Novel Genus of Phages Infecting Streptococcus thermophilus: Genomic and Morphological Characterization

Despite decades of research and adapted antiphage strategies such as CRISPR-Cas systems, virulent phages are still a persistent risk for the milk fermentation industry worldwide, as they can cause manufacturing failures and alter product quality. Phages P738 and D4446 are novel virulent phages that infect the food-grade Gram-positive bacterial species Streptococcus thermophilus. These two related viruses represent a fifth group of S. thermophilus phages, as they are significantly distinct from other known S. thermophilus phages. Both phages share similarities with phages infecting nondairy streptococci, suggesting their recent emergence and probable coexistence in dairy environments. These findings highlight the necessity of phage surveillance programs as the phage population evolves in response to the application of antiphage strategies.

ABSTRACT

Streptococcus thermophilus is a lactic acid bacterium commonly used for the manufacture of yogurt and specialty cheeses. Virulent phages represent a major risk for milk fermentation processes worldwide, as they can inactivate the added starter bacterial cells, leading to low-quality fermented dairy products. To date, four genetically distinct groups of phages infecting S. thermophilus have been described. Here, we describe a fifth group. Phages P738 and D4446 are virulent siphophages that infect a few industrial strains of S. thermophilus. The genomes of phages P738 and D4446 were sequenced and found to contain 34,037 and 33,656 bp as well as 48 and 46 open reading frames, respectively. Comparative genomic analyses revealed that the two phages are closely related to each other but display very limited similarities to other S. thermophilus phages. In fact, these two novel S. thermophilus phages share similarities with streptococcal phages of nondairy origin, suggesting that they emerged recently in the dairy environment.

IMPORTANCE Despite decades of research and adapted antiphage strategies such as CRISPR-Cas systems, virulent phages are still a persistent risk for the milk fermentation industry worldwide, as they can cause manufacturing failures and alter product quality. Phages P738 and D4446 are novel virulent phages that infect the food-grade Gram-positive bacterial species Streptococcus thermophilus. These two related viruses represent a fifth group of S. thermophilus phages, as they are significantly distinct from other known S. thermophilus phages. Both phages share similarities with phages infecting nondairy streptococci, suggesting their recent emergence and probable coexistence in dairy environments. These findings highlight the necessity of phage surveillance programs as the phage population evolves in response to the application of antiphage strategies.

INTRODUCTION

As the most abundant and diverse biological entities, phages adapt and thrive in a vast array of ecological niches (1). The dairy industry is no exception, and the emergence of new phages represents an important threat for milk fermentation processes (2). Streptococcus thermophilus is a lactic acid bacterium widely used as a starter culture for yogurt and cheese production. Infection of this bacterium by strictly lytic (virulent) phages is the main factor impairing milk fermentation, thereby reducing factory productivity and product quality. All S. thermophilus phages characterized so far belong to the Siphoviridae family of the Caudovirales order, which are phages with double-stranded DNA (dsDNA) genome and noncontractile tails (3, 4). Historically, S. thermophilus phages have been classified into two main genera based on their mode of DNA packaging and the composition of their structural proteins (5). These groups include the so-called cos-type phages (Sfi21dt1virus) with cohesive genome extremities and the pac-type phages (Sfi11virus) that package their DNA via a headful mechanism (5). In addition, two new groups of phages infecting S. thermophilus have recently emerged, the 5093 phage group, which shares homologies with nondairy streptococcal prophages (6), and the 987 group (7). Interestingly, the 987 group is likely the result of recombination events between phages infecting S. thermophilus and Lactococcus lactis, another lactic acid bacterium widely used by the dairy industry to make cheeses (7). The emergence of the 5093- and 987-like phages with genes likely originating from lactococcal or nondairy streptococcal phages highlights the genomic mosaicity and diversity of S. thermophilus phages (6, 7).

Comparative genomic analyses have suggested that dairy streptococcal phages evolve via modular exchange (8–11). Such exchange between phages by homologous, illegitimate, or site-specific recombination drives genomic evolution (12–14). Diversity assessment is essential for investigating phage-host interactions and for identifying strategies to limit phage proliferation in industrial settings. Particular attention is given to understanding the host range of these phages for appropriate selection of S. thermophilus strains (15–17).

