Genomic Analysis of Clostridium perfringens Bacteriophage φ3626, Which Integrates into guaA and Possibly Affects Sporulation

Abstract

Two temperate viruses, φ3626 and φ8533, have been isolated from lysogenic Clostridium perfringens strains. Phage φ3626 was chosen for detailed analysis and was inspected by electron microscopy, protein profiling, and host range determination. For the first time, the nucleotide sequence of a bacteriophage infecting Clostridium species was determined. The virus belongs to the Siphoviridae family of the tailed phages, the order Caudovirales. Its genome consists of a linear double-stranded DNA molecule of 33,507 nucleotides, with invariable 3′-protruding cohesive ends of nine residues. Fifty open reading frames were identified, which are organized in three major life cycle-specific gene clusters. The genes required for lytic development show an opposite orientation and arrangement compared to the lysogeny control region. A function could be assigned to 19 gene products, based upon bioinformatic analyses, N-terminal amino acid sequencing, or experimental evidence. These include DNA-packaging proteins, structural components, a dual lysis system, a putative lysogeny switch, and proteins that are involved in replication, recombination, and modification of phage DNA. The presence of genes encoding a putative sigma factor related to sporulation-dependent sigma factors and a putative sporulation-dependent transcription regulator suggests a possible interaction of φ3626 with onset of sporulation in C. perfringens. We found that the φ3626 attachment site attP lies in a noncoding region immediately downstream of int. Integration of the viral genome occurs into the bacterial attachment site attB, which is located within the 3′ end of a guaA homologue. This essential housekeeping gene is functionally independent of the integration status, due to reconstitution of its terminal codons by phage sequence.

Clostridium perfringens is an anaerobic, gram-positive, spore-forming rod that can be isolated from the environment and is frequently found in the intestines of humans and domestic and feral animals. Spores can persist in soil, sediments, and areas subject to human or animal fecal pollution. The organism is the causative agent of histotoxic diseases spanning a range from gas gangrene (clostridial myonecrosis) to necrotic enteritis found in humans (pig-bel disease) or animals, especially poultry (46, 58, 60). Even subclinical C. perfringens infections in broiler flocks may be responsible for an impaired production performance (39). C. perfringens is also responsible for a significant number of food poisoning cases, due to its ability to produce enterotoxin (46). In addition, these strains may cause non-food-borne human gastrointestinal illnesses such as antibiotic-associated diarrhea (45). The surveillance reports of the Centers for Disease Control and Prevention ranked C. perfringens as one of most common causes of food-borne disease in the United States (5, 51).

Symptoms of a C. perfringens infection are caused by mostly extracellular enzymes and toxins produced by the organism (52). The genes for these enzymes can be chromosomally or plasmid encoded or, like the enterotoxin cpe gene, be located on a transposon (52). It has been reasoned that the similarities of some toxins of C. perfringens with toxins found in other organisms are due to horizontal gene transfer based on conjugative plasmids, transposons, or bacteriophages (52). In other clostridial species, toxins are known to be bacteriophage encoded; prominent examples are the neurotoxins BoNT/C and BoNT/D of Clostridium botulinum (41). Although the existence of bacteriophages infecting C. perfringens has been reported (43) and a certain phenotypic effect of temperate phages of this organism has demonstrated (61), we were surprised to find that no sequences or other molecular data on C. perfringens phages were available, except for a preliminary mapping of the integration sites of two phages (7). In general, Clostridium phages seem to be very poorly characterized, and so far they have escaped the advent of automated sequencing.

The aim of our present study was to gain essential information on phages infecting C. perfringens, with respect to basic morphological characteristics, nucleotide sequence, and potential effect on lysogenized host cells. Two novel temperate phages were isolated, one of which was chosen for detailed genetic and molecular analysis. We analyzed its genes and genome organization and localized the attachment sites for integration of the virus into the bacterial chromosome. Two genes which are possibly involved in regulation of sporulation in lysogenized C. perfringens were found.

MATERIALS AND METHODS

Organisms and plasmids.

Fifty-one C. perfringens strains were obtained from various sources (Table 1). Escherichia coli strain DH5αMCR (Invitrogen) in combination with plasmid pBluescript II SK(−) (Stratagene) was used for cloning.

TABLE 1.

