In Yersinia pestis, biofilm formation is stimulated by HmsT, a GGDEF-domain containing protein that synthesizes cyclic-di-GMP (c-di-GMP), and inhibited by HmsP, an EAL-domain protein. Only the EAL-domain portion of HmsP is required to inhibit biofilm formation. The EAL domain of HmsP was purified as a 6XHis-tag fusion protein and demonstrated to have phosphodiesterase activity using bis(p-nitrophenyl) phosphate (bis-p NPP) as a substrate. This enzymatic activity was strictly manganese dependent. A critical residue (E506) of HmsP within the EAL domain, that is required for inhibition of biofilm formation, is also essential for this phosphodiesterase activity. While the proposed function of EAL-domain proteins is to linearize c-di-GMP, this is a direct demonstration of the required phosphodiesterase activity of a purified EAL-domain protein.
Many bacterial species form biofilms, a community of aggregated cells surrounded by extracellular polymeric substance [1,2]. Biofilms serve a variety of functions which include intercellular adherence as well as protection against various environmental stress factors or antimicrobial agents. There are several steps in biofilm formation: attachment of bacteria to surfaces, microcolony formation and maturation . As a rule many bacteria produce exopolysaccharides as an important component of a mature biofilm . The clumping of bacteria in liquid medium and the capacity of cells to bind certain dyes like Congo red (CR)  and Calcofluor [5,6], which can interact with polysaccharides, are indicative phenotypes of biofilm formation.
It has been known for many years that certain strains of Yersinia pestis bind CR and form cell aggregates when grown at 26°C. Autoaggregation of Y. pestis cells in liquid medium and formation of pigmented colonies on media containing hemin  or CR  correlates with the ability to block fleas [9–11]. Blockage of the flea proventriculus, a valve between the esophagus and midgut, by Y. pestis is required for effective transmission of bubonic plague to mammals [12,13]. It has been observed that clumps of pigmented variants were tightly attached to proventricular spines while nonpigmented bacteria grew as unstable loose masses [9,14]. The hmsHFRS operon has been shown to be required for the CR-binding phenotype [15–17] and for blocking the flea proventriculus [10,14]. The hmsHFRS gene products show moderate similarity to pgaABCD (formerly ycdSRQP) genes products [17,18]. The later are required for biofilm formation and biosynthesis of poly-β-1,6-N-acetyl-d-glucosamine in Escherichia coli.
We have identified six hms genes that affect biofilm formation in Y. pestis. HmsH is an outer membrane β-barrel protein while HmsS is associated with the inner membrane; currently, there are no conserved domains associated with these proteins to suggest their roles in biofilm formation. HmsR and HmsF have glycosyltransferase and polysaccharide deacetylase domains, respectively . HmsT, a GGDEF-domain protein, is required for the CR-binding or Hms+ phenotype [18,22] Overexpression of HmsT leads to increased biofilm production and the GGEE residues of HmsT are essential for this phenotype. HmsP, which contains GGDEF and EAL domains, has a negative effect on CR-binding and biofilm formation in Y. pestis. The GGDEF domain of HmsP is likely inactive since SKT(EF) residues are present rather than the essential GGDE(F) residues. In HmsP, the E506 and L508 residues are essential for its negative effects on biofilm formation .
Open reading frames (Orfs) containing GGDEF and/or EAL domains are encoded in a large number of bacterial genomes . GGDEF and EAL domain proteins regulate the biosynthesis of cellulose production in Gluconacetobacter xylinus (formerly Acetobacter xylinum) through turnover of cyclic-di-GMP (c-di-GMP) . In G. xylinus, c-di-GMP serves as an allosteric activator of cellulose synthase . In E. coli, Salmonella typhimurium, Pseudomonas fluorescens and Pseudomonas aeruginosa GGDEF proteins are required for biofilm-associated phenotypes [26–29]. The response regulator PleD, required for stalk formation in Caulobacter crescentus, possesses a GGDEF domain and has diguanylate cyclase activity in vitro . Additionally, GGDEF proteins VCA0965 and AdrA caused an increase in the cellular level of c-di-GMP in Vibrio cholerae and S. typhimurium, respectively [31,32]. Finally, six purified GGDEF proteins along with two individual GGDEF domains were shown to convert GTP into c-di-GMP .
