Multiple Roles for Enterococcus faecalis Glycosyltransferases in Biofilm-Associated Antibiotic Resistance, Cell Envelope Integrity, and Conjugative Transfer

Abstract

The emergence of multidrug-resistant bacteria and the limited availability of new antibiotics are of increasing clinical concern. A compounding factor is the ability of microorganisms to form biofilms (communities of cells encased in a protective extracellular matrix) that are intrinsically resistant to antibiotics. Enterococcus faecalis is an opportunistic pathogen that readily forms biofilms and also has the propensity to acquire resistance determinants via horizontal gene transfer. There is intense interest in the genetic basis for intrinsic and acquired antibiotic resistance in E. faecalis, since clinical isolates exhibiting resistance to multiple antibiotics are not uncommon. We performed a genetic screen using a library of transposon (Tn) mutants to identify E. faecalis biofilm-associated antibiotic resistance determinants. Five Tn mutants formed wild-type biofilms in the absence of antibiotics but produced decreased biofilm biomass in the presence of antibiotic concentrations that were subinhibitory to the parent strain. Genetic determinants responsible for biofilm-associated antibiotic resistance include components of the quorum-sensing system (fsrA, fsrC, and gelE) and two glycosyltransferase (GTF) genes (epaI and epaOX). We also found that the GTFs play additional roles in E. faecalis resistance to detergent and bile salts, maintenance of cell envelope integrity, determination of cell shape, polysaccharide composition, and conjugative transfer of the pheromone-inducible plasmid pCF10. The epaOX gene is located in a variable extended region of the enterococcal polysaccharide antigen (epa) locus. These data illustrate the importance of GTFs in E. faecalis adaptation to diverse growth conditions and suggest new targets for antimicrobial design.

INTRODUCTION

Biofilms are broadly defined as communities of single-species or multispecies microbial cells attached to surfaces. Mature biofilms form complexes encased in a bacterium-produced extracellular matrix composed of polysaccharides, proteins, nucleic acids, and lipids (1, 2). Clinical diseases manifest from biofilm-associated infections on indwelling medical devices (e.g., orthopedic implants), venous and urinary catheters, and damaged heart valves (3). It has been estimated that 65% of bacterial infections are associated with biofilms (4). Biofilm populations can become up to 1,000-fold more resistant to antibiotics than planktonic cells, enhancing their prevalence in clinical settings in which antibiotic selection pressure is high (5). Enterococci are the second most common nosocomial pathogens, with Enterococcus faecalis being the primary enterococcal species identified in health care-associated infections (6). E. faecalis is intrinsically resistant to numerous antibiotics, has the propensity to acquire antibiotic resistance via horizontal gene transfer, and readily forms biofilms. These traits enable the microbes to persist in hospital environments.

Multiple genes that play roles in E. faecalis biofilm development have been identified (reviewed in reference 7); however, genes involved in biofilm-associated antibiotic resistance have not been reported. The mechanisms that determine biofilm resistance to antibiotics are not completely understood. It has been suggested that the extracellular matrix provides a physical or chemical barrier that prevents antibiotics from penetrating and accessing the cells located within the matrix. In addition, it has been speculated that a subpopulation of cells within the biofilm are metabolically inactive and resistant to the effects of antibiotics. There is increasing evidence suggesting that biofilm populations also activate distinct mechanisms of resistance during biofilm development. Specific genetic determinants and similar putative mechanisms are responsible for biofilm antibiotic resistance in Pseudomonas aeruginosa, Candida albicans, and Escherichia coli (8–10). Cyclic-β-(1,3)-glucans in P. aeruginosa and C. albicans have been shown to interact directly with antibiotics and antifungals, respectively, resulting in a putative antimicrobial sequestration mechanism (9, 10). In E. coli, there is a proposed dual mechanism by which the helicase-like protein RapA upregulates a multidrug-resistant pump and a putative protein that affects cell wall composition, both of which render the biofilm resistant to antibiotics (8). These findings suggest that increased antibiotic resistance of many bacterial biofilms, including those produced by E. faecalis, is also likely mediated by specific genetic determinants.

Since enterococci exhibit the ability to acquire resistance to most antibiotics encountered, recent studies have investigated the mechanisms of E. faecalis resistance to two relatively new classes of antibiotics, namely, cyclic lipopeptides such as daptomycin (DAP) and oxazolidinones such as linezolid (Lz). These antibiotics are the preferred compounds for the clinical treatment of vancomycin-resistant enterococci (11, 12). Shortly after the introduction of DAP and Lz, resistant E. faecalis isolates were reported (13–16). Therefore, research has primarily focused on mutations conferring resistance to DAP or Lz (16–18) and on conjugative transfer of resistance plasmids contributing to Lz resistance in E. faecalis (19). Clinical and laboratory-derived isolates associated with high levels of DAP resistance include mutations in the following genes: liaFSR (three-component stress response pathway), cls (cardiolipin synthase), and gdpD (glycerophosphodiesterase) (17, 20–24). A correlation between enhanced E. faecalis DAP resistance and increased biofilm formation has been reported, but the genetic basis for the phenotype was not determined (21). While we have shown that expression of plasmid-encoded antibiotic resistance determinants is increased in biofilm cells (25), to our knowledge, there are no other reports demonstrating specific E. faecalis genes involved in biofilm resistance to antibiotics.

In this study, we sought to identify determinants responsible for E. faecalis biofilm-associated antibiotic resistance in the core genome and to characterize their mechanisms. Here we show that components of the fsr quorum-sensing system (fsrA and gelE) and two glycosyltransferase (GTF) genes (epaI and OG1RF_11715) enable biofilm formation in the presence of antibiotic concentrations that are subinhibitory for parent strain growth and biofilm development. Analysis of the respective mutant strains demonstrated multiple roles for the GTFs in resistance to detergent and bile salts, maintenance of cell envelope integrity, cell shape determination, polysaccharide production, and conjugative transfer of the pheromone-inducible plasmid pCF10. We also provide evidence supporting expansion of the polysaccharide biosynthesis gene cluster epa (enterococcal polysaccharide antigen) to include the GTF OG1RF_11715, renamed epaOX.

