|Home | About | Journals | Submit | Contact Us | Français|
The bacterial phosphoenolpyruvate phosphotransferase system (PTS) is a highly conserved phosphotransfer cascade that participates in the transport and phosphorylation of selected carbohydrates and modulates many cellular functions in response to carbohydrate availability. It plays a role in the virulence of many bacterial pathogens. Components of the carbohydrate-specific PTS include the general cytoplasmic components enzyme I (EI) and histidine protein (HPr), the sugar-specific cytoplasmic components enzymes IIA (EIIA) and IIB (EIIB), and the sugar-specific membrane-associated multisubunit components enzymes IIC (EIIC) and IID (EIID). Many bacterial genomes also encode a parallel PTS pathway that includes the EI homolog EINtr, the HPr homolog NPr, and the EIIA homolog EIIANtr. This pathway is thought to be nitrogen specific because of the proximity of the genes encoding this pathway to the genes encoding the nitrogen-specific σ factor σ54. We previously reported that phosphorylation of HPr and FPr by EI represses Vibrio cholerae biofilm formation in minimal medium supplemented with glucose or pyruvate. Here we report two additional PTS-based biofilm regulatory pathways that are active in LB broth but not in minimal medium. These pathways involve the glucose-specific enzyme EIIA (EIIAGlc) and two nitrogen-specific EIIA homologs, EIIANtr1 and EIIANtr2. The presence of multiple, independent biofilm regulatory circuits in the PTS supports the hypothesis that the PTS and PTS-dependent substrates have a central role in sensing environments suitable for a surface-associated existence.
Attachment of a free-swimming bacterium to a surface, which is termed biofilm formation, is the result of a complex decision tree that occurs when a bacterium encounters a surface (19). Environmental signals dictate the decisions made at each branch point. The advantages afforded to a bacterium by surface attachment depend on the environmental niche of the bacterium being studied, and the environmental signals that induce biofilm formation reflect this.
Formation of a multilayer bacterial biofilm often requires synthesis of an extracellular matrix composed of biological polymers that enhance interbacterium attachment. These extracellular polymers may be proteins, polysaccharides, and/or DNA (7). Synthesis of the biofilm matrix is often tightly regulated. The environmental activators of many signaling pathways that modulate multilayer biofilm accumulation have been identified. These activators include specific carbohydrates, quorum-sensing molecules, nucleic acids and their precursors, and polyamines (1, 6, 9, 13, 15, 17, 18, 26, 31, 33, 35, 41, 45, 53). However, there are many known biofilm regulatory pathways for which no environmental activator has been identified yet.
The phosphoenolpyruvate phosphotransferase system (PTS) is a multicomponent phosphotransfer cascade that mediates transport and phosphorylation of selected sugars, such as glucose, sucrose, mannose, and N-acetylglucosamine (10). In addition, it has been implicated in the formation of biofilms by diverse organisms (1, 2, 17, 31, 42). Phosphate enters the PTS through transfer from phosphoenolpyruvate to the first PTS component, the phosphoenolpyruvate-protein phosphotransferase or enzyme I (EI). EI in turn transfers the phosphate group to another component of the PTS, histidine protein (HPr). Many bacterial genomes also encode a protein homologous to HPr termed FPr, which is preferred for transport of fructose through the PTS. HPr and FPr transfer phosphate to a number of enzymes II, which are multisubunit, membrane-associated complexes that carry out transport and phosphorylation of specific PTS substrates. Because transport of PTS substrates rapidly depletes the PTS of phosphorylated intermediates, the phosphorylation states of PTS components serve as cytoplasmic reporters of environmental nutrient availability. These reporters then modulate cellular functions such as chemotaxis (60), uptake and catabolism of PTS-independent carbohydrates (1, 12, 39), and glycogen breakdown (54, 55).
The genomes of many Gram-negative organisms contain genes encoding another phosphotransfer cascade that is homologous to the carbohydrate-transporting PTS. Because these genes are close to rpoN, which encodes the sigma factor involved in transcription of many genes required for nitrogen assimilation, this phosphotransfer cascade is termed the nitrogen-related PTS or PTSNtr (22, 49, 50). The components of this phosphotransfer cascade include EINtr, NPr, and EIIANtr. Unlike the carbohydrate-transporting PTS, PTSNtr does not include membrane-associated components and, therefore, does not participate directly in transport of nutrients into the cell. Rather, its primary function in the cell is thought to be regulatory (22, 24, 25, 32, 44). While the breadth and overarching goal of the PTSNtr have not been defined, transfer of phosphate between the two PTS cascades may be one mechanism by which the PTSNtr influences cellular function (46).
