Phage Cocktails Constrain the Growth of Enterococcus.

Phages that infect pathogenic bacteria present a valuable resource for treating antibiotic-resistant infections. We isolated and developed a collection of 19 Enterococcus phages, including myoviruses, siphoviruses, and a podovirus, that can infect both Enterococcus faecalis and Enterococcus faecium. Several of the Myoviridae phages that we found in southern California wastewater were from the Brockvirinae subfamily (formerly Spounavirinae) and had a broad host range across both E. faecium and E. faecalis. By searching the NCBI Sequence Read Archive, we showed that these phages are prevalent globally in human and animal microbiomes. Enterococcus is a regular member of healthy human gut microbial communities; however, it is also an opportunistic pathogen responsible for an increasing number of antibiotic-resistant infections. We tested the ability of each phage to clear Enterococcus host cultures and delay the emergence of phage-resistant Enterococcus. We found that some phages were ineffective at clearing Enterococcus cultures individually but were effective when combined into cocktails. Quantitative PCR was used to track phage abundance in cocultures and revealed dynamics ranging from one dominant phage to an even distribution of phage growth. Genomic characterization showed that mutations in Enterococcus exopolysaccharide synthesis genes were consistently found in the presence of phage infection. This work will help to inform cocktail design for Enterococcus, which is an important target for phage therapy applications. IMPORTANCE Due to the rise in antibiotic resistance, Enterococcus infections are a major health crisis that requires the development of alternative therapies. Phage therapy offers an alternative to antibiotics and has shown promise in both in vitro and early clinical studies. Here, we established a collection of 19 Enterococcus phages and tested whether combining phages into cocktails could delay growth and the emergence of resistant mutants in comparison with individual phages. We showed that cocktails of two or three phages often prevented the growth of phage-resistant mutants, and we identified which phages were replicating the most in each cocktail. When resistant mutants emerged to single phages, they showed consistent accumulation of mutations in exopolysaccharide synthesis genes. These data serve to demonstrate that a cocktail approach can inform efforts to improve efficacy against Enterococcus isolates and reduce the emergence of resistance.

INTRODUCTION

Enterococcus species are regular members of vertebrate gut microbiomes and are pathogens of enormous clinical significance due to their high potential for antibiotic resistance (e.g., widespread vancomycin-resistant Enterococcus) (1). Antibiotic treatment often leads to a high abundance of Enterococcus spp. in the gut, as many other members of the gut microbiome are more sensitive to antibiotics (2–5). There are multiple species of Enterococcus that cause human disease, but the most common are E. faecium and E. faecalis. Both E. faecium and E. faecalis develop resistance to vancomycin, and they are known to cause disease, particularly in individuals who have undergone multiple rounds of antibiotics. The development of alternative therapies for Enterococcus infections is vital for providing effective treatment options for recalcitrant long-term infections, such as endocarditis.

Surprisingly, there are relatively few characterized Enterococcus phages. Bacteriophage (phage) therapy is an alternative treatment to antibiotics that has shown promise for treating Enterococcus in vitro and in animal models (6–8). However, there is a paucity of basic research into phage safety, mechanisms of action, and best practices of use. Because of the propensity for Enterococcus to develop antibiotic resistance, and the elevated abundance in critically ill or antibiotic-treated patients, Enterococcus spp. are a logical focus for phage therapy.

Although phages hold much promise for overcoming antibiotic resistance, bacteria can also evolve resistance to phage infection. Bacteria exist in a constant evolutionary battle with phages and thus have evolved many systems to resist phage infection, including preventing phage binding, restriction-modification systems, CRISPR-Cas9 immunity, and abortive infection (9, 10). Given strong selective pressure from a single phage, bacteria often quickly evolve resistance to that phage in laboratory settings (11). Despite the potential for evolved host immunity, unlike antibiotics, phages can coevolve to combat and circumvent bacterial resistance mechanisms (12). Changes that arise during coevolutionary exchanges can be tracked through experimental evolution, sequencing, and analysis of the consistently arising mutations.

Cocktails of phages are often more effective at treating infection than any one phage, because many phages have narrow host ranges. Multiple phages can be used in combination to increase the likelihood that several or all strains of the target bacteria will be killed (13, 14). Theoretically, phage cocktails also decrease the chance that a phage-resistant mutant will arise, as such a development would likely require the simultaneous evolution of multiple orthogonal resistance mechanisms. The genetic changes that lead to phage resistance can also provide a fitness disadvantage, for example, by making the bacteria more susceptible to infection by other phages (15). Similar to phage cocktails, combinations of antibiotics are used to treat tuberculosis infections, and combinations of antivirals are used to treat HIV (16, 17). The design principles for effective phage cocktails are an exciting frontier which may benefit from knowledge derived from experimental coevolution.

