Genetically distant bacteriophages select for unique genomic changes in Enterococcus faecalis

Abstract The human microbiota harbors diverse bacterial and bacteriophage (phage) communities. Bacteria evolve to overcome phage infection, thereby driving phage evolution to counter bacterial resistance. Understanding how phages select for genetic alterations in medically relevant bacteria is important as phages become established biologics for the treatment of multidrug‐resistant (MDR) bacterial infections. Before phages can be widely used as standalone or combination antibacterial therapies, we must obtain a deep understanding of the molecular mechanisms of phage infection and how host bacteria alter their genomes to become resistant. We performed coevolution experiments using a single Enterococcus faecalis strain and two distantly related phages to determine how phage pressure impacts the evolution of the E. faecalis genome. Whole‐genome sequencing of E. faecalis following continuous exposure to these two phages revealed mutations previously demonstrated to be essential for phage infection. We also identified mutations in genes previously unreported to be associated with phage infection in E. faecalis. Intriguingly, there was only one shared mutation in the E. faecalis genome that was selected by both phages tested, demonstrating that infection by two genetically distinct phages selects for diverse variants. This knowledge serves as the basis for the continued study of E. faecalis genome evolution during phage infection and can be used to inform the design of future therapeutics, such as phage cocktails, intended to target MDR E. faecalis.


| INTRODUCTION
Enterococcus faecalis is a Gram-positive bacterium naturally residing as a commensal in the gastrointestinal tracts of animals, including humans (Lebreton et al., 2014). Immune suppression and/or antibiotic treatment promotes E. faecalis to outgrow and become a dominant member of the microbiota, leading to opportunistic infections (Ubeda et al., 2010). Strains of E. faecalis and Enterococcus faecium have acquired traits that allow them to survive host and environmental stresses, contributing to their success as pathogens (Lebreton et al., 2017;Tyne & Gilmore, 2014). The overuse of antibiotics in both medical and agricultural settings supports enterococcal pathogenesis by driving multidrug-resistant (MDR) phenotypes (Gawryszewska et al., 2017;Munoz-Price et al., 2005). As MDR E. faecalis infections continue to persist worldwide, there is a need to find alternative therapeutics capable of bypassing existing modes of antibiotic resistance (Esmail et al., 2019;Farman et al., 2019;Remschmidt et al., 2018).

