Identification of novel stage-specific genetic requirements through whole genome transcription profiling of Vibrio cholerae biofilm development

Summary

Bacterial biofilm formation has been described as a developmental process. This process may be divided into three stages: the planktonic stage, the monolayer stage and the biofilm stage. Bacteria in the planktonic stage are not attached to each other or to a surface; bacteria in the monolayer stage are attached to surfaces as single cells; and bacteria in the biofilm stage are attached to surfaces as cellular aggregates. In a study limited to the Vibrio cholerae flaA, mshA and vps genes, we previously demonstrated that transcription in monolayer cells is distinct from that in biofilm cells and that the genetic requirements of monolayer formation are distinct from those of biofilm formation. In this work, we sought to identify additional stage-specific genetic requirements through microarray analysis of the V. cholerae transcriptome during biofilm development. These studies demonstrated unique patterns of transcription in the planktonic, monolayer and biofilm stages of biofilm development. Based on our microarray results, we selected cheY-3 as well as two previously uncharacterized genes, bap1 and leuO, for targeted mutation. The ΔcheY-3 mutant displayed a defect in monolayer but not biofilm formation, suggesting that chemotaxis plays a stage-specific role in formation of the V. cholerae monolayer. Mutants carrying deletions in bap1 and leuO formed monolayers that were indistinguishable from those formed by wild-type V. cholerae. In contrast, these mutants displayed greatly decreased biofilm accumulation. Our microarray analyses document modulation of the transcriptome of V. cholerae as it progresses through the stages in biofilm development. These studies demonstrate that microarray analysis of the transcriptome of biofilm development may greatly accelerate the discovery of novel targets for stage-specific inhibition of biofilm development.

Introduction

Most microbes in the natural environment live in surface-attached communities called biofilms (Costerton et al., 1995). Genetic and microscopic studies of biofilm formation by several Gram-positive and Gram-negative organisms have suggested that development of a mature, three-dimensional biofilm involves the following consecutive, discrete stages: the planktonic stage, the monolayer stage and, finally, the biofilm stage (O’Toole et al., 2000). Free-swimming planktonic cells encountering a surface become transiently attached to it. Permanent immobilization of these cells on the surface results in the formation of the monolayer. Induction of extracellular matrix biosynthesis by cells in the monolayer leads to the development of a multi-layered biofilm through the formation of intercellular contacts. Biofilm cells have been shown to be physiologically and transcriptionally different from planktonic cells (Whiteley et al., 2001; Schembri et al., 2003; Stanley et al., 2003; Zhu and Mekalanos, 2003; Meibom et al., 2004).

Vibrio cholerae is both the agent of the diarrhoeal disease cholera and a natural inhabitant of aquatic environments. The major virulence determinants of V. cholerae human infection are cholera toxin (CTX), which is carried on the CTX phage, and the toxin co-regulated pilus (TCP) (Mekalanos, 1985; Taylor et al., 1986; 1987; Waldor and Mekalanos, 1996; Tacket et al., 1998). A number of regulatory elements including TcpP and TcpH are responsible for activation of ctx transcription upon entry into the human host (Carroll et al., 1997; Hase and Mekalanos, 1998; Murley et al., 1999).

Vibrio cholerae is also found in marine, estuarine and fresh water environments in association with zooplankton, phytoplankton, crustaceans, insects and plants (Huq et al., 1983; 1986; Tamplin et al., 1990; Colwell and Spira, 1992; Colwell and Huq, 1994; Fotedar, 2001). Surface attachment is thought to be required for colonization of both the human intestine and the aquatic environment. As has been observed for other organisms, V. cholerae passes through the planktonic and monolayer stages prior to forming a biofilm. Free-swimming planktonic cells are characterized by the presence of flagella, and the flagellar genes are actively transcribed in this stage. Transient interactions with the surface are observed in the planktonic stage, and these are mediated by the mannose-sensitive haemagglutinin (MSHA), a type IV pilus. The interaction of MSHA with the surface is blocked by mannose or by α-methylmannoside (AMM), a non-metabolizable analogue of mannose. Surface association leads to repression of flagellar gene transcription, and this, in turn, leads to permanent attachment of cells to the surface in a monolayer. Once formed, these permanent attachments are distinguished from transient attachments by their resistance to the action of AMM. The flagellar mutant monolayer is also resistant to the action of AMM. This supports the hypothesis that flagellar motility must be absent for permanent attachment to occur (Moorthy and Watnick, 2004).

Exposure of a wild-type V. cholerae monolayer to monosaccharides, either by supplementation of the bathing medium or by degradation of a polysaccharide surface to which the cells are attached, activates transcription of the vps genes, which are responsible for synthesis of the VPS exopolysaccharide (Yildiz and Schoolnik, 1999; Kierek and Watnick, 2003; Moorthy and Watnick, 2004). The vps synthesis genes, which include vpsA (VC0917) and vpsL (VC0934), are located within the vps island encompassing loci VC0916–VC0941. Synthesis of the VPS exopolysaccharide leads to formation of a mature biofilm consisting of bacterial pillars attached to a surface (Watnick and Kolter, 1999; Yildiz and Schoolnik, 1999). Thus, progression from the planktonic to the biofilm stage involves changes in gene transcription, extracellular matrix composition and three-dimensional structure.