Phage therapy has shown promise in in vitro and in vivo mouse experiments (7, 8, 18). Additionally, Enterococcus phages have been used to disrupt Enterococcus biofilms, which are generally much harder to treat than planktonic cells, because antibiotics have trouble penetrating biofilms (8). Enterococcus phages have also been used to treat humans. Two phage cocktails sold by the Eliava Institute of Bacteriophages, Microbiology, and Virology in Georgia contain phages against Enterococcus spp. as well as phages against other common pathogens (19). In vitro, Enterococcus phage combinations have been shown to be more effective than single phages at preventing the growth of both antibiotic- and phage-resistant Enterococcus mutants (20). For example, Morisette et al. recently reported that, when used in combination with daptomycin, a two-phage cocktail exhibited a substantially improved capacity to eradicate E. faecium and prevent the emergence of phage resistance, while resistance did emerge with either phage by itself (21).While these examples are encouraging, they represent only limited examples of cocktail design. Thus, part of our goal in this work was to further compare the impact of different phage combinations on bacterial growth.

Our goals were the following: (i) characterize and test a diverse panel of phages for their ability to reduce Enterococcus growth; (ii) determine if and how host growth changes in the context of different phage combinations; and (iii) elucidate the underpinnings of phage resistance by identifying mutations in the Enterococcus host genome that are repeatedly found in the presence of phage infection.

RESULTS

Phage characterization and host range evaluation.

We isolated a collection of 18 Enterococcus-infecting phages from Southern California wastewater influent samples. Genome sequencing showed 8 Myoviridae, 10 Siphoviridae, and 1 Podoviridae phages. A list of the 18 newly isolated phages along with one already known Myoviridae phage, EfV12-phi1 (V12), is shown in Table S3 in the supplemental material, together with genome size and GenBank accession numbers. Plaque assays were used to test the phage susceptibilities of clinical isolates of E. faecium and E. faecalis, including both vancomycin-resistant Enterococcus (VRE) and vancomycin-susceptible Enterococcus (VSE). Eleven of 19 phages were able to lyse several strains of both E. faecalis and E. faecium (Fig. 1). Most phages from the Myoviridae family showed broad host ranges, especially Ben, Bop, and V12, which lysed almost all isolates.

FIG 1

Host ranges of Enterococcus phages as determined using spot assays (see Materials and Methods). Complete lysis is indicated by the orange boxes, and white boxes represent no lysis. Several vancomycin-sensitive (light gray boxes around strain names) and vancomycin-resistant (dark gray boxes around strain names) Enterococcus isolates were used for this study. GenBank accession numbers for the 18 additional phages discovered as part of this study along with the Myoviridae phage, EfV12-phi1 (V12) obtained from the Félix d'Hérelle Reference Center for Bacterial Viruses at the Université Laval are listed in Table S3 in the supplemental material.

Phage information. The following phages were isolated and included in these experiments, with the phage name, family, genus, genome size, and GenBank accession numbers for the sequencing data we deposited, with the exception of EfV12-phi1, which was already available and which we used in our initial coevolution study (Wandro et al., 2019 [35]). The naming convention for each phage begins with vB_OCPT for virus of bacteria and Orange County Phage Team (OCPT). Download Table S3, DOCX file, 0.02 MB.

Of the 10 Siphoviridae phages, 9 belonged to the genus Saphexavirus. Genome alignments showed that these phages share a core genome of 54 genes and approximately 43 accessory genes per phage (Fig. 2A and B). Their genomes were approximately 57 kb in length, and only 3 genes could be annotated (Fig. 2C).

FIG 2

Comparative genomics of major families Myoviridae (Kochiodavirus and Schiekvirus genera) and Siphoviridae (Saphexavirus genus). (A) Phylogenetic tree made from the core genome of Siphoviridae phages. (B) Number of genes in the Siphoviridae core genome and average number ± standard error of the mean (SEM) of accessory genes per phage genome. (C) Layout of genes in a representative Siphoviridae genome (phage SDS1), with known genes annotated. Colors indicate numbers of other phages that also contained each gene (50% BLASTp similarity). (D) Phylogenetic tree made from the core genome of Myoviridae phages. (E) Number of genes in the Myoviridae core genome and accessory genome per phage. (F) Layout of genes in a representative Myoviridae genome (phage phiV12) with known genes annotated. Colors indicate numbers of other phages that also contain each gene (50% BLASTp similarity).

Of the 8 Myoviridae phages, all belonged to the Brockvirinae subfamily, but these were divided into two genera, Kochikohdavirus and Shiekvirus (Fig. 2D). Their genomes were approximately 150 kb in length with a core genome containing ~66 genes shared among all phages in the subfamily and an accessory genome of approximately 133 genes per phage (Fig. 2E and F). The genomes were organized into two portions, shown by representative phage V12 in Fig. 2F The first portion contained shorter genes that were not always present in members of the subfamily, and the second portion contained longer genes that were more conserved among members of the subfamily. The second portion also contained the known structural genes, although most genes remained unannotated.

