|Home | About | Journals | Submit | Contact Us | Français|
The small heat shock gene (shsp) cluster of Drosophila buzzatii was sequenced and the gene order and DNA sequence were compared with those of the shsps in Drosophila melanogaster. The D. buzzatii shsp cluster contains an inversion and a duplication of hsp26. A phylogenetic tree was constructed based on hsp26 genes from several Drosophila species of the Sophophora and Drosophila subgenera. The tree shows first a separation of the Sophophora and the Drosophila subgenera and then the Drosophila subgenus is divided into the Hawaiian Drosophila and the repleta/virilis groups. Only the latter contain a duplicated hsp26. Comparing the gene organisation of the shsp cluster shows that all the Drosophila subgenus species contain the inversion. Putative heat shock elements (HSE) were found in the promoters of all the shsp and putative regulator elements for tissue specific expression were found in the promoter of hsp23, hsp27 and one of the hsp26 genes. hsp23 was found to be polymorphic for four non-synonymous changes that all lead to exchange of a Valine. The duplicated hsp26 gene in D. buzzatii (phsp26) was polymorphic for two non-synonymous changes. The allele frequencies of these variants were determined in nine D. buzzatii populations covering most of its distribution in Australia using high-resolution melting curves. The allele frequencies of one of the hsp23 variants showed a significant linear regression with longitude and the pooled frequency of the four Valine changes of hsp23 in the nine populations showed a significant linear regression with longitude and with a composite measure of climatic variables.
The online version of this article (doi:10.1007/s12192-009-0140-y) contains supplementary material, which is available to authorized users.
The cactophilic species Drosophila buzzatii has a wide geographical distribution, and it breeds and feeds on decaying tissues of Opuntia cacti. D. buzzatii originates from South America, from where it has colonised and spread to the Old World and Australia. D. buzzatii belongs to the mulleri complex of the repleta group within the subgenus Drosophila and due to its rather broad geographic distribution, it has been a model for evolutionary adaptation to different environmental conditions. Several studies on this species have described latitudinal and altitudinal variation in different traits (longevity, fecundity, senescence, stress resistance and chromosome polymorphism; Hasson et al. 1985; Sorensen et al. 2005; Norry et al. 2006) and genetic variation in allozyme markers (Barker et al. 1985) and microsatellites (Frydenberg et al. 2002; Barker et al. 2009). However, variation of candidate genes for traits associated with climatic adaptation has only rarely been studied. In Drosophila melanogaster, allele frequencies in candidate genes have been successfully correlated to climatic variables and latitude, including shsps (Frydenberg et al. 2003), hsr-omega (Collinge et al. 2008) and Adh (Fry et al. 2008). Heat shock genes are highly conserved, which makes these genes also potential candidate genes in other Drosophila species, especially in species such as D. buzzatii, where several life stages occur inside rotting cladodes that are often exposed to the sun.
In D. melanogaster, there are seven shsp genes located closely in a linked cluster on chromosome three in the order hsp27, hsp23, hsp67Ba, hsp26, hsp22, hsp67Bb and hsp67Bc (Ayme and Tissières 1985). The sHsps have many different functions, both under stress and under non-stress conditions (Mestril et al. 1986; Michaud et al. 1997; Michaud and Tanguay 2003). Under stress conditions they have chaperone functions, protect the cytoskeleton and modulate the apoptotic process. The sHsps have a cell-specific pattern of expression, likely reflecting different chaperone activities (Morrow et al. 2006). In the absence of stress, the cell-specific expression of the shsps is tightly regulated, suggesting specific functions during development.
In D. melanogaster, the regulation of each of the shsps has been thoroughly investigated, either using promoter deletions to locate important regulator regions (Cohen and Meselson 1985; Riddihough and Pelham 1986; Frank et al. 1992; Michaud et al. 1997), or regulation by known tissue specific transcription factors (Michaud and Tanguay 2003) or search for consensus transcription factor binding sites. The expression after heat shock is regulated by the heat shock factor (HSF) binding to the heat shock element (HSE) and HSE have been found in the promoter of most hsp genes (Mestril et al. 1986; Sandaltzopoulos et al. 1995; Riddihough and Pelham 1986).
