Multilocus Characterization Scheme for Shiga Toxin-Encoding Bacteriophages

Abstract

Shiga toxin-producing Escherichia coli (STEC) strains are food-borne pathogens whose ability to produce Shiga toxin (Stx) is due to integration of Stx-encoding lambdoid bacteriophages. These Stx phages are both genetically and morphologically heterogeneous, and here we report the design and validation of a PCR-based multilocus typing scheme. PCR primer sets were designed for database variants of a range of key lambdoid bacteriophage genes and applied to control phages and 70 stx⁺ phage preparations induced from a collection of STEC isolates. The genetic diversity residing within these populations could be described, and observations were made on the heterogeneity of individual gene targets, including the unexpected predominance of short-tailed phages with a highly conserved tail spike protein gene. Purified Stx phages can be profiled using this scheme, and the lambdoid phage-borne genes in induced STEC preparations can be identified as well as those residing in the noninducible prophage complement. The ultimate goal is to enable robust and realistically applicable epidemiological studies of Stx phages and their traits. The impact of Stx phage on STEC epidemiology is currently unknown.

The emergence of Shiga toxin-producing Escherichia coli (STEC) strains as food-borne pathogens has become a worldwide public health concern. STEC infections can result in diarrheal symptoms that may develop into hemorrhagic colitis and, in severe cases, progress to hemolytic uremic syndrome or thrombotic thrombocytopenic purpura, both of which are potentially fatal complications (15) caused by Shiga toxin (Stx). The genes encoding Stx (stx genes) are located within prophage or remnant prophage sequences in all STEC strains, and the horizontal transfer of stx genes is facilitated by bacteriophages (Stx phages) (35). Over the last 5 years, bacterial genome sequencing has revealed that many bacterial pathogens possess a significant amount of prophage and remnant prophage DNA (2). STEC strains are no exception, and the two sequenced E. coli O157:H7 strains, one isolated from an outbreak in Sakai, Japan (19), and another from an outbreak in Michigan (36) carry an additional ∼1 Mbp of DNA compared to E. coli K-12 (6). Approximately 40% of the variation between the K-12 and O157:H7 strains is due to remnant and inducible prophages (14, 36).

The Stx phages are lambdoid (2), having the ability to infect a host cell and then either replicate or integrate into the bacterial genome, and they share a distinct genetic organization (37) (Fig. 1) with bacteriophage λ (8). Lambdoid bacteriophage genomes have been shown to possess high levels of mosaicism, although their genome organization and orientation remain similar (8, 11). Stx phages are intimately involved in the pathogenic profile of their bacterial lysogens, the survival and dissemination of the stx genes in the environment, and the emergence of new Stx-producing pathogens (2). A classification method for monitoring Stx phage dispersal would enable the generation of epidemiological information to address the relationship between STEC disease outbreaks and the distribution of zoonotic STEC in livestock and the farm environment. There is a wealth of information on STEC strains in this context but little on the occurrence, distribution, and identity of the bacteriophages that disseminate the stx genes and potentially provide a vehicle for their survival in the environment (2, 10, 23, 30). Traditionally, phages have been characterized by morphological characteristics according to a universal system for virus taxonomy (32, 38, 39); but morphologically similar phages may be completely unrelated at the nucleotide sequence level, and morphologically distinct phages can possess large regions of sequence identity or similarity (2). Characterization of bacteriophages can be further complicated by the high levels of recombination that occur between inducible and remnant bacteriophage genomes within a bacterial lysogen (2, 7, 24). Not only do recombination events occur frequently, they are also up-regulated by bacteriophage-encoded proteins; of these, the λ Red recombinase system is used commercially (13, 52) and has also been identified in Stx phages (2). A multilocus typing system designed to identify genetic similarity and disparity would provide the ability to compare and contrast genetic variation in inducible phages without the need to sequence entire phage genomes. The ability to type Stx phages induced from STEC isolates will enable definition of the level of heterogeneity among Stx phages and identification of specific phage genes that are disseminated across a bacterial population or even enable the discovery of phage-borne genes that are associated with enhanced pathogenicity/fitness of their bacterial hosts.

FIG. 1.

