The adjacent ATP-binding protein-encoding genes of the Enterococcus faecalis phosphate-specific transport (pst) locus have non-overlapping cellular functions

TL;DR

This study shows that two similar ATPase genes in Enterococcus faecalis have distinct roles in phosphate import, with one being essential and the other having a supporting role.

Contribution

The first study to demonstrate non-overlapping functions of tandem pstB paralogs in any microbe.

Findings

pstB2 is essential for phosphate import, while pstB1 has an accessory role.

Overexpression of pstB2 can compensate for the loss of pstB1 but not vice versa.

Deletion of either gene increases the expression of the other.

Abstract

The widely conserved pst-phoU operon encodes a low-velocity, high-affinity, ATP-dependent importer for inorganic phosphate (Pi). The pstB gene encodes the ATPase that powers the import of Pi into the cell. In some Firmicutes, including the gastrointestinal commensal and opportunistic pathogen Enterococcus faecalis, the pst-phoU locus contains adjacent pstB genes. In this work, we compared the functionality of E. faecalis pstB1 and pstB2. E. faecalis pstB1 and pstB2 share sequence similarities with verified PstB ATPases from Escherichia coli and Streptococcus pneumoniae and only share ~60% amino acid identity with each other. Deletion of pstB1 was associated with a growth defect in low Pi-containing chemically defined medium (CDM), reduced Pi uptake, and a moderate increase in alkaline phosphatase (AP) activity. Deletion of pstB2 fully inhibited growth in CDM regardless of inorganic phosphorus source but did not hinder growth in rich, undefined medium. The ΔpstB2 mutant also exhibited a significant increase in AP activity that was associated with extracellular Pi accumulation. Overexpression of pstB2 in the pstB1 mutant was sufficient to restore growth in low-Pi CDM, Pi uptake, and AP activity, but this was not recapitulated with overexpression of pstB1 in the ΔpstB2 mutant. Deletion of either pstB paralog increased expression of the tandem paralog, and overexpression of pstB2 in ΔpstB2 reduced pstB1 expression. These results suggest that the E. faecalis pstB2-encoded ATPase is required for Pi import, while the pstB1-encoded ATPase has an accessory role in Pi import that can be duplicated by the presence of excess PstB2.

Phosphate is critical for all microbial life. In many bacteria, inorganic phosphate (Pi) is imported by the high-affinity, low-velocity Pst-PhoU system. The pstB gene encodes the ATPase that powers Pi import. The pst-phoU operon in many Firmicutes, including the human commensal and opportunistic pathogen Enterococcus faecalis, contains adjacent pstB genes, pstB1 and pstB2. No studies on the relative biological contributions of tandem pstB paralogs in any microbe have been published. This genetic study indicates that E. faecalis pstB1 and pstB2 do not have equivalent functions. The pstB2 gene encodes an ATPase that is required for Pi import, while the ATPase encoded by pstB1 has an accessory role in Pi import that can be duplicated by the presence of excess PstB2.

Figures

The pst-phoU operon is well conserved throughout the bacterial kingdom. The genomic organization of pst-phoU loci from selected gram-negative and gram-positive bacterial species is shown. The genomes of Vibrio cholerae, Bacillus anthracis, Enterococcus faecalis, Streptococcus pneumoniae, and Clostridioides difficile contain pst genes in more than one location. The pst genes were identified in three to five strains of each species shown, including the strains listed above; in all strains of each species evaluated, the genomic organization of the pst genes was consistent with what is shown. Genome accession numbers and locus tags for each strain shown are listed in Table S1.

The E. faecalis pst-phoU locus is transcribed as an operon and encodes PstB1 and PstB2, which are likely ATPases. (A) Primers were designed to amplify the intergenic regions of the E. faecalis OG1RF pst-phoU locus (pstS2 to phoU), as well as the upstream and downstream flanking genes (OG1RF_11471 and liaX, respectively). Amplification of cDNA from the intergenic regions located between pstS2 and phoU was observed. There was no amplification of cDNA from the intergenic regions between OG1RF_11471 and pstS2 or phoU and liaX. The diagram above the gel image shows the organization of the pst-phoU region in E. faecalis OG1RF. The sizes of the intergenic regions and the approximate regions amplified are indicated. Gel image is representative of four biological replicates. (B) Alignment of amino acid sequences for PstB from E. coli BW25113 strain K-12 and Streptococcus pneumoniae strain R6, which are known ATPases, and PstB1 and PstB2 from E. faecalis strain OG1RF. Sequence motifs characteristic of ATPases are identified by colored boxes as follows: orange, Walker A motif; brown, Walker B motif; black, LSGGQ motif; and purple, H-loop. Individual amino acids are colored as follows: goldenrod, hydrophobic side chain; teal, positively charged side chain; gray, negatively charged side chain; and purple, polar uncharged side chain. The alignment was generated with Geneious Prime software (version 2023.0.4).

ΔpstB1 exhibits a growth defect in low-Pi CDM, and CDM does not support growth of ΔpstB2 at 37°C. Bacterial growth was monitored over time for strains grown in BHI (top row), CDM containing 0.2 mM Pi (low-Pi CDM; middle row), and CDM containing 80 mM Pi (high-Pi CDM; bottom row). Graphs in the first column show only strains that carry the empty vector (pPLK2). The second and third columns show strains that express pstB1 or pstB2, respectively, in trans from pPLK2. Each data point is the mean of three biological replicates. Error bars show the standard deviation. Statistical analyses for the middle and bottom rows are found in Tables S5 and S6, respectively.

Deletion of either pstB1 or pstB2 significantly increases expression of the other pstB gene, and overexpression of pstB2 in the ΔpstB2 strain is associated with decreased expression of pstB1. RNA isolated from bacterial strains incubated in 10% BHI for 6 hours was reverse transcribed and analyzed via qPCR for expression of (A) pstB1 or (B) pstB2 relative to the reference gene gyrB. The solid black line indicates a relative gene expression level of 1. The dashed black line indicates the lowest possible value for relative gene expression based on the limit of detection of the target genes (see Materials and Methods). Each data point represents an independent biological replicate. Horizontal bars show the mean; error bars indicate the standard deviation. One-way analysis of variance with Tukey’s correction: *, P < 0.05; *, P < 0.001; , P < 0.0001.

ΔpstB1 exhibits diminished Pi uptake compared to wild-type, while medium in ΔpstB2-containing samples rapidly accrued more Pi than was initially added. Cultures of each strain were grown in 25% BHI for (A) 4 hours or (B) overnight, then incubated in Pi-free medium for 2 hours. Following this Pi starvation step, all strains were mixed with K2HPO4 and incubated for 1, 5, 10, 30, or 60 minutes. Cells were removed by filtration, then exogenous Pi remaining in the medium was measured. Pi uptake was calculated as described in the Materials and Methods. Data and error bars at each time point show the mean and standard deviation, respectively, of three biological replicates. Figure S3D shows the same data with the strains separated by empty vector, pPLK-pstB1, and pPLK-pstB2.

AP activity and phoZ expression are highly increased in the ΔpstB2 strain compared to the wild-type strain. (A–C) Colorimetric-based phenotypes of AP activity on (A) 10%, (B) 25%, or (C) 100% BHI agar containing 100 µg/mL XP. A darker blue color indicates more AP activity. The image shown is representative of three biological replicates. (D) RT-qPCR measurement of relative expression of E. faecalis phoZ, which encodes AP, as compared to the reference gene gyrB. RNA was collected from strains incubated in 10% BHI for 6 hours. The solid black line indicates a relative gene expression level of 1. The dashed black line indicates the lowest possible value for relative gene expression based on the limit of detection of the target genes (see Materials and Methods). Each symbol represents an independent biological replicate. Horizontal bars show the mean; error bars indicate the standard deviation. (E) Quantitative measurement of AP activity in cell-associated fractions by cleavage of the colorimetric substrate pNPP. Each symbol represents an independent biological replicate. Horizontal lines show the mean; error bars indicate the standard deviation. For (D) and (E), one-way analysis of variance with Tukey’s correction: *, P < 0.05; , P < 0.0001.

