Planktonic Interference and Biofilm Alliance between Aggregation Substance and Endocarditis- and Biofilm-Associated Pili in Enterococcus faecalis

Most bacteria express multiple adhesins that contribute to surface attachment and colonization. However, the network and relationships between the various adhesins of a single bacterial species are less well understood. Here, we examined two well-characterized adhesins in Enterococcus faecalis, aggregation substance and endocarditis- and biofilm-associated pili, and found that they exhibit distinct functional contributions depending on the growth stage of the bacterial community. Pili interfere with aggregation substance-mediated clumping and plasmid transfer under planktonic conditions, whereas the two adhesins structurally complement one another during biofilm development. This study advances our understanding of how E. faecalis, a ubiquitous member of the human gut microbiome and an opportunistic pathogen, uses multiple surface structures to evolve and thrive.

Ebp are composed of 3 subunits, EbpA, EbpB, and EbpC, where EbpC is the major pilus subunit with EbpB at the base, and EbpA is at the tip of the pilus (2,9). The three subunits are cotranscribed at the ebpABC locus and are positively regulated by the transcriptional regulator EbpR, which is encoded upstream of ebpABC (10). Polymerized Ebp exist as high-molecular-weight polymers (Ͼ200 kDa), and the length of the pilus may reach 10 m (5). Pili are only expressed on a subset of cells in the population, suggesting that they may be phase variable, and pilus expression can be induced by exposure to serum, glucose, or bicarbonate (2,(11)(12)(13). The tip adhesin EbpA mediates attachment to host fibrinogen and collagen and contributes to UTI, CAUTI, and endocarditis (2,3,13). Mutations within the N-terminal domain of EbpA prevented Ebpassociated biofilm formation both in vitro and in vivo, as well as CAUTI in mice (2,3,14).
AS is a 137-kDa protein encoded by prgB on the pheromone-responsive pCF10 plasmid (15). In the absence of the cCF10 pheromone, the expression of AS and most of the pCF10 plasmid-carried genes is inhibited by the small plasmid-encoded and constitutively expressed peptide iCF10 (16). Alternatively, the expression of AS can be induced by albumin-lipid complexes in the bloodstream that sequester or degrade iCF10, resulting in the activation of autocrine pheromone signaling (17). Scanning electron microscopy experiments demonstrated that AS is only expressed on a subset of cells in a population, even at saturating concentrations of the cCF10 pheromone (18). When expressed, AS contributes to biofilm formation, cellular aggregation required for conjugative plasmid transfer, and increased virulence in endocarditis models (1,(19)(20)(21). In addition, AS facilitates the adherence of E. faecalis to renal tubular cells and intestinal epithelial cells, as well as binding to and survival in neutrophils (22).
Most bacteria encode multiple adhesins; however, they are not always expressed at the same time, and this differential expression can arise via cross-regulation. For example, Escherichia coli Pap pili and type 1 fimbriae are cross-regulated, as are flagella and type 4 pili in Pseudomonas aeruginosa (23)(24)(25). Despite an increasing number of characterized and predicted adhesins in E. faecalis, we do not know whether or how adhesin expression is coordinated within a population. In this study, we used the well-characterized E. faecalis adhesins Ebp and AS to understand how the two adhesins may be differentially expressed on different cellular subsets, which could give rise to partitioned adhesive functions within a population. While we detected no transcriptional cross-regulation between the two adhesins, we observed rapid expression of AS in pheromone-induced cultures on nearly all cells, with Ebp coexpression on a subset of those cells. Simultaneous expression of Ebp and AS on the same cells prevented AS-mediated clumping required for conjugative plasmid transfer and, consequently, reduced horizontal gene transfer (HGT). Within biofilms, we demonstrate distinct functional contributions of Ebp and AS to biofilm development and structure, working synergistically to promote biofilm formation.

