Biodiversity of Phages Infecting the Dairy Bacterium Streptococcus thermophilus

Abstract

Streptococcus thermophilus-infecting phages represent a major problem in the dairy fermentation industry, particularly in relation to thermophilic production systems. Consequently, numerous studies have been performed relating to the biodiversity of such phages in global dairy operations. In the current review, we provide an overview of the genetic and morphological diversity of these phages and highlight the source and extent of genetic mosaicism among phages infecting this species through comparative proteome analysis of the replication and morphogenesis modules of representative phages. The phylogeny of selected phage-encoded receptor binding proteins (RBPs) was assessed, indicating that in certain cases RBP-encoding genes have been acquired separately to the morphogenesis modules, thus highlighting the adaptability of these phages. This review further highlights the significant advances that have been made in defining emergent genetically diverse groups of these phages, while it additionally summarizes remaining knowledge gaps in this research area.

1. Introduction

Streptococcus thermophilus is a dominant starter bacterial species applied in global dairy fermentations, particularly in the production of yogurt and Italian and Swiss-type cheeses. The continuous and intensive application of dairy starter cultures renders them susceptible to infection by bacteriophages that may be present in the starting product, i.e., milk, or in the processing environment. A recent comparative genome analysis of 23 S. thermophilus strains revealed that such strains can be divided into two major clusters based on genome size, the presence or absence of prtS, CRISPR (clustered regularly interspaced short palindromic repeats) 2, 3 and/or 4, and the distribution patterns of restriction-modification systems, among other features [1]. This apparent homogeneity within the species creates significant opportunities for phage proliferation, while it also highlights the dependence of this species on the activity of CRISPR-Cas systems for protection against phage infection [2,3,4]. Furthermore, phage adaptation and evolutionary strategies to evade CRISPR-Cas and other barriers to infection require considerable genome plasticity, particularly considering that “natural” evolutionary processes are expedited through the industrial scale, production intensity, and consistent provision of hosts in dairy fermentation environs [5,6].

A previous study of 81 dairy streptococcal phages derived from cheese and yogurt factories identified a greater diversity among cheese-associated isolates than those isolated from yogurt, perhaps reflective of the lower diversity of strains that were applied in yogurt fermentations [7]. This study represented one of the first significant efforts to classify phages of this species and applied serological, morphological, and molecular tools to achieve this, as well as defining the host range and lifestyle of the phages (lytic versus temperate). While phage taxonomy is now highly reliant on sequence data and sequence relatedness, historical approaches to phage taxonomy were based on a range of phenotypic attributes, such as morphology and the number of major structural proteins, leading to a limited interpretation of the extent of biodiversity in phages of this species. The natural reservoir for S. thermophilus phages is still not precisely defined. The presence of prophages among dairy streptococci was, until recently, presumed to be low (2%), being based on prophage induction trials [7,8], and the evolutionary path of these phages remains unclear. More recently, the abundance of genome data for this species has permitted detailed genome analyses and revealed that intact prophages may be present in up to 13 % of strains, while prophage remnants may be present in a majority of strains, thereby creating a genetic reservoir for recombination with infecting virulent phages [1]. There has been an exponential increase in the number of phage genome sequences available in public databases permitting their comparison and classification. In the present review, we will (i) provide an account on the genetic and morphological diversity of the major groups of dairy streptococcal phages, (ii) discuss the role of various structural proteins in facilitating host interactions, and (iii) identify genomic regions that are associated with diversification of these phages. We highlight the major advances in dairy streptococcal phage biodiversity and classification while simultaneously reporting on gaps in our current understanding of the mechanisms that underpin the observed genomic plasticity and interactions with their hosts.

2. Current Taxonomy of Dairy Streptococcal Phages

2.1. Biodiversity and Morphological Characteristics of Dairy Streptococcal Phages

Until 2011, just two groups of dairy streptococcal phages were known to exist. These were classified based on their mode of DNA packaging, i.e., cos (cohesive ends) and pac (headful packaging method), and the number of major structural proteins that they contained [9]. These groups have recently been renamed as the Moineauviruses (formerly called cos group) and Brussowviruses (pac group) [10]. Subsequently, three additional and genetically distinct phage groups were identified, designated the 5093 [11], 987 [12,13] and the P738 [14] groups that appear to be hybrid phages or derived from phages of distinct bacterial species.

