Role of a Microcin-C–like biosynthetic gene cluster in allelopathic interactions in marine Synechococcus

Abstract

Competition between phytoplankton species for nutrients and light has been studied for many years, but allelopathic interactions between them have been more difficult to characterize. We used liquid and plate assays to determine whether these interactions occur between marine unicellular cyanobacteria of the genus Synechococcus. We have found a clear growth impairment of Synechococcus sp. CC9311 and Synechococcus sp. WH8102 when they are cultured in the presence of Synechococcus sp. CC9605. The genome of CC9605 contains a region showing homology to genes of the Escherichia coli Microcin C (McC) biosynthetic pathway. McC is a ribosome-synthesized peptide that inhibits translation in susceptible strains. We show that the CC9605 McC gene cluster is expressed and that three genes (mccD, mccA, and mccB) are further induced by coculture with CC9311. CC9605 was resistant to McC purified from E. coli, whereas strains CC9311 and WH8102 were sensitive. Cloning the CC9605 McC biosynthetic gene cluster into sensitive CC9311 led this strain to become resistant to both purified E. coli McC and Synechococcus sp. CC9605. A CC9605 mutant lacking mccA1, mccA2, and the N-terminal domain of mccB did not inhibit CC9311 growth, whereas the inhibition of WH8102 was reduced. Our results suggest that an McC-like molecule is involved in the allelopathic interactions with CC9605.

The production of secondary metabolites to inhibit either competitors or predators, known as allelopathy, is thought to play an important role in shaping microbial communities (1–3). The allelopathic potential of cyanobacteria was discovered in the 1970s (4, 5), and subsequent studies have continued to focus largely on freshwater cyanobacteria (6–10). The cyanobacterial repertoire of secondary metabolites is large and synthesized through diverse biochemical pathways (11). Recently, there has been increased interest in one class of compounds called bacteriocins, which are ribosomally synthesized peptides that become modified and are typically exported. Cyanobactins and Prochlorosins are examples of these types of compounds produced by cyanobacteria (12–14). Bioinformatic analyses of cyanobacterial genomes have shown the widespread occurrence of the potential to make bacteriocins of diverse types in almost every cyanobacterium (15).

Despite the recognized ecological importance of marine cyanobacteria, very little is known to date about their allelopathic interactions with each other or other marine bacteria (11). Lanthipeptides called Prochlorosins have been discovered in the marine cyanobacteria Prochlorococcus, although their function is unclear, because antibiotic activity has not been shown (12, 14). Some marine intertidal cyanobacterial strains have been shown to produce substances with inhibitory effects on Gram-positive bacteria or Artemia (16, 17), but the identity of these compounds is not known. Furthermore, certain marine bacteria may produce compounds that positively or negatively affect the growth of marine cyanobacteria, which has been shown for Prochlorococcus (18).

Unicellular marine Synechococcus are cyanobacteria found throughout the world’s oceans, and they are particularly abundant in coastal environments. Their discovery in 1979 (19, 20) soon led to a realization of their important contribution to primary productivity (21, 22). It has also become clear that there is significant diversity in the group (21), including distinct genetic clusters with distinct distributions in the water column (23–27), implying the presence of distinct species with different ecological niches. One might wonder whether allelopathic interactions are involved in this differentiation. In fact, antibiotic interactions within microbial communities have been proposed as an effective way of maintaining bacterial diversity (28).

The complete genomic sequences of a number of these abundant photosynthetic microbes are available and have been analyzed in detail (29–33). Recently, metagenomic studies have also added to our picture of marine Synechococcus diversity (34). Collectively, genomic and metagenomic studies suggest that these organisms have a core genome (shared by all Synechococcus), potentially a clade-specific genome, and a variable, possibly strain-specific genome dominated by horizontal gene transfer.