Deletion of phoZ reduces the amount of Pi that accrues in the medium of ΔpstB2-containing samples. (A) Colorimetric-based phenotypes of AP expression on agar containing 100 µg/mL XP for ΔphoZ mutant strains. The image shown is representative of three biological replicates. (B) Overnight cultures of each strain were starved of phosphate for 2 hours, then were mixed with K2HPO4 and incubated for 5, 30, or 60 minutes. One milliliter was removed from each incubated time point and filter sterilized. Exogenous Pi remaining in the medium was measured, and Pi uptake was calculated as described in the Materials and Methods. Data and error bars at each time point show the mean and standard deviation, respectively, of three biological replicates.

Identification of phosphorus-containing compounds that support growth of the wild-type, ΔpstB1, or ΔpstB2 strains in CDM. Biolog PM4A plates containing 59 different phosphorus-containing molecules were inoculated with either wild-type, ΔpstB1, ΔpstB2, ΔphoZ, ΔpstB1ΔphoZ, or ΔpstB2ΔphoZ suspended in CDM with 0.0 mM Pi. Bacterial growth was measured by reading the OD600 nm of each well after incubation in a humidity chamber at 37°C for 16 hours. The 59 separate compounds are grouped as inorganic phosphorus molecules (black) and organic phosphorus molecules that are further subdivided into carbohydrates (blue), amino acids (red), nucleotides (green), and phosphonates and phosphites (purple). N = 3–4 biological replicates for OG1RF, ΔpstB2, and the three phoZ deletion strains, and 4–5 biological replicates for ΔpstB1. Outliers identified by the application of the Grubb’s test were excluded from the data set.

Full text

INTRODUCTION

Phosphorus is the 11th most abundant element present in the Earth’s crust and is necessary for the formation of macromolecules that are critical for all terrestrial life (1). Phosphorus is primarily found in its most oxidized form, PO_4_^3-^, which is commonly referred to as the orthophosphate or inorganic phosphate (Pi) anion (2). Unsurprisingly, organisms have evolved multiple mechanisms to take advantage of the abundance of Pi in order to meet cellular needs for phosphorus. One such mechanism that bacteria use to acquire Pi directly from their environment and transport it into the cytoplasm of the cell is the phosphate-specific transport (Pst) system. The Pst system, which is encoded by the pst-phoU operon, is a high-affinity, low-velocity Pi importer that is most active when external Pi levels fall below a certain threshold (3, 4). Pst signals through the PhoB (response regulator)-PhoR (histidine kinase) two-component system, and disruption of the pst system results in a variety of phenotypic changes across bacterial species. For example, Cronobacter sakazakii Δpst mutants exhibited decreased biofilm formation and increased adhesion when grown in low Pi (5). Deletion of the pst operon in avian pathogenic Escherichia coli resulted in reduced virulence in an infected chicken model (6). In uropathogenic E. coli, deletion of pst resulted in diminished colonization of the urinary tracts of mice and rendered mutants less invasive in human bladder epithelial cells, likely due to decreased expression of type 1 fimbriae (7).

The pst-phoU operon is well conserved throughout the bacterial kingdom, but the genes that comprise the operon vary among species (8). In its most basic form, the operon contains pstS, pstC, pstA, pstB, and phoU. pstS encodes a Pi-binding protein that delivers Pi to the importer, where it is actively transported across the membrane and into the cytoplasm (9, 10). pstC and pstA encode hydrophobic membrane-spanning components of the Pst system that heterodimerize, forming a channel for Pi to enter the cell (11–13). The gene product of pstB is identified as an ATP-binding cassette (ABC), which binds and hydrolyzes ATP in order to power the active transport of Pi across the membrane (12, 14). PhoU connects the PstSCAB Pi importer with the phosphate-responsive PhoB/PhoR two-component system and serves as a negative regulator of phosphate signaling, likely through its known interactions with PstB and PhoR (15–17). Interestingly, some species have partial or complete additional copies of the operon located elsewhere in their genomes (8). In addition, a number of gram-positive bacteria have pst-phoU operons with two adjacent pstB genes (Fig. 1) (8). The roles of these duplicate pstB genes, and whether or not they perform identical or dissimilar functions, are not understood (8).

The pst-phoU operon is well conserved throughout the bacterial kingdom. The genomic organization of pst-phoU loci from selected gram-negative and gram-positive bacterial species is shown. The genomes of Vibrio cholerae, Bacillus anthracis, Enterococcus faecalis, Streptococcus pneumoniae, and Clostridioides difficile contain pst genes in more than one location. The pst genes were identified in three to five strains of each species shown, including the strains listed above; in all strains of each species evaluated, the genomic organization of the pst genes was consistent with what is shown. Genome accession numbers and locus tags for each strain shown are listed in Table S1.

Enterococcus faecalis is a gram-positive gastrointestinal commensal and opportunistic pathogen of humans, composing roughly 0.1% of the human gut microbiota (18). Having likely emerged during the Paleozoic Era, during the period that some animal species first left the oceans for land, E. faecalis has evolved to exist and thrive in multiple environments (19). Enterococci are highly resistant to desiccation, UV radiation, detergents, disinfectants, bile salts, heat, and high salinity (19–22). Despite the importance of Pi in microbial metabolism and virulence, neither Pi uptake nor the pst-phoU locus has been characterized to date in E. faecalis. Furthermore, the E. faecalis pst-phoU locus has adjacent genes annotated as pstB1 and pstB2 (Fig. 1). In this work, we sought to compare and contrast the functionality of pstB1 and pstB2 to better understand the role of each gene in E. faecalis strain OG1RF with respect to growth, pst-phoU locus expression, and Pi uptake. Our results demonstrate that deleting either pstB gene from E. faecalis results in vastly different phenotypic outcomes, suggesting that these two genes have non-overlapping functions.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and reagents

E. faecalis strains used in this study are listed in Table S2. Plasmids are listed in Table S3. The E. faecalis strain used in this study, OG1RF, contains no mobile genetic elements, so discoveries made in this strain can likely be extrapolated to other E. faecalis strains (23). E. faecalis strains were cultured in brain heart infusion (BHI) broth (Becton-Dickinson and Co., Franklin Lakes, NJ), on BHI containing 1.5% agar, or in chemically defined medium (CDM) (24, 25). Ten percent and 25% BHI broths were prepared by 1:10 and 1:4 dilutions (vol/vol) of sterile BHI broth, respectively, in sterile Milli-Q-filtered water. Overnight cultures of strains were incubated at 25°C or 37°C under static conditions in ambient air for 16–20 hours, unless otherwise stated. E. coli strain DH5α, which was used to propagate plasmids, was grown in Luria Broth (LB), containing 10 g/L tryptone, 5 g/L yeast extract, and 10 g/L sodium chloride (ThermoFisher Scientific, Waltham, MA). LB or Super Optimal broth with Catabolite repression (SOC) was used as recovery medium for E. coli following chemical transformation.

Plasmid-containing strains were grown in broth culture or on agar plates containing antibiotics for selection at the following concentrations, unless otherwise specified: 20 µg/mL chloramphenicol, 50 µg/mL carbenicillin, or 100 µg/mL erythromycin. All antibiotics were purchased from Sigma-Aldrich (St. Louis, MO). Restriction enzymes and other enzymes used for recombinant DNA methods were purchased from New England Biolabs (Ipswich, MA). PfuUltra II Fusion DNA polymerase (Agilent, Santa Clara, CA) or high-fidelity Phusion Hot Start II DNA Polymerase (ThermoFisher Scientific) were used for all PCR amplifications performed for strain and plasmid construction. DL-2-amino-3-phosphonopropionic acid (APP), 2-aminoethylphosphonic acid (AEP), (±)-2-amino-4-phosphonobutyric acid (APB), methylphosphonic acid (MPP), and polyphosphate were purchased from Sigma-Aldrich.