A schematic of the well-studied Stx bacteriophage 933W genome, detailing the functional genes that were identified as targets for the multilocus PCR characterization scheme. The target genes used as part of this study, from left to right, include int, cIII, N, cI, cro, cII, Q, stx, capsid and packaging genes, and tail spike/host recognition proteins.

Multilocus sequence typing is an established molecular tool for typing bacteria through the generation of an allelic profile. Here, we take a similar approach to produce a scheme based on lambdoid phage genome organization using loci that represent key modules involved in phage infection and propagation. This system makes it possible to identify inducible phage genes that are present in an STEC background and the genes that impinge upon the fluidity of the mosaic Stx phage genome. The core functional genes controlling the biology of Stx phages (int, cIII, N, cI, cro, cII, Q, stx, capsid structural genes, packaging genes, and tail spike/host recognition protein genes) (Fig. 1 and Table 1) were identified from published genomic Stx phage sequences (NC_004813, NC_000924, AF034975, NC_003525, NC_004914, NC_000902, and NC_003356), as well as lambda phage. The sequences of these target genes were subjected to BLASTN analysis (4) against sequenced STEC strains. Matches from these analyses, from both inducible and remnant Stx prophages, were then subjected to further rounds of BLASTN (4) analyses to identify all complete, phage-related genes within the genetic databases. Sequences for each gene were aligned (21, 48) and grouped into clades (27), and oligonucleotide primers capable of differentiating between variants of each target gene were designed. These primer pairs (Table 1), upon which the multilocus typing scheme is based, were validated by PCR amplification of target genes from bacteriophages possessing the specific sequence for which the primer pair was designed, as well as against phages that did not harbor the particular gene variant, to ensure specificity of the primer pair (data not shown). All amplification products from these controls were sequenced by Macrogen, Inc., Seoul, Korea, to confirm their identity. DNA amplification (30 cycles) was performed using Phusion (New England BioLabs) according to the manufacturer's instructions (1× HF buffer containing a 200 μM concentration of each deoxynucleoside triphosphate, a 0.5 μM concentration of each oligonucleotide primer, and 0.02 U of Phusion DNA polymerase). Amplification reactions involved denaturation at 98°C, annealing for 30 s (temperatures listed in Table 1), and an extension period of 1 min at 72°C. Amplification products were produced from PCR using 2 or 4 μl of each phage preparation as the template source. Failure to amplify the E. coli housekeeping gene gapA (encoding glyceraldehyde-3-phosphate dehydrogenase) from the phage preparations indicated that host genomic DNA did not contaminate the phage preparations to any significant level.

TABLE 1.