Fighting diverse and ubiquitous enemies is no easy task, and the dairy industry has developed many strategies to limit phage outbreaks. To control S. thermophilus phages, these strategies mostly rely on the natural selection of bacteriophage-insensitive mutants (BIMs), as well as the establishment of a rotation of bacterial strains that have nonoverlapping phage sensitivities (17). BIMs are spontaneous phage-resistant derivatives that have survived exposure to virulent phages (18). The selection of these mutants suggests taking advantage of different phage resistance mechanisms. For example, some of them may have mutations at the phage receptor level (19) or other host factors (20). However, many of these BIMs are obtained via their CRISPR-Cas systems (21). CRISPR-Cas is an adaptive system that acts as an immune memory, allowing recognition and cleavage of foreign nucleic acids (22). Two type II-A systems, CRISPR1 (CR1) and CRISPR3 (CR3), are usually active in S. thermophilus strains (23), with the CR1 system being found in almost all strains. The exploitation of these natural defense mechanisms led to the development of industrial strains resistant to a wide variety of phages. Some S. thermophilus strains may also be partially resistant to lytic phages due to the presence of innate defenses such as restriction-modification systems (24, 25). Superinfection exclusion (Sie) systems (26) or cryptic prophages in the bacterial genome (27) can also provide some antiviral protection.

Streptococcal phages can overcome the above resistance mechanisms through a variety of means. Nucleotide mutations or deletion as well as modular exchanges may allow a phage to circumvent the protection provided by the CRISPR-Cas system (21, 28–30). Phages can also produce anti-CRISPR proteins that block CRISPR-Cas activity (31, 32).

Here, we describe two members of a new genus of S. thermophilus phages, including their morphology, genomic organization, and phylogenetic relationships with other phages.

RESULTS

Phage isolation and plaque size.

Phages P738 and D4446 were isolated independently. Phage D4446 was isolated from a dairy sample from Senegal using S. thermophilus strain DGCC7891 as the host. Phage P738 was similarly isolated in Germany using the S. thermophilus host strain S4. Both S. thermophilus strains are sensitive to both phages. These phages were propagated at 42°C on their original host strain and produced different plaque sizes depending on their host. Phage P738 produced small turbid plaques (diameter, <1 mm) on strain S4 at 42°C, but when the incubation temperature was lowered to 30°C, larger plaques were observed (diameter, ≥3 mm). Phage D4446 exhibited medium-sized plaques (1 to 3 mm) on strain DGCC7891 at both temperatures. The plaque size phenotype was host-dependent, as phage P738 exhibited medium-size plaques on DGCC7891 at 30°C and 42°C, while phage D4446 produced smaller plaques on S4 at 42°C. It is unclear at this time what the reasons are for the observed temperature effect on phage plaque development, but it is likely related to the host biology and growth parameters affecting the phage lytic cycle.

Phage morphology.

A sample of each CsCl-purified phage was analyzed using transmission electron microscopy (TEM), showing that both phages belong to the Siphoviridae family (Fig. 1). They have similar capsid diameters, estimated at 56 nm connected to a noncontractile tail of 124 nm long (Fig. 1). The tail length is similar to that of S. thermophilus phages of the 987 type, which also have a shorter tail (120 to 150 nm) (8) compared to the longer tails (200 to 250 nm) found in members of the three other S. thermophilus phage groups (5, 6). Of note, phages P738 and D4446 also possess unique twisted tail fibers (length, 48 to 50 nm) composed of 2 or 3 subfibers.

FIG 1.

TEM micrographs of CsCl-purified P738 (left) and D4446 (right) phage particles stained with uranyl acetate (bar indicates 100 nm) with general characteristics for each phage. The arrows indicate the distal ends of the thin (twisted) tail fibers.

Genome organization and annotation.

We sequenced the double-stranded DNA (dsDNA) genomes of S. thermophilus phages P738 and D4446 using 454 and Illumina technology, respectively. The genomes are composed of 34,037 bp and 33,656 bp for P738 and D4446, respectively (Table 1). The average G+C content of the genomes was 37.2% for P738 and 37.0% for D4446, similar to other S. thermophilus phages (38.3 ± 1.1%, 100 NCBI genomes) and similar to the host species S. thermophilus (38.6 ± 1.2%). Annotation of these two genomes revealed a modular organization, as is observed with most dsDNA phages. Totals of 48 and 46 open reading frames (ORFs), preceded by a putative ribosome-binding sequence, were identified in phages P738 and D4446, respectively. No gene coding for an integrase was found in these genomes, confirming their lytic lifestyle, and no tRNA gene was detected. Based on BLASTp analysis, functions could be assigned for 21/48 (phage P738) and 18/46 (phage D4446) of the deduced phage proteins (see Tables S1 and S2 in the supplemental material). We noticed the presence of six ORFans (i.e., ORFs for which no homolog is found in public databases) in phage P738 and four in phage D4446. The four ORFans present in the D4446 genome are also found in P738.

TABLE 1.