MATERIALS AND METHODS

Growth conditions.

E. faecalis strains were cultured overnight at 37°C in brain heart infusion (BHI) medium (Becton, Dickinson and Co., Franklin Lakes, NJ) under static conditions, unless stated otherwise. M9-YE (26) medium was used for liquid mating assays. Tryptic soy broth without dextrose (TSB^−d) (Becton, Dickinson and Co.) was used for all other experiments. Escherichia coli strains were cultured in Luria-Bertani (LB) broth (Life Technologies, Grand Island, NY) or BHI medium. Antibiotics were added to the medium as appropriate, i.e., erythromycin at 100 μg/ml in BHI medium for E. coli or 10 μg/ml for E. faecalis, spectinomycin at 50 μg/ml for E. coli or 250 or 1,000 μg/ml for E. faecalis, carbenicillin at 100 μg/ml, tetracycline at 10 μg/ml, rifampin at 200 μg/ml, fusidic acid at 25 μg/ml, daptomycin (Fisher Scientific, Pittsburgh, PA) at 2 to 6 μg/ml with 50 mg/liter CaCl₂, gentamicin (Gm) at 37.5 μg/ml, linezolid at 0.25 μg/ml, ampicillin (Amp) at 0.125 μg/ml, or nisin at 25 ng/ml. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise.

Strain and plasmid construction.

Strains and plasmids are shown in Table 1. E. faecalis strains were derived from the parent strain OG1RF. Strains containing markerless gene deletions (JD100, JD102, and JD106) were created using allelic exchange and counterselection, as described previously (27). Briefly, sequences surrounding each gene for deletion were amplified and cloned into pGEM-T Easy (Promega, Madison, WI) by using either PstI or SacI to fuse the upstream and downstream fragments. The fused sequences were subcloned into pCJK47 using BamHI and XmaI, creating each deletion vector. E. coli strain EC1000 was used for propagation of pCJK47 derivatives. Primers used to delete genes in strains are shown in Table 2.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmida	Descriptionb	Reference
Strains
OG1RF	Parent strain; Rifr Fusr	69
OG1SSp(pCF10)	Conjugative donor strain; Smr Spr Tetr	70
JD100	OG1RF ΔfsrA	This work
JD102	OG1RF ΔepaOX	This work
JD106	OG1RF ΔepaI	This work
TX5264	OG1RF ΔgelE	56
Plasmids
pMSP3535	Nisin-inducible expression vector; Emr	29
pCI372spc	Shuttle vector; Spr	This work
pJLD3	fsrA complementation vector; pCI372spc derived	This work
pJLD4	epaOX complementation vector; pCI372spc derived	This work
pMSP3614	gelE complementation vector; pMSP3535 derived	51
pCQP1	epaI complementation vector; pMSP3535 derived	This work
pCJK205	P23::lacZ constitutive reporter vector; pMSP3535 derived	47

^aOG1SSp(pCF10) required 250 μg/ml spectinomycin; E. faecalis strains harboring pCI372spc derivatives required 1,000 μg/ml spectinomycin. E. faecalis isolates harboring pMSP3535 derivatives were cultured with 25 ng/ml nisin except when used in MIC assays.

^bRif^r, rifampin resistant; Fus^r, fusidic acid resistant; Sm^r, streptomycin resistant; Sp^r, spectinomycin resistant; Tet^r, tetracycline resistant; Em^r, erythromycin resistant.

TABLE 2.

Primers

Primer	Sequence (5′ to 3′)	Descriptiona
JD321	CCCGGGCGTTCGGAAGCCAAAAACTAC	fsrA deletion, upstream (XmaI)
JD324	TGCACTGCAGCATTTGTTCACTCATCCCTTTCT	fsrA deletion, upstream (PstI)
JD318	GGATCCGTCAGCGCATAAATCAACCAAG	fsrA deletion, downstream (BamHI)
JD323	TGCACTGCAGGTCAAGGCGCTATTTCTTACTTAG	fsrA deletion, downstream (PstI)
JD315	CCCGGGGTTCTAGTGAAGGAGACTGT	epaOX deletion, upstream (XmaI)
JD326	TGCACTGCAGCACACTTTTCCCTCATTTTATTAAT	epaOX deletion, upstream (PstI)
JD312	GGATCCCCACCTCTCAGTATAGCTAC	epaOX deletion, downstream (BamHI)
JD325	TGCACTGCAGTAAGGAGTTAAAGATGTCAGAAATT	epaOX deletion, downstream (PstI)
JD334	GGATCCGAAACACATTCGTGGCTGAT	epaI deletion, upstream (BamHI)
JD348	ATCCTGCAGGCCTTTATGAAGGAGGGAGA	epaI deletion, upstream (PstI)
JD337	CCCGGGCGTGCAGATAAAGAAGCTGT	epaI deletion, downstream (XmaI)
JD349	ATCCTGCAGTTCCTCATTGTAAGCAGGAATG	epaI deletion, downstream (PstI)
JD354	GGATCCCAGATTGGAGATATTCACTATG	dnaK complementation (BamHI)
JD355	CCCGGGTTATTTGTCATCACCATTTACTTC	dnaK complementation (XmaI)
JD285	GGATCCTCCTGTAAAAACAACGACTG	fsrA complementation (BamHI)
JD286	CCCGGGCGTTCGGAAGCCAAAAAC	fsrA complementation (XmaI)
JD287	GGATCCATTCTGGGAATTACTCTTTTCC	epaOX complementation (BamHI)
JD288	CCCGGGGTGTTCCATAGTCAATGGAGA	epaOX complementation (XmaI)
JD316	CTCGAGTTAATCTCCCTCCTTCATAAAGGC	epaI complementation (XhoI)
JD317	GGATCCGTGTAAGTGAGGGGTTTTGTTAG	epaI complementation (BamHI)

^aFragment amplified, with respect to the open reading frame start codon (upstream or downstream sequences).

