Isolation and Optimization of Phages Infecting Members of the Streptococcus bovis/Streptococcus equinus Complex

Abstract

Background: Cattle production is a cornerstone of U.S. agriculture but faces increasing pressure to balance profitability with environmental sustainability. Optimizing the ruminal microbiome to enhance feed efficiency could help address both challenges. Members of the Streptococcus bovis/Streptococcus equinus complex (SBSEC) are key contributors to ruminal acidosis and related digestive disorders due to their role in carbohydrate fermentation and lactic acid production. Bacteriophages targeting this bacterial group present a promising approach to mitigate this problem with high precision and without promoting the spread of antibiotic resistance. Methods: A collection of SBSEC-targeting bacteriophages were isolated from cattle rumen fluid and feces and further characterized. Characterization included host-range evaluation, whole genome sequencing, and growth inhibition assessment via optical density measurements. Selected bacteriophages underwent training to enhance infectivity. Results: Eleven lytic and one lysogenic phage were isolated. Several phages demonstrated sustained bacterial growth suppression, showing efficacy against SBSEC bacteria from diverse sources despite narrow host ranges. Co-evolutionary training was done in a subset of phages to improve bacteriolytic activity but had an inconsistent effect on the ability of phages to inhibit the growth of their naïve host. Genomic sequencing and phylogenetic analysis revealed uniqueness and clustering into three distinct groups that matched phenotypic characteristics. Conclusions: This study demonstrates the potential of bacteriophages as precise biological control agents, with successful isolation and enhancement of phages targeting SBSEC bacteria. Eleven lytic genome-sequenced phages show promise for development as cattle feed additives, though further research is needed to optimize their application in agricultural settings.

1. Introduction

Cattle production is a cornerstone of US agricultural activity, representing 88.4 billion in total cash receipts for agricultural commodities in 2023 [1]. However, the industry faces dual growing pressures: the need for increased production to meet rising demand while improving long-term sustainability and reducing impacts to the environment [2,3,4,5]. Beef production faces intense scrutiny due to the resource intensiveness of cattle production practices [6,7] thus technologies that can efficiently increase beef production while maintaining or improving economic and environmental sustainability are critically needed.

Enhancing feed efficiency in cattle is widely regarded as a crucial strategy for improving the overall sustainability of the beef industry [8,9,10]. It has been estimated that a 10% increase in feed efficiency can lead to a substantial 43% increase in profits. Accordingly, there has been substantial research effort into improving cattle feed efficiency [11,12]. Though the factors associated with variation in feed efficiency are still far from being completely understood, many contributing processes have been identified, including feeding behavior and physical activity, digestibility, energy metabolism, protein turnover, body composition, the endocrine system, and lastly, rumen microbiota [13,14,15,16,17,18].

Rumen microbiota both influences and is influenced by diet, shaping cattle feed efficiency [19,20,21]. Despite significant advancements in identifying specific rumen microorganisms and understanding their functions, our knowledge of rumen microbiology and the ecological networks required to degrade diverse substrates remains incomplete [22,23]. Therefore, new technologies are urgently needed to enable microbiome engineering and capitalize on recent discoveries.

Top-down microbiome engineering strategies, widely used in the cattle industry, involve modifying pre-existing microbial communities. These include dietary adjustments, antibiotics, prebiotics, essential oils, rumen transfaunation, probiotics (direct-fed microbials), and vaccines, all aimed at improving animal performance through microbiome modulation [24,25,26]. Generally, these strategies lack specificity, yield inconsistent results or produce temporary improvements that are challenging to maintain [27,28,29,30,31,32]. While the need for precise microbiome manipulation tools is widely acknowledged, few cost-effective solutions have been developed for animal agriculture [22]. Bacteriophages (phages), viruses that selectively and exclusively infect and lyse bacteria, offer a highly specific, self-amplifying, and environmentally friendly alternative to traditional antimicrobials [33]. Unlike broad-spectrum antibiotics, phages precisely target bacterial hosts without disrupting beneficial microbial communities, making them an attractive tool for microbiome modulation. First described by Félix d’Hérelle in the early 20th century [34], phage research has undergone a renaissance in recent years, with renewed interest in their applications across medicine, agriculture and biotechnology [35].

As such, we have isolated and characterized phages infecting several problematic ruminal bacterial species [36,37], including phages targeting a group of facultatively anaerobic gastrointestinal streptococci, the Streptococcus bovis/Streptococcus equinus complex (SBSEC). Streptococcus is a genus of diverse spherical Gram-positive lactic acid bacteria, containing commensal species native to the microbial communities of humans and animals but also pathogenic species responsible for numerous diseases. In ruminants, SBSEC have been observed to outgrow other ruminal bacterial flora under optimal growth conditions, producing a large amount of lactate and capsular polysaccharide causing acute ruminal acidosis and bloat [38,39]. Feedlot cattle are commonly fed ionophores (such as monensin and other antibiotics) to increase feed efficiency [40], stop SBSEC overgrowth and prevent the subsequent drop in ruminal pH [41]. However, this practice raises serious industrial and public health concerns highlighting the need for other methods of microbial control and ruminal community modulation. Given the importance of SBSEC, the first suggestion of phage therapy for microbial control in the rumen was for “S. bovis” [42] and substantial progress was made previously to isolate and characterize SBSEC phages and ruminal phages [42,43,44,45,46]. However, many lysogenic phages infecting SBSEC were identified, and only a few were genetically characterized. Recently, whole genome sequences of lytic ruminal phages have been reported and new SBSEC phages have been isolated [47,48,49]. Further experimental in vitro and in vivo studies with viral isolates are expected to remain crucial for the elucidation of biological properties required for the development of effective phage therapies and answer outstanding questions in rumen virus research [50].

