We have analyzed five complete genome sequences of phylogenetically distant T4-like bacteriophages. This analysis is the first part of an ongoing comparative genomics project on T4-like phages. At present this project has generated single contiguous sequences for 12 divergent T4-like genomes. Of these sequences, five genomes were selected for in depth analysis on the basis of their phylogenetically diversity 10. Among completed genomes that are not dealt with here are the Aeromonas phages 31 and 25, since they are both close relatives of 44RR2.8t and thus do not add significantly to the sequence diversity of the group. Five other genomes are considered draft quality (coliphages RB16 and phi-1, Vibrio phage nt-1, Acinetobacter phage 133, and Aeromonas phage 65) and are not included in this analysis but are available through the Tulane T4-like Genome Website http://phage.bioc.tulane.edu. The five genomes presented here share between 61 and 67 percent amino acid similarity to each other among ~100 conserved open reading frames. T4 is most closely related to RB69, with which it shares 81% amino acid similarity over 207 ORFs. T4 exhibits about the same level of similarity to the other 4 genomes as they do to each other.

A summary of this analysis is presented in Table 1. The sizes of these five genomes range between 164 kb and 233 kb. The genome of Aeh1 had been predicted to be significantly larger than the other genomes, based on pulse field gel electrophoresis of genomic DNA 10. This genome (233234 bp) is in fact nearly 40% larger than the average of T4 and the other four genomes presented here; the genomes of KVP40 7 and P-SSM2 9 are larger still (244 kb and 252 kb, respectively). All genomes have low %GC, although to a lesser degree than T4. ORFs were identified using GeneMarkS 1112 and ORFs orthologous to T4 genes were identified by blastp mutual best hits to predicted proteins in the GenBank accession for the T4 genome. The probable significance of matches was assessed by expected value (E-value) scores. Most ORFs scored well below the 10^-4 cutoff for significant matches. A conserved core of 82 ORFs (T4-like genes) was found in all 5 genomes analysed here. There are 106 T4-like genes conserved among at least 4 of these 5 genomes; Aeh1 shared the fewest of these conserved genes (94) and the average similarity of the T4 orthologs of the conserved genes was lowest in this phage as well (49%). The conserved genes are generally clustered in several large blocks throughout each genome. Interspersed between these conserved blocks are segments containing blocks of predicted novel ORFs, most of which are unique to the genome that harbours them. Novel ORFs represent between 20% and 54% of the total coding capacity of the 5 genomes analyzed.

The conserved genes are generally localized in large clusters. The gene order among the clusters is highly collinear between most phages, as depicted in Figure 1: a higher resolution version is also available (see additional file 1). In T4, early and middle expressed genes are transcribed in a leftward direction (counterclockwise on the circular map), while late genes are primarily transcribed in the opposite direction. The genomes of RB69, RB49, and 44RR display a high degree of synteny with T4 and maintain essentially all of the clustering of related genes seen in T4. Synteny with T4 conserves the gene orientation with respect to time of expression during the infectious cycle. The genome of Aeh1 is also syntenous with T4, although small rearrangements of individual genes can be seen in Figure 1. Only RB43, with at least two substantial genome rearrangements, displays a significant break in synteny with T4 and the other T4-like phage genomes. The predicted transcription pattern appears more complex for RB43, with smaller clusters of genes predicted to be co-transcribed and some orthologs of T4 early and middle genes are transcribed from the opposite strand used in T4 13. A discussion of genes conserved in all T4-like phages can be found in a companion manuscript 13, as well as an earlier work 9.