Bacteria, phages, and plasmids used in this study

Strain(s), phage, or plasmid	Genotype or relevant properties	Sourcea
Bacteria
E. coli DH5α MCR	F−mcrA Δ(mrr-hsdRMS-mcrBC) φ80dlacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 supE44 λ−thi-1 gyrA96 relA1	Invitrogen
C. perfringens WS 2981 to 3002		22 isolates from various food sources
C. perfringens WS 3006 to 3008		3 isolates from the cecum/ileum of poultry with necrotic enteritis
C. perfringens WS 2955 to 2963		9 clinical isolates from humans
C. perfringens ATCC 3626 and 3628	Toxin type B and C type strains	ATCC (Manassas, Va.)
C. perfringens NCTC 528, 3110, 6719, 6785, 8081, 8346, 8533, 10240, 10612, 10614, 10719, and 11144	Strains with various toxin types (A, B, C, E)	NCTC (London, United Kingdom)
C. perfringens DSM 756T, 798, and 2943		DSMZ (Braunschweig, Germany)
Phages
φ3626	From C. perfringens ATCC 3626	This work
φ8533	From C. perfringens NCTC 8533	This work
Plasmid pBluescript II SK(−)	3.0-kb cloning vector; Ampr	Stratagene

^aATCC, American Type Culture Collection; NCTC, National Collection of Type Cultures; DSMZ, Deutsche Sammlung von Mikroorganismen.

Media and growth conditions.

C. perfringens strains were grown at 37°C in tryptone-yeast extract (TY) medium (16) in anaerobic jars (Oxoid), using the Anaerocult A system (Merck). Most of the handling was performed in a flexible vinyl glove chamber (Coy Laboratories, Grass Lake, Mich.), which contained a 95% N₂-5% H₂ atmosphere. Luria-Bertani medium was used for the incubation of E. coli cells at 37°C. Ampicillin (100 μg ml⁻¹) was used for the selection for plasmid-bearing cells.

Isolation and purification of bacteriophages.

A total of 51 clostridial strains were screened for lysogeny by UV irradiation as previously described (35). Exponentially growing cells (10 ml) were exposed to UV light (254 nm; 0.02 J cm⁻²) for 5 min. After 3 h of incubation at 37°C in the dark, cultures were centrifuged (10 min, 8,000 × g), and supernatants were cleared by filtration (0.2-μm-pore-size filter). Phage activity was tested by the spot-on-the-lawn method against all available C. perfringens strains.

The soft-agar layer technique (2) was used for the purification and propagation of the phages. Dilutions of the supernatants displaying lytic activity were added to 3.5 ml of molten soft agar inoculated with 0.1 ml of log-phase culture of the propagation strain. The mixture was poured on TY plates and incubated overnight. Single plaques were picked and placed into 0.45 ml of TY medium. After 4 h of incubation at 4°C, the phage-containing solution was filter sterilized and used for a second round of purification.

Determination of the lytic range of the bacteriophages.

The ability of the two phages to lyse C. perfringens strains was tested by the drop-on-the-lawn-technique. Ten microliters of the prepared phage stocks (10⁷ PFU/ml) was placed on the plates inoculated with C. perfringens strains. The lytic activity was observed after overnight incubation.

Propagation and purification of φ3626.

For high-titer stocks (>10⁹ PFU ml⁻¹), liquid cultures were used. Cultures were infected at an optical density at 600 nm of 0.1 at a multiplicity of infection of 1. Afterwards, growth was monitored photometrically, and following lysis, phages were harvested by centrifugation (10,000 × g, 10 min) and sterile filtration of the culture supernatant.

Purification of viruses from high-titer stocks has been described earlier (67). Briefly, phages were concentrated by polyethylene glycol 8000 precipitation, stepped CsCl density gradient centrifugation, and dialysis (54).

Electron microscopy.

Phage particles were examined by electron microscopy as reported before (67). Briefly, a small drop of the CsCl solution (5 μl) was placed on top of a carbon film fixed on a copper grid (400 mesh) for 1.5 min, to allow phage to attach to the carbon film. Excess solution was removed. The surface of the grid was repeatedly washed with water and finally negatively stained with 2% uranyl acetate. Pictures of the virus particles were taken with a transmission electron microscope (Zeiss EM-10A) at an acceleration voltage of 60 kV with a magnification of ×100,000.

Cloning, nucleotide sequencing, and identification of the cos site.

The DNA of φ3626 was extracted and purified by using standard techniques (54), and the construction of genomic libraries of φ3626 was performed essentially as described previously (36). Here, limited digests with Tsp509I (New England Biolabs) and complete digests with HindIII (MBI Fermentas) or TaqI (Roche) were performed, and fragments of 1 to 2 kb in length were ligated into pBluescript and transformed into E. coli. Blue-white screening on X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside)-containing agar plates was used for the identification of insert-bearing clones. Plasmids from small-scale cultures were digested with PauI (MBI), and 58 clones carrying inserts of various sizes were identified by agarose gel electrophoresis. These plasmids were used for sequencing with IRD-800-labeled primers complementary to sequences flanking the multiple cloning site. Sequencing was performed using a heat-stable polymerase (SequiTherm EXCEL II; Epicentre Technologies) on an automated DNA sequencer (4200 IR²; LI-COR). The sequences obtained were edited and aligned using the software DNASIS (version 2.1; Hitachi). Gaps were closed by direct sequencing of φ3626 chromosomal DNA, with the aid of specific primers derived from the contigs. Distinct chain termination signals were generated at the ends of the molecule, i.e., the putative single-stranded ends (cos sites). The genome sequence was finalized by determination of the sequence of the cos site overlaps, by PCR amplification of DNA from lysogenic host bacteria (see below), using primers complementary to sequences upstream and downstream of the cos site.

