Bacteriophage Genomics

Introduction

Because of their relatively small size and simplicity of isolation, bacteriophages were the first complete genomes to be sequenced, beginning with the 5,386bp single-stranded DNA (ssDNA) phage φX174 in 1977 [1]. The first complete sequence of a double-stranded DNA phage was that of the lambda 48,502bp genome [2], and the sequence of the 39,936bp phage T7 genome was reported shortly after [3]. The first complete sequence of a dsDNA tailed-phage genome of a virus infecting a non-Escherichia coli host was that of mycobacteriophage L5 almost ten years later [4], and the number of sequenced bacteriophage genomes has exploded since that time as DNA sequencing methodologies have advanced.

The application of epifluorescence techniques to the enumeration of phage particles in environment samples in the late 1980′s [5-7] contributed to a significant re-thinking of the roles of bacteriophages in biology and in the environment. Viral densities in ocean samples are 10⁶ − 10⁷ particles per milliliter, and given the size of the oceans and determinations of similar terrestrial viral populations [8], the total phage population is estimated to be around 10³¹ particles [9,10]. This is validated by independent estimates of 10³⁰ bacterial cells in the biosphere [11], along with the finding that environmental viral-bacterial ratios are typically 5−10:1 [10,12]. The size of this viral population is stunning, and suggest that prokaryotic viruses are a majority of all biological forms [13]. Furthermore, experimental studies show that the population is highly dynamic, with an estimated 10²³ infections/second globally [12].

The advances in bacteriophage genomics have been fueled with two primary goals. First, the abundance and turn over of the phage population poses questions as to how phage genomes are related to each other and to their hosts, and what evolutionary mechanisms shaped this population. Secondly, genomics has advanced the use of phages for development of genetic, biotechnological, and clinical tools, and a large variety of utilities and approaches have been described. In this review I will largely focus on what we have learned about the structure and content of bacteriophage genomes, and how they have evolved, focusing on developments since the review by Casjens in 2005 [14]. Several additional reviews and resources on phage genomics might be of interest [15-20].

Sequenced bacteriophage genomes

At the time of writing, the NCBI phage genome database contains 500 completely sequenced genomes, corresponding to a total of 22.3Mbp of sequence information. Although phage genomes are on average only approximately 1% the size the bacterial chromosomes, the number of sequenced bacterial genomes in the NCBI bacterial chromosome list (741) is substantially greater in number and a 100-fold larger in total sequence information (2.42Gbp). This is somewhat surprising given the high diversity of the phage population and its abundance, and it seems likely that the next few years will see a substantial shift in the ratio of sequence phage and bacterial genomes. There are a large number of resident prophages in sequenced bacterial genomes [21,22], possibly outnumbering sequenced phage genomes, although the proportion of these retaining potential for lytic growth is unclear.

Sequenced phage genomes vary considerably in size from Leuconostoc phage L5 (2,435bp) to Pseudomonas phage 201phi2−1 (316,674b). However, there is not a uniform distribution of genome sizes across this spectrum (Fig. 1), and there is a predominance of genome sizes in three separate ranges. The largest is a peak of genomes in the 30−50kbp interval, corresponding to nearly 50% of all phages (Fig. 1). A second group (abut 20% of total) is those smaller than 10kbp, and the third group contains those in the 100−200 kbp interval (6% of the total). This overall distribution likely reflects the phage isolation methods employed and the sequencing technologies available at that time, and is not a true reflection of size distributions of phages in the environment. For example, the group of genomes smaller than 10kbp has a median size value of only 5.5kbp and many were characterized prior to automated sequencing methodologies. In contrast, the dearth of larger phage genomes (>250kbp) is not from technological availability, but because bacteriophages with larger heads and corresponding larger genomes form extremely small plaques on standard agar plates and are thus easily overlooked [23]. The use of alternative microbiological substrates and electron microscopy suggests there are many large phages in the environment that have yet to be explored.

