Index INTRODUCTION MATERIALS AND METHODS Fly stocks SNP mapping Analysis of the candidate region by PCR Cloning of CG18468 and transformation of flies Sequence comparisons RESULTS Genetic mapping of sd, jba and Lhr SNP mapping of Lhr Structure and sequence analyses of the Lhr strain Transformation experiments with D. simulans CG18468 DISCUSSION Mapping of Lhr Identification of Lhr Gene expression References

INTRODUCTION

In sexually reproducing organisms, speciation is generally accomplished by reproductive isolation (Dobzhansky, 1937; Mayr, 1942; Coyne and Orr, 2004). The reproductive barrier results from the accumulation of genetic incompatibilities due to epistatic effects between genes of incipient species (Dobzhansky, 1936; Muller, 1942; Orr and Turelli, 2001). However only a few genes involved in hybrid incompatibilities have been identified, and mostly in Drosophila species (Orr et al., 2004; Wu and Ting, 2004). Such genes termed as “speciation genes” are acting in hybrids between Drosophila melanogaster and D. simulans, leading to their lethality or sterility (Sawamura, 2000). The discovery of mutants that rescue hybrid viability or fertility indicated that genetic incompatibilities were not as complicated as it was thought and gave the reasonable hope to elucidate the genetic basis of the reproductive isolation in Drosophila (Sawamura et al., 1993).

The hybrid males produced by the cross between D. simulans males and D. melanogaster females usually die at a larval or prepupal stage (Sturtevant, 1920). This lethality is due to a genetic incompatibility between the D. melanogaster X chromosome and the D. simulans autosomes (Sturtevant, 1920; Yamamoto, 1992). Several mutants are known to restore the viability of hybrid males that otherwise die. They are the Lethal hybrid rescue (Lhr¹ and Lhr²) of D. simulans and several mutants of D. melanogaster: Hybrid male rescue (Hmr¹ and Df(1)Hmr^–); In(1)AB and Df(1)EP307-1-2 (Watanabe, 1979; Hutter and Ashburner, 1987; Hutter et al., 1990; Barbash et al., 2004; Brideau et al., 2006; Barbash and Lorigan, 2007). Interestingly the rescuing mutations of D. melanogaster are all located on the X chromosome while Lhr has been mapped on the second chromosome of D. simulans. This situation correlates with the expected position of the interacting incompatible loci of both species. It was then supposed that Lhr might be the autosomal speciation gene of D. simulans that interacts in a deleterious way with the D. melanogaster X chromosome and possibly with Hmr (Brideau et al., 2006). The Hmr gene has been the first one to be identified at the molecular level and was shown to actually induce the hybrid lethality (Barbash et al., 2003). More recently Brideau et al. (2006) proposed an identification of the Lhr gene. However the gene they have cloned does not produce genetic incompatibility in transgenic male of D. melanogaster, contrarily to the two Dobzhansky-Muller genes interaction model and therefore they proposed several reasons including the requirement of a hybrid genetic background. They genotyped only ten recombinants between Lhr and the distal visible marker jabara (jba) and consequently defined a candidate region of “several hundred kilobases centromere-proximal to qkr54B” (Brideau et al., 2006). The mapping being insufficient to identify the gene they estimated CG18468 to be a good candidate gene on the base of sequence characteristics. Their identification of the Lhr gene with CG18468 was strengthened by the observation of mutations in that gene in both Lhr mutant stocks. In the Lhr¹ mutant they observed the presence of a repetitive retrotransposed sequence in the 5' UTR region. However the gene is not disrupted and it is still expressed in a lesser amount (Brideau et al., 2006). The rescue effect of the Lhr² mutant was suggested to result from the absence of a small internal duplication commonly observed in wildtype CG18468 (Brideau et al., 2006). However it was recently suggested that the absence of the duplication in CG18468 as observed in the Lhr² mutant was not sufficient to rescue the hybrids (Nolte et al., 2008). Further transgenic analyses in D. melanogaster suggesting the complementation of the Lhr¹ mutation in hybrids led them to conclude that CG18468 is Lhr (Brideau et al., 2006). However alternative interpretations could lead to a different conclusion. Genes producing hybrid incompatibility are not rare and CG18468 could be one of them in addition to Lhr. For example a small introgression of the D. simulans 2L chromosome into D. melanogaster led to male hybrid lethality by suppressing the rescuing effect of the Lhr mutation (Sawamura, 2000). In the same order, Presgraves (2003) showed that at least 20 small independent chromosomal regions of D. simulans when uncovered by a deficiency on the counterpart D. melanogaster chromosome were able to kill the hybrid males otherwise rescued. Furthermore, he estimated that 191 hybrid-lethal incompatibilities separated D. melanogaster and D. simulans (Presgraves, 2003). Moreover the use of the GAL4/UAS system in Brideau et al. (2006) to drive the expression of the transgene might have resulted in an ectopic and/or elevated expression causing the hybrid lethality, independently of the normal expression of the gene. The sequence characteristics of CG18468 suggest functional similarity with Hmr, which may explain that it is naturally involved or ectopically forced to act in the same process of incompatibility. The observation that the forced expression of the CG18468 transgene killed the hybrid males despite the presence of the Lhr¹ rescuing mutation but not in the presence of the Df(1)Hmr^– rescuing mutation was not sufficient to prove that CG18468 was Lhr. Indeed, a transgene containing a gene acting in the same incompatibility pathway but downstream to Lhr would not give distinguishable result.