All dairy streptococcal phages that have been isolated to date possess long non-contractile tails and thus belong to the Siphoviridae family of tailed phages. Though analysis of the morphological diversity of dairy streptococcal phages by electron microscopy may not discern these phages [15], it is now timely to revisit this point as phages infecting this bacterial species are more diverse than was previously known. All phages of S. thermophilus exhibit isometric capsids while their tails exhibit variable lengths with distinct structures at the tail tip in certain cases. For example, the tails of members of the Moineauvirus, Brussowvirus and 5093 phage groups typically have a length that exceeds 200 nm, while those of the P738 and 987 groups are much shorter with lengths of 120–140 nm [11,13,14,16]. Additionally, the tail tip of members of the 987 phages possesses large appendages similar to those observed in certain lactococcal P335 phages, whereas the tail tip of virions belonging to the 5093 group contains extending globular structures (Figure 1). These distinctive features, combined with tail length observations, are now useful in identifying members of at least some of the five dairy streptococcal phage groups. However, these morphological features are not suitable to easily discern between members of the most problematic and prevalent Moineauvirus and Brussowvirus groups.

2.2. Genetic Diversity of Dairy Streptococcal Phages

To evaluate the robustness of the current classification system, a clustalW alignment of the deduced genomes and concatenated proteomes of 182 dairy streptococcal phages (available in the NCBI database) commencing at the small terminase subunit protein (and following the genomic order) was performed. All protein sequences of the 182 phages were included. The alignment was employed to generate a phylogenetic tree using the “iTOL” software (http://itol.embl.de/) (accessed on 2 April 2021), applying the neighbor-joining method. This analysis confirmed the overall classification of all assessed phages among one of five major genotypes of S. thermophilus phages (A representation of the phylogenetic tree based on the whole nucleotide sequences is presented in Figure S1), being consistent with a recent analysis of 100 dairy streptococcal phage genomes at nucleotide level [14]. It further highlights that most sequenced phages currently described are classified among the Moineauvirus species (61% or 111 of 182 analyzed) or Brussowviruses (25% or 46 of 182 phages analyzed). These statistics, based on genome analysis, compare well to a recent survey of an industrial collection of S. thermophilus phages using PCR-based typing methods in which the Moineau- and Brussowviruses were observed to represent the dominant genotypes in a collection of 1179 phages (69% and 29%, respectively) [19].

While the overall genome/proteome analysis supports the existence of five distinct phage groups, perhaps what is less clear is the presence of shared functional modules between distinct groups that may unravel the evolutionary pathway of phages of this bacterial species. It has been noted that some dairy streptococcal phages represent hybrids of (pro)phages of distinct bacterial host species. For example, the 987 group phages are suggested to represent genomic hybrids containing a replication module similar to the Moineauvirus, 7201, and a morphogenesis-associated module with similarity to lactococcal P335 phages (being consistent with the associated morphological features, which will be discussed later) [13]. Similarly, the P738 phages (P738 and D4446) appear to be genetic hybrids containing a limited number of dairy streptococcal phage structural protein-encoding genes (believed to be involved in host interactions) and a replication module that is derived from non-dairy streptococcal phages including those of Streptococcus pyogenes [14]. In addition to these hybrids at host species level, it is also worth mentioning that several reports of module shuffling and DNA exchange between dairy streptococcal phage groups have been published. For example, the replication module of certain 5093 group phages was observed to share homology to those present in members of the Brussowvirus, Moineauvirus and the 987 group [12,13,20]. Similarly, a significant level of sequence relatedness was observed across the replication modules of the Moineauvirus P7573 and the Brussowvirus P7572. The region encoding the tail structural components, which include elements presumed to be involved in host interactions, is also similar between these phages (P7572 and P7573), highlighting the significant plasticity of these phage genomes [20]. This mosaicism is not a new concept and, in fact, was highlighted in 1999 in a comparative genome analysis of five S. thermophilus phages [21]. This early observation of insertions and deletions across the genomes of these phages has been corroborated in recent studies and allows us to now investigate the possible source of such divergences, thereby providing insights into their evolutionary pathway(s).

