In this paper we investigated the structure and general arrangement of hsp70 genes from two Stratiomyidae species from strikingly contrasting habitats.
According to field observations and measurements, S. singularior has highly variable and instable larval habitats in the arid steppe zone of the Crimea. Strong changes in the shore larval habitats of S. singularior were observed since 2005 to 2009, both within the season and between the years. These changes were related mostly to hydrological regime, i.e. varying precipitation and the water level of the lakes. Generally, in the periods of the higher water table, many larval habitats were flooded or strongly influenced by hypersaline lake water (mineralization reached 150 - 280‰ and higher). The decreasing water level of the lake coincided with a decrease of mineralization (up to 80 - 35‰ in some larval habitats); in these periods, the respective habitats were moistened mostly with groundwater of lower salinity. It should be also mentioned that the abundance of S. singularior larvae varied greatly along the shoreline and between the (micro)habitats. In favorable places, e.g. in some groundwater-fed pools above the lake water margin, the larvae reached high abundance (hundreds and thousands individuals per m2), but typically, they were completely absent or found only as single individuals along the most part of the shore.
It seems that in the study area, S. singularior is represented by a number of micropopulations mostly confined to some shore localities, which may be at a distance of several km from each other. Many of these favorable breeding places are temporary and completely dry out in drier and warmer years. However, in this case adult flies are able to cover the distance of several kilometers easily in search of new places for oviposition.
Therefore, our preliminary observations on the Crimean larval habitats of S. singularior testify to a variability in the conditions between the habitats, strong variation of conditions within the habitats occurring with time, together with the temporary nature of many habitats and their possible isolation from each other. All these features may result in a high polymorphism of hsp70 demonstrated in this particular species.
We demonstrated a high level of polymorphism by Southern blot hybridization with genomic DNA digests, depending on the number of the larvae taken for analysis (Figure 2). Furthermore, sequencing of several phages containing the same hsp70 genes revealed polymorphism in the intergenic regions including long deletions. In a few cases we detected deletions of entire individual hsp70 genes (in phages 52, 33 and 51), likely a result of unequal recombination.
These deletions may result from an evolutionary trend to make the hsp70 cluster in this species more compact. Previously we have described more compact organization of the hsp70 genes cluster in a more thermotolerant low-latitude species Drosophila virilis in comparison with the closely related higher-latitude D. lummei from the temperate climatic zone  and, therefore, the observed deletions may represent certain pattern. Alternatively, the deletions described in S. singularior may be due to a high overall rate of DNA loss in this species, a phenomenon previously demonstrated for two Drosophila species .
In contrast to the S. singularior biotope, the studied larval habitat of O. pardalina - a cold spring near St. Petersburg observed from 2004 to 2010 - was very stable in mineralization, temperature, water regime, and in other major conditions both within season and interannually. This spring and headwater stream seems to be an island habitat for the O. pardalina population. Another similar habitat is situated at about 3 km, and it is unlikely that O. pardalina adults migrate regularly between these localities. The abundance of larvae is not high, not exceeding 10 - 50 individuals per m2 of semiaquatic shore substrate (moss etc.). Thus, habitats of the two species differ dramatically from each other not only in the average parameters such as temperature and mineralization, but also in the spatial organization and in the degree of variability in time.
Therefore, in the case of the latter species we have a comparatively small and probably highly homozygous population. Southern blot hybridization analysis (Figure 2) did not reveal any polymorphisms in O. pardalina hsp70 genes and, hence, confirms this conclusion. Furthermore, individual phages containing the same fragments of DNA did not exhibit any polymorphisms either in coding or in intergenic regions of the same hsp70 genes. Since these phages were isolated from two independently obtained genomic libraries they cannot contain the same DNA but rather likely represent highly homologous sibs.
Previously, when we investigated inbred laboratory strains of D. virilis and D. lummei, we also demonstrated apparent identity of the overlapping hsp70 -containing clones, which is typical for highly homozygous strains .
In spite of strikingly different habitats, the larvae of all Stratiomyidae species studied have very high tolerance to elevated temperature in comparison with most of other Diptera species investigated . Furthermore, although tolerances were in general correlated with the typical habitat temperatures of the species, even O. pardalina larvae inhabiting cold springs with constant temperature regime (5 - 10°C), significantly exceeded all Drosophila species in tolerance and exhibited a huge gap between the natural temperatures (5 - 10°C) and the critical temperature (43°C).