A major limitation for phage therapy is that phages against any specific pathogen are often not readily available. To gain a better understanding of how the phages we isolated are distributed in nature, and how common they are to isolate when hunting for candidate phages, we assessed the distribution of representative phages from the Brockvirinae subfamily (Myoviridae V12) and the Saphexavirus genus (Siphoviridae CCS3). We queried these genomes against 67,429 publicly available metagenomes in NCBI’s Sequence Read Archive (Table 1) (22). Metagenomes with positive hits were downloaded and aligned to representative Brockvirinae genomes to ensure most of each genome was covered. Brockvirinae phages were found to be globally distributed in human fecal metagenomes, including those sampled in the United States, Europe, the Middle East, and Asia. Saphexavirus phages were only observed in two metagenomes. Sequences aligning to Brockvirinae phages were also found in nonhuman fecal metagenomes from condors, pigs, and bats. Brockvirinae phages were also found to be highly abundant in two phage cocktails from the Eliava Institute designed to treat intestinal issues, the Intestiphage cocktail and the PYO phage cocktail. These phage cocktails contain many different phages targeting a wide range of bacterial hosts, including Enterococcus.

TABLE 1

Enterococcus phages from the Sequence Read Archive

Phage	SRA ID	Title	Location	Sample type
phiV12	SRP077952	INTESTI bacteriophage cocktail genome sequencing and assembly	Georgia	Phage cocktail
phiV12	PRJEB23244	PYO phage cocktail	Georgia	Phage cocktail
phiV12	ERP017091	Gut microbiome in Crohn’s disease and modulation by exclusive enteral nutrition	Guangdong, China	Human fecal
phiV12	ERP006678	Gut and oral microbiome dysbiosis in rheumatoid arthritis	Beijing, China	Human fecal
phiV12	SRP071229	Gymnogyps californianus microbiome raw sequence reads	USA (Los Alamos National Laboratory)	California condor fecal
phiV12	ERP006046	Virus_Discovery_for_Vietnam_Initiative_on_Zoonotic_Infections__VIZIONS_	Vietnam	Viral metagenome
phiV12	ERP001956	Diagnostic metagenomics: a culture-independent approach to the investigation of bacterial infections	Germany	Human fecal
phiV12	SRP051511	New York City MTA subway samples metagenome	USA (New York City)	Subway samples
phiV12	ERP012929	Towards personalized nutrition by prediction of glycemic responses	Israel	Human fecal
phiV12	SRP040146	Clostridium difficile FMT	USA (Massachusetts)	Human fecal
phiV12	SRP115494	Longitudinal multi’omics of the human microbiome in inflammatory bowel disease	USA (Massachusetts)	Human fecal
phiV12	SRP099123	Metagenomic analysis of gut microbiota in sows and piglets	Freie University of Berlin	Pig fecal
CCS3	SRR1438030	Metagenomic identification of novel enteric viruses in urban wild rats and genome characterization of a group A rotavirus	Berlin, Germany	Rat fecal
CCS3	ERR2737461	Virus discovery for Vietnam Initiative on Zoonotic Infections (VIZIONS)	Vietnam	Viral metagenome

Phage cocktails prevent Enterococcus growth longer than single phages.

Many bacterial hosts quickly develop resistance to infecting phages. Experimentally, this can be detected by resumption of bacterial growth after a period of little or no growth following phage infection. Thus, we measured bacterial growth over 72 h in liquid medium to observe whether phage cocktails could reduce bacterial growth for longer than single phages, which would suggest that cocktails are more likely to prevent resistance evolution. Four E. faecalis strains were chosen for testing the efficacy of phage cocktails, with strain Yi6-1 allowing for the most combinations because it was susceptible to 17 of the 18 phages in our collection (Fig. 3; see also Fig. S1 in the supplemental material). The multiplicities of infection (MOIs) of phage cocktails tested on E. faecalis strains, including 0.1, 0.01, and 0.001, did not consistently affect the resumption of bacterial growth (see Fig. S2). Therefore, we chose to infect strains at the highest MOI of 0.1. To ensure that the effect on host growth was not simply due to adding more phage, the amount of each phage was halved relative to the single-phage condition when two phages were used and reduced to one-third when three phages were used.

FIG 3

Phage cocktails cleared cultures and prevented growth of E. faecalis for 72 h. Data points indicate the final OD600 of replicate bacterial cultures after 72 h of incubation with phage. Boxplots represent medians and interquartile ranges. Combinations of one, two, or three phages were added to susceptible E. faecalis Yi6-1 (A) and E. faecalis V587 (B) cultures in exponential growth phase, and mixtures were incubated for 72 h. M, Myoviridae phage; S, Siphoviridae phage; P, Podoviridae phage.

Effect of phage cocktails on the growth of E. faecalis EF06 (A) and E. faecalis EF11 (B) for 72 h. Combinations of one, two, or three phages were added to susceptible cultures in exponential growth phase, and after 72 h bacterial growth was measured based on the OD600. M, Myoviridae phage; S, Siphoviridae phage; P, Podoviridae phage. Download FIG S1, TIF file, 0.3 MB.

Effect of MOI on bacterial growth in phage cocktails. Three different MOIs (0.1, 0.01, 0.001) were used to study the efficiency of single phages and phage cocktails against four different E. faecalis isolates over the 72-h incubation period. (A) Yi6-1; (B) EF06; (C) EF11; (D) V587. Download FIG S2, TIF file, 0.4 MB.