Regulation of VPS synthesis has been partially elucidated through the work of several laboratories. Environmental signals such as monosaccharides and nucleosides have been identified as activators of vps gene transcription and biofilm formation (Haugo and Watnick, 2002; Kierek and Watnick, 2003), while high cell density has been identified as an inhibitor of vps gene transcription through the action of HapR (Hammer and Bassler, 2003; Vance et al., 2003; Zhu and Mekalanos, 2003; Yildiz et al., 2004). VpsT, VpsR and VieA are additional regulators of biofilm formation that respond to as yet unidentified environmental signals (Yildiz et al., 2001; Casper-Lindley and Yildiz, 2004; Tischler and Camilli, 2004).

A complete understanding of the developmental pathway leading to formation of the bacterial biofilm requires exploitation of global whole genome approaches. In this work, we have used whole genome transcriptional analysis to describe modulation of the V. cholerae transcriptome during passage through the planktonic, monolayer and biofilm stages of biofilm development. Through these experiments, we have made the observation that the transcriptomes of the V. cholerae monolayer and biofilm are, indeed, distinct with the exception of a few similarly regulated genes. Furthermore, we have demonstrated stage-specific roles for CheY-3 in monolayer formation and for two newly identified proteins, Bap1 and LeuO, in formation of the extracellular biofilm matrix. Genetic requirements such as these present novel targets for development of stage-specific biofilm inhibitors.

Results

Distinct modulation of the V. cholerae transcriptome characterizes entry into the monolayer and biofilm stages

Wild-type V. cholerae forms a monolayer when grown in minimal medium (MM) alone. When mannose is added to MM, the monolayer develops into a biofilm. We have previously used these growth conditions to demonstrate that transcription levels of genes involved in flagellar and exopolsaccharide synthesis are different in wild-type V. cholerae monolayers and biofilms (Moorthy and Watnick, 2004). In the present experiments, our goal was to use microarray analysis to obtain a genomic perspective of modulation of gene transcription in response to monolayer and biofilm formation. To achieve this, we performed the following three microarray experiments (Fig. 1). To determine the effect of monolayer formation on the V. cholerae transcriptome, we compared levels of gene transcription in wild-type V. cholerae monolayer cells with those in planktonic cells grown in MM (monolayer experiment). In order to separate the contributions of mannose supplementation and surface attachment to modulation of the V. cholerae transcriptome during biofilm formation in MM supplemented with mannose, we performed the microarray analysis of biofilm formation in the following two steps: (i) a comparison of the transcriptome of planktonic cells grown in MM supplemented with mannose with that of planktonic cells grown in MM alone (planktonic experiment) and (ii) a comparison of the transcriptome of biofilm cells with that of planktonic cells grown in MM supplemented with mannose (biofilm experiment). Each comparison was performed in triplicate. The goal of our microarray analysis was to identify as many stage-specific genes as possible rather than to accurately measure levels of differential transcription. Thus, we did not base inclusion in our lists of differentially regulated genes on a test of statistical significance. Rather, genes that were induced or repressed by twofold or greater in all three microarray replicates were included in the lists of differentially regulated genes (see Tables S1–S4).

Fig. 1. Schematic representation of the state of V. cholerae under each of the conditions in this work.

A. Monolayer experiment. Test: Monolayer (−M)/Reference: Planktonic (−M).

B. Planktonic experiment. Test: Planktonic (+M)/Reference: Planktonic (−M).

C. Biofilm experiment. Test: Biofilm (+M)/Reference: Planktonic (+M).

D. vpsA monolayer experiment. Test: ΔvpsA mutant monolayer (+M)/Reference: ΔvpsA planktonic (+M).

+M and −M refer to the presence and absence of mannose in the growth medium respectively.

The monolayer, planktonic and biofilm experiments identified 150, 128 and 383 differentially regulated genes respectively. Most of these genes were differentially regulated under the conditions of only one of the three experiments. This result supports our previous hypothesis that the V. cholerae monolayer and biofilm are distinct developmental stages.

To obtain an overview of our microarray results, we classified genes according to the genome annotation and then tabulated the numbers of differentially regulated genes in selected functional categories (Table 1; Heidelberg et al., 2000). We have previously shown that addition of mannose to planktonic and monolayer cells activates vpsL gene transcription and formation of intracellular contacts (Moorthy and Watnick, 2004). This suggests that synthesis of an extracellular matrix occurs in the presence of mannose. Construction of an extracellular matrix requires: (i) transporters to import the necessary substrates for extracellular matrix synthesis; (ii) enzymes to metabolize these substrates; and (iii) cell envelope enzymes to catalyse synthesis of the extracellular matrix. In fact, many induced genes were identified in the categories of transport and binding proteins, carbohydrate metabolism and cell envelope synthesis in the planktonic and biofilm experiments. As is the case for many microarray experiments, hypothetical and conserved hypothetical genes constituted the largest category of differentially regulated genes. This supports the widely held view that there is yet much to be learned about biofilm development and the bacterial genome, in general.

Table 1.

Categories of differentially regulated genes in monolayer, planktonic and biofilm microarrays.