Bacterial viruses, bacteriophages (phages), exist in high numbers
in the intestinal tract where they infect and sometimes kill host bacteria, likely influencing the structure of the microbiota Minot et al., 2013;Wandro et al., 2019). Due to their narrow host specificity and ability to lyse bacteria, phages are becoming an essential resource for the treatment of MDR bacterial infections (Kortright et al., 2019). Phage therapy offers many advantages over traditional antibiotics. For example, specificity can be tailored to target only the desired bacteria, leaving native microbes largely unaffected (Brives & Pourraz, 2020;Sulakvelidze et al., 2001).
Additionally, phage replication is restricted to the abundance of the host, thus upon host exhaustion phages are depleted from the population (Moelling et al., 2018). In contrast, conventional antibiotics lack specificity, killing resident bacteria, and the compounds can remain in the patient after the infection has cleared (Langdon et al., 2016). There is an ever-growing repertoire of phages that infect E. faecalis (Cook et al., 2021), making these promising candidates for phage therapy.
The development of successful phage therapies will require a complete understanding of the genetic interactions between phages and bacteria. Although phage therapy holds promise for the treatment of E. faecalis infections (Gelman et al., 2018;Khalifa et al., 2015), the molecular mechanisms of enterococcal phage infection and the bacterial host response to phage infection are understudied. Phage tail protein-receptor interactions underpin the molecular basis for phage strain specificity of the bacterial cell surface (Kabanova et al., 2019;Tu et al., 2017). To date, only the transmembrane protein PIP EF (phage infection protein of E. faecalis) has been identified as a bona fide enterococcal phage receptor (Duerkop et al., 2016). Both phages VPE25 and VFW bind to E.
faecalis by engaging with cell surface polysaccharides, and infection proceeds following viral DNA entry which requires PIP EF (Duerkop et al., 2016). Studies in E. faecium have identified cell wall polysaccharides, secreted antigen A, and RNA polymerase to be involved in phage infection (Canfield et al., 2021;Wandro et al., 2019). Other studies have identified the enterococcal polysaccharide antigen (Epa) as a coreceptor for E. faecalis phages (Chatterjee et al., 2019;Ho et al., 2018).
To further our understanding of phage-enterococcal interactions and their impact on genome evolution, we cocultured E. faecalis SF28073, an MDR strain resistant to vancomycin, gentamicin, and erythromycin, with two genetically distant phages, VPE25 and phage 47 (phi47) (Chatterjee et al., 2019;Duerkop et al., 2016). Both phages are long noncontractile-tailed siphophages with double-stranded DNA genomes and rely on the enterococcal polysaccharide antigen (Epa) for infection (Chatterjee et al., 2019(Chatterjee et al., , 2020. Epa of SF28073 has only been characterized based on predicted gene annotations (Palmer et al., 2012), and the biochemical structure has not been determined.
The biochemical structure of Epa from a different MDR E. faecalis strain V583 has been determined (Guerardel et al., 2020). VPE25 has been confirmed to be a lytic phage (Duerkop et al., 2016), and we hypothesize that phi47 is lytic due to its lack of integrases, yet direct experiments need to be performed. Although both phages infect E. faecalis strain SF28073, nucleotide alignment revealed that their genomes only share 37.3% nucleotide identity, indicating they are genetically distinct. Orthologous protein clustering confirmed that these phages belong to unique enterococcal phage lineages. Based on these observations, we hypothesized that E. faecalis SF28073 may gain single-nucleotide polymorphisms (SNPs) in genes involved in cellular pathways and macromolecule production that are specific to infection by either phage. To test this hypothesis, we ran two parallel coculturing experiments. In this report, predation by phages VPE25 and phi47 selects for unique genetic subpopulations of E. faecalis SF28073. We identified mutations in known macromoleculeencoding genes previously demonstrated to be necessary for E. faecalis phage infection; however, numerous novel mutations were also identified within a lower percentage of the E. faecalis population. Our work shows that surface-associated factors are the major driver of E. faecalis phage resistance, yet genetic alterations emerge that implicate diverse metabolic pathways in the E. faecalis response to phage infection. Additionally, our data suggest that the ratio of phage to bacteria is an important factor when studying phage-bacterial coevolution in vitro, as bottlenecks during serial passage may favor phage extinction.
2 | MATERIALS AND METHODS 2.1 | Routine bacterial culture E. faecalis SF28073 (isolated from urine in 2003) (Oprea et al., 2004) was cultured in brain heart infusion (BHI, BD) medium at 37°C. Individual phi47-resistant colonies were identified by inoculating washed bacterial samples from whole cultures into BHI and grown overnight at 37°C, normalizing overnight cultures to an optical density (OD) of 0.1, and serial dilutions were spotted onto THB agar containing 10 mM MgSO 4 and 1 × 10 7 PFU/ml of phi47. Plates were incubated overnight at 37°C and resistant isolates were picked, grown overnight in BHI at 37°C, and DNA was extracted and sequenced following the methods described below.

| Coevolution assay
Individual colonies of E. faecalis SF28073 were grown overnight. The

| Testing bacterial cross-resistance
From glycerol stocks of whole culture samples (bacteria and phage), BHI was inoculated, and the culture was grown overnight at 37°C with shaking at 250 rpm. Overnight cultures were washed three times with SM-plus buffer to remove any extracellular phage. Bacterial pellets were resuspended in SM-plus buffer and normalized to an OD of 0.1. Ten-fold serial dilutions were spotted on THB agar plates containing 10 mM MgSO4 and 1 × 10 8 PFU/ml of VPE25 or phi47. Bacteria coevolved with phi47 were spotted on VPE25 plates while bacteria coevolved with VPE25 were spotted on phi47 plates. CFU/ml was enumerated after overnight incubation at 37°C. Relative viability was calculated by dividing the CFU/mL of bacteria grown on phage agar plates divided by CFU/ml of bacteria grown in the absence of phage.

