New insertion sequences of Sulfolobus: functional properties and implications for genome evolution in hyperthermophilic archaea

Recovery of active ISs from natural Sulfolobus populations

Heterotrophic thermoacidophiles were isolated from Yellowstone National Park and Lassen Volcanic National Park in the western USA, and the Kamchatka peninsula of eastern Siberia (Whitaker et al., 2003). Although the clonally purified strains examined in this study exhibited a range of growth phenotypes, they appeared, by molecular criteria, to be conspecific. For example, nucleotide sequences representing the 5′ end of the pyrB gene, the pyrEF promoter, the pyrE gene or the pyrF gene from several representative isolates all yielded pair-wise nucleotide identities of about 99%. This nucleotide divergence lies well below that observed within most prokaryotic species as typically defined (Palys et al., 1997). In addition, a 1400 nt interval encompassing this region was found to be highly divergent from the corresponding chromosomal regions of S. tokodaii (55.2% nt identity) and Sulfolobus acidocaldarius (55.5% nt identity), less divergent from S. solfataricus strain P2 (90.9% nt identity), and nearly identical to Icelandic isolate REN1H1 (98.9% nt identity over the pyrE gene) (Zillig et al., 1994). These relationships agree with the extensive multilocus sequence typing of 75 similar isolates drawn from this same collection (Whitaker et al., 2003). The multilocus analyses indicated that the corresponding Sulfolobus isolates from these regions are conspecific with those isolated by Zillig and co-workers from Iceland and informally designated ‘Sulfolobus islandicus’ by them (Zillig et al., 1994).

In order to screen for transpositionally active ISs, liquid cultures were plated on growth medium containing 5-fluoro-orotic acid (FOA) plus uracil to select spontaneous loss-of-function mutants of the pyrE and pyrF genes or their common promoter. The corresponding region of the chromosome of the mutants was then amplified by polymerase chain reaction (PCR) to detect enlarged loci. This is the approach used by Martusewitsch et al. (2000) for S. solfataricus strain P1, and represents an example of the more general ‘gene trap’ strategy for recovering ISs (Gay et al., 1985; Solenberg and Bergett, 1989; Szeverenyi et al., 1996; Beck et al., 2002). Enlarged loci were common among the mutants; most were 0.8 to about 2 kb larger than the native gene, although a few cases of enlargement by less than 0.2 kb were observed.

Molecular features of insertion sequences

Sequencing of enlarged PCR products revealed seven distinct ISs of differing lengths and structural features, which are described in Table 1. Sequence analysis permitted identification of putative open reading frames (ORFs), and blast searches using predicted transposase primary sequences enabled most of the new ISs to be assigned to families that have been identified previously (Mahillon and Chandler, 1998). The molecular features of the seven ISs may be described as follows.

ISC735.

ISC735 represents the smallest IS yet identified in Sulfolobus. blast searches revealed no significant similarity to any catalogued nucleotide sequences other than two small fragments of 77 and 25 nt in the S. solfataricus genome. A single putative ORF occupies 88% of the element's length and is predicted to encode a transposase of 214 amino acids. A region between aa41 and aa208 was found to encode a conserved COG3316 protein domain (Fig. 1). This domain is characteristic of transposases related to IS240-A, a member of the IS6 family. Phylogenetic analysis of transposases accordingly placed ISC735 into this family (Fig. 2).

Non-IS insertions

In the course of screening spontaneous mutations by PCR, mutant loci were sometimes observed to be enlarged only slightly. In most of these cases, sequencing revealed tandem duplications of the native sequence; similar duplications are relatively common among spontaneous pyrE mutations in S. acidocaldarius (Grogan et al., 2001). In one case, however, sequencing revealed an insertion of 138 bp near the 3′ end of the pyrE gene, which included a 10 bp duplication of the site of insertion. The 128 bp segment within this TSD had ends very similar to those of ISC1057 and ISC1058 (Fig. 3A), but the rest of the insertion exhibited only low similarity to other portions of these ISs (data not shown). This organization, in which the ends correspond to those of a known IS, but the central region does not, is a defining characteristic of type II MITEs (Oosumi et al., 1996).

