In order to determine whether Lsp1α is the only gene within its chromosomal region to escape dosage compensation, the dosage compensation status of the genes flanking Lsp1α was examined. Previous work indicated that the gene 5' of Lsp1α, CG2560, is dosage compensated 31. Several other genes in the region near Lsp1α were identified after publication of the Drosophila genome sequence.

Five putative genes exist in the region immediately surrounding Lsp1α as predicted by cDNA evidence from the Drosophila genome project 32: CG15926, CG2560 (referred to as L12 by 31), CG15730, CG2556 and CG11146 (Figure 1A). Since CG11146 is separated from Lsp1α by two intervening genes, it was not examined in this study. The developmental stage in which Lsp1α, CG15926, CG2560 and CG2556 are expressed was determined by Northern RNA hybridisation analysis (Figure 1B and 1C). As expected Lsp1α is very highly expressed and easily detected in total RNA from third instar larvae (Figure 1B). CG2560 is expressed in all larval stages, as previously reported. Four transcripts were detected for CG2556, which is downstream of Lsp1α (Figure 1C). These transcripts are expressed differentially throughout development but importantly are present in third instar larvae. Thus, genes 5' and 3' of Lsp1α are expressed at the same stage of development. Transcripts from CG15730, the gene immediately 3' of Lsp1α, were not detected at any of the developmental stages analysed using 2 μg of poly(A)⁺ mRNA, or in adults using RT-PCR (data not shown). It is possible that this intron-less gene may be expressed in only a few cells, or may require stimuli not present under standard conditions. Since expression could not be detected, it was not examined further in this study. CG15926 transcripts are present in ~2-fold higher levels in hemisected (head plus thorax) adult males compared to females as shown by RNase protection (data not shown), thus the dosage compensation status of this gene could not be determined. There were slightly higher transcript levels of the rp49 loading control in whole adult females than males (Figure 1C) as shown previously 33, possibly due to strong ovarian expression. For this reason, hemisected adults or sexed larvae were used in this study for determining a gene's dosage compensation status.

Multi-probe quantitative RNase protection analysis was used to determine the dosage compensation status of CG2560 and CG2556 genes in first and third instar larvae respectively. Control probes detected transcripts from the X-linked dosage compensated 6-phosphogluconate dehydrogenase (Pgd) 3435 and constitutive autosomal ribosomal protein 49 (rp49) 33 genes. First instar larvae were sexed using a stock in which only the male larvae express GFP. Female to male CG2560 and Pgd transcript ratios were normalised to rp49, as the 3 probes were analysed simultaneously. A female to male transcript ratio of one indicates that a gene is compensated, whereas a female to male ratio of two suggests that a gene is not compensated. CG2560 and Pgd have female to male transcript ratios of 1.02 ± 0.08 and 0.84 ± 0.20 respectively (Figure 1E) indicating that these genes are both dosage compensated. This concurs with the CG2560 transcript ratios obtained in second instar larvae using an alternative method 31. Single probe RNase protection of CG2556 and Lsp1α was conducted in male and female y w staged-third instar larvae relative to Pgd and rp49, as both transcripts are present at this developmental stage (Figure 1B,C). Female to male CG2556, Lsp1α and Pgd transcript ratios were not normalised to rp49, as the 4 probes were analysed separately due to the similar size of the protected RNAs. Rp49 and Pgd have female to male transcript ratios of 0.79 ± 0.27 and 0.91 ± 0.42 respectively (Figure 1F), demonstrating equivalent RNA levels are present in both sexes. CG2556 and Lsp1α have female to male transcript ratios of 0.99 ± 0.15 and 1.81 ± 0.14 respectively (Figure 1F), indicating that CG2556 is dosage compensated but Lsp1α is not. These results show that two of the genes flanking Lsp1α are dosage compensated, and suggest that Lsp1α is unique within its chromosomal domain in escaping regulation by the MSL complex.