Comparison of S. thermophilus phages P738 and D4446 and S. pyogenes prophage T12^a

P738			D4446			nt ID (%)	aa ID (%)	Predicted function	S. pyogenes phage T12 BLASTp analysis with P738 (aa ID) (%)
Gene no.	Gene length (bp)	No. of aa	Gene no.	Gene length (bp)	No. of aa
1	480	159	1	480	159	97	99	Terminase small subunit
2	1,278	425	2	1,278	425	93	99	Terminase large subunit	73
3	1,293	430	3	1,293	430	93	97	Portal protein	64
4	816	271	4	816	271	94	96	Capsid protein
5	165	54	5	165	54	88	94	Hypothetical
6	588	195	6	588	195	93	95	Scaffold protein	45
7	909	302	7	903	300	91	94	Capsid protein	59
8	153	50	8	153	50	93	94	Hypothetical
9	327	108	9	327	108	91	95	Hypothetical
10	312	103	10	312	103	97	99	Hypothetical
11	351	116	11	351	116	97	97	Hypothetical
12	411	136	12	411	136	96	99	Hypothetical	63
13	498	165	13	501	166	90	92	Major tail protein	72
14	315	104	14	315	104	97	98	Hypothetical	54
15	294	97	15	294	97	96	97	Hypothetical	57
16	2,217	738	16	2,217	738	93	98	Tail tape measure protein	52
17	2,322	773	17	2,322	773	94	98	Distal tail protein	39
18	3,972	1,323	18	3,972	1,323	93	97	N-Acetylmuramoyl-l-alanine amidase	42
19	2,001	666	19	2,001	666	93	96	Structural protein	40
20	273	90	20	258	85	96	88	Hypothetical
21	351	116	21	351	116	96	97	Hypothetical
22	192	63	22	192	63	96	98	Hypothetical
23	747	248	23	747	248	96	99	Peptidoglycan hydrolase	29
24	267	88	24	282	95	90	77	Hypothetical
25	126	41	25	126	41	93	90	Hypothetical
26	228	75	26	228	75	93	99	XRE family transcriptional regulator
27	363	120	27	363	120	83	83	Hypothetical
28	567	188						DUF1642 domain-containing protein
29	522	173						DUF1642 domain-containing protein
			28	546	181			Hypothetical
30	180	59						Hypothetical
31	147	48	29	147	48	95	91	Holliday junction resolvase RusA
32	168	55						Hypothetical
			30	162	53			Hypothetical
33	342	113	31	342	113	94	97	Hypothetical
34	228	75	32	228	75	93	95	Hypothetical
35	465	154	33	465	154	95	98	Hypothetical
36	1,188	395	34	1,188	395	91	95	Type III restriction endonuclease subunit
37	648	215	35	621	206	94	99	Recombinase
38	423	140	36	423	140	80	70	No hit
39	339	112	37	339	112	95	98	Hypothetical
40	825	274	38	825	274	93	98	Primase
41	1,350	449	39	1,350	449	92	98	Helicase
42	291	96	40	288	95	85	81	Hypothetical
43	174	57	41	174	57	85	88	No hit
44	444	147	42	453	150	93	93	Hypothetical
45	351	116	43	321	106	95	95	Hypothetical
46	279	92	44	279	92	95	99	No hit
47	339	112	45	339	112	95	98	No hit
48	1,170	389	46	1,170	389	95	97	Hypothetical

^aaa, amino acids; nt ID, nucleotide identity; aa ID, amino acid identity.

Comparative genomic analysis.

Genome sequencing revealed that, despite their independent isolation from distinct geographical areas, these new phages are very similar. The P738 and D4446 genomes share 94% nucleotide identity over 96% of their length (Fig. 2). BLASTn analysis of the complete genomic nucleotide sequences revealed that the most closely related phage was Streptococcus pyogenes phage T12 (GenBank accession no. KM289195), with 67% identity across only 15% of the genome (Fig. 2). Thirteen ORFs shared amino acid identity, and 8 of the 13 ORFs shared at least 50% identity with those of phage T12. These ORFs encode a terminase large subunit (ORF2), portal (ORF3), capsid (ORF7), tail protein (ORF13), tail tape measure protein (ORF16), and three hypothetical proteins (ORF12, ORF14, and ORF15). The amino acid identities between the ORFs of phages P738 and T12 are shown in Table 1.

FIG 2.

Pairwise BLASTn comparison of the two new S. thermophilus phages P738 and D4446 and S. pyogenes prophage T12. Analysis shows high relatedness between phages P738 and D4446 and homology with prophage T12. Genes identified by bioinformatics analysis as belonging to specific modules are color-coded. Selected genes are annotated in the figure, encoding conserved proteins. TerS, terminase small subunit; TerL, terminase large subunit; TMP, tail tape measure protein; Dit, distal tail protein; Tal, tail-associated lysin.

