The Phosphoenolpyruvate Phosphotransferase System Regulates Vibrio cholerae Biofilm Formation through Multiple Independent Pathways

Abstract

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 EI^Ntr, the HPr homolog NPr, and the EIIA homolog EIIA^Ntr. 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 (EIIA^Glc) and two nitrogen-specific EIIA homologs, EIIA^Ntr1 and EIIA^Ntr2. 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 PTS^Ntr (22, 49, 50). The components of this phosphotransfer cascade include EI^Ntr, NPr, and EIIA^Ntr. Unlike the carbohydrate-transporting PTS, PTS^Ntr 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 PTS^Ntr have not been defined, transfer of phosphate between the two PTS cascades may be one mechanism by which the PTS^Ntr 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. 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.

FIG. 1.

Schematic diagram of the 25 V. cholerae PTS components. Homologs of the EI, HPr, and EII proteins are indicated by the same colors. Furthermore, proteins belonging to the same EIIA superfamily are color coded (blue, glucose superfamily; violet, mannitol-fructose superfamily; yellow, lactose superfamily). Sugar specificities are indicated for PTS components if they are known (16). Parentheses indicate presumptive specificities. The VC1824 HPr-like domain is surrounded by a dashed line to indicate its low level of similarity to HPr. The sugar specificity of VC1281 is based on previously described data (3). NAG, N-acetylglucosamine.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are listed in Table 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.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Genotype and/or phenotype	Reference
E. coli SM10λpir	thi thr leu tonA lacY supE recA::RP4-2-Tc::Mul pir R6K; Kmr	38
V. cholerae strains
PW357	MO10 lacZ::vpsLp→lacZ; Smr	15
PW751	MO10 lacZ::vpsLp→lacZ ΔPTS; Smr	17
PW961	MO10 lacZ::vpsLp→lacZ ΔEI; Smr	17
PW964	MO10 lacZ::vpsLp→lacZ ΔHPr; Smr	17
PW954	MO10 lacZ::vpsLp→lacZ ΔFPr; Smr	16
PW965	MO10 lacZ::vpsLp→lacZ ΔHPr ΔFPr; Smr	16
PW836	MO10 lacZ::vpsLp→lacZ ΔEIIAGlc; Smr	17
PW989	MO10 lacZ::vpsLp ΔEIIBCGlc; Smr	17
PW988	MO10 lacZ::vpsLp ΔEIIAGlc ΔEIIBCGlc; Smr	This study
PW996	MO10 lacZ::vpsLp ΔEIIBCGlc ΔMlc; Smr	This study
PW994	MO10 lacZ::vpsLp ΔEIIAGlc ΔEIIBCGlc ΔMlc; Smr	This study
PW993	MO10 lacZ::vpsLp ΔMlc	This study
PW992	MO10 lacZ::vpsLp→lacZ ΔEIIANtr1; Smr	16
PW991	MO10 lacZ::vpsLp→lacZ ΔEIIANtr2; Smr	16
PW1002	MO10 lacZ::vpsLp→lacZ VC1283::pGP704; Smr Apr	16
PW1003	MO10 lacZ::vpsLp→lacZ VC1820::pGP704; Smr Apr	16
PW1004	MO10 lacZ::vpsLp→lacZ VC1822::pGP704; Smr Apr	16
PW1005	MO10 lacZ::vpsLp→lacZ VC1826::pGP704; Smr Apr	16
PW1006	MO10 lacZ::vpsLp→lacZ VCA0245::pGP704; Smr Apr	16
PW1007	MO10 lacZ::vpsLp→lacZ VCA1045::pGP704; Smr Apr	16
Plasmids used in rescue experiments
pBAD-TOPO-EI	pBAD-TOPO carrying the coding sequence of VC0965	17
pBAD-TOPO-EI(H189A)	pBAD-TOPO carrying the coding sequence of VC0965 with an H-to-A mutation at position 189	17
pBAD-TOPO-EIIAGlc	pBAD-TOPO carrying the coding sequence of VC0964	17
pBAD-TOPO-EIIAGlc(H91A)	pBAD-TOPO carrying the coding sequence of VC0964 with an H-to-A mutation at position 91	This study
pBAD-TOPO-EIIAGlc(H91D)	pBAD-TOPO carrying the coding sequence of VC0964 with an H-to-D mutation at position 91	This study
pBAD-TOPO-mlc	pBAD-TOPO carrying the coding sequence of VC2007	This study
pBAD-TOPO-EIIANtr1	pBAD-TOPO carrying the coding sequence of VC2531	This study
pBAD-TOPO-EIIANtr1(H66A)	pBAD-TOPO carrying the coding sequence of VC2531 with an H-to-A mutation at position 66	This study
pBAD-TOPO-EIIANtr1(H66D)	pBAD-TOPO carrying the coding sequence of VC2531 with an H-to-D mutation at position 66	This study
pBAD-TOPO-EIIANtr2	pBAD-TOPO carrying the coding sequence of VC1824	This study
pBAD-TOPO-EIIANtr2(H172A)	pBAD-TOPO carrying the coding sequence of VC1824 with an H-to-A mutation at position 172	This study
pBAD-TOPO-EIIANtr2(H172D)	pBAD-TOPO carrying the coding sequence of VC1824 with an H-to-D mutation at position 172	This study