The genes flanking Lsp1α

a

The precise chromosomal location of these X-linked transgenes was determined by inverse PCR. The I-arm-lacZ-I2:19C3 transgene has inserted between the X-linked CG1631 and CG15462 genes that are uncharacterised with respect to expression and are part of an approx. 140 kb gene poor region of the chromosome (19C2 to 19C5). The nearest genes that showed significant binding of MSL1 and MSL3 in embryos are Rab10 (19C1) and l(1)G0004 (19C6) that are approximately 95 kb upstream and 45kb downstream respectively from the site of insertion of the I-arm-lacZ-I2 transgene 2123 (Additional file 1; Panel B). The arm-lacZ:10D8 transgene has inserted in the first intron of the X-linked inaF (CG2457) gene, which encodes a protein with calcium channel regulator activity involved in rhodopsin mediated signalling that is expressed in the head and eye of adult flies 38. Although, it has not been reported if inaF is dosage compensated, significant binding of MSL1 is detected at the 3' end of this gene in embryos 23.

There was no binding of MSL1 or MSL3 to Lsp1α in embryos or of MSL3 to Lsp1α in SL2 cells 212223 (Additional file 1; Panel D). This would be the expected result since Lsp1α is not dosage compensated. In SL2 cells, ~90% of MSL3 binding clusters were within expressed genes, with an enrichment in the middle and 3' end 21. Since Lsp1α is not actively expressed in SL2 cells it was possible that MSL complex could be binding to Lsp1α in the tissue in which the gene is expressed, namely third instar larval fat body. As acetylation of H4 at lysine 16 is dependent on the MOF component of the MSL complex, we measured the relative level of H4K16ac on Lsp1α in larval fat body nuclei by chromatin immunoprecipitation analysis. Chromatin from hand-dissected male y w larval fat body was immunoprecipitated with antibody against H4K16ac. The X-linked, fat body-expressed Lipid storage droplet-2 (Lsd-2) and Pgd genes showed 3 – 10 fold enrichments after immunoprecipitation compared to the autosomal gene, Glycerol-3-phosphate dehydrogenase (Gpdh), which had a relative enrichment of one (Figure 2). The differential levels of enrichment within Pgd have been observed previously 39. Lsp1α exhibited no enrichment for H4K16ac, confirming the prediction that actively expressed Lsp1α would not be acetylated by MOF as it is not dosage compensated.

Since the majority of MSL target genes are widely expressed 212223, we next investigated if the Lsp1α gene was not enriched for H4K16ac because of the high activity and tissue specificity of its promoter. The Lsp1α gene promoter was fused to the lacZ reporter gene and two X-linked lines containing this gene construct were obtained. Chromatin from male Lsp1α (-573 to +20)-lacZ:9B4 and Lsp1α (-573 to +20)-lacZ:19E7 third instar larval fat bodies was immunoprecipitated with antibody against H4K16ac (Figure 2). As for the y w strain, the X-linked Lsd-2 and Pgd genes were enriched for H4K16ac. The two X-linked Lsp1α (-573 to +20)-lacZ transgenes also showed 3-fold enrichments within lacZ. Thus the lack of enrichment of H4K16ac within Lsp1α is not because of the tissue-specificity of the gene promoter.

There are two possible explanations for why X-linked Lsp1α-lacZ transgenes were enriched for H4K16ac in larval fat body. Either the transgenes have inserted near MSL complex target genes or alternatively MSL complex is directly recruited to the lacZ gene. The latter would seem less likely as the gene is of bacterial origin and MSL complex is not recruited to autosomally integrated lacZ transgenes 37. The Lsp1α (-573 to +20)-lacZ:9B4 transgene has inserted in the first intron of the CG15309 gene. While there are no MSL1 or MSL3 binding sites within CG15309, there are clusters of sites within the closely adjacent l(1)G0230 gene in embryos, SL2 and Clone 8 cells (Additional file 1; Panel C). The Lsp1α (-573 to +20)-lacZ:19E7 transgene has inserted between the CG1529 and Ntf-2 (CG1740) genes. There are clusters of MSL1 and MSL3 binding sites within both the CG1529 and Ntf-2 genes in embryos, SL2 and Clone 8 cells 2123 (Additional file 1; Panel A). While it is not known if CG15309, CG1529 or Ntf-2 gene are actively expressed and targeted by the MSL complex in third instar larval fat body, Ntf-2 is likely to be expressed in this tissue based on mutant phenotypes 40. That MSL complex distribution appears to remain largely stable across development 2122 would suggest that MSL complex would be bound to the CG15309, CG1529 and Ntf-2 genes in larval fat body.