In this publication, we describe 12 phages isolated from ruminal contents and cattle feces infecting SBSEC isolates and evaluate their suitability as antimicrobial agents, possibly for future use as feed additives to reduce or prevent SBSEC mediated acidosis.

2. Materials and Methods

2.1. Sampling and Storage of Ruminal Contents and Cattle Fecal Matter

Over 100 rumen fluid samples were collected from a beef processing plant in Amarillo, Texas, the Beef Cattle Research Center (Kansas State University) and the USDA Livestock Issues Research Unit (Lubbock, TX, USA) over the course of several years. Some rumen fluid samples were from cattle exposed to acidotic diets [51], while others were sourced from tylosin-fed feedlot cattle, and from cannulated cattle not fed tylosin. Samples shipped overnight on ice were processed within 24 h and stored at −80 °C in 10% dimethyl sulfoxide. Fresh fecal samples were sourced from grazing land in central Texas and stored at 4 °C or frozen at −80 °C in 10% dimethyl sulfoxide until use.

2.2. Bacterial Host and Bacteriophage Isolation

Samples were either streaked directly or enriched for bacterial isolation. Enrichments were performed to improve the isolation rate of both SBSEC strains and bacteriophages. Briefly, 200 µL of bovine rumen fluid or stool were added to a minimal medium consisting of 20 g/L pancreatic digest of casein, 10 g/L yeast extract, 0.1 mg/L resazurin, 1 g/L L-cysteine, 4 mg/L CaCl₂, 8 mg/L MgSO₄, 20 mg/L KH₂PO₄, 20 mg/L K₂HPO₄, 40 mg/L NaCl, 160 mg/L NaHCO₃. The medium was sparged with an anaerobic gas mix consisting of 10% CO₂, 5% H₂ and balancing N₂ and was supplemented with 1% maltose or 1% starch [52,53,54]. After overnight incubation at 37 °C, bacterial strains were isolated by streaking enrichments on Modified Edwards Medium (Oxoid) in anaerobic conditions. The species of SBSEC bacteria was confirmed by sequencing of their 16S rRNA gene [55]. After isolation, all SBSEC isolates were grown and maintained in brain heart infusion (BHI) broth in aerobic conditions. After enrichment and SBSEC isolation, all subsequent work was done under aerobic conditions. A list of the bacterial isolates used in this study can be found in the Supplementary Information (Table S1). For bacteriophage isolation, lysates from the enrichment culture were produced by centrifugation (10,000× g for 10 min) and filter sterilization using 25-mm diameter, 0.22-µm pore size PVDF hydrophilic syringe filter (Choice™, Thermo Scientific™, Waltham, MA, USA). The resulting lysates were spotted on lawns of the bacterial hosts in double layer assays (DLA), further detailed below [56].

2.3. Double Layer Assay and Bacteriophage Propagation

DLA were prepared with Brain Heart Infusion (BHI) (RPI Corp., Mt. Prospect, IL, USA) agar that was overlaid with BHI top agar (0.4% agarose) supplemented with 5% glycerol, 1 mM MgCl₂, 1 mM CaCl₂ and 5% of host bacteria and incubated aerobically. Clear plaques or zones of clearance were serially diluted and spotted again until single plaques could be picked using a sterile 1-mL micropipette tip. Agar plugs containing single plaques were resuspended in SM buffer (50 mM Tris-HCl (pH 7.5), 0.1 M NaCl, 8 mM MgSO₄, 0.01% w/v gelatin) and amplified in the corresponding host. Phages were amplified using DLA or using liquid cultures infected with phages under aerobic conditions. Phage titers were assayed via serial dilution and spotting on DLA. Clarified phage stocks were concentrated by the addition of 1:2 volume of sterile polyethylene glycol (PEG) 3× solution (30% PEG 8000, 3 M NaCl) and centrifugation at 12,000× g for 30 min after overnight incubation at 4 °C. The resulting pellet was suspended in SM buffer. Phage stocks were stored at 4 °C under aerobic conditions.

The host range of the isolated phages was evaluated using a library of SBSEC bacterial isolates. Phages stocks at high titer (1 to 5 × 10⁷ PFU/mL) were spotted on DLA lawns of SBSEC isolates and their ability to cause lysis was recorded.

2.4. Phage Genetic and Phenotypic Characterization

2.4.1. Whole Genome Sequencing

High molecular weight DNA of each bacteriophage was extracted using a traditional phenol chloroform DNA extraction method [57] with the following differences: after PEG precipitation phages were resuspended in SMB, phenol chloroform extraction was done 3 times, and nucleic acids were resuspended in water instead of TE buffer in the final step. The DNA quantity and quality were assessed using Take 3 microvolume plate and BioTek Synergy H1 microplate spectrophotometer (Agilent Technologies, Inc., Santa Clara, CA, USA). DNA concentrations were also determined using the fluorescent dsDNA quantification kit Broad Range BioDynami (BioDynami, Huntsville, AL, USA), and molecular weight of the samples was verified by electrophoresis of 200 ng DNA on a 1% agarose gel using a Quick-Load^® Purple 1 kb Plus DNA Ladder (New England Biolabs, Ipswich, MA, USA). If required, size selection was done with magnetic beads using 0.5 or 0.45 volumes of beads, according to the manufacturer’s instruction (Omega Bio-Tek, Inc., Norcross, GA, USA).