| DNA extraction for population sequencing
Genomic DNA was isolated from 1 ml culture aliquots from coevolution cultures consisting of phage and bacteria using a previously described protocol for E. faecalis (Manson et al., 2003). Briefly, samples were treated with 5 mg/ml lysozyme for 30 min at 37°C. In all, 0.5% SDS, 20 mM EDTA, and 50 μg/ml Proteinase K were added and incubated at 56°C for 1 h. Samples were cooled to room temperature before adding an equal volume of phenol/chloroform/isoamyl alcohol and extracted by shaking. Samples were centrifuged at 17,000 rcf for 1 min, and the aqueous layer was extracted with an equal volume of chloroform. Again, samples were centrifuged at 17,000 rcf for 1 min, and nucleic acids were precipitated from the aqueous layer by adding 0.3 M NaOAc [pH 7] and an equal volume of isopropanol. Nucleic acid was pelleted by centrifuging at 17,000 rcf for 30 min at 4°C, washed with 70% ethanol, and centrifuged at 17,000 rcf for 10 min. Finally, the pellet was dried and resuspended in sterile water. Genomic DNA was sequenced using the Illumina NextSeq. 2000 platform to 300 Mbp depth at the Microbial Genome Sequencing Center (MiGS).

| Hybrid assembly of the E. faecalis SF28073 genome
The E. faecalis SF28703 genome was sequenced using Oxford Nanopore Technology (ONT) as described previously (Jain et al., 2016;Wick et al., 2017). Briefly, 1.5 μg genomic DNA was mechanically sheared into 8 kb fragments with a Covaris g-tube per the manufacturer's instructions before library preparation with the ONT Ligation Sequencing Kit 1D (SQK-LSK108). Libraries were base called with MinKNOW (v3.5.5) to generate FASTQ and fast5 sequence JOHNSON ET AL. | 3 of 18 reads. Illumina reads were obtained from MiGS as described above.
Incomplete assemblies were manually completed as described in the "Unicycler tips for finishing genome" page (https://github.com/ rrwick/Unicycler/wiki/Tips-for-finishing-genomes). Briefly, Bandage (v0.8.1) was used to visualize the completion status of the assembly (Wick et al., 2015), and unassembled contig sequences were extracted. Using these unassembled contig sequences as baits, long reads from MinION sequences were gathered for incomplete regions using minimap2 (v2.11-r797) and an in-Bandage BLAST search was performed with the long reads against the graph (Li, 2018). If long reads supported the continuity of two unassembled contigs, then the Bandage graph editing function was used to duplicate, delete edges, and merge contigs. The complete assembly sequence was saved from Bandage in FASTA format.

| Analysis of serially passaged bacterial populations using Illumina sequencing
Illumina reads from the bacterial populations obtained from MiGS were mapped to the assembled E. faecalis SF28073 chromosome (GenBank accession number CP060804) and the three endogenous plasmids (pSF1, CP060801; pSF2, CP060802; and pSF3, CP060803) using CLC Genomics Workbench (Qiagen) with default settings. Detailed read mapping statistics were generated using the "QC for read mapping" tool in CLC Genomics Workbench with default settings to obtain the range of coverage and zero coverage regions in each assembly. The "Find low coverage" tool in CLC Genomics Workbench with the low coverage threshold set at 0 was used to manually inspect the regions found by the quality analysis to contain regions with 0 coverage. Sequence variants were identified using the "Basic variant detection" tool with a minimum coverage of 100, the minimum frequency of 30% (count/coverage] × 100), and a ploidy of 0. All variants identified were manually examined, and mutations not altering the amino acid sequence (silent) were excluded from the analysis. Variants present in the bacteria-only controls were also excluded from further analysis.