Vibrio cholerae is a Gram-negative bacterium whose natural habitat includes environments with low and intermediate salinities, such as ponds and estuaries (8). There is some evidence that V. cholerae forms a multilayer, exopolysaccharide-based biofilm in freshwater environments (20). The exopolysaccharide is referred to as VPS (Vibrio polysaccharide), and the region of the V. cholerae genome containing many of the genes required to make the biofilm matrix is known as the vps island (64).
When pathogenic V. cholerae is ingested by humans in contaminated food or water, the diarrheal disease cholera results. While many studies of human and murine infection have suggested that proteins required for biofilm matrix synthesis are expressed in vivo (14, 30), no study has found that these proteins are essential for colonization of the mammalian intestine (16, 23, 63).
The V. cholerae genome encodes 25 PTS components, including two EI homologs, three HPr homologs, and nine EIIA homologs (Fig. (Fig.1).1). A role in carbohydrate transport has been established for a subset of these components (16). We recently discovered that supplementation of minimal medium (MM) with PTS sugars activates transcription of the vps genes and formation of a multilayer biofilm (40). Furthermore, we found that phosphorylated form of EI represses V. cholerae biofilm formation in minimal medium (17). Subsequent biofilm assays conducted with Luria-Bertani (LB) broth suggested that there are additional PTS-based regulatory pathways. Here we define three PTS-based biofilm regulatory pathways that are present when this medium is used. These studies illustrate the complexity of the regulation of V. cholerae biofilm accumulation by PTS components and underscore the importance of carbohydrates as signals in the decision of V. cholerae cells to join a biofilm.
The bacterial strains and plasmids used in this study are listed in Table Table1.1. V. cholerae strain PW357 was used for biofilm assays. Bacteria were cultivated in Luria-Bertani broth or minimal medium supplemented with 0.5% (wt/vol) glucose or pyruvate (Sigma) (17). Where indicated below, the medium was also supplemented with streptomycin (100 μg/ml), ampicillin (50 or 100 μg/ml), or l-arabinose (0.04% wt/vol). Where specified below, a 0.1 M phosphate-buffered saline solution (PBS) (pH 7.0) was used.
The plasmid used to make the Δmlc mutant, pKEKΔmlc, was generously provided by J. Reidl (3). All other plasmids used in the construction of insertion and in-frame deletion mutants were available in our laboratory (16, 17). Suicide plasmids were used to generate insertion and in-frame deletion mutants by single and double homologous recombination, respectively.
Plasmids used for rescue experiments, which are listed in Table Table1,1, were constructed as previously described (17, 61) using the primers listed in Table Table2.2. Briefly, the native sequence of the targeted gene was amplified by PCR and cloned into pBAD-TOPO (Invitrogen) using the manufacturer's protocol. Point mutation sequences were generated by amplification of two gene fragments using internal primers with overlapping sequences containing the desired mutation. The two fragments were joined using the splicing by overlap extension (SOE) technique and cloned into pBAD-TOPO (Invitrogen) using the manufacturer's protocol. The sequence of the inserted fragment was confirmed by amplification and sequence analysis. The rescue plasmids were then introduced into V. cholerae strains by electroporation, and transformants were selected on LB agar supplemented with ampicillin. A pBAD-TOPO vector containing an antisense fragment of the coding sequence for EIIAGlc(H91A) was used as a control plasmid for rescue experiments. Expression of the cloned protein was evaluated by Western blot analysis as previously described (16).
Quantification of surface association was performed as described previously (20). Briefly, the strains were grown overnight on LB agar plates at 27°C. The following morning, the resulting colonies were used to inoculate borosilicate tubes filled with 300 μl of LB broth or MM supplemented with glucose or pyruvate. After incubation for 18 to 24 h at 27°C, each planktonic cell suspension was collected, and the planktonic cell density was estimated by measuring the optical density at 655 nm (OD655) using a Benchmark Plus microplate spectrophotometer (Bio-Rad). To quantify the surface-attached cells, 300 μl of PBS and a small number of 1-mm glass beads were added to the surface-attached cells remaining in the borosilicate tube, and the cells were dispersed by vortexing. The OD655 of the resulting cell suspension was measured. Each total growth value represented the sum of the OD655 recorded for the planktonic and surface-associated cell suspensions. All values are the means of at least three experimental replicates, and standard deviations were determined. Statistical significance was calculated using a two-tailed t test.
