From Farm to Fork: Streptococcus suis as a Model for the Development of Novel Phage-Based Biocontrol Agents

Bacterial infections of livestock threaten the sustainability of agriculture and public health through production losses and contamination of food products. While prophylactic and therapeutic application of antibiotics has been successful in managing such infections, the evolution and spread of antibiotic-resistant strains along the food chain and in the environment necessitates the development of alternative or adjunct preventive and/or therapeutic strategies. Additionally, the growing consumer preference for “greener” antibiotic-free food products has reinforced the need for novel and safer approaches to controlling bacterial infections. The use of bacteriophages (phages), which can target and kill bacteria, are increasingly considered as a suitable measure to reduce bacterial infections and contamination in the food industry. This review primarily elaborates on the recent veterinary applications of phages and discusses their merits and limitations. Furthermore, using Streptococcus suis as a model, we describe the prevalence of prophages and the anti-viral defence arsenal in the genome of the pathogen as a means to define the genetic building blocks that are available for the (synthetic) development of phage-based treatments. The data and approach described herein may provide a framework for the development of therapeutics against an array of bacterial pathogens.


Introduction
The global livestock industry is a major contributor to food security and economic development, with a value of about 1.4 trillion dollars [1]. The Food and Agriculture Organisation (FAO) estimates that the livestock sector accounts for 40% of the overall agricultural output, with about 1.3 billion people depending on the sector for livelihood and food security [2]. However, infectious diseases threaten the sustainability and growth of the industry. Viruses such as porcine reproductive and respiratory syndrome virus (PRRSV) of swine, bovine herpesvirus 1 of cattle, and infectious bronchitis virus (IBV) of poultry cause severe contagious diseases within the livestock industry. Notwithstanding, bacteria are the key aetiological agents implicated in animal microbial diseases. Bacterial infections such as those arising from Campylobacter spp., Salmonella spp., E. coli, Clostridium spp., Listeria monocytogenes and Streptococcus suis take a significant toll on animal welfare and exert enormous financial losses on the industry [3]. It has been estimated that between 2000 and 2010, zoonotic infections caused about $20 billion and $200 billion in direct and indirect losses, respectively [4,5]. In Australia, it was reported that sales losses and disposal of affected pigs had an associated cost of AUD$ 10-30 million. These losses accounted for a 16-37% reduction in the gross income in the affected regions [6].  While the use of phages has not been approved for commercial use in most countries, they are used on a case-by-case basis in treating patients for whom currently approved therapeutics have failed (compassionate phage therapy). Table 2 below summarises some of the recent compassionate uses of phages in human infections. The next subsections explore some recent applications of phages (as single phage preparations or combinations of multiple phages in a "cocktail") in agriculture and the food industry. Table 2. Some applications of phages in human infections (2017)(2018)(2019)(2020)(2021)(2022).
Dyspnea resolved and cough reduced Increased lung function [49] Crohn's disease/MDR Klebsiella pneumoniae 3 week cycle single phage treatment 10 6 pfu/mL p.o., 10 6 pfu/mL rectal 15 days after first PT treatment, no MDR K. pneumoniae was isolated from patient's stools, rectal [50]  While the use of phages has not been approved for commercial use in most countries, they are used on a case-by-case basis in treating patients for whom currently approved therapeutics have failed (compassionate phage therapy). Table 2 below summarises some of the recent compassionate uses of phages in human infections. The next subsections explore some recent applications of phages (as single phage preparations or combinations of multiple phages in a "cocktail") in agriculture and the food industry.

Phages in Agriculture and Food Industry
The world is currently moving toward "greener" crops and animal food products, with an attitudinal shift in consumer preference for antibiotic-free and sustainably farmed and or processed products [59]. On farms, the restriction of antibiotic use and AMR are indicative that exploring alternative infection control is imminent. Thus, the antibacterial and environmentally friendly properties of phages support their potential use across the agricultural supply chain, such as in the "farm to fork strategy", which aims to provide safe, sustainable and eco-friendly food systems.