RESULTS
AS and Ebp are coexpressed on the same cells after pheromone induction. Previous studies reported that neither AS nor Ebp are expressed on all cells within a population, but both adhesins contribute to biofilm formation (2,(18)(19)(20). We hypothesized that Ebp and AS expression may be coordinated within a population such that different population subsets express different adhesin repertoires for optimal colonization, virulence, or biofilm architecture development. To address how the expression of the two adhesins is coordinated, we first quantified the individual expression of AS and Ebp within a population by diluting overnight cultures of E. faecalis strain OG1RF (a rifampin-and fusidic acidresistant derivative of OG1) into fresh medium containing the cCF10 pheromone (0.12 ng/ml). After 30 min of pheromone exposure, 82% of the cells expressed AS on the cell surface, and this number increased to 95% by 90 min of pheromone exposure (Fig. 1A). The fraction of AS-expressing cells was higher than what we expected based on earlier reports for the same strain in which representative scanning electron micrographs (SEM) of exponentially grown pheromone-induced bacteria showed that only ϳ75% of cells expressed AS (18); this difference which may be due to the increased sensitivity of AS detection by immunofluorescence microscopy (IFM) compared to SEM. After 18 h of growth in the presence of cCF10, the percentage of AS-expressing cells in the population dropped significantly to 4% (Fig. 1A). Since AS transcription peaks between 30 and 60 min after the addition of pheromone and returns to uninduced levels by the end of the second hour, the drop in the number of AS-expressing cells is likely due to an accumulation of inhibitor iCF10, along with division and dilution of AS-expressing cells by AS-negative daughter cells (17,26). Within the same population, Ebp were expressed on only 20 to 40% of cells, and Ebp expression was independent of cCF10 exposure or AS expression (Fig. 1B), indicating that Ebp could be coexpressed with AS on a subset of cells. Interestingly, independent of pheromone exposure, we noticed a significant decrease in the number of piliated cells 30 min after overnight subculturing compared to that in the overnight culture itself. We speculate that in newly replicated early log-phase cells, a sufficient number of pili have yet to be assembled and accumulated on the surface to be detected. Finally, we performed coimmunostaining using AS and EbpC antisera on pheromone-induced cells and observed that the two adhesins were displayed on the same cells and colocalized at the same hemispherical areas of the cell ( Fig. 1C; see also Fig. S1 in the supplemental material).
Ebp interfere with AS-mediated clumping. AS was originally described for its association with cellular aggregation, and Ebp have also been shown to contribute to cellular aggregation (27,28). While we did not observe a difference in Ebp expression between induced AS-expressing and uninduced AS-nonexpressing populations (Fig. 1B), we noticed that not all cells in the cCF10-induced cultures aggregated, but they instead separated into clumped cells that settled at the bottom of the tube as a pellet and cells that remained in suspension. Separation became obvious 1.5 to 2 h after the addition of cCF10 to the cultures ( Fig. 2A). We hypothesized that the clumping we observed in pheromoneinduced cultures might be both AS and Ebp dependent, where pilus-expressing cells might facilitate AS-mediated clumping. To test this, we manually separated the clumped pellet from the suspended cells, stained them with Ebp antiserum, and performed IFM. Contrary to our hypothesis, we observed fewer Ebp-expressing cells in the clumped pellet and more in the suspension than in the total (mixed) induced culture, where pellets and suspended cells were not separated ( Fig. 2B and S2).
We next performed reverse transcription-quantitative PCR (RT-qPCR) on the pellet cells and confirmed that ebpC transcript levels were lower in the pellet cells than those in the total population, with a similar transcript reduction of ebpR, a positive transcriptional regulator of ebp genes (Fig. 2C). Given that Ebp expression levels within the total population were unchanged upon cCF10 pheromone exposure (Fig. 1B), we speculated that pilus-expressing cells may physically segregate to the suspension and be excluded from the aggregated pellet where pilus-mediated steric hindrance interferes with AS-mediated clumping.