Mutations were complemented by introducing the native gene in trans on a plasmid. To complement the ΔfsrA strain, a 2,746-bp region, including the native promoter and ribosome binding site (RBS), was amplified using the JD285/JD286 primer pair. The amplified product was cloned into the BamHI and XmaI restriction sites of pCI372spc, generating pJLD3. pCI372spc was generated by our laboratory and is a derivative of pCI372 (28) in which the chloramphenicol selectable marker was replaced with spectinomycin using NcoI and EcoRI restriction sites. Similarly, the ΔepaOX mutant strain was complemented by amplifying a 4,120-bp region, including the native promoter and RBS, using the JD287/JD288 primer pair and cloning the product into the BamHI and XmaI restriction sites of pCI372spc to generate pJLD4. The ΔepaI strain was complemented by amplifying the epaI gene, including its native RBS, using the JD316/317 primer pair and cloning the product into the BamHI and XhoI restriction sites of pMSP3535 (29), creating pCQP1. E. coli strain DH5α was used for propagation of pCI372spc and pMSP3535 derivatives.

Biofilm growth.

Ninety-six-well polystyrene plate-based biofilm assays were performed as described previously (30). Briefly, overnight cultures were diluted 1:100 in TSB^−d with or without antibiotics, and 100 μl was dispensed into 8 wells per strain. The 96-well microtiter plates (Corning Costar 3595) were incubated in a humidified chamber for 24 h at 37°C. Cell growth was measured as the optical density at 600 nm (OD₆₀₀), using a Modulus microplate multimode reader (Turner Biosystems, Sunnyvale, CA). The culture medium was removed, and the plates were washed three times using double-distilled water and air dried for a minimum of 2.5 h. Wells were stained using 100 μl of 0.1% safranin for 20 min, rinsed five times using double-distilled water, and air dried. Safranin-stained wells were quantified by measurement of the OD₄₅₀ (Modulus microplate multimode reader). Biofilm production was calculated as an index of biomass stained with safranin (OD₄₅₀ values) normalized to cell growth (OD₆₀₀ values). Relative biofilm biomass values were calculated by further normalizing the biofilm index values of parent cultures to those of mutant cultures.

Genetic screening of transposon mutants.

Prior to screening of the E. faecalis arrayed transposon (Tn) mutant library, the antibiotic concentrations used in the screen were determined empirically. MIC assays for daptomycin, linezolid, gentamicin, and ampicillin were performed as described below, using TSB^−d as the culture medium. Growth curves were determined using 2- to 16-fold less antibiotic, relative to the determined MICs, until a concentration that was subinhibitory for OG1RF growth and biofilm formation was identified. The following subinhibitory antibiotic concentrations were used in the screens: 6 μg/ml daptomycin (supplemented with 50 mg/liter CaCl₂), 0.25 μg/ml linezolid, 37.5 μg/ml gentamicin, or 0.125 μg/ml ampicillin. For initial screening of the transposon library, overnight starter cultures of the parent strain and transposon mutants were prepared, and 96-well microtiter plates containing 200 μl BHI medium were inoculated with cells from thawed transposon library stock plates by using a 96-pin microplate replicator (Boekel Scientific, Feasterville, PA). A parent OG1RF starter culture was prepared by inoculating 20 ml of BHI medium with cells from a freezer stock. Both starter cultures were incubated statically at 37°C for approximately 16 h. The OG1RF starter culture was distributed (200 μl/well) into 96-well microtiter plates. Biofilm cultures were prepared by using the starter cultures to inoculate a 96-well polystyrene microtiter plate containing TSB^−d with or without subinhibitory antibiotic concentrations. The 96-pin microplate replicator was used for all inoculations and is estimated to transfer 2 to 3 μl of starter culture. The microtiter plates were incubated in a humidified chamber for 24 h at 37°C and then were washed and stained as described above. Transposon mutants that exhibited no defect in planktonic growth and decreased biofilm biomass (50%, compared to the parent strain) in the preliminary screen were retested individually, as described in “Biofilm growth.” The transposition loci were identified using either inverse PCR or a modified version of the INSeq method (31), followed by PCR-specific validation of disrupted loci. Markerless in-frame deletions in the genes of interest that reproducibly demonstrated the desired phenotype and could be complemented in trans were constructed and used in all follow-up assays.

Antibiotic MIC and bile/detergent sensitivity assays.

Assays were performed by diluting overnight cultures to an OD₆₀₀ of 0.05 in TSB^−d and incubating the cells statically at 37°C. For MIC experiments, cells at an OD₆₀₀ of 0.1 were diluted 1:100 in fresh TSB^−d and 100 μl of cell suspension was added to 100 μl of medium containing 2-fold dilutions of antibiotic. MICs were recorded as the lowest antibiotic concentrations with no growth. For detergent and bile sensitivity assays, 10-fold serial dilutions of cultures at an OD₆₀₀ of 0.5 were plated on BHI medium or BHI medium supplemented with 0.0125% SDS, 4% sodium cholate, or 0.06% deoxycholate. All plates were incubated at 37°C under static conditions.

Chlorophenol red-β-d-galactopyranoside β-galactosidase assay.

Overnight cultures were diluted to a starting OD₆₀₀ of 0.05 in TSB^−d supplemented with erythromycin and 25 μg/ml chlorophenol red-β-d-galactopyranoside (CPRG) and were incubated statically at 37°C. One-milliliter samples were obtained at the lag (0 h), exponential (2 h), transition (4 h), and stationary (6, 8, and 24 h) phases of growth. Cell growth was measured using OD₆₃₀ values, and CPRG hydrolysis was measured using OD₅₇₀ values determined after centrifugation and removal of bacteria.

Scanning electron microscopy.