If X-linked Lsp1α-lacZ transgenes are enriched for H4K16ac because they have inserted near MSL target genes then why is Lsp1α not enriched for H4K16ac as the flanking genes are dosage compensated? Since there is a strong correlation between MSL complex binding and gene transcription 21, one possibility is that the flanking genes are not transcribed in third instar larval fat body. Both CG2560 and CG2556 are expressed in third instar larvae (Figure 1C), but their tissue distribution is unknown. In order to determine whether CG2560 and CG2556 are expressed in the fat body, real-time RT-PCR of cDNA from male fat body and whole larvae was conducted on CG2560 and CG2556 relative to the fat body specific genes Lsp1α 27 and Lsd-2 41, and the constitutively expressed genes Gpdh and rp49. Lsp1α and Lsd-2 show 2.24 and 1.82-fold enrichments respectively in fat body cDNA compared to whole third instar larval cDNA, while Gpdh and rp49 show 0.78 and 0.79-fold enrichments respectively (Figure 1D). CG2560 demonstrates 0.03-fold enrichment in fat body cDNA compared to whole third instar larval cDNA, indicating that it is not expressed in fat body tissue. This is consistent with its proposed function as a structural component of the larval cuticle 42 and specific expression in dorsal and ventral epidermis in late embryos 43. CG2556 shows 0.92-fold enrichment in fat body cDNA compared to whole third instar larval cDNA, indicating that it is expressed in both this tissue and other parts of the larvae. Transcripts for the gene immediately 3' of Lsp1α, CG15730, were not detected in third instar larvae hence it is unlikely that the MSL complex is targeted to this gene in fat body tissue.

Since CG2556 is also expressed in fat body we performed ChIP experiments with isolated fat body nuclei and anti-H4K16ac antibody. There was no enrichment with a 5' UTR fragment (fold enrichment of 1.10 ± 0.09 relative to Gpdh in y w male fat bodies). Further, the 3' UTR of CG2556 is only moderately enriched for this histone modification (fold enrichment of 1.93 ± 0.77 relative to Gpdh in y w male fat bodies), suggesting MSL complex is not present in high levels in this region of the chromosome in third instar larval fat body, although the gene is clearly dosage compensated. The moderate enrichment of H4K16ac at the 3' end of CG2556 is consistent with the high-density ChIP-chip profiles that found that the MSL complex binds to intragenic regions, particularly the 3' end of X-linked genes 2123.

It had been suggested that Lsp1α has arisen from a duplication and translocation of an autosomal Lsp1β gene 44. The sequencing of the genomes of twelve Drosophila species 45 allowed us to address when Lsp1α evolved. As expected homologues of Lsp1β and Lsp1γ were found in all Drosophila species examined. Lsp1α homologues, however, are only present in the melanogaster subgroup of species, which are thought to have descended from a common ancestor 8 – 12 million years ago 46 (Figure 3). In these species Lsp1α lies between homologues of CG2560 and CG15730, while in the remainder of the species these genes are immediately adjacent (including D. ananassae, which is part of the melanogaster group but not sub-group). Thus it would appear that Lsp1α arose relatively recently within the melanogaster subgroup of species and so may not have yet evolved MSL binding sites. However, a tree based on maximum likelihood analysis of LSP1 sequences (Methods) suggests that Lsp1α arose before the divergence of Drosophila species (Figure 3). If so, then Lsp1α has been precisely lost on at least four separate occasions (ancestor of: D. ananassae, obscura group, willistoni group and Drosophila subgenus). More likely some residues within LSP1α proteins may not be under the same functional constraints as in LSP1β proteins leading to the observed divergence. The Lsp1β gene seems to be particularly prone to duplication events as duplicated Lsp1β genes were found in D. ananassae, D. grimshawi and D. willistoni genome sequences (Figure 3). The two Lsp1β genes are immediately adjacent to each other in these three species. The maximum likelihood analysis suggests the duplication events have occurred recently within each of the three species.

