Introduction

Alkaliphilic microorganisms, which grow optimally at pH values between 9 and 11, are widespread in nature. They have been isolated from many naturally occurring alkaline environments (such as soda lakes) man-made alkaline environments (e.g., bauxite processing wastes), as well as many other sites (e.g., soils and feces). Both bacteria and archaea have alkaliphilic representatives. Curiously, in spite of the widespread distribution of alkaliphilic microorganisms, reports of viruses active against these organisms are extremely rare. Indeed, isolation of viruses active against most extremophilic microorganisms, with the exception of thermophilic and hyperthermophilic bacteria and archaea (Rachel et al. 2002; Prangishvili 2003), is uncommon. Recently, the first complete sequence of a virus, ϕCh1, infecting the haloalkaliphilic archaeon Natrialba magadii, was reported (Klein et al. 2002). The sequence of the halotolerant alkaliphilic bacterium Oceanobacillus iheyensis (Takami et al. 2002) contains a defective prophage, called Bha35X (Mantri and Williams 2004a, 2004b) or Bh1 (Canchaya et al. 2003). Two phages of alkaliphilic bacilli have been reported, phage A1-K-I active against a not further characterized Bacillus sp. (Horikoshi and Yonezawa 1978) and, more recently, the temperate bacteriophage BCJA1, which infects the obligately alkaliphilic species Bacillus clarkii (Jarrell et al. 1997).

Bacteriophage BCJA1c is a member of the Siphoviridae family with B1 morphology. It possesses an isometric head that measures 65 nm in diameter and a noncontractile tail of 195 nm in length. Analysis of the protein composition of the phage revealed approximately ten structural proteins, with the major protein species having apparent molecular masses of 36.5 and 28 kDa. The latter were considered as possibly the major head and tail proteins, respectively (Jarrell et al. 1997). The genome was estimated to be between 32.1 and 34.8 kb in length, with a mol% G+C content of 45.6. This contribution represents the first complete genome sequence of a bacteriophage active against an obligately alkaliphilic bacterium.

Materials and methods

Organism and growth medium

Bacillus clarkii JaD is an obligate alkaliphile isolated from alkaline red mud from bauxite processing waste (Agnew et al. 1995). It was grown at 37°C in the growth medium (pH 10) recommended for Bacillus alcalophilus (Slepecky and Hemphill 1992). The clear plaque mutant of bacteriophage BCJA1, BCJA1c, was used in this study as it routinely grew to a higher titer than the wild-type version of the bacteriophage (Jarrell et al. 1997). Bacteriophage BCJA1c (accession number HER 428) and host B. clarkii (accession number HER 1406) have been deposited in the Felix d’Herelle Reference Centre for Bacterial Viruses, Faculty of Science, Laval University, Que., Canada.

Isolation of bacteriophage BCJA1c DNA

DNA was isolated from CsCl-purified bacteriophage BCJA1c as previously described (Jarrell et al. 1997).

Sequencing procedure

A combination of three procedures was used to determine the sequence of BCJA1c DNA. In the first case, the DNA was partially digested with a mixture of four restriction endonucleases that recognize 4-bp sequences and produce blunt-ended fragments. These were AccII (Amersham Biosciences, Baie d’Urfé, Que., Canada), HaeIII, AluI (New England Biolabs, Beverly, Mass., USA), and HpyF44III (MBI Fermentas, Burlington, Ont., Canada). After preparative agarose gel electrophoresis, fragments of 1.5–3 kb were recovered using Prep-A-Gene matrix (Bio-Rad Laboratories, Philadelphia, Penn., USA) and ligated into pUC18 digested with SmaI and dephosphorylated with bacterial alkaline phosphatase. The constructs were electroporated into Escherichia coli DH5α [F− ϕ80dlacZM15 (lacZYA-argF) U169 recA1 endA1 hsdR17(r − k , m + k ) phoA supE44 λ− thi-1 gyrA96 relA1], and colonies were selected in Luria agar (Difco) containing ampicillin (100 μg/ml, Sigma-Aldrich Canada, Oakville, Ont., Canada) and 40 μg/ml X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactoside). Individual clones were grown in Terrific Broth (Difco), and plasmid DNA was isolated using the alkaline lysis procedure (Sambrook et al. 1989). The DNA inserts were sequenced at the McGill University and Genome Québec Innovation Centre (Montreal, Que., Canada). Gap closure was accomplished using primer walking off the phage DNA at the Robarts Research Institute (London, Ont., Canada) and by PCR amplification with specific primers and amplicon sequencing at the Centre for Applied Genomics (Toronto, Ont., Canada). The sequence was stripped of poor quality and vector data, using SeqMan (DNASTAR, Madison, Wis., USA) and assembled into contigs.