Wild-type V. cholerae was cultured for 18 h in LB broth or MM supplemented with pyruvate (0.5%, wt/vol). The cells were then pelleted by centrifugation, and total RNA was isolated using an RNeasy kit (Qiagen), followed by RNase-free DNase I treatment to remove contaminating DNA. Reverse transcription-PCR (RT-PCR) was performed using 1 to 2 ng of total RNA with a SuperScriptIII first-strand kit (Invitrogen). Subsequently, 15 ng of the resulting cDNA was used as the template for quantitative PCR using iTaq SYBR green Supermix with ROX (Bio-Rad) and 5 pmol of the relevant primers (Table (Table2)2) in a 20-μl reaction mixture. The level of the clpX (VC1921) transcript was used to normalize all measurements. Template-free reaction mixtures were included as controls to confirm the absence of contaminating DNA or RNA. The experiments were conducted with a StepOnePlus PCR system (Applied Biosystems) using the following steps: (i) 95°C for 10 min, (ii) 40 cycles of denaturation for 15 s at 95°C and annealing and extension for 1 min at 60°C, and (iii) dissociation curve analysis using temperatures from 60°C to 90°C. Data were analyzed using the StepOne V2.0 software (Applied Biosystems). Measurements were obtained in triplicate.
Strains were inoculated into LB broth to obtain an initial OD655 of approximately 0.05. Cultures were incubated at 27°C for 4 h with shaking until the OD655 was approximately 0.5. The OD655 was determined and subsequently used for normalization of measurements of β-galactosidase activity. Fifty or 100 μl of each culture was moved to a white 96-well plate (Nunc). Three freeze-thaw cycles were performed, and an equal volume of the β-Glo luminescent substrate (Promega) was added to each well. After incubation in the dark at room temperature for 30 min, luminescence was measured with an Infinite 200 spectrophotometer (Tecan). Three experimental replicates were included each time that the experiment was performed, the experiment was repeated three times, and similar results were obtained in the three experiments. Values for a representative experiment are reported below and are the means of three experimental replicates; standard deviations were also determined, and statistical significance was calculated using a two-tailed t test.
We first compared biofilm formation in LB broth and minimal medium supplemented with glucose by a ΔEI mutant, a ΔEIIAGlc mutant, and a ΔPTS mutant lacking the genes encoding EI, HPr, and EIIAGlc (Fig. (Fig.2).2). Similar to what was observed in minimal medium, in LB broth a ΔEI mutant exhibited increased surface accumulation compared with wild-type V. cholerae, while a ΔEIIAGlc mutant exhibited decreased surface accumulation. As previously noted (17), surface accumulation by a ΔPTS mutant lacking EI, HPr, and EIIAGlc was increased in minimal medium supplemented with glucose. In contrast, the surface accumulation by the ΔPTS mutant was similar to that by wild-type V. cholerae in LB broth. This suggested that the biofilm phenotype of the ΔPTS mutant in LB broth might reflect the additive effects of two opposing regulatory pathways.
In both minimal medium and LB broth, EI represses biofilm formation (17). To determine if the phosphorylated form of EI is required for this repression, we measured biofilm formation by a ΔEI mutant rescued with a control plasmid, with a plasmid carrying a wild-type EI allele, or with a plasmid carrying a mutant EI allele in which phosphorylated histidine 189 had been changed to alanine. As shown in Fig. Fig.3A,3A, while the biofilm phenotype of a ΔEI mutant was rescued by a wild-type EI allele provided in trans, the EIH189A allele was unable to rescue the biofilm phenotype despite adequate expression of the protein (Fig. (Fig.3B).3B). This suggests that, similar to what was observed in minimal medium (17), EI-P represses V. cholerae biofilm formation in LB broth. We previously reported that in minimal medium supplemented with glucose, repression of biofilm accumulation by EI becomes apparent only during the stationary phase of growth. This corresponds to the time at which the glucose in the growth medium is depleted (17) and phosphorylated PTS intermediates begin to accumulate. Because PTS substrates are less abundant in LB broth (20), we predicted that during growth in LB broth, repression of biofilm accumulation by EI-P would be observed much earlier in the growth cycle. As predicted, the difference in biofilm accumulation between wild-type V. cholerae and the ΔEI mutant occurred during the exponential phase of growth and increased in early stationary phase (Fig. (Fig.3C3C).