RNA was extracted using TRIzol reagent (Invitrogen) and RNA secure (Ambion). Poly(A) RNA was isolated using oligo(dT) cellulose (Roche) and Northern hybridization analysis conducted as described in 50.

RNA from 10 male third instar larvae or 10 male third instar larval fat bodies was treated with Turbo DNase (Ambion) and reverse transcribed (Roche). Quantitative real-time PCR was conducted in triplicate (variation <15%) using the LightCycler FastStart DNA Master^PLUS SYBR Green I reaction mix (Roche) in a LightCycler Instrument (Roche) on 2 μl of 10-fold or 100-fold diluted cDNA in a 10 μl reaction volume. Information about the primers used in this study is available upon request. An annealing temperature of 55°C and an extension of 12 s were used. The crossing point (CP) was automatically determined by the LightCycler software (Roche). Fold enrichment was determined by 2^{^}(CP whole larvae – CP fat body). The primer sets used were; Lsp1α (5'CTCGCTGACGGACAAC and 5'GGGCTCAGTAAGGTCCA), Rp49 (5'CGGTTACGGATCGAACA and 5'CGATCTCGCCGCAGTAAA), Lsd-2 (5'AGTGTACTAGCCGATACG and 5'TCTGACTCCCGGATCT), CG2560 (5'ATGGCAATGCTTACGGT and 5'GAGGTGGCTGATAATCGTAG), CG2556 (5'TGGTAATGGCGGCCTAAA and 5'TGCGAGTGTTCAGCTTG), Gpdh (5'GTGCCCGACCTGGTTGAG and 5'CTTGCCTTCAGGTGACGC).

Quantitative RNase protection was conducted on 3 – 4 separate collections of matched female and male FM7I, P{w [+mC]=Act GFP}JMR3/c(1)DX,y¹f¹ first instar larvae or y w blue food-staged third instar larvae 51 using the RPA III kit (Ambion). Antisense RNA probes for CG2560, Pgd, rp49, Lsp1α and CG2556 were synthesized with T7, T3 or SP6 RNA polymerase (Roche). The relative radioactivity of the probes was adjusted by increasing the concentration of [α-³²P]CTP and decreasing the concentration of unlabeled CTP in the reaction cocktails. Unincorporated radionucleotide was removed with the NucAway spin column (Ambion). The CG2556 probe was gel purified (Qiagen). 3 – 10 fold molar excess of probe was added to 4 μg (CG2560, Pgd, rp49 and Lsp1α) or 20 μg (CG2556) of DNase-treated phenol/chloroform purified RNA and annealed overnight with RNA in the presence of pellet paint co-precipitant (Novagen); protected RNA probes were detected and quantified on 5% polyacrylamide urea gels with the Storm 860 phosphorimager (Molecular Dynamics) or the FLA-5000 phosphorimager (Fujifilm). The quantification value of each protected RNA species was corrected for the background value of the sample of yeast RNA hybridized to probe treated with RNase. The mass of RNA used in each assay was determined to be within the linear range of the RNase protection assay for each protected RNA. The sizes of the protected RNAs were 366 nt for CG2560, 391 nt + 294 nt for Lsp1α, 268 nt for CG2556, 171 nt + 43 nt for Pgd and 312 nt for rp49.