Phages in Primary Food Production
Salmonella spp. remains a leading aetiological agent of human food poisoning and gastrointestinal infections in swine, poultry and other livestock. As an enteric pathogen, intestinal contents are usually affected, which predisposes carcasses to contamination during slaughter and packing. To prevent this, controlling colonisation of the pathogen prior Viruses 2022, 14, 1996 6 of 27 to slaughter is practised. Thanki et al. determined the efficacy of a two-phage cocktail in reducing Salmonella enterica colonisation in piglets. Weaner meals infused with spray-dried phages were fed to experimentally infected piglets over five days. Phage treatment significantly reduced Salmonella carriage in the stomach tissue by 1 × 10 1 CFU/g, duodenum tissue (1.05 × 10 2 CFU/g), colon content (1 × 10 1 CFU/g) and caecum (1 × 10 1 CFU/g). A 16 S rRNA gene sequencing analysis demonstrated no deleterious effect on the overall microbiome following phage treatment. However, a group that received a phage diet yielded higher abundance of Prevotellaceae-a Bacteroidete that has been associated with disease resilience and growth performance in pigs [60]. This study is possibly the first report of scale-up data in a mainstream commercial spray dryer [61]. A similar study reported the clearance of Salmonella in the caecal samples (95%) and ileal samples (90%) when pigs were orally administered a microencapsulated phage cocktail post-bacterial challenge [62]. Gebru et al. also demonstrated that anti-Salmonella Typhimurium phage in diets significantly reduced bacterial shedding in pig faeces while increasing growth performance [63]. Assessment of phage efficacy in Salmonella and other infections associated with poultry has been extensively reviewed by Mosimann et al. [64]. Contrastingly, very few studies have reported the potential of phages in bacterial infections of cattle. Apart from some in vitro and mice model studies, no reports of farm trials were found in a non-exhaustive search using the keywords "phage", "cattle" and "biocontrol" in Google Scholar, Scopus and PubMed, [65][66][67]. This may be due to the cost associated with enrolling enough cattle in a study that would be suitable for making statistically sound inferences, as well as the space required to house the animals for the period of the study.
Avian pathogenic Escherichia coli (APEC) is an extra-intestinal E. coli that causes local and systemic infections in avian species such as chicken, turkey and ducks. Some common colibacillosis caused by APEC include egg peritonitis, pericarditis, salpingitis, airsacculitis, cellulitis and osteomyelitis [68]. Collectively, these infections contribute to a reduction in egg production (20%) and mortality of 20-40%, with birds between 4-6 weeks of age being the most susceptible [69]. APEC may be the singular pathogen in an infection or a preceding, concomitant or secondary infection in birds [70,71]. Tawakol et al. evaluated the efficacy of phages in a single APEC infection and a mixed APEC and infectious bronchitis virus (IBV). Challenging birds with APEC and/or IBV resulted in a mortality rate of 16% and 29% by day 7 and day 8, respectively. Conversely, intratracheal administration of phage preparation during the course of the study prevented mortality in birds challenged with APEC and APEC + IBV and decreased IBV and APEC shedding in these two treated groups [72]. A similar observation was made in a more recent study in which chickens were challenged with a multidrug-resistant and strong biofilm producer, E. coli O78. Birds that were treated with phages following challenge presented less severe clinical manifestations compared to the challenged untreated group. In addition, administration of phages offered complete protection, whereas in the untreated group, a mortality rate of 26.7% was recorded [73]. The superior efficacy of phage cocktails compared to a single phage preparation was demonstrated in colibacillosis in quails. A reduction in mortality from 26.5% to 13.6% was reported following an intramuscular injection with a single phage and a cocktail, respectively. Both phage-treated groups resulted in better clinical and survival outcomes compared to the challenged untreated quails (46.6% mortality rate) [74].
Campylobacter is a zoonotic pathogen responsible for~25% of all human diarrheal illnesses globally [75]. The bacterium is prevalent in several food animals, particularly in the avian gut, and could contaminate carcasses during postharvest processing. Richards et al. demonstrated that a two-phage cocktail could selectively infect and reduce C. jejuni populations in the gut of broiler chickens by 2.4 log CFU/g without disturbing the resident gut microbiome [76]. One major hurdle in the phage application in C. jejuni infection is the reported in vitro and in vivo emergence of phage-resistant isolates (up to 13% of isolates) [77]. However, rational selection of phages with diverse genetic characteristics can be used to reduce the challenge. Historically, based on genome size, morphology, and the host receptor employed during infection, Campylobacter phages were grouped into I, II and III [77]. In a 31-day trial, administering one group III phage followed by a group II phage significantly decreased C. jejuni counts in broiler chickens by 3.0 log units compared to a two-phage cocktail made of only group III phages (1.0 log unit). In addition to the higher reduction, combining group II and III resulted in lower levels of phage-resistant isolates compared to using a single phage or a homogenous phage cocktail [78]. The efficacy of this multi-group Campylobacter phage cocktail design has been validated in vitro using several group II and III phages [79]. The application of phage therapy in C. jejuni infection in chickens has been reported by other studies with varying degrees of efficacy [80][81][82].
Clostridium perfringens Type A strains form part of the resident microbiota of animals; however, toxin-producing strains are implicated in a number of livestock diseases, including necrotic enteritis in poultry, necrohaemorrhagic enteritis in cattle, haemorrhagic diarrhoea in swine, and pigbel in humans. The multi-host nature of this pathogen makes it an ideal target for phage therapy. The phage cocktail "INT-401" was examined for its effectiveness in chicken necrotic enteritis (NE). As part of the study, three methods of oral administration were tested. Throughout the 42 days of observation, all phage treatments significantly (p < 0.05) reduced mortalities due to NE to 0-14.0% compared to the mortality rate among the challenged untreated birds (64%) or the antibiotic/medicated group (50%). In-water delivery of phages was the most effective delivery method, with 0% mortality due to NE [83]. In a recent study, 14-day-old broiler chickens were experimentally infected with C. perfringens via oral gavage. Phage treatment via similar routes significantly improved clinical parameters and reduced gross lesions compared to the infected untreated birds. The mortality rate was 30% in the phage-treated group compared to the infected untreated group (55%) [84].
Aquaculture is one of the fastest growing industries in agriculture. The FAO reported that fish accounted for 17% of all animal protein consumed globally in 2017 [85]. Aquaculture production reached an all-time high of 84.1 million tonnes in 2018, higher than bovine meat or ovine meat production [86]; however, microbial infections threaten the growth and sustainability of this industry. Phage therapy has been used as a control strategy in bacterial diseases of fishes. In the terrestrial livestock setting, the delivery, diffusion and access of phages to infection sites or surfaces sometimes limit treatment outcomes. However, the liquid medium of aquaculture allows easy access of phages to pathogens in the water (farm environment) and to fishes through the gills [87]. This has been demonstrated in Aeromonas hydrophila infection in Nile tilapia. A. hydrophila is a zoonotic pathogen reported as the leading cause of septicaemia in freshwater fish [88]. In their study, Dien et al. demonstrated that successive introduction of the inoculum and phage (multiplicity of infection, MOI 1) in water significantly improved fish survival rates (up to 80%) compared to the 25% survival rate among fish in the infected untreated water. Furthermore, phage treatment reduced A. hydrophila in fish as well as stimulated the development of specific IgM against the MDR A. hydrophila [89]. The immunogenic and immunostimulatory capability of phages in humans and animals have been described by others [90][91][92]. However, an MOI of 100 was required to achieve a similar survival rate (85%) in turbot intraperitoneally infected with Vibrio harveyi and administered in-feed phage [93]. The differences in the efficacies of phage treatment between the two studies may be due to the delivery method utilised in the studies. Infusing phages in food pellets may have locked phage particles on food substrate and hence therapeutic phages were available to turbot mostly by feeding ad libitum. By dispensing directly in water, free phages became readily accessible to the Nile tilapia, and thus a comparatively lower MOI was required to produce a similar level of protection recorded in the turbot study. Other studies make reference to MOI when reporting phage dosages. However, factors such as target bacterium, animal species, phage infection kinetics and environmental conditions in the experimental setup could also account for the phage dosage required to offer protection [94]. It is likely that lowering the challenge inoculum by 10-fold would not be resolved by treating with a 10-fold lower phage dose. Thus, reporting of phage dosages could include more descriptive information, such as Poisson distribution and killing titres, which can improve reproducibility and ensure consistency in phage therapy and in the biocontrol literature [95]. Other factors that affect the efficacy of phages include the method of delivery. Phages can be inactivated by the acidic environment of the stomach and enzymatic degradation or neutralised by the immune system, which impacts efficacy of treatment in farm animals [96]. Encapsulating phages in biomaterials such as biopolymers and the use of nano-carriers have been applied for improved stability and accessibility of infection sites and the overall half-life of phages compared to free phages [97,98]. Phages for disease control in aquaculture have been extensively reviewed by Culot et al. and other groups [87,[99][100][101].