Construction of mutants.

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.

Construction of rescue plasmids.

Plasmids used for rescue experiments, which are listed in Table 1, were constructed as previously described (17, 61) using the primers listed in Table 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 EIIA^Glc(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).

TABLE 2.

Primers used in this study

Protein	Primer sequences
Primers used for construction of expression plasmids
Primers for wild-type alleles
Mlc	TAAGAGGAATAATAAATGTACATGGCTCAA CCCGG, TCCTTCGACCACTTTCATTAA GAG
PtsN	CAATTGAGCGAAGTACTTTCATTGGACTG, TTGGTTAACCATGATGTTGTACAGC
VC1824	GCGTGACATCGCGTACATCAG, CATGAATTCCAAGTCACCTTCTTGG
Internal primers for construction of mutant alleles
EIIAGlc(H91A)	GCGTTGAGCTGTTTGTTGCCTTCGGTATCG, CGATACCGAAGGCAACAAACAGCTCAACGC
EIIAGlc(H91D)	GCGTTGAGCTGTTTGTTGACTTCGGTATCG, CGATACCGAAGTCAACAAACAGCTCAACGC
PtsN(H66A)	ATAGCCATCCCAGCCGCACGCATGT, ACATGCGTGCGGCTGGGATGGCTAT
PtsN(H66D)	ATAGCCATCCCAGACGCACGCATGT, ACATGCGTGCGTCTGGGATGGCTAT
VC1824(H172A)	GGGATTGCTATTCCCGCTGTGATGTTTGC, GCAAACATCACAGCGGGAATAGCAATCCC
VC1824(H172D)	GGGATTGCTATTCCCGATGTGATGTTTGC, GCAAACATCACATCGGGAATAGCAATCCC
Primers used in quantitative RT-PCR
EI	ACCGTTATCGAAGAGCAAGCCACT, TCTGCGCAGTTTCAGAAGGCGTTA
EIIAGlc	GCGATCAAGCCTGCTGGTAACAAA, AAGCCTTCACCTTTCAGCTCAACC
HPr	TGTTCAAGCTGCAAACGCTAGGTC, GCAACTAGGTGCTCAACAGCTTCT
EIIANtr1	GCTATTGCAGTGTGAAGAGCCGAT, CTGCGTTACGAAGCTGTTTGAGGA
EIIANtr2	TCGGTATTGATGCTGAACTCGCCT, GGGCTTTGGCATAGTGCCATTTGA
ClpX	AGAGTTCATTGGTCGTCTGCCTGT, AACAACGCCGCATACTGTTTGGTC

Quantitative analysis of total growth and biofilm formation.

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.

Quantification of PTS component transcripts.

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 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.

Measurement of vpsL transcription using a β-galactosidase reporter.

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.

RESULTS

The biofilm phenotypes of ΔEI and ΔEIIA^Glc mutants in LB broth are additive.

We first compared biofilm formation in LB broth and minimal medium supplemented with glucose by a ΔEI mutant, a ΔEIIA^Glc mutant, and a ΔPTS mutant lacking the genes encoding EI, HPr, and EIIA^Glc (Fig. 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 ΔEIIA^Glc mutant exhibited decreased surface accumulation. As previously noted (17), surface accumulation by a ΔPTS mutant lacking EI, HPr, and EIIA^Glc 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.

FIG. 2.

PTS mutant displays distinct biofilm phenotypes in LB broth and minimal medium (MM) supplemented with glucose. Total growth and biofilm accumulation were determined for wild-type V. cholerae (WT), a ΔEI mutant, a ΔEIIA^Glc mutant, and a ΔPTS mutant in minimal medium supplemented with 0.5% (wt/vol) glucose (MM + glucose) and LB broth (LB). The values are the averages of four experimental replicates, and the error bars indicate standard deviations. There was a statistically significant difference (P = 0.0062)between biofilm formation by wild-type V. cholerae and biofilm formation by the ΔPTS mutant in minimal medium supplemented with glucose (indicated by an asterisk) but not in LB broth (P = 0.116).