Individual phages and phage cocktails had varying success in preventing host growth over 72 h, with some phages and cocktails preventing growth and others failing to do so (Fig. 3). Successful two-phage cocktails were usually composed of multiple phage families (for instance, Myoviridae and Siphoviridae), but not all cocktails composed of two different families were successful at preventing growth. Several successful two- and three-phage cocktails against E. faecalis Yi6 included phages that were unable to prevent host growth alone, suggesting that synergy between the phages was responsible for limiting host growth. All combinations of two- and three-phage cocktails against E. faecalis strain V587 (VRE) were successful in preventing host growth. E. faecalis strain V587 was only susceptible to Myoviridae phages in our cocktails showing that synergy can occur within the same family of phages.

Phage growth dynamics by qPCR.

The dynamics of phage growth and phage-phage interactions in a cocktail were examined by quantitative PCR (qPCR), which provides an estimate of the relative abundance of each phage throughout infection. We formulated cocktails containing two or three phages that were ineffective alone but successful together in preventing bacterial growth over 72 h, and we chose E. faecalis strain Yi6-1 as the host. Phage DNA was extracted immediately after phage and host inoculation (0 h), as well as at 24, 48, and 72 h postinoculation. Changes in phage abundances over time in each of eight cocktails were compared with a phage-only control that contained no bacteria. E. faecalis strain Yi6-1 growth without any phage is shown for comparison, as well as results for each individual phage infecting E. faecalis strain Yi6-1 alone (Fig. 4A and B).

FIG 4

Phage abundances and E. faecalis Yi6-1 growth with phage cocktails. Phage was added at hour 0, and the cultures were grown for 72 h. Individual phage abundances were measured by qPCR and plotted as the replicate mean ± SEM of the log PFU per milliliter increase from hour 0. (A) Bacterial growth when E. faecalis Yi6-1 was grown alone (host only), with each phage separately (1 phage), or with phage cocktails (2 phages, 3 phages). The replicate mean ± SEM OD600 is shown. (B) Each phage was added separately to growing E. faecalis Yi6-1 cultures. The phage abundance is plotted in the top panel, and the bacterial growth curve is shown in the bottom panel. (C to J) Phage cocktails were added to growing E. faecalis Yi6-1 cultures. The abundance of each phage cocktail is plotted in the top panel, and the bacterial growth curve is plotted in the bottom panel.

When grown in culture without any phage, E. faecalis strain Yi6-1 grew to an optical density at 600 nm (OD600) of 0.7. When these individual phages were added (remembering that these phages were chosen because they failed to totally inhibit growth), growth was inhibited to an OD600 of ~0.15 on average (Fig. 4B). When used in two- or three-phage cocktails, the OD600 was reduced to nearly zero (Fig. 4A and C to J). Using qPCR to measure phage abundances, we determined which phages were contributing most to bacterial growth inhibition.

We monitored the concentration of each phage in two- and three-phage cocktails against E. faecalis strain Yi6-1 (Fig. 4C to J). In seven of eight of the cocktails, the abundances of the phages were similar, suggesting that all phages present were contributing to bacterial lysis at a similar level. The one exception was in two-phage cocktail 6, in which phage Ump showed nearly a 6-fold increase in abundance over phage Carl (Fig. 4H).

Phage infection selects for mutations in Enterococcus exopolysaccharide synthesis genes.

Host mutations that emerge across replicates after phage infection indicate genes with the potential to confer phage resistance. To identify these genes, we sequenced several strains of Enterococcus that could grow in the presence of phage infection. These included E. faecalis strains DP11, Yi6, B3286, and TX2137 as well as E. faecium strain TX1330. Target strains were also grown in identical conditions in the absence of phage to ensure that observed mutations were specific to evolution with phages. Sequencing reads were compared to the original strain genome to determine if any genes had acquired mutations. All observed mutations are summarized in Table 2, with the position of mutations within the Epa locus shown in Fig. S3 in the supplemental material. Enterococcus strains consistently acquired point mutations in genes involved in exopolysaccharide synthesis. The four E. faecalis strains acquired mutations in the Epa exopolysaccharide synthesis locus, while E. faecium TX1330 acquired mutations in the Yqw exopolysaccharide synthesis locus. In the host-only controls, mutations in the Epa locus were not observed; however, we did observe Yqw mutants, indicating that these mutations may arise without the addition of lytic phage.

TABLE 2

Mutations observed in Enterococcus genomes after culturing with individual phages