Sequence analysis

Open reading frames were identified using Kodon (Applied Maths, Austin, Tex.). Protein masses and isoelectric points were calculated using EditSeq (DNASTAR). Potential homologues were identified using BLASTP (Altschul et al. 1990) or PSI-BLAST (Altschul et al. 1997) at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). Where homologues were identified, pairs of sequences were compared using the Institute of Human Genetics’ program ALIGN at its Web site (http://www2.igh.cnrs.fr/bin/align-guess.cgi). Conserved protein motifs were identified as part of BLASTP analyses include Pfam (Bateman et al. 2002), Smart (Letunic et al. 2002; Schultz et al. 2000; Hofmann et al. 1999), CDD (Marchler-Bauer et al. 2003), and COG (Tatusov et al. 2003) databases. To predict transmembrane domains, TMHMM (Sonnhammer et al. 1998) at the Center for Biological Sequence Analysis at the Technical University of Denmark (http://www.cbs.dtu.dk/services/TMHMM-2.0/) was employed. Helix–turn–helix motifs were identified using the Pôle Bio-Informatique Lyonnais Network Protein Sequence Analysis server at http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_hth. html.

Potential integration host factor (IHF)-binding sites were assessed using PromScan (Studholme and Dixon, 2003) at http://molbiol-tools.ca/promscan/, while potential transcriptional terminators (Brendel et al. 1986; Brendel and Trifonov 1984) were assessed using GeSTer (Unniraman et al. 2002). Promoters were predicted using Softberry’s BPROM program at http://www.softberry.com/berry. phtml?topic=promoter. Repeat sequences were found using PHIRE (Lavigne et al. 2004).

Nucleotide sequence accession number

The BCJA1c sequence has been deposited with GenBank (accession number AY616446).

Results

DNA sequence analysis

Based upon restriction analysis the predicted size of BCJA1 DNA was 32.1–34.8 kb (Jarrell et al. 1997). DNA sequencing indicates that the unique sequence is 41,092 bp, and that this phage possesses terminal repetitious sequences of about 0.35 kb. For ease of presentation, the nonredundant genomic sequence was opened adjacent to a 33-bp bidirectional ρ-independent terminator (AAAAAAAGAGCCCGGTTAATTCCGGGCTTTTTT) with a calculated ΔG of −7.3 kcal/mol located downstream of gene 59 (lysin).

The overall base composition of the viral DNA (41.7 mol% G+C) was almost 4% lower than the published value (45.6%) determined from the melting profile (Jarrell et al. 1997). The current value is very similar to that of the host bacterium, Bacillus clarkii (42.7 mol% G+C, Nielsen et al. 2004), which is a not unexpected observation, since the base composition of temperate phage genomes usually closely match that of their hosts (A.M. Kropinski, unpublished results).

Description of selected protein coding sequences

Four criteria were used to define potential coding sequences (CDSs): either they had to exhibit sequence similarity to existing genes in the databases or they had to (1) contain >30 codons, (2) be preceded by a sequence displaying similarity to the consensus ribosome-binding site TAAGGAGGT (Shine and Dalgarno 1974, 1975), and (3) usually employ ATG or GTG as initiation codons. As with other phage genomes, the genes of BCJA1c were densely packed with many incidences of overlapping gene sequences.

A total of 59 potential CDSs were discovered, of which 68% encoded polypeptides that showed significant sequence similarity to proteins in the GenBank databases (Table 1; Fig. 1). In approximately 40% of the cases where homologues exist, they are to uncharacterized or hypothetical proteins. The properties of some of the CDSs that had identified function are discussed in the following sections, with special emphasis to the genes involved in replication, morphogenesis, integration, lysis, and regulation.