The T4 genome has 132 predicted ORFs of unknown function. Eleven of these ORFs are conserved among the five T4-like genomes and orthologs to 93 T4 ORFs are found in at least one of these genomes. Although the conserved ORFs were not identified as essential in T4 by genetic methods 14, their preservation among phages suggests that they must be advantageous for survival in nature. In most instances the functions provided by these conserved ORFs remains obscure, but matches to Pfam motifs provide some clues about the function for a few of these ORFs, as shown in Table 2. For example, ORF vs.6 has a highly significant match to the Gly_radical Pfam accession, which is also found in the nrdD anaerobic nucleotide reductase. Thus, the vs.6 gene product may play a role in phage-induced nucleotide metabolism. Another conserved ORF, vs.1, exhibits marginally significant similarity to the SLT lytic transglycosylase domain, suggesting some role in cell lysis. These results corroborate PSI-BLAST matches previously reported for the T4 vs.1 and vs.6 ORFs to lysozyme and glycyl radical domains 15. Overall, the match of vs.1 to the SLT domain is conserved; four of the six phage vs.1 orthologs match SLT with E value <0.05 and the other two orthologs match more marginally, with E< 0.75. The nrdC.10 ORF is conserved in 3 of 6 phages, and all 3 of these match the AAA ATPase motif, with E values ranging from 0.082 to 0.16. Another conserved ORF, 5.4, displays a less probable, although conserved, match to the PAAR membrane associated motif. However, such low probability matches must be interpreted with caution, but they could provide starting points for the identification of the functions for conserved proteins. Functional assignments for vs.1, vs.6, and nrdC.10 were corroborated by BLAST matches to the Conserved Domain database 16. In addition, Conserved Domain BLAST searches identified matches for 4 of 6 tk.4 orthologs to the A1pp phosphatase domain and 5 of 6 nrdC.11 orthologs to the COG3541 nucleotidyltransferase domain.

Only recently has the conserved ORF uvsW.1 been recognized 17 in T4. Previously this sequence was believed to encode the C-terminal 76 amino acids of the UvsW protein. For all 5 of the genomes analyzed here, the coding region corresponding to T4 uvsW was divided into 2 ORFs, uvsW and uvsW.1. Concurrent crystallography on the UvsW protein from T4, showed that it too lacked the region similar to uvsW.1 and subsequent resequencing of this region in T4 confirmed the presence of the two distinct ORFs, uvsW.1 and uvsW 17. Although uvsW.1 is conserved among T4 and all 5 genomes studied here, its function remains unknown.

Each phage genome includes a surprisingly large number of ORFs that have no matches in T4. We term these ORFs "novel ORFs" and their numbers range from 230 in Aeh1 (54% of the genome) to 62 (20% of the genome) in RB69. Similarly, 64 T4 ORFs (15% of the genome) have no apparent ortholog in RB69, its closest relative in this analysis; these 64 ORFs are novel to T4 (see Table 1). Locations of the novel ORFs appear to be non-random, with most clustered in groups between blocks of conserved genes. In a few instances, however novel ORFs are found singly between conserved genes (see Figure 1). The direction of transcription of the novel ORFs is almost invariably the same as flanking conserved genes. This suggests that the novel ORFs are subject to the same regulatory constraints as the rest of the phage genome, with early expressed genes being transcribed primarily counterclockwise and late genes being transcribed clockwise. Nearly 90% of the novel ORFs are clustered among early and middle gene orthologs, suggesting that these genes are expressed at the beginning of the infectious cycle, along with the flanking conserved genes (see also below). The novel ORFs do not appear to differ significantly in codon bias from conserved genes. They share the same strand bias of the third codon position seen in T4 18 and do not vary significantly in codon adaptation index 19 from conserved genes (data not shown). These observations argue that the novel ORFs are not recent acquisitions of host genes.