Phages as Growth Promoters
Phages have been demonstrated to play a significant role in growth promotion and improving gut health. Upadhaya et al. [102] assessed the effects of a commercial multispecies phage cocktail on growth performance and other production traits as well as gut health in healthy broiler chickens. A higher body weight gain was observed in birds that received phage-supplemented feed than in birds fed with only the basal diet. The effects of the antibiotic control diet and multi-species phage cocktail supplemented diet on growth performance, barrier function and gut microbiota were described in another study. A key observation was that at 200 mg/kg, there was no significant difference in growth performance compared to the antibiotic diet. Moreover, a 400 mg/kg and 600 mg/kg phage diet increased the daily feed intake, average daily gain and final body weight (p < 0.05) and reduced the incidence of diarrhoea in weaned piglets (p < 0.001) [103]. Studies that report on the effects of dietary phage supplements in healthy unchallenged food animals are scarce, but many recent studies in phage application in food and animal production include parameters that measure growth performance in their experimental design [61,63,83,104].

Phage Therapy in Crop Production
Phages have also been evaluated as biocontrol agents in crop production. Pseudomonas syringae, Pectobacterium spp., Ralstonia solanacearum, Dickeya solani and Xanthomonas spp. are some of the leading causes of bacterial diseases in crops. Phages that target these plant pathogens have been isolated and characterised as safer alternatives to conventional broad spectrum chemotherapeutics. Bacterial canker caused by Pseudomonas syringae is an economically important disease in farming. The efficacy of single phages in controlling canker was evaluated by spraying the leaves of two-year-old cherry trees with P. syringae. Phage treatment almost completely cleared the bacterial population (down to < 10 CFU) by the fifth week [104]. Dickeya solani and Pectobacterium spp. account for half of all Enterobacteriaceae soft rot (SRE) in angiosperms, including food and ornamental crops. A six phage cocktail was used to control soft rot in potato tubers following a 5-day incubation. Washing tubers in a phage preparation significantly reduced (p < 0.0001) maceration in D. solani infection [105]. Phage treatment significantly reduced disease severity (64%) in tubers and incidence of soft rot (61%) among potato tubers Pectobacterium atrosepticum [106].