In minimal medium, repression of biofilm formation by EI requires phosphotransfer to either HPr or FPr (16). Furthermore, this repression occurs at the level of transcription. To determine if phosphotransfer from EI to HPr or FPr contributed to repression of biofilm accumulation in LB broth, we compared biofilm accumulation by wild-type V. cholerae and biofilm accumulation by ΔHPr, ΔFPr, and ΔHPr ΔFPr mutants. As shown in Fig. Fig.4A,4A, biofilm accumulation by a ΔHPr mutant was approximately 2-fold greater than that by wild-type V. cholerae, while biofilm formation by a ΔFPr mutant was not significantly different. To demonstrate that EI is upstream of HPr/FPr in the signaling pathway, we documented that a wild-type EI allele provided in trans repressed biofilm formation by a ΔEI mutant but was unable to repress biofilm formation in the absence of HPr and FPr (Fig. (Fig.4B).4B). Lastly, we measured activation of the vps genes in wild-type V. cholerae and ΔEI, ΔHPr, ΔFPr, and ΔHPr ΔFPr mutants using a chromosomal lacZ reporter fusion to the vpsL promoter. As shown in Fig. Fig.4C,4C, deletion of EI and HPr but not FPr increased transcription of the vps genes. Thus, the regulatory pathway leading to repression of biofilm formation and vps gene transcription through phosphorylation of EI and HPr is present in both minimal medium and LB broth. This pathway is referred to in this paper as pathway 1 (see Fig. Fig.1414).
To further define regulation of biofilm formation by the PTS, we compared biofilm accumulation in LB broth by wild-type V. cholerae and biofilm accumulation in LB broth by strains having mutations in the FPr gene, which encodes both EIIA and HPr domains, as well as in each of the nine genes in the V. cholerae genome encoding EIIA homologs (Fig. (Fig.1).1). As shown in Fig. Fig.5,5, we identified three EIIA mutants with altered biofilm phenotypes compared with wild-type V. cholerae. First, approximately 3-fold fewer ΔEIIAGlc mutant cells accumulated in a biofilm than wild-type V. cholerae cells. Second, mutations in VC2531 and VC1824 led to increases in biofilm accumulation. VC2531 (EIIANtr1) and VC1824 (EIIANtr2) both contain EIIA domains homologous to EIIANtr, and VC1824 also includes an N-terminal domain which is homologous to HPr but is not as closely related as the other V. cholerae HPr homologs (Fig. (Fig.1).1). Our observations suggested that at least two independent PTS pathways control biofilm formation at the level of EIIA in LB broth.
Because independent control of biofilm formation by an EIIA component was not observed in minimal medium (16), we hypothesized that transcription of EIIAGlc, EIIANtr1, and EIIANtr2 might be activated in LB broth compared to their transcription in minimal medium. To test this hypothesis, we used quantitative RT-PCR to compare the transcript levels of EI, HPr, FPr, EIIAGlc, EIIANtr1, and EIIANtr2 in LB broth and minimal medium supplemented with pyruvate (Fig. (Fig.6).6). Pathway 1 was present in both minimal medium and LB broth. However, the transcript levels of the pathway 1 components EI and HPr were lower in LB broth. The transcription of FPr, which does not appear to play a role in regulation of biofilm accumulation in LB broth, was elevated in LB broth compared with minimal medium. The transcription of EIIAGlc was approximately 6-fold greater in LB broth, suggesting that activation of biofilm formation by EIIAGlc in LB broth may be due at least in part to increased transcription of EIIAGlc. Transcription of the PTS components EIIANtr1 and EIIANtr2 was not increased in LB broth. Therefore, differential transcription of these components does not underlie the observation that they regulate biofilm formation in LB broth but not in minimal medium.
Phosphotransfer through EI and HPr is predicted to lead to EIIAGlc phosphorylation at histidine 91. To determine whether regulation of biofilm formation by EIIAGlc requires phosphorylation of this residue, we studied biofilm accumulation by a ΔEIIAGlc mutant containing an expression plasmid carrying either a control sequence, a wild-type EIIAGlc allele, an EIIAGlc allele with a histidine-to-alanine mutation at position 91, or an EIIAGlc allele with a histidine-to-aspartate mutation at position 91. As shown in Fig. Fig.7A,7A, while these EIIAGlc alleles did not affect the total growth, all of them activated biofilm accumulation in the ΔEIIAGlc mutant background. To further establish that this rescue was not dependent on either EI or HPr and, therefore, was not part of regulatory pathway 1, we measured activation of biofilm accumulation by the same EIIAGlc alleles in a ΔPTS mutant background, which lacks EI, HPr, and EIIAGlc (Fig. (Fig.7B).7B). If EIIAGlc were dependent on EI or HPr for activation of biofilm formation, an EIIAGlc allele would be unable to activate biofilm accumulation in the absence of EI or HPr. However, we observed that rescue of biofilm accumulation by the ΔPTS mutant was possible with each of the three EIIAGlc alleles but not with the control plasmid. Lastly, we demonstrated by Western analysis that all rescue constructs resulted in expression of a full-length protein in the ΔEIIAGlc mutant background (Fig. (Fig.7C).7C). Therefore, we concluded that activation of biofilm accumulation by EIIAGlc does not require phosphorylation of the conserved histidine at position 91 and that it represents a second, independent PTS-based biofilm regulatory pathway (regulatory pathway 2).
