Genetic divergence with distance in Sulfolobus species

The extremely thermoacidophilic archaeon Sulfolobus islandicus colonizes acidic hot springs throughout the Northern Hemisphere and has diversified extensively, both within local populations and as a function of the distance between them (Whitaker et al., 2003). The observed divergence with distance implies that immigration leading to gene flow between local populations is severely restricted. This contrasts sharply with the efficient passive dispersal of most micro-organisms (Finlay 2002), and has made S. islandicus an influential model of prokaryotic allopatric diversification (Martiny et al., 2006).

However, the pattern of genetic divergence does not, in itself, define the basis of gene-flow restriction. The S. islandicus barrier is commonly attributed (sometimes implicitly) to the fact that the environmental conditions required by Sulfolobus species (Huber and Prangishvili, 2006) occur in geothermal sites that are extremely limited in area and separated by extremely large distances (Fenchel, 2003; Hahn and Pöckl, 2005; Knittel et al., 2005; Cohan, 2006; Ramette and Tiedje, 2007; Pagaling et al., 2009). Successful immigration thus requires Sulfolobus cells to (i) be shed from relatively small populations, (ii) retain viability during prolonged transport and (iii) enter small islands of geothermal habitat. Although the hypothesis thus seems plausible that genetic isolation of S. islandicus populations results from excessive dilution and death of propagules (reflecting the interaction of organismal physiology and physical geography), this has not been tested. Importantly, it predicts that all Sulfolobus species will diverge similarly with distance, as the growth requirements that define the genus also define similar geothermal habitats having a similar spatial distribution for all its species.

Three geographically distinct strains of S. acidocaldarius have been reported to date: (i) the type strain 98-3 isolated from the Norris Geyser Basin of Yellowstone National Park (Brock et al., 1972, ii) strain N8 isolated from the Jigokudani thermal field near Norboribetsu, on the Japanese island of Hokkaido (Kurosawa et al., 1995) and (iii) strain Ron 12/I cultured from self-heating mining waste near Ronneburg in eastern Germany (Fuchs et al., 1995). The average distance between these sites (that is, 8209 km) exceeds the largest distances represented by sequenced S. islandicus genomes by about 2000 km (Reno et al., 2009), and thus should measure primarily the impact of geographical separation, with only a minor contribution from local Sulfolobus diversity (Whitaker et al., 2003; Reno et al., 2009). We therefore investigated divergence by distance in S. acidocaldarius by determining the complete genome sequences of Sulfolobus strains N8 (NBRC 15159) and Ron 12/I. We re-purified the strains on plates, extracted genomic DNAs, and submitted the DNAs for 76 extension cycles on an Illumina (San Diego, CA, USA) GAII instrument (DNA Analysis Facility, Iowa State University). About six million reads per genome were assembled (220-fold average coverage) and the resulting alignments were analyzed, using Genomics Workbench (CLC bio, Aarhus, Denmark). We filled small assembly gaps at repeats and low-coverage sites by Sanger sequencing of PCR products, to yield closed circular genome sequences.

Conservation of the S. acidocaldarius genome

Despite the global-scale distances represented, the genomes of strains N8, Ron12/I and the type strain 98-3 proved to be nearly identical, averaging fewer than 26 differences (polymorphisms) per pair (Table 1). Using S. islandicus as a reference, the average pairwise nucleotide divergence for shared genes of strains from Kamchatka vs those from North America is 1.11 × 10−2, whereas the average for strains from within each of the two regions is 2.7 × 10−3 (Reno et al., 2009). This latter divergence (24% of the total) thus estimates local diversity, and subtracting it from the total divergence yields the divergence of S. islandicus that can be attributed specifically to large geographical separation, that is, 8.4 × 10−3.

Table 1 Measures of S. acidocaldarius genome divergence
Full size table

The average nucleotide divergence of the S. acidocaldarius genomes is 1.12 × 10−5. Assuming the same relative contribution of local diversity as in S. islandicus (24%) yields 8.5 × 10−6 as the divergence of S. acidocaldarius genomes that can be attributed to long-range geographical separation. When this value is normalized with respect to separation, the rate of S. acidocaldarius divergence with distance is 1300 times lower than that in S. islandicus (1.03 × 10−6 vs 1.37 × 10−3 per 1000 km).