Phages as Biopreservatives and Biosanitisers
After harvest, both animal and plant products are susceptible to bacterial contamination and spoilage while in storage or undergoing postharvest processing. The FAO estimates show that the global economy loses $220 billion annually due to plant diseases [107]. Moreover, in their latest (2015) global report, the WHO's Foodborne Disease Burden Epidemiology Reference Group (FERG) reported that over 600 million cases of foodborne illness were implicated in 420,000 deaths [108]. Besides colonising food products, bacteria can attach to materials, including stainless steel (SS), polyethylene, rubber and glass by forming an extracellular polymeric matrix (biofilms). The persistence of biofilms on food contact surfaces presents a major challenge in the food industry by damaging equipment, and contaminating food products, thereby increasing production costs. Thermal and non-thermal treatments are routinely used to remove biofilms from surfaces; however, in addition to their cost, these methods can alter the nutritional and organoleptic profiles of food [109,110]. Moreover, several studies have reported disinfectant-resistant biofilms at various stages of food production [111][112][113]. The antibacterial activity and the natural appeal of phages have been exploited in food processing to extend the shelf life of products, prevent food spoilage and decontaminate surfaces ( Table 3). The application of phages in food preservation has been reviewed by others in detail [114,115]. Several phages against ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter species) bacteria have been characterised and assessed in vivo for their potential in veterinary medicine and the food industry, with several commercial phage preparations (Table 4) already approved for use in food safety [130]. However, there are other animal bacterial pathogens of One Health importance that are poorly explored as targets for phage therapy. One such bacterial agent is Streptococcus suis, a major cause of respiratory and systemic infections in swine herds.

Streptococcus suis and Its Phages: A Case Study
S. suis is a gram-positive pathobiont of swine that commonly colonises the upper respiratory tract [131] and, to a lesser extent, the digestive and genital tracts of these animals [132]. However, some S. suis strains have evolved to become virulent, and in 1954, the first case of invasive infection was reported among piglets [133,134]. S. suis strains have been classified into 35 serotypes (1 to 34 and 1/2) based on the antigenicity of their capsular polysaccharides (CPS) [135]. DNA homology and biochemical profiling of the serotypes have led to the reclassification of serotypes 20, 22 and 26 as Streptococcus parasuis, 32 and 34 as Streptococcus orisratti, and 33 as Streptococcus ruminantium [136]. For diagnostic and surveillance purposes, all 35 serotypes are generally reported as S. suis. The prevalence of serotypes varies among different locations; however, most clinical cases worldwide have been associated with serotype 2 [137].
In addition to the CPS, which is a critical virulence factor of S. suis, proteins such as suilysin (Sly), DNase, muramidase-release protein (Mrp), surface antigen protein, VirA, and extracellular protein factor (Epf) contribute to the overall pathogenicity of the bacterium [138][139][140][141]. Infection in pigs by pathogenic and opportunistic pathotypes may arise from vertical transmission from sow to piglet during or soon after parturition, or by horizontal transmission such as in nose-to-nose contact among herd members [142]. Colonisation often goes undetected since it is not associated with symptomatic expression of disease. However when established, the bacterium may disseminate systemically, resulting in early manifestation, including fever, inappetence, depression, nystagmus and spontaneous death among infected piglets. Lesions due to polyarthritis, meningitis and endocarditis may also be observed [143].
S. suis is the most prevalent bacterium implicated in systemic infections in swine [144]. Regarded as a production disease, reporting of S. suis infections is not obligatory in many countries and hence data on global prevalence is limited [145]. This hampers efforts to estimate economic cost and the structuring of standardised control measures within the industry. The incidence of diseases among a given herd is typically lower than 5% and is a function of the extent of the prophylactic application of antibiotics [146]. An evaluation of the economic impact of S. suis on Spanish, Dutch and German farms determined that the cost per pig was € 0.60-1.30. Using a stochastic model, the study reported higher morbidities and mortalities in nurseries than in any other production stage. This reflected the higher cost of metaphylactic (treatment of animals with no signs of disease that have been in close contact with other animals with clinical manifestation) measures in piglets compared to that of autogenous vaccines for sows [147]. In 2005, Bennett et al. estimated that losses arising from S. suis type II infections alone were £100,000-1.3 million annually in Great Britain [148]. The loss is higher in Vietnam, where $370,000-500,000 went into direct costs and $2.27-2.88 million to indirect costs annually from 2011-2014 [149].
As is the case with many viral and bacterial pathogens, the threat posed by S. suis is not limited to the porcine industry [137]. Within 15 years of the first report of S. suis infection among piglets, a report of a human infection was identified among meningitis patients in Denmark [134,150]. In the same year, cases of S. suis meningitis and septicaemia among patients were recorded in the Netherlands and other European countries, suggesting that the pathogen may have crossed the species barrier long before it was first reported [151]. Recent time-dated phylogenomic studies employing analyses of single-nucleotide polymorphisms (SNPs) in S. suis genomes suggest the emergence of human-associated clade from swine isolates originating from western Europe. Using representative samples from six continents, Dong et al. traced the most common recent ancestry of human-associated clade to 1802-1855 [152]. A population expansion of this virulent zoonotic clade in the mid to late 1900s correlates with the extensive implementation of intensive pig rearing-where farmers were in close contact with suis-carrying pigs in enclosed and often poorly ventilated spaces-and the export of selectively bred pigs from Europe and United States to different continents [153]. Following the transcontinental exportation of zoonotic S. suis strains, new epidemic lineages appear to have emerged in Asia, where reports of human and porcine infections remain prevalent [154]. Cases of community-acquired S. suis infections among humans with no direct contact with pigs are becoming increasingly common [155][156][157][158]. Furthermore, cases of S. suis infections and S. suis isolates have been recently identified in cattle, lamb, dogs and other animals [159][160][161][162].