| Phage 47 genome sequencing and analysis
Phi47 genomic DNA was isolated following the methods described above. The phi47 genome was assembled de novo from Illumina reads using CLC Genomics Workbench following the default parameters, with the largest contig forming the full genome. Genome annotation was performed using RAST and NCBI BLASTp (v2.0) (Altschul et al., 1990;Aziz et al., 2008). During coevolution with E. faecalis SF28073, culture media was filtered through a 0.45-μm filter.
DNA was extracted from filtered media using the proteinase K and phenol/chloroform method described above and sequenced using Illumina technology at MiGS. These reads were mapped to the wildtype phi47 genome and SNPs were identified using CLC Genomic Workbench with a minimum coverage of 15, a minimum frequency of 30%, and a ploidy of 0.

| OrthoMCL analysis
Enterococcal phage phylogeny was determined using OrthoMCL (Li et al., 2003) as described previously (Canfield et al., 2021). Enterococcal phage genomes were downloaded from the INPHARED phage genome database (Cook et al., 2021). As of July 1, 2021, there were 126 enterococcal phage genome sequences available in addition to our inclusion of the phi47 genome. Proteomes determined using Prodigal (Hyatt et al., 2010) were used as input into an Or-thoMCL MySQL database. A cluster inflation value of 1.5 was used and the resulting matrix was input for ggdendro and ggplot2 packages in R version 3.6.3. The dendrogram was determined using the average linkage method for hierarchical clustering of Manhattan distance metrics.
EasyFig comparison of the phage genomes revealed only three shared genes (lines, Figure 1a) (Sullivan et al., 2011). These genes exhibit 67% or greater identity at the nucleotide level.
The phi47 genome is 57,289 base pairs in length, consisting of 101 predicted open reading frames (ORFs) and two tRNAs. Using RAST genome annotation (Aziz et al., 2008)   Comparative genomic analysis of phi47 was performed with all publicly available enterococcal phage genomes using OrthoMCL ( Figure 1c) (Cook et al., 2021). This algorithm generates a phylogenetic tree of clustered phage genomes (orthoclusters) based on orthologous proteins (Bolocan et al., 2019;Canfield et al., 2021;Li et al., 2003). Of the 11 known orthoclusters (Bolocan et al., 2019;Canfield et al., 2021), phi47 is placed into cluster III, while VPE25 is in cluster X. Together, these genetic analyses demonstrate the lack of common genes between phages VPE25 and phi47, making them genetically distinct. Nonsynonymous, unique bacterial SNPs were observed in all experimental replicates, except in one of the VPE25-challenged replicates where the mutation frequencies were below our 30% population-wide cutoff. Interestingly, the mutations selected in E.