Evidence for repression of biofilm accumulation by EI-P and HPr-P in LB broth (regulatory pathway 1).

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. 3A, while the biofilm phenotype of a ΔEI mutant was rescued by a wild-type EI allele provided in trans, the EI^H189A allele was unable to rescue the biofilm phenotype despite adequate expression of the protein (Fig. 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. 3C).

FIG. 3.

EI-P represses biofilm formation and vps gene transcription in LB broth (biofilm regulatory pathway 1). (A) Total growth and biofilm accumulation in LB broth by wild-type V. cholerae (WT) or a ΔEI mutant rescued with either a control pBAD plasmid (pCTL), a pBAD plasmid expressing a wild-type EI allele (pEI), or a pBAD plasmid expressing an EI allele encoding an H-to-A mutation at position 189 (pEI^H189A). Protein expression was induced by addition of 0.04% l-arabinose. The values are the averages of four experimental replicates, and the error bars indicate standard deviations. Biofilm formation by the ΔEI mutant rescued with a wild-type EI allele was not statistically different from biofilm formation by wild-type V. cholerae (P = 0.290), while biofilm formation by a ΔEI mutant rescued with the EI^H189A allele was statistically different (P = 0.0007). (B) Western blot analysis demonstrating that the wild-type EI and EI^H189A alleles are well expressed in the ΔEI genetic background. (C) Quantification of total growth and biofilm formation over time for wild-type V. cholerae and a ΔEI mutant. The values are the means of two experimental replicates. The ΔEI mutant demonstrated more biofilm accumulation than wild-type V. cholerae throughout the course of the experiment.

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. 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. 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. 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. 14).

FIG. 4.

EI and HPr are in the same biofilm regulatory pathway. (A) Total growth and biofilm accumulation by wild-type V. cholerae (WT), a ΔHPr mutant, a ΔFPr mutant, and a ΔHPr ΔFPr double mutant. Biofilm accumulation by a ΔHPr mutant is significantly different from that by wild-type V. cholerae (P = 0.002), while biofilm accumulation by a ΔHPr ΔFPr is not significantly different from that by ΔHPr (P = 0.758). (B) Total growth and biofilm accumulation by wild-type V. cholerae (WT), as well as ΔEI and ΔHPr ΔFPr mutants rescued either with a control pBAD plasmid (pCTL) or a pBAD plasmid expressing a wild-type EI allele (pEI). Protein expression was induced by addition of 0.04% l-arabinose. Deletion of HPr and FPr blocked the ability of EI to repress biofilm formation. (C) Measurement of vps gene transcription in wild-type V. cholerae and various PTS mutants harboring a chromosomal vpsL-lacZ fusion. vpsL transcription in the ΔEI (P = 0.0002), ΔHPr (P < 0.0001), and ΔHPr ΔFPr (P = 0.0001) mutants is significantly different from that in wild-type V. cholerae. The error bars indicate standard deviations. RLU, relative light units.

FIG. 14.

Schematic diagram of three pathways in the PTS network that regulate biofilm formation. Pathway 1 requires phosphorylation of the PTS components EI and HPr or FPr to repress transcription of the vps genes. Pathway 2a is comprised of the components EIIBC^Glc, Mlc, and EIIA^Glc. EIIA^Glc activates vps transcription in a phosphorylation-independent manner. Mlc potentiates the action of EIIA^Glc but may also activate vps gene transcription through an EIIA^Glc-independent route (pathway 2b). EIIBC^Glc interferes with the action of Mlc. Pathway 3 includes EIIA^Ntr1and EIIA^Ntr2, whose functions display some redundancy in repression of vps transcription.

Three EIIA homologs regulate biofilm accumulation.

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. 1). As shown in Fig. 5, we identified three EIIA mutants with altered biofilm phenotypes compared with wild-type V. cholerae. First, approximately 3-fold fewer ΔEIIA^Glc 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 (EIIA^Ntr1) and VC1824 (EIIA^Ntr2) both contain EIIA domains homologous to EIIA^Ntr, 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. 1). Our observations suggested that at least two independent PTS pathways control biofilm formation at the level of EIIA in LB broth.

FIG. 5.

