Suppressive subtractive hybridization detects extensive genomic diversity in Thermotoga maritima

Abstract

Comparisons between genomes of closely related bacteria often show large variations in gene content, even between strains of the same species. Such studies have focused mainly on pathogens; here, we examined Thermotoga maritima, a free-living hyperthermophilic bacterium, by using suppressive subtractive hybridization. The genome sequence of T. maritima MSB8 is available, and DNA from this strain served as a reference to obtain strain-specific sequences from Thermotoga sp. strain RQ2, a very close relative (approximately 96% identity for orthologous protein-coding genes, 99.7% identity in the small-subunit rRNA sequence). Four hundred twenty-six RQ2 subtractive clones were sequenced. One hundred sixty-six had no DNA match in the MSB8 genome. These differential clones comprise, in sum, 48 kb of RQ2-specific DNA and match 72 genes in the GenBank database. From the number of identical clones, we estimated that RQ2 contains 350 to 400 genes not found in MSB8. Assuming a similar genome size, this corresponds to 20% of the RQ2 genome. A large proportion of the RQ2-specific genes were predicted to be involved in sugar transport and polysaccharide degradation, suggesting that polysaccharides are more important as nutrients for this strain than for MSB8. Several clones encode proteins involved in the production of surface polysaccharides. RQ2 encodes multiple subunits of a V-type ATPase, while MSB8 possesses only an F-type ATPase. Moreover, an RQ2-specific MutS homolog was found among the subtractive clones and appears to belong to a third novel archaeal type MutS lineage. Southern blot analyses showed that some of the RQ2 differential sequences are found in some other members of the order Thermotogales, but the distribution of these variable genes is patchy, suggesting frequent lateral gene transfer within the group.

FIG. 1.

Small-subunit rRNA gene phylogenies of all the Thermotogales strains (with the exception of F. pennivorans, for which there is no sequence available in the GenBank database) included in this study (a) and the strains most closely related to T. maritima MSB8 and Thermotoga sp. strain RQ2 (b). The value on a branch is the number of times the branch was recovered in a bootstrap analysis using 100 bootstrap replicates. Both trees were estimated in PAUP∗ version 4.08ba (53) by using the heuristic search option with 10 stepwise additions. The tree in panel a was estimated by using logdet distances and rooted by the two Aquifex sequences. The tree in panel b was estimated by using Kimura 2P distances and rooted by midpoint rooting.

FIG. 2.

Comparison of G+C contents of clones with (divergent and false-positive clones) and without (differential clones) a DNA match in the T. maritima MSB8 genome.

FIG. 3.

Phylogeny of prokaryotic MutS homologs. Two eukaryotic MutS homologs, human MSH3 from the MutS-I group and human MSH4 from the MutS-II group, are included as references. The tree is a Fitch tree made from distances estimated by using a JTT+ Γ +Ι model in PHYLIP version 3.6 (14). The α parameter and the proportion of invariable sites (I) were estimated in PUZZLE version 4.0 (52). The maximum-likelihood tree calculated by using Proml in PHYLIP version 3.6 (with a substitution matrix provided by E. Tillier [personal communication] and the same Γ + Ι model as used in the distance analysis) and parsimony trees estimated by using PAUP∗ (53) gave very similar topologies. The overall topology also agrees with the tree presented by Eisen (12), who used a larger alignment. In all of these analyses, 10 random additions of the sequences and global rearrangements were used. The value at a node is the number of times the node was recovered in 100 bootstrap replicates. Black dots on branches indicate that the bootstrap support was greater than 90% in all of the analyses, while gray dots indicate greater than 70% support. Only the most conserved part of the alignment was used to construct trees (120 amino acids covering motifs I to IV). Some genomes contain multiple copies of the same type of MutS-encoding genes (the Thermoplasma and Halobacterium genomes), and for these, the gene number is indicated. Archaeal sequences are in boldface type. Some possible MutS homologs were not included in the tree, as they formed extremely long branches in a preliminary analysis (H. pylori and Campylobacter jejuni) or did not contain the part of the sequence used in the alignment (Deinococcus radiodurans).

FIG. 4.

Occurrence of RQ2-specific sequences in other members of the order Thermotogales as observed by Southern blot analysis. (A) V-ATP-B clone probe made from T. neapolitana LA4 clone (Nesbø and Doolittle, unpublished). (B) V-ATP-D from Thermotoga sp. strain RQ2 clone 3a1 (the signal from RKU-1 was weak). (C) Pectin methyltransferase-like protein from Thermotoga sp. strain RQ2 clone 2a10. (D) ABC transporter periplasmic substrate-binding protein from Thermotoga sp. strain RQ2 clone 3F3. (E) MutS3 from Thermotoga sp. strain RQ2 clone D5. (F) RmlC from Thermotoga sp. strain RQ2 clones E9 and 2a4. (G) Probable arabinosidase BH1878 from Thermotoga sp. strain RQ2 clone A8. (H) Probable arabinosidase BH1878 from Thermotoga sp. strain RQ2 clone A12 (the signal from F. nodosum was weak). (I) Methyl-accepting chemotaxis protein from Thermotoga sp. strain RQ2 clone B5. (J) Two hypothetical Methanobacterium thermoautotrophicum genes (MTH323 and MTH324) from Thermotoga sp. strain RQ2 clone 2C12 (the signal from most strains was weak). (K) Alcohol dehydrogenase from Thermotoga sp. strain RQ2 clone 2B8. A schematic representation of the small-subunit phylogeny in Fig. 1 is on the left. F. pennivorans is represented by a dotted line because we had no small-subunit sequences from this species. Since the phylogeny of the different T. neapolitana and Fervidobacterium strains is unresolved, loss or gain involving strains from these groups was counted as one event. For instance, for probe D, we assumed one transfer involving a common ancestor of SL7 and RQ7. For the probes where a single most parsimonious pattern could be resolved, the pattern of gain and loss is shown. Genes gained are in boldface type, and genes lost are in italics.

FIG. 5.

Histogram of the number of RQ2 subtractive sequences with a significant DNA match in the MSB8 genome plotted against percent similarity to the MSB8 homolog. The number of clones is plotted along the y axis, and percent similarity is plotted along the x axis. Low-quality sequences were excluded. Where multiple clones covered the same T. maritima MSB8 gene, an average was calculated. Sequences of clones that covered more than one gene were divided into the respective number of subsequences. This resulted in 222 comparisons between MSB8 and RQ2 coding regions.