Bioinformatic approaches have identified gene clusters with homology to the biosynthetic gene cluster of the antibiotic Microcin C (McC; also called Microcin C7, C51, or C7/C51) of Enterobacteriacea in two marine cyanobacteria: Synechococcus strains CC9605 and RS9916 (35) (Fig. 1). Microcins are small (less than 10 kDa) bacteriocins produced by Escherichia coli and its close relatives (36, 37) that inhibit the growth of closely related bacterial species by targeting essential functions, such as DNA replication, transcription, and translation (35). Some microcins, such as the object of this study (McC), are initially ribosomally synthesized as a core peptide and heavily posttranslationally modified by dedicated maturation enzymes (35). The peptide moiety is required for the entry of unprocessed McC into sensitive cells, where it must be cleaved inside the target cell by peptidases to generate the inhibitory aminoacyl-nucleotide part of the drug (processed McC), which mimics the aspartyl adenylate and inhibits the aspartyl tRNA synthetase, thus blocking protein synthesis at the translation step (38). The McC-like compound from Synechococcus strains CC9605 and RS9916 is structurally and functionally different from known ribosomally synthesized and posttranslationally modified peptides synthesized by other cyanobacteria (39) [for example, the cyanobactins produced by Microcystis and other cyanobacterial strains (13) or the lanthipeptides produced by Prochlorococcus MIT9313 (12)]. McC was not found using comprehensive cyanobacterial genome bioinformatic analyses for bacteriocins (15), possibly because it is missing a double glycine processing site and a conserved leader peptide.

Fig. 1.

Genetic organization of McC biosynthetic gene cluster in E. coli and predicted McC-like biosynthetic gene clusters in Synechococcus strains. Block arrows denote genes, with arrows pointing in the direction of transcription. Matching colors indicate homologous genes. Accession numbers for protein sequences from the National Center for Biotechnology Information Entrez database are indicated below each gene [Synechococcus sp. CC9605 ORFs syncc9605_2504 (mccD), syncc9605_2505 (mccE1), syncc9605_2506 (mccE2), and syncc9605_2507 (mccB)]. NA, not annotated. At the top is a schematic showing the McC biosynthesis operon along with the functions of individual mcc gene products based on previously published information. Modified from refs. 35 and 48.

In E. coli, the McC biosynthetic cluster consists of the plasmid-encoded mccABCDE operon (Fig. 1). mccA encodes the heptapeptide McC precursor. In both Synechococcus strains, mccA heptapeptide ORFs were not annotated in the gene cluster, but two (sp. CC9605) and three (sp. RS9916) direct repeats, each potentially encoding 56- to 57-aa polypeptides, could be found (Fig. 1). It was suggested that the products of these ORFs might be subject to posttranslational modification and act as microcins (35). In E. coli, MccB adenylates the MccA heptapeptide, whereas MccD and the N-terminal domain of MccE are required for phosphate modification with propylamine (40). McC in E. coli functions through a Trojan horse mechanism of action (38, 41).

In this work, we describe an example of an allelopathic interaction occurring between marine Synechococcus strains. We focused on the inhibitory effect that strain CC9605 has on the growth of strains CC9311 and WH8102 and the hypothesis that it is because of a cyanobacterial version of an McC-like biosynthetic gene cluster encoded by Synechococcus sp. CC9605. Additionally, we studied the effect of purified E. coli McC on several marine Synechococcus strains.

Results

Allelopathic Interactions Between Strains of Marine Synechococcus.

Because it is not possible to distinguish Synechococcus strains from one another by microscopy or epifluorescence microscopy techniques, we characterized the allelopathic interactions between different Synechococcus strains in liquid cocultures by measuring the relative abundance of each Synechococcus strain using the relative abundance of the rpoC1 marker gene and specific primers and conditions for each strain as described by Tai and Palenik (ref. 25; Table S1). Synechococcus sp. CC9605 always dominated when cocultured with CC9311 or WH8102 (Fig. 2). Synechococcus sp. CC9311 dominated the coculture when growing with Synechococcus sp. WH8102 (Fig. 2). These effects were also seen in solid medium. When a spot of CC9605 was plated on an existing lawn of CC9311 or WH8102, a zone of clearing developed, which was fivefold larger in the case of WH8102 (Fig. 3). The inhibition of CC9311 over WH8102 was also observed in solid medium. Both the liquid and plating approaches, thus, showed that CC9605 inhibits the growth of CC9311 and WH8102, likely through a bacteriocidal mechanism.