Analysis and amino-terminal sequencing of φ3626 structural proteins.

The isolation and purification of the structural proteins were performed as described earlier (37, 67). Virion proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), followed by densitometric scanning. Proteins were electroblotted on a polyvinylidene difluoride membrane and stained with Coomassie blue, and three major bands were excised from the membrane. The first 10 amino acids (aa) of each of the individual proteins were determined using an automated sequencer (Applied Biosystems Procise 492-01).

Identification of attPP′ and attBB′.

Chromosomal DNA of the C. perfringens ATCC 3626 lysogen was isolated by a mutanolysin-lysozyme incubation step, followed by phenol-chloroform extraction and ethanol precipitation of the bacterial DNA (17). The material was then used as a template for the identification of the attachment site by inverse PCR (50). attPP′ was expected to be located in a noncoding region immediately downstream of int. A Sau3AI restriction site is present within the int gene, and this enzyme (Roche) was used for complete digestion of the bacterial DNA. Fragments were treated with T4 DNA ligase to obtain self-ligated circular molecules. Divergent primers Att3up (5′-CTCAAATGATAGCAACAACAGG-3′) and Att3dw (5′-CTTTTACTTTTAGGAGTTTGGG-3′), complementary to an area located within int, were designed and used for PCR amplification of ligated fragments. The products were purified and sequenced using the same primers. The obtained sequence contained the attBP′ site, and the nonprophage part of the sequence displayed 100% identity over 663 nucleotides (nt) to a sequence of the unfinished C. perfringens genome available from The Institute for Genome Research (TIGR) (http://www.tigr.org). Additional sequence was obtained in order to design a primer, attB1 (5′-GACAATCATATTAAAATGACTGCC-3′), that in combination with a primer complementary to the prophage DNA, att5dw (5′-CTCAAATGATAGCAACAACAGG-3′), produced a fragment containing the attPB′ site, which was also purified and sequenced.

Bioinformatics.

The program DNASIS and the Husar Analysis Package (version 4.0; http://genome.dkfz-heidelberg.de) were used for analysis of the nucleotide and amino acid sequences. The BLAST algorithms (4) were used for similarity searches in the databases available through the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov) or the Husar Analysis Package. COILSCAN (40) was used to predict probabilities to form a coiled-coil structure. The HMMscan program (Pfam database, release 5; http://pfam.wustl.edu) identifies protein families by using the hidden Markov model (15), and TmHMM (version 2.0) protein analysis uses this model for prediction of transmembrane domains (59).

Nucleotide sequence accession numbers.

The DNA sequences reported here will appear in the EMBL, GenBank, and DDBJ databases under accession no. AY082069 and AY082070.

RESULTS

Isolation of novel C. perfringens phages.

Lytic activity was observed in UV-induced culture supernatants of two of the tested strains (ATCC 3626 and NCTC 8533), which was shown to be due to the presence of bacteriophages. Their lytic ranges on 51 C. perfringens strains were different: phage φ3626 was lytic against 11 strains (21.6%), while phage φ8533 was able to lyse only 4 strains (7.8%), which were also all sensitive to φ3626. The optimal propagation hosts were determined to be NCTC 3110 for φ3626 and ATCC 3628 for φ8533.

Since phage φ3626 displayed a broader host range and, in our hands, was easier to propagate, it was chosen for further studies. Electron microscopy (Fig. 1) revealed that the φ3626 virion has an isometric capsid (diameter, 55 ± 2 nm [seven particles were measured]) and a long, flexible, noncontractile tail (length, 170 ± 5 nm). Therefore, φ3626 belongs to the Siphoviridae family of double-stranded DNA bacterial viruses in the order Caudovirales (1).

FIG. 1.

(A) Electron micrograph of φ3626, showing its isometric capsid and flexible tail. Bar, 100 nm. (B) SDS-PAGE showing the protein profile of φ3626 (lane φ) and molecular mass marker proteins (lane M). N-terminal amino acid sequences from selected structural proteins are shown between panels A and B. The arrows point to the respective SDS-PAGE bands and individual, corresponding viral components, as deduced by bioinformatic analysis.

Nucleotide sequencing and determination of the cos sites.

The genome of φ3626 was sequenced by a shotgun approach. Sequencing of the various plasmid inserts yielded contigs which allowed the design of primers to close the gaps between the contigs. The sequence was finalized by determination of the cos site core sequence from amplified bacterial chromosomal DNA (Fig. 2). Predicted restriction maps of the φ3626 DNA were in perfect agreement with the experimentally achieved pattern, indicating that the sequences were assembled correctly (results not shown). The complete genome has a size of 33,507 nt with 3′-protruding, single-stranded cohesive ends of 9 nt (Fig. 2). Its average molar G+C content of 28.4 mol% is slightly higher than the 24 to 27 mol% reported for its host (21) or for the clostridial plasmid pIP404 (25 mol%).