Here, we report a more extensive mapping of the Lhr¹ mutation with the use of the proximal visible marker spread (sd) and then completed by the use of SNPs, The resulting candidate region spans less than 35 kb and contains only 10 ORFs. The retrotransposed sequence insertion upstream to CG18468 is the most noticeable change in the candidate region. Unlike the former characterization of Lhr we have cloned the genomic coding sequence of CG18468 together with its potential native promoter for transgenic experiments in D. simulans. We thus observe the deleterious effect of the transgene on the male hybrids in spite of the presence of the Lhr¹ rescuing mutation. These results prove that CG18468 is the Lhr gene.

MATERIALS AND METHODS

Fly stocks

All stocks are maintained on standard cornmeal-yeast-glucose-agar medium at 23–24°C. Mutant stocks of D. simulans: Lhr (ks67), sd jba (M43B + 1; isogenic for the second chromosome), w^S2-6, w^ak; sd Lhr jba (for phenotypic description, Sturtevant, 1929; Watanabe, 1979; Yamamoto et al., 1997 and Flybase, http://flybase.org). w^ak (for akebi) is a newly isolated white mutant. Stocks of D. melanogaster: y w/y⁺Y, Df(2R)Jp1 (51C3; 52F9; with the CyO balancer that does not complement sd in hybrids) and several chromosome deficiencies provided by the Bloomington Stock Center and Kyoto Stock Center (DGRC): Df(2R)50C-101 (50C21; 50D1-5), Df(2R)BSC18 (50D1; 50D2-7), Df(2R)Exel7130 (50D4; 50E4), Df(2R)BSC11 (50E6-F1; 51E2-4), Df(2R)Exel6066 (53F9; 54B6), Df(2R)BSC44 (54B1-2; 54B7-10), Df(2R) 02B10w-08 (54E8; 54F3-4), Df(2R)14H10w-35 (54E5-7; 55B5-7), Df(2R)Pcl11B (54F6; 55C13), Df(2R)PC4 (55A; 55F), Df(2R)Exel7153 (55B9; 55C1).

SNP mapping

Lhr was crossed to sd jba to allow recombination between their second chromosomes. Female progeny was backcrossed to sd jba and both sd and jba recombinant sons were selected. From each recombinant male a line was temporarily established and maintained heterozygous by backcrossing the males at each generation. For each recombinant line, DNA of 5–10 flies was extracted according to the DNeasy 96 protocol of Qiagen. SNP loci between the sd jba (M43B + 1) and Lhr (ks67) strains were identified by short sequences and recombinants were subsequently genotyped by Takara Bio Inc. (Japan). The list of primers used for SNP typing is provided in Table 1.

View Details

Table 1 List of primers used to define and type SNPs with their chromosomal locations

Analysis of the candidate region by PCR

Genomic DNA was extracted from D. simulans Lhr and sd jba stocks, and PCR was performed with ExTaq (Takara Bio Inc. Japan). Primers used for PCR and sequence:

cnkU2: TGGTGGACAGACAGGGAAGAC;

l(2)k01209U1: CGAAAACCATGTACGGGGCCAA;

l(2)kR1: GTAGCGAGAAGTGCGAGGTG;

l(2)k1183f: CTACAAGGACCAGCCAATGCC;

CG6550U1: GCGATTGGACGAGGTGTGCAAA;

CG6550-930f': CTGTATCCTCGGAAGTAAGCC;

CG4802-456r: GTGCTTTGCCATCGTAGAAGG;

CG4802-483f: CGATGGCAAAGCACAAAGTCC;

CG4802R1: CATTGCTATGGGCGGATCAC;

Bap55-71r' GAAAAGTATTCGGAACGCCCC;

Bap55-778f': CTGTCGTGAAAGGAAGTCCCC;

CG18468-1132r: AGCGTGGTGTTTCTATCGGTC;

EDTP-627r': TGCCTACGACTGCCTGAGTGG;

CG18468U1: TATACAGTTGAAGTAGTACACAA;

CG18468R1: GCAAGCGAAACATAAAATGCGA;

Bap55R1: TGATGTCCGGGATCGAAGACCA;

CG18468R2: CGCTGTCGGTACTCATTTTGAT.