Several proteins are known to interact with EIIAGlc. Two of these proteins are EIIBCGlc and Mlc. EIIBCGlc is downstream of EIIAGlc in the glucose-specific PTS phosphotransfer cascade. Furthermore, in Escherichia coli, the DNA-binding protein Mlc has been shown to repress transcription of EIIBCGlc when PTS substrates are scarce (21). Unphosphorylated EIIBCGlc interacts directly with Mlc to relieve this repression (27, 43). The V. cholerae Mlc and EIIBCGlc proteins are very similar to the corresponding proteins of E. coli. We therefore asked whether EIIBCGlc and Mlc might play a role in regulation of V. cholerae biofilm formation by EIIAGlc. We first measured biofilm accumulation by wild-type V. cholerae, a ΔEIIBCGlc mutant, a ΔMlc mutant, and a ΔEIIAGlc mutant, as well as by strains with combinations of the mutations. As shown in Fig. Fig.8,8, the total levels of growth were comparable for all of the strains studied. However, biofilm formation by the ΔMlc and ΔEIIAGlc mutants was significantly decreased, while biofilm accumulation by the ΔEIIBCGlc mutant was increased compared with wild-type V. cholerae biofilm accumulation. Furthermore, deletion of either Mlc or EIIAGlc in a ΔEIIBCGlc mutant background was sufficient to decrease the biofilm accumulation almost to the levels observed for the ΔEIIAGlc mutant. This led us to hypothesize that the biofilm regulatory pathway shown in Fig. Fig.9A9A is present. In this signal transduction pathway, EIIBCGlc inhibits activation of EIIAGlc by Mlc. EIIAGlc, in turn, activates transcription of the vps genes by an as-yet-unidentified mechanism.
To further evaluate the presence of this signal transduction pathway, we measured vpsL transcription in wild-type V. cholerae and ΔEIIBCGlc, ΔMlc, and ΔEIIAGlc mutants, as well as in strains with combinations of these mutations. The results of this experiment are shown in Fig. Fig.9B9B along with the components of the signal transduction pathway present in each genetic background and the relative amount of vps transcription expected based on functional pathway components. The data support a model in which Mlc activates ΔEIIAGlc and is repressed by EIIBCGlc. Furthermore, the level of vps transcription in a ΔEIIAGlc ΔEIIBCGlc mutant is higher than that in a ΔEIIAGlc ΔEIIBCGlc ΔMlc mutant, suggesting that Mlc activates vps transcription even in the absence of EIIAGlc. Therefore, an additional pathway must be postulated, in which Mlc activates vpsL transcription independent of EIIAGlc. These pathways are referred to in Fig. Fig.9A9A as pathways 2a and 2b, respectively.
In general, the measurements of biofilm accumulation correlated well with measurements of vpsL transcription. However, differences were found. First, there was no difference in biofilm accumulation between ΔEIIAGlc and ΔMlc mutants. Second, biofilm formation by strains having mutations in EIIBCGlc as well as in EIIAGlc and/or Mlc all formed biofilms that were slightly but significantly and reproducibly larger than those formed by ΔEIIAGlc and ΔMlc mutants. One possible explanation for these observations is that EIIBCGlc also represses biofilm accumulation at the posttranscriptional level. The complex effect of EIIBCGlc on biofilm formation is currently under investigation.