Using the 128 bp sequence, blastn identified two short sequences of the S. solfataricus P2 genome that are nearly identical to each other and marginally similar to the 128 nt MITE (P-values of about 10⁻⁶); these were later determined to be two copies of SM3A (Redder et al., 2001). No other prokaryotic genomes, including those of S. tokodaii and Aeropyrum pernix, yielded blastn matches. The first 11 bp of the two S. solfataricus sequences matched the 128 nt MITE exactly and represent the region of most extensive identity to it (data not shown). The two S. solfataricus sequences exhibited only low similarity to ISC1057 and ISC1058b.

Rates of transposition

Arguably the most important functional property of a TE is its ability to transpose to a new region of DNA. To our knowledge, however, rates of transposition have not been measured for any TE of HA. In order to quantify the rates of transposition of the new Sulfolobus ISs, we first measured rates of mutation to FOA resistance in the corresponding host strains, using small-scale fluctuation tests. These rates represent the sum of all spontaneous mutations that inactivate either the pyrE gene, the pyrF gene or their common promoter (Martusewitsch et al., 2000; Thia-Toong et al., 2002). Using PCR, we then determined the proportion of these loss-of-function mutations that resulted from IS transposition into these loci. The rates that we measured (Table 2) ranged from about 1 × 10⁻⁸ to about 2.6 × 10⁻⁶ insertions per cell division and were often characteristic of the IS. ISC796, for example, transposed at relatively low frequencies in all tests, whereas ISC1058b exhibited uniformly high frequencies. The data also suggest variation among host strains, however, as seen by the fact that ISC1205 transposed at low rates in two of the strains evaluated but at high rates in the other three (Table 2). A genetic consequence of the variation in rates is that gene inactivation (spontaneous mutation) was dominated by IS transposition in some isolates but not in others.

Target-site selectivity

Another functional property affecting the impact of TEs on a host genome is the specificity that determines the site of an insertion. To evaluate this property, we selected a number of independent transpositions into the mutational target region of about 1250 bp (defined by the pyrEF promoter and the two structural genes) and identified the sites of insertion. The locations of the independent insertions (Table 3) showed that the Sulfolobus ISs differ both quantitatively and qualitatively with respect to the site specificity of their transposition. ISC735 was the most specific; only one target site was used in a total of six independent transposition events analysed by sequencing. ISC1205 and ISC1057 exhibited lower selectivity, using certain sites repeatedly but not exclusively.

In addition to exhibiting differing levels of sequence specificity, these ISs also exhibited differences in the nucleotide sequence of the preferred target. Except for the closely related ISC1057 and ISC1058b, target sites used by the different ISs were all distinct, so that no insertion site was used by two unrelated ISs (Table 3). In addition, alignment of the different sites of insertion did not reveal any obvious sequence motifs common to the various ISs (Table 3). However, the data did suggest that certain regions of DNA attracted transposition that was not specific to a particular nucleotide sequence. For example, various target sites used by ISC1057 and ISC1058b tended to cluster in an interval 15–20 nt ahead of the pyrE coding sequence, into which ISC735 also inserted (Table 3). By comparison to the homologous region of the S. acidocaldarius chromosome (Thia-Toong et al., 2002), this interval is about halfway between the pyrEF promoter (boxA) and the mRNA start site.

Precise excision

In bacteria, the various processes that remove TEs are broadly classified as either ‘precise’ or ‘imprecise’ excision. Precise excision, which includes all events that delete the IS and one equivalent of any TSD, is not necessarily more frequent than imprecise excision, but it has important consequences for the host because it restores the sequence, and therefore function, of the interrupted gene. Precise excision accordingly represents the only mechanism that completely reverses an insertion mutation in one step. As non-essential genes inactivated by TEs can be nevertheless beneficial to the host, correspondingly strong selective pressure can exist for precise excision of these insertions which does not apply to imprecise excision. Precise excision also differs mechanistically from imprecise excision in being mediated in most cases by host functions rather than TE-encoded functions (Eigner and Berg, 1981; Lundblad et al., 1984).

