Involvement of Chromosomally Encoded Homologs of the RRNPP Protein Family in Enterococcus faecalis Biofilm Formation and Urinary Tract Infection Pathogenesis

Enterococcus faecalis causes health care-associated infections and displays resistance to a variety of broad-spectrum antibiotics by acquisition of resistance traits as well as the ability to form biofilms. Even though a growing number of factors related to biofilm formation have been identified, mechanisms that contribute to biofilm formation are still largely unknown. Members of the RRNPP protein family regulate a diverse set of biological reactions in low-G+C Gram-positive bacteria (Firmicutes). Here, we identify three predicted structural homologs of the RRNPP family, EF0073, EF1599, and EF1316, which affect biofilm formation and CAUTI pathogenesis.

ABSTRACT

Enterococcus faecalis is an opportunistic pathogen capable of causing infections, including endocarditis and urinary tract infections (UTI). One of the well-characterized quorum-sensing pathways in E. faecalis involves coordination of the conjugal transfer of pheromone-responsive plasmids by PrgX, a member of the RRNPP protein family. Members of this protein family in various Firmicutes have also been shown to contribute to numerous cellular processes, including sporulation, competence, conjugation, nutrient sensing, biofilm formation, and virulence. As PrgX is a plasmid-encoded RRNPP family member, we surveyed the genome of the multidrug-resistant strain V583 for additional RRNPP homologs using computational searches and refined those identified hits for predicted structural similarities to known RRNPP family members. This led us to investigate the contribution of the chromosomally encoded RRNPP homologs to biofilm processes and pathogenesis in a catheter-associated urinary tract infection (CAUTI) model. In this study, we identified five such homologs and report that 3 of the 5 homologs, EF0073, EF1599, and EF1316, affect biofilm formation as well as outcomes in the CAUTI model.

IMPORTANCE Enterococcus faecalis causes health care-associated infections and displays resistance to a variety of broad-spectrum antibiotics by acquisition of resistance traits as well as the ability to form biofilms. Even though a growing number of factors related to biofilm formation have been identified, mechanisms that contribute to biofilm formation are still largely unknown. Members of the RRNPP protein family regulate a diverse set of biological reactions in low-G+C Gram-positive bacteria (Firmicutes). Here, we identify three predicted structural homologs of the RRNPP family, EF0073, EF1599, and EF1316, which affect biofilm formation and CAUTI pathogenesis.

INTRODUCTION

Enterococcus faecalis is a Gram-positive commensal of most mammalian gastrointestinal tracts, including that of humans, and its origin in animal digestive systems dates back to the terrestrialization of land animals nearly 500 million years ago (1). This evolutionary history has likely allowed enterococci to also adapt to harsh environmental conditions present in clinical settings (desiccation, starvation, antimicrobial therapy, and disinfection), contributing to its rise as a hospital-associated pathogen. This association as a nosocomial pathogen is further bolstered by the ready acquisition of antibiotic resistance through horizontal gene transfer events, rendering it a formidable and serious threat to public health (https://www.cdc.gov/drugresistance/biggest-threats.html).

As a leading opportunistic pathogen in hospitals, E. faecalis causes bloodstream infections, surgical site infections, and urinary tract infections (UTI) (2). E. faecalis predominantly utilizes biofilm formation as one of the strategies to colonize its human host at both commensal and infectious sites (3–5). Gene products identified to date that contribute to biofilm formation are primarily involved in early-stage adhesion events in the developmental process and include the Ebp pilus (6–10) and surface adhesins involved in cell aggregation during conjugal transfer of pheromone-responsive plasmids (11). PrgX, which controls the synthesis of E. faecalis conjugal transfer adhesins, is a prototype of the RRNPP protein family and the first for which a protein structure was solved (12, 13). The broader RRNPP family (14–16) is comprised of the prototype Rap phosphatases and the DNA binding transcription factors Rgg, NprR, PlcR, and PrgX. An attribute of this protein family is the ability to engage in quorum sensing of small linear peptides through tetratricopeptide repeats (TPRs) present in the C-terminal domain. The TPR forms a structural motif via tandem repeats of 34-amino-acid patterns forming an alpha helix-rich solenoid region (17). Members of the RRNPP family have different numbers of TPRs (Rap phosphatase, 7; NprR, 9; PlcR, 5) (13, 18–20) and play a major role in protein dimerization/oligomerization and peptide binding. PrgX (13) and Rgg (20) proteins possess a TPR-like domain that is rich in alpha helices and forms a similar secondary and tertiary structure facilitating protein dimerization/oligomerization and peptide binding. TPR and TPR-like domains are known to possess key amino acids, including asparagine residues that facilitate peptide binding and adjacent hydrophobic amino acids that form a pocket for peptide docking, as well as key features important for dimerization and tetramerization (13, 18, 19). Bioinformatic searches with the primary amino acid sequence of RRNPP family proteins that are transcriptional activators, including PrgX as recent evidence now suggests that it also responsible for transcriptional activation when bound to cCF10 (21), reveal that they all have a well conserved N-terminal helix-turn-helix (HTH) DNA binding domain belonging to the Cro/Ci family of DNA binding domains. Rap phosphatases, which lack DNA binding activity, possess a 3-helix bundle instead. Members of the RRNPP protein family have been shown to contribute to a wide array of cellular processes, including sporulation and competence by Rap phosphatases in Bacillus spp. (22), as well as toxin production (mediated by PlcR) (23) and protease production (controlled by NprR) (24) in the Bacillus cereus group. Work in Enterococcus (PrgX) and Streptococcus (Rgg) has shown that RRNPP members contribute to conjugation control (25), biofilm formation (11, 26), and virulence (27). It is noteworthy that this family of proteins regulate such a diverse set of biological functions, given that they all possess similar N- and C-terminal protein arrangements and a common peptide recognition system. Hence, it is necessary to understand how different proteins in this family regulate their molecular targets. Table S1 in the supplemental material briefly describes the mechanism by which these RRNPP proteins regulate various functions, emphasizing structural features and their cognate peptide maturation pathway.

Because of the broad conservation of the RRNPP family among the Firmicutes and knowing that PrgX is a plasmid-encoded trait, we hypothesized that additional family members would be present on the chromosome of E. faecalis, similar to what is observed in Bacillus spp. and Streptococcus spp. The annotated gene nomenclature of V583 gene loci has recently been updated. However, for consistency, we have continued to refer to the older annotation using the previous nomenclature and have mentioned the new gene locus tags in Table S2 in the supplemental material. Previous bioinformatic analysis, using streptococcal Rgg homologs as the query, predicted seven putative RRNPP homologs encoded in the E. faecalis V583 genome (28, 29), with two of those seven residing on the pheromone-responsive plasmids, pTEF1 and pTEF2, present in V583 and that appear to be functionally related to PrgX. One of the chromosomal homologs, ElrR (EF2687), has previously been shown to be involved in resistance to phagocytic uptake by macrophages through positive regulation of the leucine-rich protein ElrA (30). The roles of ElrR and the other chromosomally encoded RRNPP homologs in biofilm formation have not been characterized to date. It was therefore of interest to determine whether these putative regulators contribute to biofilm development and pathogenesis in the catheter-associated UTI (CAUTI) model.