Enterococcus strain	Phage(s)	Mutated host gene	Locus ID	Mutationtype	Exopolysaccharidelocus	V587 locus tag (EPA only)	AA change
E. faecalis DP11	Ump, SDS1	NAD-dependent epimerase/dehydratase	HOCGOLEH_00595	SNP	Epa gene	EF_2165	G11V, G279E
E. faecalis DP11	Ump	Bacterial sugar transferase	HOCGOLEH_00585	Nonsense	EpaR	EF_2177	E249*
E. faecalis DP11	SDS2	TagF gene. glycerol glycerophosphotransferase	HOCGOLEH_00593	SNP	Epa gene		A113E
E. faecalis Yi6	Bop	Bacterial sugar transferase	UMS_01916	SNP, nonsense	EpaR	EF_2177	N705Y
E. faecalis Yi6	Bop	UTP-glucose-1-phosphate uridylyltransferase	UMS_01646	DEL			K22*
E. faecalis Yi6	Bop	UDP-N-acetylglucosamine 1-carboxyvinyltransferase 1	UMS_01161	SNP			A210T
E. faecalis Yi6	Bop	Isoleucyl-tRNA synthetase	UMS_00984	DEL			W31*
E. faecalis Yi6	phiV12	30S ribosomal protein S7	UMS_00243	SNP			G82D
E. faecalis Yi6	phiV12	DNA-directed RNA polymerase subunit alpha	UMS_00275	SNP			G29V
E. faecalis B3286	Bop, phiV12	NAD-dependent epimerase/dehydratase	SQ1_02166	SNP, nonsense	Epa gene	EF_2165	A123E, Q204*
E. faecalis B3286	Bop	Glucose-1-phosphate thymidylyltransferase;	SQ1_02191	SNP	EpaE	EF_2194	R217C
E. faecalis B3286	Bop	Endonuclease III	SQ1_01204	SNP			A130V
E. faecalis B3286	phiV12	Bacterial sugar transferase	SQ1_02177	SNP	EpaR	EF_2177	Y336*
E. faecalis B3286	phiV12	ATP-dependent Clp protease ATP-binding subunit ClpE	SQ1_00765	SNP			I13F
E. faecalis TX2137	Bop	Epimerase/dehydratase	HMPREF9494_02361	SNP	EpaW	EF_2171	D229Y, G192V
E. faecalis TX2137	Bop	Phosphocarrier protein HPr	HMPREF9494_02513	SNP			V209L
E. faecalis TX2137	Bop, phiV12	Bacterial sugar transferase	HMPREF9494_02367	SNP	EpaR	EF_2177	R400C, G251R
E. faecalis TX2137	phiV12	DNA ligase (NAD+)	HMPREF9494_02501	SNP			A434E
E. faecium TX1330	Ben	DNA gyrase subunit A (EC 5.99.1.3)	HMPREF0352_0587	SNP			D116A
E. faecium TX1330	Ben	DNA-directed RNA polymerase beta' subunit (EC 2.7.7.6)	HMPREF0352_2730	SNP			A924V
E. faecium TX1330	Bob	Response regulator transcription factor	D3Y30_RS07660	SNP			E192K
E. faecium TX1330	Bop	Polysaccharide biosynthesis protein	D3Y30_RS11110	SNP			L222S
E. faecium TX1330	Bop, Ben, Bill, no phage	CpsD/CapB family tyrosine-protein kinase	D3Y30_RS11120	SNP	Yqw		P136S, P26Q, L62E
E. faecium TX1330	Carl, phiV12, no phage	Tyrosine protein kinase	D3Y30_RS11125	SNP	Yqw		A147T, V153A

Each row indicates an Enterococcus gene in which one or more mutations were observed when that Enterococcus strain was cultured with the indicated phage. When more than one phage is listed the mutation occurred separately for each phage host pair and the phages were not used in combinations, the phages were not used in combinations. Locus ids refer to the genomes listed in Table S2 of the supplemental material. SNP, single-nucleotide polymorphism; DEL, deletion.

Mutations arise in E. faecalis Epa locus genes following phage infection. The genes comprising the Epa locus of E. faecalis are shown. Black, green, and red ticks represent the locations of nonsynonymous mutations observed in E. faecalis B3286, TX2137, and Yi6, respectively, as they coevolved with Brockvirinae phages. Detailed information about these mutations can be found in Table 2. Download FIG S3, TIF file, 0.2 MB.

List of bacterial strains used in this study and the antibiotics to which they are resistant. Download Table S2, DOCX file, 0.02 MB.

DISCUSSION

We set out to discover phages that can infect Enterococcus spp., characterize their host ranges and genomes, evaluate their interactions as single phages and in cocktails, and use experimental coevolution to define where selection pressures lead to accumulation of functionally equivalent mutations. Several of the myoviruses we found in Southern California wastewater were from the Brockvirinae subfamily (formerly Spounavirinae), and were able to infect both E. faecium and E. faecalis. To better understand the biology of these phages in the context of phage therapy, we examined their prevalence in public sequence data. We evaluated fecal metagenomes and found that Brockvirinae phages are globally distributed in animal and human microbiomes. We also showed that phage infection dynamics change in the presence of multiple phages and that cocktails of phages restrict bacterial growth for at least 72 h.

While using phage cocktails that include a diversity of phages targeting multiple disparate bacterial types has been a common practice in phage therapy, there is no standard for how many phages should be used, and the answer will likely change for different bacterial hosts. Some phage cocktails are effective at killing bacteria and preventing growth of resistant mutants, but the most effective phage combination is unknown for the vast majority of pathogenic bacteria. For example, the pyophage (PYO) cocktail from the Georgian Eliava Institute of Bacteriophages contains approximately 30 different phages targeting multiple bacterial hosts (19). Recent uses of phage therapy designed to target a single strain of a bacterium have included between one and six phages (23–25). Often, there is no obvious rationale behind the number of phages chosen for phage therapy, although lack of access to phages against the specific host of interest is often a limitation. Here, we show that combinations of two phages are effective at inhibiting growth of several Enterococcus strains over 72 h. Theoretically, using more phages in a cocktail would increase the chances of choosing phages that displayed synergy in reducing the host growth. However, increasing the number of phages could also lead to antagonistic interactions between phages (26). Antagonistic interactions may arise from competition for finite host resources, competition for host receptors, or the production of phage repressors.