Despite this special genetic significance, precise excision had not been investigated, to our knowledge, in any of the HA. We therefore assayed for the precise excision of several ISs by selecting phenotypic revertants of corresponding pyrE insertion mutants in fluctuation tests. As summarized in Table 4, the observed rates of precise excision were below detection for all but ISC1926. This result was unexpected, given the molecular diversity of the ISs being evaluated (1, 2; Table 1) and the potential of these assays to detect reversion rates below 5 × 10⁻¹⁰ per cell division. The result could not be attributed to unfavourable plating conditions: (i) the same cultures yielded colonies on uracil-supplemented medium with high efficiency of plating under these conditions, (ii) selective plates were incubated twice as long as necessary to yield visible colonies on the same medium supplemented with uracil, (iii) the revertant colonies grew to a readily visible size (>1 mm) under these conditions and (iv) cross-feeding caused faint haloes to develop around the revertant colonies in the later stages of incubation, demonstrating that the pyrimidine-starvation conditions were not lethal for the cell population over this period of time. The lack of phenotypic revertants could also not be attributed to a special property of one IS or one insertion site, because, with the sole exception of ISC735, at least two distinct insertions of each of these ISs were evaluated for reversion.

Related sequences in other genomes

In the course of analysing the new ISs, we found that several of them had extensive similarity to sequences in Sulfolobus genomes. In S. tokodaii, for example, blastn with the ISC796 sequence identified five complete copies of a closely related IS, plus at least five partial copies; several partial copies were similarly found in S. solfataricus. blastn with the ISC1205 sequence revealed only partial copies in both genomes. blastn with ISC735 revealed two small fragments in S. solfataricus but no related sequences in S. tokodaii. These various copies had not been annotated as IS or IS fragments in either genome (Kawarabayasi et al., 2001; She et al., 2001).

We analysed the ISC1205-related sequences in greater detail in order to gain insight into the molecular processes that form partial IS copies. Eight such sequences were found in S. tokodaii, and two in S. solfataricus; together, these 10 partial copies represent three classes of mutated IS. The first and largest class (five copies in S. tokodaii and one in S. solfataricus) consists of varying lengths of the 3′ end of the element (as defined by the orientation of its transposase; Fig. 3B). The sequences seem consistent with a single large deletion (or multiple overlapping smaller deletions) extending past the 5′ end of each of six ancestral IS copies of the ISC1205 relative. The postulated deletion events appear to have been independent, as each partial copy exhibits a different end-point. In addition, the various copies have numerous substitutions with respect to each other, and three have small deletions of the extreme 3′ end.

The second class of partial IS includes the remaining three copies in S. tokodaii, which can be explained by three similar but distinct deletions of the central portion of the ISC1205 relative, leaving about 90 nucleotides intact at each end (Fig. 3C). This is the structure of type I MITEs, which can be represented as internal deletion mutants of functional ISs (Oosumi et al., 1996). Examples of other type I MITEs have been found in HA (Redder et al., 2001). The type I MITEs found in S. tokodaii are slightly divergent from each other (79–86% sequence identity) and from ISC1205, but the alignment (Fig. 3C) suggests a common deletion event relating all three to an ISC1205-like ancestor. It should be noted that, while the S. tokodaii genome contains no complete IS whose sequences match those of the MITEs shown in Fig. 3C, it does have a complete copy of ISSt1145, a distant relative of ISC1205 with low overall sequence similarity but very similar IRs. This raises the possibility that these MITEs may remain active in S. tokodaii via ‘heterologous’ complementation, i.e. complementation by a rather distantly related IS. It is also interesting to note that Redder et al. (2001) independently identified a number of type II MITEs (designated ‘SM2 elements’) in S. solfataricus that seem to be derived from ISC1217, the closest known functional relative of ISC1205.