The sequences coding for ORF1 to ORF19 of phages P738 and D4446, which correspond to structural proteins, shared 94% to 99% nucleotide identity. Only two ORFs, ORF17 and ORF18, showed significant similarity between the deduced phage proteins and those found in other S. thermophilus phages, and they are likely involved in host recognition (see below). We observed a higher degree of variation between the rest of the ORFs, namely, ORF20 to ORF48_P738/ORF46_D4446. The most variable genomic region is located between ORF27 and ORF33 (Fig. 2), where there is more similarity with phages infecting nondairy streptococci than with dairy phages.

Tail proteins and host recognition.

The ORF16 of both phages was annotated as the tail tape measure protein (TMP) based on several observations: (i) the genomic position of orf16, located next to genes coding for phage tail proteins, (ii) its similarity (based on BLASTp analysis) to proteins described as TMP, (iii) the presence of six trans-membrane regions (based on single-molecule real-time [SMART] analysis) (33), often observed in TMPs (34), and (iv) the correlation between tail length and its number of amino acids (35). ORF17 of phage P738 (ORF17_P738) shares homology with a putative distal tail protein (Dit) (Tables S1 and S2). It could be aligned with homologs found in other S. thermophilus phage groups, sharing 22.9% amino acid identity with ORF51_Sfi11, 22.7% identity with ORF19₂₉₇₂, and 18% identity with ORF17_DT1 and similar low identity with corresponding ORFs of other phages (Fig. 3). Similarities mostly occur at the N and C termini of the homologs, particularly with ORF19 of phage 2972 (36). ORF19₂₉₇₂ is also known as a hybrid structural protein, with a central section conserved between cos and pac phages (36). HHpred analysis revealed that the Dit-encoding genes of P738 and D4446 phages harbor two insertions of carbohydrate-binding domains (CBDs) (Fig. 4A), a feature observed in Lactobacillus phage J-1 (34). One of the CBDs shares structural similarity with the CBD of the evolved Dit of Lactobacillus phage J-1 (37). Evolved Dit has also been observed in S. thermophilus cos group phages (8) and in a large number of Lactococcus lactis phages of the Skunavirus genus (38, 39), with only one insertion. This new configuration of Dit also differs from the one of 5093 and 987 phages (7), which does not contain any CBDs.

FIG 3.

Schematic illustration representing the alignment of the S. thermophilus phage P738 genome with representative ORFs of other S. thermophilus phages. VR1, VR2, and VR3 indicate variable regions 1, 2, and 3, respectively. TerS, terminase small subunit; TerL, terminase large subunit; TMP, tail tape measure protein; Dit, distal tail protein; Tal, tail-associated lysin.

FIG 4.

Full-sequence HHpred analysis of the insertions of evolved Dit (A) and Tal (B) proteins of phages D4446 and P738. The HHpred output in red means that the HHpred probability is over 98%. The label indicates which Protein Data Bank (PDB) entry is closest to the query.

ORF18_P738, the largest deduced phage protein, shows similarity with two domains involved in peptidoglycan hydrolysis, namely, LYZ2 (lysozyme; E value, 1.2e–13; amino acid positions 264 to 401) and CHAP (cysteine, histidine-dependent amidohydrolase/peptidase; E value, 1.1e–9; amino acid positions 456 to 594) (40). These domains are often associated with tail-associated lysins (Tal) (41, 42), which usually follow the Dit-encoding gene, and have been previously observed in other S. thermophilus phages (8). Tal forms a structural trimer with a conical shape, as observed in phage p2 ORF16 (43). In many phages, this structural trimer extends through a fiber-shaped structure harboring hydrolases (phage TP901-1) (39) or antireceptor domains, such as in S. thermophilus DT1 phage VR1 and VR2 domains (36). HHpred analysis of ORF18_P738 and ORF18_D4446 reveals a striking feature (Fig. 4B); the ORF18 (or Tal) N-terminal structural domain harbors two insertions, with the fold of a hydrolase (CBM1) and a lysin (CBM2). The N terminus is followed by an extension harboring a unique carbohydrate-binding domain (CBM3) with the fold of the BppA domain of phage Tuc2009 (44), followed by a putative cell adhesion domain at the N terminus (CBM4) (Fig. 5).

FIG 5.