In this study, we investigated a role for each of the putative chromosomally encoded RRNPP homologs in biofilm formation. Here, we report the effect of 3 of the 5 RRNPP homologs (EF0073, EF1599, and EF1316) on biofilm formation using a drip-flow biofilm reactor (DFBR), as well as their contributions to CAUTI pathogenesis.

RESULTS

The Enterococcus faecalis V583 genome encodes 7 structural RRNPP homologs.

It has been previously reported that there are 5 chromosomally encoded Rgg-like proteins in the E. faecalis genome (28, 29) along with plasmid-encoded homologs present on each of the pheromone-responsive plasmids (pTEF1 and pTEF2) resident in V583. The initial study utilized streptococcal Rgg systems as prey sequences to identify putative Rgg homologs (29). To get a more comprehensive view of RRNPP family members in E. faecalis, we queried the V583 genome sequence with representative RRNPP family sequences (Streptococcus pyogenes Rgg, Bacillus cereus PlcR, and E. faecalis PrgX and ElrR) using BLASTP. These queries yielded 44 putative proteins that belonged to the Cro/CI family of transcription factors based on similarity to a conserved helix-turn-helix DNA binding domain. To further delineate these protein predictions, we performed a DELTA-BLAST (domain enhanced-lookup time accelerated Basic Local Alignment Search Tool) search using the same query proteins (Rgg, PlcR, PrgX, and ElrR) with a default query coverage of 80% and three iterations to better define position-specific amino acid alignments. This refinement narrowed down the number of protein candidates to 7. We also performed PHYRE2 (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index) prediction on these 7 putative homologs, and all 7 were found to exhibit structural homology to known RRNPP family proteins at over 96% confidence (Table 1). EFA_0004 and EFB_0005 are encoded on plasmids pTEF1 and pTEF2, respectively. The remaining 5, i.e., EF0073, EF1224, EF1316, EF1599, and EF2687 (ElrR), are encoded on the chromosome, and this prediction is in agreement with the analysis performed by Fleuchot et al. (29), in which they identified the same set of proteins using only Rgg as the query.

TABLE 1.

PHYRE2 protein structure prediction for the putative PrgX homologs

V583 gene locus	New gene locus	% confidence for homology to RRNPP	% coverage
EFB_0005	EF_RS15800	100	97
EF1224	EF_RS05915	100	93
EF1316	EF_RS06340	100	97
EF0073	EF_RS00325	100	96
EFA_0004 (traA)	EF_RS16180	100	94
EF1599	EF_RS07705	100	97
EF2687 (elrR)	EF_RS12740 (elrR)	96.7	94

The chromosomal homologs that were predicted in V583 from the analysis were conserved among various sequenced strains of E. faecalis, including the commonly used plasmid-free strain OG1RF, and were not found to be associated with known plasmid remnants integrated into the genome (31), the pathogenicity island (32), or prophage elements (33), suggesting that these gene products are part of the core genome as they were not localized to annotated mobile elements. It was therefore of interest to understand the phylogenetic relationship between these homologs and the archetypical members of the RRNPP family of proteins. As shown in Fig. 1, the phylogenetic tree corroborated the previous finding that NprR and RapA phosphatase were not related phylogenetically to Rgg, PrgX, or PlcR and grouped separately with 100 bootstrap support (15). The tree also revealed that all the V583 RRNPP homologs likely clustered independently of PlcR with 62.1 bootstrap support. One interesting note from this analysis is that EF1316, EF1599, and EF1224 appear to cluster independently from Rgg and PrgX, whereas EF0073, EF2687, EFA_0004, and EFB_0005 strongly cluster with Rgg and PrgX with 87.4 bootstrap confidence. The tree also showed that EFB_0005 was closely related to PrgX with high bootstrap support (90.5) and reiterated the finding that EF2687 (ElrR) grouped closely with Rgg, albeit with modest bootstrap support (61.1) (28, 29). Overall, this analysis shows the probable evolutionary relationship between RRNPP proteins in E. faecalis.

FIG 1.

Phylogenetic analysis of E. faecalis V583 RRNPP proteins. A RAxML phylogenetic tree was constructed using the maximum-likelihood method. The protein sequences of homologs and the RRNPP family of proteins were aligned using ClustalW. The branch distance of 1.0 represents 1 amino acid substitution per site.

sORFs adjacent to RRNPP homolog genes.

It is well established that binding to cognate peptides is an integral part of how RRNPP proteins regulate cellular responses (18, 24, 34). In many instances, the genes encoding the RRNPP family members and their corresponding propeptides are closely positioned within their genomes. These propeptides are either cotranscribed, as in the case with Rap phosphatases and the inhibitory Phr peptides in B. subtilis (35), oriented in the same direction, as in the case of PlcR and its associated peptide PapR (36), or divergently transcribed, as in the case of prgX and the gene encoding the inhibitory peptide iCF10, prgQ (34), as well as Rgg2/3 and the cognate peptides Shp2/3 in S. pyogenes (26). We examined the genetic region around the chromosomally encoded enterococcal RRNPP family members to identify small open reading frames (sORFs). As shown in Fig. 2A, we were able to identify predicted sORFs adjacent to ef0073, ef1316, and ef1599 that are divergently transcribed from the RRNPP homolog. There was no predicted sORF found adjacent to either ef1224 or elrR (28). Amino acid lengths of peptide precursors encoded by RRNPP archetypical members range from 8 to 48 amino acids (aa) (PhrR, 38 to 44 aa; SIP and Shp, 8 to 23 aa; NprX, 27 aa; PapR, 48 aa; PrgQ, 23 aa) (24, 26, 34, 37, 38), and these peptides are processed upon secretion to their active forms, which range in size from 5 to 8 amino acids. Similarity to the lengths of the sORF adjacent to the RRNPP homologs in E. faecalis predicts that these 3 are likely to encode functional peptides (EF1315 [EF_RS16545], 58 aa; EF0072, 15 aa; and EF1600, 20 aa).

FIG 2.

Small open reading frames encoded adjacently to RRNPP homologs. (A) Schematic representation of chromosomal RRNPP homolog genes (with gene lengths in base pairs) and their putative peptides (with gene and predicted amino acid lengths). The spacing between the predicted start codons of the RRNPP-encoding genes and the putative peptide genes is shown, and the computationally predicted ribosome binding sites (consensus sequence of AGGAGG) for the putative peptides are highlighted in red. (B) Quantification of the LacZ expression using modified Miller assays from strains grown to mid-log and stationary phases. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical analysis was performed using Student’s two-tailed unpaired t test with confidence intervals of 95%. In all figures: *, 0.01 < P < 0.05; **, 0.001 <P < 0.01; ***, P < 0.001.