Table 1 List of gene identified in phage BCJA1c with the properties of the protein products, related proteins and their functions, if known. CDS Coding sequences
Full size table
Fig. 1
figure 1

Genome map of phage BCJA1c with integrase (black), regulatory genes (red), genes involved in replication or recombination (dark blue), DNA packaging (green), morphogenesis (brown), and lysis (pink). Coding sequences with undefined homologues are in light blue, while unique genes are displayed in outline. ρ-Independent terminators are shown as ball-on-stick figures. Regions of homology to Streptococcus pyogenes prophage 370.1 are shown as filled purple bars, while regions with similarity to Bacillus halodurans proteins are illustrated by purple boxes. With the exception of the fourth line, each line in the diagram represents 11.2 kb

Full size image

Replication and recombination

The replication of this phage has not been studied, but it probably involves genomic circularization mediated through recombination between the redundant ends of the molecule, followed by expansion involving θ-type replication and ultimately rolling circle, or σ replication, providing the substrate for packaging. Bacteriophage genomes often contain genes with homology to primases or helicases, and BCJA1 is no exception. Gene 16 specifies an ATP-dependent DNA helicase containing the NTP-binding domain T31 GCGKT36 [TGxGK(T/S), Walker box A, Walker et al. 1982] and a D122 EAH125 motif involved in Mg2+ binding and catalysis (Walker box B). What sets BCJA1 apart from most phages is the presence of a RepA homologue. This protein, often associated with plasmid replication, is involved in binding to the origin of replication and recruiting replication proteins. Only two other phages possess RepA proteins, and these are the Escherichia coli plasmid prophages of N15 (Ravin et al. 2003) and P1 (Chattoraj et al. 1985; Abeles et al. 1984). This suggests that in the lysogenic state BCJA1 also exists extrachromosomally.

RepA binds to iterated sequences (iterons) on N15 and P1 DNA, and these are characteristic of many phage replication origins. The use of base compositional skew analysis has been used to define replication origins (Ori) (Lobry 1999; Kowalczuk et al. 2001; Grigoriev 1998), but unfortunately, in this case it did not prove informative. In addition to iterons, Ori regions contain protein-binding motifs for replication proteins such as DnaA, HU, FIS, and IHF, the latter three of which induce DNA bending (Betermier et al. 1994; Grove et al. 1996; Swinger et al. 2003). A sequence (TTTTCCACA) was found on the minus strand centered at 10780, i.e., within gene 17, which is identical to the consensus DnaA-binding site (TTWTNCACA) (http://www.nmr.chem.uu.nl/~mike/ html/dna.html). There are two strong IHF-binding sites in CDS 16. Numerous direct repeats lie in the CDS 17–22 region including two copies of ATTCGAAGCAT, three copies of GGCAAAAAG, and four copies of TTGAAGGA. These all lie within approximately 4 kb of each other. A site (TCACAGAATTACTCAACAAAAAAGGA) that bears strong sequence similarity to the consensus FIS-binding site (TN2YN2AAWTN7AAWWRA) is found between 9622 and 9647. These finding suggest that the BCJA1 Ori lies somewhere between CDSs 17 and 22.

Other CDSs which may participate in DNA replication or recombination include RecF (CDS 13) and RusA endonuclease (Holliday junction resolvase) homologues. The latter is sufficiently similar to the E. coli enzyme to permit molecular modeling.

Lysis

Bacteriophage-induced lysis usually involves a two-gene lysis cassette composed of a holin and an endolysin (murein hydrolase). The holin creates pores in the inner or cytoplasmic membrane, permitting the endolysin to access the peptidoglycan layer in the periplasm, resulting in cell lysis and release of progeny viruses (Young and Bläsi 1995; Young 1992). With some exceptions the endolysin gene is preceded or overlapped by a gene encoding a holin. The product deduced from CDS 59 is a polypeptide of 355 amino acids that displays sequence similarity to a variety of putative bacterial or prophage lytic proteins classified at N-acetylmuramoyl-L-alanine amidases. This enzyme rather than lysozyme is also used by Streptococcus pneumoniae prophages MM1 (Obregon et al. 2003) and EJ-1 (NC_005294), Bacillus subtilis phages SPβc2 (NC_001884) and SPP1 (Alonso et al. 1997), and Bacillus cereus prophage phBC6A51 (Ivanova et al. 2003).