Libraries for genome sequencing were prepared from high molecular weight DNA using a Rapid barcoding sequencing kit 24 V14 (SQK-RBK114.24, Oxford Nanopore Technologies, Oxford, UK) and sequenced on a MinION sequencer (Oxford Nanopore Technologies, Oxford, UK) using a Flongle adaptor and a Flongle flow cell (R.10.4.1 chemistry) according to the manufacturer’s instructions. Reads were base called (super-accurate basecalling) and demultiplexed using MinKNOW and filtered by size with SeqKit v 2.9.0 [58,59]. Assemblies were generated using Flye v 2.9.5 [60] and also compared to assemblies generated by Unicycler v 0.5.0 [61,62] and Canu v 2.2 [63,64] to ensure the same assembly was produced by at least two methods. A subset of phage genome sequencing was performed by Plasmidsaurus using the same technique: Oxford Nanopore Technology for linear PCR/sequencing (25 to 125 kbp) with custom assembly (Plasmidsaurus Inc., South San Francisco, CA, USA).

2.4.2. Genome Annotation and Phylogenetic Analysis

Assembled viral genomes were annotated using Pharokka v1.4.0 [65] and manually refined using BLASTx (v 2.16.0) [66]. BLASTx was done using the NCBI web browser BLAST services, using translated nucleotide queries against the standard non-redundant protein sequences (nr) database and the default threshold of 0.05. Genomes were submitted to NCBI GenBank under the following accession numbers: PY1, PQ621712; PY2, PQ621713; PY3, PQ621714; PY4, PQ621715; PY7, PQ621716; PY9, PQ621717; 36L, PQ621707; Bsingle, PQ621708; Pika, PQ621709; PYS40, PQ621710; CSJC, PQ621704; Taco, PQ621705; Mushu, PQ621706; and Vroast, PV146553. Genomes and associated annotations were visualized with Proksee [67]. Alignments of similarity of the genomes were generated using clinker in the online CompArative GEne Cluster Analysis Toolbox (CAGECAT) [68] and OrthoANI [69]. Phylogenetic analysis including other published Streptococcus phage genome sequences was carried out using the Virus Classification and Tree Building Online Resource (VICTOR) web service https://victor.dsmz.de (accessed on 13 February 2025), a method for genome-based phylogeny and classification of prokaryotic viruses [70]. VICTOR analysis conducts all pairwise comparisons of the nucleotide sequences using the Genome-BLAST Distance Phylogeny (GBDP) method [71] under settings recommended for prokaryotic viruses [70]. Taxon boundaries at the species, genus and family level are estimated with the OPTSIL program [72], the recommended clustering thresholds and an F value (fraction of links required for cluster fusion) of 0.5 [70,73]. Phylogenetic analysis was done including full Streptococcus phage genomes. They consist of all complete “Streptococcus phage” sequences found in the NCBI nucleotide database that had SBSEC Streptococcus as a host. It also included phage genomes that had greater than 50% identity over more than 45% query cover to the genome of the phages from this study as found by NCBI BLAST. We also included 1 Mitis group and 2 Pyogenic group Streptococcus phages for comparison to dissimilar Streptococcus species phages. The phage genomes included in the phylogenetic analysis [47,48,49,74,75,76,77,78] are listed in Supplementary Table S2.

2.4.3. Transmission Electron Microscopy (TEM)

Phage particles were imaged by transmission electron microscopy (TEM) as described in [36]. Briefly, glow discharged formvar Carbon 400 mesh grids (Ted Pella Inc., 01754-F, Redding, CA, USA) were prepared using 7.5 mg/mL uranyl formate (Polysciences Inc., 24762, Warrington, PA, USA) as the staining solution. Three µL of concentrated phage stock (which was filtered to remove cellular debris) was applied to each grid for 1 min before removal using a filter paper. After drying overnight, grids were imaged using a JEOL JEM-1400Flash transmission electron microscope (JEOL Ltd., Tokyo, Japan) equipped with a tungsten filament and 15-megapixel AMT NanoSprint15 sCMOS camera (Advanced Microscopy Techniques, Woburn, MA, USA).

2.4.4. Assessing Inhibition of Bacterial Growth

Bacterial isolates were grown in BHI broth until mid to late log phase of growth was reached (approximately 5 h until OD₆₀₀ = 0.5 to 0.7). Cell density was adjusted to 2 × 10⁷ CFU/mL in double-strength BHI broth. Phage lysates of known titer were adjusted to 2 × 10⁷ PFU/mL and serially diluted ten-fold in SMB to assay phage activity at various multiplicity of infections (MOI). Then equal volumes of 100 µL of cells and 100 µL phage lysates were mixed in duplicate wells of a 96-well culture plate. Each experiment was done in triplicate. Bacterial growth at 37 °C was monitored by measuring absorbance at 600 nm using a BioTek synergy H1 plate reader at 10-min intervals with discontinuous shaking for 20 h (10 s double orbital shaking 807 cm before each read). Data from the BioTek Synergy H1 microplate reader was acquired with the Gen5 software version 3.10.06 (Biotek Instruments, Agilent Technologies, Inc., Santa Clara, CA, USA). Growth curve experiments were performed with a minimum of two biological replicates. Statistical analysis and data visualizations were performed either in Microsoft Excel (Microsoft Office 16) or Python 3.10, using Matplotlib version 3.10. or Seaborn 0.13.2 plotting software. The graphical abstract was made using Biorender. Area under the curve (AUC) was calculated using Simpson’s rule [79,80]. This metric evaluates phage performance over time; a higher AUC indicates more bacterial host growth and a lower AUC indicates robust host growth inhibition. Results were visualized using bar plots that displayed the mean normalized responses with propagated error bars. Significance markers (***, **, *, ns) were added above bars based on the p-values (p < 0.001, p < 0.01, p < 0.05, p ≥ 0.05). Error bars represent propagated standard errors to account for variability in replicates and normalization. Independent two-sample t-tests assuming unequal variances (Welch’s t-test [81]) were performed to compare the normalized responses of untrained and trained phages for each condition.