In order to characterise the shsp genes in D. buzzatti and analyse their genetic variation, we have obtained sequences of the gene region of D. buzzatii which contains the genes coding for the sHsps. The gene organisation in this region is compared with the homologous region of D. melanogaster and an inversion in the region is described. Comparative genomic regions of the D. buzzatii and the D. melanogaster sequences are used to define phylogenetic footprinting that may identify the tissue-specific transcription factor binding sites and the HSEs.
The sequence variation of the D. buzzatii hsp23, phsp26 and hsp26 is analysed. The allele frequencies of four non-synonymous SNP in hsp23 are determined in nine Australian populations using a high resolution melting (HRM) method and the variation is assessed for association with geographical location and climatic variables.
The sequences of the shsps from D. melanogaster were found in Fly Base and aligned. We designed two primers that are located in the α-crystallin domain, conshspF GTGGGCAAGGATGGCTTCCAG and conshspR TTGCTGGATCCGGACGATGCGCTC, in order to get the α-crystallin domain from shsp in D. buzzatii. A 300 bp fragment was amplified under standard polymerase chain reaction (PCR) conditions. The fragment was sequenced (Macrogen Genomics, Korea) and found to contain part of hsp67Ba. From this sequence two forward and two reverse specific primers were designed. Using respectively the two reverse primers and the two forward primers as specific primers in the DNA Walking SpeedupTM kit (Seegene, USA), we walked out in the genome in both directions. This procedure was repeated and combined with primers designed from published sequences. The coding sequences of D. melanogaster hsp23, hsp26 and hsp27 were used for a search for related sequences in the Drosophila mojavensis gene library using the sequences available http://rana.lbl.gov/drosophila/. The D. mojavensis genome assembly was produced by Agencourt Bioscience Corporation. Identified D. mojavensis hsp23, hsp27 and hsp26 genes were then used to design PCR primers for D. buzzatii DNA. The fragment was isolated and sequenced (Macrogen Genomics, Korea) with the PCR primers and internal primers.
The chromosome regions between CG4461-RA and CG4080-RA from D. melanogaster, Drosophila ananassae, Drosophila pseudoobscura, Drosophila willistoni, Drosophila grimshawi, Drosophila virilis and D. mojavensis were available at http://www.ncbi.nlm.nih.gov/genome/guide/fly/ (Adams et al. 2000; Richards et al. 2005; Flybase Consortium 2002).
A phylogenetic tree of the hsp26 genes was constructed using the MEGA version 3.1 (Kumar et al. 2004) by the Minimum Evolution method with the distance calculated by the Tamura and Nei method (Tamura and Nei 1993). Bootstrap values were calculated from 1,000 replications.
The promoter region of the shsp genes from D. melanogaster, D. pseudoobscura, D. virilis and D. mojavensis were available at http://www.ncbi.nlm.nih.gov/genome/guide/fly/ (Adams et al. 2000; Flybase Consortium 2002).
Local blocks of homologous sequences were found with the use of Chaos/DiAlign (Brudno et al. 2004; http://dialign.gobics.de/chaos-dialign-submission), DNA block aligner (DBA; Birney and Durbin 1997; http://www.ebi.ac.uk/Wise2/dbaform.html) and the VISTA programme (Mayor et al. 2000; http://genome.lbl.gov/vista/index.shtml) with a 10 bp 70% window.
The DNA sequence variation of the coding region of hsp26, phsp26 and hsp23 was examined in D. buzzatii by sequencing hsp26, phsp26 and hsp23 in a total of 40 flies from four different populations (Cafayate, Trancas, Chumbicha and Santiago de Estero) from Argentina (Sorensen et al. 2005). hsp26 was amplified by primers Dbuz26vaF (ATTCATCAAGCGAACAACTTC) and DbuzvaR (TGGAATGAGGATGACACAGG) and DbuzvaR was used as sequence primer. phsp26 was amplified using primers Dbp26pro (CAAAGAACTTCGCTGAGCGA) and Dbp26RS (GGAATGCGATTGACTTTGGTT). The PCR products were sequenced directly using the primer Dbp26pro1 (AAAACTACATAACCTTAAATC) (Macrogen Genomics, Korea). hsp23 was amplified by the primers Db23F1 (TACTGTTTGAAAAAACCAAG) and Db23R (TAGATATTGACTTCAATAATT). The PCR products were sequenced by Db23R (Macrogen Genomics, Korea).