To ascertain the modular diversity and removing the bias of overall genome content, we selected 25 phages representing a spread across observed clusters in the overall proteomic tree (Figure S1) and representing those where previous observations of modular shuffling had been reported (Moineauviruses Sfi19, Sfi21, 9A, 128, Abc2, DT1, SW6, V2, STP1, CHPC572, P7573 and 7201; Brussowviruses 2972, Sfi11, P7572, ALQ13.2, TP-J34, P7955, and O1205; P738 phages P738 and D4446; 5093 phages 5093 and P0092; and 987 phages 9871 and SW22). Since these 25 phages are spread across the phylogenetic tree based on overall nucleotide content, we considered these to be representative of the overall cohort of dairy streptococcal phages. Secondly, the genomes were considered to contain three major gene clusters, i.e., those involved in replication, morphogenesis, and lysogeny (Figure 2). The morphogenesis modules were considered to encompass the genomic region associated with encoding proteins involved in DNA packaging and capsid/tail morphogenesis. Between the morphogenesis modules and the presumed replication modules are genes whose encoded products are predicted to be involved in host lysis and lysogeny. Lysogeny and so-called “lysogeny replacement modules” appear to be universally present among dairy streptococcal phages [21] and are highly diverse. The replication modules were arbitrarily defined as all genes downstream of those encoding lysis/lysogeny (replacement) functions and including those with identifiable DNA replication functions (Figure 2). A major defining feature of the 5093 and P738 phage groups is the genetic novelty associated with the structural modules among dairy streptococcal phages. Furthermore, the replication modules of dairy streptococcal phages have recently been reported to fall into one of two major groups [20]. Since these two functional modules are present in all dairy streptococcal phage genomes and with at least some conserved elements to allow comparison, they were selected as the basis of a focused genetic analysis. A phylogenetic analysis of the structural proteomes and replication proteomes (and excluding the lysogeny/lysogeny replacement modules) as distinct modules was undertaken to ascertain the diversity of modules that may act as a reservoir for the evolution of dairy streptococcal phages.

2.2.1. Replication Modules Display Inter-Group Relatedness

Significant genetic diversity was observed among the replication modules of the 25 analyzed phages; however, members of distinct phage groups were regularly observed to harbor at least one protein with sequence similarity above 50 % (Figure 3). Based on this analysis, there appear to be four Moineauvirus-specific replication modules (designated MV1 through to MV4) and five Brussowvirus-specific replication modules (named BV1-5) as well as one replication module type that is shared between members of both phage groups (called MV/BV), notwithstanding that the groups are arbitrarily established based on the cladogram presented in Figure 3. Additionally, one Brussowvirus replication module bears similarity to that of 5093 and another that is shared with 987 phages (Figure 3). These observed genetic “overlaps” are suggestive of evolution/emergence of the 5093 and 987 groups through recombination with temperate Brussowviruses. Interestingly, the replication module of phage P7955, a Brussowvirus that was previously reported to possess a unique replication module (i.e., that did not align with Brussowvirus O1205 or Moineauvirus 7201) according to Stanley et al. [22]), clusters with those of the 987 phages (Figure 3). While it is not possible to discern which is the ancestral phage lineage, it is tempting to speculate that the Brussowvirus group is likely the ancestral phage, given their prevalence in industrial fermentation environs for several decades. Phages P738/D4446 are confirmed as being unique among dairy streptococcal phages within the replication module.

2.2.2. Morphogenesis Modules Are Highly Group-Specific

Based on the analysis of 25 representative phage morphogenesis modules, it was observed that the modules clustered well according to the five major groups with subgroups observed among the Moineauvirus (MV) and Brussowvirus (BV) modules (Figure 4). Interestingly, the 5093-like morphogenesis module aligned most closely with those of the Brussowviruses, while those of the P738 and 987 phage groups formed separate clades on the tree (Figure 4). Based on this analysis, three Moineauvirus (MV1-3) and two Brussowvirus (BV1-2) phylogroups of structural modules could be discerned. Interestingly, the MV3 morphogenesis module appeared to be phylogenetically closer to BV1 than the MV1 and MV2 modules, which may represent an evolutionary link between the Moineauvirus and Brussowvirus modules. Furthermore, the 5093 phages and Brussowvirus phages morphogenesis modules cluster closely in the phylogenetic tree, highlighting a likely historical evolutionary link between these groups.