Cloning of CG18468 and transformation of flies

Genomic DNA of the sd jba stock was used to amplify a fragment containing the full sequence of CG18468. Amplification was done using KOD Plus polymerase (Toyobo Co. Ltd., Japan) and primers Bap55-778f' and EDTP-627r'. Amplified DNA was digested by PstI and cloned in pBlueScript previously treated by PstI and EcoRV. A clone was digested by NotI and XhoI and the resulting DNA fragment was subcloned in pCaSpeR4. Concentration was adjusted to 250 ng/μl with 50 ng/μl helper DNA in injection buffer. The construct was injected in w^S2-6 of D. simulans. Transgenic lines were crossed to w^ak; sd Lhr jba and successively backcrossed for introducing the transgene under Lhr background. Insertion sites have been determined by inverse PCR using primers internal to the pCaSpeR4 described in Berkeley Drosophila Genome Project Homepage (http://www.fruitfly.org/index.html).

Sequence comparisons

Sequence reactions have been performed according to ABI protocols and run on ABI capillary sequencer. Sequences of CG18468 with flanking regions were produced for both Lhr and sd jba strains. The coding sequence of CG18468 has also been sequenced for the recombinant jba154. Sequences were aligned by GENETYX-MAC Ver. 12 (Genetyx Co. Ltd., Japan) and subsequently adjusted by hand in SeqPup (http://iubio.bio.indiana.edu/soft/molbio/seqpup/).

RESULTS

Genetic mapping of sd, jba and Lhr

The Lhr mutation was previously mapped between sd and jba (Yamamoto et al., 1997). The corresponding genes for sd and jba have not been identified in D. melanogaster and their position is cytologically uncertain. The closest marker genes available in D. simulans and precisely identified are vestigial (vg) and plum (pm), which are homologous to vg and Punch (Pu; Sugaya and Fuyama, 1995) of D. melanogaster (Fig. 1A–B). The genes vg and Pu are about 7800 kb apart. Second chromosomes are cytologically homosequencial between species (Aulard et al., 2004). Consequently the arrangement of genes was expected to be conserved and gene positions to be reflected in the recombination map. However, complementation tests using chromosomal deficiencies of D. melanogaster showed the jba gene to be in the region 55A1-55B5-7, indicating a discrepancy between the recombination map of D. simulans and the cytological map of D. melanogaster (Fig. 1A–B). We took advantage of a previously identified CyO balancer chromosome that should carry a sd mutation of D. melanogaster (Yamamoto et al., 1997) to map the sd gene of both species. The location of sd turned out to be in 50D4-50D7 (Fig. 1B).

View Details

Fig. 1 Mapping of the Lhr gene. (A) Recombination map in D. simulans. (B) Cytological map according to D. melanogaster. Deficiencies are depicted by a black bar or a white bar whether or not they complement the mutation. (C) Mapping by SNPs. Each SNP is numbered in a square above the cytological map. Each row is a different round of SNP genotyping. Number and location of recombinants are shown in arrowhead boxes directed toward the putative site of Lhr. *Triple recombinant sd25. **Inconsistent recombinant jba154. (D) Gene map of the candidate region. CG18468 is marked by a black box. PCR amplified DNA fragments are indicated below the gene organization. (E) The insertion in the Lhr mutant is shown at the putative transcriptional start of CG18468. The wild type DNA fragment used for transgenic experiment is represented below. Rectangular cases depicted coding exons while circular ones are untranslated sequences of mRNA. Note that mRNAs of CG18468 and EDTP are expected to partially overlap.

From sd jba and Lhr chromosomes, 983 recombinants were recovered and successfully analyzed for the survival of hybrids to determine whether they carried Lhr or not. The numbers of recombinants obtained between sd – Lhr and Lhr – jba were 821 (374 sd Lhr plus 447 jba), and 162 (73 sd plus 89 Lhr jba), respectively, indicating the location of the Lhr gene in the subdivision 54B between sd and jba (Fig. 1B). The candidate region contains 20 ORFs from mbl to qkr54B, including nine CG genes. Deficiencies in that region (Df(2R)Exel6066 and Df(2R)BSC44) have no effect on hybrid lethality when mated with Lhr. As usual, Lhr⁺ hybrid males are lethal when made hemizygous by any of these deficiencies. Consequently, these complementation tests do not help to refine the location of Lhr.