EIIA homologs EIIA^Glc, EIIA^Ntr1, and EIIA^Ntr2 regulate biofilm formation: quantification of total growth and biofilm accumulation by wild-type V. cholerae, as well as strains having mutations in each of the 10 genes encoding EIIA domains. The biofilm accumulation by the ΔEIIA^Glc (P < 0.0001), ΔEIIA^Ntr (P < 0.0001), and ΔVC1824 (P = 0.0004) mutants was significantly different from that by wild-type V. cholerae (WT). The error bars indicate standard deviations.

Because independent control of biofilm formation by an EIIA component was not observed in minimal medium (16), we hypothesized that transcription of EIIA^Glc, EIIA^Ntr1, and EIIA^Ntr2 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, EIIA^Glc, EIIA^Ntr1, and EIIA^Ntr2 in LB broth and minimal medium supplemented with pyruvate (Fig. 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 EIIA^Glc was approximately 6-fold greater in LB broth, suggesting that activation of biofilm formation by EIIA^Glc in LB broth may be due at least in part to increased transcription of EIIA^Glc. Transcription of the PTS components EIIA^Ntr1 and EIIA^Ntr2 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.

FIG. 6.

Differential transcription of PTS components is not solely responsible for the patterns of biofilm regulation observed in LB broth compared with those observed in minimal medium. The EI, HPr, FPr, EIIA^Glc, EIIA^Ntr1, and ΔEIIA^Ntr2 transcript levels in wild-type V. cholerae grown in LB broth were compared to the transcript levels in MM supplemented with pyruvate. The data were analyzed in triplicate using the ΔΔC_t method. clpX was used as a standard. The error bars indicate standard deviations.

Activation of biofilm formation by EIIA^Glc does not require phosphorylation of the conserved histidine at position 91 (regulatory pathway 2).

Phosphotransfer through EI and HPr is predicted to lead to EIIA^Glc phosphorylation at histidine 91. To determine whether regulation of biofilm formation by EIIA^Glc requires phosphorylation of this residue, we studied biofilm accumulation by a ΔEIIA^Glc mutant containing an expression plasmid carrying either a control sequence, a wild-type EIIA^Glc allele, an EIIA^Glc allele with a histidine-to-alanine mutation at position 91, or an EIIA^Glc allele with a histidine-to-aspartate mutation at position 91. As shown in Fig. 7A, while these EIIA^Glc alleles did not affect the total growth, all of them activated biofilm accumulation in the ΔEIIA^Glc 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 EIIA^Glc alleles in a ΔPTS mutant background, which lacks EI, HPr, and EIIA^Glc (Fig. 7B). If EIIA^Glc were dependent on EI or HPr for activation of biofilm formation, an EIIA^Glc 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 EIIA^Glc 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 ΔEIIA^Glc mutant background (Fig. 7C). Therefore, we concluded that activation of biofilm accumulation by EIIA^Glc 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).

FIG. 7.

Activation of biofilm formation by EIIA^Glc does not require phosphorylation and is independent of EI and HPr (biofilm regulatory pathway 2). (A) Quantification of total growth and biofilm accumulation by a ΔEIIA^Glc mutant rescued with a pBAD plasmid carrying a control sequence (pCTL) and pBAD plasmids with a wild-type EIIA^Glc allele (pEIIA^Glc) and a variety of mutant EIIA^Glc alleles. Protein expression was induced with 0.04% arabinose. Biofilm accumulation by the ΔEIIA^Glc mutant rescued with a wild-type EIIA^Glc glucose allele (P = 0.0024), an EIIA^Glc allele encoding a histidine-to-alanine substitution at position 191 (pEIIA^GlcH91A) (P < 0.0001), or an EIIA^Glc allele encoding a histidine-to-aspartate substitution at position 191 (pEIIA^GlcH91D) (P < 0.0001) was significantly increased compared with biofilm accumulation by the ΔEIIA^Glc mutant rescued with a control sequence. (B) Quantification of total growth and biofilm accumulation by a ΔPTS mutant (lacking EI, HPr, and EIIA^Glc) rescued with a pBAD plasmid carrying a control sequence (pCTL), a wild-type EIIA^Glc allele (pEIIA^Glc), an EIIA^Glc allele encoding a histidine-to-alanine substitution at position 191 (pEIIA^GlcH91A), or an EIIA^Glc allele encoding a histidine-to-aspartate substitution at position 191 (pEIIA^GlcH91D). Protein expression was induced with 0.04% arabinose. Biofilm accumulation by the ΔPTS mutant rescued with any of the EIIA^Glc alleles was significantly increased compared with biofilm accumulation by the ΔPTS mutant rescued with a control sequence (P ≤ 0.0002 for all comparisons). The error bars indicate standard deviations. (C) Western blot demonstrating that the wild-type and mutant EIIA^Glc alleles used in these experiments are well expressed and produce full-length proteins.