Fig. 2.

Allelopathic interactions in liquid cocultures of Synechococcus strains. The proportion of each Synechococcus strain in liquid cocultures was determined by Q-PCR after 10 d and a subsequent 10 d (day 20) after retransfer.

Fig. 3.

Study of allelopathic interactions between Synechococcus strains on solid medium. Plates containing lawns of each Synechococcus strain—WH8102, CC9311, and CC9605—with drops of the same Synechococcus strains spotted in the middle of each plate.

Synechococcus CC9605 Is Resistant to E. coli McC.

Because our results could be explained by an McC-like compound potentially produced by CC9605, we characterized the toxic effect of McC purified from E. coli (38). Synechococcus strains WH8102 and CC9311 were very sensitive to E. coli McC, showing clear growth inhibition when McC was added up to 100 µM (CC9311) and 10 µM (WH8102) (Fig. S1). In contrast, Synechococcus strains CC9605 and RS9916 were resistant to E. coli McC (Fig. S1).

Cloning mccE2 from Synechococcus sp. CC9605 into an McC-Sensitive E. coli Strain Confers Resistance to McC.

To determine whether the Synechococcus sp. CC9605 mccE2 gene could confer resistance to McC, we cloned this gene into an McC-sensitive E. coli strain, TOP10. E. coli TOP10 carrying the plasmid pPY8 and containing the CC9605 mccE2 gene became fully resistant to the McC concentrations assayed (Fig. S2), whereas the controls, the TOP10 strain alone, and TOP10 cells carrying plasmid pPY9, which lacks the promoter and beginning of mccE2 (Table S2), were sensitive to McC (Fig. S2), showing a visible growth inhibition at McC concentrations ranging from 1 mM to 20 µM.

Expression Pattern of the mcc Genes.

In Synechococcus CC9605, the mcc genes are in the order mccD, mccE1, mccE2, mccA1, mccA2, and mccB and were expressed under exponential growth as seen by RT-PCR of individual genes (Fig. S3). Cotranscription of consecutive genes was also analyzed by RT-PCR, a common method for analysis of cotranscription (42). RT-PCR amplification products, indicating cotranscription of mccA2 and mccB and cotranscription of mccE2 and mccA1 (Fig. S4), were obtained with the primer pairs Mic-RT16/Mic-RT23 and Mic-RT5/Mic-RT8, respectively (Fig. S4). In contrast, no amplification product was observed with Mic-RT1/Mic-RT10, Mic-RT1/Mic-RT20, Mic-RT3/Mic-RT6, and Mic-RT14/Mic-RT8 (Fig. S4), indicating no cotranscription of the whole gene cluster or regions mccB-mccE2, mccE1-mccE2, or mccA1-mccA2, respectively. The results are summarized in the gene expression scheme shown in Fig. S4.

Synechococcus sp. CC9311 Induces the Expression of Some mcc Genes in Synechococcus sp. CC9605.

Using quantitative RT-PCR and internal primers specific for each ORF, we analyzed the expression level of the ORFs associated with the biosynthesis of the McC-like molecule in Synechococcus sp. CC9605. Because of the identical DNA sequence of the two mccA copies, we were unable to differentiate the expression level of each one separately; thus, we determined the expression of both mccA ORFs together, and mccA will be referred to as a single gene. Among the ORFs composing the McC-like biosynthetic gene cluster in CC9605, the mccA and mccD genes were expressed in axenic exponentially growing cultures and strongly (more than threefold) up-regulated in response to the addition of CC9311. Induction of mccA and mccD was observed after 6 h of mixed incubation, and maximum expression occurred after 24 h for both mccA and mccD genes (Fig. 4A), increasing up to 8.82 ± 0.93- and 13.11 ± 2.51-fold, respectively. In the case of the other three genes, only mccB underwent a significant change in expression but only after 24 h (2.91 ± 0.78-fold), whereas mccE1 and mccE2 did not exhibit any significant change in gene expression (Fig. 4A).

Fig. 4.