The allele frequencies of non-synonymous mutations in hsp26, phsp26 and hsp23 were analysed in nine populations (30 individuals from each) from Australia (Fig. 1), Mulambin Beach (Mub; 23°19′ S, 150°78′ E, 0 km distance from coast (further on called dist)), Isla Gorge (Isg; 25°25′ S, 149°94′ E, 230 km dist), Grandchester (Grd; 27°68′ S, 152°45′ E, 75 km dist), Hickey Island (His; 29°43′ S, 153°36′ E, 0 km dist), Baradine (Bar; 30°95′ S, 149°07′ E, 345 km dist), Tambar Springs (Tam; 31°31′ S, 149°94′ E, 250 km dist), Gerongar Point (Ger; 34°56′ S, 150°81′ E, 0 km dist), Maldon (Mal; 36°98′ S, 143°77′ E, 125 km dist), Bulla (Bul; 37°65′ S, 144°77′ E, 25 km dist; Barker et al. 2009).
A 100-bp PCR fragment including both SNP601 and SNP610 in phsp26 was amplified with the primers Dbp26HRM1 (TGGTTCAAATACAGCAAGTG) and Dbp26R (CTACTTGGCGTTGTTTCCATT) using the LightCycler 480 High Resolution Melting Master kit (Roche). The reactions were run in 10-μl reaction volume with 3.0 mM MgCl2 and annealing temperature 55°C. A 146-bp fragment including SNP58 and SNP127 in hsp23 was amplified by the primers Db23F (ATGGCAAATCTTCCACTTC) and Db23HRMR (CAGCTGCTGGTCAAGAGGAA) and a 99-bp fragment including SNP166 and SNP184 was amplified with the primers Db23HRMR1 (ATGCACACCTGGAATCCGTC) and Db23HRMF (TTCCTCTTGACCAGCAGCTG). Both the hsp23 fragments were amplified under the same conditions as the phsp26 fragment. The genotypes which were known from the sequenced hsp23 and phsp26 genes were used as standard in the melting curves. Within each group of melting curves four flies were sequenced to see if the melting curve groupings correlate to the expected sequences.
The HRM curves were analysed using the LC480 software. In some cases the melting curves of normal homozygote and a mutant homozygote were identical, so in order to identify mutant homozygotes, the homozygote samples were mixed with a sample known to be a normal homozygote. The mutant homozygote samples give then a melting curve as heterozygote.
The deletion of Val18 and insertion of Asp-Lys in hsp26 were scored as microsatellites using primers located around the indels. The PCR products were analysed on an ALF express DNA sequencer together with internal and external size standards.
Nucleotide diversities were calculated using DNASP version 4 (Rozas et al. 2003). The associations between allele frequencies and geographical location and climatic variables were analysed using JMP v.7 (2007). Geographical location was defined using latitude, longitude and distance to the coast. Distance to the coast was included because climatological zones in eastern Australia tend to run parallel to the coast (Barker et al. 2005). Principal component analysis has been applied to 35 climatic variables for each locality estimated using the BIOCLIM programme of the ANUCLIM 5.1 package (Barker et al. 2005; Houlder et al. 2000). The climate was characterised by the scores for the first four principal components (PC1, PC2, PC3 and PC4), which account for 93% of the climatic variation. Of this, PC1 accounts for 43.0% and PC2 for 31.8% (Barker et al. 2005).
The structure and the sequences of the shsps in D. buzzatii were determined by a combination of DNA walking and PCR derived from amplifying genomic DNA with primers designed from D. mojavensis sequences. A nucleotide sequence of 12,545 bp was determined (Genbank accession no. EU196734). By comparison of the sequence with those of the shsp genes in D. melanogaster and using the same nomenclature for the homologous genes, we found that the region contains hsp67Ba, hsp23, hsp27 and hsp26 in two copies: we call one phsp26 and the other hsp26 (Fig. 2). In D. melanogaster hsp26, hsp67Ba, hsp23 and hsp27 are located between the genes CG4461-RA and CG4080-RA with hsp27 located closest to CG4080-RA (Fig. 2). In D. buzzatii hsp27 is located closest to CG4461-RA and the hsp26 gene is located between hsp27 and CG4461-RA. This suggests that the region containing phsp26, hsp67Ba, hsp23 and hsp27 is inverted in the D. buzzatii genome.