2.4.5. Lifestyle Characterization

Lifestyle prediction analysis was performed on all sequenced phage genomes using the several algorithmic tools: PhageTerm (Galaxy version 1.0.12) [82,83], PhaTYP 0.3.0 [83] within PhaBox2 2.1.10 [84] and PhageAI (version 1.0.2) [85]. For 7 selected phages with ambiguous lifestyle predictions, representing both phage types under investigation, experimental validation of lysogenization capacity was conducted [86]. The temperate phage Vroast, though only partially characterized, was also included as a positive control for lysogenization. To detect lysogenized phages in bacterial survivors of infection, [86] two approaches were employed. First, spot tests were incubated extensively up to four days to promote mesa formation. Mesas are zones in the center of lysis spots which, after a prolonged incubation, show confluent bacterial growth due to the bacteria becoming resistant to the phage, possibly (but not exclusively) due to homoimmunity [86]. Bacteria from these mesas were isolated on BHI agar, and five single colonies per mesa were selected for analysis. When mesas yielded no bacterial growth, liquid infections were performed using initial bacterial concentrations of 1 × 10⁸ CFU/mL at MOI = 1, conditions known to enhance lysogeny rates [87,88]. All surviving clones underwent three successive streak purifications to eliminate residual free phage. Survivors were tested for phage immunity through double-layer agar assays, where the original phage was spotted onto bacterial lawns of the surviving isolates [86].

The presence or absence of phage DNA in bacterial genomes was then verified by PCR using phage-specific primers (Supplemental Material Table S3). Furthermore, surviving bacterial isolates were also subject to mitomycin C induction [89], using a concentration of 1.49 µM and exposure time of 14 h. The resulting lysates were spotted onto double-layer agar plates containing the naïve host to check for plaque formation.

2.4.6. Stability of Phages in Rumen Fluid

Stability of SBSEC phages in rumen fluid was evaluated in a subset of 8 phages presented in this study. Phage suspensions were diluted in 5 mL sterile rumen fluid (Bar Diamond Inc., Parma, ID, USA) at a concentration of 1 × 10⁷ PFU/mL and incubated at 37 °C under anaerobic atmosphere. The fraction of surviving phages was quantified by spotting serially diluted phages in rumen fluid on DLA. All experiments were done in duplicate and spots on DLA were done in triplicate.

2.5. Phage Training to Increase Killing Efficiency

Three of the isolated bacteriophages were subject to 30 rounds of serial passaging with their host adapting methods used by others [90,91,92,93]. Briefly, mid log host bacterial cultures were used to inoculate 5 mL BHI broth at 10⁸ CFU/mL and infected with the corresponding phage at MOI = 0.01, followed by incubation at 37 °C statically, overnight. Serial passaging was done by transferring 100 µL of the infected culture to 5 mL of fresh BHI medium and incubating again for 8 h or overnight until 30 rounds were completed or phage loss was observed. For every 10 cycles, lysates were made by centrifugation (10,000× g for 10 min) and filter sterilization using a 25 mm diameter, 0.22 µm pore size PVDF hydrophilic syringe filter (Choice™, Thermo Scientific™, Waltham, MA, USA). The viability of the phages was confirmed by spotting lysates on a DLA of the corresponding naïve host.

3. Results

3.1. Genetic Analysis of Isolated Phages Reveals Two Clusters of Unique Phages That Have Strong Intercluster Homology

We isolated and sequenced 12 novel Streptococcus phages ranging between 33.8 and 40.5 Kbp in size, with GC content between 37% and 40% (

Table 1. A summary of genetic characteristics of novel Streptococcus bacteriophage isolates.

Isolation Source	Cluster	Phage	Production Host	Genome (bp)	GC Content (%)	Predicted CDS	CDS Hypothetical
Fecal	1	CSJC	OC2C	33,868	37.16	58	32
Fecal	1	Mushu	B+	36,293	37.1	67	34
Fecal	1	Taco	OC1D	35,445	37.16	66	37
Rumen	2	Pika	MEM36	40,558	39.5	68	39
Rumen	2	36L	MEM36	40,696	39.41	68	39
Rumen	2	B-single	C5D10	40,564	39.5	67	38
Rumen	2	PYS40	MEM35	40,560	39.49	68	39
Rumen	2	PY1	MEM7	40,576	39.49	72	41
Rumen	2	PY2	MEM7	40,589	39.49	71	38
Rumen	2	PY3	MEM7	40,590	39.5	67	38
Rumen	2	PY4	MEM7	40,595	39.5	72	40
Rumen	2	PY7	MEM35	40,588	39.5	70	39
Rumen	2	PY9	MEM7	40,592	39.49	72	37
Fecal	-	Vroast	MP2-7	32,968	40.6	58	31

). Annotation of the phage genomes showed clustering of genes by functional groups and all genes necessary for full phage functionality are present (Figure 1). Based on genomic similarity and phylogenetic relatedness, these phages segregated into two distinct clusters (Figure 2 and Figure 3).