Ebp impede AS-facilitated horizontal gene transfer. Since AS-mediated clumping facilitates conjugative transfer of the pCF10 plasmid to recipient cells (20), nonclumped suspended cells, which express both AS and Ebp, could display a reduced ability to undergo AS-mediated plasmid transfer compared to the pellet cells. To test this, we quantified conjugation frequency using suspension (Ebp hi containing AS [AS ϩ ]) or pellet (Ebp lo AS ϩ ) cells using OG1SS/pCF10 as a donor (a streptomycin-and spectinomycinresistant derivative of OG1 carrying pCF10, which encodes tetracycline resistance). After 2 h of pheromone exposure, we collected the top 2 ml of each culture containing suspended cells and the bottom 1 ml of clumped cells containing the clumped pellet, out of an overall 5-ml culture. We normalized the bacterial cell numbers and mixed them with OG1RF ΔebpABC srtC recipient cells that lack pCF10 (rifampin and fusidic acid resistant). We used Ebp-deficient recipient cells to avoid pilus-mediated interference when incubated with the donor cells. Thirty minutes of coincubation of Ebp lo AS ϩ pellet donor cells and Ebp-null recipient cells yielded approximately 10 times more transconjugants than the Ebp hi AS ϩ suspension donor cell mixture (Fig. 3A), indicating that Ebp interferes with AS-mediated conjugative plasmid transfer. Each experiment was performed in triplicate using the ΔΔC T method to assess expression changes, with gyrA as the reference housekeeping gene. Statistical analysis was performed by unpaired t test using GraphPad.*, P Ͻ 0.05; **, P Ͻ 0.01, ***, P Ͻ 0.001; ****, P Ͻ 0.0001, n.s., P Ͼ 0.05.
To further test the hypothesis that pili interfere with AS-mediated HGT, we used OG1RF/pCF10 or OG1RF ΔebpABC/pCF10 (ebp-null) as the donor strain and compared pCF10 conjugation rates to that of an OG1X ΔsrtC recipient strain (streptomycin resistant) that fails to polymerize pili (Fig. 3B). After 30 min of coincubation, the ΔebpABC mutant was approximately 10-fold more efficient in plasmid transfer to OG1X ΔsrtC than was OG1RF. In another study, similar conjugation experiments performed after 2 h of coculture revealed no differences in HGT for Ebp-null donor strains and OG1RF (20). To address this, we extended our assays from 30 min to 2 h and observed that HGT rates equalized (Fig. 3). This delayed equalization of recovered transconjugants may be due to a second round of transfer of the plasmid from new donors to recipient cells, a switch to "off" piliation from originally "on" piliation donor cells, and/or enhanced cellular density that facilitates cell-to-cell contact and HGT.