Overnight E. faecalis cultures were diluted 1:100 in a 24-well polystyrene plate (Corning Costar 3524) containing 1 ml TSB^−d and Aclar fluoropolymer coupons (Honeywell International, Morristown, NJ) and were cultured for 24 h at 37°C with gentle agitation (50 to 100 rpm). Reagents were purchased from Electron Microscopy Sciences (Hatfield, PA) unless otherwise noted. Samples were prepared using the cationic dye stabilization methods originally developed by Erlandsen et al. (32) and modified as we reported previously (33). Briefly, the Aclar coupons were rinsed three times with potassium phosphate-buffered saline (KPBS) and fixed (primary fixation) for 22 h in cacodylate-buffered solution (135 mM) containing methanol-free, electron microscopy (EM)-grade 2% formaldehyde and 2% glutaraldehyde, 4% sucrose, and 0.15% Alcian Blue 8GX (Sigma-Aldrich, St. Louis, MO). Primary fixative was removed, and the Aclar coupons were rinsed three times with sodium cacodylate buffer and fixed (secondary fixation) for 1 h in cacodylate-buffered solution (150 mM) containing 1% osmium tetroxide and 1.5% potassium ferrocyanide. Fixed samples were chemically dried using a graded ethanol series (25, 50, 70, 85, 95 [two times], and 100% [two times]), processed in a CO₂-based critical point dryer (Tousimis, Rockville, MD), and sputter coated with a 1- to 2-nm layer of platinum (DV-502; Denton, Moorestown, NJ). Low-voltage scanning electron microscopy (SEM) (3 kV) was performed using a Hitachi S-4700 field emission instrument.

Purification and analysis of polysaccharides.

E. faecalis parent and isogenic mutant strains were cultured in 25 ml TSB^−d at 37°C under static conditions to an OD₆₀₀ of 0.4 to 0.6. Cells were pelleted for 10 min at 1,520 × g (TJ-6; Beckman Coulter) and washed with 2 ml of sucrose solution (25% sucrose, 10 mM Tris-HCl [pH 8]). Pellets were resuspended in 750 μl sucrose solution supplemented with 1 mg/ml lysozyme and 10 U/ml mutanolysin and were incubated overnight at 37°C with gentle agitation. The cellular fraction was removed by centrifugation (16,600 × g) for 20 min. Supernatant fractions were collected and treated with 100 μg/ml RNase A, 10 U/ml DNase, 2.5 mM MgCl₂, and 0.5 mM CaCl₂ for 4 h at 37°C to remove contaminating nucleic acids. Protein impurities were removed by adding 50 μg/ml proteinase K to the supernatant and incubating the mixture overnight at 37°C. Debris was pelleted at 16,600 × g for 20 min, and the remaining impurities were extracted using 500 μl chloroform. The aqueous phase was transferred to a new tube following centrifugation (16,600 × g) for 5 min. Polysaccharides were precipitated by the addition of ethanol to a final concentration of 75% and incubation at −70°C for 30 min, followed by centrifugation (16,600 × g) for 20 min. Precipitated pellets were washed using 75% ethanol, allowed to air dry, and resuspended in 100 μl of Tris-NaCl solution (50 mM Tris-HCl [pH 8], 150 mM NaCl). Analysis of the polysaccharide composition was performed using 25 μl of resuspended sample electrophoresed on a 10% native polyacrylamide gel (ratio of acrylamide to bisacrylamide, 29:1; Fisher Scientific) in Tris-borate-EDTA (TBE) buffer (89 mM Tris base, 89 mM boric acid, 2 mM EDTA [pH 8]). Detection of polysaccharides utilized staining with Stains-all (Sigma-Aldrich), following the manufacturer's protocols.

Conjugation assay.

Donor [OG1SSp(pCF10)] and recipient (OG1RF derivative) strains were statically cultured overnight at 37°C in 5 ml M9-YE medium. Overnight cultures were diluted 1:10 in 10 ml fresh M9-YE medium and incubated for 1 h at 37°C under static conditions. Cultures were pelleted for 10 min at 1,520 × g (TJ-6; Beckman Coulter), washed twice with 1 ml KPBS containing 2 mM EDTA, centrifuged for 1 min at 16,600 × g (Heraeus Picofuge), and resuspended in fresh M9-YE medium to an OD₆₀₀ of approximately 0.25. Mating mixtures were created by combining OD₆₀₀-normalized donors and recipients in a 1:9 volume ratio, followed by static incubation at 37°C. Samples were removed at designated time points, and 10-fold serial dilutions were plated on selective medium to quantify the total number of CFU/ml of donors (tetracycline and spectinomycin), recipients (rifampin and fusidic acid), and transconjugants (rifampin, fusidic acid, and tetracycline).

RESULTS

Development of a genetic screen to identify E. faecalis mutants contributing to biofilm formation in the presence of antibiotics.

Previously, we described the construction of an arrayed library containing approximately 15,000 E. faecalis OG1RF mariner-based transposon mutants (71). In the present study, our objective was to identify transposon (Tn) insertion mutations that reduced biofilm growth in the presence of antibiotic concentrations that were subinhibitory for parent OG1RF biofilms while having minimal impact on biofilm growth in the absence of antibiotics. We devised a 96-well plate-based biofilm screen in which the parent strain and Tn mutants were cultured in the absence or presence of daptomycin (DAP), linezolid (Lz), gentamicin (Gm), or ampicillin (Amp). The concentrations of antibiotics used in the screen were determined empirically, as described in Materials and Methods. Here we describe five mutants that showed decreased biofilm production in the presence of one or more antibiotics but were similar to the parent strain in antibiotic susceptibility in planktonic culture and in biofilm growth in the absence of antibiotic. The mutated genes that produced the desired phenotypes encode components of the fsr quorum-sensing system and glycosyltransferases (GTFs) (Table 3). Once the phenotypes of the Tn mutants were confirmed, we generated markerless in-frame deletions in the genes of interest and used these mutants for most of the follow-up studies described.

TABLE 3.

E. faecalis transposon mutants identified in this study

Mutant	OG1RF annotationa	Descriptionb	Coordinate 5′ to 3′ position (%)c	% biofilm
				Without antibiotic	With antibioticd
13 B10	OG1RF_11529 (fsrA)	Quorum-sensing response regulator	1592156 (3)	77 ± 16	39 ± 20 (DAP), 51 ± 7 (Lz), 87 ± 4 (Gm)
20 F7	OG1RF_11527 (fsrC)	Quorum-sensing histidine kinase	1590604 (1)	119 ± 22	59 ± 14 (DAP)
100 E5	OG1RF_11526 (gelE)	Gelatinase	1588069 (64)	113 ± 9	20 ± 1 (DAP)
124 E8	OG1RF_11730 (epaI)	Group 2 glycosyltransferase	1813805 (84)	84 ± 22	23 ± 4 (DAP)
6 A7	OG1RF_11715 (epaOX)	Glycosyltransferase	1794475 (82)	76 ± 7	53 ± 5 (DAP), 71 ± 9 (Gm)

^aOG1RF annotation designated by the NCBI (INSDC accession no. CP002621.1) for the transposon-disrupted gene.