Evidence that EIIBC^Glc and the transcription factor Mlc participate in regulatory pathway 2.

Several proteins are known to interact with EIIA^Glc. Two of these proteins are EIIBC^Glc and Mlc. EIIBC^Glc is downstream of EIIA^Glc in the glucose-specific PTS phosphotransfer cascade. Furthermore, in Escherichia coli, the DNA-binding protein Mlc has been shown to repress transcription of EIIBC^Glc when PTS substrates are scarce (21). Unphosphorylated EIIBC^Glc interacts directly with Mlc to relieve this repression (27, 43). The V. cholerae Mlc and EIIBC^Glc proteins are very similar to the corresponding proteins of E. coli. We therefore asked whether EIIBC^Glc and Mlc might play a role in regulation of V. cholerae biofilm formation by EIIA^Glc. We first measured biofilm accumulation by wild-type V. cholerae, a ΔEIIBC^Glc mutant, a ΔMlc mutant, and a ΔEIIA^Glc mutant, as well as by strains with combinations of the mutations. As shown in Fig. 8, the total levels of growth were comparable for all of the strains studied. However, biofilm formation by the ΔMlc and ΔEIIA^Glc mutants was significantly decreased, while biofilm accumulation by the ΔEIIBC^Glc mutant was increased compared with wild-type V. cholerae biofilm accumulation. Furthermore, deletion of either Mlc or EIIA^Glc in a ΔEIIBC^Glc mutant background was sufficient to decrease the biofilm accumulation almost to the levels observed for the ΔEIIA^Glc mutant. This led us to hypothesize that the biofilm regulatory pathway shown in Fig. 9A is present. In this signal transduction pathway, EIIBC^Glc inhibits activation of EIIA^Glc by Mlc. EIIA^Glc, in turn, activates transcription of the vps genes by an as-yet-unidentified mechanism.

FIG. 8.

Evidence for involvement of Mlc and EIIBC^Glc in pathway 2: quantification of total growth and biofilm accumulation by wild-type V. cholerae and various pathway 2 mutants. Biofilm accumulation by the ΔEI (P = 0.0008) or ΔEIIBC^Glc (P = 0.0001) mutant was significantly greater than that by wild-type V. cholerae (WT). Biofilm accumulation by the ΔEIIA^Glc or ΔMlc mutants was significantly less than that by wild-type V. cholerae (P < 0.0001). Biofilm accumulation by the ΔEIIA^Glc ΔEIIBC^Glc mutant (P = 0.04) or the ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutant (P = 0.001) was significantly greater than that by the ΔEIIA^Glc mutant. Biofilm accumulation by the ΔEIIBC^Glc ΔMlc mutant (P = 0.0006) or the ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutant (P = 0.0005) was significantly greater than that by the ΔMlc mutant. The error bars indicate standard deviations.

FIG. 9.

Pathway 2 regulates biofilm accumulation at the transcriptional level. (A) Schematic diagram of pathways 2a and 2b. In this model, Mlc activates vps gene transcription through both EIIA^Glc-dependent and EIIA^Glc-independent pathways. EIIBC^Glc interferes with the action of Mlc. The numbers in circles indicate the predicted relative effect of each component in the regulatory cascade on vps gene transcription. (B) All the strains tested have vpsL-lacZ promoter fusions in the V. cholerae lacZ locus. The measurements of β-galactosidase activity reflect vpsL transcription. The measurements for the ΔEIIA^Glc (P < 0.0001), ΔEIIBC^Glc (P = 0.04), and ΔMlc (P = 0.0154) mutants were significantly different from that for wild-type V. cholerae. The vpsL transcription in the ΔEIIBC^Glc ΔMlc mutant was significantly different from that in the ΔEIIBC^Glc mutant (P = 0.0003) but not from that in the ΔMlc mutant (P = 0.766). The vps transcription in the ΔEIIA^Glc ΔEIIBC^Glc mutant was significantly different from that in the ΔEIIA^Glc mutant (P = 0.032) or the ΔEIIBC^Glc mutant (P < 0.0001). The vps transcription in the ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutant was significantly different both from that in the ΔEIIA^Glc ΔEIIBC^Glc mutant (P = 0.0086) or the ΔEIIBC^Glc ΔMlc mutant (P < 0.0001) but not from that in the ΔEIIA^Glc mutant (P = 0.415). The models on the right show the predicted impact of each genetic background on vps gene transcription. The relative contributions of the regulatory components to vps gene transcription as shown in panel A were added to obtain the predicted impact of the genetic background on vps gene transcription. WT, wild type; RLU, relative light units.