Holins are characterized by their relatively small size (71–161 amino acid residues), contain two to three membrane-spanning helices and a charged C terminus, and exhibit poor sequence identity to other members of this group of functionally similar proteins (Young 1992; Grundling et al. 2000; Young and Bläsi 1995). The predicted product of CDS 57 is likely the holin, since it is a protein containing 87 amino acids arranged into two transmembrane domains with a high concentration of basic amino acids at the C terminus.

Integration

In the lysogenic state, prophage genomes are mostly found integrated into the host genome. Integration is brought about through site-specific recombination between homologous phage (attP) and bacterial (attB) sites, catalyzed by the phage protein integrase (Int) in conjunction with the host-encoded accessory protein IHF (Campbell 1992). In the case of BCJA1, two proteins show homology to integrases. These are the products of CDSs 3A and 3B, respectively. Since BLASTX analysis of this region of the genome showed that a potential mistake had been made in the sequencing, this region was amplified by PCR and resequenced. However, the sequence proved to be correct. Interestingly, CDS 3B contains a potential ribosome slippage site (Pande et al. 1995; Alam et al. 1999; Harger et al. 2002), which, if utilized, would result in the synthesis of a fusion protein with 369 amino acid residues containing the N terminus of CDS 3B and the C terminus of CDS 3A. Immediately downstream of this site is a stem-loop structure that would result in ribosomal pausing, but it lacks the characteristic pseudoknot structure (Alam et al. 1999; Reeder and Giegerich 2004). While this type of translational regulation occurs in the synthesis of transposase (Gertman et al. 1986), this is the first time that it has been observed in a bacteriophage integrase.

The Int family of recombinases (tyrosine recombinases) has been the subject of considerable research activity, revealing which residues are conserved and which are involved in attP binding, dimerization, and catalysis (Nunes-Duby et al. 1998; Esposito and Scocca 1997; Bankhead and Segall 2000). While these proteins show a poor level of overall sequence similarity, motif analysis against the the CDD database (Marchler-Bauer et al. 2003) revealed the presence of cd01189. This motif is defined as INT_phiLC3_C (phiLC3 phage and phage-related integrases, site-specific recombinases), which contains three conserved oligopeptides, I223 NKTW227, H309 GLRHTHAS317, and Y328 VSERLGHADI338. The active site lies within the tetrad E344 YAH347. The closest homologues to the BCJA1c integrase are to be found in Streptococcus thermophilus bacteriophages Sfi21 (Brüssow and Bruttin 1995) and ϕO1205 (Stanley et al. 1997).

The attP sites contain regions for the binding of Xis, Fis, and IHF proteins. While we found no evidence for a Xis homologue, examination of the region downstream of gene 3 resulted in the identification of two pairs of the sequence TTTTACACA within a 228-bp region, which we propose may represent arm-type integrase-binding sites (Nash 1990). They overlap with two IHF-binding sites bearing strong sequence similarity to the consensus and two potential Fis sites. Finally, attP regions are usually AT-rich, and the 282-bp downstream of int has an average 67% A+T content.

Immunity region

BCJA1 is a temperate phage, and our analysis has revealed that central regulation probably involves, as in coliphage λ, opposing repressor (CDS 5) and antirepressor (cro, CDS 6) genes. Both contain helix–turn–helix motifs associated with DNA binding.

We currently do not know whether the wild-type phage BCJA1 is UV inducible; however, this is doubtful, because the repressor protein lacks both Ala–Gly or Cys–Gly motifs that are associated with RecA-stimulated autodigestion of the repressor proteins in phages such as λ, ϕ80, and P22 (Little 1991; Craig and Roberts 1980; Roberts et al. 1977; Raymond-Denise and Guillen 1991). It also lacks a C-terminal protease domain associated with repressor cleavage and induction.