Schematic depiction of the identical architecture of the distal tail and Tal regions of phages D4446 and P738. (based on HHpred analysis). The Dit structural ring and galectin domains (top) are colored green. The Dit CBMs 1 and 2 are colored yellow and purple, respectively. The Tal structural parts are colored brown/beige. The Tal N-terminal CBM insertions 1 and 2 are colored orange and red, respectively, and those on the Tal extension (CBMs 3 and 4) are colored blue and beige, respectively.

In addition, the size and function of ORF18 suggest that it forms the straight tail fiber, which paves the way for the TMP to pass through the peptidoglycan layers of the host bacterium. ORF18_P738 shows similarities to host-specificity proteins found in other S. thermophilus phages (45). Indeed, we could align ORF18_P738 with homologs in phages CHPC929, Sfi11, SW13, Sfi21, and 2972 (Fig. 3), highlighting the presence of VR1 and VR2 regions. HHpred analysis of ORF20 revealed that its C terminus harbors a phage TP901-1 RBP-like domain (see Addendum in Proof).

Proteomic analysis.

We next identified the major components of the viral structure of phage D4446 using liquid chromatography/tandem mass spectrometry (LC-MS/MS) to make the link between the bioinformatic predictions and phage structural proteins. A total of 21 proteins were detected using CsCl purified phages (Table 2). These proteins included the small (ORF1) and large (ORF2) terminase subunits, as well as the phage structural proteins—portal protein (ORF3), capsid proteins (ORF4 and ORF7), tail protein (ORF13), tail tape measure protein (ORF16), distal protein (ORF17), and Tal and tail fiber proteins (ORF18 and ORF19, respectively). The phage endolysin (ORF23), recombinase (ORF35), and helicase (ORF39) were also detected. Analysis also revealed eight phage proteins (ORF9, ORF10, ORF11, ORF12, ORF14, ORF21, ORF34, and ORF46) of unknown function. It is still possible that some of these proteins are not found in the complete virion particles but were inadvertently copurified.

TABLE 2.

Identified peptides for phage D4446 by liquid chromatography/tandem mass spectrometry (LC-MS/MS) and predicted functions

ORF	Identified proteins (predicted function)	Theoretical mol wt (kDa)	Exclusive unique peptide count
1	Terminase small subunit	19	10
2	Terminase large subunit	49	19
3	Portal protein	49	25
4	Capsid protein	31	18
7	Major capsid protein	32	33
9	ORF9	13	7
10	ORF10	12	11
11	ORF11	13	7
12	ORF12	16	8
13	Major tail protein	18	3
14	ORF14	12	2
16	Tail tape measure protein	77	34
17	Distal tail protein	87	36
18	Tail-associated lysin	148	75
19	Fiber protein	73	25
21	ORF21	13	3
23	Peptidoglycan hydrolase	27	5
34	ORF34	46	20
35	Recombinase Erf	23	3
39	Helicase	52	3
46	ORF46	45	21

P738 and D4446 belong to a new genus (P738) of streptococcal phages.

As comparative genomic analysis did not reveal additional similarities with other S. thermophilus phages, beside the tail regions implicated in host specificity described above, we sought to better define the taxonomic position of S. thermophilus phages P738 and D4446. We examined proteomes of 100 S. thermophilus phages from the ones available in the National Center for Biotechnology Information (NCBI) database (Fig. 6). The previously described four groups of S. thermophilus phages were clustered with bootstrap values over 98. We observed a significant separation between the two new phages described in this study and the other S. thermophilus phages, confirming that they belong to a newly emerging phage group infecting this dairy bacterium. We propose to name this new genus after the newly isolated phage P738. This new lineage is supported by a bootstrap value of 97. A higher level of amino acid sequence identity was observed between these newly emerging phages and other streptococcal phages, for example the S. pyogenes prophage T12, which was included to root the phylogenetic tree. This observation suggests that these two novel S. thermophilus phages may have originated, at least partially, from a phage infecting nondairy streptococci.

FIG 6.

Proteome-based phylogenetic tree using the Genome-BLAST Distance Phylogeny (GBDP) method with VICTOR under settings recommended for prokaryotic viruses. S. thermophilus phages are colored according to their genus, and S. pyogenes phage T12 was included to root the tree.

Host strain features.

We observed differences in plaque size when propagating phages P738 and D4446 on strains S4 and DGCC7891. In addition, the efficiencies of plaquing of both phages were lower on strain S4 when propagated on DGCC7891 but were restored when plaques were recovered and amplified again on strain S4, likely underlying the presence of a host modification system such as restriction-modification (46). Isolation of these two new phages was made possible by using two host strains, and we investigated their relatedness by comparing their CRISPR arrays and those of other S. thermophilus strains, as the S4 and DGCC7891 genomes are not available. We amplified and sequenced the CRISPR arrays of the two active type II-A CRISPR-Cas systems (CR1 and CR3) of both host strains S4 and DGCC7891. We could not amplify by PCR the CR3 locus of strain S4 using our set of primers. The absence of CR3 in some S. thermophilus strains has been noted previously (21).