Oligonucleotide primers comprising the multilocus identification scheme

Gene target	Phage namea	Primer name	Sequence (5′ → 3′)	Gene type	Tmb	Annealing temp	Amplicon size (bp)	Reference or source
int	Φ24B	Φ24Bint 5′	GCCAGGCTTTCTGAGCTACG	24B int		55.0		16
		Φ24Bint 3′	GCCTAAAATCATGCGTTCTCC			55.0		16
	lambda, H19J,	1F	GTTACMGGGCARMGAGTHGG	1 int		50.0		5
	HKO22, P434 (group 1)	1R	ATGCCCGAGAAGAYGTTGAGC					5
	P21, e14 (group 2)	2F	GTTACTGGWCARCGKTTAGG	2 int		50.0		5
		2R	GATCATCATKRTAWCGRTCGGT					5
	ST64B (group 3)	3F	AATGGARATWKCYTATYTVTGTGC	3 int		50.0		5
		3R	TCRTARTCTGARATYCCYTTBGC					5
	P27, 933W, Fels1, BP4795, Gifsy2, EH297 (group 4)	4F	CTBGCMTGGGARGATATHGA			50.0		5
		4R	GMCCAGCABGCATARGTRTG					5
	P4, HK620, Sf6 (group 5a)	5AF	TGGRAKRAMKTCGAYTTYGA	5A int		50.0		5
		5AR	CAGTTGCMYYTCWATMGCGTCA					5
	P4, epsilon 15	5BF	TWGTKCGTWMMAGTGAATT	5B int		50.0		5
	(group 5b)	5BR	TKGWTRTATRCCGCWCGYAC					5
	Phi80, Gifsy1	5CF	GGRMARTYATAAAACKSG	5C int		50.0		5
	(group 5c)	5CR	TGCCCGAGCAKCWTYTCA					5
	P2, L413C, Wphi	6AF	CTGAGYACWGGAGSAMGWTGG	6A int		50.0		5
	(Group 6a)	6AR	CCBCCRTTMATCATRAARTG					5
	Fels2, PSP3, P186	6BF	TVGCWACYGGCGCMMGRTGG	6B int		50.0		5
	(Group 6b)	6BR	CCBCCRTTMATCATRAARTG					5
	P22, SfV, SfII,	7F	AACATYATMAAYCTKGARTGGCA	7 int		50.0		5
	ST64T	7R	CGAACCATTTCKATRGACTCCCA					5
	P1	8F	TGCTTATAACACCCTGTTACGTAT	8 int		50.0		5
		8R	CAGCCACCAGCTTGCATGATC					5
cIII	All listed	cIII 5′	ATGCAATATGCCATTGCA	cIII	43.0	53.0	168	This study
		cIII 3′	TTAGTCTGGATAGCCATA		43.0			This study
N		N (1) 5′	ATGACACGCAGAACTCAG	N1	51.4	50.0	384	This study
		N (2) 5′	ATGCAATGCCGAAGCAAC	N2	56.5	57.5	295	This study
		N 3′	TYACCTYGCYGTCAGTTG		54.4			This study
cI	BP4795, H19B	cI (1a) 5′	ATGGAAAACAAAGATATTCGC	cI 1a	59.4	54.0	705	This study
	933W, Stx2(I)	cI (1b) 5′	ATGGTTCAGAATGAAAAAGTG	cI 1b	58.1	61.0	708	This study
	BP4795, H19B, 933W, Stx2(I)	cI (1) 3′	TCACGAACTTTTCAGCCACTC		64.2			This study
	HK97, Lambda	cI (2a) 5′	ATGAGCRCAAAAAAGAAACCA	cI 2a	60.2	60.0	714	This study
	Nil 2	cI (2b) 5′	ATGAAATGGTATGAACTGGCT	cI 2b	59.9	60.0	654	This study
	Stx2(II)	cI (2c) 5′	ATGGATGGTTCCAGTACAGAG	cI 2c	60.4	59.0	598	This study
	VT2-Sa	cI (2d) 5′	GTGGTGTTTAAATACCTTGGT	cI 2d	57.0	59.0	510	This study
	HK97, Lambda, Nil 2, Stx2(II), VT2-Sa	cI (2) 3′	TYAACCAAACGTCTCTTCAGG	cI 2d	59.3		510	This study
	HK620	cI (3) 5′	ATGGAAAATAAAAAATCACTG	cI 3	54.9	53.0	714	This study
		cI (3) 3′	TCAAACCAGCCTTAGTTTTGT		61.3			This study