If CDS 6 encodes the repressor, we would expect that repressor-binding sites might be located nearby. Indeed, we identified two 16-bp hyphenated inverted repeats with the half consensus sequence AGCTAA in the CDS 5–6 intergenic region. In almost all cases, phage operators are 14 [e.g., Mu (Goosen and van de Putte 1987)]- to19-bp [e.g., ϕ80, (Ogawa et al. 1988)] hyphenated palindromes. Since these sites were not found elsewhere in the BCJA1 genome, this also suggests that the major transcripts of this phage originate from the CDS 5–6 intergenic region (Fig. 2b).

Fig. 2
figure 2

a The sequence immediately downstream of the BCJA1c integrase gene showing potential sites involved in integration. b The intergenic region between repressor and antirepressor (cro) genes of BCJA1. Putative repressor-binding sites are boxed, while potential promoters are underlined. Note that there are four P Cro sequences and a single PRep

Full size image

Transcription

The most obvious region for promoters is the intergenic region between the repressor and cro genes. Analysis of this region, using Softberry’s BPROM program and by visual screening, revealed sequences that exhibit significantly similarity to the consensus promoter TTGACA (N15-17) TATAAT (Fig. 2). The repressor gene contains a single potential promoter (P Rep ), while in the case of cro, four promoters have been tentatively identified. Two of the putative promoters (P Rep and PCro4) additionally exhibit the extended −10 region “TGN” (Burr et al. 2000; Mitchell et al. 2003; Fig. 2b). In addition to the bidirectional ρ-independent terminator mentioned above, another terminator (ΔG −6.9 kcal) is located between nucleotides 1129 and 1158, which presumably restricts transcriptional readthrough from gene 2 into the integrase. Three additional terminators were discovered elsewhere in the genome (Fig. 1).

Morphogenesis

Since the genome is terminally redundant, we assume that BCJA1 packages DNA by a head-full mechanism with the terminase complex initiating packaging at a pac site; unfortunately, the location of the latter is unknown. We propose, on the basis of sequence similarity, that the products of genes 33 and 34 encode the small and large subunits of the terminase complex, respectively. Interestingly, while the small subunit shows greatest similarity to prophage terminases in Bacillus and Clostridium species, the large subunit is most closely related to S. pneumoniae phage MM1 terminase. A high percentage of the genes involved in morphogenesis show homology to other phages facilitating their functional identification. CDSs 35–45 are involved in capsid morphogenesis, while CDSs 46–53 are associated with tail assembly. Polyacrylamide gel electrophoretic analysis of denatured BCJA1 virion proteins revealed at least ten structural proteins with masses ranging from 17–120 kDa (Jarrell et al. 1997). The two major proteins were 36.5 and 28 kDa and were predicted to be the major capsid and tail proteins, respectively. In silico analysis of CDS 40 reveals a 34.9-kDa protein with less than 40% sequence identity to the major capsid proteins of S. pneumoniae phage MM1 and Lactococcus lactis phage ul36 (Labrie and Moineau 2002). These results suggest that the BCJA1 capsid protein is not proteolytically modified at the time of or after prohead assembly.

The most likely major tail protein is encoded by CDS 46. Unfortunately, its molecular weight (16.9 kDa) is significantly less than the protein tentatively identified as the major tail protein on the basis of its molecular mass (gp 50, 28 kDa). On the basis of comparative mass, homology, and synteny, we propose that CDS 46 represents the major tail protein of BCJA1.

The product of CDS 53 most probably encodes a host-specificity protein or phage tail fiber. Interestingly, like the tail fiber proteins of the T7-like coliphages, the N-terminal region is much more conserved than the C-terminal ligand-binding domain (Kovalyova and Kropinski 2003). While the closest overall sequence similarity is to a Bacillus halodurans protein, iterative PSI-BLAST analysis revealed relationships in the carboxy region to large proteins in B. cereus phage phBC6A51 (NP_831679), Streptococcus agalactiae prophage λSa2 (NP_688832), Lactobacillus johnsonii prophage Lj965 (NP_958595, Ventura et al. 2004), and Streptococcus mitis phage SM1 (NP_862890, Ventura et al. 2004). In the latter, the similar sequence “pblB” was experimentally characterized as a platelet-binding protein (Bensing et al. 2001a, 2001b).