The predominant bacteriophage morphological forms are tailed phages containing dsDNA genomes. However, few of these if any have genomes smaller than 10kbp, and the vast majority are larger than 15kbp. This is consistent with the finding that the virion structure and assembly genes typically encompass at least 15 kbp of genome space; moreover, the Siphoviruses (characterized by a long flexible non-contractile tail) all contain a tapemeasure protein gene whose length corresponds to the length of the phage tail, and ranges from 1.5 − 6 kbp in length. Therefore the majority of phages with Siphoviral morphologies have genomes longer than 20 kbp. In contrast, most of the phages with larger genomes (>125kbp) are Myoviruses (with contractile tails), and the largest Siphoviral genome is that of Bacillus phage SPBc2 (134,416 bp). The reasons for the lack of larger Siphoviruses is not clear.

Phage genome sequence diversity

The genetic diversity of the bacteriophage population is remarkable. In general, the nucleotide sequences of genomes derived from phages with non-overlapping host ranges rarely share sequence similarity, although this may not be surprising if the bacterial hosts are distantly related. For example, there is no obvious or extensive nucleotide sequence similarity between the four published genomes of Streptomyces phages and a collection of 50 mycobacteriophage genomes, even though the Mycobacterium and Streptomyces hosts both are members of the Actinomycetales. Presumably, phages of non-overlapping hosts enjoy at least a transient genetic isolation from each other.

Since bacteriophages infecting a common bacterial host are in genetic contact with each other, it is not surprising that they sometimes share common nucleotide sequences. More than 30 phage genomes have been isolated for Pseudomonas (33), Staphylococcus (48) and Mycobacterium (50) hosts, and there are many examples where phages of a common host enjoy related sequences [24-26]. Perhaps somewhat more surprising is that many phages within these groups share little or no sequence similarity, as illustrated by nucleotide sequence comparisons of mycobacteriophages and Pseudomonas phages (Fig. 2). In general, phages with different virion morphotypes have different genome organizations and greater sequence diversity; genome architecture may therefore impose constraints on genetic exchange. It is noteworthy that the high genetic diversity among phages of a common hoss coupled with the still limited number of available sequences suggests there is an abundance of new viral genome sequences as yet unidentified.

Phage genome architectures

Comparative analysis of phage genomes reveals some common architectural themes. In general, in phages with Siphoviral morphotypes there is clear synteny among genes encoding the virion structure and assembly functions [14]. The genes for head and tail assembly are arranged together with the head genes 5′ to the tail genes. The head genes typically include one or two terminase subunits, the portal protein, a prohead protease, a scaffold protein, and the major capsid subunit; the tail genes include the major tail subunit, two overlapping open reading frames expressed via a programmed translational frameshift [27] and the tapemeasure protein, followed by the minor tail proteins [14]. The length of the tapemeasure protein gene corresponds to the length of the phage tail, and it thus commonly the largest gene in the genome [28,29].

The syntenic relationship of the virion structure and assembly genes is conserved in phages that not only have no nucleotide sequence similarity but also in genomes in which none of the predicted protein amino acid sequences are detectably related [30], and presumably represents a relatively early feature in bacteriophage evolution. This notwithstanding, there are many architectural elaborations of this theme. For example, in the mycobacteriophages, the lysis genes are found in different locations relative to the structure and assembly genes. In some phages [e.g. L5; [4]], the lysis genes are to the left of the terminase gene, between terminase and the leftmost cos site; in other phages [e.g. TM4; [31]], the lysis genes are located to the right of the minor tail protein genes. Many phages also contain a cassette of integration functions (int, attP), which are typically located close to the center of the genome, even though genome length can vary by more than two-fold [32]. In one instance, the integration cassette appears to have relocated to a position within the tail genes [33], a process that might have occurred though use of a secondary attachment site within the genome. Comparison of mycobacteriophage genomes shows that while the structure and assembly genes typically occupy the ‘left arm’ region between a centrally located integration cassette and the leftmost cos site, the size of this region can vary by more than two fold [29]. In the largest instance (phage Omega), the gene order of the structure and assembly genes is conserved, but there are numerous additional genes of unknown function distributed between them [29].

Phages with somewhat larger genomes (>125kbp) have much less well conserved arrangements of gene functions, although there is still considerable grouping of virion structural genes. Many of these – including phages of several different hosts – are related to phage T4 and these contain conserved core groups of genes that are interrupted with hyper plastic regions [34]. The genes in these hyperplastic regions are relatively small and mostly of unknown functions, although they include a subset of virion internal proteins that are implicated in adsorption and protection from DNA modification by the host [35].