AS-clumped cells mediate microcolony formation and biofilm development. While both AS and Ebp contribute to biofilm formation, pheromone induction of OG1RF/pCF10 gives rise to thicker biofilms than with OG1RF without the plasmid (20). Given that AS-expressing OG1RF/pCF10 cells clump after 2 h of pheromone exposure, and those clumps largely exclude Ebp-expressing cells, we explored the possibility that initial biofilm attachment is facilitated by AS-mediated clumping to form microcolonies, while Ebp may be important to enhance biofilm maturation. We therefore compared the number of Ebp-expressing cells between OG1RF and OG1RF/pCF10 that have attached and been incorporated into early biofilms after 2 h of pheromone exposure by scraping the cells from biofilm chambers and performing an indirect immunofluorescence assay (IFA) with an EbpC antibody. Twelve percent of OG1RF/pCF10 and 22% of OG1RF early biofilm cells displayed Ebp on their cell surface, suggesting that if AS is present and induced, it will promote the initial attachment of biofilm without the aid of Ebp-expressing cells (Fig. 4A and S3). To further explore the contributions of Ebp and AS, together and individually, to biofilm development, we assayed biofilm formation by OG1RF (Ebp-expressing [Ebp ϩ ], AS-nonexpressing [AS Ϫ ]), OG1RF/pCF10 (Ebp ϩ AS ϩ ), OG1RF ΔebpABC srtC (Ebp-nonexpressing [Ebp Ϫ ], AS Ϫ ) and OG1RF ΔebpABC srtC/pCF10 (Ebp Ϫ AS ϩ ) strains (Fig. 4B). Consistent with previous reports (2), we observed a 30% reduction in biofilm formation by OG1RF ΔebpABC srtC compared to OG1RF and a 54% increase in the biomass of the induced OG1RF/pCF10 strain that expresses both AS and Ebp compared to OG1RF (20) (Fig. 4B). Moreover, we observed similar biofilm biomass for OG1RF and induced OG1RF ΔebpABC srtC/pCF10, suggesting that in the absence of Ebp, AS alone is sufficient to revert the biomass of the ebpABC-null strain to wild-type levels (Fig. 4B). To determine how Ebp and AS contribute to biofilm structure and organization, we grew biofilms for 24 h in the presence of cCF10, stained the DNA of all the cells with Hoechst dye, and performed confocal laser scanning microscopy. We observed distinct differences in the biofilm structure and thickness between the strains that express one, both, or neither of the adhesins (Fig. 5). In the absence of AS, OG1RF forms uniform and tightly packed biofilms, while AS-expressing OG1RF/pCF10 cells formed additional 3-dimensional microcolonies, suggesting that AS promotes the formation of structured biofilm (Fig. 5). Moreover, while we observed uniform tightly packed biofilms in Ebp-expressing OG1RF and OG1RF/pCF10, the ΔebpABC srtC and ΔebpABC srtC/pCF10 Ebp-null mutant strains appeared to be sparsely packed (Fig. 5), suggesting that pili are important for tight interactions within biofilms. Furthermore, we observed that AS drives microcolony formation through cellular clumping, since we noticed elevated microcolony clusters only for pCF10-containing AS-expressing strains (Fig. 5B). Finally, pili contributed to the development of thicker biofilms in OG1RF and AS-expressing OG1RF/pCF10 strains than that in their Ebp-null counterparts (Fig. 5B). We therefore conclude that despite accumulating similar overall biofilm biomass (Fig.  4B), OG1RF and OG1RF ΔebpABC srtC/pCF10, displaying opposite repertoires of Ebp and AS, exhibit divergent biofilm development, suggesting that the two adhesins differentially contribute to biofilm structure.

DISCUSSION
Although most bacteria encode a variety of adhesins for surface attachment and colonization, the number and variety of adhesins expressed in the population and on each cell can vary to maximize adhesive capacities but avoid immune clearance and cross-adhesin interference (23,29). In E. faecalis V583, 17 of 41 predicted sortase A substrate proteins contain MSCRAMM motifs, and at least 7 are expressed within the human host, as antibodies for them are readily detected in the serum (5). This observation raises questions as to when, to what extent, and how promiscuously each adhesin is expressed within an enterococcal population. In the present study, we addressed the relationship between two of these enterococcal adhesins, Ebp and AS.