To determine whether these sORFs (putative peptides) are expressed, we generated transcriptional fusions to LacZ and performed modified Miller assays to assess relative expression during logarithmic and stationary-phase growth. We chose this approach because quantitative reverse transcription-PCR (qRT-PCR) as a means for assessing gene expression was not feasible for such short open reading frames. LacZ expression was observed only in strains containing the promoter-fusion plasmids and not in the strain harboring the empty vector (pKS12A) (Fig. 2B), suggesting that ef0072, ef1315, and ef1600 are expressed. ef0072 was seen to possess the strongest promoter activity, producing 20 Miller units of LacZ expression during the stationary phase, with ef1315 and ef1600 producing ∼5 and 4 Miller units, respectively (Fig. 2B). However, ef0072 (40 Miller units) and ef1315 (10 Miller units) promoter expression doubled during the exponential phase, as seen in Fig. 2B. It was surprising that ef1600 promoter expression did not change drastically during logarithmic phase (5.2 Miller units) compared to stationary phase (Fig. 2B).

The SignalP 5.0 peptide processing and secretion server was used to predict the possibility of processing by signal peptidase 1 (SP1) or SP2 and secretion of these peptides by Sec/Tat pathways (39). With all three putative peptides, i.e., EF0072, EF1315, and EF1600, a sequence pattern consistent with processing by either SP1 or -2 and secretion by the canonical Sec pathway was not observed. Other known cognate peptides associated with the RRNPP family, such as iCF10 and SHP2, showed similar outcomes and were also not predicted to be dependent on the Sec pathway.

RRNPP isogenic mutants do not exhibit planktonic growth differences.

We sought to understand the physiological role of these RRNPP homologs in E. faecalis. Markerless deletion mutants of the predicted chromosomally encoded RRNPP genes were constructed to compare their growth and biofilm characteristics. Growth of the mutants and the parental strain in MM9YEGC confirmed that there were no identifiable planktonic growth differences (see Fig. S1 in the supplemental material) compared to the V583 parental strain.

Multiple RRNPP homologs affect biofilm formation and urinary tract colonization.

Our previous interest in biofilm mechanisms led us to investigate the contribution of these RRNPP proteins to biofilm formation. The effect of RRNPP homologs on biofilm formation was assessed using directed competition assays. V583 and RRNPP mutant strains were fluorescently labeled by stable integration of genes encoding either mCherry or teal fluorescent protein (TFP) ectopically in the genome. Figure S2 in the supplemental material shows a view of a Todd-Hewitt broth (THB) agar plate containing equal concentrations of parental strain V583 labeled with mCherry fluorescent reporter and the V583 Δef0073 strain labeled with teal fluorescent reporter. Similar analysis was done with all mutants (data not shown) to confirm that the presence of either reporter (mCherry or TFP) did not alter the growth of the organism during planktonic growth. We next examined the contribution of the chromosomally encoded RRNPP mutants to biofilms. We compared each mutant along with the parental V583 strain in static “plate-based” biofilms as well as by using competition assays in a drip-flow biofilm reactor. Figure S3 in the supplemental material shows that in a plate-based assay (noncompetitive), only V583 Δef0073 showed a significant change in its biofilm phenotype and resulted in enhanced biofilm formation compared to that of the parental strain. V583 Δebp (ebpA to -C deletion mutant) was used as a negative control, as this strain lacks a surface pilus previously shown to be involved with the initial attachment phase in biofilm development (9), and, as expected, showed a significant decrease in biofilm formation. Drip-flow conditions were chosen to closely mimic flow rates known to be present in catheterized patients (40). We directly enumerated the CFU to get a more quantitative measure of the differences exhibited by mutation of the RRNPP homologs (Fig. 3A). As expected, the ebp mutant decreased 26-fold relative to the parental strain. In contrast, ef0073 and ef1599 deletion mutants showed increases of 2.7-fold and 2-fold relative to the parental strain, respectively. Of the 5 chromosomal RRNPP homologs, only the ef1316 deletion mutant displayed a biofilm defect resulting in a 4-fold decrease compared to the parental strain. The remaining two mutants with deletions of ef1224 and ef2687 (elrR) did not show significant differences from the parental strain in biofilm competition assays. For the RRNPP mutants showing phenotypic differences, we were able to fully complement those observed changes, as reintroduction of the deleted gene at the native locus restored biofilm formation of the respective mutants back to wild-type levels.

FIG 3.

Multiple RRNPP homologs affect biofilm formation and urinary tract colonization. (A) Competitive indices of teal-labeled mutant (and the respective complement) biofilms grown using the drip-flow biofilm reactor compared to mCherry-labeled V583 strains, performed in a mixed biofilm environment. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical analysis was performed using Student’s two-tailed unpaired t test with confidence intervals of 95%. No asterisks, not significant. (B and C) Competition experiments performed using the CAUTI murine model with mCherry-labeled strain V583 and teal-labeled RRNPP mutant strains in bladders (B) and catheters (C) from infected mice. Statistical significance was evaluated according to the Kruskal-Wallis test.

We were also able to observe similar enrichments in urinary tract colonization by the Δef0073 and Δef1599 mutants using the CAUTI in vivo model, as bacteria recovered from implanted catheters and bladder tissue paralleled differences observed in the drip-flow condition (Fig. 3B and C). The Δef0073 and Δef1599 mutants were increased in number compared to the parental strain during the course of a 2-day infection, displaying 2.9-fold and 10-fold increases in bladder tissue, respectively. These same mutants were also more abundant on the catheter surface, showing 6-fold (Δef0073) and 31-fold (Δef1599) increases compared to the parental strain. In contrast, the Δef1316 mutant was reduced in number on the catheter and bladder tissue, displaying an 8.5-fold decrease on the catheter and a 7.0-fold decrease from bladder tissue relative to the parental strain in the in vivo competition assay. These changes with respect to bladder and catheter colonization were found to be statistically significant, with P values of <0.0001 (Kruskal-Wallis test).

Expression of RRNPP homologs across growth phases.

To assess the expression of these RRNPP homologs across various stages of growth, including stationary phase and exponential phase, as well as cells harvested from a 3-day drip-flow biofilm, we performed a quantitative gene expression study to determine the level of expression of these homologs compared to that of the housekeeping gyrB gene. As observed in Fig. 4A, relative to gyrB expression, all three RRNPP homologs were expressed less abundantly. The expression of ef0073 was 6.5-fold less than that of the endogenous control gyrB, whereas ef1599 and ef1316 showed 14- and 65-fold drops in expression relative to that of gyrB, respectively. During exponential-phase growth, we observed an 8-fold increase in expression for ef1316 compared to that in stationary phase. In contrast, expression for ef0073 did not change, and ef1599 increased only 2-fold relative to stationary phase (Fig. 4B). The expression of ef0073, ef1316, and ef1599 from cells harvested from a biofilm state was 4-, 13-, and 2-fold more, respectively, than their stationary-phase expression.

FIG 4.

Expression of RRNPP homologs across growth phases. (A) qPCR of RRNPP homolog genes on RNA isolated from parental strain V583 grown to stationary phase. gyrB was used as the endogenous constitutively expressed control, and expression of RRNPP genes was normalized to gyrB expression. Relative fold expression values were calculated by the ΔC_T method. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical significance was evaluated by using one-way ANOVA, and Bonferroni’s test was performed post hoc. (B) qPCR of RRNPP homolog genes on RNA isolated from parental strain V583 grown to mid-log phase and under drip-flow biofilm conditions. Expression of RRNPP genes was normalized to gyrB expression in the respective phases and expression values of RRNPP genes in stationary phase. Relative fold expression values were calculated by the ΔΔC_T method. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical significance was evaluated by using one-way ANOVA for individual bars, and Bonferroni’s test was done post hoc.