Then, we compared the composition of the spacers with the other strains (Fig. 7) (47). For consistency in locus comparison, we used the amplicons generated in silico. The CR1 and CR3 arrays of DGCC7891 were identical to those of S. thermophilus ND03 (48). Strains DGCC7891 and ND03 both also clustered with strains APC151 and KLDS3.1012 (Fig. 7A), as their CR1 loci are identical and only a few variations are present in their CR3 loci. Conversely, S. thermophilus S4 appears to be atypical with a short CR1 array composed of six unique spacers (Fig. 7B). Sequences of the spacers in the CRISPR1 of strain S4 are available in Table S3. Best matches revealed that this strain had acquired spacers from three S. thermophilus phage groups but not from 987 phages. Moreover, spacers of the two host strains did not match the P738 or D4446 genomes, suggesting that these phages only started interacting recently with S. thermophilus strains DGCC7891 and S4.

FIG 7.

CRISPR array comparison. (A) A set of 35 S. thermophilus strains are displayed in clusters of similar arrays. CRISPR 1 and CRISPR 3 are shown from right to left. The color-coding distinguishes the different groups after clustering, with the array clustering method implemented in CRISPRStudio. (B) Unique spacers present only once across the data set are colored, and spacers present at least twice are grayed out to show the S4 array’s originality.

Given the fact that the two virulent phages showed similarity with prophages integrated in the genome of other strains of the Streptococcus genus and that they formed a new taxonomic group, distant from the other S. thermophilus phages, we investigated whether their optimal host could actually be another Streptococcus. We used WIsH (49), a bioinformatic tool that predicts the bacterial host of phages based on genomic sequences, to compare the best host among a set of S. iniae, S. pyogenes, S. agalactiae and S. parauberis strains, but the analysis did not permit us to robustly predict potential additional hosts for the two newly isolated phages.

In addition, the host range of these new phages was determined using a panel of 154 S. thermophilus strains. This host range revealed 14 additional strains (isolated from the dairy environment) sensitive to both phages. The results with a subset of 30 strains, including the two host strains, 14 new sensitive hosts, and 14 other strains (insensitive) that have been characterized in the literature, are presented in Table 3.

TABLE 3.

Host-range of S. thermophilus phages P738 and D4446^a

Strain	Origin/reference	P738	D4446
S4	Dairy product isolate, Senegal	+ (host for propagation)	+
DGCC7891	Dairy product isolate, Germany	+	+ (host for propagation)
SC16-1	Starter culture strain	+	+
SC16-2	Starter culture strain	+	+
SC16-4	Starter culture strain	+	+
SC18-6	Starter culture strain	+	+
SC19-0	Starter culture strain	+	+
SC19-1	Starter culture strain	+	+
SC19-7	Starter culture strain	+	+
SC19-9	Starter culture strain	+	+
SC20-1	Starter culture strain	+	+
SC20-4	Starter culture strain	+	+
SC20-5	Starter culture strain	+	+
SC24-4	Starter culture strain	+	+
MBT-OM1	Dairy product isolate, Germany	+	+
MBT-CN1	Dairy product isolate, Germany	+	+
LMG18311	68	–	–
CNRZ1066	69	–	–
DGCC7854	32	–	–
DGCC782	20	–	–
DGCC7796	20	–	–
LMD-9	70	–	–
DGCC7710	21	–	–
SMQ-301	71	–	–
UY01	72	–	–
UY02	72	–	–
UY03	72	–	–
SMQ-308	73	–	–
SMQ-312	73	–	–
SMQ-316	73	–	–

^a+, clear lysis; –, absence of lysis.

DISCUSSION

Here, we show that the virulent phages P738 and D4446 are novel among S. thermophilus phages, with little relatedness to other known phages infecting strains of this species. The high sequence similarity between these two phages prompts us to define a fifth group of S. thermophilus phages. The prophage T12 of S. pyogenes appeared to be related to these two new S. thermophilus phages, suggesting a common ancestor. This putative host shift has also been mentioned after isolation of 5093-like phages, which appeared to have originated from extensive interspecies horizontal gene transfer events and intraspecies genetic recombination events with phages that infect both dairy and nondairy streptococci (6).