Phylogenetic analysis, incorporating our 12 sequenced phages and previously sequenced Streptococcus phages (listed in Supplemental Material Table S2), grouped our phages in two different clusters and confirmed their novelty (Figure 2). All phages are predicted to belong to a common family along with other Streptococcus phages. Cluster 1, where Mushu, Taco and CSJC are, grouped them with high phylogenetic similarity with SB01 (S. equinus phage) and, to a lesser extent, the Streptococcus pasteurianus phages SG586P1 and SG586P3, all members of the SBSEC streptococci [94]. Notably, two Streptococcus thermophilus phages, D4446 and phage P738, also demonstrated close genetic relatedness to the Cluster 1 phages. Victor analysis assigned Cluster 1 phages into the same genus, though identifies them as different species. The next eight phages (as shown highlighted in Figure 2) sequenced all demonstrate high genetic similarity, leading us to consider them as a group, Cluster 2. All Cluster 2 phages share the same predicted family, genus, and species, as would be expected of highly genetically similar isolates. This group also shows some relatedness to the final phage, a singleton lysogenic phage Vroast which bears closest resemblance to the SBSEC prophage Javan 220 [78]. The SBSEC phages included in our analysis, vB_SbRt-pBovineB21, phage vB_SbRt-pBovineS21 and ImqsRe26_1, are closely related to phage C1 whose host belongs to the Lancefield group C of streptococci [75], a different clade than SBSEC. Unsurprisingly, other phages from different streptococcal species were found to have phylogenetic relatedness, such as phages ImqsRe26_1 (Streptococcus equi) and phage C1 (Group C streptococci). Altogether the species of the bacteriophage hosts did not strongly correlate with phylogenetic clustering.

Clinker analysis (Figure 3) aligning the annotations of the sequenced genomes showed that Cluster 1 phages have similar genomes and the arrangement of gene functional groups is maintained in orthologous blocks, though there were clear differences between the members of the group indicated by regions without homologues identified. Interestingly, this was not the case for Cluster 2 phages. The differences between the individual genomes spanned the entire length of the sequence and seemed to concentrate around functional genes, in particular, genes related to tail subunits and assembly.

Analysis using OrthoANI (Orthologous Average Nucleotide Identity) (Figure S3) matched the segregation of phage genomes in two clusters obtained from VICTOR phylogenomic analysis (Figure 2). OrthoANI measures the overall similarity between two genome sequences by fragmenting the genomes into 1020 bp-long pieces and calculating the average nucleotide identities in an orthologous manner and assigning a percentage similarity. According to this analysis, Mushu, CSJC and Taco are between 96.14 and 99.91% similar to each other, while cluster 2 phages are between 99.87 and 99.98% orthologically identical. Cluster 2 phages are about 86% OrthoANI similar to Javan 220, a temperate phage described elsewhere, which also showed high similarity to Vroast genome.

Imaging via TEM of a selection of the isolated phages revealed that they all present a siphovirus morphology (Figure 4), with capsids of icosahedral or almost round shape and long, thin, flexible tails. This was consistent with predictions provided by PhageTerm [82] as most siphoviruses are headful packaging phages. Analysis of phage genomes and raw sequencing reads using PhageTerm yielded predictions for packaging mechanism and classification (Supplemental Material Table S6). PhageTerm, predicted CSJC, Vroast and PY9 to be headful packaging phages. Interestingly, PY3 was classified as T5 type, Direct Terminal repeats (DTR) class whereas PY9 (which is 99.91% OrthoANI similar to PY3) was classified as P1 type and headful (PAC) class. All other phages were classified as new or unknown class and unknown type. PhageTerm determined all phage ends to be redundant and most to have permuted genomes, which is consistent with headful packaging phages.

3.2. The Isolated Streptococcus Bacteriophages Have a Limited Host Range but Are Generally Effective at Inhibiting Bacterial Growth In Vitro, and Retain Activity When Exposed to Ruminal Fluid In Vitro

The phages presented in this study demonstrated infectivity against various Streptococcus strains, including bacteria isolated from both related sources and from cattle separated by significant geographical and temporal distances. The host range profiles of all 12 bacteriophage isolates are presented in Figure 5.

We evaluated each bacteriophage’s ability to inhibit the growth of their isolation host by measuring the OD₆₀₀ in liquid cultures at various MOIs (Figure 6, Supplemental Figure S1). All 12 fully sequenced bacteriophages demonstrated the ability to suppress the growth of their host to some extent, some for up to 20 h. Despite the genomic similarity found among phages, their ability to inhibit bacterial growth was observed to be very different. CSJC outperforms similar phages Mushu and Taco at inhibiting bacterial growth at all MOIs for 14 h and causing minimal growth for up to 20 h. Phage PY9 sustains better suppression of bacterial growth at all MOIs than genetically similar phages Pika, Bsingle, 36L, PY1, PY3 and PY7. All phages, except Vroast, completely inhibit growth at MOI = 1, MOI = 0.1 and sometimes even at MOI = 0.01, for at least 10 h while most perform poorly at MOI = 0.001. Excluding Taco, after 20 h the max OD₆₀₀ or phage-treated cultures at MOI = 1 was only 32% of the control bacteria in the worst performing phage (PY3) and 2% in the best (CSJC) (Supplemental Material Table S4).