All recombinant DNA manipulations were carried out using standard procedures 50 unless otherwise specified. The 883 bp region between Lsp1α and CG2560 including the Lsp1α promoter (I) and 4596 bp region between the 3' end of Lsp1α and the 5' end of CG2556 (I2) were amplified by PCR from genomic y w D. melanogaster DNA, and cloned into pGEM-T Easy (Promega). The primer sequences used are available upon request. PstI-NotI and EcoRI-StuI linkers were inserted into the PstI site 3' of lacZ-SV40 and the EcoRI site 5' of armadillo in pCaSpeR-arm-βgal (arm-lacZ) 47. The blunt-ended NotI (I) and NotI (I2) fragments were cloned into the StuI and NotI sites of this plasmid respectively, generating I-arm-lacZ-I2. The 593 bp Lsp1α promoter (-573 to +20) was amplified by PCR from genomic y w D. melanogaster DNA, and cloned into pGEM-T Easy (Promega). The blunt-ended NotI promoter fragment was cloned into the StuI site of pCaSpeR-arm-βgal in which the EcoRI/Asp718 armadillo promoter fragment had been replaced with a linker containing EcoRI,BamHI, NheI, StuI and Asp718 sites, generating the Lsp1α (-573 to +20)-lacZ construct. Transgenic flies carrying these constructs were generated from y w stock using standard procedures 52. The site of transgene integration was determined where possible using inverse PCR, and all transgenic lines consist of single insertions as determined by Southern hybridisation analysis.

β-galactosidase assays were performed on hemisected adults as described in 36. Assays were performed in triplicate on 3 separate collections unless otherwise stated. Means and standard deviations of activities and ratios were calculated from the 3 separate collections.

Chromatin immunoprecipitation was based on the procedures described in 535455 with additional modifications suggested by Edwin Smith (Emory University, 2005, personal communications). Fat bodies manually dissected from 200 male third instar larvae were quick frozen then ground in liquid nitrogen, and homogenized in 5 ml of 10 mM Hepes [pH 7.6], 1 mM EDTA, 150 mM NaCl, 0.6% Triton X-100, 4 mM DTT, 10 mM sodium butyrate with protease inhibitors (Roche). After centrifugation at 500 × g for 30 s at 4°C, the supernatant was stored on ice for 5 min, followed by centrifugation at 1500 × g for 10 min at 4°C. Pelleted nuclei were resuspended in 360 μl of nucleus isolation buffer (NIB: 0.25 M sucrose, 10 mM Tris-HCl [pH 7.4], 3 mM CaCl₂, 10 mM sodium butyrate, protease inhibitors) and incubated with formaldehyde at a final concentration of 1% for 10 min at room temperature with shaking. Nuclei were pelleted at 1500 × g for 10 min at 4°C, resuspended in 360 μl of NIB, and pelleted. Nuclei were resuspended in 200 μl of 1% SDS, 50 mM Tris-HCl [pH 8.1], 10 mM EDTA, 10 mM sodium butyrate with protease inhibitors for 10 min on ice, followed by sonication in the presence of 425–600 mm acid-washed glass beads (Sigma) for 6 × 30 s pulses at power level 1.5 (VirSonic). This sonication produced DNA fragments with a mean size of 500 bp. The sonicated chromatin lysate was diluted with 1.8 ml of chromatin immunoprecipitation buffer (CIB: 25 mM Tris-HCl [pH 8.0], 137 mM NaCl, 2.7 mM KCl, 1% Triton X-100, 1 mM EDTA, 10 mM sodium butyrate) and centrifuged at 14,000 × g for 10 min at 4°C. Input DNA was purified from 200 μl of this supernatant by incubation with 10 μl 10 mg/ml RNase for 10 min at 37°C, and 20 μl 10 mg/ml proteinase K for 6 h at 37°C. Crosslinks were reversed by incubation for 6 h at 65°C. DNA was purified using the QIAquick PCR purification kit (Qiagen).