	Monolayer		Planktonic		Biofilm
Tabulation of selected categories	Induced	Repressed	Induced	Repressed	Induced	Repressed
Hypothetical proteins	47	24	22	25	58	85
Transcription regulation	4	7	4	2	9	13
Response regulators	4			3	2	7
Transport and binding proteins	5	1	10	10	17	18
Carbohydrate metabolism		5	2		5	5
Cell envelope synthesis			6	2	6	6
Chemotaxis and Motility	8		1	6	5	6
Total number of differentially regulated genes	91	59	68	60	171	212

The ΔvpsA monolayer is similar to the wild-type biofilm

In MM alone, wild-type V. cholerae forms a monolayer on surfaces. When mannose is added, transcription of the vps genes is activated, leading to biofilm formation. In contrast, even in mannose-supplemented medium, a ΔvpsA mutant forms a monolayer attributed to the inability to synthesize exopolysaccharide (Moorthy and Watnick, 2004). This led us to hypothesize that gene transcription in a ΔvpsA monolayer formed in the presence of mannose would be most like that in a wild-type monolayer formed in the absence of mannose. To evaluate this hypothesis, we compared gene transcription in ΔvpsA mutant monolayers formed in the presence of mannose with that in ΔvpsA mutant planktonic cells grown in MM supplemented with mannose (vpsA monolayer experiment). We performed a K-means clustering of the expression values of the monolayer, planktonic, biofilm, and vpsA monolayer experiments using standard distance as the measure of similarity (Fig. 2). Our cluster analysis highlighted the close parallels in modulation of gene transcription as a result of biofilm formation by wild-type V. cholerae and monolayer formation by a ΔvpsA mutant. In fact, 123 similarly regulated genes were detected between the wild-type biofilm and vpsA monolayer experiments, while only 10 similarly regulated genes were detected between the wild-type monolayer and vpsA monolayer experiments. Thus, the ΔvpsA monolayer is more closely related to the wild-type biofilm than to the wild-type monolayer.

Fig. 2.

Cluster analysis of the monolayer, planktonic, biofilm and vpsA monolayer experiments described in this work. The intensity of red or green represents the extent of gene induction or repression respectively. Yellow represents no change in gene transcription.

Extracellular matrix gene transcription is activated by planktonic growth of wild-type V. cholerae in the presence of mannose and repressed by ΔvpsA mutant monolayer formation

As shown in Fig. 3, most genes in the vps island, including vpsL, were activated in the planktonic experiment and repressed in the vpsA monolayer experiment. This observation is in agreement with previous quantitative polymerase chain reaction (PCR) experiments demonstrating that transcription of vpsL is repressed upon ΔvpsA mutant monolayer formation in the presence of mannose (Moorthy and Watnick, 2004). This suggests that, when exposed to biofilm-activating conditions, surface-associated cells are able to sense a block in extracellular matrix synthesis and, through an unknown negative feedback loop, repress transcription of genes involved in extracellular matrix synthesis.

Fig. 3.

Modulation of transcription of vps genes and similarly regulated genes in the planktonic and vpsA experiments. The intensity of red or blue represents the extent of gene induction or repression respectively. Yellow represents no change in gene transcription. Putative gene products and functions are listed in a table below.

We identified eight additional genes outside the vps island that were activated in the planktonic experiment and repressed in the vpsA experiment (Fig. 3). On detailed analysis, all of these genes clustered with one or more of the vps genes. We hypothesized that these genes might also be involved in synthesis of the extracellular biofilm matrix.

Because our primary goal was to identify stage-specific genes rather than to accurately quantify stage-specific transcription, we proceeded directly to construction of mutants. Two genes, located at loci VC1888 and VC2485, were selected for deletion. The gene at locus VC1888 is one of three encoding haemolysin-related proteins that are co-regulated with the vps island. We have named this protein Bap1 for biofilm-associated protein 1. It has a predicted signal peptide at the N-terminus but no membrane-spanning domains, suggesting that it is a secreted protein. It also contains four 30-amino-acid repeats that are homologous to FG-GAP domains. These domains are also found in eukaryotic membrane-associated integrins, which mediate adhesion of eukaryotic cells to proteins in the extracellular matrix such as fibrinogen and fibronectin (Baneres et al., 2000). The protein encoded at locus VC2485 is homologous to transcriptional regulators of the LysR family, with 50% identity and 75% similarity to LeuO, a positive regulator of the leu operon in Escherichia coli. No better match to LeuO is encoded in the V. cholerae genome. Furthermore, the V. cholerae leuABCD operon is located near VC2485 at loci VC2490–VC2493. This arrangement of the leu genes is similar to that found in the E. coli genome. Thus, we have named this protein LeuO.

In order to determine if Bap1 and LeuO play a stage-specific role in biofilm development, we quantified surface association in the monolayer and biofilm stages by wild-type V. cholerae as well as Δbap1, ΔleuO and ΔvpsA mutants. Because biofilm accumulation is sparse in minimal medium and therefore difficult to quantify by macroscopic methods, we chose to study Δbap1 and ΔleuO biofilms formed in LB, a medium in which the predominant surface-associated state is the vps-dependent biofilm (Kierek and Watnick, 2003). As shown in Fig. 4A, the monolayers made by wild-type V. cholerae, the Δbap1 mutant, the ΔleuO mutant and a ΔvpsA mutant covered comparable areas of the substratum (Fig. 4A). In LB, both the Δbap1 and ΔleuO mutants showed reduced biofilm accumulation, demonstrating that these mutants are defective for biofilm formation (Fig. 4B). Interestingly, both these mutants demonstrated greater surface accumulation in LB than a ΔvpsA mutant, suggesting that VPS synthesis does occur in these mutants and that synthesis of VPS contributes to surface attachment.