Strain construction

Oligonucleotides are listed in Table S4. Genomic DNA was extracted from E. faecalis OG1RF with the DNeasy Blood and Tissue Kit (Qiagen, Inc., Germantown, MD), according to the manufacturer’s instructions. In-frame markerless deletion strains were constructed using previously described allelic exchange methods (26). The ΔpstB1 deletion construct used for allelic exchange was generated with overlap extension PCR by first amplifying two ~1 kb fragments from OG1RF genomic DNA with primer pairs EF1755(pstB1)-2stepF/EF1755(pstB1)-downR and EF1755(pstB1)-upF/EF1755(pstB1)-2stepR. The two products were annealed together, and second-step amplification was performed with primers EF1755(pstB1)-2stepF/EF1755(pstB1)-2stepR. The resulting ~2 kb product was gel purified, A-tailed with Taq DNA polymerase in ThermoPol buffer (New England Biolabs), and ligated into pGEM T-EZ (Promega, Madison, WI) to generate pGEM-pstB1. pGEM-pstB1 was amplified in E. coli DH5α, and the pstB1 deletion construct was verified with Sanger sequencing. The 2 kb deletion construct was released from the pGEM T-EZ backbone with EcoRI restriction digest, gel purified, and ligated with T4 DNA ligase into pCJK47 that had been digested with EcoRI and dephosphorylated with calf intestinal phosphatase (26). The ligated plasmid was propagated in E. coli EC1000 grown on BHI agar or in BHI broth with erythromycin.

The ΔpstB2 and ΔphoZ deletion constructs used for allelic exchange were generated with overlap extension PCR by first amplifying two ~1 kb fragments from OG1RF genomic DNA with primer pairs pstB2-upF_BamHI/pstB2-downR_SphI and pstB2-2stepF/pstB2-2stepR (for deletion of pstB2) and phoZ-upF_BamHI/phoZ-downR_SphI and phoZ-2stepF/phoZ-2stepR (for deletion of phoZ). The two products were then annealed together, and a second round of amplification was undertaken using primer pairs pstB2-upF_BamHI/pstB2-downR_SphI for the ΔpstB2 construct and phoZ-upF_BamHI/phoZ-downR_SphI for the ΔphoZ construct. These products were digested with restriction enzymes BamHI and SphI. The digested products were ligated with T4 DNA ligase into pCJK218 digested with the same enzymes and propagated in E. coli strain DH5α grown on chloramphenicol (27). The pstB2 and phoZ deletion constructs were verified with Sanger sequencing.

For complementation plasmid generation, pstB1, pstB2, and phoZ were amplified from OG1RF genomic DNA using primer pairs pstB1 Forward Cloning/pstB1 Reverse Cloning, pstB2 Forward Cloning/pstB2 Reverse Cloning, and phoZ Forward Cloning/phoZ Reverse Cloning, respectively. The pPLK2 plasmid, which contains the strong constitutive p23 promoter from Lactococcus lactis and confers chloramphenicol resistance, was used for complementation (28). pPLK2 and the pstB1 and pstB2 PCR products were digested with XbaI and HindIII-HF, gel purified, and ligated with T4 DNA ligase. The phoZ PCR product was A-tailed with Taq DNA polymerase in ThermoPol buffer and ligated into pGEM T-EZ, which was then digested with SacI, gel purified, and ligated with T4 DNA ligase into similarly digested and purified pPLK2. The resulting plasmids were propagated in E. coli DH5α grown in LB broth or on LB agar supplemented with chloramphenicol. Constructs were verified with Sanger sequencing.

E. faecalis electroporation

Cells were grown overnight in Todd-Hewitt broth (THB) (Fisher Scientific, Waltham, MA) at 37°C. The following day, cultures were diluted 1:10 or 1:20 in fresh THB and incubated at 37°C until an OD_600 nm_ of 0.5–1.0 was reached. Cultures were chilled on ice for 15–20 minutes, then pelleted at 3,452 × g at 4°C for 15–30 minutes in a Sorvall Legend RT centrifuge (ThermoFisher Scientific). Cells were resuspended in 500 µL lysozyme solution (10 mM Tris pH 8.0, 20% sucrose, 10 mM EDTA, 50 mM NaCl, and 25 µg lysozyme [Sigma-Aldrich]) and incubated at 37°C for 20 minutes. Cells were washed three times in 1 mL ice-cold electroporation buffer (0.5 M sucrose, 10% glycerol), then resuspended in the same buffer and stored at −80°C. Cells were electroporated with purified plasmid DNA using a Gene Pulser Xcell with PC module (Bio-Rad, Hercules, CA), set to 1.6 kV, 200 Ω, 25 µF. Following electroporation, cells were resuspended in 200 µL THB supplemented with 17.1% sucrose and incubated statically at 37°C for 2 hours. Following incubation, cells were spread on BHI agar with appropriate antibiotic selection. Agar plates with bacteria transformed with pCJK218 derivatives were incubated at 28°C–30°C; agar plates for all other plasmid transformations were incubated at 37°C.

Growth of E. faecalis cultures for RNA extraction

Overnight cultures of strains grown in BHI broth with antibiotic selection were diluted to an OD_600 nm_ of 0.01 in 10% BHI. Three milliliters of diluted cells were pipetted into each well of a six-well plate (Corning Inc., Corning, NY) and incubated statically at 37°C for 6 hours. Following incubation, supernatants from each well were removed and pooled together. An aliquot of the pooled supernatants was serially diluted and plated to enumerate colony forming units (CFU) per milliliter. The remaining pooled supernatants were centrifuged at 3,452 × g for 20–30 minutes in a Sorvall Legend RT centrifuge. Cell pellets were resuspended in 600 µL of 1× Tris-HCl, pH 8.0. Cells were treated with RNAprotect Bacteria Reagent (Qiagen) per the manufacturer’s instructions. Cell pellets were stored at −80°C until RNA extraction.

RNA extraction, DNase treatment, cDNA generation, and quantitative PCR

Cell pellets were resuspended in 200 µL of planktonic lysis solution (30 mg/mL lysozyme, 500 units/mL mutanolysin, 10 mM Tris-HCl pH 8.0, and 1 mM EDTA pH 8.0) and incubated at 37°C for 10 minutes. Following incubation, RNA was extracted with the RNeasy Mini Kit (Qiagen) per the manufacturer’s instructions; buffer RLT was prepared with β-mercaptoethanol. Purified RNA was DNase treated using a Turbo-DNA Free kit (ThermoFisher Scientific), and cDNA was generated with the SuperScript III First-Strand Synthesis System for RT-PCR (ThermoFisher Scientific). qPCR was performed using a BioRad CFX96 C1000 Touch Thermal Cycler with SsoAdvanced Universal SYBR Green Supermix (Bio-Rad). Each reaction was performed in technical triplicate, and threshold cycle (Ct) values were averaged for a single biological replicate. Relative gene expression with respect to the reference gene gyrB was calculated using the 2^ΔCt^ equation. A Ct value of 30 was set as the limit of detection for all target genes. The greatest Ct value obtained for gyrB, the reference gene, was 25. For analysis purposes, all target gene reactions with Ct values greater than 30 were set to 30.

Amplification of intergenic regions in the pst-phoU locus

PCR primers (Table S4) were designed to generate amplicons of ~100 bp–300 bp that span the intergenic regions between the genes predicted to comprise the E. faecalis OG1RF pst-phoU locus, as well as the upstream and downstream flanking genes. RNA was isolated from mid-log phase planktonic OG1RF cells grown in BHI. RNA was DNase treated and used for cDNA generation, as described above. The resulting cDNAs were used as templates in PCR reactions with Taq DNA Polymerase in ThermoPol buffer (New England Biolabs). Amplicons were separated on a 1.5% Tris-Acetate-EDTA (TAE) agarose gel stained with ethidium bromide and visualized with UV light.