RRNPP homologs affect biofilms at different stages of development.

Since the impact on biofilm development by the deletion mutants of RRNPP homologs was observed in a drip-flow setting, we hypothesized that a time course quantification of biofilm using the drip-flow system would identify specific stages in development at which parental and mutant biofilms diverged. As the drip-flow experiment spanned 72 h, we chose 0 h, which is immediately before initiation of drip-flow, and 4, 8, 12, 24, 36, 48, 60, and 72 h as time points after initiation of flow to evaluate the stage of biofilm formation at which the differences begin to appear. As seen in Fig. 5, we did not observe any difference in biofilms at the earliest stage (0 and 4 h). The impact of ef1316 mutations on biofilm formation was evident as early as 8 h after flow initiation, where there was approximately a 2-fold decrease in the competitive index. The competitive indices at later time points for ef1316 mutants were reduced 2.5- to 5-fold compared to those of the V583 parental strain. The Δef0073 and Δef1599 mutants began to show changes in competitive indices only after 24 h of flow conditions, as the competitive indices were consistently more than 2-fold and reached as high as 3.2-fold compared to parental biofilm growth. The biofilm differences observed after 8 h for the Δef1316 mutant and after 24 h in the case of the Δef0073 and Δef1599 mutants were observed to be statistically significant using a two-way analysis of variance (ANOVA). These data suggest that EF1316 regulates genes important for early-phase biofilm development, whereas EF0073 and EF1599 are involved in later biofilm phases.

FIG 5.

Time course biofilm quantification of RRNPP mutants. Competitive indices of teal-labeled mutant (and the respective complement) biofilms grown using the drip-flow biofilm reactor compared to mCherry-labeled V583 strains were determined in a mixed biofilm environment at different time points after the adherence of bacteria to the glass slide and start of flow. Competitive indices were calculated at 0 (after adherence), 4, 8, 12, 24, 36, 48, 60, and 72 h after flow initiation. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical significance was analyzed using 2-way ANOVA and was found to be significant (for the Δef1316 mutant after 8 h and for the Δef0073 and Δef1599 mutants after 24 h) with a P value of <0.0001.

RRNPP mutants affect biofilm formation in the absence of competition.

As the change in biofilm phenotype was observed with RRNPP mutants in competition assays, we examined the biofilm formation of these mutants under monopopulation drip-flow growth conditions and compared their biofilm formation to that of the parental strain using confocal microscopy (Fig. 6A). ImageJ software was used to visualize the Z-stacks using direct and orthogonal views, and COMSTAT was used to quantify the maximum thickness of the biofilm. The results demonstrate that the overall biofilm thickness is reduced 1.5-fold in the ef1316 mutant compared to the parental strain, whereas the Δef0073 and Δef1599 mutants have 1.5- and 2-fold-increased biofilm formation, respectively (Fig. 6C). We also took a direct measure of the wet mass of the biofilm from V583 and the mutants for ef0073, ef1316, and ef1599 and found that a similar trend of an overall decrease in the biomass of biofilms was observed for the Δef1316 mutant (1.8-fold decrease), whereas biofilms formed by the Δef0073 (1.6-fold increase) and Δef1599 (1.3-fold increase) mutants displayed an overall increase in biofilm biomass (Fig. 6B).

FIG 6.

RRNPP mutants affect biofilm formation in the absence of competition. (A) Confocal images of 3-day biofilms formed by E. faecalis V583 (i), V583 Δef0073 (ii), V583 Δef1316 (iii), and V583 Δef1599 (iv). Strains express the teal fluorescent reporter. Z-stacks were analyzed using orthogonal views in ImageJ software for the xz and yz plane images showing the thickness of the biofilm. (B) Net wet mass of the biofilm directly measured after scraping the biomass into a microcentrifuge tube, plotted as grams of biofilm per microscopic slide. (C) Z-stack analysis done using the COMSTAT plugin and maximum thickness of the area of the Z-section of biofilm calculated. Statistical analysis was done over three different planes of analysis. Values shown are the averages from 3 biological replicates for Z-stack analysis and 9 biological replicates for net wet mass estimation. Error bars represent standard errors of the mean. Statistical analysis was performed using Student’s two-tailed unpaired t test with confidence intervals of 95%.

RRNPP homologs regulate biofilms independent of attachment, eDNA, lysis, and Epa polysaccharide.

The literature so far points to the involvement of adhesions or surface proteins such as Ebp pili as an important marker for biofilm formation in E. faecalis (9, 10, 41–44). Another trait known to affect biofilm formation is the ability of cells to autolyze, mediated by the major autolysin AtlA (45, 46), to allow establishment of an extracellular DNA (eDNA) matrix in the early stages of biofilm development (47). Epa cell wall carbohydrates have also been shown to contribute to biofilm development (48, 49). We therefore wanted to assess the involvement of common mechanisms that may be involved in biofilm regulation by these homologs. To investigate if these homologs regulated an adhesin and altered biofilm patterns via an attachment defect or enrichment, an attachment assay using the drip-flow biofilm reactor was performed. All the strains displayed similar attachment phenotypes after allowing the bacterial cells to adhere, except for V583 Δebp(A-C) which showed its documented reduction in the attachment phase of biofilm development, as represented in Fig. 7.

FIG 7.

Attachment phenotype of RRNPP mutants. The assay was performed with fluorescently labeled V583::mCherry and V583 Δef0073::Teal, V583 Δef1316::Teal, V583 Δef1599::Teal, and V583 Δebp::Teal. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical analysis was performed using Student’s two-tailed unpaired t test with confidence intervals of 95%.

It is known that eDNA is an important component of the biofilm matrix and contributes to biofilm maturation. When eDNA release assays were conducted, no significant eDNA depletion or enrichment was seen between the mutants and the parental strain (Fig. 8A). A significant decrease was seen, however, in the previously investigated atlA mutant strain (45). This observation was confirmed by conducting autolysis assays with the respective mutants. We observed similar rates of lysis for the parental strain and the RRNPP mutants (Fig. 8B); only the aforementioned atlA mutant displayed altered rates of lysis. Analysis of cell wall polysaccharides extracted from parental and mutant strains failed to identify any alterations in polysaccharide abundance; notably, levels of the Epa and Cps polysaccharides appeared to be similar across all strains examined (Fig. 9). While these assays fail to show significant correlation of these mutants toward known biofilm formation mechanisms, they reveal that EF0073, EF1599, and EF1316 may contribute to biofilm formation in novel ways.

FIG 8.

Autolysis and eDNA release by RRNPP mutants. (A) eDNA release assay with supernatants of strains V583, V583 Δef0073, V583 Δef1316, V583 Δef1599, and V583 ΔatlA. Relative fluorescence units, which are directly proportional to eDNA released, are plotted. Values shown are the averages from three biological replicates, and error bars represent standard errors of the mean. Statistical analysis was performed using Student’s two-tailed unpaired t test with confidence intervals of 95%. (B) Autolysis assay showing spontaneous lysis of strains V583, V583 Δef0073, V583 Δef1316, V583 Δef1599, and V583 ΔatlA, with the OD₆₀₀ plotted over time. Data depict averages from three biological replicates, and error bars shown represent the standard errors of the mean.