Phages of S. suis
Data from studies that have characterised phages that infect S. suis are limited. To date, only five lytic phages (SMP [163] and SS1, SS2, SS3, SS4 [164]) have been isolated against the pathogen. SMP is the first and only isolated virulent phage of S. suis with associated sequence data. Therefore, several studies have been conducted to characterise the phage at in vitro and genomic levels. The SMP phage was determined to have a narrow host range, as it could only infect 2 out of 24 serotype 2 strains included in the screening [163]. Other studies have chemically induced temperate phages from S. suis isolates [165][166][167]. In addition, S. suis genomic studies have contributed to the few available S. suis phage sequences [165,[167][168][169]. However, the effectiveness of phages in preclinical animal models of S. suis infections has yet to be reported.
Understanding host-phage interactions is key in the evaluation of phages as therapeutic agents against specific bacteria. These interactions include the preservation of phage populations through integration (prophage) in host genomes and the defence mounted by host bacteria via the expression of anti-phage systems. To shed light on the S. suis phage landscape, we analysed 133 publicly available whole genome sequences of S. suis for the presence of prophages and anti-phage defence systems. The sequences/strains (supplementary data Table S1) were randomly selected, and antiviral defence systems were identified using the web tool PADLOC (https://padloc.otago.ac.nz/padloc/) (accessed on 22 June 2022) [170]. Detection of candidate prophage regions was performed using PHASTER [171]. The predicted regions were categorised (scored) into full-length prophage (>90), putative full-length prophage  or incomplete prophage (<70) based on the number of coding sequences (CDS) and phage-related genes. To verify predicted regions, intact prophage sequences ≥10 kb were annotated using RAST [172]. A phage proteomic tree was generated using ViPtree [173] and visualised in iTOL.
The 133 S. suis genomes encoded a total of 1890 anti-viral proteins representing 20 distinct anti-viral systems (33, including system subtypes, Supplementary data Table S2). All strains encoded at least one defence system and a maximum of up to 10 systems in a single genome. The average number of systems per genome was calculated as 5.1, with 69.2% of strains encoding five or more systems. Restriction-modification (RM) systems (particularly RM type I) are the most abundant, nearly ubiquitous (90.2%) antiviral defence system family in the S. suis genomes. This was higher than the overall abundance (83%) of RM systems in most prokaryotic genomes [174]. Abortive infection systems, including AbiD, AbiE, AbiG, AbiO, AbiQ and AbiZ, were present in 74.4% of the analysed genomes, followed by the less-characterised tmn system (39.1%). Intriguingly, CRISPR-Cas systems (type I and II) were detected in only 28.6% of the genomes. Other recently identified systems such as Hachiman, Thoeris, AVAST and Stk2 were each only identified in less than 7% of the genomes. The different anti-viral systems identified in S. suis genomes could be grouped into three categories: DNA/RNA degradation, abortive infection, and systems of unknown mechanisms ( Figure 2C). The innate RM and adaptive CRISPR-Cas immune systems target and degrade viral nucleic acids while the signalling systems CBASS, Stk2, Paris, and PrrC, are involved in "cell suicide". Altogether, of the 1890 predicted proteins, 74.8% were functionally associated with degradation of viral nucleic acids.

High Prophage Prevalence and Diversity in S. suis Genomes
The 133 genome sequences of S. suis isolated from both diseased and healthy pigs or humans from 11 countries were queried for the presence of prophage sequences. Of the analysed genomes, 91.7% harboured at least one prophage (Figure 3). In total, 501 phage regions were identified, which included 330 incomplete regions, 100 putative full-length prophages, and 71 full-length prophages (Supplementary data Table S3). The number of prophage sequences in prophage-carrying genomes ranged from 1 to 10, with an average of 3.8 prophages per genome. Full-length prophages were detected in 42.9% of the genomes and carried essential phage genes representing a reservoir of genetic material for the development of S. suis therapeutic phages or derived enzymes. The average size of the full-length prophages is 45.5 kb, which is larger than the 36.0 kb of the S. suis SMP phage. Only full-length prophages were included in further analyses.