Expression of mcc-like genes in cocultures. (A) mccA and mccD genes are induced in the presence of CC9311 to a greater extent than the other genes. (B) mccA gene expression is induced by Synechococcus CC9311 and a mixed population of heterotrophic marine bacteria of unknown types but not supernatant from CC9311 cultures or some other bacteria, such as Synechococcus WH8102, Vibrio, and E. coli, or glass beads. *Significant expression changes.

mccA induction was proportional to the concentration of CC9311 added to a coculture. When different initial concentrations of CC9311 were used in 24-h induction experiments, the highest concentration (5 ± 1 × 10⁸ cells mL⁻¹) resulted in an 8.82 ± 0.93-fold change of the mccA expression, whereas the mRNA level only changed 3.88 ± 2.8- and 1.01 ± 0.016-fold at lower cell concentrations (5 ± 1 × 10⁷ and 5 ± 1 × 10⁶ cells mL⁻¹, respectively) (Fig. S5).

To determine whether expression of the mccA gene was induced by other microorganisms, we carried out 14-h incubations of Synechococcus sp. CC9605 with Synechococcus sp. WH8102, E. coli strain TOP10, Vibrio cholerae T2123, and a mix of unknown heterotrophic marine bacteria isolated from a culture of the heterotrophic nanoflagellate Pteridomonas (Table S2). We did see a 2.83 ± 0.64-fold expression change with the mix of unknown marine heterotrophic bacteria (Fig. 4B), but we did not see a significant change in the mccA gene expression (Fig. 4B) when CC9605 was cocultured with the other organisms, including WH8102. In addition, 1.03-µm glass beads did not generate any change in the mccA gene expression (Fig. 4B).

Cloning the McC-Like Biosynthetic Gene Cluster from Synechococcus sp. CC9605 into Synechococcus sp. CC9311 Confers Resistance to Synechococcus sp. CC9605 and Purified E. coli McC.

We cloned the McC-like biosynthetic gene cluster from CC9605 into the comEC gene of CC9311, resulting in strain PY28 (Fig. S6 and Table S2).

In liquid competition experiments between strain PY28 and Synechococcus sp. CC9605, PY28, unlike its WT parent CC9311, was able to grow in coculture with CC9605 (Fig. 5E). This ability was also evident in plating experiments, where after 14–20 d of solid coculture, an inhibition zone was not observed in the PY28 lawn (Fig. 5B), whereas CC9311 was still inhibited (Fig. 5A). We tested the sensitivity of strain PY28 to purified E. coli McC on plates, and as expected, it was fully resistant (Fig. 5D).

Fig. 5.

Sensitivity of Synechococcus strain CC9311-PY28 to CC9605 and purified E. coli McC. Synechococcus strain CC9311-PY28 resists CC9605 in (B) solid medium and (E) liquid conditions as well as (D) purified McC from E. coli. WT CC9311 was sensitive to CC9605 in (A) solid medium and liquid conditions (Fig. 2) as well as (C) purified McC from E. coli (concentrations of McC are indicated).

Partial Deletion of McC-like Biosynthesis Genes in Synechococcus sp. CC9605 Abolishes Growth Inhibition of Synechococcus sp. CC9311 and Reduces Growth Inhibition of Synechococcus sp. WH8102.

The region of mccA1, mccA2, and the N-terminal domain of mccB in Synechococcus sp. CC9605 was substituted with a kanamycin resistance cassette, resulting in strain PY35 (Fig. S7 and Table S2). PY35 did not inhibit the growth of CC9311 (Fig. 6D) and still inhibited the growth of WH8102 (Fig. 6B), but it produced an area of inhibition 57% smaller than the area produced by WT CC9605 (Fig. 6A).

Fig. 6.

Sensitivity of Synechococcus strains WH8102 and CC9311 to Synechococcus strains CC9605 and CC9605-PY35. Synechococcus strain WH8102 was inhibited by both (A) CC9605 and (B) the mutant strain CC9605-PY35, although the diameter of the zone of inhibition by the mutant was 56.7% smaller than the diameter produced by the WT strain. Unlike WH8102, the growth of CC9311 was only inhibited by (C) the WT strain CC9605, whereas (D) PY35 did not inhibit CC9311 growth.