Genome mosaicism and its origins

One of the most striking features of bacteriophage genomes is their apparent mosaic structure; in essence, each genome can be considered as a unique combination of modules that are exchangeable among the population. The size of the modules, their rates of exchange, and the phage genomes carrying them all vary greatly, with phages of different virion morphology, size, and host-range all participants in an orgy of recombination [36]. This mosaicism is by no means unique to the phage population and is also prevalent in bacteria, where genes are acquired by horizontal genetic exchange (predominantly transduction, transformation and conjugation) [37]; but the extent of mosaicism in phage genomes is remarkable and has become even clearer as the number of genomes available for comparative analysis has grown.

Phage genome mosaicism can be viewed at two levels. The first is by comparison of nucleotide sequences, described initially by DNA heteroduplex mapping [38], and subsequently by sequence comparison [39]. At the resolution of DNA sequence information, precise junctions can be observed corresponding to the boundaries of two DNA segments that clearly have distinct evolutionary origins. One of the most notable features of these functions is that they predominantly correspond to the boundaries of open reading frames (Fig. 3A). There are two types of models to explain the recombination mechanisms that give rise to these patterns. The first proposes that there are short conserved boundary sequences at gene junctions that serve to target exchange events catalyze by homologous recombination, using either host- or phage-encoded recombinases [40]. Some examples of short conserved sequences have been reported [41], although their role is not clear, and this model would not seem to account for the vast majority of exchange events. A second model proposes that the recombination events are not targeted and occur randomly (or with only short sequence preferences), such that most events give rise to non-functional genomic trash [13]. Most of the progeny would not be viable, except for those that have an appropriate genome size and that retain gene functions, thus accounting for the correlation with gene borders. Productive genetic exchange may require multiple recombination events, and while the overall process is expected to occur at low frequency, this process is potentially highly creative, yielding new combinations of genes, as well as new combinations of protein domains. A recent report suggests that homeologous recombination mediated by phage-encoded RecE/T-like recombination systems may play an important role in generating mosaicism [42], and bacterial acquisition of CRISPR sequences could represent an additional mechanism [43].

Comparing the predicted amino acid sequences of phage gene products generates an alternative manifestation of mosaicism [20,24,29]. This is an informative approach since many groups of phages – including those that infect common hosts – may not share any nucleotide sequence information; protein sequence data reveals genes that share much older ancestry. Sequence comparison and phylogenetic reconstruction shows that different genes, groups of genes, or segments of genes, all share different ancestry, and thus represent modules within a mosaic genome. However, the phylogenetic comparisons can be complex because the diversity is sufficiently high that none or few of the participating genomes in the analysis are in common. One alternative method or representing these relationships is through the use of phamily circles [24] in which all of the participating genomes are represented, even those that do not contribute to that particular phylogenetic relationship (Fig, 3B). The extent of phage genomic mosaicism thwarts the simple determination of phylogenetic relationships of whole phage genomes as units of evolution, and their histories can best be considered as the totality of the evolutionary routes taken by each constituent module.

Phage metagenomics

Phage metagenomics provides an alternative view of phage genome structure that is different but complementary to whole genome sequence determination. A variety of phage metagenomic studies have been reported [44] and one of the most notable features is that they contain a high proportion of unknown sequences, consistent with the high genetic diversity in the completely sequenced genomes. These studies offer considerable insight into the geographic distribution of phages and phage types, but as yet provide relatively little insight into whole phage genome structure and evolution, because rarely do whole genome sequence emerge. However, phage metagenomics is well-suited to the ultra high throughput sequencing methodologies just now becoming available, and large scale viral metagenomics will be a crucial asset in understanding phage genomics in the next few years.

Conclusions

In summary, enormous advances have been made in phage genomics over the past few years and while we do not yet have detailed answers as to the extent of phage diversity and evolutionary pathways, the questions have become clearer and the groundwork well established. The technologies for sequence determination are no longer limiting and it is reasonable to expect a deluge of new information in the next few years. Hopefully a clearer picture will emerge of phage genomic structure along with new questions as to what functions the many newly discovered phage genes do, and how they may be exploited for biological and biotechnological ends.