|
faecalis SF28073 challenged with phage VPE25 largely differed from mutations in E. faecalis SF28073 challenged with phi47. As expected, we observed mutations in pip EF at one or more time points in all VPE25 challenged replicates, except the culture mentioned above which did not meet our read mapping cutoff (Table A1). We detected epa mutations in four of the cultures infected with phi47 (Table A2).
These two macromolecules have been previously reported to be essential for successful infection of these phages; the integral membrane protein PIP EF is the receptor for VPE25, while both VPE25 and phi47 rely on the enterococcal polysaccharide antigen (Epa) for adsorption (Chatterjee et al., 2019(Chatterjee et al., , 2020Duerkop et al., 2016).
We identified ccpA as the only common gene mutated when SF28073 was challenged with either VPE25 or phi47 (Tables A1   and A2). When exposed to phage VPE25, ccpA had mutations in two replicates which appeared at different time points (Days 7 and 14, Table A1), while exposure to phi47 resulted in one replicate harboring a ccpA mutation on Day 14 (Table A2). In E. faecalis, catabolite control protein A (CcpA) plays a key role in regulating the transcription of proteins involved in carbon source utilization (Gao et al., 2013). Moreover, both experimental groups had mutations arise in different components of the SUF system, which is involved in the iron-sulfur (Fe-S) cluster assembly pathway (Srour et al., 2020) (Tables A1 and A2). One VPE25-challenged replicate had a mutation in sufU, encoding a sulfur relay protein (Srour et al., 2020), while one phi47-challenged replicate had a mutation in sufD. Both mutations were identified on Day 7 and were maintained until Day 14. This suggests that both phages may utilize bacterial iron-sulfur complexes during infection. We also found a mutation in the gene manX which was identified in one replicate challenged with VPE25. manX is a component of the manXYZ operon encoding a protein complex involved in the transportation of mannose as well as other carbohydrates (Plumbridge, 1998).
Additionally, one replicate challenged with phage VPE25 had mutations in genes encoding a putative restriction-modification (R-M) system, located within the two specificity (S) subunit genes, specifically genes H9Q64_13860, and H9Q64_13845 (Table A1). In R-M systems, the S subunit genes, composed of two target recognition domains, recognize specific DNA sequences thereby providing target specificity to the R-M complex (Gao et al., 2011). On Day 7, both subunits shared missense mutations resulting in amino acid changes from leucine to phenylalanine and lysine to glutamine. Surprisingly, on Day 14 these mutations were no longer detected, suggesting that they rendered these cells less fit in the population.
Other nonsynonymous mutations specific to each S subunit gene were also found on Day 7. These two mutations were observed again on Day 14 at higher frequencies, which on the contrary, suggest that these mutations provided a fitness advantage to the population.  Figure 1a. Lollipops represent genes with identified nonsynonymous mutations found in each culture. Lollipop color corresponds with culture number. (c) OrthoMCL was used to compare the phage 47 genome to all publicly available enterococcal phage genomes. A phylogenetic tree was generated from OrthoMCL. Height is the average linkage of hierarchical clustering with 1000 iterations using the Manhattan distance metric. In all, 126 enterococcal phage genomes from the INPHARED database were used for comparison to phi47 (in red text). Distinct phage orthoclusters are represented by colored boxes. Roman numerals next to shaded boxes designate the orthocluster number Lastly, Day 14 revealed two additional mutations in both S subunitencoding genes that were not present on Day 7. The numerous amino acid changes observed across the S subunits of the R-M system suggest these mutations may be increasing the specificity of the subunit S towards recognizing the VPE25 genome.
H9Q64_01755, a predicted transposase, was mutated in two replicates. Both replicates had the same mutation resulting in a change from arginine 144 to leucine. H9Q64_09795 was also mutated in two replicates. This gene, epaAC, is a predicted epimerase/dehydratase (Guerardel et al., 2020). Lastly, H9Q64_09850 was mutated in three replicates. This gene is epaR, the final gene in the rhamnose-sugar biosynthesis locus of epa that is a predicted priming glycosyltransferase (Guerardel et al., 2020). Epa is involved in phage adsorption (Chatterjee et al., 2019;Ho et al., 2018), making this gene essential for successful phi47 infection.
We looked closer at the distribution of these mutations among single clones to better understand their distribution in the population.
We sequenced a total of seven colonies, all of which are resistant to phi47, corresponding to Day 7 Cultures 1, 4, and 5 (Table A3). We selected Day 7 since it was the last sample where the bacteria were cocultured with phi47, and these cultures were selected due to their different phi47 kinetics on Day 7 (Figure 2). We observed epa mutations in all seven colonies, including those derived from Culture 4, indicating that the frequency of epa mutations in Culture 4 was below our limit of detection by whole-genome sequencing but can be resolved when analyzing single clones. All of the isolates from Culture 1 shared mutations in ptsP, sufD, and a transposase.
To determine if the mutations protected SF28073 from infection by other phages, we challenged bacteria from Day 7 cultures with the phage it was not challenged with ( Figure 3).
Bacteria from Day 7 cultures were grown overnight, normalized to an OD of 0.1, and spotted on agar plates that contain phage, or agar without phage. Bacteria that were coevolved with phi47 are resistant to infection by VPE25 (Figure 3a), while bacteria that were coevolved with VPE25 are susceptible to phi47 infection (Figure 3b).