To further evaluate the presence of this signal transduction pathway, we measured vpsL transcription in wild-type V. cholerae and ΔEIIBC^Glc, ΔMlc, and ΔEIIA^Glc mutants, as well as in strains with combinations of these mutations. The results of this experiment are shown in Fig. 9B 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 ΔEIIA^Glc and is repressed by EIIBC^Glc. Furthermore, the level of vps transcription in a ΔEIIA^Glc ΔEIIBC^Glc mutant is higher than that in a ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutant, suggesting that Mlc activates vps transcription even in the absence of EIIA^Glc. Therefore, an additional pathway must be postulated, in which Mlc activates vpsL transcription independent of EIIA^Glc. These pathways are referred to in Fig. 9A 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 ΔEIIA^Glc and ΔMlc mutants. Second, biofilm formation by strains having mutations in EIIBC^Glc as well as in EIIA^Glc and/or Mlc all formed biofilms that were slightly but significantly and reproducibly larger than those formed by ΔEIIA^Glc and ΔMlc mutants. One possible explanation for these observations is that EIIBC^Glc also represses biofilm accumulation at the posttranscriptional level. The complex effect of EIIBC^Glc on biofilm formation is currently under investigation.

We hypothesized that if Mlc were partially dependent on EIIA^Glc for its effect on biofilm formation, the absence of EIIA^Glc 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, ΔEIIBC^Glc ΔMlc, and ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutants. As shown in Fig. 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 ΔEIIBC^Glc ΔMlc mutants increased biofilm accumulation approximately 4-fold. In contrast, introduction of the mlc allele into a ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutant increased biofilm accumulation only 2-fold. This result supports our model for pathway 2 in which Mlc activates biofilm accumulation through EIIA^Glc-dependent and -independent pathways.

FIG. 10.

Activation of biofilm accumulation by Mlc is partially dependent on EIIA^Glc: quantification of biofilm accumulation and total growth by wild-type V. cholerae (WT) and ΔMlc, ΔEIIBC^Glc ΔMlc, and ΔEIIA^Glc ΔEIIBC^Glc ΔMlc mutants rescued with either a control vector (pCTL) or a plasmid carrying a wild-type mlc allele (pMlc). Biofilm formation by ΔMlc (P < 0.0001), ΔEIIBC^Glc ΔMlc (P < 0.0001), and ΔEIIA^Glc ΔEIIBC^Glc ΔMlc (P = 0.002) mutants rescued with the plasmid carrying the wild-type mlc allele was significantly greater than that by mutants carrying a control vector. The error bars indicate standard deviations.

EIIA^Ntr1 and EIIA^Ntr2 repress biofilm accumulation (regulatory pathway 3).

The gene encoding EIIA^Ntr, which is located close to rpoN, has been found in many bacterial genomes (36, 48, 51). Mutation of the two V. cholerae EIIA^Ntr homologs, EIIA^Ntr1 (60% identity and 75% similarity to E. coli EIIA^Ntr) and EIIA^Ntr2 (26% identity and 42% similarity to E. coli EIIA^Ntr), resulted in increased biofilm accumulation (Fig. 5). The conserved histidine residues at position 66 of EIIA^Ntr1 and position 172 of EIIA^Ntr2 are predicted to be phosphorylated by phosphotransfer from the HPr homolog NPr, which in turn receives a phosphate from EI^Ntr. We first determined whether phosphorylation of the conserved histidine was required for the function of EIIA^Ntr1 by comparing biofilm accumulation data for ΔEIIA^Ntr1 mutants carrying either a plasmid encoding a control sequence, a wild-type allele of EIIA^Ntr1, 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 ΔEIIA^Ntr1 mutant was observed with both mutant alleles, as well as with the wild-type allele of EIIA^Ntr1 (Fig. 11A). Furthermore, all alleles of EIIA^Ntr1 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 EIIA^Ntr1. Based on this observation, we also predicted that the upstream components of the PTS^Ntr would not be required for repression of biofilm accumulation by EIIA^Ntr1. To test this hypothesis, we constructed strains having in-frame deletions in EI^Ntr 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 ΔEI^Ntr and ΔNPr mutants was not significantly different from that by wild-type V. cholerae (Fig. 11C).

FIG. 11.