The third class of ISC1205 fragment is represented by a single segment in the S. solfataricus genome. This sequence has about 90% nt identity to ISC1205, and displays a deletion of 309 bp near the centre of the element. In addition, it is interrupted by about 2600 nt of DNA, segments of which display homology to the putative ISs ISC1212 and ISC1359 found elsewhere in the P2 genome (Brügger et al., 2002). Examination of the boundaries of these various segments indicates an order of events leading to this complex structure (Fig. 3D). The ISC1205 relative, for example, is flanked by DRs of 6 bp, consistent with the TSDs observed for ISC1205. This indicates that the ISC1205 relative transposed into this region of the chromosome before the centre of its transposase gene was deleted. Similarly, the ISC1359 segment is flanked by DRs of 4 bp of the ISC1205 ORF, implying that ISC1359 transposed into the ISC1205 relative in its current location. Finally, the copy of ISC1212 is intact and flanked by 6 bp DRs of ISC1359, indicating that ISC1212 was the last to transpose, thereby inactivating the ISC1359 copy.

IS diversity in Sulfolobus spp.

Despite the extreme conditions of their habitats and relative isolation of individual populations (Whitaker et al., 2003), HA have proven to be a surprisingly rich source of transposable genetic elements (Brügger et al., 2002). For example, despite identification of at least 25 ISs in the genome of S. solfataricus strain P2 (She et al., 2001), a completely distinct IS was discovered fortuitously in another S. solfataricus strain cultivated from the same geothermal area (Ammendola et al., 1998). Results like these suggest that Sulfolobus populations harbour an extensive diversity of ISs and that many more of these TEs remain to be discovered by sequencing and genetic screening of additional Sulfolobus isolates. Our experimental results reinforce this idea, and more than double the number of ISs demonstrated experimentally to transpose in HA. Of seven ISs recovered from one widely distributed Sulfolobus species, we considered only one to be an isoform of a previously identified IS. Furthermore, one of the remaining six, ISC1205, could not be assigned to any recognized IS family, although it exhibited distant relationships to several ISs that also have not been assigned to any family (Fig. 2). These homologues, all identified by blast, include ISC1217 of S. solfataricus (32% transposase identity/48% similarity over 369 aa), ISSt1145 of S. tokodaii (33%/54% over 282 aa), and hypothetical ISs in Ferroplasma acidarmanus (28%/45% over 169 aa), Bradyrhizobium (30%/52% over 100 aa), a Desulfitobacterium sp. (27%/46% over 129 aa) and Caenorhabditis briggsae (25%/49% over 136 aa). These data suggest that ISC1205 belongs to a yet unnamed IS family with diverse representatives in all three domains of life.

Do different Sulfolobus species, or conspecific isolates, have different complements of active ISs?

Nucleotide sequences of the pyrBEF region suggest that the host strains examined in this study belong to a Sulfolobus species related to S. solfataricus but distinct from it. The pattern of transposition that we observed among these isolates thus provides an informative comparison to the pattern reported for S. solfataricus strain P1 (Martusewitsch et al., 2000). For example, only one of the ISs observed to transpose in S. solfataricus P1 (ISC1058) is closely related to any of the ISs we observed to transpose. This demonstrates that closely related Sulfolobus species can differ greatly with respect to their complement of active ISs. Furthermore, we observed differences with respect to the active IS complement among different strains of the same species. For example, aside from ISC1057 and ISC1058b, which are close relatives of each other, we observed very few cases in which more than one IS was observed to transpose in a given strain, despite examining numerous isolates and as many as 10 independent transposition events per isolate. This contrasts dramatically with S. solfataricus, in which a single isolate (strain P1) yielded four unrelated ISs in only seven transposition events (Martusewitsch et al., 2000). Although our results could, in principle, be explained by a lower activity of ISs than in S. solfataricus, we note that several of the ISs were capable of frequent transposition. We similarly doubt that many of the isolates in our study contain only one IS in an intact form. Although this cannot ruled out without exhaustive analysis of each isolate, we note that Southern blotting detected multiple copies of ISC796 that were conserved among strains from the Lassen population, and that IS-specific PCR detected full-length copies of at least two different ISs in about 90% of more than 100 isolates screened from all three geographical regions (Z.D. Blount, unpubl.).

Do Sulfolobus ISs differ with respect to target-site selectivity and other functional properties?