^bOpen reading frame description assigned by the NCBI.

^cCoordinate positions were mapped to the OG1RF reference genome by the University of Minnesota Genomic Center. Percentages indicate the position of the transposon insertion from the 5′ end of the respective gene.

^dDAP, daptomycin; Lz, linezolid; Gm, gentamicin.

The fsr quorum-sensing system and the GelE protease contribute to biofilm growth in the presence of antibiotics.

We utilized isogenic fsrA and gelE deletion strains in biofilm assays to confirm that these genes contributed to the antibiotic resistance of biofilm cells. fsrA is the quorum-sensing response regulator that positively controls expression of the virulence-associated protease gelE, in addition to components of the fsrABDC locus (35). We did not delete fsrC, the quorum-sensing histidine kinase, although it was identified in the biofilm screen (Table 3), since FsrC is required for phosphorylation and activity of FsrA (36, 37) and a ΔfsrC mutant strain would likely exhibit a phenotype similar to that of the ΔfsrA strain. It was shown previously that insertions in the fsr regulon, including fsrA and gelE, negatively affect biofilm formation to different degrees depending on the E. faecalis strain and the culture conditions (30, 38–40). Using our culture conditions, biofilm biomass decreased 11% for the ΔfsrA mutant and 30% for the ΔgelE mutant when these strains were cultured in the absence of antibiotics. Much greater impairment in biofilm formation was observed in all instances when antibiotics were added to the culture medium. Biofilm biomass decreased between 51 and 67% for the ΔfsrA mutant, compared to the parent strain, in cultures with DAP, Lz, or Gm (Fig. 1A). No differential effects on the planktonic growth of mutant versus wild-type strains were observed at the antibiotic concentrations used. Deletion of gelE resulted in similar, albeit larger (54 to 90%), decreases in biofilm biomass when cultured in the presence of DAP, Lz, or Gm. In addition, all decreases in biofilm biomass could be complemented by introducing the native locus in trans. Since FsrA positively controls the expression of gelE, we wanted to determine whether the ΔfsrA phenotype was gelE dependent. Complementation of the ΔfsrA mutant with gelE in trans restored biofilm formation to parent strain levels in the presence of antibiotics (see Fig. S1 in the supplemental material). These results indicate that a ΔfsrA mutant produces decreased biofilm biomass in a gelE-dependent manner. This is the first report demonstrating that the fsr locus contributes to biofilm growth in the presence of antibiotics.

FIG 1.

Biofilm production in the absence and presence of antibiotics. (A) E. faecalis strains cultured for 24 h in 96-well polystyrene plates using medium alone (−), 6 μg/ml daptomycin (DAP) supplemented with 50 mg/liter CaCl₂, 0.25 μg/ml linezolid (Lz), or 37.5 μg/ml gentamicin (Gm). WT, wild type. (B) ΔepaI and ΔepaI::epaI strains cultured using medium alone (−) or 2 to 4 μg/ml DAP supplemented with 50 mg/liter CaCl₂. The mean absorbance of safranin-stained wells (OD₄₅₀) normalized to cell growth (OD₆₀₀) was calculated as a biofilm index (BI). The relative biofilm biomass equals (mutant BI/wild-type BI) × 100. Results are representative of a minimum of three independent biological replicates. Error bars, standard deviations of the means. ⭑, little to no planktonic growth. Statistical analysis using Student's t test indicated that all deletion mutants produced significantly less biofilm in the presence of antibiotic, compared to no antibiotic. *, P < 0.05.

Two glycosyltransferases are important for biofilm production in the presence of antibiotics.

epaI and OG1RF_11715 are annotated as glycosyltransferases that are located within or near the epa locus described by Teng et al. (41). The epa locus is annotated as a cluster of 18 genes, epaA to epaR, involved in polysaccharide biosynthesis (41–43). OG1RF_11715 is located 6 genes (5.3 kb) downstream of epaR (Fig. 2). Recently, Rigottier-Gois et al. (44) identified a glycosyltransferase gene located downstream of epaR in an E. faecalis V583 derivative, and they named the gene epaX (Fig. 2). The V583 epaX gene contains no significant nucleotide sequence similarity to OG1RF_11715; however, BLAST analysis showed that the two proteins are 52% similar. We propose naming OG1RF_11715 epaOX, to indicate its proximity to the epa locus annotated by Teng et al. (41) and to differentiate it from epaX found in E. faecalis V583 (44), since it has not been established that the two genes are functionally equivalent.

FIG 2.

E. faecalis epa locus extended-region comparison. The OG1RF annotated epa locus is shown in white. V583 epa genes upstream of epaR are not illustrated. A variable region between OG1RF and V583, extending downstream of the epa locus, is indicated in gray. Numbers within or above the arrows represent OG1RF or V583 gene designations. Transposon insertion sites are denoted by black triangles, and the disrupted gene is indicated in parentheses.