Discussion

While phages are the most abundant and probably the most diverse life forms on Earth (Rohwer 2003), relatively few viruses have been isolated and even fewer fully characterized against the extremophilic bacteria. This is not due to the lack of bacterial species growing in extreme environments. Quantitatively, the greatest numbers of sequenced tailed phages are for members of the bacterial phyla Proteobacteria and Firmicutes. What has emerged from phage genomic studies is the realization of the extent to which nonhomologous recombination has created genomic mosaics (Hendrix et al. 1999; Hendrix 2002) and has complicated our understanding of their taxonomy (Lawrence et al. 2002; Proux et al. 2002). This is also borne out from our analysis of BCJA1, which one might expect to be ecologically isolated and thus, less subject to horizontal gene transfer.

We initially predicted that most homologues of the BCJA1 proteins would occur among the five sequenced “Bacillus” genomes: B. anthracis (Read et al. 2003); B. cereus (Ivanova et al. 2003); B. halodurans (Takami et al. 2000); B. subtilis (Kunst et al. 1997) and Oceanobacillus iheyensis (Takami et al. 2002) or the fully characterized Bacillus phages, which include B. subtilis phages SPP1 (Alonso et al. 1997), PZA (Paces et al. 1989), B103 (Pecenkova et al. 1997), SPβc2 (Ravantti et al. 2004), and GA-1 (Salas 2004); and B. thuringiensis phage Bam35. This was not the case, though BCJA1 CDS 3–8 are collinear with B. halodurans prophage Bha35X/Bh1 genes BH3551–BH3546. The possible exception is the Cro homologue of BCJA1 (CDS 6), which does not display sequence homology to B. halodurans gene BH3548. Interestingly, the latter encodes a basic protein of 98 amino acids, containing a helix–turn–helix motif. This unannotated gene is spatially arranged relative to the putative Bha35X/Bh1 repressor gene in an identical orientation to that of BCJA1 CDS 6–7, suggesting that BH3548 probably encodes a Cro homologue. The other region displaying B. halodurans homologues is defined by BCJA1 CDS 50 (BH0961) CDS 53 (BH0962), and CDS 59 (BH0963). These genes in B. halodurans are found upstream of an amidase–holin pair, suggesting the presence of another defective prophage in the bacterial genome. This cluster of genes also occurs in O. iheyensis.

The high degree of similarity between Bacillus phage BCJA1c and Streptococcus phage and prophage genes was unexpected. Similarity is particularly evident (Table 1) in the genes involved in DNA replication, recombination, and morphogenesis. This is practically apparent using a dotplot comparison of the nucleotide sequence of BCJA1 with Streptococcus pyogenes prophage 370.1 (Ferretti et al. 2001; Fig. 1), which, in typical phage evolution format, shows blocks of homology separated by regions that do not share a common ancestor. Canchaya et al. (Canchaya et al. 2003) have proposed that prophage 370.1 is a member of the Sf11-like group of Siphoviridae employing pac-site DNA packaging. It has also been shown that the morphogenesis genes of prophage Bha35X/Bh1 are homologous to those of S. pyogenes prophage 315.5 (Banks et al. 2002;Beres et al. 2002). This adds further proof to the argument that horizontal gene transfer has occurred freely among the members of the bacterial phylum Firmicutes.

Proteins achieve conformational stability by means of covalent bonds (Cys-Cys), electrostatic forces, hydrogen bonds, and van der Waals interactions. These are all influenced by extremes of pH. We had hoped that a physicochemical comparison of alkaliphilic versus nonalkaliphilic viral structural homologues would provide us with some understanding of the nature of protein structure and its adaptation to high pH environments. Phage BCJA1 virions are completely stable from pH 6–11, but lose 75% of their titer after 1 h at pH 4 (Jarrell et al. 1997). Using the portal, major capsid, and major tail proteins as indicators, there was no significant difference in the relative concentrations of strongly acidic (D, E), strongly basic (K, R), hydrophobic (A, I, L, F, W, V) or polar amino acids (N, C, Q, S, T, Y) when comparing BCJA1 and its closest homologues. While within the window of pH stability, only the ionic interactions between charged side chains would be affected, other protein stabilizing interactions would function to achieve virion structural stability. These forces function in both alkaliphilic and nonalkaliphilic systems.