FIG 9.

Cell wall polysaccharides of RRNPP mutant strains. Cell wall preparations from V583 and isogenic RRNPP mutants were analyzed by PAGE and visualized by staining with 0.005% Stains-All dye. Labels depict capsule, Epa polysaccharide, and predicted teichoic acid segments of the cell wall.

DISCUSSION

Members of the RRNPP family of proteins are broadly present in different bacterial species within Firmicutes and regulate a diverse set of functions, including sporulation, virulence, conjugative transfer of plasmids, and biofilm formation. Over the last decade, various studies have been reported on the importance of RRNPP homologs in bacilli and streptococcal species (26, 50–53) and recently in Clostridioides difficile (54). In enterococci, PrgX was the first to be discovered due to its linkage to conjugation control for pheromone-responsive plasmids (12). More recently, Dumoulin et al. (28) characterized a role for ElrR in positively regulating ElrA, a leucine-rich surface protein involved in macrophage escape, and ElrA overexpression studies showed enhanced resistance (55), suggesting that modulating the activity of ElrA is the primary purpose for ElrR.

It is known that even though RRNPP proteins are found in a diverse set of bacteria in Firmicutes, some of the members are predominantly distributed in only certain bacterial orders (15). For example, NprR and Rap phosphatase have homologs in Bacillales, whereas Rgg and PrgX are present in Lactobacillales. Our results confirm these predictions, as each of the E. faecalis RRNPP homologs groups closely phylogenetically with Rgg and PrgX, and are consistent with a prior study (29), wherein only Rgg was used to identify RRNPP homologs in E. faecalis.

When the amino acid sequences were analyzed in silico, we found that the overall secondary structural arrangements were in agreement with the general trend of helical arrangement in the RRNPP family of proteins, using a secondary structure prediction tool called JPred 4.0 (56). As seen in Fig. S4 in the supplemental material, the N-terminal domains were predicted to have a Cro/C1 family of HTH DNA binding domain using a domain prediction and functional classification tool called InterPro (57). EF1316, EF1599, and EF1224 were also predicted to have TPRs according to TPRpred, a functional prediction software for proteins with TPR domains (58), but no TPRs were predicted to be present in EF0073 and ElrR. However, it should be noted that some of the structurally characterized RRNPP proteins, including PrgX and Rgg, do not possess a predicted TPR based on bioinformatic predictions but their solved C-terminal domain crystal structures formed an alpha-helical superstructure similar to ones formed by known TPR-containing proteins (20).

The enterococcal RRNPP homologs encoded on the chromosome appear to be highly conserved across the publicly available E. faecalis genomes and as such are likely to regulate a suite of core functions in all strains. In this study, we have successfully identified homologs of the RRNPP family (EF0073, EF1316, and EF1599) that contribute to the regulation of biofilm. It is noteworthy that of the 3 proteins, 2 (EF0073 and EF1599) repress aspects of biofilm formation, as deletion of those genes resulted in enhanced biofilm formation. Conversely, EF1316 is positively correlated with the ability to form biofilms, as deletion of ef1316 results in attenuated biofilm formation. These observations parallel what was observed by Chang et al. (26) in group A streptococci, wherein Rgg2 positively regulates biofilm and Rgg3 negatively regulates biofilm development, as well as lysozyme resistance (59). We also show that aspects of biofilm development controlled by these RRNPP family members likely happen via novel pathways, as known indicators of biofilm formation in enterococci were not significantly altered by the presence or absence of the RRNPP homologs. Bacterial attachment, eDNA release, and rates of autolysis were unaltered in the mutants. Moreover, examination of cell wall polysaccharides in the parental V583 strain and isogenic RRNPP mutants did not reveal gross changes in polysaccharide expression. Recent studies have shown an important role for the Epa polysaccharide is shaping the overall structure and strength of the biofilm in E. faecalis (48, 49, 60), but these pathways do not appear to be regulated by any of the RRNPP family members in E. faecalis V583. Time point experiments point toward a possibility of EF1316 being important for early-phase biofilm development, whereas EF0073 and EF1599 might play crucial roles in either growth of biofilm biomass or dispersal mechanisms of biofilm formation. Because of a gap in knowledge of biofilm determinants in E. faecalis except for adhesins, eDNA, and cell wall modifiers, elucidating the regulatory mechanisms of these RRNPP homologs will widen the knowledge base on biofilm formation in these bacteria.

It was interesting to identify putative peptide-coding sequences whose promoter regions induced LacZ expression in promoter fusion experiments. Even though these predicted small ORFs have a near-consensus ribosome binding site for translation, the presence of a peptide precursor or active peptide and their involvement in biofilms remain unexplored areas of investigation.

Recently, RopB (Rgg) from S. pyogenes was found to have an unusual cognate peptide (speB-inducing peptide [SIP]) that is only 8 amino acids in length, lacks a signal leader peptide, and does not require canonical oligopeptide permease pathways for peptide uptake (61). The fact that ef0072 and ef1600 encode putative peptides of 15 and 20 amino acids, respectively (Fig. 2A), suggests that they may follow the group A Streptococcus SIP signaling pathway. Consistent with its phylogenetic relationship to PrgX and Rgg, the genetic loci proximal to the enterococcal RRNPP family members from this study appear to express the short open reading frames that are divergently transcribed from their respective RRNPP genes.

The functional significance of the growth phase during which peptide production was stimulated has been highlighted in previous reports. Production of SIP, the peptide that is a cofactor for RopB-mediated regulation by group A streptococci, is stimulated only at the late exponential phase, owing to the optimal acidic conditions produced by an increase in the number of cells (61, 62). Expression of PhrR, the peptide precursor that regulates Rap phosphatase activity in bacilli, is activated at the early stationary phase, when cells initiate sporulation (63). Shp2 and Shp3, signaling peptides in group A streptococci that coordinate the function of Rgg2 and Rgg3, respectively, are expressed abundantly during log phase and are implicated in biofilm production (26). cCF10, the conjugation-stimulating pheromone produced by E. faecalis, is constitutively produced at a low level, leading to accumulation at higher cell densities in order to activate PrgX-dependent transcription and counter the balance of the inhibitory peptide, iCF10 (34, 64). This is also the basis for how E. faecalis prevents autostimulation of its cognate conjugation machinery in the absence of a neighboring recipient cell, as cCF10 produced from a donor cell is insufficient to overcome the inhibitory activity of iCF10. NprX, the stimulatory peptide regulating the activity of NprR, is produced abundantly at a late stationary phase and is regulated via multiple promoters to ensure the active regulation of NprR-regulated genes at later phases of growth (65).