| Phi47 acquires mutations in tail and hypothetical genes during coevolution with E. faecalis SF28073
Phi47 is an uncharacterized phage; little is known about genes required for phi47 infection other than its dependence on Epa (Chatterjee et al., 2019). To further explore the mechanisms of phi47 infection, we enumerated both phi47 and E. faecalis SF28073 to determine the population kinetics for each experimental replicate ( Figure 2) and sequenced the phi47 population throughout our experiment. We observed different phage abundance patterns across the five replicates, despite each being treated identically. While all replicates had an expected spike in phi47 titer on Day 1 after 24 h of replication in a completely susceptible population, and a reduction in titer on Day 3, phage abundance differed for each replicate on Day 7 ( Figure 2). Culture 1 phi47 titer spiked and was followed by a continuous decline until phi47 was no longer detectable in the culture via plaque assay by Day 11. Cultures 2 and 3 had no detectable phi47 on Day 7. Cultures 4 and 5 had low phage titers on Day 7. Culture 4 had a phi47 spike on Day 9 followed by a decline until it was no longer detected on Day 12, and phi47 was undetectable in Culture 5 by Day 9.
Due to the differences observed in phi47 abundance throughout these experiments, we sequenced the phage population of each replicate on Days 0, 1, 3, and 7. Day 14 was excluded from analysis because no replicate had detectable phages by plaque assay. Sequencing of phage DNA revealed that each culture had a unique mutation profile. Table 3 in the Appendix (Table A3) includes all observed mutations, including those in intergenic regions and synonymous changes. In our analysis, we were most interested in SNPs that result in nonsynonymous changes. Phages from Cultures 2 and 3 acquired the most SNPs, but viable phages were undetectable at F I G U R E 2 Phi47 kinetics differ in each experimental replicate. Bars represent the CFU/ml shown on the left Y-axis, lines represent the PFU/ ml shown on the right Y-axis. Lines and bars of the same color represent E. faecalis SF28073 and phi47 titers, respectively. Bars represent the mean of three technical replicates, while lines are a single biological replicate. Additional plaque assays were performed to identify when phi47 became undetectable in each culture Day 7, suggesting that these acquired mutations in the various tail and hypothetical protein-encoding genes may have been deleterious to the phages' ability to overcome bacterial resistance mutations.

Phages from Cultures 3 and 4 developed identical tail SNPs on Day 7.
While there were no infectious phages detectable by plaque assay in Culture 3 on Day 7, we were still able to recover phage DNA in culture media, allowing us to perform genetic analyses. Culture 5 phages only developed one SNP in the gene encoding the tail fiber protein on Day 3, which was maintained on Day 7. Culture 1 phages developed no SNPs. Interestingly, there are three hypothetical genes, the major tail protein gene, and one tRNA that were mutated in phages across multiple cultures (Table A4). Despite the different mutations observed across replicates, phi47 was not detectable in any of the cultures by the end of the passaging, indicating that these phages were unable to subvert phage resistance leading to the inability to detect them by our methods.