High Prophage Prevalence and Diversity in S. suis Genomes
The 133 genome sequences of S. suis isolated from both diseased and healthy pigs or humans from 11 countries were queried for the presence of prophage sequences. Of the analysed genomes, 91.7% harboured at least one prophage (Figure 3). In total, 501 phage regions were identified, which included 330 incomplete regions, 100 putative full-length prophages, and 71 full-length prophages (Supplementary data Table S3). The number of prophage sequences in prophage-carrying genomes ranged from 1 to 10, with an average of 3.8 prophages per genome. Full-length prophages were detected in 42.9% of the genomes and carried essential phage genes representing a reservoir of genetic material for the development of S. suis therapeutic phages or derived enzymes. The average size of the full-length prophages is 45.5 kb, which is larger than the 36.0 kb of the S. suis SMP phage. Only full-length prophages were included in further analyses.  The DNA sequences of all full-length prophages were aligned, and gene similarity and prediction was performed. Although the pairwise identity ranged from 0-100% about 72.4% of prophage-against-prophage identities were less than 10% (13.5% overal average similarity, Supplementary Materials Table S4). Nonetheless, a proteomic heatmap revealed nine clusters (identity ≥ 50%) and four singletons (Figure 4). Similarly, a phylogenetic tree using phage SMP as the root showed a similar clustering pattern ( Figure 5) In the case of strains harbouring more than one full-length prophage, there was a wide distribution of the prophages among clades, suggesting low similarity and high diversity ( Figure 5). Prophages within a clade were retrieved from hosts isolated from different locations and diverse serotypes, which suggests wide prevalence among S. suis serotypes and geographical regions. The DNA sequences of all full-length prophages were aligned, and gene similarity and prediction was performed. Although the pairwise identity ranged from 0-100%, about 72.4% of prophage-against-prophage identities were less than 10% (13.5% overall average similarity, Supplementary Materials Table S4). Nonetheless, a proteomic heatmap revealed nine clusters (identity ≥ 50%) and four singletons (Figure 4). Similarly, a phylogenetic tree using phage SMP as the root showed a similar clustering pattern ( Figure 5). In the case of strains harbouring more than one full-length prophage, there was a wide distribution of the prophages among clades, suggesting low similarity and high diversity ( Figure 5). Prophages within a clade were retrieved from hosts isolated from different locations and diverse serotypes, which suggests wide prevalence among S. suis serotypes and geographical regions.

Protein-Encoded S. suis Prophages
Full-length prophages encoded an average of 64 CDS (coding sequence) per sequence (ranged 13-141 CDS), among which, an average of 49.7% could be assigned a putative function. Bacterial proteins associated with virulence, such as toxin/antitoxin, efflux pumps, zeta toxin, and virulence-associated protein E, were detected in the prophages. In addition, anti-phage defence genes such as RMs were detected in prophage sequences, indicating that some bacterial defence systems are prophage-encoded (Supplementary Materials Table S5). The functional analysis revealed that most of the predicted genes are associated with phage morphogenesis, replication and transcription, lysis, and lysogeny. Representative prophages (8), one from each clade, were aligned. A modular arrangement of genes into various phage functions was observed ( Figure 6). The majority of prophage genes were transcribed in the same direction, except the lysogeny modules, which were generally transcribed in the opposite direction. Although alignment scores were very low among different clusters, members within the same cluster showed high gene similarities (Supplementary Materials Figure S1).

Relationship between Bacterial Genome Size, Number of Anti-Viral Defence, and Prevalence of Prophages
The number of prophages in a genome was correlated with the total anti-viral systems identified per genome ( Figure 7A). There is a moderate positive correlation between the two parameters (Spearman r = 0.40, p < 0.0001). The influence of host genome size on the association between prevalence of prophages and number of defence systems was controlled using a correlation matrix (p < 0.0001). Similarly, a moderately positive correlation was observed between the host genome length and the number of prophages (Spearman's r = 0.56, p < 0.0001). Up until 2.5 Mb, the number of prophages increased with increasing genome length ( Figure 7B).

Protein-Encoded S. suis Prophages
Full-length prophages encoded an average of 64 CDS (coding sequence) per sequen (ranged 13-141 CDS), among which, an average of 49.7% could be assigned a putat function. Bacterial proteins associated with virulence, such as toxin/antitoxin, eff pumps, zeta toxin, and virulence-associated protein E, were detected in the prophages. addition, anti-phage defence genes such as RMs were detected in prophage sequenc indicating that some bacterial defence systems are prophage-encoded (Supplementa Materials Table S5). The functional analysis revealed that most of the predicted genes a associated with phage morphogenesis, replication and transcription, lysis, and lysogen Representative prophages (8), one from each clade, were aligned. A modular arrangeme of genes into various phage functions was observed ( Figure 6). The majority of propha genes were transcribed in the same direction, except the lysogeny modules, which w generally transcribed in the opposite direction. Although alignment scores were very l among different clusters, members within the same cluster showed high gene similarit (Supplementary Materials Figure S1).

Relationship between Bacterial Genome Size, Number of Anti-Viral Defence, and Prevalence of Prophages
The number of prophages in a genome was correlated with the total anti-viral systems identified per genome ( Figure 7A). There is a moderate positive correlation between the two parameters (Spearman r = 0.40, p < 0.0001). The influence of host genome size on the association between prevalence of prophages and number of defence systems was controlled using a correlation matrix (p < 0.0001). Similarly, a moderately positive correlation was observed between the host genome length and the number of prophages (Spearman's r = 0.56, p < 0.0001). Up until 2.5 Mb, the number of prophages increased with increasing genome length ( Figure 7B).