Stability of SBSEC phages in rumen fluid was evaluated in a subset of 8 of the 12 phages presented in this study. Phage suspensions in 5 mL sterile rumen fluid (1 × 10⁷ PFU/mL) were incubated at 37 °C under anaerobic atmosphere and the fraction of surviving phages was quantified. The titer of phages dropped 0 to 0.45 log after 2 h in rumen fluid and continued exposure for 4, 24 and 48 h further decreased phage viability but not by more than 1.36 log (Taco) and remained around less than 1 log drop for the other 7 phages tested (Figure S4). Despite this considerable loss in phage viability, >1 × 10⁶ PFU/mL remained after 48 h. In actual application of phages as a feed additive there would be additional forces at play, while some would be detrimental, phage infection of SBSEC hosts leading to the release of more phage particles is also anticipated. Animal studies would provide true insights into the viability of these phages to modulate SBSEC proliferation in the rumen. Although the shelf stability of the phages from this study was not evaluated, it was noted that titers of phage stocks stored for 4 months at 4 °C in SMB presented minimal titer loss, but longer storage had variable effects in phage stability sometimes dropping 3 log units in 10 months (Taco and PY7) and sometimes maintaining similar titers for up to 12 months (CSJC and 36L) (Supplemental Material Table S5).

3.3. All but One of the Sequenced Phages Are Lytic, but Further Analysis Is Required for Cluster 2 Phages

Analysis of phage genomes using PhaTYP and PhageAI yielded lifestyle predictions that varied among the different algorithms. With the exception of lysogenic phage Vroast, all phages were predicted to be virulent by at least one algorithm, with some receiving virulent predictions from both PhaTYP and PhageAI (Supplemental Material Table S7). Genomic analysis revealed no canonical lysogeny-associated genes (integrase, excisionase, repressor or Cro protein) in any of the 11 lytic phages. In contrast, Vroast carries an integrase, superinfection exclusion and transcriptional regulator genes encoded within the same region. In the 11 lytic phages, we found other genes that can potentially be associated to lysogeny however all of them are also found in strictly lytic phages. A comprehensive list of all the genes relevant for considerations of therapeutical use that were identified in the genomes of all the phages from this study is found in Supplemental Material Table S8. They include a transcriptional regulator upstream a HNH endonuclease in Cluster 2 phages, recombinase-like proteins in Cluster 1 and Cluster 2 phages, a single strand binding protein found in both Cluster 1 and Cluster 2 phages, and a holiday junction resolvase found in Cluster 2 phages. Interestingly, a ParB-like partition protein is present and highly conserved in the genomes of Cluster 2 phages. No antibiotic resistance genes or virulence genes were found in the genomes of the phages presented here. A sequence which is predicted to be a phosphoadenosine phosphosulfate (PAPS) reductase is also present in Cluster 2 phages.

A subset of 7 Streptococcus phages was used for the infection of host bacteria under conditions known to promote lysogeny [87,88]. The assessment included the temperate phage isolate Vroast as a positive control for lysogenization, CSJC and Mushu from Cluster 1 and PY1, PY7, 36L and Pika from Cluster 2. Five single colonies that survived infection were isolated per phage exposure. PCR screening using cluster-specific primers revealed no detectable phage DNA present in any infection survivors (Supplemental Material Table S6). Due to lack of sequencing information, no PCR-detection of phage Vroast could be done; however, mitomycin treatment to induce potential prophages resulted in 4 Vroast-exposed survivors producing infectious particles capable of lysing the naïve host. Phage sensitivity testing of surviving colonies provided additional insight into potential lysogenization. Of the 4 Vroast survivors that produced phages following mitomycin induction, 3 also showed phage immunity, confirming successful lysogenization. Some PY1 and Mushu survivors exhibited immunity to phage infection, despite negative PCR and mitomycin induction results, suggesting an alternative resistance mechanism present in 4 and 3 of the 5 survivor isolates, respectively.

3.4. Phage Training by Serial Passaging Had Inconsistent Results

The phage training protocol of sequential co-culture by serial passaging that we applied to train three bacteriophages relies on the bacteria-bacteriophage arms-race dynamics that select for phages better at infecting bacteria, while at the same time selecting for bacteria that are better suited to survive phage infection. The ability of PY7 to inhibit bacteria growth after 16 rounds of training (after this PY7 lost its ability to plaque) was statistically worse than the original PY7 on naïve bacteria at MOIs of 0.001, 0.1 and 1 (p < 0.05). At the lowest MOI, a 32% reduction in bactericidal effect was observed. Mushu presented a 9% increase in killing efficiency at the lowest MOI, 0.001 (p < 0.05) and no difference at other MOIs. Growth kinetics of trained phage PY1 demonstrate a strong and significant improvement at the lowest MOIs of 0.001 and 0.01, improving performance as compared to untrained phage by 26% and 93%, respectively (p < 0.001) (Figure 7, Supplemental Material Figure S1).