Importantly, larvae of all four Stratiomyidae species investigated have exhibited high constitutive levels of Hsp70 proteins which are induced only 2 - 3-fold after temperature elevation . In this respect, Stratiomyidae species resemble Locusta migratoria, an insect species adapted to arid tropical and subtropical climate, where temperature frequently exceeded 40°C .
Additionally, in contrast to Drosophila, the accumulation of Hsp70 continues in the cells of the treated larvae of all Stratiomyidae species investigated for many hours after heat shock and plateaus only approximately 24 - 36 h after the treatment .
The results obtained may be explained by constitutive weak transcription (leakage) of hsp70 genes under normal physiological conditions, and possibly by comparatively higher stability of Hsps in Stratiomyidae species in comparison with other insect species studied.
In order to further characterize the molecular basis underlying this pattern of hsp70 expression and differences in thermotolerance in the Stratiomyidae and other Diptera, as well as between the Stratiomyidae species, we compared the general arrangement of hsp70 genes clusters of the two Stratiomyidae species from thermally different habitats. Surprisingly, the analysis revealed quite different organization of hsp70 genes in the two species compared.
Although the hsp70 copy number does not differ significantly between less tolerant O. pardalina and more thermotolerant S. singularior, the general organization of hsp70 genes is divergent in these two species. While in S. singularior all hsp70 genes are confined to one genomic section of about 29 kb in length, in O. pardalina most of hsp70 genes are scattered in the genome and apparently located rather far from each other if linked at all. However, in both species we have found two hsp70 copies in an inverted orientation (a fundamental and probably ancient hsp70 gene unit) . Furthermore, in O. pardalina we cloned a fragment carrying one copy of inducible hsp70 and one copy of presumptive hsp68 gene in tandem orientation separated by only 3.5 kb.
Previously we described two hsp68 copies in inverted orientation in species of the virilis group of Drosophila . We speculated that hsp68 in these species arose by duplication of a basic unit of hsp70 genes typically arranged in Diptera species in inverted orientation. However, in the virilis group species hsp70 and hsp68 are located in the same chromosome but are separated by a large distance . Presumptive hsp68 gene of O. pardalina is closely linked to hsp70 gene (Figure 1B) and exhibits higher homology with hsp70 genes of this species than with hsp68 genes from other Diptera species (data not shown). These data enable us to suggest that O. pardalina hsp68 also represents a duplication of hsp70 gene. At the present time we can not discriminate between two hypotheses explaining the appearance of hsp68. The appearance of the hsp68 by duplication of hsp70 may occur very early in Diptera evolution and predate the split of Drosophilidae and Stratiomyidae. Alternatively hsp68 may appear independently in the evolution of Drosophila and Stratiomyidae and acquire similar functions and structure by convergence.
It is noteworthy that even for the genes located in inverted orientation, which apparently represents the primitive state of hsp70 in Diptera, the distance between the inverted copies in O. pardalina is four times as large as between the inverted copies in S. singularior (Figure 1A and 1B).
The different spacing of hsp70 genes in the species compared may resemble different abilities of their genome to efficiently respond to stress. The homogenized coding sequences and highly compact organization of hsp70s in S. singularior may be necessary for fast and efficient transcriptional response and thereby thermotolerance of the larvae to extreme and rapidly changing conditions of the environment.
On the other hand, O. pardalina inhabiting cold running waters with comparatively stable thermal conditions may not require such tight hsp70 cluster organization and, hence, underwent dispersal and pseudogenization of a few hsp70 copies in the process of evolution. Two hsp70 pseudogenes detected in the genome of this species in our studies may represent the remains of previously active genes. Previously we have shown that larvae of O. pardalina produce significantly less Hsp70 in response to heat shock in comparison with all other Stratiomyidae species investigated . The observed drastic differences in the organization of hsp70 genes in S. singularior and O. pardalina may account for the differences in the thermotolerance and Hsp70 levels after HS treatment observed between these species.
Furthermore, HSE1 of O. pardalina is not canonical and contains only two 5 bp units. Therefore, the first canonical element (HSE2) in O. pardalina is located at significantly larger distance from the transcription start than in S. singularior (Figure 3). The differences in the number of canonical HSEs and their spacing may also contribute to different activity of hsp70 genes in the two species  and explain why O. pardalina exhibits lower thermotolerance in comparison with all other Stratiomyidae species studied so far .