FIG. 2.

Identification of single-stranded cohesive ends in φ3626 DNA. (A) Chromatogram showing the sequence from the right end of the genome towards the cos site (coordinates 33481 to 33498). (B) Chromatogram of the sequence from the left end of the noncircular genome towards the cos site (shown inversed; coordinates 1 to 17). (C) Sequence of a PCR product from φ3626 prophage, spanning the entire cos site. (D) Corresponding sequence of the ligated cohesive ends, joining the left and right arm of the DNA molecule. The single-stranded 3′-protruding ends are in boldface.

Identification and organization of φ3626 ORFs.

Bioinformatic analysis revealed the existence of 50 putative protein-coding regions on the φ3626 genome (Table 2), covering 94.1% of the sequence. The criteria for the characterization of a potential open reading frame (ORF) were the existence of an ATG, GTG (five present), or TTG (three present) start codon and a minimum coding capacity of 40 aa. Except for ORF41, all ORFs were preceded by a recognizable ribosome binding site with a sequence complementary to the 3′ end of C. perfringens 16S rRNA (18).

TABLE 2.

Features of bacteriophage φ3626 ORFs, gene products, functional assignments, and amino acid sequence homologies

ORF	Start position	Stop position	gP		Functional assignment	Homology(ies)b
			kDa	pIa
1	54	557	19.2	7.2	Terminase, small subunit	φ105 (gp21), Sfi19 (gp161), bIL309 (gp38)
2	550	2295	67.7	5.9	Terminase, large subunit	φ105 (gp22 and gp23), φSLT (gp563), D3 (gp2)
3	2296	3564	48.7	8.3	Portal protein	φ105 (gp25), φSLT (gp412), D3 (gp4)
4	3554	3715	6.4	9.7
5	3720	4325	23.1	4.7	Putative prohead protease	φPV83 (gp41), φC31 (gp35), φPVL (gp5a)
6c	4366	5631	47.7	4.9	Major capsid protein (Cps)	φ105 (gp27), bIL285 (gp44), φadh (gp395)
			34.3	4.7	Processed form, cleaved between R114 and D115
7	5641	5919	10.8	4.9		φ105 (gp30)
8	5912	6235	13.2	8.8
9	6236	6685	17.0	8.7		φ105 (gp32)
10	6687	7037	14.0	4.4
11c	7054	7650	22.5	4.9	Major tail protein (Tsh)	φ105 (gp34)
12	7665	7979	11.9	9.7
12.1d	7665	8485	32.2	4.3
13	8515	11400	103.8	10.0	Tail tape measure protein (Tmp)	φSLT (gp1374), TP901-1 (TMP), φPVL (gp15)
14	11401	12102	27.4	4.8		PSA (gp13), bIL285 (gp53)
15	12102	15053	110.0	6.2		PSA (gp14), bIL285 (gp54), φg1e (gp1608)
16c	15106	16584	55.1	5.5	Minor structural protein	φ105 gp42
17	16597	16905	12.1	4.8
18	16914	17669	28.2	4.8		bIL170 (gpl12, accession number AF009630)
19	17684	18061	14.3	9.6	Holin	φ105 (gp45)
20	18149	19192	38.8	5.9	Endolysin	pIP404 (ORF10), C. perfringens (ORF10C), PSA (Ply), 12826 (Ply12)
21	19382	19804	15.9	4.8
22	21119	20070	41.3	9.9	Integrase	T12 (Int), PSA (gp24), T270 (Int)
23	21587	21135	17.6	5.2		PSA (gp26), φSLT (gp153), A118 (gp35)
24	22330	21608	27.4	5.1	Repressor	φSLT (gp101a), φgle (gp132), PBSX (Xre)
25	22504	22722	8.3	9.2	Cro	A2 (Cro)
26	22749	23645	34.5	8.7		φSLT (gp81a) φadh (Tec)
27	23680	23856	6.8	9.6	Transcriptional regulator	φETA (gp13 [64]), φPV83 (gp12)
28	23903	24082	6.7	9.8
29	24093	24299	7.9	9.2
30	24319	24537	8.5	10.3
31	24527	24853	1.1	4.5
32	24867	25412	21.3	9.4	Sigma factor	C. acetobutylicum, B. subtilis, and C. perfringens
33	25428	25547	4.6	8.9
34	25563	26315	29.3	9.4		φETA (gp22), φSLT (gp256), rlt (gp11 [63])
35	26325	27608	48.3	8.3	Helicase	P1 (Ban), SPP1 (gp40), D3 (gp74)
36	27625	27825	7.9	4.4
37	27825	28061	9.3	4.9
38	28058	28270	7.9	6.4
39	28263	28679	15.4	4.9	SSBe	UI36 (gp141), bIL286 (SSB), A118 (SSB)
40	28706	28942	9.1	4.1
41	29020	29295	10.9	10.0
42	29304	29576	10.5	10.9	SpoIIID	C. acetobutylicum, B. subtilis, and other bacilli
43	29638	30168	21.1	8.7
44	30515	30865	13.5	9.5
45	30858	31424	21.4	4.7
46	31428	31973	21.4	10.2	Recombinase	L2 (gp5 [44]), Wφ (Int [34]), 186 (Int [24])
47	32153	32500	13.4	9.5
48	32533	32718	7.4	4.6
49	32735	33058	12.6	9.9
50	33039	33467	16.6	9.8		φSLT (gp104b), bIL286 (gp39), φadh (gp170)