Discussion

Most studies of allelopathic interactions between cyanobacteria have been performed in freshwater environments (11), and very little is known about this phenomenon among marine cyanobacteria, especially Synechococcus. Using a quantitative PCR strategy to quantify marine Synechococcus clades within a liquid medium coculture (25), our results revealed that Synechococcus sp. CC9605 clearly impairs the growth of strains CC9311 and WH8102 under the conditions assayed (Figs. 2 and 3). WH8102 seems to be fivefold more sensitive than CC9311 to CC9605 in solid medium. Interestingly, CC9605 and WH8102 co-occur in oligotrophic environments, where CC9605 is thought to be more abundant than WH8102 (25, 43). We also found that Synechococcus sp. CC9311, which does not encode an McC-like gene cluster, noticeably inhibits the growth of Synechococcus sp. WH8102 (Figs. 2 and 3), presumably by the production of a different allelochemical. These results are evidence of allelopathic interactions between marine Synechococcus strains.

Considering the zone of inhibition produced by CC9605 over the other two strains on plates (Fig. 3), it is likely that the CC9605 McC is secreted into the medium, like it is in E. coli, and then taken up by the target cells. WH8102, CC9311, and CC9605 all encode genes homologous to the ATP-binding cassette (ABC) transporter responsible for taking up McC in E. coli (44) that might be involved in the uptake of the CC9605 McC-like molecule. E. coli YejA, YejB, YejE, and YejF homologs are SYNW0709, SYNW0708, SYNW1683, and SYNW2325, respectively, in WH8102; sync_0939, sync_0938, sync_0676, and sync_2703, respectively, in CC9311; and Syncc9605_1959, Syncc9605_1960, Syncc9605_800, and Syncc9605_2456, respectively, in CC9605. However, unlike E. coli, CC9605 does not encode any homologs to the E. coli mccC efflux transporter (Fig. 1). We hypothesize that the three ORFs upstream of the CC9605 McC biosynthetic gene cluster might be involved in CC9605 McC secretion. They are annotated as components of an ABC efflux transporter and are also expressed under exponential growth in CC9605 (Fig. S3).

Allelopathic interactions involving antibiotics have been described for marine heterotrophic bacteria (1). Because we hypothesized that CC9605 dominance over WH8102 and CC9311 was due to an McC-like antibiotic, we tested the sensitivity of all of these strains to McC purified from E. coli (38). McC inhibited the growth of CC9311 and WH8102 but not CC9605 or RS9916 (Fig. S1), which are both known to carry an McC-like biosynthetic gene cluster (35). Therefore, E. coli McC can inhibit cyanobacteria.

Synechococcus sp. CC9605 has at least one resistance gene, the mccE2 gene, which we showed here can confer McC resistance in E. coli (Fig. S2) and presumably provides self-immunity against its own McC-like antibiotic. Synechococcus sp. RS9916 carries a putative mccF-like immunity gene instead of an mccE-like gene (Fig. 1).

McC is produced by members of the family Enterobacteriaceae, and its biosynthetic pathway is thought to be horizontally transferred within the family (45). We believe that the two unrelated marine Synechococcus strains, CC9605 and RS9916, potentially acquired the McC-like biosynthetic gene cluster by horizontal gene transfer. The nucleotide sequence also supports our hypothesis, because the G+C content determined for both McC-like gene clusters was 42.6% (CC9605) and 45.1% (RS9916), values significantly lower than the values found for their genomes: 59.2% (CC9605) and 59.8% (RS9916). Horizontal gene transfer of the McC-like gene cluster into some Synechococcus strains would explain why it has not been found in other sequenced cyanobacterial genomes to date.