The observation that ef1315 promoter fusions increased LacZ expression during exponential phase is consistent with the observed increase in expression of ef1316 relative to that in stationary phase. Furthermore, the early-phase biofilm defects observed in the ef1316 mutant indicate that this system regulates early biofilm processes. The ef1600-LacZ promoter fusions produced similar levels of expression during exponential- and stationary-phase growth, likely resulting in a steady state of peptide accumulation. With ef1599 expression also varying little over growth phase, a prediction can be made that the expression of a peptide from ef1600 serves as a potential ligand for EF1599. The observed alterations in biofilm development for the ef1599 mutant suggest that these divergently transcribed gene products act at later stages of biofilm development, possibly through unknown dispersal mechanisms, but this remains to be determined experimentally. The ef0072-LacZ promoter fusion was the most abundantly expressed of the 3 short open reading frames at all stages of growth. Its potential binding partner, EF0073, was observed at a low steady-state level, and it is unclear if these factors combine to regulate later stages of biofilm development, but the ef0073 mutant diverges from the parental biofilm after 24 h, which also suggests an involvement in later stages, analogous to what was observed for EF1599.

Our in silico analysis of the processing and secretion predictions for the divergently transcribed short open reading frames (EF0072, EF1315, and EF1600) that we predict serve as cognate peptides for the respective RRNPP family members was inconclusive, suggesting that these peptides are not processed and secreted though canonical pathways. Further investigations are required to know whether these peptides are processed for secretion in a manner similar to that for other known RRNPP quorum peptides requiring Eep processing and active transport via PptAB (66–68) or follow a noncanonical pathway like the group A streptococcal SIP (61).

During the course of this work, we were able to develop tools to perform a directed competitive assay that not only is sensitive for detection but can also readily distinguish teal from mCherry fluorescent protein expression due to the absence of spectral overlap. The use of an ectopic integration system allows stable expression of fluorescent proteins in the absence of antibiotic selection. The addition of unlabeled cells also allows for combinations of up to 3 distinct strains (teal, mCherry, and unlabeled) in competition studies and does not rely on differential antibiotic resistance for selection.

A recent report called into question the usage of polystyrene plate-based static assays in enterococcal biofilm formation (69). In this study, we were not only able to successfully optimize and use a drip-flow biofilm reactor for studying biofilms but were also able to confirm a biofilm phenotype for V583 Δef1316 and V583 Δef1599 that was not observable in static plate-based assays, and the significance of these findings showed a strong correlation with the in vivo results observed using the murine catheter-associated UTI model.

It was also interesting to observe the development of biofilm using the drip-flow biofilm reactor over a time course, as it allowed us to differentiate when a particular mutant diverged from the parental strain, and this information will aid in future studies to determine the genes regulated by these RRNPP transcription factors. Most of the initially adherent cells are washed away upon initiation of flow conditions, leading to a decrease of 3 to 4 logs down to 10² to 10³ CFU during the first 4 h (see Fig. S5 in the supplemental material); this observation would suggest that only a subset of cells in the original inoculum survive the presence of flow conditions, which is in stark contrast with a static system and will be relevant to guide future studies to investigate biofilm determinants in E. faecalis. Biofilm assays using the drip-flow system also allowed us to inspect the biofilm using confocal laser scanning microscopy (CLSM), directly quantify biofilms using the net wet mass, and obtain a quantifiable CFU readout that is reproducible. Thus, the drip-flow biofilm reactor (DFBR) can be used as an effective replacement of the conventional plate-based biofilm assay and may reveal previously unidentified biofilm candidates not observed under static conditions.

Since ElrR was shown to positively regulate ElrA to enhance virulence (28, 55), it was surprising that in our in vivo model we did not observe a role for ElrR. This could be due to differences in the infection model (peritonitis versus UTI) or strain differences, as previous studies were conducted in plasmid-free OG1RF and we used the multidrug-resistant strain V583. We also observed a more dramatic difference in colonization of bladder and catheter tissues by the respective RRNPP mutants relative to the parental strain compared to the in vitro grown biofilms. This may indicate that additional host cues contribute to the regulation governed by RRNPP family members in vivo. The basis for this difference will require additional study. A study by Hoover et al. (52) showed that an RRNPP family member in Streptococcus pneumoniae controls the expression of a lantibiotic operon but also requires the host sugar galactose as an additional cue for maximal induction, suggesting that host cues may provide an additional layer of regulation.

Finally, we establish that 3 out of the 5 RRNPP homologs regulated biofilm formation in vivo and in vitro. We believe that each of the homologs follows a different regulatory pathway to affect biofilm, as there is an opposing biofilm pattern observed for these mutants. As the previously described enterococcal biofilm characteristics appear to be unaffected in V583 Δef0073, V583 Δef1316, and V583 Δef1599, it will be important to identify the regulons of these transcription factors to identify the pathways each system regulates to contribute to overall biofilm development and pathogenesis.

MATERIALS AND METHODS

Bioinformatic analysis.

DELTA-BLAST was used to find the possible homologs of the RRNPP protein family in the E. faecalis V583 genome. We used protein sequences from Streptococcus pyogenes MGAS8232 Rgg (accession no. AAL98562.1)¸ Bacillus cereus ATCC 14579 PlcR (CAB69810.1), and E. faecalis plasmid-encoded pCF10 PrgX (AAA65845.1) and ElrR (AAO82391.1) as query sequences. These results were refined by setting the query coverage at 80% and performing three iterative searches with an E-value cutoff of 0.001. We used the results from the combined queries to generate a list of possible homologs that were subsequently examined for phylogenetic analysis.

Phylogenetic analysis.

Phylogenetic analysis was performed using the amino acid sequences of all 7 identified E. faecalis RRNPP homologs in the V583 genome, including the previously characterized ElrR (28) along with Rgg from S. pyogenes¸ PlcR from B. cereus, PrgX from E. faecalis pCF10, RapA phosphatase from Bacillus subtilis 168 (CAB13100.1), and NprR from Bacillus thuringiensis (AEQ53965.1).

ClustalW (70) using default parameters was used to align the 12 amino acid sequences. A maximum-likelihood phylogenetic tree was constructed using the GAMMA BLOSUM62 substitution model implemented in RAxML (version 8.2.11) (71) with 100 bootstrap replicates and no defined outgroup. Rooting of outgroups (NprR and RapA phosphatase) was done post hoc. Both alignment and tree building were performed in Geneious (72) using the appropriate plugins.

Bacterial growth.

E. faecalis strains were routinely cultured in Todd-Hewitt broth (THB) and grown at 37°C overnight with appropriate antibiotics as specified (250 μg/ml gentamicin; 15 μg/ml chloramphenicol). Planktonic growth curve studies were done in MM9YEGC medium (73) by diluting 2 μl of overnight culture in 198 μl of freshly prepared medium and seeding into 96-well tissue culture plates, with growth monitored by optical density at 600 nm (OD₆₀₀) every 30 min for 12 h without shaking at 37°C using an Infinite M200 Pro plate reader (Tecan Instruments).

Escherichia coli cultures were grown in Luria-Bertani broth at 30°C with shaking at 225 rpm with antibiotics where specified (10 μg/ml chloramphenicol). Strains used in this study are listed in Table S2 in the supplemental material, and the plasmids used during the study are listed in Table S3 in the supplemental material.