In addition to exhibiting different transposition frequencies, the ISs were also diverse with respect to qualitative properties such as the degree of target-site specificity. Within the pyrEF region, ISC735 used only one site in six independent transpositions, whereas the closely related ISC1057 and ISC1058b showed a moderate preference with regard to sequence. The latter two ISs also exhibited a preference for inserting into particular intervals of the chromosomal target, despite variation in the specific position. One of the more prominent of these hot-spots for insertion lies halfway between the TATA box and the transcription start site predicted by transcript analysis of the homologous region in S. acidocaldarius (Thia-Toong et al., 2002). Interestingly, the sequence used exclusively by ISC735 also occurs in this interval, and the homologous interval in S. solfataricus strain P1 serves as the preferred insertion site for ISC1058 (Martusewitsch et al., 2000).

This pattern resembles the tendency of many bacterial ISs to insert in or near promoters (Mahillon and Chandler, 1998) and cannot be explained readily by some property of the FOA selection in Sulfolobus spp. that enables insertions in or near the promoter region to support faster growth than insertions in other parts of the pyrE–pyrF interval. In particular, as the basis of the selection is loss of function, insertions into the promoter region would not be favoured by virtue of a partial phenotype. This is confirmed experimentally by the fact that in S. acidocaldarius, leaky pyrE alleles are at a growth disadvantage in the selection, and are not recovered under conditions of higher stringency (Grogan et al., 2001). Alternatively, there is no obvious mechanism to explain how insertion mutations before the coding sequences, whether they disrupt the promoter itself or exert polarity on the pyrF gene, would result in less enzymatic activity than insertions into the pyrE or pyrF coding sequences. The insertional preference observed in this region of Sulfolobus genomes may thus represent a functional similarity of IS transposition in HA and mesophilic bacteria involving the increased accessibility or reactivity of certain DNA regions. The high level of target-site specificity we observed for ISC735 would seem to make it unique within the IS6 family, however, as no other members of this family have displayed such specificity (Chandler and Mahillon, 2002).

Do MITEs retain the ability to transpose in Sulfolobus genomes?

MITEs are extremely short repetitive sequences with structures corresponding to ‘ends-only’ TEs. They are found in all three domains, and can be extremely abundant in eukaryotic genomes (Brügger et al., 2002; Yang and Hall, 2003). MITEs can also be abundant in HA genomes, as demonstrated by S. solfataricus strain P2 (Redder et al., 2001; Brügger et al., 2004). To our knowledge, all identifications of MITEs have been based on genome sequences, and none have been confirmed experimentally to transpose. Our observation of a MITE insertion event in Sulfolobus thus provides the first experimental demonstration of MITE transposition, and suggests that such transposition can be sufficiently frequent to affect genome evolution in Sulfolobus. As the host strain has an active copy of a cognate IS (ISC1057), the observed insertion is consistent with the hypothesis that MITEs are mobilized by transposase produced from complete ISs having the corresponding IRs. In addition, the MITE we recovered by transposition has a type II structure in which IRs are separated by intervening sequence lacking homology to any known IS. Simple internal deletion of a functional IS therefore cannot readily explain this class of MITEs (Oosumi et al., 1996), and it is not yet clear how these structures originate. It should also be noted that similar ‘gene trap’ screening of new S. solfataricus isolates has revealed transposition of another, unrelated type II MITE (Z.D. Blount, unpubl. results), suggesting considerable diversity and activity of these TEs in Sulfolobus genomes. The present study also detected new, putative type I MITEs in the genome of S. tokodaii based solely on their relatedness to ISC796 and ISC1205 (see below).