The most likely location of the shsp locus on the D. buzzatii chromosomes was found by comparing the locations of mapped D. buzzatii RAPD fragments from Laayouni et al. (2000) and D. buzzatii microsatellites (Barker et al. 2009) with the corresponding location in the D. mojavensis genome scaffolds. The D. mojavensis shsp region which is homologous to the D. buzzatii shsp region is located around D. mojavensis scaffold 6680: 7.190.000, whereas the D. buzzatii RAPD 70.09.6sts is located at D. mojavensis scaffold 6680: 7.007.768 and the D. buzzatii microsatellite M142 (now Db142) is located at D. mojavensis scaffold 6680: 7.131.400 (Frydenberg et al. 2002). As the D. buzzatii RAPD 70.09.6sts and Db142 are located on D. buzzatii chromosome 4 outside inversion 4s this is probably also the case for the D. buzzatii shsp locus (Laayouni et al. 2000; Barker et al. 2009).
Drosophila species are classified into the Sophophora subgenus and the Drosophila subgenus. The Sophophora subgenus is then divided into the willistoni, the obscura and the melanogaster species groups, whereas the Drosophila subgenus is divided into the Hawaiian Drosophila, the repleta group and the virilis group. In order to date the inversion and the duplication, we analysed the gene order of the homologous genes of CG4080-RA, the shsps and CG4461-RA in seven Drosophila species: D. melanogaster and D. ananassae from the melanogaster group, D. pseudoobscura from the obscura group, D. willistoni from the willistoni group, D. grimshawi from the Hawaiian Drosophila, D. virilis from the virilis group and D. mojavensis from the repleta group, using the Flybase consortium database. The results show that all the Sophophora species and D. grimshawi contain one copy of hsp26, while D. virilis and D. mojavensis both contain two hsp26 genes as in D. buzzatii and that all species from the Drosophila group have the inversion. A molecular phylogenetic tree of the hsp26 genes was constructed by the ME method (Fig. 3). The tree is identical to the tree described above constructed from the sequence of other Drosophila genes (Drosophila 12 Genomes Consortium 2007).
In order to localise putative response elements in the promoter region of the shsp genes of D. buzzatii, we analysed the conserved sequences in the promoter region between the shsp genes of D. buzzatii and D. melanogaster in more detail using three different alignment methods: DiAlign, DBA and VISTA.
We found two possible HSEs and a TATA-box conserved in the intergenic region between hsp67Ba and hsp23 (Supplemental data Table S1). Two regions are overlapping with regions earlier found to be important for hormone induction of hsp23 (Mestril et al. 1986; Rogulski and Cartwright 1995; Dubrovsky et al. 2001) and three regions contain a proneural transcription factor binding site CAGCTG sequence (E-box) (Powell et al. 2004).
The D. buzzatii phsp26 promoter and D. melanogaster hsp26 promoter homologous regions contain two regions located in the region important for ovarian expression. One region represents an alignment of the D. melanogaster distal GAGA element to a possible D. buzzatii GAGA element and two regions contain a possible HSE and a TATA-box, respectively (Supplemental data Table S2).
Searching for homologous regions between the D. buzzatii hsp26 and D. melanogaster hsp26 promoters only reveals few homologous regions (Supplemental data Table S3). One region is a possible TATA-box, one is a possible D. buzzatii HSE and one is an alignment of the first part of proximal D. melanogaster GAGA elements to a possible proximal D. buzzatii GAGA element.
When the D. melanogaster and D. buzzatii hsp27 promoter regions were aligned two regions known to be involved in the ecdysone induction of hsp27 were found. One of them contains a possible D. buzzatii ecdysone response element (EcRE). Two regions contain possible HSE and one contains a TATA-box (Supplemental data Table S4).