We searched the sequences of novel ORFs for matches to phage genomes and the Swissprot database by using blastp, and Pfam motifs (HMMer). We identified a total of 750 ORFs from the 5 genomes that lacked T4 orthologs. Of these, only 64 showed matches to Pfam functional domains (Table 3) or to proteins of known function in GenBank. Although novel ORFs are not orthologs of T4-like genes, some appear to be paralogous duplications of adjacent, conserved genes, such as RB69ORF010c with motB, and RB49ORF183c, 44RRORF188c and T4 ORFs alt.-1 alt.-2, with alt. An additional ORF, 44RRORF187c, appears to be a full-length duplication of alt, but displays only 54% similarity to 44RR alt. Although none of the remaining novel ORFs showed any similarity to T4, 89 of them matched other novel ORFs from one of the other five T4-like genomes in this study. A subset of ORFs in phages 44RR, Aeh1, and RB43 appear to be orthologs of a pyrimidine salvage pathway, previously described in the T4-like phage KVP40 7. This pathway includes an NAPRTase and a bifunctional NUDIX hydrolase/nucleotidyl transferase, which is distinct from the monofunctional NUDIX hydrolase, nudE, found in T4 20; nudE orthologs were also predicted for Aeh1, RB43 and RB69. It thus appears that Aeh1 and RB43 possess both the bifunctional NUDIX protein and the T4-like monofunctional NudE protein. It is unclear whether these observations reflect a functional redundancy for RB43 and Aeh1, or if nudE and the bifunctional NUDIX/transferase provide different functions in the phage-infected cell. Conversely, RB49 does not appear to encode either nudE or the bifunctional NUDIX protein.

Several other novel ORFs may be involved in nucleotide modification and synthesis. These include DNA methylase, nucleotidyl transferase, nucleotide triphosphatase and sugar isomerase domain functions identified by Pfam matches. In addition, phylogenetic analyses suggest that phage 44RR appears to have acquired ribonucleotide reductase and thioredoxin genes from a bacterial host, rather than through conservation of the T4-like orthologs 13. A number of the predicted ORFs likely to be involved in gene regulation were also identified, including DNA binding proteins, polyADP-ribosylases and -hydrolases, DNA helicases, an excision repair endonuclease and homing endonucleases, as indicated in Table 3. Other putative functions identified include membrane proteins, peptidases, ATPases, an exotoxin, and a putative DnaJ-type protein chaperone. Several ORFs that do not match known genes in GenBank do match GenBank environmental sample sequences. It is unclear if these matches are to uncharacterized bacterial hosts, or to unknown bacteriophages.

All ORFs were also searched for matches to signal peptide 21 and transmembrane motifs 22. Tables of ORFs matching these motifs for each genome are available (see additional file 2).

The T4 genome encodes a number of mobile DNA elements, including 3 group I introns with integrated ORFs encoding homing endonucleases as well as the freestanding homing endonucleases genes (HEGs), mob and seg 3. No group I introns were detected among any of the T4-like genomes sequenced here. However, two ORFs bearing similarity to the mob genes of T4 were identified in Aeh1 and RB43. An ORF similar to T4 segD has also been described for KVP40 7. Thus, T4 seems to carry many more mobile elements than the genomes analyzed here. Interestingly, both RB49 and RB43 exhibit matches to a recently identified class of HEGs, AP2-HNH mobile DNA elements, which are related to the AP2 DNA transcription factor in plants 23 (also see 13). This class of HEGs has been postulated to have transferred from bacteriophages into plant genomes via the chloroplast genome 23.

The similarities of genome organization to T4 suggested that T4 transcriptional regulatory circuits might be conserved for many T4-like phages in nature. However, phages 44RR and Aeh1 replicate in different hosts than T4 and coliphage RB43 has a substantially rearranged genome compared to the T4 prototype. The relevance of these differences to gene regulation was analyzed by prediction of transcriptional promoter elements in each genome. Consensus nucleotide sequences have been described for three temporal classes of promoters in T4: genes expressed early, middle and late in the infectious cycle 3. Each of the five T4-like genomes was searched for matches to these T4 transcriptional regulatory signals.