| DISCUSSION
E. faecalis is a commensal and nosocomial pathogen and is becoming increasingly resistant to last-resort antibiotics (Turner, Lee, et al., 2021). In this study, we show the coevolution of E. faecalis SF28073 with two genetically distinct phages, VPE25 and phi47. Our results reveal that each phage selects for different E. faecalis mutations, indicating that genetically unique phages select different genetic compositions within the population. In particular, for both phages, E. faecalis developed missense mutations in genes encoding cell wall macromolecules, specifically PIP EF and Epa, that are required for successful infection by VPE25 and phi47, respectively (Chatterjee et al., 2019;Duerkop et al., 2016). VPE25 has recently been shown to depend on Epa (Chatterjee et al., 2020), most likely for adsorption to the cells. Because PIP EF and Epa are essential for successful phage infection, mutations in these genes prevent phage infection. Despite this, we observed no epa mutations in cultures challenged with VPE25, suggesting that pip EF mutations are dominant and potentially more advantageous than epa mutations, likely due to fitness costs (Chatterjee et al., 2019;Ho et al., 2018). SNPs in pip EF and epa indicate that phage receptor and coreceptor mutations are common to prevent phage infection. Additionally, many enterococcal phages rely on Epa for adsorption to cells (Al-Zubidi et al., 2019;Canfield et al., 2021;Chatterjee et al., 2019Chatterjee et al., , 2020Ho et al., 2018). In this study, we show that epa mutations are abundant when the bacteria are challenged with phi47. Mutations in epa genes also protect bacteria against VPE25 infection (Figure 3a). Therefore, if phi47 were to be used in a therapeutic cocktail with other Epa-dependent enterococcal phages, epa mutations could reduce the effectiveness of the cocktail. If the bacteria develop dominant mutations in alternative phage receptors, such as pip EF , Epa will remain intact. This could promote infection by Epa-dependent phages that utilize receptors other than pip EF (Figure 3b). Our data suggest that VPE25 and phi47 could be useful in combination as a therapeutic cocktail, yet there remains a need to identify Epa-independent phages for inclusion in future therapeutic phage cocktails.
Mutations in ccpA were found in cultures challenged with both VPE25 and phi47. CcpA, a transcriptional regulator, plays a central role in the catabolite control mechanism, regulating transcription in response to carbon source availability (Carvalho et al., 2011;Gao et al., 2013;Muscariello et al., 2001). Phage infection and development are determined by bacterial host factors such as nutritional state (Li et al., 2003). A recent study showed that carbon source is an important factor in phage production (Silva et al., 2021). A previous study using Bacillus subtilis showed that phage 29 infection downregulated CcpA-dependent genes. The authors propose that repression of genes involved in the utilization of specific carbon sources keeps B. subtilis in a metabolic state ideal for phage development (Li et al., 2003), suggesting a link between CcpA and optimal phage production. It is possible that nonsynonymous mutations in ccpA alter carbohydrate metabolism. Studies focused on transcription during bacteria and phage coevolution could reveal deeper insights into the (a) (b) F I G U R E 3 Resistance to phi47 protects bacteria from VPE25 infection. Serial dilutions of Day 7 bacterial communities were challenged with the opposite phage that the bacteria was coevolved with. (a) E. faecalis SF28073 that was coevolved with phi47 was challenged with VPE25.
(b) E. faecalis SF28073 that was coevolved with VPE25 was challenged with phi47. Multiple comparisons using a one-way ANOVA. **p ≤ 0.01, ****p ≤ 0.0001 impact of ccpA mutation. Studies in streptococci have established a role for CcpA in the control of virulence factors and carbohydrate metabolism (Carvalho et al., 2011). We cannot rule out the possibility that CcpA may be directly involved in the regulation of epa transcription or indirectly involved by regulating genes responsible for the production of Epa precursors. Either circumstance could influence the production of Epa thereby affecting phage infection.
A mutation in the gene manX, a component of the manXYZ operon, was found in one replicate infected with VPE25. The ManXYZ protein complex is a transporter for mannose as well as other molecules, such as N-acetylglucosamine (GlcNAc) (Plumbridge, 1998).
Interestingly, analysis of Epa from two E. faecalis strains revealed the presence of GlcNac (Guerardel et al., 2020). In addition, a coevolution study using phage ΦSA012 and its host, Staphylococcus aureus SA003, demonstrated that the phage receptor-binding protein binds to α-GlcNAc in wall teichoic acids (Takeuchi et al., 2016). and their phages (Peters et al., 2020;Takeuchi et al., 2016) and the model organism E. coli and its phages. For example, in one study E.
coli and phage T3 were coevolved in a chemostat, allowing for a controlled experimental environment. Under this condition, the authors observed common bacterial mutations at the gene level and phage mutations at the codon level across experimental replicates (Perry et al., 2015). We believe that our current study both supports and contradicts these findings.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.

ETHICS STATEMENT
None required. Note: NA indicates no change in amino acid. a Genes mutated across multiple cultures.