2.3. Evolutionary Pathways of Dairy Streptococcal Phages

Based on the phylogenetic analysis of the replication and morphogenesis modules of 25 representative dairy streptococcal phages, eleven and eight clusters of replication and morphogenesis module types, respectively, were arbitrarily identified. To assess the number of modular arrangements present among the analyzed phage proteomes, the derived proteome types of the morphogenesis and replication modules were collated (

Table 1. Summary of replication and morphogenesis module types based on structural and replication module phylogenetic analysis.

Phage	Phage Group	Replication Module Type	Morphogenesis Module Type
9A	Moineauvirus	MV1	MV2
V2, DT1, 128	Moineauvirus	MV2	MV2
P7573	Moineauvirus	MV2	MV3
SW6	Moineauvirus	MV3	MV2
STP1, CHPC572, Abc2	Moineauvirus	MV4	MV2
7201	Moineauvirus	MV4	MV3
Sfi19, Sfi21	Moineauvirus	MV/BV	MV1
Sfi11	Brussowvirus	MV/BV	BV2
O1205	Brussowvirus	BV1	BV2
P7572, 2972	Brussowvirus	BV2	BV1
ALQ13.2	Brussowvirus	BV3	BV1
TP-J34	Brussowvirus	BV4/5093	BV2
P7955	Brussowvirus	BV5/987	BV2
9871, SW22	987	BV5/987	987
5093, P0092	5093	BV4/5093	5093
P738, D4446	P738	P738 (S. pyogenes phage T12)	P738

). Sixteen combinations of replication and morphogenesis modules were identified using this approach. While many of the possible combinations of replication and morphogenesis modules present in currently sequenced phage genomes may be represented among the sixteen combinations described in this review, it is likely that as additional phage genomes are sequenced, that this number will rise considerably. The sample group of 25 phages highlighted the high level of diversity of the currently sequenced phage genomes and the combinations of replication and morphogenesis modules. Among the Moineauviruses, seven combinations of the structural and replication modules were observed with all except one module being unique to the Moineauviruses (Sfi19 and Sfi21-like; See Table 1). As mentioned above, certain Brussowviruses possess shared genetic content with phages of the 5093 and 987 groups, and this is reflected in the diversity of combinations observed among the Brussowvirus, 5093, and 987 phage genomes (Table 1). Finally, the P738 group appears to be unique among dairy streptococcal phages and likely evolved separately from all other dairy streptococcal phage groups.

Currently available sequence data are heavily biased in favor of the Moineauvirus and Brussowvirus members as these have been studied at the sequence level since the 1990s, being in stark contrast to the more recently identified phage groups. However, based on current data, two dominant Moineauvirus (MV)-specific morphogenesis modules exist (MV1 and MV2). These MV-specific modules appear to recombine preferentially with replication modules of other Moineauvirus phage genomes and, to a lesser extent, with certain Brussowvirus members (Table 1). In contrast, there is a greater prevalence of Brussowvirus-related replication modules among members of four of the five phage groups, indicating that Brussowvirus members contribute significantly to the overall evolution of dairy streptococcal phages possibly through recombination with temperate members.

