated that Synechococcus could help Prochlorococcus indirectly by stimulating EZ55. The potential for allelopathic interactions between picocyanobacteria has also been noted (86)(87)(88).
Our study provides a new dimension to picocyanobacterium-heterotroph and picocyanobacterium-picocyanobacterium interactions: the ability of one phototroph (Prochlorococcus) to drive a shift from coexistence to competition between a second phototroph (Synechococcus) and a heterotroph. Christie-Oleza et al. (59) found that Synechococcus and heterotroph strains coexist during prolonged coculture in unamended seawater and that upon N addition, cross-feeding could occur by the conversion of N substrates unusable by the other microbe: the heterotroph strain could convert organic nitrogen (peptone) to ammonia, while WH7803 could convert nitrate to dissolved organic nitrogen. In our study, both the heterotroph and phototroph could utilize nitrate and nitrite, and unless the former was mutated in its ability to utilize these resources, the heterotroph could apparently outcompete the Synechococcus strain for this resource when fed organic carbon by Prochlorococcus.
While usually found at abundances of 10 4 cells mL 21 or lower in the open ocean (89)(90)(91), Alteromonas was chosen as a proxy for the heterotrophic community because of previously described interactions with Prochlorococcus. The tripartite interaction that influenced the success of Prochlorococcus over Synechococcus is likely due to the nutrient utilization capabilities of the heterotrophic bacteria rather than an adaptation to nutrient-limited growth. However, to explore this interaction further, a future direction of this work will be to observe tripartite outcomes upon the inclusion of dominant oligotrophic heterotrophs, such as SAR11 Pelagibacter, to determine if these metabolic interactions occur between numerically dominant members of each trophic level (92,93).
Conclusion. This study demonstrates that metabolic interactions between trophic groups can influence relative abundances within trophic groups. The prediction that Prochlorococcus outcompetes rival phytoplankton, including Synechococcus, under nutrient limitation is largely confirmed, but this outcome may require the ability of Prochlorococcus to energize heterotrophic bacteria to outcompete their photosynthetic rivals for resources that they themselves do not use. If our results can be extrapolated to the natural environment, they highlight an important connection between carbon and nitrogen availability and suggest that complex microbial interactions can benefit streamlined, efficient genera such as Prochlorococcus to the detriment of their competition.

MATERIALS AND METHODS
Strains and culturing. Axenic cultures of Prochlorococcus strains MIT9215, MIT9312, and MED4 and Synechococcus strains WH7803, CC9605, and WH8102 were used in this study. Stock cultures of cyanobacteria were initially maintained in an artificial seawater medium, AMP-A (12,94,95), and were inoculated and serially maintained (for up to 2 years) in AMP-MN (this study) (described below) to prevent the introduction of excess nitrogen (N). The axenicity of cyanobacterial stocks and experimental cultures was tested routinely by diluting a small volume of the culture into 1/10Â Prochlorococcus AC (ProAC; Difco) and yeast tryptone sea salts (YTSS) media and incubating these cultures in the dark at room temperature for up to 6 weeks to monitor any increase in turbidity indicating the presence of heterotrophic bacteria (35). All experiments were carried out at 24°C in I36VLX incubators (Percival, Boone, IA) with modified controllers that allowed gradual increases and decreases of cool white light to simulate sunrise and sunset, with a peak midday light intensity of 150 mmol quanta m 22 s 21 on a 14-h/10-h light/dark cycle (96). Ammonium (NH 4 1 ) was the N amendment in all experiments, unless otherwise stated, as it can be used by all strains in this study. Experiments that included different NH 4 1 concentrations were performed with NH 4 1 amendments to the AMP-A derivative AMP-MN (minus nitrogen), which is identical to AMP-A except that no N source is included. Stepwise amendments of NH 4 1 to AMP-MN and subsequent regression analysis of maximal Prochlorococcus abundances indicated that the residual N bioavailable to Prochlorococcus and Synechococcus was approximately 0.164 mM (see Fig. S1 in the supplemental material) (R 2 = 0.9729).
While all culture media were sterilized by autoclaving, sterilized spent or Prochlorococcus-conditioned medium was generated by culturing Prochlorococcus strain MIT9215 in large volumes of AMP-MN (;300 mL). At stationary phase (25 to 30 days), these cells were removed by gentle filtration (27 inHg) in a 1-L filter tower (Nalgene) using 0.2-mm-pore-size GTTP isopore membrane filters (MilliporeSigma, Burlington, MA). Previous studies indicated that low-pressure filtration does not cause detectable rupture of Prochlorococcus cells during filtration (12). The sterility of this conditioned medium was determined by flow cytometry alongside the experiments in which it was utilized, in addition to the purity assay detailed above.
Quantification of cyanobacterium and heterotroph abundances. The abundances of cyanobacteria were quantified by flow cytometry using a Guava EasyCyte 8HT flow cytometer (Millipore, Burlington, MA) with populations of Prochlorococcus and Synechococcus differentiated in cocultures by their red and red/yellow fluorescence, respectively (35,100). Heterotrophs in mono-and coculture experiments were quantified by viable counting with serial dilutions on YTSS-1.5% agar plates incubated at 24°C.
Transposon mutagenesis. Mutants of Alteromonas macleodii strain EZ55 incapable of growing on nitrate (NO 3 2 ) as a sole N source were generated by transposon mutagenesis using a mini-Himar1 Mariner transposon carrying a kanamycin resistance-selectable marker (101). The RB1 plasmid vector containing the transposon was propagated in Escherichia coli strain WM3064, a pir 1 and 2,6-diaminopimelic acid (DAP) auxotroph donor strain (102). Cultures of the donor strain grown overnight were inoculated from cryopreserved stocks (280°C in LB plus 10% glycerol) into 5 mL of LB amended with 10 mg/ mL of kanamycin and 150 mL of 100 mM DAP (Alfa Aesar, Haverhill, MA) and incubated with shaking at 37°C. Conjugations with EZ55 were performed by plating both the donor and recipient onto YTSS agar plates for 8 h. Exconjugants were selected on plates containing YTSS plus 10 mg/mL kanamycin. Selected colonies were screened for NO 3 2 utilization by replica plating (103) on AMP-A agar with 1.5% Noble agar (Difco) amended with 500 mM sodium pyruvate (Sigma-Aldrich) and either 400 mM NH 4 1 or 882 mM NO 3 2 as the nitrogen source. Replica-plated colonies growing solely on plates containing NH 4 1 were transferred again into tubes of AMP-A with excess carbon and different nitrogen sources to confirm that the mutants were unable to grow on nitrate or nitrite. The insertion location of the Mariner transposon within the nirB gene was verified by arbitrary PCR (104), Sanger sequencing, and BLAST comparisons with the EZ55 genome (IMG accession number 2785510739).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.