The remaining 1.8 ml of sonicated chromatin lysate was pre-incubated with 60 μl 50% Protein A Sepharose Bead suspension (Sigma), protease inhibitors and 9 μl 10 mg/ml salmon sperm DNA (ssDNA) for 1 h with shaking at 4°C. 6 μl of rabbit anti histone H4 (Ac16) antibody (AHP417; Serotec) was pre-bound to 120 μl 50% Protein A Sepharose Beads suspension in 1 ml of CIB with protease inhibitors, 10 μl 10 mg/ml ssDNA and 20 μl 10 mg/ml BSA (NEB) for 1 h with shaking at 4°C. The pre-cleared chromatin lysate was incubated with the pre-bound antibody-beads for 3 h at 4°C with shaking. Following this the beads were washed successively in: 150 mM NaCl, 20 mM Tris-HCl [pH 8.0], 2 mM EDTA, 1% Triton X-100, 0.1% SDS, 10 mM sodium butyrate; 500 mM NaCl, 20 mM Tris-HCl [pH 8.0], 2 mM EDTA, 1% Triton X-100, 0.1% SDS, 10 mM sodium butyrate; 250 mM LiCl, 10 mM Tris-HCl [pH 8.0], 1 mM EDTA, 1% sodium deoxycholate, 1% IGEPAL CA-630 (Sigma), 10 mM sodium butyrate; TE buffer. Beads were rinsed and suspended in TE Buffer. Immunoprecipitated DNA (ChIP DNA) was incubated with 1 μl 10 mg/ml RNase for 10 min at 37°C, followed by incubation with 5 μl 10% SDS and 2 μl 10 mg/ml proteinase K for 6 h at 37°C. Crosslink reversal and DNA purification were conducted as described for input DNA.

Quantitative real-time PCR was conducted as described for real-time RT-PCR assays. Undiluted immunoprecipitated DNA (2 μl) and 100-fold diluted input DNA (2 μl) were assayed with each primer set in triplicate (variation <15%). The primer pairs used were; Lsp1α-1184 (5'CTCGCTGACGGACAAC and 5'GGGCTCAGTAAGGTCCA), Lsp1α-2439 (5'CTTCAAGTTGTAGATCTGATAACATTCCGA and 5'GCTTAACAGAGAAATCTGAATACGTTGG), Pgd-543 (5'GGATAAGCAGGTGGAAGTAGGAAG and 5'ACACTTGTGGTTACGGTTTTCG), Pgd-2303 (5'GAAGGGCACGGGCAAGTG and 5'CAATGCCGCCGTAATTAAGTCTC), Lsd-2 (5'GAAACACACGCACACGG and 5'TCCCAGCGAGCGTACAA), Gpdh (5'GTGCCCGACCTGGTTGAG and 5'CTTGCCTTCAGGTGACGC), lacZ (5'GCGCGAATTGAATTATGGCCC and 5'GCCATGTGCCTTCTTCCG). The Pgd and Gpdh primer sets are identical to those used in a previous study 39. The identity of the PCR products amplified by each primer combination was confirmed by sequencing. Fold enrichment was determined by 2°CP input X-linked gene – CP ChIP X-linked gene)/2°CP Gpdh – CP ChIP Gpdh).

The Drosophila melanogaster LSP1α (GenBank accession no. NP_511138), LSP1β (GenBank accession no. NP_476624) and LSP1γ (GenBank accession no. NP_523868) protein sequences were used to search the nucleotide sequences available in "DroSpeGe: Drosophila Species Genomes" 45 using tBlastn . The scaffolds on which the LSP1 homologues were identified are described in additional files 2 and 3. LSP1β and LSP1γ had independently been identified in D. buzzatii and D. pseudoobscura 44, as had LSP1γ from D. simulans (GenBank accession no. AAB71667). An alignment of the dataset was performed in ClustalX 56, ambiguous bases and gaps were removed from the alignment using PAUP*4.10b 57. Using ProtTest 58, the optimal model of sequence evolution was determined to be WaG 59. A maximum likelihood analysis was performed on the data in Phyml using the WaG model with 100 bootstrap replicates 60. The D. simulans LSP1α sequence is incomplete due to a gap in the genomic sequence.