A Possible Heterodimeric Prophage-Like Element in the Genome of the Insect Endosymbiont Sodalis glossinidius

Abstract

Extrachromosomal element pSOG3 (52,162 nucleotides) in the genome of Sodalis glossinidius contains redundant phage-related gene pairs, indicating that it may have been formed by the fusion of two ancestral phage genomes followed by gene degradation. We suggest that pSOG3 is a prophage that has undergone genome degeneration accompanying host adaptation to symbiosis.

Bacterial symbionts that live obligately in animal hosts are not expected to provide a significant reservoir for bacteriophages because they have little or no contact with bacteria outside of their host. This explains why the genomes of ancient primary insect symbionts, such as Buchnera aphidicola, Wigglesworthia glossinidia, and Blochmannia spp., are devoid of prophage sequences (1, 5, 11). However, studies have shown that prophage sequences are present in the genomes of recently derived facultative insect symbionts, such as Wolbachia spp., Sodalis glossinidius, and “Candidatus Hamiltonella defensa” (7, 9, 10, 12, 14). In the present study, we report on phage-related sequences found in an extrachromosomally replicating element (pSOG3) in S. glossinidius, a facultative symbiont of the tsetse fly.

The sequence of pSOG3 was obtained by the shotgun method from extrachromosomal DNA isolated by alkaline lysis of a laboratory culture of S. glossinidius from Glossina morsitans morsitans. The pSOG3 sequence was assembled at an average of 11-fold coverage from shotgun reads using the SeqMan assembler in the Lasergene package (DNASTAR, Madison, WI). Open reading frames (ORFs) were predicted using GeneMark (6) and by translating BLAST searches. The sequence of pSOG3 encompasses 52,162 bp comprising 50 putative coding sequences and five pseudogenes (Fig. 1). Based on our analysis, we hypothesize that pSOG3 is an extrachromosomal element that arose from the integration of two distinct ancestral phage genomes and subsequently degenerated in the adaptation towards symbiosis. Another sequence containing phage-related genes (pSG3) has been obtained from an S. glossinidius strain isolated from Glossina palpalis palpalis (4). It comprises only about 40% of the pSOG3 genome (Fig. 1).

FIG. 1.

Representation of the genomes of prophage extrachromosomal elements pSOG3 of S. glossinidius from Glossina morsitans morsitans and pSG3 of S. glossinidius from Glossina palpalis palpalis (4). pSOG3 is drawn to emphasize its composition as two sets of phage genes. The first set extends from the HK620-like virion structural genes at the extreme left to the int-like gene at nucleotide position 30000; the second set extends from the epsilon15-like structural genes at position 33000 to the int-like gene at the extreme right. ORFs are shown as open arrows pointing in the direction of transcription. Filled arrows represent putative pseudogenes. The names given to some of the genes or their putative proteins are indicated above the arrows representing them. Two regions of protein amino acid sequence homology are shown as open rectangles with dashed lines embracing the homologous ORFs. The region of HK620 homology is also homologous to Sf6 and APSE1 (2). A nucleotide scale is indicated at the bottom of the figure.

Phage genomes are described as mosaic because they consist of DNA sequences or modules inherited from distinct ancestral phages (6). Typically, each mosaic phage has a single copy of each of the genes necessary for the phage lifestyle. Thus, pSOG3 is unusual because it has a striking heterodimeric structure composed of two contiguous regions harboring genes derived from two ancestral phages related to extant phages HK620, a relative of P22 (3), and epsilon15 (Fig. 1). The most parsimonious explanation for the formation of pSOG3 is the coalescence or heterodimerization of two ancestral phage genomes, resulting from homologous recombination between orthologous sequences from distinct ancestral prophages. In these circumstances, natural selection would be expected to favor the inactivation of genes that are functionally redundant in the resulting heterodimer. In the case of pSOG3, the region derived from a P22-like ancestor has a complete set of virion structural genes and no replication genes, while the region derived from an epsilon15-like ancestor has intact genes encoding proteins presumed to be involved in replication alongside inactivated structural genes (Fig. 1). Both regions include ORFs encoding putative proteins homologous to RecT, RecE, HU, Int, and the lysis triad holin, lysozyme, and Rz (Fig. 1). One copy of each of the recE and holin genes also appears to have been inactivated.

On the left of the pSOG3 map (Fig. 1) is a segment of 15.2 kb containing 14 ORFs whose putative proteins share substantial identity (50 to 80%) with virion structural proteins of temperate phages HK620 (Escherichia coli), Sf6 (Shigella flexneri), and APSE-1 (“Candidatus Hamiltonella defensa”). Virions of these phages are morphologically similar to those of phage P22, and their virion structural proteins show substantial identity to those of Salmonella enterica phage P22 (2, 3, 15). By comparison, in the middle of the map (Fig. 1) is a region of 7.3 kb that shares substantial sequence identity with a region of phage epsilon15, which is another temperate phage of Salmonella enterica that is distinct from P22 (M. R. McConnell, personal communication; GenBank accession no. NC004775). The putative structural genes in the epsilon15-like region of pSOG3 are clearly pseudogenes, because they have coding sequences that are truncated by frameshifts and/or deletions.