Fig. 4. Monolayer formation, biofilm formation and vpsL transcription by wild-type V. cholerae, a Δbap1 mutant, a ΔleuO mutant and a ΔvpsA mutant.

A. Total surface area covered by wild-type and mutant monolayers formed over 3 h.

B. Quantification of wild-type and mutant biofilm accumulation in LB.

C. Quantification of vpsL transcription in wild-type and mutant V. cholerae harbouring a chromosomal fusion of the vpsL promoter to the lacZ gene.

Because decreased biofilm formation by V. cholerae mutants has previously been correlated with decreased transcription of the vps genes, we measured vpsL transcription in LB by wild-type V. cholerae, a Δbap1 mutant and a ΔleuO mutant carrying chromosomal vpsL–lacZ reporter fusions. The levels of vpsL transcription in LB broth were comparable for wild-type and mutant strains (Fig. 4C). This indicates that decreased biofilm formation by the mutants is not the result of decreased vpsL gene transcription.

The monolayer and biofilm stages have 12 commonly regulated gene clusters

We identified 27 genes that were similarly regulated in the monolayer and biofilm experiments. In contrast, only four differentially regulated genes were common to the planktonic and monolayer experiments, and only three were common to the planktonic and biofilm experiments. This suggested to us that a small subset of differentially regulated genes might be used to describe the footprint of the V. cholerae surface-associated state. To define this footprint more precisely, we searched for genes that were similarly regulated in the monolayer, biofilm and vpsA monolayer experiments as well as similarly regulated genes that either were in common operons, were in spatial proximity, or were related in function (Table 2). Twelve clusters of similarly regulated genes were identified. Although the transcription patterns of these gene clusters were not confirmed by quantitative PCR, in all cases they were observed in more than one microarray experiment comparing transcription in a surface-associated state to transcription in the corresponding planktonic state. The function of most of these gene clusters remains to be elucidated. However, two interesting themes emerged from this tabulation. First of all, the gene cluster including tcpP and tcpH, which is involved in virulence gene regulation, was repressed in all three surface-associated states. Futhermore, the CTXΦ genes rstR, ctxA and ctxB were repressed in the wild-type and vpsA monolayer experiments. These data suggest that, under the conditions of these experiments, surface association may repress virulence gene transcription.

Table 2.

Similarly regulated gene clusters in the monolayer, biofilm and vpsA monolayer experiments.

Gene ID	Product	Monolayer	Biofilm	vpsA Monolayer
Upregulated
VC1075	Conserved hypothetical protein	3.80		8.23
VC1077	Hypothetical protein			9.02
VC1330	Hypothetical protein		7.58
VC1332	Conserved hypothetical protein		83.90	26.08
VC1333	Hypothetical protein	8.37	80.21	20.61
VC1334	Conserved hypothetical protein	5.46	124.97	3.10
VC1335	Transcriptional regulator, GntR family	6.46	29.63	12.09
VC1336	Carboxyphosphonoenolpyruvate phosphonomutase	8.45	21.70	27.76
VC1338	Aconitate hydratase 1			2.34
VC1339	Conserved hypothetical protein		3.12
VC1514	Hypothetical protein	5.58	5.03
VC1515	Chaperone, formate dehydrogenase-specific, putative	8.58	7.82
VC1516	Iron-sulphur cluster-binding protein	3.37	3.63
VC1517	Hypothetical protein	3.98
VC1518	Hypothetical protein	3.32
VC1519	Formate dehydrogenase accessory protein	5.51
VC2705	Sodium/solute symporter, putative	3.95	21.33
VC2706	Conserved hypothetical protein			5.96
VC2708	Guanylate kinase	8.74
VC2709	DNA-directed RNA polymerase, omega subunit			13.85
VC2711	ATP-dependent DNA helicase RecG		4.03	4.75
VCA0676	Iron-sulphur cluster-binding protein NapF	9.93	3.51	4.13
VCA0677	napD protein		6.19
VCA0678	Periplasmic nitrate reductase		5.77
VCA0680	Periplasmic nitrate reductase, cytochrome c-type protein		3.18
VCA0682	Transcriptional regulator UhpA		28.09
VCA0684	Regulatory protein UhpC	8.86	5.86
VCA0685	Iron(III) ABC transporter, periplasmic iron-compound-binding protein	6.67
Downregulated
VC0826	Toxin co-regulated pilus biosynthesis protein P	0.31a		0.29a
VC0827	Toxin co-regulated pilus biosynthesis protein H	0.23	0.26a	0.35a
VC0940	Conserved hypothetical protein		0.10	0.12
VC0943	Lipoic acid synthetase		0.30
VC0944	Lipoate-protein ligase B	0.33
VC1443	Hypothetical protein	0.24
VC1444	Hypothetical protein	0.33
VC1445	Sensor histidine kinase/response regulator	0.32	0.38
VC1455	Transcriptional repressor RstR	0.42
VC1456	Cholera enterotoxin, B subunit, ctxB			0.21
VC1457	Cholera enterotoxin, A subunit, ctxA	0.25a		0.34a
VC2698	Aspartate ammonialyase		0.12	0.29
VC2699	C4-dicarboxylate transporter, anaerobic	0.40
VCA0817	Hypothetical protein	0.11	0.30
VCA0818	Magnesium transporter MgtE, putative	0.37
VCA0820	Chaperonin, 60 Kd subunit			0.29
VCA1045	PTS system, mannitol-specific IIABC component	0.39	0.29	0.20
VCA1046	Mannitol-1-phosphate 5-dehydrogenase	0.16	0.38	0.28

^aDenotes that the expression value is an average of two replicates.