4. Discussion

A total of 12 SBSEC bacteriophages were isolated from cattle feces or ruminal fluid and were comprehensively characterized. Despite great efforts to isolate lytic SBSEC phages, including processing of many ruminal fluid and fecal samples, only one lysogenic phage and 11 lytic phages belonging to two truly distinct phage types were isolated, congruent with findings by others and with the observation that despite the high numbers of phages in the rumen, most may exist in harmony with their hosts in a state of lysogeny or pseudolysogeny [95]. The host range of SBSEC phages presented here was limited, similar to what has been previously described by others for this bacterial species, however some phages lysed bacterial isolates from distinct geographic and temporal sources suggesting phage adaptation to common bacterial features of this species [44,46]. Most of the phages presented exerted long-term inhibition of bacterial growth in pure culture in some instances for up to 20 h, making them potential candidates as precise biocontrol agents. Although our investigation was limited to in vitro experiments, during actual applications, in vivo dynamics often amplify the effectiveness of phages due to limited resources, competitive exclusion, and environmental stresses [96,97]. While previous publications have reported the isolation and characterization of SBSEC phages, many of the early isolated SBSEC phages were lysogenic or lack full genomic sequence characterization [44,45,46,48,49,75,95,98,99]. However, these investigations laid a strong foundation in understanding the prevalence and role of bacteriophages in the ruminal ecosystem and how SBSEC phages could be applied to control bacterial proliferation in the rumen [42]. The phages isolated in this study contribute significant diversity to known Streptococcus phages and as such they may represent unique adaptations or functionality not previously described. Our results demonstrate that these phages are lytic and further investigation of their safety and performance in more complex systems or in vivo is warranted. Note, however, that phages exclusively infect bacteria and cannot infect animal or plant cells, making them inherently safe from that perspective.

Genomic sequencing along with phylogenetic and OrthoANI analysis showed that 11 of the 12 phages sequenced clustered in two different groups, here named Cluster 1 and Cluster 2, plus a singleton Vroast which were consistent with the observed differences in host range. Within each cluster there were high levels of genomic identity, particularly in Cluster 2, despite observed differences in the host range and the dissimilar ability of phages to control bacterial growth in vitro. Further resolution of single basepair differences between phage genomes within each cluster could be obtained by the use of hybrid assemblies, however assemblies generated using ONT data can be sufficient (achieving raw read accuracies of approximately 99%) especially when the goal is to generate de novo assemblies that allow identification of genes present or absent in a genome [62,100,101]. The observed phenotypic differences between highly similar phages are either related to the small genomic differences observed between genomes but could also be due to the differences in bacterial hosts used for amplification that could impart the phage lysate additional characteristics [102]. Other reports of SBSEC phages have also found high similarity between their isolated phages, perhaps indicating the prevalence of specific phage types within the ruminal ecosystem [48,49].

Similarities to other phages with lytic activity against other SBSEC bacteria were present, as expected, however phages with hosts not belonging to SBSEC were also clustered together. This similarity between Streptococcus species infecting phages has been observed previously and seems to indicate adaptation to features common to streptococci for phage attachment and to ensure protein function in closely related bacterial strains [47]. Similarities to other genus-specific phages is expected considering that phages show complex evolutionary relationships that do not follow hierarchical phylogeny due to prevalent mosaicism [103]. Phages CSJC, Mushu and Taco were grouped in Cluster 1 closely together with 3 Streptococcus phages, including S. equinus (SBSEC) phage Sb01 which is described to be lytic based on the production of clear plaques and clear bacterial cultures and due to the lack of integrase gene [47] and phages P738 and D4446 which are also deemed to be lytic [77]. The phages belonging to Cluster 2 are grouped separately from other Streptococcus phage sequences used in this analysis, but as expected from their genomic similarity, highly related to each other. Cluster 2 phages have phylogenetic relatedness and about 86% OrthoANI similarity to Streptococcus phage Javan 220. The full-length prophage Javan 220 was identified in the genome of S. equinus strain Sb02, but has not been isolated and propagated. In contrast to the 11 lytic phages from this study, Javan 220 carries an integrase and a repressor [78], as does Vroast, which are crucial functions for the temperate lifestyle of lysogenic phages.

None of the phage genomes presented in this study carry any antibiotic resistance genes or virulence genes that could render them unsuitable for any potential therapeutical applications. Cluster 1 phages carry a YopX-like protein. Although Yersinia outer proteins (Yops) are a well described group of virulence factors of plasmid origin, they are also found in the replication module of a variety of Streptococcus and Staphylococcus epidermidis bacteriophages and their function is considered to be unrelated to virulence [104,105]. Cluster 2 phages carry a potential moron (at term referring to an extra gene within a prophage genome which confers benefits the host [106]): a phosphoadenosine phosphosulfate (PAPs) reductase. Considering the importance of this protein to the metabolic function of bacteria it has been suggested that this gene could provide an advantage to its host by facilitating inorganic sulfate assimilation [107]. However, PAPs are widely distributed among phages and could also be used to recruit cysteine for phage assembly [108,109].

None of the typical hallmarks of lysogeny were identified in the genomes of the phages presented, specifically, no integrase, excisionase or repressor genes, with the exception of Vroast. However, the two tools used for lifestyle prediction, PhaTYP and PhageAI, showed conflicting results for most phages, possibly due to the identification of plausible bacterial genes. The various tools available are trained on different sets of data and as a result the outcomes can be dramatically different when analyzing the same type of data. It is important to note that all phages presented in this work were predicted to be virulent by at least one tool except for Vroast. Another consideration is the dearth of lifestyle classification of many available phage genome sequenced; therefore, the ambiguity of lifestyle classification tools may be due to flawed inputs yielding unreliable results. We looked for genes which could be indirectly related to maintaining a lysogenic lifestyle and found a limited number of CDS annotated as genes with functions that can be important for lysogeny, though they also may play a role in the lytic lifestyle of virulent phages (Supplemental Material Table S8).