Polymorphic sites in the coding region of hsp23, hsp26 and phsp26 were identified by sequencing the genes in 40 flies from four different D. buzzatii populations from Argentina. In hsp23 we found 15 synonymous sites and four non-synonymous sites. Three of the non-synonymous mutations are changing of a Valine. SNP58 was changing Val20 to Met20, SNP127 Val43 to Ileu43 and SNP166 Val56 to Ileu56. Val20 and Val56 are conserved between the D. melanogaster Hsp23 and the D. buzzatii Hsp23. The last non-synonymous mutation SNP153 is a change of GluN51 to His51 and was found only in one individual. In hsp26 there were 22 synonymous sites and two indels, a deletion of GCT at 52–54 that leads to deletion of Val18 and insertion of GATAAG at 606 that leads to an Asp-Lys repeat. In phsp26 there were 44 synonymous sites and three non-synonymous sites. The three non-synonymous mutations in phsp26 were SNP601, SNP607 and SNP610. SNP601 was changing Pro101 to Ser101, SNP607 Thr103 to Ala103 and SNP610 Thr104 to Ala104.
The levels of nucleotide heterozygosity of hsp23, hsp26 and phsp26 were estimated by means of π and θ and compared to π and θ values of other D. buzzatii genes (Table 1; Gómez and Hasson 2003; Piccinali et al. 2004, 2007). The total π and θ values for hsp23 and hsp26 were lower than that for Xdh, α-E5 and Est-A, whereas for phsp26 it was in between the values for the other genes. But looking at the values for synonymous and non-synonymous variation separately, the differences between the hsp genes and other genes are more striking as the values of πns and θns for all the hsp genes are low as compared with the values for other genes. The same is the case when the percentages of non-synonymous mutations are compared.
The allele frequencies of the non-synonymous mutations in hsp23 and the two non-synonymous mutations in phsp26 were determined in nine populations from Australia (Fig. 1). The fragment which was used to determine the SNP166 in hsp23 was found also to include a non-synonymous mutation at 184 in some of the Australian samples. SNP184 was a change from Val62 to Leu62. Val62 is conserved between D. buzzatii and D. melanogaster. SNP153 was not found in the Australian populations. The allele frequencies of the four other hsp23 mutation (SNP58, SNP127, SNP166 and SNP184) and the three phsp26 mutations (SNP601, SNP607 and SNP610) are shown in Table 2. The presence of the indels in hsp26 was determined in the same populations using fragment size variation. The two indels in hsp26 were only found once in the Australia populations and therefore not analysed further.
The allele frequencies of the different SNP in phsp26 and hsp23 were compared to geographical location (longitude, latitude and distance from the coast) and the four principal components (PC) which account for 93% of the climate variability and of this PC1 accounts for 43%. None of the polymorphic SNP in phsp26 show any effect of either geographical or climatic variables. When the hsp23 mutations were analysed, SNP127 showed a linear association with longitude (R2=0.40, P<0.038, Fig. 4a). Historical records and microsatellite variation (Barker et al. 2009) indicate that the southern populations represent a secondary colonisation from the north. If the two southern populations Bul and Mal were omitted, the SNP127 showed an even stronger linear association with longitude (R2=0.93, P<0.0002, Fig. 4a). Since SNP58, SNP127, SNP166 and SNP184 all are changes of a Valine, we analysed the total frequencies of these four SNP (hsp23Val). hsp23Val showed a linear association with longitude (R2=0.66, P<0.0048, Fig. 4b) and with the climatic PC1 (R2=0.87, P<0.0002, Fig. 4c). If the two southern population are removed the fit is increased with longitude (R2=0.84, P<0.002, Fig. 4b) but not with PC1 even though it still is significant (R2=0.81, P<0.004, Fig. 4c). Finally, the frequencies of hsp23Val show a linear association with distance to coast (R2=0.80, P<0.004, Fig. 4d).
In this study we sequenced the chromosome region containing the phsp26, hsp67Ba, hsp23, hsp27 and hsp26 genes in D. buzzatii, a member of the repleta group from the Drosophila subgenus. The gene region was inverted as compared with species of the Sophophora subgenus, and the hsp26 gene is duplicated in the virilis and the repleta groups. Some inversions arise by simple cut and paste events, others arise by more complicated events which may result in duplication of coding sequences at the break points (Matzkin et al. 2005; Sharakhov et al. 2006). Since D. buzzatii phsp26 and D. buzzatii hsp26 are located at the borders of the inversion, it seems likely that the hsp26 duplication originated together with the inversion event, but there could be several explanations of the origin of the inversion/duplication events.