3. Interactions of Dairy Streptococcal Phages and Their Hosts

3.1. Phage Adhesion Device

The interactions of dairy streptococcal phages with their hosts are known to be multi-factorial, encompassing carbohydrate binding domains that are present in structural proteins associated with the distal tail structure which is termed the adhesion device [23,24,25]. The adhesion device incorporates (part of) the tail tape measure protein (TMP), the distal tail protein (Dit), the tail-associated lysin (Tal) and a receptor binding protein (RBP) in the case of the 5093 and 987 phages. In addition, phages belonging to the Moineauvirus, Brussowvirus, and P738 groups possess an additional and fully conserved structural protein (referred to as Bpp or Tal+1). While the role of variable domains within the so-called Tal protein has long since been established [23], additional structural proteins have also been implicated in phage binding by model phages, such as the Moineauvirus DT1 [24]. The variable regions (VR) of Tal have been implicated in the recognition and binding to the host cell of the Moineauvirus DT1, and VR2 sequences have been observed to largely correlate with host strain irrespective of phage group, i.e., Moineauvirus or Brussowvirus; for example, it was observed that phages with distinct host ranges may also group, suggesting that other structural components are additionally involved in the binding process [19]. Furthermore, the Dit of many analyzed Moineauvirus and Brussowvirus members are “evolved”, implying that they contain a carbohydrate binding domain, while the 987 and 5093 phage members harbor classical Dits, being devoid of such domains. For a detailed review of the domain architecture of the adhesion device proteins of dairy streptococcal phages, see [25]. Structural homology analysis tools are essential for accurate functional annotations of proteins, particularly in the absence of sequence homology. Here, we analyzed the Bpp amino acid sequences of 22 representatives of Moineauvirus, Brussowvirus and P738 phages, as well as the RBP of representative 5093 and 987 group phages using the structural prediction software HHpred [26].

The RBP of 9871 (987 group) encompasses an N-terminal domain with structural similarity to the upper baseplate protein (BppU) of the lactococcal P335 phage TP901–1 (4V96_AQ with 99.8% probability) and with an extended C-terminal region that has previously been reported to bear similarity to the Bacillus phage Phi29 RBP [25]. This protein architecture is reminiscent of the lactococcal P335 phage RBP subgroup II that appear to have a fused “upper baseplate” and receptor binding protein [27]. The 987 phages recognize an exopolysaccharide (EPS) receptor on the host cell surface [28]. Not all S. thermophilus strains produce EPS, and the loci associated with their biosynthesis are diverse [19,29], two key factors that underpin the specificity of the 987 phages. The 5093 phage RBP has previously been reported to exhibit an esterase-like domain at the C-terminal end of the protein (Figure 5) [25]. Several saccharidic moieties in the Gram-positive cell wall (EPS, CWPS, and teichoic acids) may incorporate (phosphodi)ester-linked components; thus it is plausible that the esterase domain is functional with respect to one or more of these cell wall polysaccharides [30,31]. All analyzed Bpps encoded by Moineauvirus and Brussowvirus members harbor RBP domains with structural relatedness to those of the lactococcal phage TP901–1 (PDB 4I0S_A), Listeria phage PSA (6R5W_C), and/or Staphylococcus phage phi80 alpha (C), with at least 97% probability (Figure 5). Therefore, it is assumed that the Bpp of these phages performs the RBP function and acts as the primary determinant of host-recognition specificity, an activity that is enhanced and supported by host-binding activities of carbohydrate binding domains present in Dit and/or Tal proteins, where appropriate. This proposal is supported by the observation of higher binding affinity/fluorescent labelling of the Bpp carbohydrate binding domain compared to that of Tal host cells of the Moineauvirus STP1 [24]. For detailed reading on the structural proteins involved in phage–host interactions, please see [19,25].

3.2. RBP Phylogeny Is Not Necessarily Linked to That of the Overall Morphogenesis Module

To develop an understanding of the diversity of binding modules within the encoded RBPs or Bpps of the 25 representative phages and to determine if they evolved with the overall morphogenesis modules, an alignment of the identified binding domains was performed. Multiple sequence alignments of the deduced amino acid sequences of the RBP/Bpp carbohydrate binding domains, whose boundaries were identified using HHpred analysis described in the previous section and typically representing the C-termini of the respective proteins, were performed using ClustalW software. The alignment was employed to generate an unrooted phylogenetic tree using the “iTOL” software (http://itol.embl.de/) (accessed on 26 April 2021), applying the neighbor-joining method. This analysis suggests that the RBPs have not necessarily co-evolved with the overall morphogenesis modules, as the RBP/Bpp host binding domain phylogenetic groupings do not necessarily correlate with the overall morphogenesis module phylogroups presented in Table 1 (Figure 6). The identified phylogenetic groups show that there is considerable overlap between RBP binding domains encoded by members of the Moineauvirus and Brussowvirus groups, supporting the role of this domain in host specificity rather than being linked with the genetic lineage of the phage. It also highlights the genetic plasticity of this genomic region and is congruent with the observation of evolved structural proteins, including Dit and Tal proteins, among dairy streptococcal phages [14,25]. Such genomic plasticity facilitates the incorporation of multiple or alternative carbohydrate (or receptor) binding domains, thereby enhancing the binding affinity/avidity, or altering/expanding the host range of the phage.