Some phages in our collection were effective against E. faecalis and E. faecium, including both VRE and VSE isolates, suggesting that they would be good candidates for phage cocktails against a broad range of Enterococcus spp. If the phages in a cocktail target different bacterial proteins as binding sites, the cocktail will be more effective at lysing a pathogen even if a mutation in a single protein arises (27). As shown in Fig. 3A, combinations of phages that individually cannot clear a culture of E. faecalis Yi6-1 are able to do so when used in combinations. This shows that the evolutionary advantage of phage-resistant mutants could be diminished with the use of well-designed cocktails (28, 29).

Comparing the abundances of each phage over time in the eight phage cocktails shown in Fig. 4 yields a wide range of outcomes, from one phage greatly outpacing the other (cocktail 6), to relatively even abundances (cocktails 1, 7, and 8). Each of the five phages was ineffective at clearing Enterococcus cultures alone, yet all cocktails, with their varied phage dynamics, resulted in clearing of the Enterococcus cultures. This shows that there is more than one path to an effective phage cocktail. Even if one phage appears to dominate, the other phage may still be necessary to prevent the emergence of a phage-resistant mutant.

When under selective pressure from Brockvirinae phages, Enterococcus produce mutations primarily in exopolysaccharide synthesis genes, suggesting that phage resistance may evolve by preventing phage recognition and initial binding. E. faecium and E. faecalis both contain the highly conserved Epa capsule synthesis locus, in which genetic variation has been observed consistently for E. faecalis strains (30, 31, 32). Mutations in the Epa locus have been observed previously during coevolution with Brockvirinae phages; these mutations impaired Enterococcus host colonization and increased antibiotic sensitivity (7, 33). Consistent with the results presented herein, Canfield et al. observed mutations to the E. faecium epaR and epaX genes during bacteriophage infection (34, 35). In addition to the Epa locus, E. faecium encodes the Yqw exopolysaccharide synthesis locus, which is not present in E. faecalis. Mutations were observed in the Yqw locus of E. faecium TX1330, but not in the Epa locus. Reproducible mutations arising across replicates within genes of the Yqw locus were previously observed during coevolution between phage and Enterococcus (36). However, in the current experiments, mutations in Yqw locus genes occurred in both TX1330 host control cultures (which lacked phage) and cultures containing phage. Therefore, we cannot attribute these mutations to phage evolutionary pressure.

The need for alternative therapies for antibiotic-resistant bacterial infections continues to grow. Bacteriophages present an alternative therapeutic route with the potential to replace or supplement antibiotics, but there is a significant knowledge gap that reduces the utility of phages. Challenges facing phage therapy include the emergence of host resistance, limited information on how phages may collectively eradicate their hosts, and a lack of guidance on the number of phages necessary for the formulation of successful phage cocktails. Enterococcus is a good model bacterium because it is a diverse genus containing multiple species capable of causing debilitating human infections, and it has the ability to acquire significant antibiotic resistance. Using this model, we have demonstrated that simple two-phage cocktails have significant potential to kill their hosts and reduce the emergence of resistant isolates. For Enterococcus, our data suggest that host killing was not substantially increased by adding more than two phages to cocktails, but a larger study is needed to confirm these findings. Rationally designing phage cocktails with knowledge of phage-host interactions from experimental evolution has the potential to significantly advance phages as antibiotic alternatives for the treatment of human pathogens such as Enterococcus.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The bacterial strains used in this study are listed in Table S2 of the supplemental material. E. faecalis and E. faecium strains were added to our strain collection from the patient population at UC San Diego Health; all personally identifying information was removed. Each of these isolates was identified using a biotyper instrument based on matrix-assisted laser desorption ionization–time of flight spectroscopy technology (Bruker, Billerica, MA, USA), and antibiotic susceptibilities were assessed using broth microdilution techniques on the BD Phoenix instrument (Becton, Dickinson, Franklin Lakes, NJ, USA). Four of the strains were obtained from the Human Microbiome Project repository at the Biodefense and Emerging Infections Research Resources Repository (BEI: www.beiresources.org). E. faecalis and E. faecium strains were cultured in brain heart infusion (BHI) broth. All strains were grown at 37°C in liquid medium overnight with shaking. Solid medium was prepared with 1.5% agar when culturing bacteria or 1.0% bottom agar and 0.3% top agar for plaque assays.

Phage isolation propagation and storage.