Repression of biofilm formation by EIIA^Ntr1 does not require phosphorylation of the conserved histidine at residue 66 and is independent of pathway 1 (biofilm regulatory pathway 3). (A) Quantification of total growth and biofilm accumulation by wild-type V. cholerae or a ΔEIIA^Ntr1 mutant rescued with a control plasmid (pCTL), a plasmid carrying a wild-type EIIA^Ntr1 allele (pEIIA^Ntr1), a plasmid carrying an EIIA^Ntr1 allele resulting in an alanine substitution for the histidine at position 66 (pEIIA^Ntr1H66A), a plasmid carrying an EIIA^Ntr1 allele resulting in an aspartate substitution for the histidine at position 66 (pEIIA^Ntr1H66D), and a plasmid carrying a wild-type EIIA^Ntr2 allele (pEIIA^Ntr2). Only biofilm accumulation by the ΔEIIA^Ntr1 (pCTL) (P = 0.0003) and ΔEIIA^Ntr1 (pEIIA^Ntr2) (P = 0.0004) mutants was significantly different from that by wild-type V. cholerae (WT). The error bars indicate standard deviations. (B) Western blot demonstrating that the wild-type and mutant EIIA^Ntr1 alleles used in these experiments are well expressed and produce full-length proteins. (C) Quantification of total growth and biofilm accumulation by wild-type V. cholerae and strains having mutations in the other two components of the PTS^Ntr, EI^Ntr, and NPr. Biofilm accumulation by the two mutants was not significantly different from that by wild-type V. cholerae. (D) Quantification of total growth and biofilm formation by a ΔEI mutant or a ΔEI ΔEIIA^Ntr1 mutant rescued with a pBAD plasmid carrying a control sequence (pCTL), a wild-type EI allele (pEI), or a wild-type EIIA^Ntr1 allele (pEIIA^Ntr1). Biofilm accumulation by the ΔEI ΔEIIA^Ntr1 mutant rescued with a wild-type EI allele was significantly different from that by the ΔEI ΔEIIA^Ntr1 mutant carrying a control sequence (P = 0.0102).

Because biofilm formation is also repressed by EI, we asked whether EI and EIIA^Ntr1 function independently. As shown in Fig. 11D, return of a wild-type EI allele to a ΔEI ΔEIIA^Ntr1 mutant partially rescued the increased biofilm phenotype, suggesting that EI regulates biofilm formation independent of EIIA^Ntr1. Return of a wild-type EIIA^Ntr1 allele did not reduce biofilm formation by a ΔEI ΔEIIA^Ntr1 mutant. This suggests that the ΔEI mutation is dominant in the ΔEI ΔEIIA^Ntr1 mutant. While we have not ruled out a model in which one function of EI is to act downstream of EIIA^Ntr1 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 EIIA^Ntr2. Biofilm accumulation was measured for ΔEIIA^Ntr2 mutants harboring plasmids containing either a control sequence, a wild-type allele of EIIA^Ntr2, an allele of EIIA^Ntr2 resulting in substitution of the conserved histidine for alanine at position 172, or an allele of EIIA^Ntr2 resulting in substitution of the conserved histidine for aspartate at position 172. As shown in Fig. 12A, the biofilm phenotype of a ΔEIIA^Ntr2 mutant was rescued by the wild-type EIIA^Ntr2 allele, as well as by both nonphosphorylatable mutant alleles. Furthermore, all rescue constructs were expressed as full-length proteins (Fig. 12B).

FIG. 12.

Repression of biofilm formation by EIIA^Ntr2 does not require phosphorylation at histidine 172 and is rescued by EIIA^Ntr1 (biofilm regulatory pathway 3) (A) Quantification of total growth and biofilm accumulation by wild-type V. cholerae or a ΔEIIA^Ntr2 mutant rescued with a pBAD plasmid carrying a control sequence (pCTL), a wild-type ΔEIIA^Ntr2 allele (pEIIA^Ntr2), a ΔEIIA^Ntr2 allele encoding an H-to-A point mutation at conserved residue 172 (pEIIA^Ntr2H172A), a ΔEIIA^Ntr2 allele encoding an H-to-D point mutation at conserved residue 172 (pEIIA^Ntr2H172D), or a wild-type EIIA^Ntr1 allele (pEIIA^Ntr1). Protein expression was induced with 0.04% arabinose. Biofilm accumulation by a ΔEIIA^Ntr2 mutant rescued with wild-type EIIA^Ntr2 (P = 0.263), EIIA^Ntr2H172A (P = 0.776), EIIA^Ntr2H172D (P = 0.126), or EIIA^Ntr1 (P = 0.514) was not significantly different from that by wild-type V. cholerae (WT). The error bars indicate standard deviations. (B) Western blot demonstrating that the wild-type and mutant EIIA^Ntr2 alleles used in these experiments are well expressed and produce full-length proteins.