Deletion of epaOX resulted in an approximately 60% decrease in biofilm biomass, compared to the parent strain, in the presence of Gm (Fig. 1A). Deletion of either GTF gene, epaOX or epaI, also resulted in enhanced susceptibility of planktonic cells to concentrations of DAP that were subinhibitory for the parent strain (Table 4). Therefore, biofilm production in the presence of DAP could not be assessed (Fig. 1A). These data differed from the DAP susceptibility phenotype of the planktonic cells of the respective Tn mutants identified in the initial screens. This discrepancy is likely explained by the Tn insertion being near the 3′ end of each GTF gene, potentially resulting in a partially functional protein (Fig. 2 and Table 3). Since 6 μg/ml DAP inhibited the growth of the ΔepaOX and ΔepaI strains, we decreased the concentration of DAP to one subinhibitory for planktonic growth of ΔepaOX and ΔepaI mutants and reexamined biofilm production, compared to the parent strain. As shown in Fig. 1B, biofilm production by the ΔepaI mutant decreased in a concentration-dependent manner, compared to that by the parent strain. Biofilm production by the ΔepaOX mutant strain was comparable to that of the parent strain using DAP concentrations ranging from 2 to 4 μg/ml (see Fig. S2 in the supplemental material). All GTF mutants produced biofilms similar to that of the parent strain in the absence of antibiotic and could be complemented in trans. In total, these results indicate that the GTF genes epaOX and epaI contribute to biofilm formation in the presence of antibiotics. More specifically, epaOX enables proper biofilm formation in the presence of the aminoglycoside Gm, while epaI contributes to biofilm growth in the presence of the lipopeptide DAP.

TABLE 4.

E. faecalis parent strain and isogenic mutant MICs of antibiotics used in biofilm assays

Strain	MIC (μg/ml)a
	Daptomycin	Linezolid	Gentamicin
OG1RF	16	2.5	100
ΔepaOX	8	2.5	100
ΔepaI	8	2.5	100
ΔfsrA	16	2.5	100
ΔgelE	16	2.5	100

^aThe MICs were determined by broth microdilution using TSB^−d as the culture medium.

epaOX functions to maintain cell envelope integrity.

E. faecalis is a commensal organism of the gastrointestinal tract and frequently encounters harsh environments, including elevated bile salt levels (45). Since the GTF mutants exhibited increased sensitivity to the cell membrane-targeting antibiotic DAP, we hypothesized that the cell envelopes of these mutants were perturbed. To determine the cell envelope integrity of the GTF mutants, we plated serial dilutions of cultures grown in biofilm-conducive medium onto the detergent SDS and the primary and secondary bile salts sodium cholate and deoxycholate, respectively (46). The ΔepaOX mutant failed to grow in the presence of SDS or sodium cholate and showed a 1-log decrease in growth with deoxycholate (Fig. 3). In addition, the ΔepaOX mutant produced smaller colonies than the parent strain when cultured on BHI medium. Growth was restored in the complemented strain. Deletion of epaI resulted in growth comparable to that of the parent strain under all test conditions. Similarly, the ΔfsrA and ΔgelE strains showed growth comparable to that of the parent strain in the presence of SDS, sodium cholate, and deoxycholate. These results indicate that the GTF gene epaOX promotes E. faecalis resistance to detergent and bile salts.

FIG 3.

Growth defect of the ΔepaOX mutant in the presence of detergents and bile salts. Overnight cultures were diluted into fresh TSB^−d and cultured to an OD₆₀₀ of ∼0.1. Tenfold serial dilutions were plated onto BHI medium or BHI medium containing 0.0125% SDS, 4% sodium cholate, or 0.6% deoxycholate.

To quantify differences in cell envelope integrity, we performed β-galactosidase assays utilizing the nonpermeable reagent chlorophenyl red-β-d-galactopyranoside (CPRG) (47, 48). A constitutive lacZ reporter construct (47) was introduced into the E. faecalis isogenic mutants, and the strains were cultured in biofilm-conducive medium supplemented with CPRG. The CPRG substrate does not penetrate the cell envelope; therefore, it can be hydrolyzed only if the cell envelope integrity is perturbed and CPRG enters the cell cytosol or if cytoplasmic LacZ is released into the medium. β-Galactosidase activity increased in the ΔepaOX mutant over the 24-h time course, resulting in 3.5-fold greater CPRG hydrolysis, compared with the parent strain (Fig. 4). The β-galactosidase activity of the ΔepaI strain increased at 24 h but was more variable and was not statistically different from that of the parent strain (P = 0.15), whereas the ΔfsrA and ΔgelE strains phenotypically mirrored the parent strain. Notably, none of the mutants had growth defects, as measured by OD₆₃₀ values (Fig. 4). In total, these data indicate that the cell envelope integrity is disrupted in the ΔepaOX mutant, and they support a role for epaOX in E. faecalis cell envelope stabilization.

FIG 4.

Cell envelope integrity defects associated with deletion of epaOX. E. faecalis strains containing the constitutive lacZ reporter construct pCJK205 were cultured in TSB^−d supplemented with 25 μg/ml CPRG and were incubated at 37°C. At the designated time points, samples were removed to measure cell density (OD₆₃₀) and CPRG hydrolysis (OD₅₇₀) in the bacterium-free culture supernatant fraction. *, Statistically significant increases versus the parent strain (P < 0.01, Student's t test).

To determine whether the GTF mutations affected cellular morphology, we used low-voltage scanning electron microscopy (SEM) of biofilm cells cultured on Aclar fluoropolymer membranes (Fig. 5). Parent OG1RF cells produced typical oval diplococci, whereas the ΔepaOX strain produced small spherical cells that readily formed chains and clumped. Deletion of epaI resulted in an intermediate phenotype consisting of both parent-like and spherical cells. Using phase-contrast and transmission electron microscopy, it was reported previously that disruptions within the epa locus resulted in spherical and mainly single cells (41, 44). Similar to reported findings, deletion of fsrA or gelE resulted in cell chaining and maintenance of the parent oval shape (data not shown). In total, these results implicate epaI and epaOX in E. faecalis shape determination and the structural integrity of the cell envelope.

FIG 5.

Effects of epa mutations on cell morphology. Scanning electron micrographs captured after 24 h of growth on Aclar fluoropolymer coupons highlight morphological differences in the OG1RF, ΔepaOX, and ΔepaI strains. Bars, 5 μm.

epa mutations affect polysaccharide content.

To determine the effects of epa mutations on cell envelope polysaccharides, we isolated and analyzed E. faecalis carbohydrate extracts from parent and isogenic deletion strains. Deletion of either GTF gene resulted in altered polysaccharide composition. No detectable polysaccharide was extracted from the ΔepaOX strain, whereas the ΔepaI mutant showed lower-molecular-weight polysaccharides than the parent strain (Fig. 6). Deletion of fsrA and gelE had no effect on polysaccharide production, compared to extracts of the parent strain. These results indicate that epaOX is part of an extended epa locus that contributes to polysaccharide production, and they suggest that the ΔepaI strain produces immature polysaccharides. The exact polysaccharide composition of each mutant is the subject of future experimentation in our laboratory.