Much less is known about EAL-domain proteins. Expression of several EAL domain-containing proteins repressed biofilm formation in P. aeruginosa[32,34], V. cholerae[31,35], V. parahaemolyticus, S. typhimurium[29,32], E. coli and Y. pestis. Moreover, functional EAL domain proteins reduced the level of cellular c-di-GMP in V. cholerae and S. typhimurium[31,32]. It is hypothesized that EAL-domain proteins degrade c-di-GMP and thus prevent biofilm formation. However, a direct phosphodiesterase activity has not been demonstrated for any purified EAL-domain.
Here, we show that the isolated EAL domain of HmsP can substitute for the full-length HmsP protein as a negative regulator of biofilm formation in Y. pestis. Additionally, the purified EAL-domain protein exhibited phosphodiesterase activity.
2Materials and methods
2.1Bacterial strains and cultivation
The relevant characteristics of all bacterial strains are given in Table 1. E. coli strains were cultured in Luria broth medium at 37°C; Y. pestis strains were grown on tryptose blood agar or CR agar plates at 26°C. For crystal violet staining assays, Y. pestis cells were cultured in the defined medium, TMH . Ampicillin (100 μg ml−1) and kanamycin (25 μg ml−1) were added to media when required. Where indicated, media were supplemented with arabinose (0.2%) to induce expression from ParaBAD promoters or with IPTG (1 mM) to induce expression from Plac promoters.
Table 1. Bacterial strains and plasmids used
Strain or plasmid
Source or reference
aY. pestis KIM6–2090.1+ is avirulent because it lacks the low-calcium response (Lcr) plasmid pCD1. It possesses an intact 102-kb pgm locus (Pgm+) but has temperature-independent expression of biofilm formation (Hmsc) due to the hmsP mutation. Cmr and Apr– resistance to chloramphenicol and ampicillin, respectively.
4.2 kb, Apr, 0.8 kb Nco I–Bam HI fragment of PCR product from pBADHmsP cloned into Nco I–Bgl II site of pQE60, IPTG-inducible expression of EAL-HmsP
4.2 kb, Apr, 0.8 kb Nco I–Bam HI frament of PCR product from pBADHmsP-E506A cloned into Nco I–Bgl II site of pQE60, IPTG-inducible expression of EAL-HmsP-E506A
2.2Plasmids and recombinant DNA techniques
To construct the plasmids expressing the HmsP EAL-domain (EAL-HmsP) and HmsP-E506A EAL-domain (EAL-HmsP-E506A) proteins, plasmid DNA from pBADHmsP and pBADHmsP-E506A (Table 1), respectively, were used as templates for PCR amplification. PCRs were performed with primers HmsPEAL-Nco I (5′-CAT GCC ATG GTA TTT GAG CCT CAT TTG ATT G-3′) and HmsP-End-Bam HI (5′-CGG GAT CCA CTT ACG TGG TGA GCG CTG CT-3′) using Proof Start DNA polymerase (Qiagen, Inc., Valencia, CA) and consisted of one 5 min cycle at 95°C followed by 30 cycles of 94°C for 15 s, 60°C for 30 s and 72°C for 1 min and one 5 min cycle at 72°C. The resultant 0.8 kb PCR products were digested with Nco I and Bam HI, and cloned into the Nco I/Bgl II sites of the pQE60 expression vector (Qiagen, Inc.). The plasmids, designated pQE-EAL-HmsP and pQE-EAL-HmsP-E506A (Table 1), were transformed into E. coli M15(pREP4). These constructs express C-terminal 6XHis-tag polypeptides lacking the first 456 amino acids of HmsP. The remaining 271 amino acids approximately constitute the EAL domain of HmsP. All PCR constructs were sequenced by Elim Biopharmaceuticals (Haywood, CA) to confirm the fidelity of the amplification.
2.3Cellular localization of HmsP
Y. pestis KIM6–2090.1(pBADHmsP)+ cells were harvested after overnight growth in the presence of 0.2% arabinose and the cell pellets were frozen on dry ice. After thawing, the cells pellets were resuspended in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0), incubated with lysozyme (1 mg ml−1) for 30 min on ice and sonicated using seven 10 s bursts with 10 s cooling periods. After centrifugation of the lysate at 10,000g for 30 min at 4°C, the pellet and supernatant represent the insoluble (membrane) and soluble cellular components, respectively. Equal amounts of proteins from these fractions were separated on an SDS–polyacrylamide gel and transferred to polyvinylidene fluoride membrane (Immobilon P; Millipore Corp., Billerica, MA). The transferred proteins were probed with antibodies raised against a C-terminal synthetic polypeptide of HmsP  and developed with the ECL Western blot detection reagent (Amersham Biosciences, Piscataway, NJ). Immunoreactive proteins were visualized on Kodak Biomax Light film (Eastman Kodak Co., Rochester, NY).