^aPredicted by computer analysis.

^bFor reference or accession numbers, see text.

^cExperimentally shown.

^dTranslational frameshift might result in two C-terminally different products (see text).

^eSSB, single-stranded-DNA binding protein.

On the whole, the probable transcriptional units of φ3626 (as deduced from bioinformatic analysis) can be organized into three major functional clusters (Fig. 3), which is also reflected by the transcriptional direction of the clustered ORFs. The first cluster, from the cos site at coordinate 1 up to position 19804, is transcribed rightward in the genomic map (Fig. 3) and represents genes encoding the structural proteins and the lysis system. These genes may be considered late genes. The second putative cluster (nt 19805 to 23645) encodes products which are likely responsible for integration and the control of lysogeny, including the att site, an integrase, the repressor, and a putative Cro-like protein. The third cluster (nt 23680 to 33507) includes rightward-facing ORFs, whose putative products most probably represent the early genes, involved in replication, recombination, and modification of phage DNA.

FIG. 3.

Schematic representation of the φ3626 genome, with its assumed ORFs, some functional assignments, and the overall genetic organization. The ORFs are numbered consecutively (see Table 2) and are indicated by arrows or arrowheads that point into the direction of transcription. Black arrows indicate rightward transcription, and grey-shaded ORFs are oriented leftwards. Their relative position on the genome (33,507 nt) is also indicated. The attPP′ and the cos sites are shown by a dashed arrows. P, specific putative sporulation-dependent promoter; T, putative rho-independent transcription terminator (see text for explanation). SSB, single-stranded-DNA binding protein.

Structural proteins of φ3626.

The proteins building up the virion particles were separated by SDS-PAGE (Fig. 1). Microsequencing yielded N-terminal sequences of three structural proteins, which enabled identification of the corresponding genes.

The major capsid component Cps (NH₂-DIMSSTNNGA, encoded by ORF6) resembles 43.3% of total phage protein. The N-terminal sequence indicated that the gene product must be posttranslationally processed between R₁₁₄ and D₁₁₅, which results in a decrease in size from 47.7 to 34.3 kDa. Cps displays a high probability to form a coiled-coil structure in the N-terminal portion removed by processing, a finding similar to what has been reported for phages Sfi21 (Streptococcus thermophilus) and HK97 (E. coli) (9, 11).

The major tail protein Tsh (NH₂-PEVVNTRRXG, encoded by ORF11) corresponds to 12.7% of the total protein, with an apparent size of 27 kDa. The predicted size was 22.5 kDa. This observation was also made for other phages, such as Listeria phage A118 (36). The amino acid sequence differs from the predicted sequence only by the absence of the initiator methionine, as is often observed in prokaryotes with proline as the penultimate amino acid (22).

A minor structural protein (gp16, 2.1% total protein content) was also isolated, and its N-terminal sequence (NH₂-MKYIQTKVVY) was in agreement with the putative product of ORF16. The predicted size of 55.1 kDa is in perfect agreement with the determined size.

Bioinformatic analysis of φ3626 gene products.

Deduced amino acid sequences of the 50 ORFs were compared with known sequences from the databases to uncover similarities to genes with known function. Functional assignments and significant homologies to other proteins are listed in Table 2, and some of the most interesting findings are described below.

(i) ORF1 and ORF2.

Fifty-four nucleotides downstream of the cos site, two ORFs have been allocated that likely represent the small and large subunits of the terminase, which introduces specific cuts into the concatemeric DNA at the cos site to initiate genome packaging. gp1 shares similarities with gp21 of Bacillus subtilis phage φ105, gp161 of phage SfiI9 of S. thermophilus (12), and gp38 of prophage bIL309 of L. lactis (8), with similarities (related amino acids) of between 43 and 47% over stretches of 132 to 169 aa. The large subunit (gp2) displayed significant similarities to the gp22 and gp23 of φ105 (55% over 416 aa and 48% over 88 aa, respectively) and to terminases from various other phages.