Surprisingly, when the competitor Synechococcus sp. CC9311 was incubated with Synechococcus sp. CC9605 in a 24-h time course, the expression of some microcin-like genes, mccD and mccA, increased significantly after only 6 h of incubation (Fig. 4A), whereas mccB was induced after 24 h. This result suggests that CC9605 can sense the presence of CC9311 and respond by inducing the expression of McC-like synthesis genes. Several molecular mechanisms could be responsible for the observed expression induction from physical contact without any kind of cell specificity to specific cell surface recognition. In the latter case, it could be a response to a broad number of microorganisms, or alternatively, it could be a more specific response to a few microorganisms. Our results suggest that the response may occur through cell surface recognition of CC9311, because we did not obtain significant induction of mccA expression when supernatant from a CC9311 culture, glass beads, or several other microorganisms was used (Fig. 4B). However, Synechococcus sp. CC9605 does not exclusively recognize and respond to Synechococcus sp. CC9311. We found a 2.83 ± 0.64-fold induction when a mixture of unidentified marine heterotrophic bacteria was used. Additional studies are necessary to identify the molecular mechanism by which CC9605 recognizes CC9311 and induces the expression of some of the mcc-like genes.

The McC-like gene cluster from Synechococcus sp. CC9605 was cloned into the McC-sensitive Synechococcus sp. CC9311. As expected, strain PY28 became resistant to Synechococcus sp. CC9605 (Figs. 5 B and E), and PY28 also acquired immunity to McC purified from E. coli (Fig. 5D). Furthermore, deletion of mccA1, mccA2, and the N-terminal domain of mccB in CC9605 (strain PY35) abolishes its inhibitory activity on strain CC9311 (Fig. 6 C and D). Thus, our findings strongly support the idea that allelopathic interactions between CC9605 and CC9311 are because of an McC-like compound. The constitutive expression of these genes in Synechococcus sp. CC9605 under exponential growth could be ecologically costly but provide an important defensive role against competitors or even predators; having the ability to further induce expression allows tuning of the defense to changing environmental conditions and co-occurring microbes.

Unlike CC9311, the growth of WH8102 was still inhibited by the mutant strain PY35, although to a lesser extent than the WT strain CC9605 (Fig. 6 A and B), indicating that, in addition to an McC-like compound, the strong growth inhibition of WH8102 by CC9605 is also caused by at least another allelochemical that CC9605 might produce. As has been shown by previous bioinformatic analyses of cyanobacterial genomes (15), CC9605 bears two gene clusters that potentially encode for two different bacteriocins. We believe that at least one of these bacteriocins might also be involved in the allelopathic interaction observed between CC9605 and WH8102.

Interestingly, CC9605 exhibits a tendency to form aggregates that can constitute from 65% to 75% of total CC9605 abundance in some culture conditions (46). If aggregation occurs in the environment as well, it would result in higher McC-like concentrations surrounding CC9605 cells. This behavior would be significant in the oligotrophic environments that CC9605 inhabits (25, 43, 47), where the total microbial cell density is low and the distance between single CC9605 cells and their nearest neighbor would otherwise be high (∼200 cell lengths, which has been suggested for Prochlorococcus) (12). It has also been proposed that antibiotic production may be a mechanism used by particle specialists to dominate the particle phase by deterring other potential colonizers (1).

We have shown that allelopathic interactions occur between marine Synechococcus strains. Our results suggest that the product of an McC-like gene cluster is involved in these interactions when CC9605 is present, and provides an example of an active McC-like gene cluster outside of enteric bacteria. In addition, we found that the growth of WH8102 was also inhibited by strain CC9311. Because this effect is not McC-dependent, additional investigations are needed to identify this compound and other new compounds produced by Synechococcus strains that might be involved in such interactions. Given the significant abundance of Synechococcus in marine waters, such allelopathic interactions are likely to be relevant in shaping the composition of the marine bacterial community as a whole.

Materials and Methods

The bacterial strains and plasmids used in this study are listed in Table S2. SI Materials and Methods discusses specific culture conditions, cyanobacterial genomic DNA and RNA extraction, quantitative real-time PCR, toxicity assay of E. coli McC on strains of Synechococcus, and genetic manipulation in E. coli and Synechococcus strains.

Expression of mcc-Like Genes in Cocultures.