Implications for the Development of Therapeutics
Persistent lysogeny and anti-viral mechanisms are critical phage-host interactions that affect the success of phage infection, particularly in the application of phages as therapeutic agents. We investigated the prevalence of prophages in S. suis and the anti-phage defence systems encoded in the bacterium's genome. All strains encoded at least one defence system, with RM systems being the most abundant. The high abundance of RM systems is consistent with a previous study of Type I RM systems in S. suis, which identified 95% of isolates encoding the defence system [175]. Interestingly, CRISPR-Cas systems were detected in a small fraction of S. suis genomes as opposed to the overall estimated abundance in bacteria (~50%) [176]. Moreover, the opposite is the case among other streptococci, such as S. thermophilus, in which all strains possibly encode the system, or in S. mutans, where 95% encode at least one type of CRISPR-Cas [177]. Recent studies have unearthed previously unknown anti-viral systems within bacterial defence islands. This suggests that current quantification of defence systems underestimates the prokaryotic defence arsenal. Except for tmn (39.1%) and gabija (21.1%), each of these newly described systems were present in less than 7% of S. suis. We categorised the defence systems on the basis of their molecular mechanism. Nucleic acid degradation systems were the most common mechanism in the S. suis defence arsenal. This is in line with the estimation for prokaryotes [174].
We identified 501 prophage regions, averaging 3.8 prophages per genome. The prevalence in S. suis is higher than has been reported for other Streptococcal species such as S. pneumoniae (1.4 per genome) [178] and S. mutans (1.5 per genome) [179]. Of the 501 prophages, 71 full-length prophages encoding essential phage genes were further analysed. It is possible that the number of full-length prophages was underestimated when using PHASTER because the majority (60.2%) of the genomes in this study were retrieved as multiple contigs. Thus, there is a possibility of prophages encoded on separate contigs, in which case they would be predicted to be incomplete. More than half of the intact prophages were detected in the genomes that were complete S. suis chromosomes. However, this observation could also be a function of the sample size. An all-against-all genomic similarity revealed little proteomic relatedness among full-length prophages suggesting a

Implications for the Development of Therapeutics
Persistent lysogeny and anti-viral mechanisms are critical phage-host interactions that affect the success of phage infection, particularly in the application of phages as therapeutic agents. We investigated the prevalence of prophages in S. suis and the antiphage defence systems encoded in the bacterium's genome. All strains encoded at least one defence system, with RM systems being the most abundant. The high abundance of RM systems is consistent with a previous study of Type I RM systems in S. suis, which identified 95% of isolates encoding the defence system [175]. Interestingly, CRISPR-Cas systems were detected in a small fraction of S. suis genomes as opposed to the overall estimated abundance in bacteria (~50%) [176]. Moreover, the opposite is the case among other streptococci, such as S. thermophilus, in which all strains possibly encode the system, or in S. mutans, where 95% encode at least one type of CRISPR-Cas [177]. Recent studies have unearthed previously unknown anti-viral systems within bacterial defence islands. This suggests that current quantification of defence systems underestimates the prokaryotic defence arsenal. Except for tmn (39.1%) and gabija (21.1%), each of these newly described systems were present in less than 7% of S. suis. We categorised the defence systems on the basis of their molecular mechanism. Nucleic acid degradation systems were the most common mechanism in the S. suis defence arsenal. This is in line with the estimation for prokaryotes [174].
We identified 501 prophage regions, averaging 3.8 prophages per genome. The prevalence in S. suis is higher than has been reported for other Streptococcal species such as S. pneumoniae (1.4 per genome) [178] and S. mutans (1.5 per genome) [179]. Of the 501 prophages, 71 full-length prophages encoding essential phage genes were further analysed. It is possible that the number of full-length prophages was underestimated when using PHASTER because the majority (60.2%) of the genomes in this study were retrieved as multiple contigs. Thus, there is a possibility of prophages encoded on separate contigs, in which case they would be predicted to be incomplete. More than half of the intact prophages were detected in the genomes that were complete S. suis chromosomes. However, this observation could also be a function of the sample size. An all-against-all genomic similarity revealed little proteomic relatedness among full-length prophages suggesting a high genetic diversity. Two patterns can be described for phage groups with low gene relatedness. Firstly, along the genomes of phages of common ancestry, several gene homologs of little similarity exist. Alternatively, phage mosaicism may arise from genetic transfer between dissimilar phages such that only a small fraction of the genome would contain highly similar homologs. The latter trend is consistent with the low inter-cluster relatedness as we qualitatively show in a schematic representation of functional genome alignment ( Figure 6). Furthermore, it has been demonstrated that temperate phages and virulent phages, employ high and low gene flux, or only low gene flux evolutionary modes, respectively. Particularly in Synechoccocus phages among which low gene flux is favoured, there is an observed clusters of genetically-distinct populations and constrained mosaicism [180]. Future investigations may explore the rates of genetic transfer and its influence on the diversity, host specificity and host range of S. suis temperate phages. Despite the low genetic similarity, all representative prophages from the clusters displayed a modular architecture such that essential phage genes could be categorised into structural, lysis and lysogeny, replication/transcription/metabolism, and packaging. Thus, it is likely that these temperate phages can be induced in vitro and explored for their potential for therapeutic applications.
The role of prokaryotic defences such as CRISPR-Cas in preventing successful phage infection has been described by others [181,182]. We hypothesised that the abundance and diversity of anti-viral defence systems would have a negative effect on the prevalence of prophages in a genome. On the contrary, we found a positive correlation between the number of anti-viral systems encoded in a genome and the number of prophages it harboured. We also found a similar association between host genome size and the prevalence of prophages. Strains with small genomes harboured no or fewer prophages compared to strains with larger genomes. This raises the questions of why and how a trifaceted association of S. suis host genome size, the prevalence of prophages, and the abundance of defence systems exists. In Salmonella and E. coli, larger genomes were associated with more integration hotspots. In particular, there seemed to be a selection for integration sites within intergenic and regions of low gene expression rather than highly transcribed regions. In highly transcribed regions, spillovers from transcribed genes could trigger phage gene expression, which can reduce host abundance and consequently the temperate phage population following a switch to lytic cycle [183]. Future investigations could examine integration sites utilised by temperate phages infecting S. suis and how this influences prophage prevalence among different strains. Following integration, temperate phages can confer fitness to the host. For instance, in the context of the correlation between defence systems and the number of prophages, annotations revealed the presence of phageencoded RM systems (Supplementary Materials Table S5). Thus, selection for such traits could lead to saturation of host chromosomes by prophages that encode fitness genes (including virulence and AMR), which will influence the diversity of the anti-viral arsenal of the host.
It has been suggested that the phage population in the gut microbiota is sustained predominantly through lysogenic replication, with spontaneous inductions rather than repetitive lytic cycles or frequent triggered prophage inductions [184]. In the lysogenic state, the prophage exerts superinfection immunity, which acts to prevent subsequent phage infections as demonstrated in mycobacterial phages [185]. We have shown that there is a high prevalence of prophages in the genomes of the bacterium; however, only five lytic phages against S. suis have been reported to date. It is possible that in this species, there is an evolutionary preference for lysogeny in the nasopharyngeal microbiota of the animal hosts. Notwithstanding, the prophages identified serve as building blocks for the development of candidate phages for therapeutic application in S. suis infections.