We hypothesized that if Mlc were partially dependent on EIIAGlc for its effect on biofilm formation, the absence of EIIAGlc would compromise the ability of an mlc allele provided in trans to rescue an Mlc mutation. To test this hypothesis, the following epistasis experiment was performed. We introduced a wild-type allele of mlc into ΔMlc, ΔEIIBCGlc ΔMlc, and ΔEIIAGlc ΔEIIBCGlc ΔMlc mutants. As shown in Fig. Fig.10,10, while a control vector had no effect on biofilm accumulation by any of these mutants, introduction of the mlc allele into the ΔMlc and ΔEIIBCGlc ΔMlc mutants increased biofilm accumulation approximately 4-fold. In contrast, introduction of the mlc allele into a ΔEIIAGlc ΔEIIBCGlc ΔMlc mutant increased biofilm accumulation only 2-fold. This result supports our model for pathway 2 in which Mlc activates biofilm accumulation through EIIAGlc-dependent and -independent pathways.
The gene encoding EIIANtr, which is located close to rpoN, has been found in many bacterial genomes (36, 48, 51). Mutation of the two V. cholerae EIIANtr homologs, EIIANtr1 (60% identity and 75% similarity to E. coli EIIANtr) and EIIANtr2 (26% identity and 42% similarity to E. coli EIIANtr), resulted in increased biofilm accumulation (Fig. (Fig.5).5). The conserved histidine residues at position 66 of EIIANtr1 and position 172 of EIIANtr2 are predicted to be phosphorylated by phosphotransfer from the HPr homolog NPr, which in turn receives a phosphate from EINtr. We first determined whether phosphorylation of the conserved histidine was required for the function of EIIANtr1 by comparing biofilm accumulation data for ΔEIIANtr1 mutants carrying either a plasmid encoding a control sequence, a wild-type allele of EIIANtr1, a mutant allele having a histidine-to-alanine substitution at residue 66, or a mutant allele having a histidine-to-aspartate substitution at residue 66. Rescue of the biofilm phenotype of a ΔEIIANtr1 mutant was observed with both mutant alleles, as well as with the wild-type allele of EIIANtr1 (Fig. 11A). Furthermore, all alleles of EIIANtr1 were well expressed as full-length transcripts (Fig. 11B). This result suggests that phosphorylation of the conserved histidine at position 66 is not required for repression of biofilm accumulation by EIIANtr1. Based on this observation, we also predicted that the upstream components of the PTSNtr would not be required for repression of biofilm accumulation by EIIANtr1. To test this hypothesis, we constructed strains having in-frame deletions in EINtr and NPr and compared the biofilm accumulation by these strains to that by wild-type V. cholerae. As expected, we found that biofilm accumulation by ΔEINtr and ΔNPr mutants was not significantly different from that by wild-type V. cholerae (Fig. 11C).
Because biofilm formation is also repressed by EI, we asked whether EI and EIIANtr1 function independently. As shown in Fig. 11D, return of a wild-type EI allele to a ΔEI ΔEIIANtr1 mutant partially rescued the increased biofilm phenotype, suggesting that EI regulates biofilm formation independent of EIIANtr1. Return of a wild-type EIIANtr1 allele did not reduce biofilm formation by a ΔEI ΔEIIANtr1 mutant. This suggests that the ΔEI mutation is dominant in the ΔEI ΔEIIANtr1 mutant. While we have not ruled out a model in which one function of EI is to act downstream of EIIANtr1 in a common signaling pathway, we maintain that pathway 1 is independent of pathway 3 because it is present in minimal medium (16), while pathway 3 is not.
We then asked whether phosphorylation of the conserved histidine at position 172 was required for repression of biofilm accumulation by EIIANtr2. Biofilm accumulation was measured for ΔEIIANtr2 mutants harboring plasmids containing either a control sequence, a wild-type allele of EIIANtr2, an allele of EIIANtr2 resulting in substitution of the conserved histidine for alanine at position 172, or an allele of EIIANtr2 resulting in substitution of the conserved histidine for aspartate at position 172. As shown in Fig. 12A, the biofilm phenotype of a ΔEIIANtr2 mutant was rescued by the wild-type EIIANtr2 allele, as well as by both nonphosphorylatable mutant alleles. Furthermore, all rescue constructs were expressed as full-length proteins (Fig. 12B).
In other systems, EIIANtr has been shown to operate at the transcriptional level (25). We used β-galactosidase assays to measure the effects of the EIIANtr homologs on vpsL gene transcription. As shown in Fig. Fig.13,13, while both EIIANtr1 and EIIANtr2 repress biofilm formation at the transcriptional level, EIIANtr1, which is more similar to the EIIANtr of E. coli, has a more pronounced effect.