The modular nature of phage RBPs and module shuffling within these proteins has been demonstrated in the development of a lactococcal phage TP901–1 chimera harboring the RBP head domain of the Skunavirus p2 [32]. Therefore, this shuffling appears to be an evolutionary (universal) strategy among phages to allow rapid adaptation to available hosts in their environment to ensure their continued success [33]. Further evidence of genomic plasticity and modular shuffling is provided by the P738 phages (P738 and D4446) [14]. The P738 phage genomes bear little sequence similarity to other dairy streptococcal phages with the exception of two genes within the morphogenesis module that encode structural proteins associated with host interactions [14]. Furthermore, these phages harbor a Bpp-like protein that possesses the same RBP domain observed among the assessed Moineauvirus and Brussowvirus phages (Figure 5 and Figure 6). Therefore, dairy streptococcal phage adhesion appears to require multiple carbohydrate-binding proteins, of which the Bpp of Moineauvirus, Brussowvirus and P738 groups apparently represents a bona fide RBP (Figure 5), which in some cases is supported by carbohydrate-binding activities of Tal- and Dit-associated carbohydrate binding activities [25].

4. Conclusions

The ever-increasing number of available genome sequences of S. thermophilus phages has been transformative to our understanding of the diversity and interactions of these phages with their hosts. The observation of hybrid phage genomes and extensive genome plasticity appears to be widespread among phages of this bacterial species. While five major groups of dairy streptococcal phages are currently identified, the extent of module shuffling highlighted in this review underscores the need for consistent monitoring of dairy fermentation facilities, as they represent evolutionary melting pots that facilitate the emergence of new phages. Moineauvirus and Brussowvirus undoubtedly remain the most persistent threat to thermophilic dairy fermentations; however, the repeated presence of 987 and 5093 phage members in recent studies [11,12,13,16,20] suggests that recently discovered phage groups may become more problematic in coming decades if the corresponding hosts are applied with increased intensity. It also seems likely that additional “hybrid” phage groups will emerge in the years to come, further endorsing the need for continuous monitoring of phage populations in industrial fermentations.

Recent studies aimed at identifying the receptor for dairy streptococcal phages have identified saccharidic moieties including rhamnose-glucose polysaccharides and exopolysaccharides as the primary receptor for several phages [18,28,34]. However, significant knowledge gaps remain in this area, and an improved functional examination of the primary determinants of phage binding will enhance our understanding of the key determinants of phage evolution in the coming decade, just as knowledge of the CRISPR-Cas (and phage-encoded anti-CRISPR) systems has informed and guided our understanding of dairy streptococcal phage evolution over the past decade [2,4,5,6,35,36,37]. Bioinformatic tools, such as HHpred, combined with increasingly populated protein structure and genome sequence databases, will continue to strengthen the predictive abilities and interpretations and will facilitate improved protein annotations. In the present review, we identified apparent module shuffling events between dairy streptococcal phage species in terms of their replication and morphogenesis modules; however, there do not appear to be limitations or obvious recombination “hotspots” within these modules apart, perhaps, from the adhesion device module. Within this region of the morphogenesis module, phages appear to acquire carbohydrate binding or host-specifying modules to enhance their adaptive potential to available host(s). In this review, the role of members of Brussowvirus species, many of which are temperate (pro)phages has been proposed as an influencer and source of dairy streptococcal phage evolution. Increasing numbers of complete streptococcal genomes are becoming available each year and with sufficient data pertaining to the prophages of this species, the mechanism that underpins this evolutionary process may become evident.