Both Cluster 1 and Cluster 2 phages carry transcriptional regulators, which are most likely related to the temporal expression of phage genomes for effective replication assembly and lysis of the host, critical in both lytic and lysogenic phages [110]. In Cluster 2 phages there is a Holliday junction resolvase (HJRs), which is a key enzyme of DNA recombination rarely involved in the lysogenic lifestyle of bacteriophages [111]. In both Cluster 1 and 2 there are single-strand annealing proteins and DNA binding proteins that are frequently encoded in the genome of many bacteriophages playing a role in DNA repair and recombination, essential for phage replication and mosaicism [112]. Sak4 recombinases (found in Cluster 2 phages) are commonly found in temperate phages more than in lytic ones [113] but their recombinatorial activity is still unconfirmed [113,114]. Although Cluster 2 phages carry an HNH endonuclease, which can be associated to genome integration in lysogeny, they are also present in lytic phages and involved in DNA packaging and defense against co-infecting phages [115,116,117].

The ParB-like partition protein gene found in the genomes of Cluster 2 phages is usually an important gene for the replication and segregation of plasmid-like prophages. Although ParB-like genes can be found in strictly lytic phages this role is less frequently described [118,119]. In Cluster 2 phages, the ParB-like partition protein was highly conserved but no ParA (ATPase) could be found which is another essential gene for plasmid-like replication [120]. Likewise, no toxin/antitoxin system was evident in our phages which is often required for maintenance of plasmid phage genomes in the bacterial host [121,122]. Together with our inability to find lysogens of Cluster 1 or Cluster 2 phages in a subset of isolates it is likely that all 11 are lytic. However, the presence of ParB-like and Sak4 genes in Cluster 2 phages suggests further investigation into their lifestyle is warranted. Interestingly only survivors of PY1 and Mushu produced some bacterial clones resistant to subsequent infection, however their lack of detectable PY1 or Mushu via PCR and lack of mitomycin inducible prophages suggests alternative mechanisms for their resistance, which often involves mechanisms of reduced phage adsorption [90,123,124]. In general, the lack of phage resistance in concert with strong killing efficiencies in vitro is encouraging, and merits further investigations in more complex systems.

Evolutionary adaptation to train our bacteriophages for improved phenotypes yielded variable results. Ideally, phage training would improve bactericidal effect at the lowest MOI, allowing for the lowest dosage of phage to be applied while maintaining bacterial growth repression. Phage training by co-evolution with their host only markedly improved the bacteriolytic activity of phage PY1 at the two lowest MOIs (out of 3 phages trained) despite multiple reports of successful applications of similar protocols to generate phages with improved infectivity and biocontrol [125,126,127,128] and strong genetic similarities with phage PY7, which demonstrated a statistically significant reduction in host killing performance at the lowest MOI. However, publication biases towards positive results make it difficult to understand how often phage training results in deleterious adaptations. The antagonistic co-evolution accelerates molecular changes and generates genetic diversity intra and inter-population in which both the bacterium and the phage are allowed to evolve adaptations and counter-adaptations [90,123,129]. For this reason, it is not surprising that phages better adapted to prevailing in the co-culture environment but not better at suppressing bacterial growth in vitro can be isolated. Furthermore, there may be host dependent factors that reduce the effectiveness of these types of training strategies, which may be revealed with more comprehensive investigation in future.

5. Conclusions

This study characterizes 12 novel phages with significant bacteriolytic activity against SBSEC bacteria, demonstrating the potential of 11 lytic phages as precise and safe biocontrol tools to ameliorate ruminal acidosis and increase feed efficiency. Genomic analyses of the lytic phages confirmed their safety through the absence of virulence and antibiotic resistance genes, while the lack of canonical lysogeny-related genes and experimental evidence further supports their suitability for application. Despite the limited host ranges of individual lytic phage isolates and clear trends between the two clusters, some phages demonstrated cross-geographic efficacy, suggesting there is potential for broader application in cattle microbiome management. We also observed limited development of resistance during experiments.

Our findings contribute to the growing diversity and understanding of Streptococcus phages, enriching the genomic datasets that underpin phage ecology and evolution. These phages represent a significant step forward in advancing phage-based biocontrol strategies, particularly in mitigating challenges like ruminal acidosis and improving cattle feed efficiency in agricultural settings. Phage training via serial co-culturing yielded variable effects on bactericidal efficiency, which underscores the complexity of phage-host interactions and the necessity for further investigation into optimization strategies. Stability assessments in rumen fluid demonstrated a moderate decline in phage viability, with most isolates retaining high infectivity after prolonged exposure, an essential consideration for future in vivo applications. While shelf stability varied among isolates, select phages maintained high titers for up to a year under refrigeration, supporting their potential as viable feed additives, though investigation into stability under other methods of storage, e.g., lyophilization, remain to be pursued.

These results underscore the importance of continued research to assess efficacy, safety, and unintended consequences of phage application and manipulation, which may ultimately contribute to the development of tailored, sustainable biocontrol solutions. As the cattle industry seeks alternatives to traditional antibiotics for managing SBSEC and improving feed efficiency, further investigation into these and other reported phages offer promising alternatives for targeted microbiome modulation. Future research should focus on several key areas: evaluating phage stability and efficacy under various environmental conditions, investigating potential synergistic effects of phage combinations, and assessing the emergence of bacterial resistance in complex microbial communities. Understanding these aspects will be crucial for developing effective delivery systems and treatment protocols. This work expands the toolbox for addressing agricultural challenges and establishes a framework for future studies in microbiome engineering and pathogen control, contributing to the broader goal of sustainable livestock production.