Secondly, the large gene clusters within VC1514–VC1519 and VCA0676–VCA0685 include proteins containing iron-sulphur clusters as well as those involved in formate dehydrogenation and nitrate reduction respectively. These types of proteins are involved in anaerobic respiration, and transcriptional activation of these gene clusters suggests that surface-attached cells may induce proteins involved in anaerobic respiration.

Chemotaxis plays a role in monolayer formation

The V. cholerae genome contains 43 homologues of methyl-accepting chemotaxis proteins (MCPs). Eight of these MCPs were activated in the monolayer experiment but none were repressed (Table 3). In the planktonic experiment, only one MCP displayed increased transcription. In the biofilm experiment, six MCPs were differentially transcribed. Three of these displayed increased transcription, and three displayed decreased transcription. Based on these observations, we questioned whether chemotaxis might play a role in V. cholerae biofilm development. In E. coli, CheY plays a central role in chemotaxis, interacting directly with the flagellar motor to change the direction of rotation. V. cholerae possesses multiple CheY homologues. Only CheY-3, however, has been shown to play a role in classical flagellum-mediated chemotaxis (Butler and Camilli, 2004). To investigate the role of chemotaxis in surface association, we constructed a V. cholerae ΔcheY-3 mutant and tested its ability to form a monolayer and a biofilm. As illustrated in Fig. 5A and quantified in Fig. 5B, after 24 h, the substratum surface area covered by the Δche Y-3 mutant monolayer was only slightly less than that covered by the wild-type V. cholerae monolayer. We have previously shown that incubation with AMM may be used to differentiate between permanently and transiently attached cells. To determine the numbers of permanently attached cells in the wild-type and Δche Y-3 mutant monolayers, monolayers formed over 24 h were treated with AMM. More than 50% of the cells in the Δche Y-3 mutant monolayer were removed by AMM treatment (Fig. 5A and B), while only a small proportion of wild-type cells were removed by this treatment. This suggests that the transition from transient to permanent attachment is retarded in the Δche Y-3 mutant. Interestingly, the ΔcheY-3 mutant displayed no impairment in biofilm formation in LB (Fig. 5C). Thus, the surface-attachment defect of the ΔcheY-3 mutant is monolayer-specific.

Table 3.

Differentially regulated methyl-accepting chemotaxis genes in monolayer, planktonic and biofilm microarrays.

	Induced		Repressed
	Gene ID	Average expression value	Gene ID	Average expression value
Monolayer	VC1248	4.5
	VC1394	4.5
	VC1405	5.0
	VC1868	3.3
	VCA0008	7.4
	VCA0031	7.8
	VCA0906	4.1
	VCA0988	3.7
Planktonic	VCA0663	6.7	VC1406	0.3
Biofilm	VC0216	6.6	VC1535	0.1
	VC0282	8.7	VCA0268	0.3
	VC1898	9.6	VCA0663	0.3

Fig. 5. Effect of AMM treatment on wild-type V. cholerae and ΔcheY-3 mutant monolayers formed over 24 h.

A. Phase contrast micrographs of monolayers before (Monolayer) and after (Monolayer + 0.1% AMM) treatment with AMM.

B. Total surface area covered by monolayer cells before and after treatment with AMM.

C. Quantification of wild-type V. cholerae and ΔcheY-3 mutant biofilm accumulation in LB.

Discussion

In this work, we have used microarray analysis to study the transcriptome of V. cholerae during each stage of biofilm development. Our results indicate that the wild-type V. cholerae monolayer and biofilm stages are transcriptionally distinct at the whole genome level and, thus, support previous assertions that the monolayer and biofilm are distinct stages in biofilm development. Based on our microarray analysis, we have identified three proteins, namely CheY-3, LeuO and Bap1, which are required in a specific stage of biofilm development.

In E. coli, the bacterial chemotaxis machinery directs cells towards favourable stimuli and away from noxious stimuli by altering the frequency of flagellar pauses and reversals of direction (Lapidus et al., 1988; Eisenbach et al., 1990). CheY, which is at the centre of the chemotactic signal transduction system, is phosphorylated by the kinase CheA (Szurmant and Ordal, 2004). This process is modulated by MCPs, which serve as the sensors of environmental stimuli. In the absence of CheY-P, the flagellum undergoes counterclockwise rotation, leading to a straight swimming phenotype. When CheY-P interacts with the flagellar motor, clockwise rotation is induced, and flagellar pauses and reversals of direction increase in frequency. Recent studies suggest that a similar mechanism is at work in V. cholerae chemotaxis (Gosink et al., 2002; Butler and Camilli, 2004).