(ii) ORF3 to ORF5.

gp3 encodes the putative portal protein, based on similarities to gp25 of φ105 and the portal proteins of φSLT (47) and D3 (29) (39 to 45% identity over 313 to 397 aa). The localization of ORF4 disrupts the pattern found in φC31, D3, and HK97, which consists of a consecutive order of portal protein, prohead protease, and an N-terminally processed major capsid protein (14, 29, 57). No similarities to any potential protein could be found for gp4. gp5 is similar to putative prohead proteases from various phages, including S. aureus phages φPV83 gp41 (25) and φPVL gp5a (26), Streptomyces phage φC31 gp35, and bIL309 gp36. Homologies range from 58 to 62% over stretches of 147 to 172 aa.

(iii) ORF12 and ORF12.1.

Tailed bacteriophages frequently have a pair of overlapping ORFs between the major tail protein gene and the tail length tape measure gene that are expressed by a translational frameshift (23). This seems to be the case for ORF12 and ORF12.1, resembling the situation in λ phage (32). ORF12 starts at nt 7665, and the obvious stop codon is located at nt 7979. However a putative “slippery sequence” (GGGTTTT) is located at positions 7947 to 7952, where ribosomes might shift frames and continue in the −1 frame until termination at nt 8485, resulting in a larger gene product.

(iv) ORF19 and ORF20.

ORF19 and ORF20 encode a dual lysis system, consisting of a holin (Hol) and an endolysin (Ply), responsible for cell lysis and release of phage progeny. gp19 is similar (50% over 105 aa) to a probable holin from B. subtilis phage φ105. Strong similarities (72 to 75% over 265 to 346 aa) of gp20 with hypothetical proteins present encoded by the C. perfringens chromosome and a C. perfringens plasmid, pIP404 (19, 42, 55), were found. Within the N terminus, gp20 displays similarities to N-acetylmuramoyl-l-alanine amidases from different sources, e.g., Listeria monocytogenes phage PSA endolysin (GenBank accession number AJ312240), Bacillus cereus phage 12862 endolysin (38), or a B. subtilis autolysin (30), with similarities of 41 to 43% over 163 to 207 aa. Taken together, these data indicated the possible roles of gp19 and gp20, which were recently confirmed by experimental approaches (M. Zimmer, N. Vukov, S. Scherer, and M. J. Loessner, submitted for publication).

(v) ORF22.

ORF22 is located immediately upstream of the attachment site (see below). A HMMscan indicated that it encodes a phage integrase belonging to the λ integrase family, responsible for the site-specific recombination of φ3626 into the C. perfringens chromosome. We also found a relationship to Int459 of the transposon-like element CW459tet(M) from C. perfringens (53) (similarity of 47% over 363 aa). Many other integrases of phage origin also displayed homology.

(vi) ORF24 and ORF25.

gp24 displays in the N terminus homologies to several repressors of phages. Strongest hits are the repressor of φSLT, the repressor Xre of the Bacillus prophage PBSX (62), and the repressor of φg1e (28) (similarities of 53 to 64% over 62 to 70 aa). In the opposite direction, ORF25 encodes a product with similarity to Cro of Lactobacillus casei phage A2 (31) (47% similarity over 70 aa). Both proteins contain putative H-T-H motifs similar to those of CI and Cro of λ (results not shown), indicating their potential to bind to DNA.

(vii) ORF32.

Both HMMscan and BLAST searches suggest that gp32 is a sigma factor, similar to sporulation-specific sigma factors of Clostridium acetobutylicum, (sigma-F) (48), B. subtilis (sigma-E) (30), and C. perfringens (sigma-K) (database accession number AF21885) (45 to 53% over 174 to 219 aa).

(viii) ORF35.

gp35 seems to be a helicase, responsible for the unwinding of DNA before replication. The strongest similarities are with DnaC of C. acetobutylicum (48) and B. subtilis (30) (51 to 55% over 424 aa), but there are also similarities to phage helicases from SPP1 (3) and others.

(ix) ORF42.

gp42 shows significant similarities (66 to 71% over 74 aa) to σ^E/σ^K-dependent transcriptional regulators, also known as stage III sporulation protein D (SpoIIID), found in C. acetobutylicum (48), in B. subtilis (30), and in other bacilli (33, 62, 65).

Identification and nucleotide sequence of the attachment sites attPP′ and attBB′.

The integration site of the bacteriophage φ3626 was identified by using an inverse PCR approach. Sequencing of the first PCR product yielded 663 nt of bacterial sequence. This (left-end) prophage-host junction was designated attBP′ (Fig. 4C). The bacterial sequence was found to be 100% identical to a portion of a contig (10.804 bp) of the unfinished genome of C. perfringens (TIGR, http://www.tigr.org). A second PCR product yielded additional sequence information on the host (183 nt), which was also identical over its full length to the C. perfringens sequence and encompassed the attPB′ site at the junction of the right arm of φ3626 and the bacterial chromosome.