	D3112	cI (4) 5′	GTGAAATCAGACACTTACGGA	cI 4	59.2	62.2	665	This study
		cI (4) 3′	CTAAACCATCCAGCGGCTAGC		66.7			This study
	P27	cI (5) 5′	ATGAAATCTTTAGGTGAACGC	cI 5	59.6	62.6	723	This study
		cI (5) 3′	TCAGAAAATATCCCACCTGGC		64.3			This study
	phi 105	cI (6) 5′	ATGACTGTAGGGCAAAGAATC	cI 6	59.6	53.0	437	This study
		cI (6) 3′	GTATTCTTGATCGTCATTTCT		54.8			This study
cro	HK97, Sakai	Cro (1) 5′	ATGGAACAACGCATAACCCTG	cro 1	65.6	59.4	201	This study
		Cro (1) 3′	TTATGCAGTTGTTTTTTTGTT		56.4			This study
	HK620	Cro (2) 5′	ATGATTCGAATGACACTTGCC	cro 2	63.7	57.0	195	This study
		Cro (2) 3′	CTATTTGTTTTTCTTGTTGCT		55.3			This study
	933W, StxII	Cro (3) 5′	ATGCAAAATCTTGATGAGCCG	cro 3	66.0	65.0	228	This study
		Cro (3) 3′	TTATGCAGCCAGAAGGTTCTT		62.6			This study
	Sakai, ST64T, VT2Sa	Cro (4) 5′	ATGAGCAAYCTWCGRAAAWWY	cro 4	56.9	59.9	216	This study
		Cro (4) 3′	TTARGCRGCWTTRWGYTCMGG		61.7			This study
	P22	Cro (5) 5′	ATGTACAAGAAAGATGTTATCGAC	cro 5	57.7	55.0	186	This study
		Cro (5) 3′	CTTCATGGTTCTTTTGCG		59.4			This study
	ST104	Cro (6) 5′	ATGACTAACAAAGCAATACAA	cro 6	54.0	54.0	216	This study
		Cro (6) 3′	TTAACTTGCTGCCAGTAAGTC		58.6			This study
	SfV	Cro (7) 5′	ATGAAAGCGTATTGGGACTCT	cro 7	61.4	54.8	201	This study
		Cro (7) 3′	TTAATCTTTCGGATAGATATC		55.0			This study
	N15	Cro (8) 5′	ATGAAACCCGAAGAACTTGTG	cro 8	62.8	56.0	216	This study
		Cro (8) 3′	CTATTTAGTTCCACTGTTATG CCC		61.4			This study
	BP4795	Cro (9) 5′	ATGAGTAATGAACTACTACGCTGG	cro 9	60.1	56.0	234	This study
		Cro (9) 3′	TTATGCAGCCGATGCTCT		62.1			This study
cII	All listed	CII - 5′	ATGRMACRARCAAGYTACAGC	cII	56.0	53.0	297	This study
		CII - 3′	TCAGAATTGCATATCAAT		50.7			This study
Q	All listed	Q ATG 5′	ATGTTCTTATGGTTCACCG	Q	53.0	55.5	435	This study
		Q 3′	TTACGATCGTAAACTATTTTT		50.0			This study
stx	All sequenced stx-genes	Degen StxA (Type I&II)	TTTGTYACYGTSAYAGCWGAAG	stx	47.2	47.5	675	This study
		Degen StxB (Type I&II)	TYMTCATTATAYTTDGWRWACT		49.0			This study
Capsid	CP933X Capsid	CP933X Capsid 5′	TGGGRCCGGSAWRACATSCTG	capX	63.3	63.3	1,572	This study
		CP933X Capsid 3′	TTACGCAGCTCTGCTGTC		60.48			This study
	CP933I Capsid	CP933I Capsid 5′	ATGACAATCCCAGAACAG	capI	55.6	54.9	753	This study
		CP933I Capsid 3′	TCATACTGCTTTCTCCTT		51.9			This study
	CP933X (3) Capsid	CP933X (3) Capsid 5′	ATGTCSRTKTACACMACYGCC	capX3	55.0	58.0	1,026	This study
		CP933X (3) Capsid 3′	TTAYGCCAGYTKKACGGASAC		58.1			This study
	CP933X (2) Capsid	CP933X (2) Capsid 5′	RTGRCAGCAGAGCTGCGT	capX2	63.8	55.0	1,204	This study