Phages were isolated from sewage using three rounds of plaque assays. Raw sewage influent was collected from wastewater treatment plants in Orange County, Redwood Shores, and Escondido, California. Sewage was stored at 4°C and used for phage isolation for several months. Sewage was centrifuged for 10 min at 10,000 × g to remove particulates, and the supernatant was used in plaque assays with various strains of Enterococcus. A 100-μL aliquot of sewage supernatant was added to 100 μL exponentially growing Enterococcus in BHI medium and incubated at 37°C for 15 min. Five milliliters of warm BHI containing 0.3% UltraPure low-melting-point agarose (ThermoFisher catalog number 16520050) was added, and the mixture was poured on a BHI agar plate for overnight incubation at 37°C. The next day, plates were examined for plaques, and any plaques were picked with a pipette tip and suspended in 50 μL SM buffer (37). Picked plaques underwent two more rounds of plaque assays in the same manner to ensure purity of the phage isolate. Pure phages were propagated by performing a plaque assay to create a plate displaying webbed lysis that was then flooded with 3 mL of SM buffer and incubated for 1 h. The SM buffer was then collected and centrifuged at 10,000 × g for 10 min. For long-term storage, phages were stored at −80°C in SM buffer containing 25% glycerol.

Genomic sequencing.

DNA was extracted from Enterococcus and phage using a Quick-DNA Microprep kit (Zymo catalog number D3020). Before Enterococcus DNA extraction, lysozyme was added to lysis buffer at a concentration of 100 μg/mL and incubated at 37°C for 30 min. For DNA extraction from coevolution cultures containing both bacteria and phage, the extractions were performed without lysozyme. Libraries were prepared using scaled-down reactions with the Illumina Nextera enzyme (24). Paired-end sequencing with a 75-bp read length was performed on the Illumina NextSeq using the Mid Output v2 reagents. Approximately 2.5 million reads were obtained for each sample.

Genomic characterization.

Phage and Enterococcus genomes were assembled de novo using the SPAdes assembler (38). Genomes were annotated using Prokka with the genus set to Caudovirales (39). Core genomes were determined and aligned using Roary (40). Phylogenetic trees were constructed from core genomes using FastTree (41). Visualizations were made using the Python matplotlib, dna_features_viewer, and Biopython (42).

Searching the Sequence Read Archive.

All metagenomes in the SRA were searched for Brockvirinae Enterococcus phages using the “searching SRA” tool with V12 and CCS3 as representative phages for the myovirus and siphovirus families, respectively (22). Briefly, the searching SRA tool searches for the query sequence in all 111,156 metagenomes currently on the SRA by subsampling 100,000 sequences from each metagenome. From the metagenome hit list, we selected only metagenomes where the average read length matching our query was over 50 bp. Alignments were manually inspected using Geneious to ensure the majority of the genome was covered (43). Information about the SRA projects with verified hits is summarized in Table 1.

Host range determination.

Phage susceptibilities were measured in several bacterial strains with established multidrug resistance serious enough to cause illness (Fig. 1). Phage susceptibility was determined using a spot assay in which 5 μL of each phage lysate was spotted on a lawn of an Enterococcus strain on a 1.5% agar plate infused with BHI. The spots were allowed to dry at room temperature for 30 min and incubated at 37°C for 24 h. The next day, plates were examined to identify the host’s susceptibility based on the size and shape of the cell lysis zones.

Although spot assays may show lysis from without due to too many phages infecting the cell simultaneously, this is not a widespread phenomenon (44). Furthermore, our MOI experiments, and also our observation of concentration-dependent host lysis with different phage titers in liquid and on plates (see Fig. S2 in the supplemental material), indicated that lysis from without was not occurring in the spot assays. Nevertheless, we cannot completely exclude the possibility that lysis from without contributed to some of the spot assay results.

Determination of MOI.

MOI is defined as the ratio of the number of phage particles to the number of target cells of each host. We tested three MOIs (0.1, 0.01, and 0.001) as follows. A single colony was picked from a streak plate and was grown overnight in BHI broth at 37°C. The next day, the bacterial culture was diluted to an OD600 of 0.05 in fresh BHI. A serial dilution series from 10−1 to 10−8 of each bacterial culture was performed using phosphate-buffered saline. For each isolate, an aliquot from serial dilutions of 10−4 and 10−5 was plated on 1.5% BHI agar plates and incubated overnight at 37°C. The next day, the number of colonies was counted on each dilution plate to determine the number of CFU per milliliter (see Table S1 in the supplemental material). Host susceptibility to each phage was determined via plaque assays. Cells from the log phase of growth were infected with different phages at different dilutions. Based on plaque assay plates, we determined the titer of each phage (in PFU per milliliter) (see Table S1). Based on observed CFU of Enterococcus isolates and the titer of each phage, the appropriate volume of isolates and phages were determined at three different MOIs, 0.1, 0.01, and 0.001, to conduct growth curve experiments.

Phage titers and the concentration of the host. The phage titers (PFU per milliliter) were determined in plaque assays. Viable counts of the hosts were measured by counting colonies formed from liquid cultures at an OD600 of 0.05. All phage titers were determined on their corresponding susceptible E. faecalis hosts: Yi6-1 (A), EF06 (B), EF11 (C), and V587 (D). Download Table S1, DOCX file, 0.4 MB.

Phage cocktails.