Several lines of evidence indicate that pSOG3 replicates extrachromosomally in S. glossinidius. First, the circular assembly of pSOG3 contains two sequences hypothesized to be involved in replication. One of these is an ORF (repA) whose putative protein product is 27% identical to the RepA protein of plasmid pEA29 from Erwinia amylovora. Upstream of the pSOG3 repA are four imperfect repeats (ACCGCTTTTTGGTCGTC) sharing similarity with the putative replicator designated for pEA29 (8). Downstream of repA is an ORF whose translated protein is 30% identical to the plasmid partitioning protein, ParA, of Corynebacterium glutamicum plasmid pGA2 (13). Southern hybridization experiments also indicate that pSOG3 is not integrated into the chromosome of S. glossinidius (Fig. 2), consistent with the complete sequence of the S. glossinidius chromosome, which does not include the prophage and is derived from the same strain (14). The copy number of pSOG3, estimated by TaqMan real-time PCR, closely matches that of the S. glossinidius chromosome (Fig. 2).

FIG. 2.

Evidence that pSOG3 replicates as a plasmid.

(A) Left panel, pulsed-field gel of S. glossinidius total DNA (lane 1) and plasmid DNA (lane 2) digested with XmaI and S. glossinidius uncut plasmid DNA (lane 3). Right panel, corresponding Southern hybridization of the gel on the left with a radiolabeled probe specific for pSOG3 gn53. Hybridization reveals the presence of a single band in XmaI digests of S. glossinidius total DNA and a band of the same mobility in the plasmid DNA (indicated by arrowheads), indicating that pSOG3 replicates as a plasmid and is not integrated into the chromosome of S. glossinidius. (B) Relative copy numbers of single-copy chromosomal genes and pSOG3 gn52 determined by TaqMan quantitative PCR assays conducted in triplicate with S. glossinidius total DNA were used to determine the copy number of pSOG3. The numbers of pSOG3 gn52 relative to single-copy chromosomal genes fusA (gp52:fusA) and rplB (gp52:rplB) are close to 1, indicating that the pSOG3 copy number is close to that of the chromosome.

Since phage virions have been identified sporadically in cultures of S. glossinidius isolated from G. m. morsitans (C. Dale, unpublished observations), we hypothesize that pSOG3 is the prophage responsible for those virions. It is notable that pSOG3 retains a full complement of ORFs required for virion assembly (Fig. 1), while no other complete set of such ORFs is present in the chromosomal sequence (14). Molecular evolutionary analyses of three structural-protein-encoding virion genes reveal that these proteins have evolved under strong purifying selection, i.e., selection that eliminates variants because they lack activity or selective advantage. Purifying selection is evidenced by low ratios of nonsilent (amino-acid-changing) substitutions (d_N) to silent substitutions (d_S) in pairwise comparisons of homologous sequences among pSOG3, HK620, Sf6, and APSE-1 (Table 1). This is consistent with the preservation of the functionality of structural virion proteins in pSOG3 by natural selection, suggesting that lysis and virion production do play important roles in vivo. However, the conditions required to induce pSOG3 lysis have not yet been defined. Notably, the pSOG3 prophage does not carry any ORFs homologous to the typical regulatory genes (i.e., cI, cro) found in lambdoid phages, although such sequences are found in the S. glossinidius chromosome (14). However, the prophage does maintain three putative coding sequences (gn18, gn26, and gn43) that could serve as transcriptional regulators.

TABLE 1.

d_N/d_S ratios for three structural-protein-encoding virion genes^a

Protein	dN/dS ratio
	Sf6	APSE-1	pSOG3
Capsid
HK620	0.063	0.049	0.063
Sf6		0.064	0.063
APSE-1			0.049
Portal
HK620	0.088	0.117	0.078
Sf6		0.115	0.075
APSE-1			0.135
Gp10
HK620	0.035	0.170	0.178
Sf6		ND	0.175
APSE-1			0.173

^aThe d_N/d_S ratios were calculated using the pairwise Kumar method in MEGA 3.1 (available at http://www.megasoftware.net/). The nucleotide sequences for the following genes encoding the proteins listed were used for the comparison: for HK620, hkbS, hkbQ, and hkbV (3); for Sf6, genes 5, 3, and 8 (2); for APSE-1, ORFs 24, 19, and 28 (15); and for pSOG3, gn05, gn03 and gn07 (this study). Gp10 is the name of a P22 protein that participates in virion tail formation and thus helps stabilize DNA in the phage head. For reference, d_N/d_S ratios of <1 are consistent with purifying selection; the strength of purifying selection increases as the d_N/d_S ratio approaches zero. The d_N/d_S ratio was not computed for the Gp10 homologues shared by Sf6 and APSE-1 because the sequences were too divergent (indicated by ND [not determined]).

Nucleotide sequence accession number.

The sequence of pSOG3 has been deposited in GenBank under accession no. DQ785801.