		CP933X (2) Capsid 3′	TTAMACTGGKGTGTTYARCAA		52.1			This study
	CP933C Capsid	CP933C Capsid 5′	ATGCMGAGAATAATCGAATTAC	capC	56.5	56.0	1,158	This study
		CP933C Capsid 3′	TTAATCGTCGTCYTSYGGCAG		63.8			This study
	CP933R (3) Capsid	CP933R (3) Capsid 5′	ATGAAACGAACGCCTGTC	capR3	61.3	53.0	1,590	This study
		CP933R (3) Capsid 3′	TTACGTCTCACGKGRTGT		54.6			This study
	CP933R Capsid	CP933R Capsid 5′	ATGGGATTGTTTACGACC	capR	57.1	53.0	1,029	This study
		CP933R Capsid 3′	TTATTTCACCTGTACCAC		50.0			This study
	CP933O/R Capsid 5′	CP933O/R Capsid 5′	ATGGTRACGAAAAMATCACTGA	capO/R	57.5	60.6	348	This study
		CP933O/R Capsid 3′	TTACGGCAGCGCYGC		63.8			This study
	P27 Capsid	P27 Capsid 5′	RYGGYTGATRTTAAAGATGTG	capP27	60.0	56.9	1,224	This study
		P27 Capsid 3′	AATTTTCAKCAGCTTRATMGC		57.0			This study
	HK620 Major Capsid	HK620 Major Capsid 5′	ATGGCCTAACAATCTCGA	cap620	51.0	51.0	1,272	This study
		HK620 Major Capsid 3′	TTACGGATTACCGAAGAA		49.0			This study
Host recognition proteins								This study
Short-tailed	933W, VT2-Sa,	VTTF1 5′	GTTGTTGTTTCGGGGACG	VTTF1	64.0	55.0	1,900	This study
Stx phages	Φ24B	VTUTF 3′	TCATTCTCCTGTTCTGCC	VTTF2	58.8			This study
		VTTF3 5′	TGCAGAGGAAAGCTCGAC	VTTF3F	54.9	55.0	800	This study
		VTTF3 3′	GCAGCCTCTTCTGCCTTT	VTTF3R	62.3			This study
Long-tailed	P27 TF	P27 (p56) (TF) 5′	ATGTCTGTAGTGATATCAGGT	TFP27	52.4	55.3	400	This study
Stx phage		P27 (p56) (TF) 3′	TCATGCCAATCCTCACAA		61.0			This study
	N15 HRP	N15 (TF) 5′	ATGGCTACATCTACTCCG	TFN15	48.0	47.0	3,200	This study
		N15 (TF) 3′	TCAAAATACCCCCGTAAT		43.0			This study
	(CP)EcS1650,	(CP)EcS1650 (TF) 5′	ATGGCAGTAAAGATTTCA	TFcp	52.3	55.8	2,916	This study
	CP933M, λ (TF) 5	(CP)EcS1650 (TF) 3′	TTATGCAAGCCTCACAAT		56.6			This study
	T7 (TF)	T7 (TF) 5′	ATGGCTTAACGTAATTAA	TFT7	50.0	48.0	1,662	This study
		T7 (TF) 3′	TTACACGTCCTCTACGGC		58.0			This study
	HK022,	HK022 (TF) 5′	TTGCCTGGAGAAAATATG	TF022	56.0	57.7	1,116	This study
		HK022 (TF) 3′	TTAGTCAACAAGCTCCCT		54.6			This study
	HK97 (31111) (TF)	HK97 (31111)(TF) 5′	ATGATTTATAGCACCGGA	TF97	55.3	58.3	795	This study
		HK97 (31111)(TF) 3′	TTATACTGCCCTTACAATGTA		54.0			This study
Terminase	933W Major	933W Mterm 5′	ATGACATTCCGGAAGAAT	Term1	56.7	51.0	1,272	This study
	Terminase	933W MTerm 3′	TCAGTGAGCCATGCAGTG		62.5			This study
	933W Minor	933W 5′ MinTerm	ATGGCAAAGCTGGACTGG	Term2	63.9	64.9	807	This study
	Terminase	933W MinTerm 3′	TCATTCTTCCGGAATGTCAT		61.9			This study