The first explanation includes three inversion events. In the molecular phylogenetic tree the two D. virilis hsp26 genes are more closely related to each other than either is to the D. buzzatii and D. mojavensis hsp26 genes, whereas the D. buzzatii and the D. mojavensis hsp26 genes are clustered together and the D. buzzatii and the D. mojavensis phsp26 genes are clustered together. This would date the inversion/duplication event after the separation of the virilis group and the repleta group and include two inversion events. But D. grimshawi also has the inversion, so the inversion could also have happened in this lineage but in this case without duplication of hsp26. The reoccurrence of inversion breakpoints is common so it is not unlikely that the three inversion events share breakpoints (González et al. 2007).
The second explanation could be that the inversion/duplication happened only once at the Sophophora/Drosophila separation. In this case D. grimshawi must have lost one of the hsp26 copies. D. grimshawi has several rearrangements including the shsp loci, which include duplication of hsp23 and part of hsp27 and moving of the hsp23-CG4461 region away from the hsp67Ba gene. Therefore, we consider a deletion of the hsp26 gene in this species a possible event. It is more difficult to explain the clustering of the D. virilis hsp26 genes in the hsp26 tree. Here one explanation could be divergent selection on each of the two hsp26 genes in the repleta group after the separation from the virilis group. This may also lead to the closer relationship between the D. grimshawi hsp26 and the two D. virilis hsp26 genes than to the repleta group hsp26 genes.
A third explanation could be that the inversion happened at the Sophophora/Drosophila separation and the duplication of the hsp26 genes happened twice, once in the repleta group and once the virilis group. However, we find this latter explanation unlikely due to the location of the duplicated hsp26 genes at each end of the inversion.
Transcription factor binding sites were localised in the shsp promoter regions on the basis of coincidence between studies of important regulator regions and CNS found in alignments of the D. buzzatii and D. melanogaster shsp promoter sequences. The hsp23 promoter contains three CNS with proneural transcription factor binding site E-boxes, each surrounded with a specific conserved sequence which may indicate that each E-box is specified for a specific proneural transcription factor which then regulates the expression of hsp23 in different tissues (Powell et al. 2004). The cell-specific and temporary expression of hsp23 in the imaginal discs has been shown to be regulated by proneural transcription factors (Alonso et al. 2005).
When the gene of a multifunctional protein is duplicated, the two copies sometimes get different functions, either by mutation in the coding region or by separation of regulator sequences (Force et al. 2005). The D. melanogaster hsp26 is expressed in spermatocytes, nurse cells and the epithelium of larvae, pupae and adults. The sequences which regulate the tissue-specific expression and the heat shock expression of D. melanogaster hsp26 have been shown to be located in different parts of the promoter. Three regions involved in spermatocyte-specific expression are identified downstream from −352, whereas the ovarian expression is dependent on the region (−522 to −352) (Cohen and Meselson 1985; Frank et al. 1992; Michaud et al. 1997).
In the comparison of the D. buzzatii hsp26 and phsp26 promoters to the D. melanogaster hsp26 promoter, we find that both the D. buzzatii genes retain the regulatory sequences for heat shock expression. In the D. buzzatii phsp26 there are two CNS located in the region important for ovarian expression, so D. buzzatii phsp26 may have retained the ovary specific expression. On the other hand the D. buzzatii hsp26 gene seems to have lost its tissue-specific expression as we found no CNS located in regions important for ovarian expression.
The D. melanogaster hsp27 gene contains an EcRE but three regions are important for ecdysone induction during oogenesis and embryogenesis (Riddihough and Pelham 1986). The D. buzzatii hsp27 seems to have retained both the heat shock and the tissue-specific expression as the promoter contains homology to both HSE and EcRE binding sites and has a CNS located in a region important for ecdysone induction.
There is a great difference between the percentages of non-synonymous replacements in D. buzzatii hsp23, hsp26 and phsp26 and all of them are lower than the percentages for D. buzzatii Xdh, α-E5 and αE5 (Table 1; Gómez and Hasson 2003; Piccinali et al. 2004, 2007). But analyses of the DNA variation in different genes within and between Drosophila species have shown that there is great variation in the percentage of non-synonymous replacements both between genes within species and between species (Moriyama and Powell 1996; Andolfatto 2001; Piccinali et al. 2004). A low level of non-synonymous mutations is probably connected to high evolutionary constraints in the hsp genes. This is in agreement with the high level of conserved sequence in hsp genes in general. Both hsp26 and phsp26 have a low number of non-synonymous mutations which makes it likely that they both are active genes in D. buzzatii.
