Exploiting biofilm phenotypes for functional characterization of hypothetical genes in Enterococcus faecalis

Abstract

Enterococcus faecalis is a commensal organism as well as an important nosocomial pathogen, and its infections are typically linked to biofilm formation. Nearly 25% of the E. faecalis OG1RF genome encodes hypothetical genes or genes of unknown function. Elucidating their function and how these gene products influence biofilm formation is critical for understanding E. faecalis biology. To identify uncharacterized early biofilm determinants, we performed a genetic screen using an arrayed transposon (Tn) library containing ~2000 mutants in hypothetical genes/intergenic regions and identified eight uncharacterized predicted protein-coding genes required for biofilm formation. We demonstrate that OG1RF_10435 encodes a phosphatase that modulates global protein expression and arginine catabolism and propose renaming this gene bph (biofilm phosphatase). We present a workflow for combining phenotype-driven experimental and computational evaluation of hypothetical gene products in E. faecalis, which can be used to study hypothetical genes required for biofilm formation and other phenotypes of diverse bacteria.

Keywords: Biofilms; Microbial genetics; Pathogens.

Fig. 1

Workflow for characterizing hypothetical gene products important for biofilm formation in E. faecalis. a An arrayed library of E. faecalis OG1RF transposon mutants was constructed using strains with insertions in poorly characterized/hypothetical genes and intergenic regions and was used in a genetic screen to identify biofilm determinants. The protein sequences of low-biofilm gene products were examined to predict structure and function. This data was used to identify top candidate genes for in vitro validation of function. b From this workflow, we identified eight genes involved in early biofilm formation, three of which are important for in vitro surface attachment. Further experiments identified OG1RF_10435 (bph) as a phosphatase

Fig. 2

Mutations in E. faecalis hypothetical genes lead to decreased biofilm formation. a Biofilm production of Tn mutants chosen from arrayed Tn library screen was measured at 6 and 24 h. Values are normalized to biofilm produced by OG1RF (black horizontal line). Statistical significance at each time point was calculated by one-way ANOVA (****q < 0.0001, ***q < 0.001, **q < 0.01, *q < 0.05 with multiple comparisons by two-stage step-up method of Benjamini, Krieger, and Yekutieli). b Gene neighborhood of bph (gray arrow). Gene numbers indicate “old locus tag” identifiers from the annotated OG1RF genome (NC_017316). Tn insertions are designated by triangles and are numbered based on position in gene relative to the 5’ end (1 = 56.3%, 2 = 91.9%). c Biofilm production and complementation in bph mutants. Strains carried either an empty vector plasmid (pVector) or complementation plasmid pBph and were cultured in microtiter dishes for 6 h. Values were normalized to parental OG1RF. Statistical significance was calculated by one-way ANOVA as in panel a. d Fluorescence microscopy of biofilms grown on Aclar substrates for 6 h and stained with Hoechst 33342 (scale bar = 20 μm). Images were obtained with a ×20 objective (×200 total magnification) and are representative of three independent experiments. e Cell envelope integrity of Δbph. Strains carried a constitutive lacZ expression vector (pCJK205) and were cultured in the presence of chlorophenyl red-β-d-galactopyranoside (CPRG). The ratio of CPRG activity (OD₅₇₀) to cell density (OD₆₃₀) was calculated at each time point. Significance was calculated by two-way ANOVA (*q < 0.05 with multiple comparisons as in panel a. In panels a, c, and e, each data point represents a biological replicate (n = 3). Error bars represent standard error of the mean

Fig. 3

bph is required for surface attachment and pili expression. a Attachment of Tn mutants to polystyrene microtiter plates was measured after 1 h. Attachment was calculated as the ratio of attached biomass (OD₄₅₀) to cell density (OD₆₀₀) relative to OG1RF. Significance was determined by one-way ANOVA (*q < 0.05 with multiple comparisons by two-stage step-up method of Benjamini, Krieger, and Yekutieli). The inset table shows attachment values (mean of three biological replicates) for the bph Tn mutants and clean deletion. b Attachment to polystyrene microtiter plates was measured as in panel a. Strains carried either an empty vector or the complement plasmid pBph. ΔahrC (light gray) and Δbph (light blue) strains had reduced attachment (significance determined as in panel a). Attachment was increased by a plasmid expressing bph (dark blue). c Strains from panel b were cultured with porcine heart valves, and tissue-associated colony forming units were quantified. ΔahrC and Δbph strains had reduced average attachment relative to OG1RF, and expressing bph from a plasmid resulted in a significant increase in attachment relative to the empty vector strain (significance determined as in panel a). d Pili protein in whole-cell lysates isolated after 2 or 4 h of growth was detected by Western blotting with an anti-EbpC antibody. e Activities of the ebpR and ebpA promoters were measured using cells carrying a vector control (pTCV-LacSpec) or promoter fusions for the indicated genes. Values represent Miller units calculated from β-galactosidase assays over three biological replicates. f The strains used in panel d were spotted on plates containing X-gal and imaged. In panels a, b, c, and e, each data point represents a biological replicate (n = 3). Error bars represent standard error of the mean