Growth curves

Overnight cultures grown in BHI with appropriate antibiotics, as needed to select for plasmids, were washed in 1× Tris-HCl, pH 8, and resuspended in either CDM or BHI to an OD_600 nm_ of 0.1 or 0.01 for strains grown at 21°C–25°C or 37°C, respectively. Two hundred microliters of either samples or blanks, with appropriate selective antibiotics added, was pipetted into tissue culture-treated, flat-bottomed 96-well microtiter plates with six wells used per strain sample (Corning). Outer wells of the microtiter plates were filled with 200 µL/well of sterile water in order to maintain humidity through the course of incubation. Plates were placed in a BioTek Synergy HTX Multi-Mode Reader (Agilent) incubated at either 21°C–25°C (room temperature) or 37°C for 16 hours, with OD_600 nm_ measurements taken every 30 minutes immediately following a 5 second shake cycle. The technical replicates of the same strains were averaged together, from which the average of the blank control technical replicates was subtracted to generate a single biological replicate of the growth curve. Three biological replicates were performed on separate days.

Pi uptake assays

Pi uptake from supernatants was measured using a modification of a previously described method (29). Bacteria from colonies grown overnight at 37°C were incubated statically either for 4 hours or overnight in 25% BHI with appropriate selective antibiotics at 37°C. Cells were washed twice with Pi-free CDM that contained no glucose, then resuspended to an OD_600 nm_ of 0.5 in Pi-free CDM containing 0.01 g/mL glucose. Cells were incubated for 2 hours at 37°C. Following incubation, 750 µL of cells were mixed with 750 µL of 10.5 µM K_2_HPO_4_ and incubated for 1, 5, 10, 30, or 60 minutes at room temperature. One hundred microliters was removed from each sample/time point, serially diluted, and plated to enumerate CFU per milliliter. A positive control of 750 µL of Pi-free CDM with no bacteria mixed with 750 µL of 10.5 µM K_2_HPO_4_ (diluted in Pi-free sterile water) and a blank of 750 µL of Pi-free CDM with no bacteria mixed with 750 µL of Pi-free sterile water were used. Positive and negative control samples were incubated for the same amount of time as the paired strain samples. Following incubation, 1 mL of cell/K_2_HPO_4_ mix was sterilized by passing through a Whatman Uniflo 13 mm 0.2 µmpolyethersulfone filter (Cytiva Life Sciences, Emeryville, CA), and the filtrate was saved. Three technical replicates of 100 µL/well of the filtered medium taken from each strain at each time point were pipetted into the wells of a flat-bottomed 96-well Corning microtiter plate. To determine the remaining Pi concentration in the filtrate, the Malachite Green Phosphate Assay Kit (Sigma-Aldrich) was used in accordance with the manufacturer’s instructions. A phosphate concentration ladder provided by the manufacturer and filtered positive and negative controls were also measured. Following incubation with the kit reagents, the OD_620 nm_ for all samples, controls, and ladder controls was collected using a BioTek Synergy HTX Multi-Mode Reader. The average optical density of the technical replicate values was calculated, then the averaged blank OD_620 nm_ was subtracted. This value was converted into a concentration through the use of the kit-provided phosphate concentration ladder and used to calculate the exogenous Pi in the medium. The medium exogenous Pi concentration was subtracted from the known starting Pi, which was based on the calculated concentration of the positive control. Three biological replicates for all strains and time points were performed.

Colorimetric detection of alkaline phosphatase (AP) activity

A modification of a qualitative AP activity assay described previously was used (30). Briefly, strains were streaked on BHI agar containing chloramphenicol and 100 µg/mL of the micronized p-toluidine salt form of 5-bromo-4-chloro-3-indoxyl phosphate (XP) (GoldBio, St. Louis, MO). Plates were incubated overnight at 37°C and then imaged with a Pixel 6 camera (Google, Mountain View, CA) with illumination from a Neewer RL-12 LED Ring Light (Shenzhen Neewer Technology, Guangdong, China).

Alkaline phosphatase activity assays

Overnight cultures of strains grown in BHI with appropriate antibiotic selection were diluted to an OD_600 nm_ of 0.2 in BHI, then incubated at 37°C for 90 minutes to reach OD_600 nm_ ~0.5. Supernatants were collected and saved, and cells were washed once in 1 M Tris-HCl pH 8.0, then resuspended in 1 mL of 1 M Tris-HCl pH 8.0. A 100 µL aliquot of the suspension was serially diluted and plated for CFU per milliliter enumeration. One hundred microliter aliquots of the cell suspension and the supernatants were diluted in 800 µL AP buffer (1 M Tris-HCl pH 8.0, 0.1 mM ZnCl_2_), mixed with 100 µL of 0.4% p-nitrophenyl phosphate (pNPP) (VWR, Radnor, PA), and incubated at 37°C for 10 minutes. One hundred twenty microliters of a 1:5 mix of 0.5 M EDTA to 1 M KH_2_PO_4_ was used to stop the reaction after 10 minutes (31). Two technical replicates, each containing 100 µL of reaction mixture from each strain, were pipetted into the wells of a tissue culture-treated, flat-bottomed 96-well microtiter plate (Corning). The OD_405 nm_ and OD_600 nm_ were measured on a BioTek Synergy HTX Multi-Mode Reader for each sample. The technical replicate values were averaged, and AP activity was calculated using the following equation modified from Zhang et al.: $[eqn]$ (32). Three biological replicates were performed.

Evaluation of phosphorus sources to support growth of OG1RF, pstB1, and pstB2 deletion mutants

To evaluate phosphorus-containing compounds capable of supporting the growth of E. faecalis pstB1 and pstB2 mutants in CDM, Phenotype MicroArray PM4A plates for phosphorus and sulfur utilization assays were procured from Biolog (Biolog Inc., Haywood, CA). Bacterial strains were grown statically in BHI broth for 16 hours, washed in 1× Tris-HCl, pH 8.0, then resuspended to an OD_600 nm_ of 0.01 in CDM containing no inorganic phosphorus. The PM4A plates were inoculated with 100 µL per well of the resuspended cultures, then incubated for 16 hours at 37°C in a sealed plastic food storage container lined with damp paper towels in order to maintain humidity. The OD_600 nm_ of each well was measured on a BioTek Synergy HTX Multi-Mode Reader after 16 hours growth. Four to five biological replicates were performed for each strain, with outlier replicates removed after application of the Grubbs Test online calculator (GraphPad Software, Boston, MA).

Growth curves with other phosphorus sources were carried out in CDM as described above. Phosphonate compounds were added to CDM lacking other phosphorus sources at a final concentration of 0.2 mM. Polyphosphate was added to CDM lacking other phosphorus sources at a final concentration of 0.7% (wt/vol).

RESULTS

Confirmation that the E. faecalis OG1RF pst-phoU locus is expressed as an operon

The pst-phoU locus in E. faecalis has not been definitively shown to be organized as an operon as it is in other bacteria, including E. coli, multiple streptococcal and Bacillus species, and Nostoc punctiforme (8, 33–35). In order to investigate this, primer pairs were designed to amplify the intergenic regions spanning from pstS2 through phoU, as well as the adjacent intergenic regions flanking the pst-phoU locus (OG1RF_11471-pstS2 and phoU-liaX; Fig. 2A). cDNA was used as a template, with the rationale that if the pst-phoU locus in E. faecalis is expressed as a polycistronic RNA, then all genes and intergenic regions should be co-transcribed. Figure 2A shows amplification of the intergenic regions from pstS2 through phoU, indicating the presence of reverse-transcribed polycistronic mRNA and supporting the hypothesis that the pst-phoU locus in E. faecalis is organized as an operon. There was no amplification of either the OG1RF_11471-pstS2 or the phoU-liaX intergenic regions, suggesting that these flanking genes are not included in the pst-phoU polycistronic transcript.