Future Perspectives
A potential approach to explore untapped (pro)phages against S. suis and other bacterial pathogens is through genetic engineering. Current genetic manipulation techniques such as recombineering, yeast artificial chromosome (YAC), CRISPR-Cas systems, and in vitro synthetic genome assembly have been leveraged to custom-design phages with desired characteristics [186]. Modifications through deletions and insertions of targeted genes or gene clusters can be employed to broaden host range of a phage, increase phage efficacy through evasion of specific host defences, and improve the safety of phages. By using a YAC platform, Ando et al. constructed synthetic T3 and T7 phages with swapped gp17-a host determinant. The resulting hybrids successfully cross-infected hosts of both phages. Using the same technology, engineered coliphage T7 could selectively eliminate Klebsiella sp. and Yersinia sp. in a mixed population [187]. Alternatively, the CRISPR-Cas system of S. thermophilus has been used to introduce specific point mutations, large deletions, and ORF replacement, with relatively high efficiencies. This model was adapted for the construction of escape phages through the introduction of a methyltransferase gene such that the resulting phage mutants evaded the R/M system of the host. This modification significantly improved the bacteriolytic efficiency of the S. thermophilus phage 2972 [188]. A synthetic genome assembly has been used to produce infectious viral particles from short phage DNA fragments [189], allowing for the deletion of lysogeny-associated genes and acquired virulence genes within phage genomes. Obligately lytic phages have been constructed from temperate phages for the control of Mycobacterium abscessus in cystic fibrosis patients [57,58]. Similar approaches can be explored in the custom design of lytic phages against S. suis using their prophages as building blocks.
Moreover, co-evolution "phage training" experiments have been adopted to adapt phages to oppose host defences and widen the host range [190]. The concept has been used to pre-adapt phages for treatment in a human K. pneumoniae infection [53]. However, this coevolutionary dynamic is yet to be validated in vivo for other clinically relevant species. Lastly, "Enzybiotics" explores the use of phage-encoded enzymes in the control of bacteria. Phage-derived enzymes and endolysins targeting S. suis have demonstrated flexibility in engineering, low toxicity profiles, and broad efficacy against the pathogen and other streptococcal species, which support their potential for translational development [191,[191][192][193][194].
Although naturally occurring phages possess a green appeal, the incorporation of genetically modified organisms (GMOs) in food may interfere with their approval by regulatory bodies. To date, the application of genetically modified phages in the treatment of infections in humans have been limited to rare cases with specific therapeutic circumstances meriting exemptions for their usage. Any additional barriers to approval will lessen commercialisation. Similarly, consumer reticence towards GMOs will decrease the likelihood of their application. Compelling research will be needed to prove the efficacy, safety and scalability of genetically engineered phages required by authorising bodies. Researchers also have a duty to engage with the public and policy groups to impress upon them the importance of phage solutions to tackle the catastrophe of antimicrobial resistance. As the prevalence of the problem continues to grow, this consumer reticence towards GMOs may become less of a challenge.

Conclusions
The emergence of antibiotic-resistant zoonotic bacteria poses a threat that transcends the food industry, affecting the overall health of the environment and other animal populations. The efficacy and safety of phages have been demonstrated in several bacterial infections, with some commercial products already approved in several countries for use in food production. However, there are other zoonotic bacteria, including S. suis, for which phage applications remain underexplored. We provide insights on the S. suis phage landscape and also paint a quantitative picture of the anti-phage arsenal of S. suis. Our findings reveal the diversity and abundance of S. suis prophages, which can be used as building blocks for synthesising safe lytic phages for application in food and medicine.

Conflicts of Interest:
The authors declare no conflict of interest.