In other systems, EIIA^Ntr has been shown to operate at the transcriptional level (25). We used β-galactosidase assays to measure the effects of the EIIA^Ntr homologs on vpsL gene transcription. As shown in Fig. 13, while both EIIA^Ntr1 and EIIA^Ntr2 repress biofilm formation at the transcriptional level, EIIA^Ntr1, which is more similar to the EIIA^Ntr of E. coli, has a more pronounced effect.

FIG. 13.

EIIA^Ntr1 and EIIA^Ntr2 operate at the level of transcription: measurements of vps gene transcription in wild-type V. cholerae, a ΔEIIA^Ntr1 mutant, and a ΔEIIA^Ntr2 mutant harboring a chromosomal vpsL-lacZ fusion. The vpsL transcription in the ΔEIIA^Ntr1 and ΔEIIA^Ntr2 mutants was significantly different from that in wild-type V. cholerae (WT) (P < 0.0001). The error bars indicate standard deviations. RLU, relative light units.

Lastly, we examined whether EIIA^Ntr1 and EIIA^Ntr2 have redundant roles in repression of biofilm formation by attempting to rescue the biofilm phenotype of a ΔEIIA^Ntr1 mutant with a wild-type EIIA^Ntr2 allele and vice versa. A wild-type EIIA^Ntr1 allele provided in trans did rescue the biofilm phenotype of the ΔEIIA^Ntr2 mutant, supporting our conclusion that these two EIIA^Ntr homologs function in the same biofilm regulatory pathway (Fig. 12A). EIIA^Ntr2 provided in trans was not able to rescue the ΔEIIA^Ntr1 mutant phenotype (Fig. 11A). We attribute the inability of EIIA^Ntr2 to rescue the phenotype of the ΔEIIA^Ntr1 mutant to its smaller effect on vps gene transcription and biofilm formation.

DISCUSSION

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 EIIA^Glc and one of which (pathway 3) results in repression of biofilm formation by EIIA^Ntr (Fig. 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 EIIA^Ntr 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. 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 EIIA^Glc and is observed in the absence of phosphorylation of the conserved histidine residue at position 91. In E. coli, EIIA^Glc has two important regulatory functions. EIIA^Glc-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 EIIA^Glc directly blocks transport of non-PTS substrates (52). Here we discovered a phosphorylation-independent regulatory role for EIIA^Glc 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 EIIA^Glc does not require phosphorylation of the conserved histidine at position 91. Therefore, the mechanism by which this function of EIIA^Glc is controlled in response to environmental cues is unclear. Because transcription of V. cholerae EIIA^Glc 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 EIIA^Glc 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 EIIA^Glc-dependent and EIIA^Glc-independent pathways. EIIBC^Glc, in turn, inhibits the action of Mlc. Because Mlc is a transcription factor, we hypothesize that it acts at the transcriptional level to increase EIIA^Glc 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 EIIA^Glc. E. coli EIIBC^Glc, 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 EIIBC^Glc may block activation of EIIA^Glc by Mlc (pathway 2a) (Fig. 14). Our data also suggest that a small portion of the observed activation of vps gene transcription by Mlc may proceed through an EIIA^Glc-independent pathway (pathway 2b) (Fig. 14).

We identified two EIIA^Ntr homologs that repress V. cholerae biofilm formation and vps gene transcription in LB broth but not in minimal medium (pathway 3) (Fig. 14). Transcription of the EIIA^Ntr 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 EIIA^Ntr1 or EIIA^Ntr2 in a ΔEIIA^Glc background activates biofilm formation (data not shown), suggesting that EIIA^Ntr1 and EIIA^Ntr2 act independent of EIIA^Glc.

While EIIA^Ntr 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 EIIA^Ntr 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 EIIA^Ntr may have a function parallel to that of its homologs in P. putida and A. vinelandii.

The mechanism by which EIIA^Ntr represses polysaccharide synthesis in P. putida and A. vinelandii has not been discovered yet. However, the mechanism of action of EIIA^Ntr 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 EIIA^Ntr 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 EIIA^Ntr1 nor EIIA^Ntr2 was required for activation of vpsL gene transcription. One possibility is that regulation of gene transcription by EIIA^Ntr in V. cholerae is governed by intracellular K⁺ levels in a phosphorylation state-independent manner. Alternatively, EIIA^Ntr 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.