The E. faecalis pst-phoU locus is transcribed as an operon and encodes PstB1 and PstB2, which are likely ATPases. (A) Primers were designed to amplify the intergenic regions of the E. faecalis OG1RF pst-phoU locus (pstS2 to phoU), as well as the upstream and downstream flanking genes (OG1RF_11471 and liaX, respectively). Amplification of cDNA from the intergenic regions located between pstS2 and phoU was observed. There was no amplification of cDNA from the intergenic regions between OG1RF_11471 and pstS2 or phoU and liaX. The diagram above the gel image shows the organization of the pst-phoU region in E. faecalis OG1RF. The sizes of the intergenic regions and the approximate regions amplified are indicated. Gel image is representative of four biological replicates. (B) Alignment of amino acid sequences for PstB from E. coli BW25113 strain K-12 and Streptococcus pneumoniae strain R6, which are known ATPases, and PstB1 and PstB2 from E. faecalis strain OG1RF. Sequence motifs characteristic of ATPases are identified by colored boxes as follows: orange, Walker A motif; brown, Walker B motif; black, LSGGQ motif; and purple, H-loop. Individual amino acids are colored as follows: goldenrod, hydrophobic side chain; teal, positively charged side chain; gray, negatively charged side chain; and purple, polar uncharged side chain. The alignment was generated with Geneious Prime software (version 2023.0.4).

The E. faecalis pst-phoU locus contains adjacent genes annotated to encode ATP-binding proteins

The E. faecalis OG1RF pst-phoU operon contains adjacent genes named pstB1 and pstB2 (Fig. 1), which encode protein products that are 60% identical/79% similar at the amino acid level based on Basic Local Alignment Search Tool (BLAST) analysis (36). PstB has been annotated as an ATPase in a number of species. ATPases contain two ATP-binding domains with characteristic motifs, called the Walker A and Walker B motifs. The Walker A motif is composed of the sequence GXXGXGK(S/T), where X can be any amino acid (37–39). The Walker B motif is composed of the sequence hhhhDE, where h denotes any hydrophobic amino acid (38, 39). ATPases also contain an ABC transporter-specific signature sequence of LSGGQ and a conserved histidine residue (H-loop) (35, 38–40). To assess whether E. faecalis OG1RF PstB1 and PstB2 (NCBI nucleotide database accession number: CP025020) both have sequence characteristics that are identifiable in known ATPases, we compared the two E. faecalis protein sequences with PstB ATPases from E. coli BW25113 strain K-12 (accession number: NZ_CP009273) and Streptococcus pneumoniae strain R6 (accession number: AE007317.1). The PstB protein of E. coli BW25113 strain K-12 is 51% identical/72% similar to PstB1 and 50% identical/72% similar to PstB2. S. pneumoniae has two pst-phoU loci, one of which contains a single copy of pstB (Fig. 1) that is 62% identical/78% similar to PstB1 and 58% identical/80% similar to PstB2. We were able to identify the four ATPase-specific sequence motifs in both E. faecalis PstB1 and PstB2 (Fig. 2B). The Walker A, LSGGQ, and H-loop motifs were identical among the four aligned sequences, while the hydrophobic amino acids in the N-terminal region of the Walker B motif sequence varied among the four proteins. These results suggest that PstB1 and PstB2 are ATPase components of the E. faecalis Pst ABC transporter.

pstB1 and pstB2 deletion mutants exhibit growth defects when grown in CDM but not BHI

We next generated in-frame, markerless single-gene deletion strains of pstB1 and pstB2 in the E. faecalis OG1RF genetic background. Wild-type copies of pstB1 and pstB2 were each expressed in trans in the wild-type and mutant strains from under the control of the strong constitutive p23 promoter (28). All strains grew well when cultured in BHI broth, an undefined nutrient-rich medium, and incubated at either 37°C or 25°C (Fig. 3A; Fig. S1, respectively).

ΔpstB1 exhibits a growth defect in low-Pi CDM, and CDM does not support growth of ΔpstB2 at 37°C. Bacterial growth was monitored over time for strains grown in BHI (top row), CDM containing 0.2 mM Pi (low-Pi CDM; middle row), and CDM containing 80 mM Pi (high-Pi CDM; bottom row). Graphs in the first column show only strains that carry the empty vector (pPLK2). The second and third columns show strains that express pstB1 or pstB2, respectively, in trans from pPLK2. Each data point is the mean of three biological replicates. Error bars show the standard deviation. Statistical analyses for the middle and bottom rows are found in Tables S5 and S6, respectively.

The ΔpstB1(pPLK2) strain exhibited a lag in growth compared to the wild-type strain in low-Pi (0.2 mM) CDM at 37°C and 25°C (Fig. 3B; Fig. S1 middle row, respectively). The growth lag was reduced, but still present, in high-Pi (80 mM) CDM at both temperatures (Fig. 3;Fig. S1, bottom row). Expression of a wild-type copy of pstB1 in trans in the ΔpstB1 strain largely restored the growth kinetics in both low- and high-Pi CDM (Fig. 3; Fig. S1, middle and bottom rows). Interestingly, providing a wild-type copy of pstB2 in trans in the ΔpstB1 strain, generating a pstB2 merodiploid strain, restored wild-type growth kinetics in both low and high Pi-containing CDM concentrations at 37°C and 25°C (Fig. 3; Fig. S1, middle and bottom rows). Unexpectedly, the ΔpstB2(pPLK2) strain was unable to grow in CDM regardless of Pi concentration or temperature (Fig. 3; Fig. S1, middle and bottom rows). Only expression of pstB2 in trans in the ΔpstB2 genetic background was capable of rescuing growth of the strain in CDM; expression of pstB1 in trans in the ΔpstB2 background was unable to rescue the growth phenotype of the mutant to wild-type (Fig. 3; Fig. S1, middle and bottom rows).

Deletion of either pstB1 or pstB2 significantly increases transcription of the remaining pstB gene

The cyanobacterium Nostoc punctiforme has four copies of pstB across three distinct loci. Deletion of the N. punctiforme pstB1 allele was shown to affect expression of two of the other three pstB genes that are located across its genome (34). Although the E. faecalis pstB1 and pstB2 alleles are in the same locus and are expected to be co-transcribed based on the data in Fig. 2A, there is evidence of alternative internal promoters in the E. coli pst operon (41). Therefore, we evaluated if deletion of pstB1 or pstB2 would affect transcription of the other pstB gene. Since the ΔpstB2 strain did not replicate in CDM (Fig. 3; Fig. S1), we extracted RNA from cells that were incubated in 10% BHI for 6 hours. Ten percent BHI has less Pi than undiluted BHI and supported a minimal number of replications of all strains (Fig. S2A), thereby creating a condition that would likely push the cells toward a phosphorus-starved state while maintaining enough nutrients to support the predicted increased expression of Pi acquisition machinery (e.g., the pst-phoU operon). Due to the low culture density resulting from the 10% BHI (Fig. S2, top row), cultures were collected after 6 hours of incubation to maximize the number of cells harvested. In the pstB2 deletion mutant, pstB1 expression was significantly increased compared to the wild-type (Fig. 4A). The converse, increased expression of pstB2 in the ΔpstB1 deletion strain, was likewise observed (Fig. 4B). The data also confirmed that strains containing pPLK2-pstB1 or pPLK2-pstB2 had significantly increased RNA levels of the corresponding gene (Fig. 4A and B). Finally, overexpression of pstB2 was associated with decreased pstB1 expression in the ΔpstB2(pPLK2-pstB2) strain relative to OG1RF(pPLK2), but the same was not observed with the OG1RF(pPLK2-pstB2) strain (Fig. 4A). Conversely, overexpression of pstB1 had no effect on expression of pstB2 in any of the strains (Fig. 4B). The reason why pstB1 expression is significantly decreased in the ΔpstB2 complementation strain but not the pstB2 merodiploid strain [OG1RF(pPLK2-pstB2)] is not fully clear. Overall, the data in Fig. 4 suggest that there may be a difference in the regulation of pstB1 and pstB2.

*Deletion of either pstB1 or pstB2 significantly increases expression of the other pstB gene, and overexpression of pstB2 in the ΔpstB2 strain is associated with decreased expression of pstB1. RNA isolated from bacterial strains incubated in 10% BHI for 6 hours was reverse transcribed and analyzed via qPCR for expression of (A) pstB1 or (B) pstB2 relative to the reference gene gyrB. The solid black line indicates a relative gene expression level of 1. The dashed black line indicates the lowest possible value for relative gene expression based on the limit of detection of the target genes (see Materials and Methods). Each data point represents an independent biological replicate. Horizontal bars show the mean; error bars indicate the standard deviation. One-way analysis of variance with Tukey’s correction: *, P < 0.05; ***, P < 0.001; ***, P < 0.0001.