Precise excision of ISC1926

The available data seem generally consistent with the hypothesis that much of the precise excision of TEs in bacteria reflects spontaneous deletion events promoted by the short IRs and DRs at the TE boundaries (Eigner and Berg, 1981; Glickman and Ripley, 1984; Schaaper et al., 1986; Perkins-Balding et al., 1999). It is thus significant that the only insertion mutation observed to excise precisely in our study has no IRs or DRs to facilitate such deletion, whereas all the ISs with such repeats, including an extensive (34 bp) IR in the case of the ISC1290 isoform, failed to excise at detectable frequencies. These results imply that short, non-tandem DRs, including those associated with IRs, do not promote frequent deletion in this Sulfolobus species. A similar conclusion was supported by a fundamentally different analysis in S. acidocaldarius (Grogan and Hansen, 2003), thus reinforcing the idea that Sulfolobus spp., and perhaps other HA, are deficient in one or more ‘pathways’ of spontaneous deletion that predominate in bacteria (Glickman and Ripley, 1984; Schaaper et al., 1986). It remains difficult to define distinct bacterial pathways of spontaneous deletion in terms of specific genes required, and genetic manipulation of the process in HA remains daunting in technical terms. However, in bacteria and yeast it has been possible to identify host mutations that accelerate the precise excision of TEs, which implicates these genes as suppressors of DR-dependent, RecA-independent pathways of deletion. The host genes include ssb, polA, topA, MutSLH, dam, uvrD, uup, and special alleles of recBC in Escherichia coli, which mediate a range of DNA transactions, and POL1 and POL3 in yeast, which encode the DNA polymerases that synthesize the lagging strand (Lundblad et al., 1984; Gordenin et al., 1992; Reddy and Gowrishankar, 2000). It must be noted, however, that tandem DRs are deleted efficiently in S. acidocaldarius, implying that the separation of DRs attenuates deletion between them (Grogan and Hansen, 2003), and that such an effect has been documented in other microbial systems, as well (Scearce et al., 1991; Chédin et al., 1994).

Comparing phenotypic reversion of pyrE::ISC1926 with spontaneous deletion formation in the S. acidocaldarius pyrE gene reveals a quantitative discrepancy, however, in that the former was about 10% as frequent as the latter (1 × 10⁻⁹ versus 1 × 10⁻⁸ per cell respectively). If deletion end-points are determined approximately randomly (as appears to be the case in S. acidocaldarius), only a very small fraction of all spontaneous deletion events within pyrE (on the order of one in 3.6 × 10⁵ deletions) would effect precise excision of any given insertion. Thus, the phenotypic reversion of a pyrE::ISC1926 mutation is not predicted to occur at an observable frequency by random, spontaneous deletion of the type documented in S. acidocaldarius. One possible explanation for our results therefore is that ISC1926 carries out or assists its own excision, a situation with precedent among bacterial TEs (Shen et al., 1987). We note, for example, that self-catalysed precise excision at about 1% of the rate of transposition into the target gene has been observed with transposon Tn10 in E. coli (Shen et al., 1987), and a similar ratio would be sufficient to explain our results. In any case, the molecular basis of ISC1926 transposition and precise excision warrants further study, particularly as TEs that have no IRs and generate no TSDs have not been studied extensively.

What do the structures of fragments reveal concerning IS removal?

In contrast to ISC1926, all other Sulfolobus ISs that we tested exhibited no phenotypic reversion of a pyrE insertion, despite strong selection. This provides experimental evidence that gene inactivation by these ISs, which are more typical of Sulfolobus ISs than ISC1926 in terms of structural features (Martusewitsch et al., 2000; Brügger et al., 2002), is largely irreversible. This has important implications for genome evolution. For example, although it does not imply permanence of the corresponding ISs, it does restrict their elimination to mechanisms such as (i) imprecise excision promoted by the IS transposase (i.e. abortive transposition) (Shen et al., 1987) and (ii) spontaneous mutation of the types occurring generally in the host genome. To the extent that mechanisms (i) leave significant portions of the IS in the genome, or are not excessively more frequent than (ii), Sulfolobus genomes should contain segments of inactivated ISs that are in the process of accumulating spontaneous mutations.