^aPhage Accession numbers: BP 4795, NC_004813; 933W, NC_000924; partial sequence H19B, AF034975; Stx II bacteriophage I, NC_003525; HK97, NC_002167; Lambda, NC_001416; Nil 2, AJ413274; Stx II bacteriophage II, NC_004914; VT2-Sa, NC_000902; HK620, NC_002730; D3112, NC_005178; P27, NC_003356; phi 105, NC_004167; ST64T, NC_004348; Sakai, NC_002695; P22, NC_002371; ST104, NC_005841; SfV, NC_003444; N15, NC_001901; EcS1650, NC_002695. All CP933 sequences are from EDL 933, NC_002655; T7, NC_001604; HK022, NC_002166; Phi 80, X04051.

^bT_m, thermal denaturation midpoint temperature. Data are presented only for the primers designed for this study.

STEC strains (463 isolates) were previously obtained as part of a long-term study that examined both horizontal transmission (44, 45) and longitudinal carriage (26) of STEC on several farms in Cheshire, United Kingdom. All STEC strains had been typed with respect to their carriage of stx1 or stx2 genes. These strains were cultured in Luria broth (optical density at 600 nm of 0.45 to 0.55) and treated with norfloxacin (1 μg ml⁻¹) at 37°C for 1 h to induce endogenous prophages (20, 22, 28). The cultures were diluted 10-fold in phage buffer (PB) (Luria broth supplemented with 10 mM CaCl₂) and allowed to recover at 37°C for 2 h. The culture liquor was serially diluted and subjected to plaque assay on mid-exponential phase cultures of indicator host strain E. coli WG5rif⁺ (optical density at 600 nm of 0.5 to 0.6) grown in PB at 37°C with shaking (at 200 rpm). WG5rif⁺ is a rifampin-resistant derivative of the E. coli strain WG5 (17), an E. coli C strain that possesses an attenuated host restriction modification system and provides increased phage sensitivity (29). WG5rif⁺ was created through passage in increasing concentrations of rifampin (5 to 500 μg ml⁻¹) (22), and its identity was confirmed by pulsed-field gel electrophoresis. Indicator host (100 μl) was incubated at 37°C with 450 μl of serially diluted broth from the norfloxacin-treated STEC cultures. After 25 min, 5 ml of PB with 0.4% (wt/vol) Difco agar and 300 μg ml⁻¹ rifampin was added to the infection mixture and poured onto PB with 1.5% (wt/vol) Difco agar. The plates were then incubated overnight at 37°C and examined for plaques. A single semiconfluent plaque plate was flooded with PB (5 ml) and incubated for 4 h at 4°C to enable the phage to diffuse from the top agar into the buffer. The resultant phage preparation was used as a template in a stx-targeted PCR. Naïve WG5rif⁺ was also used as a control in DNA amplifications to rule out the identification of remnant prophage genes from across its genome and to serve as an additional control for the carryover of host DNA.

The presence of Stx phages in the preparations was determined using a novel primer set (Table 1) capable of amplifying all known stx genes (2). Of the 463 STEC isolates screened in this manner, it was possible to detect inducible phage in 89% of the isolates and stx genes in 101 of the phage preparations, indicating a 22% carriage rate of inducible Stx phages in the screened STEC population. Of these 101 strains that possessed inducible Stx phages, a subset of 70 STEC strains was chosen to produce the validation of the multilocus typing strategy. The excluded 31 STEC isolates possessed inducible Stx phages that either were not possible to reproducibly induce or were labile upon overnight storage at 4°C, a general problem that has been reported before (31).

When this collection of 70 STEC strains was screened with the primers comprising the multilocus typing scheme, ∼50% of the phage preparations possessed an int gene identical to that carried by the sequenced Stx phage 933W (34) (Fig. 2A), ∼54% of the induced phage preparations possessed an int gene with homology to the gene associated with the Stx phage Φ24_B (3, 16, 46), and 19% possessed both int genes. The cro gene, a key element in the regulation of lysis and lysogeny, required nine oligonucleotide primer pairs (Table 1) to cover the known genetic variation among phages. Four cro gene types (cro 1, cro 3, cro 8, and cro 9) dominated those identified in the phage preparations (Fig. 2D). Two of these loci (cro 1 and cro 3) have been described in other Stx phages and STEC strains and were identified here in >80% of the phage preparations. The cro gene type cro 7, which was identified in the highest number of phage preparations (96%), has been previously identified in the bacteriophage SfV from Shigella flexneri (1). A single oligonucleotide primer pair was able to amplify the cII gene in 90% of the phage preparations, while a single cIII-specific oligonucleotide primer pair (Table 1), possessing no degeneracies, amplified a product from all of the phage preparations. The gene responsible for encoding the bacteriophage repressor cI exhibits high levels of heterogeneity among lambdoid phages (Table 1) compared to other lambdoid regulatory genes such as cIII, cII, and Q. Ten primer sets were required to cover the known cI sequence diversity, but only six of these amplified products from the 70 induced phage preparations, indicating the presence of cI genes 1b, 2a, 2b, 2d, 3, and 4 (Fig. 2C). The cI gene 1b was detected most frequently (∼89%) and has been reported previously in the genomes of Stx phages 933W and VT2-Sa. The genes encoding the antiterminators involved in controlling the expression of either the early (N) or late (Q) genes were also examined. The N gene was readily identifiable by two primer sets yielding one of two products, N1 or N2, amplified in 80% and 66% of the samples, respectively (Fig. 2B), and both have been previously identified in Stx phages. The Q gene was detected by the use of one oligonucleotide primer pair, and this amplified the Q gene in all phage preparations.