2.4Protein expression and purification
Expression of 6XHis-tag EAL-HmsP and EAL-HmsP-E506A proteins in E. coli strain M15 was induced with 0.1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 37°C for 4 h. The recombinant proteins were extracted under native conditions, according to the manufacturer's protocol (Qiagen, Inc.) and purified from the soluble fraction using Talon metal affinity resin as recommended by the manufacturer (Clontech, Palo Alto, CA). The 6XHis-tag proteins were eluted with 150 mM imidazole. Eluate fractions containing the 6XHis-tag protein, as judged by Western blot analysis, were pooled and dialyzed against 20 mM Tris–200 mM NaCl, pH 8.5. Protein concentrations were determined with the BCA Protein Assay Reagent Kit (Pierce Biotechnology, Inc., Rockford, IL). Protein purity was assessed by Coomassie blue and silver staining of SDS–polyacrylamide gels and by Western blot analysis.
2.5Crystal violet staining assay
Bacterial cells attached to an abiotic surface were detected by crystal violet staining assay [20,38] as modified by J.D. Fetherston (unpublished observations). Briefly, cells grown overnight on TBA slants were transferred in TMH medium and adjusted to an optical density of 0.1 at 620 nm. The cultures were grown in borosilicate tubes with moderate shaking for 16–18 h at 26°C, and then incubated with 0.01% crystal violet for 15 min. The liquid culture was removed and the test tubes washed three times with water. The dye retained in each test tube by the attached bacterial mass was solubilized with a mixture of 80% ethanol and 20% acetone. The amount of dye bound to the cells was measured at a wavelength of 570 nm on a Spectronic Genesys5 spectrophotometer. For the strains complemented with pBAD30 and pQE60 recombinant plasmids, TMH medium was supplemented with arabinose (0.2%) and IPTG (1 mM), respectively.
To identify possible substrates, 25 μg of purified EAL-HmsP was incubated with bis(p-nitrophenyl) phosphate (bis-p NPP), p-nitrophenyl thymidine 5′-monophosphate (p NP-TMP), p-nitrophenyl phosphate (p NPP) (Sigma–Aldrich Co., St. Louis, MO) at a final concentration of 5 mM. Standard assays were performed at 37°C for one hour in assay buffer (50 mM Tris–HCl, 1 mM MnCl2, pH 8.5) and the release of p-nitrophenol (p NP) was quantified at 410 nm (modified from ). To determine the metal dependency of the EAL-HmsP enzymatic activity, MgCl2, CaCl2, NiCl2, CoCl2 or ZnCl2 at a final concentration of 1 mM were substituted for MnCl2. Reaction mixtures with 10 mM MgCl2 were also tested for activity. Reaction mixtures were pre-incubated with 0.5 mM EDTA or 1 mM CaCl2 to test for inhibition of enzymatic activity by calcium. Determination of enzyme activities was carried out in duplicate from at least two independent assays and controls without enzyme were included in each experiment to quantify any non-enzymatic hydrolysis of the substrates.
To analyze phosphodiesterase activity in a gel, 200 μg of purified non-denaturated proteins was separated in an 8% polyacrylamide native gel using 25 mM Tris–HCl–192 mM Glycine buffer, pH 8. After electrophoresis, gels were soaked in assay buffer with 10 mM MnCl2 for 20 min and incubated with 10 mM bis-p NPP in the same buffer at 37°C until the p-nitrophenol product appeared as a yellow band . Subsequently another part of the gel was stained with Coomassie blue.
3.1The EAL domain of HmsP is sufficient for modulation of Y. pestis biofilm formation
Previously, we showed that HmsP, which has GGDEF and EAL domains, negatively affected biofilm formation in Y. pestis. In several other bacteria, a protein possessing these two domains represses biofilm-associated phenotypes. For example, PdeA1 negatively regulates cellulose biosynthesis in G. xylinus, while ScrC and MbaA abolish biofilm formation in V. parahaemolyticus and V. cholerae, respectively. In addition, these proteins have putative transmembrane domains and some have been localized to membranes [36,41]. The program TMHMM2.0  predicts two transmembrane domains for HmsP spanning amino acid residues 70–92 and 210–232. Western blot analysis of cellular fractions of KIM6–2090.1(pBADHmsP)+ showed that HmsP is predominantly associated with the insoluble or membrane fraction of Y. pestis (data not shown). Thus, HmsP is most likely an inner membrane protein.