Sequence analyses provide several lines of molecular evidence for incremental modes of IS removal from Sulfolobus genomes. For example, we found four cases in which an IS related to those we have discovered seems to have been successfully eliminated from a Sulfolobus genome: (i) an ISC796 relative and (ii) an ISC735 relative eliminated from S. solfataricus, and ISC1205 relatives eliminated from (iii) S. solfataricus and (iv) S. tokodaii. These cases are defined by the genomes having multiple fragments of the IS but no full-length copy; in all cases, the differences among the fragments suggest independent deletion events. Other cases, in which the full-length IS nevertheless remains in the genome, have been documented in genome annotations (Kawarabayasi et al., 2001; She et al., 2001). Brügger et al. (2002) noted the abundance of these partial IS copies and postulated the existence of a mechanism that specifically inactivates ISs when they become too numerous in a genome. Our experimental analysis of precise excision suggests an alternative explanation for the observed abundance of these partial copies. Without precise excision, the complete eradication of an IS copy is forced to be a slow, multistep process. In this situation, proliferation of an IS in a genome and the inevitable inactivation of individual copies by small mutations produces partially degraded intermediates in the genome, and the abundance of these relics reflects the relative rates of their production versus removal. Although comprehensive testing of either hypothesis is beyond the scope of the present study, our results supplement the available genomic sequence data by providing new examples of partial IS copies for such analyses.

Finally, in qualitative terms, our examination of the ISC1205-related fragments in sequenced genomes reveals at least two molecular processes by which IS may be inactivated and partially removed. One is a mechanism of removal, typified by most of the ISC1205 fragments in S. tokodaii, that leaves only the element's 3′ end. It will be of interest to determine whether this mode of deletion occurs with other IS and other Sulfolobus spp. The other is a mechanism of inactivation, illustrated by the nearly full-length relic in S. solfataricus P2, representing insertion of other genetic elements into the IS. Similar relics apparently generated by this inactivation mechanism occur elsewhere in the P2 genome (Brügger et al., 2002), and remain fully consistent with the non-essential nature of ISs and the fact that spontaneous mutation in S. solfataricus is dominated by IS transposition (Martusewitsch et al., 2000).

Strains and growth conditions

Samples of water and sediment were collected from a number of acidic geothermal springs in the Norris Geyser Basin, Crater Hills, and Geyser Creek areas of Yellowstone National Park in northwestern Wyoming (USA), Devil's Kitchen and Bumpass Hell areas of Lassen Volcanic National Park in northern California (USA), and Uzon Caldera, Geyser Valley, and Mutnovsky Volcano areas of the Kamchatka peninsula (Russia), as previously described (Whitaker et al., 2003). Samples were plated directly (i.e. without enrichment) on dextrin-tryptone medium and incubated for 8–15 days at 78°C. Individual colonies were cultured in liquid medium, clonally purified on plates, catalogued, and preserved at −70°C using techniques similar to those previously described for S. acidocaldarius (Grogan and Gunsalus, 1993). Isolate designations (e.g. ‘Y00 51′-90’) incorporate the following sampling and isolation data: region (Yellowstone, Lassen or Kamchatka) and year, sample number (hyphen) clone number.

Unless otherwise noted, media and growth conditions were identical to those used for S. acidocaldarius, except that 0.2%d-xylose was replaced by 0.2% Dextrin-10 (Fluka). Spontaneous pyr mutants were selected by spreading aliquots of liquid cultures on plates containing 150 g of FOA and 20 g of uracil per ml of medium. Fluctuation tests were conducted by inoculating sets of at least five tubes, each containing 3 ml of liquid medium, with one isolated colony per tube of the isolate or mutant under investigation. When the cultures reached a density of about 10⁸ cfu ml⁻¹, the cells were plated on selective media. For assays of transposition, the selective plates contained uracil and FOA, and rates of total spontaneous mutation were calculated from the distribution of the number mutant colonies, as previously described (Jacobs and Grogan, 1997). The fraction of mutants represented by insertions was determined by PCR screening of mutants from the fluctuation tests. For assays of phenotypic reversion, the cells harvested from liquid cultures were washed in sterile dilution buffer and plated on medium containing glutamine and acid-hydrolysed casein. Because most cultures in the reversion assays yielded no mutants, rates were estimated by the P₀ method of Lea and Coulson (1949).