Lastly, we examined whether EIIANtr1 and EIIANtr2 have redundant roles in repression of biofilm formation by attempting to rescue the biofilm phenotype of a ΔEIIANtr1 mutant with a wild-type EIIANtr2 allele and vice versa. A wild-type EIIANtr1 allele provided in trans did rescue the biofilm phenotype of the ΔEIIANtr2 mutant, supporting our conclusion that these two EIIANtr homologs function in the same biofilm regulatory pathway (Fig. 12A). EIIANtr2 provided in trans was not able to rescue the ΔEIIANtr1 mutant phenotype (Fig. 11A). We attribute the inability of EIIANtr2 to rescue the phenotype of the ΔEIIANtr1 mutant to its smaller effect on vps gene transcription and biofilm formation.
We previously described a PTS-dependent biofilm regulatory pathway present in minimal medium. This pathway requires phosphotransfer from EI to HPr or FPr for action (pathway 1) (16, 17). Here, we examined regulation of biofilm formation by the PTS in LB broth, a complex medium. Our studies established the action of pathway 1 in V. cholerae biofilms formed in LB broth and also identified two novel PTS-dependent pathways, one of which (pathway 2) leads to activation of biofilm formation by EIIAGlc and one of which (pathway 3) results in repression of biofilm formation by EIIANtr (Fig. (Fig.14).14). All three of these pathways regulate transcription of the vps genes, which encode the biofilm matrix biosynthetic machinery. The regulatory cascade comprising pathway 2 suggests that there are differences from the homologous PTS-dependent regulatory pathway in E. coli, while pathway 3 defines a new role for EIIANtr in regulation of bacterial biofilm formation.
In minimal medium, pathway 1 proceeds through EI-P to HPr-P and FPr-P, which have redundant functions. As a result, only a ΔHPr ΔFPr mutant forms a biofilm that is distinguishable from that formed by wild-type V. cholerae. In contrast, in LB broth, HPr is the dominant protein in the signal transduction cascade, resulting in increased biofilm accumulation by a ΔHPr mutant compared with wild-type V. cholerae. This is not due to increased transcription of HPr in LB broth compared with minimal medium (Fig. (Fig.6).6). Rather, we hypothesize that phosphotransfer through HPr is preferred in LB broth. Alternatively, if pathway 1 remains divergent downstream of HPr and FPr, subsequent components may be responsible for the differences observed in LB broth.
Signaling through pathway 2 results in activation of vps gene transcription. This activation requires EIIAGlc and is observed in the absence of phosphorylation of the conserved histidine residue at position 91. In E. coli, EIIAGlc has two important regulatory functions. EIIAGlc-P potentiates the action of adenylate cyclase to increase intracellular concentrations of cyclic AMP (cAMP), a cofactor of the global regulator, the cAMP receptor protein (CRP) (56). The unphosphorylated form of EIIAGlc directly blocks transport of non-PTS substrates (52). Here we discovered a phosphorylation-independent regulatory role for EIIAGlc in activation of the vps genes and biofilm formation in V. cholerae. It is unlikely that the regulation is due to activation of adenylate cyclase as multiple studies have shown that cAMP represses rather than activates vps transcription and biofilm formation in V. cholerae (11, 17, 28).
Regulation of biofilm formation by EIIAGlc does not require phosphorylation of the conserved histidine at position 91. Therefore, the mechanism by which this function of EIIAGlc is controlled in response to environmental cues is unclear. Because transcription of V. cholerae EIIAGlc is 6-fold greater in LB broth than in minimal medium and 10-fold greater in response to addition of a PTS substrate to minimal medium (17), one possibility is that the activity of EIIAGlc is regulated by its intracellular abundance.
The results of epistasis experiments and measurements of vpsL transcription are consistent with a model in which Mlc activates biofilm accumulation through EIIAGlc-dependent and EIIAGlc-independent pathways. EIIBCGlc, in turn, inhibits the action of Mlc. Because Mlc is a transcription factor, we hypothesize that it acts at the transcriptional level to increase EIIAGlc activity either directly or indirectly. E. coli Mlc functions as a transcriptional repressor of many PTS components (47, 59). However, it does not regulate transcription of EIIAGlc. E. coli EIIBCGlc, an integral membrane protein, blocks the action of Mlc by sequestering it at the inner membrane (27, 58). We hypothesize that a similar interaction between Mlc and EIIBCGlc may block activation of EIIAGlc by Mlc (pathway 2a) (Fig. (Fig.14).14). Our data also suggest that a small portion of the observed activation of vps gene transcription by Mlc may proceed through an EIIAGlc-independent pathway (pathway 2b) (Fig. (Fig.1414).