Fig. 4

Disrupting bph results in temporal changes in global protein expression. a Growth curves of the indicated strains in TSB-D growth medium. Data points represent the mean of three biological replicates. Error bars represent standard error of the mean. Aliquots were removed for protein analysis at time points indicated by arrows. Whole-cell lysates were separated into b protoplast and c cell wall fractions using a lysozyme treatment. d Secreted proteins were precipitated from the supernatant with trichloroacetic acid. All protein samples were subjected to SDS–PAGE and stained with Coomassie. Blue arrows to the right of gel images denote protein bands that are differentially expressed in Δbph

Fig. 5

Effect of deleting bph on arginine catabolism and response to antibiotics. a SDS–PAGE image showing an abundant protein band (boxed region) present in lysate from OG1RF but not Δbph. Gel slices corresponding to the boxed areas were excised and analyzed via LC–MS/MS. Ratios indicate values in total normalized spectral counts. b Growth of parental OG1RF (black line), the arcA null strain arcA-Tn (gray line), and Δbph (blue line) in semi-defined medium supplemented with 1% arginine. Growth was measured by absorbance at 600 nm (OD₆₀₀). arcA-Tn has reduced growth relative to OG1RF, and ΔbphA has a diauxic growth pattern. c The strains from panel b were grown in semi-defined medium supplemented with 1% glucose. All strains have similar growth patterns. d SDS–PAGE image showing an abundant protein band (boxed region) present in ΔbphA but not OG1RF. Gel slices were analyzed as in panel a. OG1RF and Δbph were grown in varying concentrations of e cefoxitin and f oxacillin. The final OD₆₀₀ value at each antibiotic concentration was divided by the OD₆₀₀ of an untreated culture. For panels b, c, e, and f, each data point represents the mean of three biological replicates. Error bars represent standard error of the mean.

Fig. 6

Bph has predicted structural homology to phosphatases and has phosphatase activity in vitro. a Predicted structural homologs identified by Phyre2. The structure of SC4828 (cyan, PDB 3EXM) was used to model Bph (gray). The overlay was generated from crystal structures of FomD (yellow, PDB 5ZDN) and SC4828 and the predicted structure of Bph. Substrates are shown as light gray sticks (CDP for FomD, GPCP for SC4828), and coordinated divalent cations are shown as dark gray spheres (Mg²⁺ for FomD, Ca²⁺ for SC4828). b Alignment (Clustal2 Omega) of active site residues in SC4828, SA1684, and Bph. Asterisks and dots (bottom) indicated amino acid relatedness. Filled squares (top) signify ligand-binding residues and open squares (top) signify residues that coordinate Ca²⁺ in SC4828. Numbers (right) indicate amino acid positions. The Tn insertions in Bph occur at Y101 (Tn1) and outside of the portion of the alignment shown (Tn2). c Purified wild-type Bph-H6 or d point mutant variants were mixed with the indicated substrates. A malachite green-based assay was used to quantify phosphatase activity. Activity on AMP (marked with a star) was below the detectable limit, so A₆₄₀ values are shown (inset, panel c). e Biofilm production at 6 h was determined for Δbph cells carrying bph complementation plasmids with the inactive point mutants from panel d. Significance was determined by one-way ANOVA (****q < 0.0001 with multiple comparisons by two-stage step-up method of Benjamini, Krieger, and Yekutieli). In panels c–e, each data point represents a biological replicate (n = 3). Error bars represent standard error of the mean

Fig. 7

Analysis of discrete steps in biofilm formation as a pathway for characterization of hypothetical gene products. OG1RF (gray, left panel) expresses cell surface factors, such as Ebp pili (black lines) that allow it to attach to substrates and grow as a biofilm in a multi-step process. Biofilm matrix is shown as a yellow shape surrounding cells. Examination of discrete biofilm steps reveals that a strain lacking bph (blue, middle panel) has altered expression of Ebp pili and reduced surface attachment, but it can still produce matrix components. Other strains with Tn insertions in hypothetical genes have low-biofilm phenotypes (various colors, right panel). These uncharacterized gene products may be important for discrete biofilm development steps such as (1) surface attachment but not matrix development, (2) elaboration of the biofilm matrix and overall matrix architecture, or (3) controlling the rate of detachment and biofilm dispersal