We showed that Ebp and AS can be coexpressed on the same cell, but that known regulatory mechanisms for each adhesin (EbpR and cCF10, respectively) do not influence the expression of the other. However, we found that at low cell density during planktonic growth, simultaneous expression of AS and Ebp interferes with AS-mediated intercellular clumping and reduces HGT rates by 10-fold, presumably due to steric interference by protruding Ebp, which could prevent AS from binding to its receptor,

Relationships between AS and Ebp in E. faecalis
Journal of Bacteriology lipoteichoic acid (LTA), on neighboring cells (30). Similarly, E. coli type 1 fimbriae can prevent cellular aggregation mediated by self-associating autotransporters (SAAT), suggesting that bulky surface proteins, such as pili, interfere with aggregation and physically mask the function of SAAT-type proteins (31,32). Therefore, pilus interference with aggregation might be a more general mechanism across bacteria than previously appreciated. Since E. faecalis pili are only expressed on a subset of cells, maintaining a subset of unclumped piliated cells after AS induction may benefit the overall population by limiting the extent to which HGT can occur. Ebp interference may be a general mechanism to protect planktonic cells from abundant plasmid intake, which in turn could contribute to genome stability and increased fitness (33). Notably, we can only detect Ebp-dependent differences in HGT rates at early time points, and the difference disappears 2 h after mating. Similarly, Bhatty et al. reported no difference in plasmid transfer rates from OG1RF or Ebp-null mutant donors after 2 h of mating, but they did not examine earlier time points, as we did in this study (20). Recently, Ebp on OG1RF donor cells was shown to confer a 1.6-fold increase in conjugative transfer of pAM␤1, after 5 h of mating, compared to ebp-null donor cells (28). While different plasmids, mating times, and experimental cell densities may explain the differing conclusions about the contribution of Ebp to plasmid transfer in these previous studies, coupled with our data, a model emerges in which Ebp may interfere with HGT at low cell densities but later promote HGT when Ebp-dependent biofilm aggregates start to form, which may instead facilitate plasmid transfer. The reasons for these densitydependent differences remain to be determined. In addition to its impact on HGT, we also demonstrate that coexpression of Ebp and AS may have unique functional contributions to biofilm development. Previously, it was shown that AS acts cooperatively with environmental DNA (eDNA) to promote biofilm formation, resulting in thicker biofilms (20). Here, we extend that model and propose that AS and Ebp differentially contribute to the structure of the biofilm, where ASmediated clumps of cells that are not expressing Ebp give rise to three-dimensional (3D) microcolonies that may be enriched for eDNA. Our data suggest that Ebpexpressing cells can then "fill in" the biofilm space, leading to a more densely packed biofilm, as well as contribute to the development of thicker 3D biofilms.
Importantly, we observed similar biomass of OG1RF and an Ebp-null strain that expresses AS, where AS restored biofilm biomass to OG1RF levels even in the absence of pili. However, despite similarities in biomass accumulation, biofilms formed in the presence of only Ebp or AS differed structurally. Taken together, our findings suggest that these adhesins may complement each other during biofilm formation and simultaneously contribute different structural roles during biofilm development.

MATERIALS AND METHODS
Bacterial strains and growth conditions. The Enterococcus faecalis strains used in this study are listed in Table 1. For planktonic growth, strains were grown statically in brain heart infusion (BHI) medium (BD Difco, USA) or on BHI agar (BHI supplemented with 1.5% agarose [1st Base, Singapore]) at 37°C. For biofilm assays, tryptone soy broth (Oxoid, UK) supplemented with 10 mM glucose (TSBG) was used. Antibiotics were used at the following concentrations, where appropriate: tetracycline (Tet), 15 g/ml; rifampin (Rif), 25 g/ml; and streptomycin (Str) 500 g/ml.