The T4 early promoter consensus 3 was used as a start point for identifying sequence similarities in the 5 T4-like genomes using the string search program fuzznuc 24. Matching sequences were scrutinized for their locations relative to the predicted translation initiation site of putative early genes or other ORFs. These sequences were then used in an iterative fashion to find additional sequences using the HMMer program, which develops a statistical model for the consensus with which more refined searches of the genome can be done. Successive rounds of sequence selection and refinement were done until the number and locations of the sequences found ceased to change. From this analysis, we derived an early gene promoter motif for each phage. The locations of the final set of putative promoters on the genome were then manually examined. In virtually all cases, putative promoter elements were identified 5' to a predicted translational start site for a predicted ORF or conserved gene and in the correct orientation for transcription of this ORF. Thus, the predicted promoters appear to be plausible transcription initiation sequences. In each case, the sequences of the presumed early promoters thus identified had similarities to the T4 early consensus, but with some distinct differences that are illustrated in Figure 2. All predicted early promoters had similarity in the -35 region sequence to the GTTTAC sequence (-36 to -31) found in T4 3, but in RB49, RB43 and Aeh1 there was a definite preference for G rather than T at position -33. In T4, this position is believed to be a preferred site of interaction of the ADP-ribosylated alpha subunit of RNA polymerase; a modification that is made in this subunit by the T4 encoded Alt protein 25. Phages RB49 and Aeh1 have putative alt genes, but in both cases the predicted Alt protein sequences are considerably diverged from the T4 sequence (data not shown); RB43 apparently lacks an alt ortholog. Position -36 is a strongly conserved G in some of the genomes analyzed but for RB43 it can be G or C; Aeh1 shows even less sequence conservation in the -36 position. All the phages frequently have an A-rich sequence from -40 to -44. This region resembles the UP element, which enhances transcription and is a site of interaction with the T4 ADP-ribosylated alpha subunit of RNA polymerase 262728.

All putative early promoters resemble the T4 consensus in the -10 region, which is recognized in the host by the σ subunit of RNA polymerase. In general, there is high conservation of T at position -7 and A residues at position -11, as seen in T4. However, in our phage conservation of the T at position -12 is variable; T is not rigidly conserved at position -12 in Aeh1, and in RB49 it can be either T or C. There is variable conservation of the GT-rich sequence 5' to position -12 exhibited by T4. 44RR shows a higher degree of conservation of A at -8 than any of the other phages. The genomes of RB69, RB49, and 44RR all show preference for C residues in the -3 to -1 region. The predicted RB49 early consensus agrees with that previously identified by 5' end mapping of RB49 early transcripts 29.

When the sites of predicted early promoters were mapped onto their respective genomes, many promoters were located 5' to orthologs of T4 early genes, as expected. Importantly, a large number of early promoters were predicted 5' to novel ORFs, including those for which no homologs exist in the sequence databases. For example, of 57 putative early promoters in RB69, 13 were upstream of novel ORFs and 45 were upstream of T4 orthologs (see example in Figure 3). These observations suggest that many novel ORFs are coordinately regulated along with the flanking conserved early T4-like genes. Early promoters were also found 5' to the tRNA genes, described below. Coordinates of putative early promoters can be found in the supplements (see additional file 3).

In the T4 infectious cycle, early transcription is followed by "middle mode" transcription, which is initiated by the binding of the phage-encoded MotA protein to its cognate recognition sequence at T4 middle promoters 303132. We used two criteria to attempt to detect conserved elements of T4-like middle mode transcriptional regulation among the five genomes studied: (a) matches to the T4 middle promoter consensus 33 and (b), matches to the T4 MotA protein sequence. The RB69 genome includes a motA ortholog (blastp E = 5X10^-48). Putative RB69 middle promoter sequences were identified using a similar strategy to that described for early promoters, but based upon the consensus sequence, (a/t)(a/t)(a/t) TGCTTtAN(11–13)TataAT 33 The RB69 middle consensus clearly resembles that of T4 (Figure 4A); with conservation of the residues at positions -12, -11, and -7 of the T4 consensus. Also, the putative RB69 middle genes exhibit extended conserved sequences from positions -13 to -16, as seen in T4. T4 middle promoters show little similarity to the -35 region of E. coli σ⁷⁰ promoters, but do possess the highly conserved GCTT motif (the T4 Mot box) at positions -30 to -27. This motif serves as the site of interaction of the T4 MotA protein with DNA. RB69 middle promoters also show similarity to the Mot box, which is presumably bound by the RB69 MotA ortholog. However, among the 4 other genomes studied, only the 44RR genome had an ortholog to the T4 MotA protein and sequence motifs similar to the T4 MotA-dependent promoters. Nine putative 44RR middle promoters were identified. They resemble the middle-mode consensus sequences of both T4 and RB69, but lack conservation at nucleotide position -11 (Figure 4A). The relatively small number of putative middle-promoters that we have detected in 44RR tempers the interpretation of their significance. However, the presence of a strong match (blastp E = 2X10^-33) to the T4 motA gene function in this Aeromonas phage is probably indicative of the presence of a 44RR-encoded middle-mode transcriptional apparatus. Previous attempts to identify a middle promoter consensus and a motA ortholog in RB49 were unsuccessful 29 as were our attempts for RB49, RB43 and Aeh1. RB69 and 44RR also possess orthologs of the MotA co-activator AsiA 34. Surprisingly, Aeh1 and KVP40, also encode AsiA proteins, which have been shown to bind T4 MotA 35, even though no ligand homologous or analogous to MotA has been identified for these genomes. AsiA can act as transcriptional inhibitor in the absence of MotA 35, or may interact with another phage protein which has yet to be identified. Coordinates of putative middle promoters can be found in the supplements (see additional file 4).