We identified two EIIANtr homologs that repress V. cholerae biofilm formation and vps gene transcription in LB broth but not in minimal medium (pathway 3) (Fig. (Fig.14).14). Transcription of the EIIANtr homologs in these two media is similar. Therefore, this regulatory pathway is not controlled at the transcriptional level. The mechanism by which this pathway is activated remains to be identified.
We have not ruled out the possibility that there is an interaction between pathways 2 and 3. However, mutation of EIIANtr1 or EIIANtr2 in a ΔEIIAGlc background activates biofilm formation (data not shown), suggesting that EIIANtr1 and EIIANtr2 act independent of EIIAGlc.
While EIIANtr homologs have not been associated with biofilm formation in other organisms, they have been implicated in regulation of diverse metabolic functions, some of which have a clear role in regulating the balance of carbon and nitrogen in the cell. Two examples of this are repression of polyhydroxyalkanoate synthesis in Pseudomonas putida and repression of β-hydroxybutyrate synthesis in Azotobacter vinelandii by EIIANtr homologs (44, 62). Both of these carbohydrate polymers are intracellular carbon storage compounds that are accumulated when the environmental carbohydrate supply exceeds the bacterial requirement. VPS, the extracellular polysaccharide which is a component of the V. cholerae biofilm matrix, may also play a role in carbohydrate balance. VPS synthesis is activated when PTS substrates are abundant and is repressed when PTS substrates become scarce (17). Thus, in repressing VPS synthesis, V. cholerae EIIANtr may have a function parallel to that of its homologs in P. putida and A. vinelandii.
The mechanism by which EIIANtr represses polysaccharide synthesis in P. putida and A. vinelandii has not been discovered yet. However, the mechanism of action of EIIANtr in transcriptional derepression of the ilvBN operon, which encodes acetohydroxy acid synthase I, a common enzyme in biosynthesis of branched-chain amino acids, has been established (25). By an unknown mechanism, transcription of this operon is inhibited by high intracellular levels of K+. The unphosphorylated form of EIIANtr inhibits the action of the K+ transporter TrkA, which keeps intracellular K+ levels low and preserves transcription of the ilvBN operon (24). In our experiments, phosphorylation of neither EIIANtr1 nor EIIANtr2 was required for activation of vpsL gene transcription. One possibility is that regulation of gene transcription by EIIANtr in V. cholerae is governed by intracellular K+ levels in a phosphorylation state-independent manner. Alternatively, EIIANtr may be involved in a regulatory interaction with a different protein.
There are several examples of V. cholerae functions that are governed by complex regulatory networks with multiple inputs. These functions include production of virulence factors (34), quorum sensing (37), chemotaxis, and regulation of biofilm formation by cyclic-di-GMP (4, 5, 29, 57). In each case, the mechanism by which various environmental signals are integrated and result in the observed phenotype is unknown. Regulation of biofilm formation in V. cholerae by the PTS also appears to be quite complex. Immediate goals include identification of the environmental signals to which pathways 2 and 3 respond, identification of downstream components of these pathways, and integration of the three pathways in regulation of biofilm formation by the PTS. In particular, it seems likely that these pathways eventually interface with members of the large family of V. cholerae proteins that modulate intracellular levels of c-di-GMP and whose activators remain largely unidentified.
The V. cholerae PTS participates in colonization of both environmental surfaces and the mammalian intestine (16). We previously demonstrated that the influence of the PTS on VPS-dependent biofilm formation does not play a role in colonization of the mammalian intestine. The regulatory pathways outlined here are hypothesized to participate in colonization of the environment. In particular, a close environmental relationship between V. cholerae and zooplankton has been demonstrated. V. cholerae is thought to degrade the chitinaceous exoskeletons of these organisms to N-acetylglucosamine, a sugar transported exclusively by the PTS (16). Therefore, accumulation of V. cholerae on the exoskeletons of zooplankton is likely to be regulated by these PTS-dependent pathways. We envision that after formation of a monolayer on a surface, augmentation of the biofilm through cell division coupled with VPS synthesis is finely tuned to the nutritive potential of the surface itself and the environment in which the surface occurs. The signaling pathways outlined here contribute to bacterial sensing of surfaces and their environments and highlight the importance of PTS substrates in this process.
We thank Joachim Reidl for generously providing plasmids for mutant construction.
This work was supported by NIH grant AI50082 to P.I.W.
Published ahead of print on 16 April 2010.