The role of chemotaxis in biofilm development varies with the bacterial species under study. Pratt and Kolter (1998) have previously shown that chemotaxis is dispensable for initiation of E. coli biofilm formation. In contrast, Aeromonas caviae chemotaxis mutants display a defect in biofilm formation (Kirov et al., 2004). V. cholerae has multiple homologues of CheY and CheA. However, only CheY-3 and CheA-2 have been shown to be involved in flagellar-based chemotaxis (Gosink et al., 2002; Butler and Camilli, 2004).

In this work, we have found the first evidence that V. cholerae monolayer formation is facilitated by the bacterial chemotactic apparatus. We have previously presented evidence that the absence of flagellar motility is required for permanent attachment (Moorthy and Watnick, 2004). One possibility is that flagellar pauses, which are observed in the presence of a functional chemotactic apparatus, facilitate initiation of the genetic changes required for permanent attachment. Alternatively, CheY-3 may enhance permanent attachment through a signalling pathway that is independent of the flagellar motor.

Transcriptional activation of several genes encoding MCPs was observed upon monolayer formation. The types of ligands bound by these MCPs have not yet been studied, and the observed transcriptional activation of these MCPs does not necessarily imply a role in monolayer formation, maintenance, or dissolution under the experimental conditions utilized here. However, this pattern of transcription suggests to us that the chemotactic apparatus of monolayer cells remains prepared for a rapid response to environmental stimuli and, hence, that monolayer cells are not fully committed to a surface-associated existence.

The transition to the biofilm stage is initiated by various environmental signals through the action of specific transcription factors (Yildiz et al., 2001; Haugo and Watnick, 2002; Hammer and Bassler, 2003; Kierek and Watnick, 2003; Zhu and Mekalanos, 2003; Casper-Lindley and Yildiz, 2004). To date, all identified transcription factors alter biofilm formation by modulation of vps gene transcription. We have identified two biofilm-specific proteins that increase V. cholerae biofilm accumulation without increasing transcription of the vps genes. Because there is little similarity between Bap1 and the V. cholerae haemolysin HlyA, it seems unlikely that Bap1 functions as a haemolysin. Rather we favour the hypothesis that, like VPS, it is transported into the extracellular milieu during biofilm formation and plays a direct role in stabilization of the extracellular matrix. However, we do not exclude the possibility that Bap1 may use a post-transcriptional mechanism to alter exopolysaccharide synthesis and transport. LeuO has been implicated in post-transcriptional repression of rpoS in E. coli and activation of virulence genes and porins in Salmonella (Repoila and Gottesman, 2001; Tenor et al., 2004). Thus, it seems likely that LeuO is a global regulator. As has been observed in E. coli, LeuO may also utilize a post-transcriptional mechanism to increase exopolysaccharide synthesis. Alternatively, it may directly regulate transcription of genes encoding elements of the extracellular matrix other than exopolysaccharide. Experiments are underway in the laboratory to determine the roles of Bap1 and LeuO in V. cholerae biofilm formation.

Biofilm development proceeds in three stages: (i) the planktonic stage, (ii) the monolayer stage and (iii) the biofilm stage. In this work, we have used microarray analysis to obtain a comprehensive description of modulation of the V. cholerae transcriptome in each stage of biofilm development. Using mutational analysis, we have identified three proteins that are uniquely required at one stage in this process. Thus, microarray analysis is a powerful tool for dissecting the transcriptome of biofilm development and for identifying stage-specific proteins that may then be used to develop stage-specific inhibitors of biofilm development.

Experimental procedures

Bacterial strains, plasmids and media

The bacterial strains and plasmids used in this study are listed in Table 4. All experiments were done in either LB broth or MM alone or supplemented with 0.2% mannose (Sigma; Moorthy and Watnick, 2004). Where specified, 0.1 M phosphate buffered saline (PBS; pH 7.0) was used to rinse the cells and AMM was added to a final concentration of 0.1%.

Table 4.

Bacterial strains, plasmids and primers.

Strain/plasmid/primers	Genotype/sequence	Source or reference
E. coli
SM10λpir	thi thr leu tonA lacy supErecA::RP4-2-Tc::MuλpirR6K; Kmr	Miller and Mekalanos (1988)
PW614	SM10λpir (pCVD443lac::ΔcheY-3)	Butler and Camilli (2004)
PW659	SM10λpir (pWM91Δbap1)	This study
PW680	SM10λpir (pWM91ΔleuO)	This study
V. cholerae
PW249	MO10 O139, Smr	Waldor and Mekalanos (1994)
PW357	MO10lacZ::vpsLp→lacZ, Smr	Haugo and Watnick (2002)
PW396	MO10 ΔvpsA, Smr	Kierek and Watnick (2003)
PW631	MO10 ΔcheY-3	This study
PW662	MO10Δbap1 lacZ::vpsLp→lacZ; Smr	This study
PW662	MO10ΔleuO lacZ::vpsLp→lacZ; Smr	This study
Plasmids
pWM91	oriR6KmobRP4 lacI pTac tnp miniTn10Km; Kmr, Apr	Metcalf et al. (1996)
pWM91Δbap1	pWM91 carrying a fragment of bap1 harbouring an internal, unmarked deletion	This study
pWM91ΔleuO	pWM91 carrying a fragment of leuO harbouring an internal, unmarked deletion	This study
Primers
P301	5′-cca acc cgc tgt atg aac tt-3′	Primers used for construction of bap1 deletion
P302	5′-tta cga gcg gcc gca gct tat cgc ggt caa cgt tt-3′
P303	5′-tgc ggc cgc tcg taa agc tgg tta acc cac aac ag-3′
P304	5′-gca gtg agt gct tcc tta cca-3′
P358	5′-ttg gca aaa agt cga ttt cc-3′	Primers used for construction of leuO deletion
P359	5′-tta cga gcg gcc gca gct ttc cat gcg gta ac-3′
P360	5′-tgc ggc cgc tcg taa atg gtg att tgt ggc gaa gt-3′
P361	5′-caa aat cag caa tcg tcc aa-3′