SNP mapping of Lhr

SNPs were searched between two second chromosomes, one bearing Lhr and the other bearing sd and jba (line M43B + 1). In total, 745 recombinants in the sd-jba interval between the two chromosomes have been used for typing. At first round, 96 recombinants were typed for five SNPs within the cytological locations from 53F to 55B (Fig. 1C). All recombinants, except sd25, indicate Lhr to exist between SNPs 1 and 2. Eight recombinants were found between these two SNPs but only one is distal to Lhr, suggesting that Lhr is closer to the SNP2. Three new SNPs, 6, 7 and 8 were chosen in consequence, to type these eight recombinants, together with newly added 88 recombinants. The second round typing showed that Lhr should be between SNPs 6 and 2, in which 12 recombinants were identified. In the third round, these 12 recombinants and additional 84 (including sd25) were examined with three new SNPs (9, 10 and 11) and Lhr was further narrowed down to between SNPs 10 and 11. Interestingly, the sd25 recombinant shows another recombination in the candidate region, implying that a triple recombination occurred. Another triple recombinant (jba132) is observed in the candidate region. In the fourth round these two particular recombinants together with 94 others were typed at the four SNPs (7, 8, 10 and 11) included in the candidate region. However no more recombinants were found in the candidate region. The recombinant jba154 was unexpected as it failed to rescue male hybrids while it should have the Lhr copy according to SNP typing, which was confirmed by further sequence analysis. This result is probably due to a progeny too small (56 females) and a lower viability of hybrid males rather than a new hybrid male lethal mutant. The interspecific cross could not be done again as the chromosome has not been kept as a stock. In the fifth round 384 more recombinants were typed for SNP10, SNP11 and a new intermediate, SNP12. Four new recombinants were found in the region but did not help narrowing it down further. None of them is informative about which side of SNP12 Lhr is located. Consequently, summing up the recombinants identified in the third and fifth rounds a candidate region can be defined by four recombinants distal to SNP10 but proximal to Lhr and four other recombinants proximal to SNP11 but distal to Lhr (Fig. 1D). Thus the SNP mapping allowed us to restrict Lhr to the region between CG30101 and CG18467, in which 10 ORFs are included in D. melanogaster genome (Fig. 1D).

Structure and sequence analyses of the Lhr strain

Our PCR examinations of the Lhr and sd jba chromosomes and the comparison of genomic sequences of D. melanogaster and D. simulans revealed that both species share the same chromosome organization in the region concerned (Fig. 1D). However as in Brideau et al. (2006) an insertion of 3 kb long has been found in the Lhr strain between Bap55 and CG18468. No other obvious size difference has been detected. We confirmed by PCR the presence of that insert consistently in 18 different Lhr sublines. The region including the insertion and CG18468 was sequenced in both Lhr and sd jba chromosomes. In the coding sequence of CG18468, there are only two synonymous substitutions. The insertion possesses a poly-A tail suggesting a retrotransposition event. Three nucleotides have been apparently deleted at the insertion site, as they exist on the sd jba chromosome. Further Blast searches against mobile element databases (FlyBase) indicate that the insert has stronger affinity to G-element and at a lesser extent to G2, hopper2, Stalker4 and G6 elements. However the homology never exceeds 10% of the sequence, suggesting that the insert might belong to another family. Clearly the insert is a 5’-truncated non-LTR G-type retrotransposon inserted at the putative transcription start of CG18468 and in the same transcriptional direction (Fig. 1E).

Transformation experiments with D. simulans CG18468

A DNA segment including the wild type allele of D. simulans CG18468 was cloned together with its 5' and 3' flanking regions as illustrated in Fig. 1E, and integrated into pCaSpeR4. Several transgenic lines were established in w^S2-6 of D. simulans. The insertion sites of all transformants were determined by inverse PCR (Table 2). Most of the transgenic lines are homozygous viable except one homozygous lethal with insertion on the second chromosome (s12F1F1). Despite the Lhr background, hybrid males with the transgene are not rescued, except for the lines s2F1F1 and s12F1F1 for which several w⁺ males emerged but with a very low viability. In both exceptional cases the transgene is inserted in intronic sequence, respectively in the Ptpmeg and Calmodulin genes and consequently it might not be correctly expressed. This transgenic experiment clearly indicates that one copy of CG18468 introduced as a transgene complements the Lhr mutation by killing hybrid males that are otherwise rescued. In other words, it demonstrates that the D. simulans gene corresponding to CG18468 is the Lhr gene. Furthermore, these transgenic experiments indicate that the Lhr mutation behaves as a loss of function mutation, and it is recessive over the wild type allele that kills hybrid males.