DNA analyses

DNA was extracted from mutant cultures using the guanidinium thiocyanate procedure of Pitcher et al. (1989). PCR was then used to amplify the pyrE and pyrF loci, their common promoter region, and the 3′ end of the pyrB gene. Primers for these amplifications were based on the S. solfataricus pyr operon sequence (Martusewitsch et al., 2000). To amplify pyrE and the promoter region-pyrB end only, the primers SsoINTER1for (5′-CGAATATTCTAAAGTAGTCATCTCTGG-3′) and SsopyrE1rev (5′-CGGGATCCATTGCTAATATTACT CTAG-3′) were used. Amplification of the entire pyrBEF region required SsoINTER1for and SsopyrF1rev (5′-TTCC TCGTGTAGATTTTCCCC). Reaction mixtures contained dNTPs, Taq DNA polymerase and buffer (Promega or New England Biolabs), ≈50 ng of genomic DNA, as well as the appropriate primers. Temperature cycling (2 min at 94°C, followed by 25 cycles of 1 min at 94°C, 1 min at 57°C and 1 min at 70°C) was carried out in an MJ Research. PTC-100 programmable thermocycler, and products were run on 1% agarose gels.

A second PCR series with nested primers was then used with those mutants displaying enlarged loci to roughly localize the insertion site within the target region. The primers used were as follows: promoter region: SsoINTER1for and SsoINTER2rev (5′-ACTAACCTTACCTGATGTTAAAACG-3′); first third of pyrE: SsopyrE2for (5′-GAAGATCTCTACGTAT GAATTTCGC-3′) and PyrEmid-2-rev (5′-CCATAGGCTCTT TAAGGTTACAAGC-3′); middle third of pyrE: PyrEmid-1-for (5′-GCTTGTAACCTTAAAGAGCCTATGG-3′) and PyrEmid-4-rev (5′-TGCGTCTGAAACTTTACCTCC-5′); last third of pyrE: PyrEmid-3-for (5′-TCCATATGAGAAAGCAACATTGG-3′) and SsopyrE1rev (5′-CGGGATCCATTGCTAATATTACTCTAG-3′). For nucleotide sequencing, the portion of the target containing the insertion was amplified, and the amplification product purified from the primers and salts and transferred to water using Millipore Microcon^® YM-100 filters. DNA sequencing was performed by the Cincinnati Children's Hospital Sequencing Facility, using ABI PRISM dye-terminator reagents. Text files were edited for miscalls by visual inspection of electropherograms.

Based on the common identity of their sequences, an interval of 1400 nt was compiled from several of the isolates. This sequence included a portion of the pyrB gene, the bidirectional promoter, the pyrE gene and the pyrF gene. Because of its much closer relationship to the individual sequences analysed, the 1400 nt sequence compiled from experimental isolates was used in the present study in place of the S. solfataricus P2 sequence as the reference for analysis of insertions and other mutations.

Assembled IS DNA sequences were used in blast searches (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi) to identify similar IS in the NCBI database, as well as fragments in sequenced genomes. Putative ORFs were identified using the ORF finder in the nebcutter program (http://tools.neb.com/NEBcutter). Putative protein sequences were then used in blast searches to identify related transposase sequences, conserved domains and family assignment. Putative protein characteristics were determined using the resources of the ExPASy site (http://us.expasy.org/). Putative transposase amino acid sequences were aligned with those of related ISs identified from blast searches or contained in the IS Finder database (http://www-is.biotoul.fr) using T-Coffee (Notredame et al., 2000). Phylogenetic reconstruction from aligned amino acid sequences was performed with paup* 4.0 for Macintosh using maximum parsimony analysis (Swofford, 2003).

Sequence designations and accession numbers

In the present study, an ‘insertion sequence’ was considered to encompass all observed sequence variants exhibiting more than 90% nt identity in the transposase-encoding region; conversely, variants with 90% or less identity were given distinct designations. Naming followed a convention similar to that of Martusewitsch et al. (2000): ‘ISC’ is followed by the length, in nucleotides, of the first isoform documented at the sequence level. Nucleotide sequences for each of the seven ISs described in the study have been deposited in GenBank under Accession No. AY671942 to AY671948.