FIG 6.

Effects of epa mutations on polysaccharide content. Purified polysaccharides were obtained from E. faecalis isolates cultured in TSB^−d and were electrophoresed through a 10% native polyacrylamide gel. Polysaccharide content was detected using Stains-all.

epaI enables conjugal transfer of the pheromone-inducible plasmid pCF10.

Given the differences in cell wall polysaccharide content of the epaOX and epaI mutants, we considered whether these strains would display altered conjugation phenotypes. Physical cell-to-cell contact, or mating pair formation, of donor and recipient cells enables efficient conjugal transfer of pCF10 (49–52). To determine whether altered polysaccharide production negatively affected the ability of ΔepaOX and ΔepaI mutants to receive pCF10 from parent donors, we compared the kinetics of plasmid transfer into parent versus mutant recipients in liquid matings. Deletion of epaI reduced pCF10 transfer during early and late liquid mating approximately 10- to 100-fold, compared to parent OG1RF recipients. Similar decreases in transconjugant/donor and transconjugant/recipient ratios were observed (Fig. 7). There was no defect in the growth of the ΔepaI recipient, measured as total CFU/ml, that would have contributed to the observed decrease in pCF10 transfer efficiency. Wild-type recipient ability was restored by complementation with epaI in trans (see Fig. S3 in the supplemental material). In addition, ΔepaI recipient strains produced parent strain levels of the conjugation pheromone cCF10, indicating that decreased transfer of pCF10 was not attributable to insufficient donor induction (data not shown). Disruption of epaOX had no effect on recipient ability; ΔfsrA and ΔgelE mutants were also not deficient in conjugal transfer of pCF10 (see Fig. S3 in the supplemental material).

FIG 7.

Conjugation kinetics of E. faecalis parent and ΔepaI recipients. Kinetic assays were performed as detailed in Materials and Methods. E. faecalis OG1SSp (pCF10) donor cells were mixed with OG1RF parent (black) and ΔepaI mutant (gray) recipients in a 1:9 volume ratio. At each time point, serial dilutions of cultures from the mating mixtures were plated on selective medium to calculate total CFU/ml of donors (triangles), recipients (squares), and transconjugants (circles). Transconjugant/donor (solid bars) and transconjugant/recipient (dotted bars) ratios were plotted for 2 h and 24 h as log₁₀ CFU/ml. All data are representative of biological triplicates. *, values fell below the limit of detection.

Since the ΔepaI mutant was defective as a recipient of pCF10 transfer, we also examined the ability of the ΔepaI strain harboring pCF10 to donate the plasmid to a parent strain recipient. The ΔepaI mutant was not a defective donor in transferring pCF10 to recipient cells (data not shown). In total, these results suggest that polysaccharide modifications produced by EpaI contribute to mating pair formation and efficient conjugal transfer of pCF10.

DISCUSSION

Several reports have identified E. faecalis genes involved in biofilm formation (reviewed in reference 7), and separate studies have determined which genes contribute to antibiotic resistance (reviewed in reference 34). Previously, we demonstrated a correlation between the expression of plasmid-encoded resistance determinants and increased resistance of E. faecalis biofilms to antibiotics (25). However, a direct link between E. faecalis genes or alleles within the core genome that enable biofilm formation in the presence of antibiotics has not been reported. In this study, we identified genes that facilitated biofilm formation in the presence of the clinically relevant antibiotics DAP, Lz, and/or Gm; three genes are part of the fsr quorum-sensing regulon (fsrA, fsrC, and gelE), and two genes are annotated as glycosyltransferases (epaI and epaOX). We did not identify any mutants that were affected by the Amp concentrations used in our study. These data suggest that there are genetic determinants that have biofilm-specific roles in antibiotic resistance, distinct from those of planktonic cells. None of the E. faecalis genes that are known to confer resistance to the antibiotics used in this screen were identified, also suggesting that distinct resistance mechanisms exist in planktonic versus biofilm cells.

This was the first report demonstrating that components of the fsr regulon were involved in biofilm-associated antibiotic resistance. It was shown previously that insertions in the fsr regulon, including fsrA and gelE, negatively affect biofilm development and are attenuated in different animal models of infection (38, 39, 53–56). We show here that deletion of either fsrA or gelE resulted in decreased biofilm formation in the presence of DAP, Lz, and Gm but that general biofilm development in the absence of antibiotics was minimal. We hypothesize that ΔgelE strain biofilms are structurally deficient and therefore intrinsically more susceptible to antibiotics. In Staphylococcus aureus, the presence of the agrABCD quorum-sensing system, a homologue of fsrABDC, results in a fitness cost to S. aureus that is enhanced when cells are exposed to sublethal antibiotic concentrations (57). S. aureus agr-negative isolates are being isolated more frequently from hospital-acquired infections and are associated with increased mortality rates due to bacteremia, suggesting that persistent antibiotic pressure is selecting for loss of agr (57). Our data suggest that loss of the E. faecalis fsr quorum-sensing system would result in decreased virulence under antibiotic pressure, which is opposite the phenotype observed with S. aureus agr mutants. A study of 215 E. faecalis clinical isolates reported that ∼40% did not produce gelatinase, predominantly due to a 23.9-kb deletion of the fsr locus, which included the genes responsible for regulation of gelE (58, 59). No correlations between the lack of gelatinase production and clinical disease in humans have been established (59); however, studies showed that the presence of GelE increased virulence in animal models (60–62).