In T4, late promoters are recognized by a phage-encoded σ factor, gp55. Contact between T4 gp55 and the DNA is facilitated by the T4 polymerase sliding clamp, gp45. A third T4-encoded gene product, gp33 forms a bridge between gp55 and gp45 36. The T4 late promoter consensus sequence is a short but highly conserved motif, TATAAATA, between nucleotide positions -13 and -6 relative to the transcriptional start site 3. Putative late promoters were found readily for four of the five phage genomes studied, using the strategy employed for early and middle promoter searches (Figure 4B). However, the T at position -13 was poorly conserved for most phages, with either A or T commonly found at this position. A similar observation was made for late promoters in an earlier description of RB49 late promoters 29, as well as in KVP40 7 and S-PM2 8.

Since our search strategy failed to detect late promoter sequences for phage Aeh1, an alternative strategy was employed to identify them. Regions upstream of ORFs orthologous to T4 late genes were analyzed with the ELPH program 37 to identify sequence motifs common to these DNA segments. The selected motifs were used as seed to identify additional late promoter sequences using HMMer. This strategy identified a conserved sequence, CTAAATA, beginning at -12 from the putative initiation site. Once identified, this putative promoter sequence was used as a seed for string search followed by HMM refinement used for late promoters of the other phages. Although the C at position -12 is a strong determinant for detection of Aeh1 late promoters, C is rarely found at this position in the putative late promoters of the other four phage genomes (Figure 4B). It should be noted that the phage Aeh1 gp55 protein, which presumably recognizes the divergent late promoter sequences of Aeh1, is itself substantially diverged from all the other phage gp55 sequences (data not shown). Coordinates of putative late promoters can be found in the supplements (see additional file 5).

Putative rho-independent terminator sequences were identified for all 5 genomes, using the TransTerm program 38. Although the locations of putative terminator sequences vary between phages, several terminators appear at conserved locations (see additional file 6). One striking example is the bi-directional terminator predicted downstream of uvsW.1that is conserved in T4 and the other 5 genomes. In all cases, the gene downstream of uvsW.1 is transcribed from the opposite strand and a bidirectional terminator is predicted between the converging transcripts. Genes 35 and 36 are transcribed rightward and a predicted terminator is located between them in all 6 genomes. Likewise, gene 23 has a terminator predicted downstream in all 6 genomes. Terminators conserved in 5 out of 6 genomes were identified downstream of Gene 32 and upstream of alt.

Comparisons between the positions of predicted terminators and transcription initiation signals allowed the identification of putative operons of gene expression. An example of operon structure from phage RB69 is shown in Figure 3. In some instances, it appears that the upstream promoters of novel genes drive expression of T4-like early genes that lack their own early promoter. In general, T4-like genes are predicted to be in operons with other T4-like genes, while novel ORFs appear to reside in operons with other novel ORFs.