Construction of isogenic deletion mutants and in-frame single-copy genetic complements.

All isogenic deletion mutants were created using vector pLT06 (74). Briefly, primers (see Table S4 in the supplemental material) with the incorporated restrictions sites were designed to amplify ∼1-kb regions flanking the gene of interest. The flanking amplicons were digested with corresponding restriction enzymes, ligated together, and reamplified by PCR using the designated P1 and P4 primers for each targeted gene. This resulting amplicon was digested with the appropriate restriction enzymes and ligated into similarly digested pLT06 vector using T4 DNA ligase (NEB). Ligated products were electroporated into either E. coli EC1000 (75) or E. coli ElectroTen-Blue electrocompetent cells (Agilent) and propagated for plasmid replication at 30°C with shaking at 225 rpm. Plasmids were purified using QIAprep spin miniprep columns (Qiagen), and purified plasmids were confirmed by restriction digest mapping and DNA sequencing. Confirmed plasmids were electroporated into E. faecalis V583 strains as previously described (76). Integration and excision of pLT06 derivatives in the E. faecalis genome to create markerless deletions were performed as previously described (74). Genetic complementation was done using the same markerless system, but the integration plasmid pLT06 contained the region that encompassed the deleted gene reintroduced at its native locus as previously described (66). Briefly, for each gene complement, corresponding P1 and P4 primers were used to amplify the flanking regions as well as the gene of interest and then subsequently cloned into pLT06 and reintroduced by electroporation in the corresponding mutant background to restore the deleted gene following integration and excision of the plasmid.

Construction of fluorescent reporter strains.

The ectopic integration region ef2238-39 (ef_rs10710-10715) in E. faecalis strain V583 was chosen based on the study by Debroy et al. (77). In the prior study, gene locus ef2238-39 was identified in OG1RF as an ideal site for ectopic integration because it possesses two convergent reading frames. Due to a number of single-nucleotide polymorphisms (SNPs) between the genomes of V583 and OG1RF at this site, we sought to recreate a vector that contains the V583 sequence to improve integration efficiency in V583. Primer pair EF2238F and EF2238R along with EF2239F and EF2239R were used to amplify the V583 genomic template. The EF2238R primer was designed to incorporate a single SmaI site in the intergenic sequence between the two genes. These two amplicons were then used as templates with phosphorylated outer primers EF2238F and EF2239R to generate a second PCR product that consisted of the amplicons fused together with a unique SmaI site between them. The resulting amplicon with phosphorylated ends was ligated into pLT06 (74) that had been digested with SmaI and SphI and treated with T4 DNA polymerase to create blunt ends. The resulting vector was named pKS500, and the orientation of the cloned insert was confirmed by PCR using primers OriF and EF2239R. Plasmid pKS500 was then digested with SmaI and ligated to the multiple-cloning site (MCS) from vector pAT28 (78) amplified using phosphorylated M13F and M13R primers. The resulting plasmid was designated pKS510. The orientation of the cloned MCS was confirmed by PCR with primer pair SeqR and M13R. To enhance the cloning functionality of the vector, pKS510 was subjected to site-directed mutagenesis to eliminate an EcoRI site at the 3′ end of the ef2238 gene. We performed site-directed mutagenesis using the QuikChange II XL site-directed mutagenesis kit and primer pair pKS510A (sense) and pKS510A (antisense) according to manufacturer’s recommendations. The resulting clone was designated pKU510, and the absence of the EcoRI site was confirmed by DNA sequencing. To generate plasmids that would allow stable integration of fluorescent reporter genes, encoding mCherry and teal fluorescent protein, into the EF2238-39 ectopic site in the V583 genome, we amplified the streptococcal codon-optimized mCherry and mTFP genes from pVMCherry and pVMTeal (79), respectively, using primer pair FP5′ and FP3′. The resulting amplicons were digested with EcoRI and SphI and ligated into pKU510 also digested with EcoRI and SphI, and the resulting clones were designated pKU511 (mCherry) and pKU512 (mTFP).

Plasmids pKU511 and pKU512 were introduced into the respective E. faecalis strains by electroporation. After integration and excision of the plasmids according to the method described by Thurlow et al. (74), we isolated stable E. faecalis fluorescent strains containing either mCherry or mTFP. Proper insertion of the fluorescent markers at the ef2238-39 ectopic locus was confirmed by PCR using primer pair EF2238Up and EF2239Down.

DFBR.

For the drip-flow biofilm reactor (DFBR), 100 μl of bacterial overnight culture grown in 3 ml modified M9 medium supplemented with 0.3% yeast extract, 1% Casamino Acids, 20 mM glucose, 1 mM MgSO₄, and 0.1 mM CaCl₂ (MM9YEGC) (73) was seeded 1:100 into 10 ml of sterile MM9YEGC. Seeded MM9YEG medium for biofilm growth was loaded onto channels in the biofilm chambers with microscope glass slides, and the DFBR was run as described previously (80). Briefly, growth chambers of the DFBR with the seeded cultures were incubated at 37ºC for 8 h for initial adherence, and then 0.1× MM9YEGC medium was fed into the growth chamber of the DFBR by inlet valves and tubing at 125 μl/min, the growth chamber was tilted 10º, and biofilm was developed for 72 h. For time point experiments, bacteria were grown for 4, 8, 12, 24, 36, 48, 60, and 72 h under flow conditions. Biofilm enumeration was done by aseptically removing the glass slides, scraping the biofilm into 5 ml of 1× phosphate-buffered saline (PBS) (1 ml of 1× PBS for the 4-, 8-, and 12-h time points), homogenizing biofilm using tissue tearor for 30 s, serially diluting, and plating on THB plates with appropriate antibiotics. Competitive indices for mixed infections were calculated according to the formula (CFU_mutant/CFU_{wild type})output/(CFU_mutant/CFU_{wild type})input. To determine cells labeled with teal fluorescent protein, colonies on agar plates were imaged with fluorescence excitation at 475 nm and an emission spectrum at 537 nm. For mCherry-labeled cells, fluorescence excitation was set to 632 nm with an emission spectrum at 710 nm using a FluorChem imager (Protein Simple, San Jose, CA). Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%. For the time point assay, statistical significance was analyzed using 2-way ANOVA, with results considered to be statistically significant (for the Δef1316 mutant after 8 h and the Δef0073 and Δef1599 mutants after 24 h) at a P value of <0.0001.

Confocal analysis of biofilm.

Confocal laser scanning microscopy (CLSM) was used to analyze 1-day-old biofilms to discern the biofilm phenotype as previously shown by Varahan et al. (66). Briefly, biofilms with teal fluorescence-labeled strains were cultivated using the DFBR for 1 day on microscopic glass slides. The center of the slide was then covered with coverslips and sealed using clear nail polish. CLSM analysis was done using an Olympus FV1000, capturing images in the ECFP channel, with each Z-slice having a depth of 0.7 μm. The COMSTAT plugin of the ImageJ software was used to quantify the maximum thickness over the area of the biofilm section analyzed. Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%.

Plate-based static biofilm assay.