We could not attribute the much stricter conservation of S. acidocaldarius genomes to a correspondingly higher replication accuracy or stronger purifying selection relative to S. islandicus. In S. acidocaldarius, replication errors occur at the pyrE and pyrF genes at the rate of about 3 × 10−7 events per cell division (Grogan et al., 2001), as compared with 0.4–16 × 10−7 (median=4.9 × 10−7) in strains of S. islandicus (Blount and Grogan, 2005). Furthermore, nearly all of the polymorphisms seen in S. acidocaldarius affect the presence or structure of protein, and thus should not be selectively neutral (about half are insertion/deletion events (Table 1), whereas most of the substitution mutations are nonsynonymous (dS/dN=3/18=0.1667)). These data imply that S. acidocaldarius genomes are free to accumulate a variety of mutations, and, accordingly, that their conservation world-wide requires relatively rapid gene flow across large distances. We also considered the question of whether the three strains could be highly similar because they all diverged from a common laboratory culture. The available evidence argues against this possibility, but extensive efforts of multiple researchers may be required to define the pattern of natural divergence more rigorously (see Supplementary Information).

Mechanisms potentially producing an inter-specific difference in gene flow

The rapid gene flow in S. acidocaldarius indicated by our data contrasts sharply with the genetic isolation documented repeatedly for S. islandicus (Whitaker et al., 2003; Grogan et al., 2008; Reno et al., 2009) and raises questions regarding what enforces this barrier in the latter species. Such a large difference between physiologically similar species suggests that biotic factors, rather than environmental ones, have critical roles, and the limited data regarding ecological parameters that we could obtain so far remain consistent with this. Properties that should facilitate the transfer of propagules among populations of a species include (i) large local populations, which would generate high fluxes of propagules from simple concentration (mass-action) effects, (ii) a large number of local populations, creating ‘stepping-stone’ networks with short distances between populations or (iii) large areas of habitat, which would capture dispersing propagules efficiently. We know of few field measurements that help define these parameters for different Sulfolobus species, but the available data generally predict that these parameters would favor propagule dispersal of S. islandicus (Supplementary Information).

One parameter that we could measure experimentally relates to the survival during dispersal. Because temperatures in the environmental matrix between geothermal islands preclude metabolism, growth and reproduction of Sulfolobus, survival between local populations would seem critical for gene flow. We found empirically that cell suspensions that were held in acidic solutions at 4° C lost viability over a period of days to weeks, thus providing a basis for measuring relative durability. Under these conditions, S. acidocaldarius cells died markedly faster than cells of three geographically matched S. islandicus strains (Table 2). This demonstrates that the intrinsic durability of propagules does differ among Sulfolobus species, but, like the other parameters we considered, the difference observed would not explain the observed difference between species with respect to gene flow.

Table 2 Long-term survival of Sulfolobus cells at 4 °C
Full size table

We are aware of one factor that would seem to explain the observed difference in gene flow between S. islandicus and S. acidocaldarius in the light of available data, namely, antagonism mediated by viruses and other mobile genetic elements; however, such antagonism would act only after a propagule is deposited into a new local population, and would not seem to depend directly on spatial distribution of habitat. Genetic elements capable of mediating severe and specific antagonisms can become established in microbial communities and create strong reciprocal selections that promote rapid coevolution of the element and host (Pal et al., 2007; Paterson et al., 2010). S. islandicus seems fully vulnerable to this, in terms of both its intrinsic ability to harbor diverse elements (Prangishvili et al., 2001) and the relatively high density of its natural populations, which favors their proliferation; S. acidocaldarius, in contrast, fails both criteria. (For further discussion of these and other mechanisms, see Supplementary Information).

Conclusion

In the time since local populations of S. islandicus were first shown to diverge with distance (Whitaker et al., 2003), several free-living mesophilic micro-organisms have been found to exhibit similar divergence, demonstrating that restricted gene flow does not require severely discontinuous habitat (Bass et al., 2007; Kuehne et al., 2007; Vos and Velicer, 2008). The evidence of global conservation of S. acidocaldarius genomes now argues, conversely, that severely discontinuous habitat is not sufficient per se to restrict gene flow of unicellular micro-organisms. Therefore, although the processes that restrict gene flow among local S. islandicus populations remain to be identified, they seem likely to have counterparts in other micro-organisms that exhibit allopatric diversification.