Several proteins containing only an EAL domain have also been shown to negatively regulate biofilm formation [31,32,34]. Consequently, we tested whether the HmsP EAL domain alone was sufficient to inhibit biofilm formation in Y. pestis. The fragment of wild type hmsP encoding amino acids 457–728, which corresponds to the EAL domain, was cloned into the pQE60 expression vector and designated pQE-PEAL (Table 1). We also constructed pQE-PEAL-E506A (Table 1), a recombinant plasmid containing the same HmsP region from HmsP-E506A. The E506A substitution in HmsP results in a protein with no inhibitory effect on biofilm formation in Y. pestis. The constructs were transformed into Y. pestis KIM-2090.1+, a ΔhmsP::cam2090.1 mutant that overproduces biofilm . While overexpression of the functional EAL domain as well as the entire HmsP protein in KIM-2090.1+ complemented the chromosomal mutation by causing the formation of white colonies on CR plates, EAL-HmsP-E506A or HmsP-E506A failed to complement the hmsP mutation yielding red colonies at 26–30°C. These CR-binding phenotypes were independent of induction by arabinose or IPTG for constructs in pBAD or pQE, respectively (data not shown). Crystal violet assays to assess biofilm formation were in agreement with the CR-binding results. While expression of HmsP and EAL-HmsP in KIM-2090.1+ showed a drastic reduction in cell mass attached to the borosilicate test tubes, the mutant variants HmsP-E506A and EAL-HmsP-E506A demonstrated robust biofilm formation in TMH medium (Fig. 1). Thus, EAL domain of HmsP is sufficient for modulation of biofilm formation in Y. pestis and E506 residue is essential for this function.
3.2The isolated EAL domain of HmsP has phosphodiesterase activity
The EAL domain of proteins is proposed to function as a phosphodiesterase that degrades c-di-GMP [23,41]. To determine whether the EAL domain of HmsP has phosphodiesterase activity, we purified our 6XHis-tag EAL-HmsP and EAL-HmsP-E506A proteins. SDS–PAGE of the purified proteins showed two bands that reacted with antiserum against HmsP. The apparent molecular masses of the bands were 32 and 64 kDa, presumably corresponding to monomeric and dimeric forms of the proteins. After heat denaturation or treatment with a reducing agent, the 32 kDa band was predominant (Fig. 2). There was a tendency for purified EAL-HmsP proteins to irreversibly multimerize upon storage for several weeks at 4°C. Subsequent denaturation converted ?50% of the apparent dimer form to a monomeric form (data not shown).
EAL-HmsP exhibited phosphodiesterase activity using the artificial substrate bis-p NPP. This activity was manganese dependent and highest at 37°C (Figs. 3(a) and (c)). Although reduced by 60% (30°C) and 73% (26°C) compared to 37°C, significant phosphodiesterase activity was retained at the lower temperatures. Negligible activity was detected with 1 mM Mg, Ni, Co, Ca and Zn (Fig. 3(a)) or 10 mM Mg (data not shown). Pre-incubation for 30 min with 0.5 mM EDTA or 1 mM CaCl2 failed to significantly inhibit phosphodiesterase activity against bis-p NPP (data not shown).
EAL-HmsP was unable to cleave either TMP-p NP and p NPP, artificial substrates detecting phosphodiesterase/nuclease and phosphatase activities, respectively (Fig. 3B), indicating that the EAL domain of HmsP acts as a Mn-dependent phosophodiesterase, but not as a phosphatase or nuclease. In contrast, EAL-HmsP-E506A failed to demonstrate significant phosphodiesterase activity (Fig. 4(a)).
EAL-HmsP and EAL-HmsP-E506A proteins were also subjected to nondenaturing gel electrophoresis and probed for phosphodiesterase activity by phenolate release from bis-p NPP in a zymogram. Whereas functional EAL-HmsP protein produced a yellow band characteristic of phenolate release, EAL-HmsP-E506A showed no enzymatic activity (Fig. 4B). In these native gels both the lower EAL-HmsP protein band, that likely corresponds to a dimeric form of the protein, and a higher, potentially multimeric form, show enzymatic activity (Figs. 4(b) and (c)).