The plate-based biofilm assay was performed as described previously (81). Briefly, E. faecalis cells were grown in 200 μl MM9YEGC medium in round-bottom 96-well microtiter plates for 24 h. After growth, the medium was discarded, and the cells were gently washed three times with PBS (pH 7.4). Adherent biofilm was fixed at 50°C for 2 h. Each well was then stained with 200 μl 1% crystal violet for 20 min, followed by a gentle tap water rinse to remove unbound dye. Bound dye was then solubilized in 200 μl ethanol-acetone (80:2.0 vol/vol), transferred to a flat-bottom microtiter plate, and read at an optical density of 550 nm (Tecan plate reader). Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%.

Net wet biofilm biomass quantification.

Net wet biofilm biomass was quantified by directly measuring the weight of biofilm scraped into a microcentrifuge tube. The average wet biomass was calculated by subtracting the mass of the microcentrifuge tube from the mass of tube with biofilm. Masses from 9 different biofilm runs were used to calculate the net wet biomass. Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%.

Autolysis assay.

The autolysis assay was performed as described previously (82). Briefly, cultures were grown in THB overnight at 37°C, diluted 1:100 in SM17 medium with 3% glycine, and grown at 37°C overnight. These samples were pelleted, washed three times with ice-cold sterile water, and resuspended in 10 mM sodium phosphate buffer (pH 6.8) to an initial OD₆₀₀ of 1. Two hundred microliters of washed cells was seeded into flat-bottom 96-well microtiter plates, and the OD₆₀₀ was recorded every 20 min for 24 h using a Tecan plate reader.

eDNA release assay.

Genomic DNA released from a cell was measured as previously described (45). Briefly, strains from 3-ml overnight cultures were spun down, and the supernatant was filter purified using a 0.2-μm syringe filter. Two hundred microliters of this taken with 1 μM SYTOX green dye (final concentration), and fluorescence (excitation, 485 nm; emission, 535 nm) was read using a Tecan plate reader. Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%.

Cell wall carbohydrate detection.

Extraction of cell wall polysaccharides from V583 and isogenic RRNPP mutants, along with analysis by PAGE and staining with Stains-All, was performed as previously described (83).

Biofilm attachment assay.

Biofilm adherence was tested using the drip-flow biofilm assay setup. Overnight cultures of fluorescently tagged V583 and mutant strains were mixed in equal concentrations, diluted 1:100 in MM9YEGC medium, and injected into the drip-flow biofilm chamber with a glass slide. The setup was allowed to incubate horizontally at 37°C for 8 h. After incubation, the glass slides were scraped, serially diluted, plated on THB agar containing appropriate antibiotics, and incubated overnight at 37°C. Input mixed-culture CFU was enumerated without the adherence step. CFU for strains that underwent the adherence step was calculated and competitive indices calculated as described for the biofilm growth and enumeration assay. Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%.

Murine model for CAUTI.

Catheter-associated urinary tract infection (CAUTI) was performed using six 7-week-old female wild-type C57BL/6 mice. Mice were anesthetized by isoflurane inhalation, and a 5- to 6-mm platinum cured silicone implant tubing (Renasil Sil025; Braintree Inc.) was transurethrally placed in the urinary bladders of the mice as previously described (84). Postimplantation, the mice were injected with 50-μl inocula of either sterile PBS or bacterial suspension (∼2 × 10⁷ CFU) by transurethral catheterization. The mice were monitored for 48 h after implantation and infection and were provided mouse chow and water ad libitum. Mice were euthanized by cervical dislocation after inhalation of isoflurane. To determine the degree of infection, the implanted silicone catheter along with kidneys and bladder was harvested aseptically, and the bacterial burden was determined using serial dilutions and plating on Todd-Hewitt agar.

For competition studies, overnight cultures of fluorescently labeled bacteria were mixed ∼1:1 and enumerated by serial dilution and agar plating to determine input numbers by fluorescent imaging as described above. Bacterial enumeration from infected animals was similarly done to determine output ratios in order to calculate the competitive index. Statistical significance was evaluated according to the Kruskal-Wallis test using Graph Pad Prism version 5.0. Animal experiments conducted during the course of this study were performed in accordance with accepted veterinary standards and with approval from the Institutional Animal Care and Use Committee (IACUC) at the University of Kansas (IACUC approval number 219-02).

RNA isolation and qPCR.

Cells were grown to stationary, mid-log, or biofilm phase and spun down at 10,000 × g for 10 min (biofilms were homogenized using a tissue tearor before centrifugation). This material was then resuspended in 1 ml TRIzol and 1 ml 0.1-mm zirconia beads. Bead beating for done for 1 min thrice, and the mixture was spun down at 10,000 × g for 10 min to remove beads and cell degradation products. The supernatant was then resuspended in equal volume of 100% ethanol until the solution turned slightly turbid. This solution was subjected to RNA isolation using the Direct-zol RNA Miniprep Plus kit from Zymo Research.

The resulting RNA (5 μg) was DNase treated using the Ambion Turbo DNase kit to remove extracellular DNA that might have been due to genomic DNA contamination from RNA isolation (this was done twice for biofilms, as DNA is a major component of enterococcal biofilms).

One microgram of DNase-treated RNA was taken and reverse transcribed using the Superscript III reverse transcriptase (RT) kit from Life Technologies. One microliter of cDNA was used in a qPCR setup using the PowerUp SYBR green master mix from Thermo Fisher Scientific and the respective gene-specific qPCR primers. The Quant Studio 3 real-time PCR system from Thermo Fisher Scientific was used to perform the qPCR, and relative quantification of expression values was done by the ΔC_T and ΔΔC_T methods. Statistical significance was tested using one-way and two-way ANOVA estimation.

Modified Miller assay (β-galactosidase assay).

V583 parental strains containing appropriate promoter fusions (or pLacZ empty vector) were grown to stationary and mid-log (OD of ∼0.5 to 0.6) phases in 5 ml THB supplemented with 500 μg/ml spectinomycin. These cultures were centrifuged at 4,000 rpm for 10 min and resuspended in an equal volume of Z buffer (60 mM disodium phosphate, 40 mM monosodium phosphate, 10 mM KCl, and 1 mM anhydrous MgSO₄ supplemented with 50 mM β-mercaptoethanol). This was resuspended in 1.5 ml Z buffer, and the cell slurry was bead beaten with 500 μl of 0.1-mm zirconia beads for 1 min. This suspension was centrifuged at 13,000 rpm for 30 s to remove beads, and the aqueous layer was centrifuged at 13,000 rpm for 10 min to remove cell debris. One milliliter of this supernatant was taken, and 250 μl of this was used with 50 μl of 4 mg/ml o-nitrophenyl-β-d-galactopyranoside (ONPG) substrate. The assay was developed for 30 min in a flat-bottom clear 96-well plate and was read at 405 nm using an Infinite M200 Pro plate reader (Tecan). These values were normalized to the total protein concentration quantified by the Coomassie Bradford Plus assay (15 μl of clarified lysate with 300 μl of Coomassie blue) using bovine serum albumin (BSA) protein standards. Miller units were calculated as the ratio between OD₄₀₅ and total protein (milligrams) per assay. Statistical significance was tested using Student’s two-tailed unpaired t test with confidence intervals of 95%.