Deletion of either GTF resulted in altered E. faecalis polysaccharide production, providing insight into putative mechanisms of biofilm resistance to antibiotics. Polysaccharides play important roles in Gram-positive cell wall integrity and the composition of the extracellular matrix of biofilms. Gram-positive bacterial cell walls are composed of three major components, namely, a peptidoglycan backbone, anionic polymers (teichoic acids and cell wall polysaccharides), and proteins (cell wall-anchored and cell wall-associated proteins). Peptidoglycan and anionic polymers, including polysaccharides, constitute a majority (∼90%) of the cell wall weight, in approximately equal proportions (63). In addition, the extracellular matrix of biofilms consists mainly of polysaccharides, proteins, nucleic acids, and lipids (1). We hypothesize that E. faecalis polysaccharides are sequestering antibiotics within or near the cell wall, preventing them from reaching their target sites. Solid-medium MIC results determined using antibiotic Etest strips, which arguably could represent colony biofilms, support our hypothesis, since epaOX and epaI mutants exhibited decreased DAP and Gm MICs, compared to the parent strain (Table 4). An epaOX mutant did not produce polysaccharides and exhibited decreased biofilm production in the presence of the aminoglycoside Gm. Enterococci are intrinsically resistant to aminoglycosides as a result of poor antibiotic uptake (64). Although the ΔepaOX strain was more permeable, planktonic cells were not more sensitive to Gm, demonstrating a specific role for epaOX in biofilm-associated antibiotic resistance. The ΔepaI mutant produced immature polysaccharide intermediates and exhibited decreased biofilm formation in the presence of DAP. DAP is a cyclic lipopeptide that is thought to target the E. faecalis cell membrane, specifically at the division septum (24). While Tran et al. (24) examined clinical isolates with increased antibiotic resistance, our studies focused on how biofilm growth affects intrinsic resistance. The production of immature polysaccharides, or no polysaccharide in the case of the ΔepaOX mutant, resulted in enhanced susceptibility to DAP, suggesting that the polysaccharides typically help block DAP from reaching the division septum. At low DAP concentrations, the ΔepaI mutant grew in the planktonic phase but was unable to develop a wild-type biofilm, suggesting a specific role for EpaI in biofilm-associated antibiotic resistance. Future biochemical studies will be performed to determine whether E. faecalis polysaccharides are capable of directly interacting with and sequestering antibiotics.

epaOX is located in close proximity to the epa polysaccharide biosynthesis genes but was not initially annotated as part of the locus. Teng et al. (41) established boundaries to the epa locus based on gene disruption analysis, focusing on phenotypes associated with biofilm formation, phage sensitivity, cell shape, and polysaccharide production. Based on these criteria, the epa locus is annotated as 18 genes, i.e., epaA to epaR. However, our results and recent work published by Rigottier-Gois et al. (44) provide evidence for an extended epa locus. We showed that deletion of epaOX in OG1RF resulted in the absence of Epa polysaccharide production and altered cell shape, among other phenotypes. Similarly, Rigottier-Gois et al. (44) identified a GTF gene (epaX) that, when deleted, affected polysaccharide production and cell shape and was involved in E. faecalis colonization of the murine gastrointestinal tract. epaX is found in the E. faecalis V583 strain downstream of epaR. In all sequenced E. faecalis strains, epaA to epaR are considered core genes, while immediately downstream of epaR there is a region of variable gene content and organization (Fig. 2); several of the sequenced strains reported by Palmer et al. (22) have downstream genes similar in sequence and organization to OG1RF, whereas others (including V583) diverge. epaOX in OG1RF and epaX in V583 are both annotated as GTFs, with 52% protein similarity. Results have demonstrated that EpaX functions to decorate the V583 Epa rhamnopolysaccharide with galactose and/or N-acetylgalactosamine (44). Since epaOX and epaX appear to confer similar phenotypes, future experiments comparing differences in total polysaccharide content of OG1RF ΔepaOX, V583 ΔepaX, and parent strains will determine functional overlap between the genes.

We also identified an additional role for the GTF epaI gene, which is located within the annotated epa locus, in recipient ability to acquire the conjugative plasmid pCF10. Conjugative transfer of pCF10 from donor cells to plasmid-free recipients requires physical cell-to-cell contact or mating pair formation. E. faecalis donor cells harboring pCF10 produce the surface protein PrgB (also referred to as aggregation substance [AS]), which promotes aggregation of donors with recipients that produce a chromosomal receptor referred to as enterococcal binding substance (EBS) (49–52, 65). The components of EBS are not completely known; however, previous studies implicated a role for lipoteichoic acid (LTA) in AS-mediated aggregation (50, 65, 66). Waters et al. (67) demonstrated that AS directly binds LTA in vitro but LTA is not sufficient for bacterial aggregation. A more complex scenario, in which PrgB mediates interactions with extracellular DNA to enhance aggregation, was recently proposed by Bhatty et al. (68). Our findings suggest that EpaI contributes to the formation of EBS, implying that the Epa polysaccharide may be a component of EBS. Deletion of epaI resulted in reduced ability to act as a recipient in liquid matings but maintained recipient ability in plate matings (data not shown), in which direct cell contact is forced. In addition, ΔepaI mutants harboring pCF10 were unable to self-aggregate in liquid culture when induced with pheromone promoting AS production (data not shown), which further supports EpaI involvement in EBS. An alternative explanation for our results is that the Epa polysaccharide is not an integral component of EBS but it affects the conformation or surface display of EBS. This model is supported by the findings that deletion of epaOX abolished production of extractable Epa polysaccharide (Fig. 6) and had more severe effects on cell envelope integrity and shape (Fig. 3 to 5) but did not reduce recipient ability (see Fig. S3 in the supplemental material).

The formation of biofilms in the presence of antibiotics is clinically relevant, since E. faecalis is the primary infectious enterococcal species identified in hospital-acquired infections (6). E. faecalis is frequently exposed to numerous antibiotics during the course of treatment regimens. Identifying targets that could render the bacteria more susceptible to clinically relevant antibiotics is beneficial. Our work provides evidence for E. faecalis genetic determinants mediating antibiotic resistance within biofilms and suggests that E. faecalis employs biofilm-specific mechanisms independent of simple extracellular matrix diffusion barriers to keep antibiotics from their targets. Future studies will be focused on identifying the exact mechanisms for fsr regulon and GTF effects on biofilm-associated antibiotic resistance. Overall, these data provide potential targets for the design of therapeutic agents that could be used in combination with antibiotics to prevent biofilm infections.