The bacteriophage T4 genome encodes eight tRNA genes 3. The other T4-like genome sequences were searched for potential tRNA genes, using tRNAscan-SE 39. The number of potential tRNA genes varied considerably among genomes (Table 1), ranging from zero in RB49 to 24 in Aeh1. Some common features were noted among the tRNA genes encoded by the phage genomes (Table 4). All genomes that encoded tRNAs had a predicted tRNA with a CAU anticodon. Although predicted to be Met tRNA by tRNAscan-SE, these tRNAs share signature sequences found in tRNAs recognized by IleRS 40. This class of Ile tRNAs is post-transcriptionally modified to lysidine at the anticodon, converting them to Ile-recognizing anticodons resembling AUA 41424344. An alignment of phage Ile and Met tRNAs is shown in Figure 5. tRNAs for Leu, Ser and Arg are among the most commonly identified putative tRNAs genes encoded in the T4-like genomes, including the previously sequenced genomes of T4 and KVP40. Other tRNAs are found more rarely, such as Ala, Pro, Gly and Val. These recognize GC rich codons, which are unusual in AT-rich T4-like genomes 18.

In bacteriophage T4, the presence of tRNA genes appears to correlate with differences in codon bias for the phage versus the E. coli host 3. The genomes sequenced here show much less correlation to differences from their laboratory hosts. A similar observation was made for the vibriophage KVP40 7. Thus, the functional role of the tRNA genes for these phages remains unclear. Nevertheless, the high degree of conservation of some tRNAs, such as the putative modified tRNA^Ile mentioned above, suggests an important functional role for at least some of these tRNAs.

Bacteriophages, bacterial hosts and growth conditions were as described 13. Phage DNA was prepared from plate lysates sequenced, and assembled as described in 13.

ORFs were detected primarily by use of the GeneMarkS program 1112. The program was chosen based upon its accuracy in ORF prediction of the T4 genomic sequence by comparison to the GenBank accession (97% of ORFs recognized). When an orthologous gene was detected in a related phage genome, the predicted translational start sites were scrutinized for additional N-terminal protein sequences with significant similarity to orthologs upstream of the predicted translational start site. In these cases, the translational start site was adjusted to maximize the length of predicted amino acid similarity. Although prediction models were not based upon similarity between genomes, generally fewer than 5% of the predicted start sites required adjustment.

GeneMarkS predictions were compared with those obtained using Glimmer 49. There was general agreement between the predictions obtained with the two programs. Glimmer predicted more ORFs per genome, but in some cases the additional ORFs predicted were inconsistent with the direction of transcription of flanking genes, which is uncommon in T4 3 and appears unusual for the genomes sequenced here. Thus, the Glimmer predictions were used primarily to adjust GeneMarkS predictions as mentioned above, or in regions where Glimmer predicted an ORF and GeneMarkS predicted an unusually long (> 200 bp) intercistronic region.

Predicted ORFs were checked for similarity to T4 genes by blastp 50 mutual similarity. Genes with mutual best hit E-values < 10^-4 to known T4 genes were designated by the T4 gene name. Putative genes without T4 orthologs were designated by their ORF numbers, with conserved gene rIIA designated as ORF001. The strand of each ORF is designated "w" for clockwise (left-to-right) transcribed genes, and "c" for counterclockwise (right-to-left) transcribed genes. In T4, the origin of the genome has been assigned to the rIIB – rIIA intercistronic region; the terminus of the genome is defined as the start of translation of the rIIB gene. The sequence origin of each genome sequenced here is defined as the termination codon of the rIIA gene.

Genomes were also searched for tRNA genes using tRNAscan-SE 39. All genomes except that of RB49 had at least one putative tRNA gene.

DNA sequences are available through GenBank [Genbank:NC_005135] (44RR), [Genbank:NC_007023] (RB43), [Genbank:NC_004928] (RB69), [Genbank:NC_005260] (Aeh1), and [Genbank:NC_005066] (RB49). Additional analyses are available through the Tulane T4-like Genome Website http://phage.bioc.tulane.edu Available data include an interactive genome browser 51, clustalW 52 alignments, EMBOSS pepstat statistics, octanol hydropathy plots 24, and HMMer Pfam matches 53.