FIG. 2.

Distribution of PCR targeted genes in phage preparations induced from 70 STEC strains. Panels present the percentage distributions of various gene types: int genes (A), N genes (B), cI genes (C), cro genes (D), capsid genes (E), and host recognition proteins (tail spikes/fibers) (F).

Phage structural genes were also analyzed. The capsid genes detected in the phage preparations had all been previously identified in prophages carried by the STEC isolate EDL933 (36) or Stx phage P27 (42); these included capX (present in 14% of phage preparations), capI (40%), capO/R (39%), and capP27 (∼2%) (Fig. 2E). The terminase genes associated with 933W, Term1 and Term2 (Table 1), involved in the packaging of DNA into the head, were present in 100 and 93% of the phage preparations, respectively. The bacteriophage host recognition factors (tail spike protein and/or tail fibers) described thus far for Stx phages were used to design oligonucleotide primer pairs (Table 1). These include the tail spike proteins from short-tailed bacteriophages (Podoviridae) such as 933W (37) and Φ24_B (3, 22); long, noncontractile-tailed bacteriophages (Siphoviridae), e.g., H19B (33); and bacteriophages with complex contractile tails (Myoviridae) like P27 (42). DNA amplification using all of these primer pairs (Table 1) indicated that 100% of the samples were positive for the tail spike protein gene, previously described (47) as associated with the short-tailed Stx phages 933W and Φ24_B (Fig. 2F). It was also possible to amplify sequences homologous to the tail spike protein gene of a remnant, long-tailed prophage (EcS1650) found within the Sakai O157:H7 genome; this gene was present in 21% of phage preparations (Fig. 2F). The high detection frequency of tail spike proteins associated with short-tailed Stx phages in the inducible phage preparations was not predicted, as there is considerable diversity in the tail fiber, base plate, or tail spike genes of coliphages, i.e., HK97, HK022 (25), and other phages such as N15 (41). The diversity of phage tails and their specific host recognition proteins has also been reported throughout the population of bacteriophages infecting the Mycobacteriaceae (18). Identification here of conservation of the tail spike protein in the inducible phages from STEC strains suggests selection pressure exerted by the conserved E. coli surface receptor for these phages (47).

A dendrogram (Fig. 3) was produced to provide an indication of the level of heterogeneity identified in the induced phage preparations from the STEC strains. The Jaccard dissimilarity (12) between the inducible phage preparations from each pair of 70 strains was calculated using the presence and absence of genes in the multilocus characterization scheme. For a pair of strains (i, j), the Jaccard dissimilarity is the proportion of genes present in i and/or j that is not present in both i and j. Thus, the shared absence of a gene in two strains is not treated as indicating similarity between them, but the shared presence of a gene is scored. The dendrogram (Fig. 3.) was constructed using single-linkage hierarchical clustering (12), and the dissimilarity between a pair of strains is represented by the height at which they join a single group. R, version 2.4.0 (40), was used for these analyses and illustrates differences between the genetic profiles of these inducible phages for each STEC strain. The patterns are indicative of a fluid gene pool and also highlight the sensitivity of the multilocus characterization scheme to identify, quickly and accurately, differences among the inducible phages from each STEC strain. This analysis shows that only two STEC strains exhibited similar inducible prophage profiles (Fig. 3), demonstrating a remarkably fluidic population of inducible phage genes.

FIG. 3.

Jaccard dissimilarity dendrogram generated from data on the presence or absence of genes in each of the induced phage preparations from 70 STEC isolates. The numbers represent the laboratory identification of each phage preparation obtained from a single STEC isolate. A value of 0.1 indicates 10% shared genes.