Pi uptake is differentially impaired in the pstB deletion mutants

We next compared the roles of pstB1 and pstB2 in E. faecalis Pi uptake with a malachite green-based assay that specifically detects Pi. Strains were grown for either 4 hours (Fig. 5A) or 24 hours (Fig. 5B) in 25% BHI, which was chosen because it supports some replication of the E. faecalis strains (Fig. S2, bottom row) while also providing less phosphorus compared to undiluted BHI. Following growth in 25% BHI, the bacterial strains were washed and further starved of phosphorus for 2 hours by incubating in Pi-free CDM. The bacterial cells were then incubated in K_2_HPO_4_, and Pi uptake was measured over time (Fig. 5). OG1RF(pPLK2), OG1RF(pPLK2-pstB1), and OG1RF(pPLK2-pstB2) displayed similar Pi uptake kinetics following both phosphorus-starvation time intervals, reaching a maximum of ~10 nmol by 60 minutes. Pi uptake in the ΔpstB1(pPLK2) strain was reduced compared to the wild-type strain under both conditions. Moreover, when ΔpstB1(pPLK2) was grown overnight in 25% BHI, uptake was close to 0 nmol for the first 30 minutes after Pi was introduced to the bacteria and thereafter increased by only a fraction of the amount that the wild-type strain increased (Fig. 5B). Expression of either pstB1 or pstB2 in trans in the ΔpstB1 strain restored Pi uptake to wild-type levels. Unexpectedly, the medium containing ΔpstB2(pPLK2) and ΔpstB2(pPLK2-pstB1) contained more Pi than the starting concentration after just 1 minute and stayed level through the duration of the experiment, indicating that neither of these strains took up Pi. The amount of Pi measured in the medium was lower for the ΔpstB2(pPLK2-pstB1) strain compared to the ΔpstB2(pPLK2) strain in the cells grown in 25% BHI for 4 hours (Fig. 5A), suggesting the possibility that pstB1 expression partially rescued the Pi uptake deficiency. However, when the same strain was grown overnight in 25% BHI (Fig. 5B), there was no difference in the amount of Pi in the medium compared to the ΔpstB2(pPLK2) strain. Expression of pstB2 in trans in the ΔpstB2 background resulted in a Pi uptake phenotype similar to the ΔpstB1 mutant in both conditions. Importantly, each time point of this assay contained an equivalent number of viable cells for each strain (Fig. S3A and B). These results indicate that the excess Pi in the ΔpstB2(pPLK2) and ΔpstB2(pPLK2-pstB1) reactions was not the result of cell lysis.

ΔpstB1 exhibits diminished Pi uptake compared to wild-type, while medium in ΔpstB2-containing samples rapidly accrued more Pi than was initially added. Cultures of each strain were grown in 25% BHI for (A) 4 hours or (B) overnight, then incubated in Pi-free medium for 2 hours. Following this Pi starvation step, all strains were mixed with K2HPO4 and incubated for 1, 5, 10, 30, or 60 minutes. Cells were removed by filtration, then exogenous Pi remaining in the medium was measured. Pi uptake was calculated as described in the Materials and Methods. Data and error bars at each time point show the mean and standard deviation, respectively, of three biological replicates. Figure S3D shows the same data with the strains separated by empty vector, pPLK-pstB1, and pPLK-pstB2.

AP activity and expression are differentially increased in the ΔpstB1 and ΔpstB2 mutant backgrounds

In E. coli, periods of Pi starvation have been shown to increase synthesis of AP (also called PhoA or Bap) over 1,000-fold (33). Based on the observation that neither the ΔpstB1 nor the ΔpstB2 mutants were able to take up Pi as efficiently as the wild type (Fig. 5), we hypothesized that AP may be upregulated in these mutants as a means to overcome a Pi deficit. Qualitative analysis of AP activity on 10%, 25%, and 100% XP-supplemented BHI agar suggested that the ΔpstB2(pPLK2) and ΔpstB2(pPLK2-pstB1) strains displayed the strongest AP activities of all the strains tested (Fig. 6A through C, sectors 3 and 6, respectively). There was no color produced by the wild type, nor the pstB1 or pstB2 merodiploid strains in the wild-type background (Fig. 6A through C, sectors 1, 4, and 7, respectively). The ΔpstB2 mutant phenotype was restored to that of the wild-type strain only when pstB2 was expressed in trans in the strain (Fig. 6A through C, sector 9); in trans expression of pstB1 from the pPLK2 plasmid in the strain did not affect the mutant phenotype (Fig. 6A through C, sector 6). The ΔpstB1(pPLK2) strain displayed a slight blue coloration (Fig. 6A through C, sector 2; most visible on 100% BHI agar in panel C), and this phenotype was not present when either pstB1 or pstB2 was expressed in trans in the ΔpstB1 genetic background (Fig. 6A through C, sectors 5 and 8).

*AP activity and phoZ expression are highly increased in the ΔpstB2 strain compared to the wild-type strain. (A–C) Colorimetric-based phenotypes of AP activity on (A) 10%, (B) 25%, or (C) 100% BHI agar containing 100 µg/mL XP. A darker blue color indicates more AP activity. The image shown is representative of three biological replicates. (D) RT-qPCR measurement of relative expression of E. faecalis phoZ, which encodes AP, as compared to the reference gene gyrB. RNA was collected from strains incubated in 10% BHI for 6 hours. The solid black line indicates a relative gene expression level of 1. The dashed black line indicates the lowest possible value for relative gene expression based on the limit of detection of the target genes (see Materials and Methods). Each symbol represents an independent biological replicate. Horizontal bars show the mean; error bars indicate the standard deviation. (E) Quantitative measurement of AP activity in cell-associated fractions by cleavage of the colorimetric substrate pNPP. Each symbol represents an independent biological replicate. Horizontal lines show the mean; error bars indicate the standard deviation. For (D) and (E), one-way analysis of variance with Tukey’s correction: *, P < 0.05; ***, P < 0.0001.

The product of the E. faecalis phoZ gene has been characterized as an AP (31). As follow-up to the qualitative AP activity assay (Fig. 6A through C), we measured expression of phoZ (locus OG1RF_12255) in our OG1RF wild-type and mutant strains (Fig. 6D). We observed increased levels of expression of phoZ in ΔpstB1 and ΔpstB2 compared to wild-type. Of the two knockout mutants, the ΔpstB2 strain displayed a greater level of phoZ expression, which is consistent with the qualitative AP activity results (Fig. 6A through C). Expression of wild-type copies of pstB1 or pstB2 from the pPLK2 plasmid in the ΔpstB1 background restored expression of phoZ to wild-type levels. However, similarly to what was observed above ([Fig. 3 to 5 and 6A through C](#F3 F4 F5 F6)), only expression of pstB2 in the ΔpstB2 strain restored phoZ expression back to wild-type levels; phoZ expression in ΔpstB2(pPLK2-pstB1) was similar to the ΔpstB2(pPLK2) strain.

In gram-negative bacteria, AP activity is highest in the periplasm (42–44). For gram-positive bacteria, AP is relegated to the outer surface of the plasma membrane (45, 46). AP can also be secreted from bacterial cells (47). In order to identify the location of enzymatically active PhoZ in E. faecalis and quantitate the relative amount of AP produced by each strain, we used a pNPP-based colorimetric assay to measure AP activity in the cell-associated fractions of mid-log phase cells (Fig. 6E) and the associated culture supernatants (Fig. S4). We observed increased AP activity in the ΔpstB1 and ΔpstB2 strains compared to wild-type, with ΔpstB2 exhibiting the highest amount of AP activity. Expression of wild-type copies of either pstB1 or pstB2 in ΔpstB1 restored AP activity to wild-type levels. The ΔpstB2(pPLK2-pstB2) strain also exhibited AP activity that was similar to that of the wild-type strain, but the ΔpstB2(pPLK2-pstB1) strain retained a high level of AP activity. The overall amount of AP activity was substantially lower in the culture supernatants (Fig. S4A), with the ΔpstB2(pPLK2-pstB1) supernatant having the most AP activity. Enumeration of CFU per milliliter from the source cultures used in the pNPP-based assay confirmed that there was no significant difference between the cell counts of the different strains (Fig. S4B). Taken together, these results indicate that the phoZ-encoded AP is primarily associated with E. faecalis cells rather than being secreted into the culture supernatant at high levels. This finding is consistent with SignalP 6.0 and previous predictions that E. faecalis phoZ is a lipoprotein (48, 49).