Biofilm formation in Y. pestis is controlled by the actions of HmsT and HmsP, with robust biofilms produced when HmsT is overexpressed or in the absence of HmsP . HmsT contains a GGDEF domain and is involved in c-di-GMP synthesis [18,20,21,55]. HmsP contains an EAL domain which has been implicated in the turnover of c-di-GMP [23,32,43–45]. In fact, phosphodiesterase activity against c-di-GMP has been detected using a purified protein (PDEA1) as well as membrane preparations from G. xylinus[24,41]. Here, we demonstrate directly that a solitary EAL domain from HmsP (EAL-HmsP) has phosphodiesterase activity (Figs. 3 and 4). This enzymatic activity was assayed with the commonly used phosphodiester artificial substrate bis-p NPP [46–48]. While the significantly higher enzyme activity of EAL-HmsP at 37°C compared to 26 or 30°C (Fig. 3C) may contribute to temperature regulation of Y. pestis biofilm formation, proteolytic degradation of HmsT remains the most likely regulatory mechanism for this control . EAL-HmsP was inactive against TMP-p NP (Fig. 3B), another artificial substrate routinely utilized for phosphodiesterase assays; several other phosphodiesterases are considerably more active against bis-p NPP than TMP-p NP [46,47,49]. Like PDEA1phosphodiesterase activity against c-di-GMP , the EAL-HmsP phosphodiesterase activity against bis-p NPP is strictly dependent on manganese (Fig. 3A and data not shown). Manganese and other divalent metals are often required for phosphodiesterase activity in a variety of enzymes and this dependency is associated in many cases with polar residues [47–51]. Unlike cleavage of c-di-GMP by PDEA1 , the enzymatic activity of EAL-HmsP against bis-p PNPP was not inhibited by calcium (data not shown). We also showed here that substitution of alanine for the glutamic acid residue, E506, an acidic polar residue, completely abrogated the phosphodiesterase activity of EAL-HmsP (Figs. 4(a) and (b)). Thus, residue E506 of HmsP may be essential for manganese or substrate binding. In line with these in vitro findings, HmsP E506 is also required for negative regulation of biofilm formation in Y. pestis. Similarly, an alanine substitution at E170 of V. cholerae VieA (in the identical position to E506 of HmsP) causes loss of function in biofilm formation . Thus, this residue as well as other highly conserved residues within EAL-domain proteins may be critical for c-di-GMP phosphodiesterase activity.
While we have not directly demonstrated degradation of c-di-GMP, it is the most likely substrate for HmsP for the following reasons. First, HmsT overexpression and an hmsP mutation lead to similar phenotypes – temperature-independent, robust biofilm formation. HmsT has recently been shown to have diguanylate cyclase activity . Second, sequence alignments show high similarity among the EAL domains of HmsP, G. xylinus PDEA1, V. cholerae VieA, and S. typhimurium YhjH  data not shown]. PDEA1 has phosphodiesterase activity against c-di-GMP and overexpression of VieA or YhjH drastically reduces cellular levels of c-di-GMP [24,31,32]. Thus, it is not unreasonable to propose that the HmsP phosphodiesterace activity serves to degrade c-di-GMP.
Many bacterial genomes have multiple genes encoding EAL domains. There are five additional EAL-domain containing Orfs in the Y. pestis genome that could have c-di-GMP phosphodiesterase activity. Although two of these (Y3389 and Y0203) have GGDEF domains, this domain is likely inactive in Y0203 since a LNSDI amino acid sequence replaces the essential GGDEF residues [20,52]. Our results with hmsP mutants 20, Fig. 1] indicate that none of these putative proteins can compensate for HmsP in regulating biofilm formation under the conditions we tested. This specificity could be due to the level of phosphodiesterase activity, environmental conditions controlling enzyme activity, or formation of specific enzyme complexes. Here, we demonstrate that HmsP is a membrane protein, probably located in the inner membrane. It is possible that all Hms membrane-associated enzymes are located in close proximity to each other in one enzymatic complex for fine regulation of polysaccharide production.
This work was supported by Public Health Service Grant AI25098 from the National Institutes of Health, USA. We cordially thank Elena Matveeva for her helpful advice on purification of the proteins and Jackie Fetherston for advice on optimizing growth conditions for the crystal violet staining procedure.