FIG. 4.

Organization of phage and bacterial attachment sites. (A) Schematic representation of the circularized φ3626 genome with its attPP′ site (see Table 2 for identity of ORFs). (B) Partial sequence of the C. perfringens genome, encompassing attBB′ and surrounding genes (see text). (C) Integrated prophage status within attBP′ and attPB′. The ORFs flanking attPP′ and attBB′ are indicated by arrows (black arrows, phage or prophage ORFs; grey arrows, C. perfringens ORFs). Partial sequences of junction fragments from the phage or prophage are in lowercase letters, the host sequence is in uppercase letters, and the homologous att site core sequence (12 bp) is boxed. Underlined sequence corresponds to the 3′ end of guaA, and the stop codon is in italics.

Alignment of the two att site flanking sequences revealed a core sequence of 12 nt (Fig. 4). On the circularized phage genome, attPP′ is located in a noncoding region of 273 bp between ORF21 and int. Sequence analysis showed that attBP′ is encompassed within the 3′ end of a C. acetobutylicum guaA homologue, encoding a GMP synthetase (89% similarity over 220 aa). The core sequence of 12 bp represents the terminal four triplets of guaA, including a TAA stop codon (Fig. 4). The region immediately downstream of attPB′ (189 nt) does not contain coding capacity.

In order to obtain a more detailed picture of the chromosomal localization of attBB′ and guaA, a 10.8-kb sequence contig of the unfinished genome of C. perfringens (TIGR) was annotated. Upstream of guaA, guaB could be identified, encoding an IMP dehydrogenase (82% similarity over 478 aa with C. acetobutylicum GuaB [48]). Downstream of attBB′, four coding regions could be identified, the products of which are similar to NarK of Pseudomonas aeruginosa (accession no. Y15252; 40% over 315 aa) and AppA, AppB, and AppC, which are part of an oligopeptide transporter operon present, for example, in C. acetobutylicum (48) (similarities of 53% over 580 aa, 63% over 288 aa, and 71% over 323 aa, respectively).

Other features.

Several potential stem-loop-forming sequences were identified at the ends of possible transcriptional units, likely representing rho-independent transcription terminators: (i) between the lysis cassette and ORF21, at position 19195 to 19230 (ΔG = −32.3 kcal mol⁻¹); (ii) between ORF21 and the lysogeny control region (19856 to 19888; ΔG = −18.0 kcal mol⁻¹); (iii) downstream of ORF42 (SpoIIID homologue) (29587 to 29620; ΔG = −17.1 kcal mol⁻¹); and (iv) in a noncoding region downstream of ORF43 (30462 to 30488; ΔG = −15.4 kcal mol⁻¹). Sequences similar to the P1 promoter upstream of the C. perfringens enterotoxin gene cpe and to the B. subtilis σ^K-dependent promoter (66) have been allocated upstream of ORF42, possibly encoding a σ^K/σ^E-regulated SpoIIID homologue (coordinates xc), but could also be found upstream of ply (18007 to 18033) and upstream of ORF43 (29688 to 29713). Interestingly, putative stem-loop structures are present downstream of all of these coding regions, suggesting a tightly regulated, spatial expression of specific genes.

DISCUSSION

Two novel bacteriophages were isolated from lysogenic strains of the pathogen C. perfringens. φ3626, a member of Siphoviridae, is the first C. perfringens phage characterized on a molecular level, and it is the first bacteriophage of the genus completely sequenced. While 19 of the 50 potential polypeptides allowed functional assignments, 22 gene products had no match in the current databases and represent new entries. Most similarities were found with proteins of other phages infecting low-G+C gram-positive bacteria, from the genera Bacillus, Streptococcus, Staphylococcus, Lactococcus, Listeria, and Lactobacillus, but also, to a more limited extent, with proteins of phages from E. coli or Pseudomonas. Intergeneric relationships of φ3626 genomic modules are particularly pronounced in the left-end module, encompassing the DNA-packaging and capsid-building machinery. The consecutive order of the genes encoding the small and large terminases, portal, prohead protease, and major head protein is common to phages from low-G+C gram-positive bacteria (13) and to lambdoid viruses in general (23). Here, it is disrupted by ORF4 with unknown function. From an evolutionary point of view, it is interesting that most amino acid sequence similarities were found to proteins from other cos site phages, similar to S. thermophilus phage Sfi21 (6). These infect low-G+C bacterial hosts, in particular B. subtilis (phage φ105), and Staphylococcus aureus (phages φSLT, φPVL, and φPV83). Here, the convincing relatedness of DNA-packaging and head proteins points to a vertical passage and evolution of at least this module (6). The conserved order of genes among these viral genomes implies a tight conservation. However, it should be noted that the similarities of φ3626 are based on amino acid sequences, whereas alignment of nucleotide sequences generally did not yield significant homologies. This observation in turn suggests a lack of recent horizontal genetic exchange and implies that evolution of φ3626 should have diverged at an earlier point. However, the virus must have had earlier access to a more common gene pool among the phages infecting low-G+C hosts, as indicated by genome composition and some amino acid sequence conservation of important elements and modules. In this context, it is important to note that clostridia occupy different ecological niches and, due to their anaerobic nature, require different growth conditions than the bacteria hosting some of the related phages. A possible hypothesis is that φ3626 may actually have coevolved with its host over time, also resulting in less frequent contact with the remaining low-G+C host gene pool. In general, our findings are in agreement with the hypothesis (20) that tailed phage genomes are genetic mosaics which have been built from a large common pool by genetic exchange. However, it is also evident that access was, and is, not uniform among the different host-dependent viruses.