We obtained the gene expression profiles of all of the genes contained in the McC-like biosynthetic gene cluster of Synechococcus sp. CC9605 after 1, 3, 6, 9, and 24 h of coculture with Synechococcus sp. CC9311. Batch cultures of CC9311 and CC9605 were grown up to late exponential/early stationary phase. Cells from both cultures were mixed at the same initial cell density (∼5 ± 1 × 10⁶ cells mL⁻¹) and grown at 25 °C under stirring conditions. A CC9605 culture was used as a control. At the indicated times, samples were taken, and RNA was extracted and used in quantitative RT-PCR analysis. The gene expression of the different genes was analyzed using the oligonucleotides primer pairs Mic-RT7/Mic-RT8 (mccA), Mic-RT9/Mic-RT21 (mccB), Mic-RT1/Mic-RT2 (mccD), Mic-RT12/Mic-RT20 (mccE1), and Mic-RT13/Mic-RT18 (mccE2) (Fig. 4A and Table S1).

The induction of mccA in response to different initial CC9311 cell concentrations was determined. The starting cell concentration of CC9311 was 5 ± 1 × 10⁸ cells mL⁻¹. We did 24-h incubations of strains CC9311 and CC9605 and varied the initial cell density of CC9311 in a series of 10-fold dilutions.

Induction of mccA after 14 h of coincubation with Synechococcus sp. WH8102, E. coli strain TOP10, Vibrio cholerae T2123, a mix of unknown heterotrophic marine bacteria (Table S2), or 1.03-µm glass beads (prewashed two times with SN medium at a final bead concentration of 5.3 × 10⁸ beads mL⁻¹; Molecular Probes Inc.) was determined.

Study of Allelopathic Interactions Between Marine Synechococcus Strains.

To study allelopathic interactions between different Synechococcus strains, duplicate 50-mL cultures consisting of two Synechococcus strains (CC9605 vs. CC9311, CC9605 vs. WH8102, and CC9311 vs. WH8102) were inoculated into SN medium at the same initial cell density (∼5 ± 1 × 10⁶ cells mL⁻¹) and grown at 25 °C with shaking; 2 mL of each coculture were collected at the beginning of the experiment (zero time point) as well as after 10 d (10-d time point). After 10 d of growth, 1 mL each coculture was transferred to 50 mL fresh SN medium, and 2 mL each coculture were collected after another 10 d of growth under the same conditions (20-d time point). DNA was extracted as described above. To calculate the relative abundance of each Synechococcus strain, quantitative PCR (Q-PCR) was carried out as described for Q-RT-PCR above but using genomic DNA instead of cDNA. Oligonucleotide primers and Q-PCR conditions specific for the rpoC1 gene of each Synechococcus strain were used as previously described (25) (Table S1). Each strain was also grown separately as a control; they were transferred using the same incubation periods as the experimental cocultures. When the control samples were collected, we mixed 1 mL each strain grown separately to recreate the combinations that we were assaying in the experiment. Growth rates were calculated for each strain growing separately. Growth was monitored by measuring phycoerythrin fluorescence (excitation = 544 nM and emission = 577 nM; Turner AU-10; Turner Designs). The specific growth rates of these cultures were determined using geometric mean linear regression of a plot of ln (culture fluorescence or cells per milliliter) vs. time during midexponential phase. These growth conditions resulted in similar growth rates for all of the cultures in the conditions assayed: 0.39 ± 0.04 d⁻¹ (WH8102), 0.40 ± 0.07 d⁻¹ (CC9311), 0.37 ± 0.06 d⁻¹ (CC9605), and 0.40 ± 0.06 d⁻¹ (PY28).

Allelopathic interactions were also analyzed in solid medium using pour plates containing lawns of each Synechococcus strain (WH8102, CC9311, or CC9605; 5 ± 1 × 10⁶ cells mL⁻¹). After incubation at 25 °C with a constant illumination of 10 microeinsteins m⁻² s⁻¹ for 1 wk, drops of each test Synechococcus strain were spotted on the plate (Fig. 3). A circle of agarose was removed from the middle of the plate, and 100 µL cells (5 × 10¹⁰ cells mL⁻¹) premixed with SN containing 0.4% (wt/vol) agarose were spotted in the hole. As a control, each Synechococcus strain was plated separately without drops of other strains.