The detection sensitivity of phage genes was assessed by PCR amplification of the Q gene from serial dilutions of a purified Φ24_B phage preparation. The data (not shown) demonstrated that PCR was capable of detecting the presence of a single PFU in a reaction mixture, providing the same sensitivity of a more conventional plaque assay. It has also been reported in the literature that STEC strains harboring more than one phage upon induction do not necessarily produce equal numbers of the different phages (3, 43, 51); therefore, the overall sensitivity of our amplification reaction, using 4 μl of phage preparation, relies on the presence of 250 phage particles ml⁻¹ (per phage genotype). Since induction rates of wild-type phages have been previously reported to be in the range of 5 × 10⁴ to 2 × 10⁹ ml⁻¹ of STEC culture following induction (3), the technique reported here should be sufficiently sensitive to characterize inducible phages directly from a lysogen, even if they cannot be further propagated on laboratory host strains. Although a bacterial lysogen usually possesses multiple phage-related segments of DNA in its genome, often only one or two infectious phage types can be detected following induction/activation of the SOS response (3, 43, 53), so even though each of the phage preparations from this study is unlikely to consist of one pure phage, there will probably not be more than a few phages. This multilocus characterization scheme also has the potential to profile remnant bacteriophages within the bacterial host genome if patterns of inducible phage genes are compared to patterns of genes amplified from the lysogen directly. This approach to phage typing has advantages over DNA amplification typing techniques such as amplified fragment length polymorphism (50) or repetitive extragenic palindromic PCR (49). Amplified fragment length polymorphism (50) might provide quite diverse lengths of amplified products due to the levels of recombination among Stx and other lambdoid phages, and the levels of recombination might also mask identification of phages that are carrying similar or identical genes. Acquisition or loss of genes flanking the amplification target will obscure the presence of genes that control the biology of phage-host interactions. Bioinformatic analyses of the annotated genome sequences for 933W (NC_000924), Stx II bacteriophage II (NC>004914), VT2-Sa (NC_000902), P27 (NC_003356), and the raw genome sequence of Φ24_B did not reveal complete repetitive extragenic palindromic or enterobacterial repetitive intergenic consensus sequences (49) as would be expected for DNA regions of ∼60 kb. The primer sets presented in this paper were designed from sequences that were first divided into clades and then subsequently checked for specificity; amplification products are therefore indicative of a specific type of gene and do not routinely require sequencing for identification of the amplification product. This level of gene identification has previously been possible only by gross sequencing and sequence alignment data. The PCR-based typing scheme described here will not distinguish single nucleotide polymorphisms but can identify an insertion or deletion encompassing a stretch of nucleotides, as has been reported for the tail spike protein (47). Not all of the primer sets amplified genes from the pool of 70 phage preparations, so these genes may be poorly represented or not present in the general Stx phage population. Therefore, the subset of primers that have amplified genes in the 70 induced preparations used here could represent the progenitor of a routinely applicable Stx phage scheme.

In conclusion, we report the use of a novel multilocus characterization scheme to simultaneously characterize the inducible phage preparations from a variety of farm environment-derived STEC lysogens. These data do not allow specific identification of the individual phages harvested from STEC strains, but they do enable description of the gene pool present in inducible phage preparations from any single bacterial lysogen. One aim is to identify prevalent phage genes that can be induced from STEC, thereby making it possible to identify the genetic background of inducible and/or noninducible prophages from clinically relevant or environmentally circulating STEC strains. Such data would highlight Stx phage-borne traits associated with clinical disease and/or particular farm environments and possibly identify more stable Stx phages that are likely to persist and disseminate stx genes. Analyses of these data by Jaccard dissimilarity (12) or other matrices together with a knowledge of the Stx phage-borne traits could identify phage profiles important in the transmission of stx or enable tracking of genes across sample study sites in order to gauge the rate of dissemination of a gene by bacteriophages. This multilocus characterization scheme can be further developed. Most Stx phages have larger genomes than the archetypal λ; 933W is 21% larger (2), and most of the additional genes have no designated function. Some of this additional DNA may influence the biology of the E. coli lysogen (e.g., fitness in the mammalian gut or colonization potential) and, therefore, be subject to positive selection (2, 9). As additional factors are identified, it will be possible to augment this multilocus typing scheme to generate further information on the genotypes of Stx phages and their epidemiological significance.