These newly isolated phages share host recognition proteins with other S. thermophilus phages, likely relevant to the capacity of infection of strains of the species. Genetic mutations during viral replication can also generate viral variants that acquire the ability to infect new hosts (50). It has been demonstrated that the abundance of potential hosts favors shifts in viral infectivity (51). In the murine gut ecosystem, which offers access to new hosts, the microbiota shifts the genetic diversity of phages, thereby promoting long-term persistence of phage populations.

Diverse species are described among the genus Streptococcus, and most of them are commensal or pathogenic in humans and animals (52). S. thermophilus is the only species of this genus to be widely used as a starter culture in the dairy industry and to have the “generally recognized as safe” status. It has colonized the milk niche, but its environmental reservoir has not been identified (53). Despite gene loss and adaptation to its niche, S. thermophilus is capable of establishing biofilms, vestiges of a commensal lifestyle (54), where horizontal and vertical phage evolution may have happened in an ancestor that probably infected other species. Interestingly, the genomes of P738 and D4446 did not match any spacer from the CRISPRFinder database (55), nor the metagenome spacer database (56), supporting the hypothesis of a very recent emergence.

It was previously reported that evolution among S. thermophilus phages is driven by modular exchanges (7–9, 13, 14), but limited shared gene modules with other S. thermophilus phages were revealed in the comparative genomic analysis of phages P738 and D4446. Such module exchange events can occur during phage coinfection or infection of a lysogenic strain (which are rare within the S. thermophilus species) or possibly after internalization of extracellular DNA through natural competence, and the newly described phages may not have recombined extensively yet with other S. thermophilus phages to adapt to their new host.

Our study also highlights the expanding genomic diversity of S. thermophilus phages as well as the importance of phage monitoring to minimize the phage risk in milk fermentation processes and to adapt control strategies.

MATERIALS AND METHODS

Phages, bacteria, and growth conditions and purification.

S. thermophilus DGCC7891 and S4 were cultivated in M17 medium (Oxoid) supplemented with 0.5% (wt/vol) lactose and 10 mM CaCl₂ for phage amplification. Cultures were incubated at 37°C without shaking to generate an overnight culture for use the following day. In all other cases, cultures were incubated at 42°C. For phage amplification, the appropriate host strain was inoculated with a scraping from a phage lysate preserved at –80°C with 15% (vol/vol) glycerol and grown until complete lysis was observed. The lysate containing phage D4446 was then filtered through a 0.45-μm polyethersulfone (PES) filter, and 100 μl was used to inoculate a bacterial culture grown to an optical density at 600 nm (OD₆₀₀) of 0.1. The resulting culture was grown until complete lysis was observed, and this second phage amplification lysate was also filtered through a 0.45-μm PES filter and stored at 4°C. To obtain high-titer phage D4446 lysate, 1 liter of phage lysate was concentrated with 10% polyethylene glycol (PEG) 8000 and 0.5 M NaCl (Laboratoire Mat) and purified on a discontinuous CsCl (Fisher Scientific) gradient followed by a continuous CsCl gradient as described previously (57). Purified phages were recovered by ultracentrifugation using a Beckman SW41 Ti rotor at 35,000 rpm (210,053 × g) for 3 h, followed by a second ultracentrifugation using a Beckman NVT65 rotor at 60,000 rpm (342,317 × g) for 18 h. The phage preparation was then dialyzed against phage buffer (50 mM Tris-HCl [pH 7.5], 100 mM NaCl, 8 mM MgSO₄). The lysate containing phage P738 was incubated with 3 μl/ml each of DNase I and RNase at 37°C for 1 h. NaCl was added to a final concentration of 1 M, dissolved by swirling, and kept on ice for 1 h. Cell debris were removed by centrifugation in a JA10 rotor (Beckman, USA) at 7,000 rpm and 4°C for 10 min. The supernatant was recovered and PEG 6000 was added to a final concentration of 10% (wt/vol). After at least 1 h on ice, precipitated phage particles were recovered by centrifugation in a JA10 rotor (8,000 rpm, 4°C, 10 min). Phage particles were purified on a discontinuous CsCl gradient (57) and recovered by centrifugation in a Beckman SW32.1 Ti rotor (Beckman, USA) at 10°C and 25,000 rpm for 20 h. The phage preparation was then dialyzed against buffer (50 mM Tris-HCl [pH 7.5], 10 mM NaCl, 10 mM MgSO₄).

DNA extraction, genome sequencing, and annotation.