Cocktails consisting of one, two, or three phages were tested against E. faecalis Yi6-1, EF06, EF11, and V587. A single colony was inoculated into BHI broth and grown overnight in a shaker at 37°C. The next day, bacterial cultures were diluted to an OD600 of 0.05 in fresh BHI. Based on the results of three MOIs (0.1, 0.01, and 0.001) (see Fig. S2 in the supplemental material), appropriate volumes of Enterococcus and phages were determined and combined in a single well inside a 96-well plate along with enough BHI to make up a total volume of 200 μL. For two-phage cocktails, one-half of the previously determined volumes of each of the two unique phage stocks were added. To evaluate three-phage cocktails, one-third of previously determined volumes of each of the three unique phage stocks was added. To avoid desiccation in wells at the perimeter of the 96-well plates, 200 μL of fresh BHI medium was added to all perimeter wells.

PCR.

Phage cocktails were grown in BHI for 72 h with their respective hosts (Yi6-1, EF06, and EF11) in 96-well plates with three biological replicates. The positive control consisted of phages without a host. The negative controls consisted of the relevant hosts grown in BHI for 72 h without phage. The samples were collected at four time points (0 h, 24 h, 48 h, and 72 h), and total genomic DNA was isolated using the DNeasy blood and tissue kit (Qiagen). For the unknown standards, we performed a plaque assay for all the phages to find their concentrations (in PFU per milliliter), and the genomic DNA of the standards was isolated using the DNeasy blood and tissue kit. Standard curves were generated with serial dilutions of phage (10−1 to 10−8). Phage-specific primers were designed using Geneious software and are listed in Table S4 of the supplemental material. For designing the primers, all the phages were aligned using the multiple alignment tool in Geneious software, and unique regions within each phage were selected. The qPCR experiment was performed in a 96-well PCR plate using an Eppendorf Mastercycler RealPlex with SYBR green PCR master mix (Eppendorf) as per the universal SYBR green qPCR protocol, where fluorescent product is detected during the last step of each cycle. The obtained melting curve data were analyzed using Eppendorf Mastercycler RealPlex to calculate the Cycle threshold (Ct) values. Ct values of the standards were then used to generate standard curves correlating the log PFU per milliliter to Ct values; this information was used to estimate the concentration of each phage in the cocktails. Changes in PFU per milliliter were plotted over time for each phage cocktail.

DNA primers (5′ → 3′) used to conduct qPCR experiments. Download Table S4, DOCX file, 0.01 MB.

Generation of Enterococcus mutants in liquid cultures.

To determine if apparent phage resistance was associated with genetic mutations, we grew five Enterococcus spp. strains in the presence of different phages and sequenced those that displayed a resistance phenotype. Both E. faecalis (B3286, TX2137, Yi6, and DP11) and E. faecium (TX1330) hosts were grown individually (only when the host was susceptible by spot assay) with Ben, Bop, Bill, Bob, Carl, EfV12-phi1, SDS2, and Ump. Enterococcus sp. strains were grown overnight in BHI, diluted to an OD600 of 0.05, aliquoted into 96-well plates with 10 μL of a highly concentrated individual phage stock (total volume of 100 μL), and incubated at 37°C. Every 24 h for 28 days, 10 μL of culture was diluted into 190 μL fresh BHI media. Wells that showed a resurgence of growth, potentially indicating evolved resistance, were frozen at −80°C in 50% glycerol for DNA extraction and sequencing.

Enterococcus mutant sequencing.

DNA was extracted from Enterococcus cultures that appeared to be resistant to phages by using the Quick-DNA Microprep kit (Zymo catalog number D3020). Before Enterococcus DNA extraction, lysozyme was added to lysis buffer at a concentration of 100 μg/mL and incubated at 37°C for 30 min. Libraries were prepared using a scaled-down protocol with the Illumina Nextera enzyme (45). Short-read-length (75 bp) paired-end sequencing was performed on the Illumina NextSeq using the Mid Output v2 reagents. Approximately 1 million reads were obtained per sample, resulting in about 10-fold coverage across the Enterococcus genome. All sequencing experiments were performed using cultured populations of bacteria, as opposed to individual colony-purified strains; therefore, each culture likely contained DNA from multiple strain variations.

Sequencing analysis.

SPAdes was used for genome assembly for phages and bacteria when reference genomes did not already exist (38). For calling bacterial mutations, DNA sequencing reads from each phage-resistant host population were aligned to their wild-type genome using Breseq, which performs short-read alignment to a reference and calls mutations (46). In cases where an existing assembly was used, preexisting mutations at the initial time point were subtracted. All mutations are reported in Table 2. To relate mutations in exopolysaccharide synthesis genes in the Epa locus among strains, genes were mapped to the E. faecalis V583 genome (GenBank GCF_000007785.1), in which the Epa locus has been well-characterized (47).

Data availability.

Data from bacterial growth assays, phage qPCR, and code for analysis and making figures are available at https://github.com/swandro/phage_cocktails. Genomes for bacterial and phage strains used in this study have been deposited with GenBank, and the available accession numbers for all phage and some bacteria can be found in Tables S2 and S3 in the supplemental material.