Phage φ3626 was isolated from a temperate C. perfringens strain, where, in the lysogenic stage, the prophage is integrated into the host chromosome within the 3′ terminus of guaA. The attPP′ site complements the disrupted coding sequence, which permits translation of guaA to terminate at the expected site (Fig. 4), independent of the integration status. guaA encodes GMP synthetase, a housekeeping protein responsible for de novo biosynthesis of the purine nucleotide GMP. guaA null mutants become guanine auxotrophs (49), and it is obvious that integration of phage must not destroy its function. Phage φ3626 exhibits a particularly elegant way to retain genetic function by duplication of terminal codons including a translational stop signal.

A putative integrase is present in φ3626, related to int459 of the transposon-like element CW459tet(M) of C. perfringens CW459 (53). Moreover, attPP′ is located immediately downstream of int. This is a common organization and forms an ideal basis for building a site-specific integration vector. No such system is available for C. perfringens, and its construction could be of interest for molecular and genetic research on this pathogen.

Posttranslational processing of the major head protein is frequently found in bacteriophages. During capsid maturation, φ3626 Cps is processed by removal of the first 114 residues. A similar processing can be observed in many other phages which lack a scaffold protein. As described for Sfi21, φPVL (11), and HK97 (9), φ3626 Cps revealed a possible coiled-coil structure in the amino-terminal part of the protein which is removed. It has been assumed by Duda and coworkers (14) that this domain might be a functional equivalent of a scaffold, fused to the capsid protein.

The length of the phage tail is thought to be determined by a ruler mechanism, dependent on the size of the so-called tape measure protein (27). Our findings support this theory, because the designated Tmp (gp13, 962 aa) is about 13% larger than the λ protein (gpH, 853 aa), and the φ3626 tail is approximately 13% longer than the λ tail (170 and 150 nm, respectively). This linear relationship was also found in other phages unrelated to λ, such as A118 from L. monocytogenes (36).

The presence of a sporulation-associated sigma factor homologue within the early genes (ORF32) suggests a possible function in programming the RNA polymerase, as shown for sigma gp28 of B. subtilis phage SPO1 (10). ORF42 encodes a protein with convincing similarity to a B. subtilis sporulation-dependent transcriptional regulator (SpoIIID), which is part of the sigma factor cascade resulting in sporulation. Moreover, several sequence motifs possibly reflecting sporulation- or σ^K-dependent promoters (66) are present in the φ3626 genome (Fig. 3). Interestingly, SpoIIID and some of the related bacterial sigma factors showing homology are mother cell specific and are made only in the late stages of sporulation, after which the mother cell lyses. Although it is not entirely clear what the precise function of the homologous factors in φ3626 is, an earlier study on the effect of lysogeny on sporulation of C. perfringens offers a possible explanation: curing of a lysogenic strain resulted in less efficient sporulation and decreased heat resistance of spores, and reintroducing the prophage reversed the effect (61). Unfortunately, no further details on the nature of this phage or its interaction with the host were reported, and we therefore do not know whether it might have been a φ3626-type virus. However, similar findings were reported for some bacilli, where spore-converting bacteriophages which enhanced the sporulation of infected cells were isolated (56). Together, these findings point towards a potential effect of lysogeny on sporulation. In fact, correlation of φ3626 prophage carrier state and sporulation ability is currently being investigated in our laboratory, and preliminary results support the above-described hypothesis.

No direct evidence for an influence of φ3626 on the pathogenicity or virulence of C. perfringens could be obtained in our study. However, it is noteworthy that the expression of the Cpe enterotoxin, which is responsible for the food poisoning effect of C. perfringens, is also dependent on sporulation events, and transcription of cpe relies on the activation of sporulation-associated promoters (66). Further studies are needed to identify the precise role of these sequences in φ3626. It is intriguing that they are homologous to the recognition sites of sporulation-dependent sigma factors from B. subtilis, which are also similar to the proposed φ3626 sigma factor gp32.