It is also evident from Figure 3 that in the case of O. pardalina, the arrangement of promoter regions of all hsp70 genes is practically identical, while in the case of S. singularior the promoters of all members of hsp70 cluster differ significantly. These differences result from fine tuning of an hsp70 "battery" necessary for adaptation of the latter species to highly variable environmental conditions. Therefore, our results indicate that O. pardalina hsp70 genes may be locally adapted, whereas higher hsp70 polymorphism may have important evolutionary consequences for S. singularior.
The evolution of hsp70 genes in the species compared is quite different in many respects. The MK tests demonstrate that the divergence between individual S. singularior hsp70 genes is very low, especially considering the age of the duplications. This is consistent with concerted evolution in this species, and also strikingly different from the divergence between paralogs in O. pardalina (See Additional file 6 : Table S1). In all cases, the number of shared polymorphisms between S. singularior hsp70 genes is highly unlikely to have occurred via parallel mutation (Table 1). The presence of multiple conversion-mediated shared SNPs and large conversion tracts in the alleles of the same hsp70 members support gene conversion as a mechanism of concerted evolution (and thus sharing polymorphism) among duplicated hsp70 genes of S. singularior species .
The overall pattern of conversion-mediated homogeneity among hsp70 genes in S. singularior that decays with distance from CDS, and polymorphic flanks that contain large indels, is consistent with previously described duplicated hsp70s in Drosophila . We cannot exclude the occurrence of gene conversion in O. pardalina, because in this species we did not isolate multiple alleles of individual hsp70 genes. However, since the level of fixed differences between hsp70 genes in O. pardalina is several fold higher than in S. singularior (Additional file 6 : Table S1), one may conclude that in the former species the conversion process is less efficient if present at all. The large distances between individual hsp70 genes in O. pardalina genome likely decrease the frequency of their conversion [35, 36]. Whether the large distances also decrease their coordinate expression is an interesting question.
Several authors justly outlined many logical and statistical problems associated with using two-species comparison for studying adaptations , and we are well aware that unless similar correlation is obtained for other Stratiomyidae species or geographical populations, any generalizations are premature. However, a similar pattern of hsp70 organization has been previously described when studying two closely-related species of the virilis group of Drosophila, which replace one another along a latitudinal cline . The less thermotolerant high-latitude species D. lummei, similar to O. pardalina, exhibits several-fold larger distance between tandemly arranged hsp70 copies in the cluster, in comparison with the closely-related more tolerant low-latitude D. virilis . Interestingly, along these lines one of the hsp70 copies described in D. lummei was shown to be a nonfunctional pseudogene lacking the first 300 nucleotides of coding sequence. Similarly, two likely nonfunctional pseudogenes have been described in the present study in the less thermotolerant O. pardalina. Dispersed genomic arrangement is not absolutely required for hsp70 degeneration. The tight hsp70 clusters of the tropical island endemic D. mauritiana segregate multiple pseudogenes, although in this case a founder effect rather than relaxation of selection may modulate hsp70 pseudogene frequencies . In spite of the approximately identical hsp70 copy number in Stratiomyidae and Drosophila virilis group species, there are several clear-cut differences in the organization and evolution of hsp70 cluster between these phyla. In species of the virilis group, not only the hsp70 transcriptional units but also the 3'-flanking sequences are highly conserved within D. virilis and between D. lummei and D. virilis. Thus, the 3'-flanking sequence of several D. virilis hsp70 genes is near identical for more than 1600 bp. Such an arrangement suggests tandem duplication of hsp70 copies by unequal crossing over in the course of Drosophila species evolution .
In contrast, our analysis detected only short, varying length regions of significant homology in the intergenic regions in S. singularior hsp70 cluster. Similarly, we detected only 650 bp of significant homology in the intergenic regions located at the vicinity of 5'-ends of hsp70 genes in O. pardalina.
At the present time, we can only speculate concerning the molecular mechanisms which may underlie the different arrangement of hsp70 genes in the species studied, although the pattern of compactness and conversion in S. singularior vs. dispersion and divergence in O. pardalina is clear. Several large deletions in S. singularior, including those spanning entire individual hsp70 genes, may be indicative of an evolutionary trend towards compactness to provide a highly coordinate response to the changing environmental conditions. The deletions may also be a consequence of conversion; the more homogenized the individual hsp70 units, the more likely they are also to participate in unequal crossover. On the other hand, the dispersed arrangement and interlocus divergence of hsp70 copies in O. pardalina may result from long-range recombination and relaxed selection on hsp70 sequences in a species living in very stable thermal conditions.
The analysis of hsp70 arrangement in other species and natural populations of Stratiomyidae and related taxa (work in progress) should help to further evaluate the evolution of hsp70 genes in the ecological context.