View Details

Table 2 P{Lhr+} insertions in D. simulans w; Lhr and hybrids from the males heterozygous for the insertion crossed to D. melanogaster yw females

DISCUSSION

Mapping of Lhr

Our results of the genetic mapping and the SNP mapping are consistent, pointing out a candidate region in 54B. The discrepancy with the previous report in 54E-F (Yamamoto et al., 1997) is largely due to the precise mapping of sd in 50D in the present study. Moreover it appears that the recombination rate is not uniform along the chromosome of D. simulans between vg and pm. It is either lower than average between sd and Lhr or higher than average between jba and pm. The lack of closer visible markers to Lhr and the subsisting uncertainty on the jba location do not allow a better genetic mapping of Lhr. The use of SNPs partially overcome this difficulty and allowed to map Lhr in 54B7-16. Despite the analysis of 745 recombinants, it was not possible to better map the mutation, possibly due to the mobile element insertion that can affect recombination closer to Lhr.

Identification of Lhr

The presence of Lhr as a gene unique to D. simulans was a possibility. However the analysis of the candidate region determined by SNP mapping revealed the same gene structure in both species. As mutations in Drosophila often result from the insertion of mobile elements (Green, 1988), the detection of a unique truncated G-type element in the candidate region of the mutant stock was the best indicator for the Lhr mutation. However the insertion does not disrupt any coding sequence, but was found between Bap55 and CG18468 as it has already been reported (Brideau et al., 2006). A rescuing effect from hybrid incompatibility can be generated by either the mutation of a speciation gene or the mutation of an unrelated gene possibly conserved in both species. In that latter case the same mutation of the orthologous gene should also rescue hybrids. But as reported by Brideau et al. (2006) the deletion of the candidate region of D. melanogaster does not rescue hybrid males and under the hypothesis that Lhr is a loss of function mutation, the mutated gene is more likely a speciation gene. Therefore the perfect conservation of the BAP55 protein between D. melanogaster and D. simulans together with the high homology in the upstream sequence were definitely excluding Bap55 as a speciation gene. On the other hand CG18468 shows a high divergence between both species, and thus fits the characteristics of a speciation gene (Brideau et al., 2006). Indeed, the few speciation genes that have been identified up to now show high divergence between species (Ting et al., 1998; Barbash et al., 2003; Presgraves et al., 2003) and this criterion was probably important in the choice of Brideau et al. (2006) for selecting CG18468 as the most probable candidate gene for Lhr.

Gene expression

As only two synonymous mutations are observed in the coding sequence of CG18468 between the Lhr and the wild type stocks, the Lhr mutation should affect the gene expression. Indeed a strong reduction of the gene expression is observed at early larval stage according to RT-PCR result (Brideau et al., 2006). The insertion of a non-LTR retrotransposon exactly at the putative initiation start for transcription of the gene, accompanied with a deletion of three base pairs are thus likely the reason. The Lhr mutant is not a complete loss of function which may also explain the absence of phenotype in D. simulans. Finally, in transgenic experiments CG18468 complemented the Lhr mutation, by killing the hybrid males that should have been rescued, confirming the result of Brideau et al. (2006). The only two exceptions might be due to position effect of the transgene insertion. It is important to note that in our experiments the transgene is under the control of its own promoter, avoiding any artifact due to ectopic expression. Furthermore it was the only way to confirm the identity of the Lhr gene. Moreover hybrids received the transgene from the D. simulans males together with the rescuing Lhr mutation, where the direction of gene transmission is the same as normal hybrid cross. Some caution has to be considered in the use of transgenic D. melanogaster as in Brideau et al. (2006), since the transgenic speciation gene is expected to produce deleterious effect in combination with D. melanogaster genome. It was thus possible that gene expression was already altered in the egg and consequently affected hybrids in an unusual way.

The identification of Lhr is an important contribution to the understanding of the molecular bases of speciation. However numerous questions are still open about the real function of the speciation genes and the exact molecular interaction that result in the hybrid incompatibilities. A priority will be to identify the determinant factor(s) allowing the completion of the deleterious interaction in which Hmr and Lhr are engaged.

We are grateful to Ms Ayako Takahashi (DGRC) for her technical assistance.