Phage D4446 genomic DNA was extracted from lysate (∼10⁹ plaque-forming units [PFU]/ml) using the Qiagen plasmid maxi kit (Qiagen, USA) with previously described modifications of the former Maxi lambda DNA purification kit (58). Sequencing libraries were prepared with the Nextera XT DNA library preparation kit (Illumina) according to the manufacturer’s instructions. The library was sequenced using a MiSeq reagent kit V2 (Illumina, 500 cycles) on a MiSeq system using paired ends (2 × 250 bp). Phage P738 genomic DNA was isolated from CsCl-purified phages according to a standard protocol, including phenol and chloroform/isoamyl alcohol purification steps (26). Sequencing was done by AGOWA (now LGC Genomics, Berlin, Germany) using 454 sequencing with an average 20-fold coverage.

De novo assembly was performed with Ray assembler version 3.0.0 (59). Nucleotide coverage was calculated with SAMtools (60). The sequences were analyzed using Geneious software version 11.0.5. Open reading frames were identified with GeneMark.hmm (61). A sequence was considered an ORF only if its starting codon was AUG, UUG, or GUG and it possessed at least 30 amino acids. Ribosomal binding sites (RBS) similar to the standard Shine-Dalgarno sequence were also identified. The BLASTp database was used to predict the function of each ORF (50). Theoretical molecular masses (MM) and isoelectric points (pI) of the phage proteins were obtained using ProtParam (http://web.expasy.org/protparam/). ARAGORN (62) and tRNAscan-SE version 1.21 (63) were used to search for tRNA genes. The nucleotide number 1 was set upstream of the beginning of the gene coding for the terminase small subunit.

Electron microscopy.

Phage lysates, purified on a cesium chloride gradient, were dialyzed against buffer (20 mM Tris-HCl [pH 7.2], 10 mM NaCl, 20 mM MgSO₄). Negative staining with 2% (wt/vol) uranyl acetate, transmission electron microscopy (Tecnai 10; FEI Thermo Fisher Scientific, Eindhoven, The Netherlands) at an accelerating voltage of 80 kV, and imaging with a MegaView G2 CCD camera (Emsis, Münster, Germany) were done as described before for S. thermophilus phages (8, 9).

Comparative genomic analysis.

All predicted ORFs from P738 were compared to ORFs of D4446 and S. pyogenes prophage T12 using Easyfig (64). The complete genome sequences of 100 S. thermophilus phages available in the GenBank database were also analyzed. S. pyogenes phage T12 was used to analyze the phylogenetic position of phages P738 and D4446. All nucleotide sequence pairwise comparisons were conducted using the Genome-BLAST Distance Phylogeny (GBDP) method (65) with the Virus Classification and Tree Building Online Resource (VICTOR) under settings recommended for prokaryotic viruses (66). The resulting intergenomic distances were used to infer a balanced minimum evolution tree with branch support via FastME (67). Branch support was inferred from 100 pseudobootstrap replicates each.

Analyses of phage D4446 structural proteins.

Phage structural proteins were detected directly from the purified phage preparations after in-solution tryptic digestion by liquid chromatography/tandem mass spectrometry (LC-MS/MS) at the Plateforme Protéomique, Centre de Génomique de Québec (Université Laval). A custom database was generated using the putative predicted proteins from the sequences of the phage genomes and compared with the peptides identified. These results were analyzed with Scaffold Proteome Software version 4.4.5.

S. thermophilus CRISPR loci analysis.

CRISPR loci CR1 and CR3 were amplified by PCR (Feldan Taq DNA polymerase) and then Sanger-sequenced. CRISPR1/CR1 analysis was performed with the primers yc70 (5′-TGCTGAGACAACCTAGTCTCTC-3′) and CR1-rev (5′-TAAACAGAGCCTCCCTATCC-3′), whereas primers CR3-fwd (5′-CTGAGATTAATAGTGCGATTACG-3′) and CR3-rev (5′-GCTGGATATTCGTATAACATGTC-3′) were used for the amplification of CRISPR3/CR3 (23). Spacer analyses were performed with Geneious version 11.0.5. CRISPRStudio (47) was used for CRISPR array visualization.

Host prediction using bioinformatics analysis.

WIsH version 1.0 was used for the prediction of bacterial hosts of the phages. Instructions on the github page of the software were followed (https://github.com/soedinglab/WIsH) (49). A total of 28 S. pyogenes strains carrying the T12 prophage, 32 S. pyogenes strains not carrying the T12 prophage, and 32 S. thermophilus strains were selected to predict the best host among them. The null model consisted of 25 coliphages known not to infect any of the streptococcal species. All the strains and phage genomes used for the prediction and null model are listed in the supplemental materials.

Data availability.

The complete genome sequences of phages P738 and D4446 have been deposited in GenBank under accession numbers MK911750 and MN938931, respectively.

ADDENDUM IN PROOF

A manuscript revisiting the adhesion devices of phages from Streptococcus thermophilus is in press (74).