Construction of the deletion mutants

The ΔleuO mutant was constructed by amplification of a 447 bp fragment at the 5′ end and a 447 bp fragment at the 3′ end of VC2485 using primer pairs listed in Table 4. Internal primers included a complementary 15 bp sequence at their 3′ and 5′ ends, respectively, as previously described (Haugo and Watnick, 2002). These two fragments were joined using the gene splicing by overlap extension technique (Horton et al., 1990; Lefebvre et al., 1995). This resulted in the construction of a fragment including only the first 66 bp and last 36 bp of leuO with a deletion of 866 bp within the coding region. The fragment containing the deletion was purified and ligated into pCR2.1TOPO. This fragment was then removed from pCR2.1TOPO by digestion with SpeI and XhoI and ligated into pWM91 to create the suicide plasmid pWM91ΔleuO. This plasmid was used to create leuO deletions in the relevant strains by double homologous recombination and sucrose selection as previously described (Haugo and Watnick, 2002). The Δbap1 mutant was constructed similarly. The bap1 deletion included 33 bp at the 5′ end and 57 bp at the 3′ end of the coding region, resulting in a 1986 bp in frame deletion. The strain SM10λpir (pCVD443lac::ΔcheY-3) was used to create the ΔcheY-3 deletion mutant (Butler and Camilli, 2004).

Phase contrast microscopy and quantitative analysis of monolayer structures

Cells were grown in sterile 24-well polystyrene microtiter dishes. Wells were filled with 300 μl of MM and inoculated with overnight cultures of the relevant strain to yield a starting A₆₅₅ of 0.001. The dishes were incubated at 27°C with gentle shaking for the indicated length of time. After incubation, the planktonic cells were removed. PBS was added to wells followed by vigorous agitation for 5 min, removal of medium and replacement with fresh PBS. This procedure was repeated three times. After removal of planktonic cells, monolayer development at the bottom surface of the well was visualized with an Eclipse TE-200 microscope (Nikon) equipped with an Orca digital CCD camera (Hamamatsu). For treatment with AMM, monolayers were formed as described above and visualized by phase-contrast microscopy. They were then rinsed, and PBS supplemented with AMM was added to the wells. The cells were incubated for three additional hours at 27°C, rinsed, and visualized again. A computer equipped with Metamorph Imaging software (Universal Imaging) was used for image acquisition, processing and quantification. This software quantifies the total surface area covered by cells. All measurements were performed in triplicate and repeated several times. Monolayer accumulation was reported as a mean measurement. Error bars represent the standard deviation.

Quantitative analyis of biofilms formed in LB

The strains to be tested were grown overnight in LB broth, pH 7.4 at 27°C with shaking. The following morning, 6 μl of these cultures was used to inoculate borosilicate tubes filled with 300 μl of fresh LB broth. These cultures were incubated statically for 24 h at 27°C. At the end of the incubation, planktonic cells were removed, and the OD₆₅₅ of this cell suspension was measured. Fresh LB and an 100 μl volume of 1 mm glass beads were added to each tube. This mixture was vortexed to disperse biofilm-associated cells, and the OD₆₅₅ of the resulting suspension was measured. All measurements were performed in triplicate and repeated several times. Biofilm accumulation was reported as a mean measurement. Error bars represent the standard deviation.

β-Galactosidase measurements of vpsL transcription

For analysis of vpsL transcription, the strains to be tested were grown to stationary phase in LB broth, pH 7.4 at 27°C with shaking. Twenty microlitres of these cultures was used to inoculate tubes filled with 2 ml of LB broth, pH 7.4. These cultures were incubated overnight at 27°C with shaking. The following morning, 100 μl of this culture was removed, and an OD₆₅₅, representing the final cell density of each culture, was measured. Of each culture 1.5 ml was placed in an eppendorf tube, spun for 5 min at 3000 r.p.m. to pellet cells, washed with 1 ml of Z-buffer, and then resuspended in 100 μl of Z-buffer (Miller, 1992). After three freeze-thaw cycles, 17 μl of a 4 mg ml⁻¹ solution of o-nitrophenyl-β-D-galactopyronoside (Sigma) was added to the lysed cells, and this preparation was incubated at 37°C for 23 h. Cell debris was then removed by centrifugation, and the OD₄₁₅ of the supernatant was measured. All β-galactosidase measurements are reported as the OD₄₁₅ of the supernatant divided by the OD₆₅₅, reflecting the final cell density of the respective culture. β-Galactosidase measurements were performed in triplicate and reported as a mean measurement. Error bars represent the standard deviation.