AP is partially responsible for the rapid accrual of external Pi observed in the growth medium of ΔpstB2-containing cultures

The results shown in [Fig. 5 and 6](#F5 F6) led us to hypothesize that the increased Pi present in the medium of the ΔpstB2 strain (Fig. 5) may be due to the strain’s increased AP activity (Fig. 6). Specifically, the increased AP activity could result in enhanced release of Pi from phosphate-containing molecules released from the cells via secretion or upon lysis. In order to test this hypothesis, we generated in-frame deletions of phoZ in the wild-type, ΔpstB1, and ΔpstB2 strain backgrounds. Differences in growth between the strains were minimal (Fig. S5). Deletion of phoZ in the ΔpstB2 background rendered the strain unable to break down XP in a qualitative assay of AP activity (Fig. 7A, sector 3 versus sector 6), while expression of phoZ in trans led to enhanced AP activity in the wild-type and both ΔpstB mutants (Fig. 7, sectors 7–9).

Deletion of phoZ reduces the amount of Pi that accrues in the medium of ΔpstB2-containing samples. (A) Colorimetric-based phenotypes of AP expression on agar containing 100 µg/mL XP for ΔphoZ mutant strains. The image shown is representative of three biological replicates. (B) Overnight cultures of each strain were starved of phosphate for 2 hours, then were mixed with K2HPO4 and incubated for 5, 30, or 60 minutes. One milliliter was removed from each incubated time point and filter sterilized. Exogenous Pi remaining in the medium was measured, and Pi uptake was calculated as described in the Materials and Methods. Data and error bars at each time point show the mean and standard deviation, respectively, of three biological replicates.

Deletion of phoZ from OG1RF and ΔpstB1 had no effect on the Pi uptake phenotypes (Fig. 7B). Similarly, expression of phoZ in trans in the same two strain backgrounds did not alter the Pi uptake phenotypes. In contrast, deletion of phoZ in the ΔpstB2 background reduced the amount of exogenous Pi compared to the ΔpstB2 parental strain; however, by the 5 minute time point, the medium had still accrued more Pi than was initially added (Fig. 7B). Expression of phoZ in trans in the ΔpstB2ΔphoZ strain restored the phenotype of the parent strain [ΔpstB2(pPLK2)]. Similar numbers of viable cells were used for each strain, indicating that the observed results were most likely due to genetic differences between the strains (Fig. S3C). Therefore, from these results, we conclude that the increased activity of the phoZ-encoded AP in the ΔpstB2 strain partially accounts for the accumulation of Pi in the culture medium that we unexpectedly observed in Fig. 5.

Investigation of alternate sources of phosphorus capable of sustaining growth of ΔpstB2 in CDM

The observation that the ΔpstB2 mutant was incapable of growth in CDM regardless of the concentration of Pi present, but was capable of reaching stationary phase in BHI broth (Fig. 3), suggests that BHI contains a phosphorus source that the ΔpstB2 mutant can use for its metabolic needs in lieu of Pi. The CDM recipe used in this work contains no other sources of phosphorus beyond Pi, except for minute traces of inorganic phosphate in the form of NADP and vitamin B_12_. Concentrations of these two molecules are roughly 0.25 mg and 0.01 mg per 0.01 L of CDM, respectively (24, 25). No growth was observed for any of the strains used in this study when they were incubated in CDM lacking any other source of phosphorus (beyond the trace amounts supplied by the base medium) (data not shown). These results led us to hypothesize that ΔpstB2 was utilizing an alternate, non-Pi source of phosphorus in order to meet its metabolic requirements when incubated in BHI.

We screened a library of 59 phosphorus-containing molecules present on Biolog PM4A plates in order to identify which phosphorus-containing compounds could support growth of OG1RF, ΔpstB1, and ΔpstB2 in CDM lacking exogenous phosphorus. The library contained both inorganic phosphorus, including Pi, and organic phosphorus compounds that were composed of carbohydrates, amino acids, nucleotides, phosphonates, and hypophosphite (Fig. 8). The wild-type and ΔpstB1 strains had similar patterns of growth when incubated in CDM with the phosphorus-containing compounds. Meanwhile, the maximum end point OD_600_ observed for ΔpstB2 for any of the compounds was ~0.2, and this only occurred in the wells containing inositol hexaphosphate, thymidine 3´-monophosphate, and uridine-2´,3´-cyclic monophosphate. However, in follow-up experiments, inositol hexaphosphate was insufficient at supporting growth of the ΔpstB2 strain when it was used as the sole phosphorus source in CDM (data not shown). Overall, these results indicate that the ΔpstB2 mutant is severely limited in the phosphorus-containing compounds it can catabolize to meet its metabolic needs.

Identification of phosphorus-containing compounds that support growth of the wild-type, ΔpstB1, or ΔpstB2 strains in CDM. Biolog PM4A plates containing 59 different phosphorus-containing molecules were inoculated with either wild-type, ΔpstB1, ΔpstB2, ΔphoZ, ΔpstB1ΔphoZ, or ΔpstB2ΔphoZ suspended in CDM with 0.0 mM Pi. Bacterial growth was measured by reading the OD600 nm of each well after incubation in a humidity chamber at 37°C for 16 hours. The 59 separate compounds are grouped as inorganic phosphorus molecules (black) and organic phosphorus molecules that are further subdivided into carbohydrates (blue), amino acids (red), nucleotides (green), and phosphonates and phosphites (purple). N = 3–4 biological replicates for OG1RF, ΔpstB2, and the three phoZ deletion strains, and 4–5 biological replicates for ΔpstB1. Outliers identified by the application of the Grubb’s test were excluded from the data set.

We also screened the OG1RF, ΔpstB1, and ΔpstB2 phoZ deletion strains on the PM4A plates in order to determine how removal of the AP gene would impact which phosphorus-containing molecules these strains were capable of catabolizing (Fig. 8). The OG1RFΔphoZ and ΔpstB1ΔphoZ mutants were unable to grow in as many phosphate-containing compounds as the parent strains. This result suggests that the compounds which supported growth of AP-containing strains but not AP-deficient strains may be processed by AP outside of the cell prior to uptake through the Pst-PhoU importer. Like the ΔpstB2 parent strain, the ΔpstB2ΔphoZ strain was incapable of substantial growth in CDM supplemented with any of the phosphorus-containing sources present on the PM4A plate.

Finally, we evaluated whether polyphosphate or the phosphonates APP, APB, AEP, and MPP could support growth of the ΔpstB2 strain in CDM lacking all other phosphorus compounds. Polyphosphates are chains of Pi molecules, sometimes hundreds of molecules long, linked by energy-rich phosphoanhydride bonds (50). Phosphonates, which contain C−PO(OR)2 groups (R = alkyl, aryl, or H), are used by some bacteria as their sole sources of phosphorus (51), and E. faecalis has been shown to take them up (52). In addition, phosphonates have been isolated from the brains of cattle, which is one of the components used to make BHI (53). Neither OG1RF, ΔpstB1, nor ΔpstB2 was able to grow in CDM supplemented with polyphosphate, APP, AEP, APB, or MPP as the sole phosphorus source (Fig. S6).