HGT assay. OG1SS/pCF10 donor cells were grown overnight in BHI medium and then diluted to OD 600 of 0.1 in 5 ml of fresh BHI medium in the presence of 0.12 ng/ml cCF10 peptide. After 2 h of shaking at 200 rpm at 37°C, visible clumps formed on the bottom of the tube. The top 2 ml represented the unclumped (suspension) fraction, and 1 ml from the bottom represented the clumped fraction. EDTA at a final concentration of 0.05 M was added to each fraction and vortexed for 15 s to separate clumped cells (19). Both fractions were normalized to OD 600 of 1 and washed with phosphate-buffered saline (PBS). Donor cells (from the suspension or clumped fraction) were mixed with recipient OG1RF ΔebpABC srtC cells at a ratio of 1:10 (donor/recipient) and incubated statically at 37°C. After 30 min and 120 min, 1 ml was removed, vortexed for 15 s, serially diluted, and plated on selective plates to quantify donor OG1SS/pCF10 cells on BHI-Str-Tet medium and transconjugants OG1RF ΔebpABC srtC/pCF10 on BHI-Rif-Tet medium. Alternatively, OG1RF/pCF10 and OG1RF ΔebpABC srtC donor cells were grown as described above and were mixed with OG1X ΔsrtC recipient cells at a ratio of 1:10 (donor/recipient). Conjugation was carried out as described above, and cells were plated on selective plates to quantify donor OG1RF/pCF10 or OG1RF ΔebpABC srtC/pCF10 cells on BHI-Rif-Tet agar and OG1X ΔsrtC/pCF10 transconjugants on BHI-Str-Tet agar. HGT rates were calculated as the number of transconjugants per donor cell (Tc/donor), as described previously (20).
Generation of polyclonal antisera. Recombinant protein fragments were designed, expressed, and purified using the Protein Production Platform (NTU, Singapore), using a technology and workflow from the former biotechnology unit at the Structural Genomics Consortium (Karolinska Institutet, Sweden). Briefly, the AS target (nucleotides 550 to 1100 from prgB [NCBI RefSeq accession no. NC_006827.2] [12883.0.16800]) and Ebp target (nucleotides 1 to 627 from NCBI RefSeq accession no. NC_017316.1 [911134.0.913017]) were amplified using a multiconstruct approach to generate 184-amino-acid and 209-amino-acid gene products, respectively, and were cloned in pNIC28-Bsa4 with an N-terminal His tag followed by a TEV protease cleavage site. The resultant plasmid was then transformed to E. coli BL21(DE3), and recombinant protein was expressed following overnight induction with isopropyl ␤-D-1-thiogalactopyranoside. Cells were lysed, and recombinant protein was purified by immobilized metalaffinity chromatography (IMAC) using the His tag, followed by size-exclusion chromatography. The purity of the recombinant protein was assessed by SDS-PAGE and the mass verified by mass spectrometry. Polyclonal antisera were generated commercially (SABio, Singapore) by immunization of rabbits or guinea pigs with purified recombinant AS or EbpC, respectively. The specificities of the immune sera were confirmed by the absence of signal on Western blots of whole-cell lysates from OG1RF ΔebpABC srtC or a pCF10-free/uninduced strain (Fig. S1).
Immunofluorescence microscopy. To image AS and Ebp on planktonically grown E. faecalis, cells were normalized to an OD 600 of 1.0, washed once with PBS, and then fixed in 4% paraformaldehyde (PFA) for 20 min at 4°C. Cells were again washed, diluted in PBS, and spread on poly-L-lysine-precoated slides (Polysciences, Inc., USA). The slides were then blocked with 2% bovine serum albumin (BSA) in PBS and incubated at room temperature (RT) for 20 min. After blocking, the slides were washed with PBS, and 20 l of primary antibody (guinea pig anti-Ebp serum and mouse anti-AS serum) was then added to the slides, followed by incubation at 4°C overnight. The next day, the slides were washed 3 times with PBS, and 20 l of fluorescence conjugated secondary antibody (Alexa Fluor 568 -goat anti-guinea pig antibody for Ebp and Alexa Fluor 488 -goat anti-rabbit antibody for AS [Invitrogen, Inc., USA]) was added and incubated in the dark for 1 h at RT. Slides were then washed, mounted with VectaShield mounting medium (Vector Laboratories, Inc., USA), and visualized on an inverted epifluorescence microscope (Axio Observer Z1; Zeiss, Germany) or Zeiss Elyra PS.1 for colocalization studies. The percentages of EbpC-and AS-stained cells were determined by comparing total number of cells viewed under phase contrast to the number of stained cells on the merged fluorescent/phase image.