Array printing

Each microarray consisted of ~4000 DNA oligonucleotides spotted onto UltraGAPS coated slides (Corning) using a Biorobotics MicroGrid II. Oligonucleotides were 70-mers designed to correspond to unique sequences within V. cholerae open reading frames (Illumina) and additional control sequences.

RNA extraction

All the strains were grown in MM or MM supplemented with mannose in sterile 90 mm Petri dishes. Briefly, for preparation of planktonic cell RNA, 10 ml aliquots of the cultures were pelleted by centrifugation. Two millilitres of TRIzol (Invitrogen) were added to each pellet, and cells were lysed by repetitive pipetting. After removal of planktonic cells, the remaining surface-associated cells were rinsed with PBS three times and then lysed in situ with 7 ml of TRIzol. The lysates of both planktonic and surface-associated cells were incubated at room temperature for 10 min. Chloroform was then added to the samples at a ratio of 0.2 ml for every millilitre of TRIzol. Samples were shaken vigorously for 15 min and then incubated at room temperature for 3 min. The samples were centrifuged at 4500 g at 4°C for 10 min to separate the aqueous layer, which was then transferred to a microcentrifuge tube. The RNA, along with 10 μg of carrier t-RNA, was incubated with isopropyl alcohol in a ratio of 0.5 ml for every millilitre of TRIzol to allow precipitation of RNA. Precipitated RNA was pelleted by centrifugation at 12 000 g for 30 min at 4°C, washed with 75% ethanol, dried, and then dissolved in 100 μl of RNase-free water. To remove contaminating DNA, the total RNA was incubated with RNase-free DNAse I (Ambion) for 30 min at 37°C. Finally, the RNeasy Mini Kit (Qiagen) was used to clean up the RNA after DNase I digestion. The RNA samples were checked for presence of residual genomic DNA by standard PCR. A measurement of absorbance at 260 nm was used to quantify RNA concentration, and RNA was subsequently stored at −80°C.

cDNA synthesis

Ten micrograms of total RNA was reverse transcribed in a final volume of 30 μl as follows. A pool of random hexamers, at a concentration of 5 μM, was added to the RNA prior to denaturation at 70°C for 10 min. To this mixture was then added (i) the first strand buffer [75 mM KCl, 50 mM Tris-HCl (pH 8.3), 3 mM MgCl₂]; (ii) a mixture containing 0.5 mM dATP, 0.5 mM dCTP, 0.5 mM dGTP, 0.2 mM dTTP and 0.3 mM aminoallyl dUTP; (iii) 10 mM DTT; and (iv) 400 units of Superscript II RNase H-Reverse Transcriptase (Invitrogen). The reaction mixture was incubated at 42°C for 2 h. The RNA was hydrolysed by incubation with 0.1 M EDTA and 0.2 M NaOH at 65°C for 15 min and then neutralized with 0.1 M HCl before purification of the cDNA with a Minelute column (Qiagen). The purified cDNA was quantified by measuring absorbance at 260 nm.

Labelling of the cDNA probes

The cDNAs were coupled to monoreactive Cy3 and Cy5 (Amersham) in the presence of 0.05 M NaHCO₃, for 60 min at room temperature. The reaction was quenched with 1.2 M Hydroxylamine and the labelled cDNAs were purified with Qiaquick columns (Qiagen). The efficiency of incorporation was determined by measuring the absorbance of the labelled cDNA at 260 nm and 550 nm (for Cy3 incorporation) and 260 nm and 650 nm (for Cy5 incorporation). Cy3-and Cy5-labelled corresponding reference and test cDNAs (see Fig. 1) were mixed in ratios such that the incorporated dyes were equal, and then dried down and resuspended in the hybridization buffer (25% formamide, 5× SSC, 0.1% SDS, 1 μg μl⁻¹ salmon sperm DNA, 2 μg μl⁻¹ yeast tRNA). Probes were denatured at 95°C for 3 min before hybridization.

Hybridization

Prior to hybridization, printed slides were cross-linked in a UV Stratalinker after brief hydration over boiling dH₂O. Slides were prehybridized at 42°C in 5× SSC, 0.1% SDS, 2% BSA and then briefly rinsed with water and isopropanol. Each array was hybridized simultaneously to differentially labelled cDNA probes corresponding to sample and control RNAs prepared previously. After hybridization for 36 h at 42°C, slides were washed with (i) 2× SSC, 0.2% SDS; (ii) 0.1× SSC, 0.2% SDS; and (iii) 0.1× SSC. Hybridized, washed slides were scanned for Cy5 and Cy3 fluorescence intensities using a Packard Scanarray 4000. Laser power and/or PMT were adjusted such that the fluorescence from Cy3- and Cy5-labelled probes was balanced.

Image quantification and data analysis

For each of the four experiments (Fig. 1), data were compiled from at least three microarray replicates. Scans were analysed with Imagene version 5.6.1 (BioDiscovery) to calculate spot intensity and background. Gene expression levels for all replicates were calculated from the background-subtracted fluorescent intensities of the sample and control channels. These data were further analysed on GeneSpring version 6.1 (Silicon Genetics). The data were normalized using intensity-dependent Lowess normalization per spot per chip. Genes were scored as differentially regulated if the ratio of test:reference transcripts was either >2.0 or <0.5 in all technical replicates of the experiment.