DISCUSSION

Phosphorus is essential for the synthesis of many organic molecules vital for life, including ATP, phospholipids, and nucleic acids. Bacteria can acquire phosphorus from the environment in the form of Pi directly through specific importers, including the low-affinity, high-velocity Pit and the high-affinity, low-velocity Pst transporters (3). The coding region of the Pst transporter, the pst-phoU operon, has been well characterized in E. coli since it was initially described in the 1970s (3, 4, 54). Since then, the locus has been associated with adherence, fimbriae production, colonization, virulence, immune evasion, and antimicrobial resistance in E. coli and other gram-negative bacterial species (6, 7, 55–62). In gram-positive bacteria, where significantly less work characterizing pst-phoU loci has been published, disruption of normal expression of the pst-phoU locus has been shown to induce nutritional immunity in Staphylococcus aureus and reduce Streptococcus mutans adhesion to abiotic surfaces (63, 64). In E. faecalis, mutations in genes of the pst-phoU locus have been identified in several genetic studies. Specifically, mutations in pstB1, pstB2, and pstC were identified in clones that were serially passaged in pH 9 medium (65). Mutations in pstB2 and pstC that were associated with small colony morphologies were also isolated from serially passaged biofilms grown in pH 9 medium (65). A pstB2 mutation co-occurred with mutations in genes encoding a hypothetical protein and N-acetylmuramoyl-L-amidase in an E. faecalis clone isolated from in vitro evolution of a strain lacking the croRS two-component signaling system-encoding genes; the isolated clone grew faster than and had altered susceptibilities to vancomycin and teixobactin compared to the parental croRS deletion strain (66). Finally, transposon insertions in phoU have been identified in screens for genes needed for growth in nutrient-rich medium and biofilm formation (67, 68). Yet, despite this collection of genetic evidence that suggests broad roles for the pst-phoU locus genes in E. faecalis physiology, no follow-up functional characterization studies have been reported. In this work, we report the first direct investigation of any genes within the E. faecalis pst-phoU operon.

Many bacteria possess multiple homologs of genes found in the canonical pst-phoU operon in distinct chromosomal locations (Fig. 1). A unique and unstudied feature of the pst-phoU locus found in some Firmicutes, including E. faecalis, is the tandem arrangement of two ATPase-encoding pstB homologs (8, 69). Our findings demonstrate that the adjacent and non-identical pstB1 and pstB2 genes in the pst-phoU operon of E. faecalis OG1RF (Fig. 2) are differentially required for growth, phosphate uptake, and AP activation ([Fig. 3, 5, and 6](#F3 F5 F6)). Overexpression of pstB2 in the ΔpstB1 strain is sufficient to restore growth in low-Pi CDM (Fig. 2), phosphate uptake (Fig. 3), and phoZ expression with corresponding AP activity (Fig. 6) to wild-type levels. In contrast, overexpression of pstB1 in the ΔpstB2 strain does not rescue any of the phenotypes we evaluated, suggesting that pstB2 is necessary for Pi uptake in E. faecalis.

Our data revealed that there is a significant change in expression of pstB1 or pstB2 when the tandem pstB homolog is deleted, and in trans overexpression of pstB2 in the pstB2 chromosomal deletion strain reduced the expression of pstB1 (Fig. 4). The same strain [ΔpstB2(pPLK2-pstB2)] also took up Pi at a level similar to the ΔpstB1 mutant, which was reduced relative to the wild-type strain (Fig. 5). We hypothesize that the decreased pstB1 expression in the ΔpstB2 complementation strain resulted in the strain phenocopying the ΔpstB1 mutant in the Pi uptake assay. Data from the filamentous cyanobacterium Nostoc punctiforme provide evidence for regulatory interactions between pstB paralogs (34). To the best of our knowledge, N. punctiforme is the only microbe in which the effect of having multiple pstB genes has been investigated. N. punctiforme possesses four pstB genes (pstB1–pstB4) that are arranged across three distinct loci; pstB1 and pstB4 are in single copy in their respective loci that each also contain single copies of pstS, pstC, and pstA, while pstB2 and pstB3 are arranged in tandem in a third locus that has single copies of pstC and pstA and lacks pstS (34). Hudek et al. generated a ΔpstB1 mutant and found that starving the strain of Pi led to increased expression of pstB2 and pstB4, while pstB3 expression remained the same when compared to the wild-type strain (34). Interestingly, they found that overexpression of pstB1 led to reduced expression of pstB3 compared to the wild-type strain following Pi starvation (34). However, the fact that N. punctiforme has four pstB paralogs and that Hudek et al. did not genetically disrupt the two copies that are tandemly arranged (pstB2 and pstB3) makes it difficult to extrapolate the relevance of these findings to E. faecalis.

The Pho regulon consists of multiple genes and operons involved in the regulation of phosphate uptake and storage (70). Mutations within the pst-phoU operon are a known cause of pho regulon dysregulation, which results in increased AP activity (6, 33, 54, 71, 72). We found that deletion of either pstB1 or pstB2 results in increased AP-based enzymatic activity compared to wild-type, with disruption of pstB2 having the highest levels of AP activity (Fig. 6). Additional experimentation will be necessary to determine the overall effect of deletion of pstB1 or pstB2 on Pho regulon activation in E. faecalis. The high AP activity of the ΔpstB2 strain was largely responsible for the increased accumulation of extracellular Pi that we observed unexpectedly in the malachite green-based Pi uptake assays ([Fig. 5 and 7](#F5 F7)). Despite this, there was still Pi accumulation in the medium from the ΔpstB2ΔphoZ sample (Fig. 7). One possible explanation for this observation is that E. faecalis may produce an as-yet-unidentified alkaline or acidic phosphatase that is capable of cleaving Pi from waste molecules. However, this explanation is unlikely given that none of the ΔphoZ strains cleaved the chromogenic substrate XP (Fig. 7A). An alternative explanation is that intracellular Pi is either leaked or actively exported from the cells into the surrounding medium. There are a number of known systems in bacteria that function through the export of Pi. For example, the systems encoded by the glp, pgt, and uhp loci are Pi antiporters that use the change in electrochemical gradient to import various organic phosphate molecules (73–75). Additionally, the yjbB gene encodes a Pi exporter, which is hypothesized to help maintain cellular phosphate homeostasis through the active export of Pi into the surrounding medium (76). Further studies will be necessary to determine if any such systems are present in the genome of E. faecalis.

We found that pstB2 is dispensable for growth in BHI, a rich undefined medium, but is required for growth in medium in which Pi is the only source of phosphorus (Fig. 3). Supplementing phosphorus-free CDM with single alternate inorganic and organic phosphorus-containing compounds did not rescue growth of ΔpstB2 (Fig. 8; Fig. S6), suggesting that deletion of pstB2 results in a mutant that is incapable of taking up phosphorus efficiently from any single phosphorus-containing source. BHI and other rich undefined media likely contain large concentrations of multiple different phosphorus-containing molecules. Our data suggest that the ΔpstB2 strain may require multiple phosphorus-containing molecules, as would be found in BHI, to meet its metabolic requirements. Our results also argue against E. faecalis OG1RF having a second functional dedicated Pi importer, such as the low-affinity, high-velocity Pit system that is present in E. coli (3, 4). Indeed, a BLAST search of the E. faecalis OG1RF genome (NCBI nucleotide database accession number: CP025020) for homologs of PitA or Pit B from E. coli strain W3110-P (NCBI nucleotide database accession number: NZ_CP084899) yielded only a single result: locus tag CVT43_09860 (OG1RF_11873), listed as encoding an “inorganic phosphate importer,” which was 32% and 29% identical to the amino acid sequences of PitA and PitB, respectively. If the CVT43_09860 (OG1RF_11873) locus does encode a Pit ortholog in E. faecalis, and it is functional, then it appears that its ability to take up Pi is insufficient to sustain growth on its own.

In conclusion, the results of this study support the hypothesis that pstB2 encodes an ATPase that is required for Pi import in E. faecalis, while the ATPase encoded by pstB1 has an accessory role in Pi import that can be duplicated when excess PstB2 is available. The data establish that E. faecalis is dependent on the pst-phoU operon to meet its Pi importation requirements. Therefore, targeting the Pst importer may be an effective strategy for future therapeutic interventions to combat antimicrobial-resistant enterococcal infections.