The frequency of SNP127 shows a linear association with longitude. If we leave out the two southern populations the strength of the association is increased. The total frequency of Valine changes in hsp23 also shows a linear association with longitude, distance to the coast and the climatic variable PC1 when we leave out the two southern populations. A study of the population differentiation and colonisation of the Australian D. buzzatii populations has shown that the three most southern populations have lower genetic variability than the other populations and it was concluded that they probably arose by a secondary colonisation from the northern populations (Barker et al. 2009). Thus it is reasonable to delete these populations from our analysis but Ger, which is the most northern population of the three, has a non-synonymous mutation SNP184 that is exclusively found in this population and which as SNP58, SNP127 and SNP166 leads to a Valine change. Therefore we leave Ger in our Valine change analysis.
The changes in hsp23 are changes of Valine to either Methonine (SNP58, Val20), Isoleucine (SNP127, Val43 and SNP166, Val56) or Leucine (SNP184, Val62), which are all hydrophobic amino acids. Valine is less hydrophobic than the three others amino acids, so all the mutations make Hsp23 more hydrophobic. Gottler et al. (2008) have shown that changing of the hydrophobicity of antimicrobial peptide protegrin-1 by Valine to Leucine changes the biological activity of the peptide. So it seems reasonable to claim that the total amount of Valine in Hsp23 could be connected to the function of the protein.
The three variables PC1, longitude and distance to the coast are all correlated but we have chosen to make a separate model for each of them to get a broader biological understanding of the result. SNP127 was significant only with longitude. PC1 and distance to the coast each have some distinctive characteristics. hsp23Val shows a significant association with PC1 both with and without the two southern populations, whereas it only shows a significant association with the distance to the coast when the two southern populations were deleted, so the climatic information seems important.
Much of the difference between the D. buzzatii hsp26 and phsp26 genes is located in the 3´end where hsp26 has a repeated sequence which codes for Asp-Lys-(Glu-Lys)4. The insertion in hsp26 leads to an increase of the repeat by Asp-Lys. The non-mutated phsp26 does not have a repeat in the 3`end but SNP601 leads to a repeat of 6 bp since TCGAAGCCGAAG changes to TCGAAGTCGAAG which gives a repeat of Ser-Lys in the protein. We have earlier found D. melanogaster hsp26 to be polymorphic for an insertion of 9 bp (GACGGCAAG, Asp-Gly-Lys) in the 3´end where 8 bp are part of a repeat. This variation showed a latitudinal cline in Australian D. melanogaster populations (Frydenberg et al. 2003). The D. buzzatii hsp26 and phsp26 repeats found in this study were very rare in the Australian populations and there seems to be no correlation with environmental variation in the frequencies of these SNP. But D. buzzatii and D. melanogaster are living in very different niches, so parallel latitudinal clines could not necessarily be expected. However, it is notable that all the hsp26 genes are polymorphic for repeats in the 3´end and all the repeats involve a Lysine.
In summary, we show that the region which includes the D. buzzatii shsp genes is inverted as compared with the D. melanogaster genome structure and the hsp26 gene is duplicated. The D. buzzatii hsp23 has four SNP that all lead to a replacement of a Valine. Summation of the frequencies of these SNPs shows a linear association with longitude, distance to the coast and climatic variables in D. buzzatii populations from Australia. In future work, the frequency of these hsp23 SNPs should be investigated in different parts of the world to look for similar associations with longitude, distance to the coast or climatic variables.
Below is the link to the electronic supplementary material.
Homologous sequences in D. buzzatii and D. melanogaster hsp23 promoters (DOC 33 kb)
Homologous sequences in D. buzzatii phsp26 and D. melanogaster hsp26 promoters (DOC 31 kb)
Homologous sequences in D. buzzatii hsp26 and D. melanogaster hsp26 promoters (DOC 29 kb)
Homologous sequences in D. buzzatii and D. melanogaster hsp27 promoters (DOC 31 kb)
The authors thank Jesper G. Sørensen for help catching some of the Australian fly samples and Fabian Norry for some of the Argentinean fly samples, Jürgen Tomiuk and Cino Pertoldi for critical reading of the manuscript and the Novo Nordisk Foundation for financial support.