An evolutionary consequence of dosage compensation on Drosophila melanogaster female X-chromatin structure?
© Zhang and Oliver. 2010
Received: 15 April 2009
Accepted: 5 January 2010
Published: 5 January 2010
Skip to main content
© Zhang and Oliver. 2010
Received: 15 April 2009
Accepted: 5 January 2010
Published: 5 January 2010
X chromosomes are subject to dosage compensation in Drosophila males. Dosage compensation requires cis sequence features of the X chromosome that are present in both sexes by definition and trans acting factors that target chromatin modifying machinery to the X specifically in males. The evolution of this system could result in neutral X chromatin changes that will be apparent in females.
We find that the general chromatin structure of female X chromosomes is distinct from autosomes. Additionally, specific histone marks associated with dosage compensation and active chromatin marks on the male X chromosome are also enriched on the X chromosomes of females, albeit to a lesser degree.
Our data indicate that X chromatin structure is fundamentally different from autosome structure in both sexes. We suggest that the differences between the X chromosomes and autosomes in females are a consequence of mechanisms that have evolved to ensure sufficient X chromosome expression in the soma of males.
Drosophila X chromosomes show peculiar features in both gene expression  and gene evolution . One of the most striking consequences of X chromosome hemizygosity in males, is dosage compensation, a process which brings X chromosome and autosome expression into balance [3–5]. Dosage compensation was probably acquired gradually in the course of sex chromosome evolution, as sex chromosomes are thought to arise by divergence of an ancestral autosome pair . Gene loss from the Y chromosome creates an increasingly aneuploid condition in males and is thought to be the driving force in the evolution of global X-chromosome dosage compensation. In the absence of dosage compensation genomic imbalance results in male lethality.
It has long been known that selective pressures applied to just one of the sexes can effect change in the other . For example, the coloration of certain birds or the nipples of mammals are advantageous to one of the sexes and are likely to be present as an evolutionary side-effect in the other. X chromosome dosage compensation might also show evidence of this type of sexual selection. X chromosome dosage compensation requires both cis and trans components . Cis changes resulting from selection of the compensation system in males will also be present in females, and might alter the character of the X chromosome in females as a secondary consequence . Indeed, we have previously noted a slight over-expression of both male and female X chromosomes relative to autosomes [4, 9], which suggests that the X chromosome is inherently more active than autosomes. We have therefore examined the structure of X chromatin in females in detail.
Expression patterns and especially X chromosome dosage compensation are mediated by chromatin modification [1, 10, 11]. Histones are nucleosome subunits required for packing DNA into the confines of the nucleus. It has long been know that chromatin structure changes are associated with transcription . For example, when chromatin is physically sheared to small fragments by sonication or enzyme digestion, shearing-bias is associated with different chromatin structures across the genome [13–15]. Histones are also modified on N-terminal tail residues to generate an expanding repertoire of histone modifications that are important modulators of transcription [16, 17]. It has become increasingly clear that specific types of modification are associated with particular transcriptional outcomes. For example, acetylation events are broadly associated with transcriptional activation, while methylation events can have either activating or repressing roles.
One of the best studied histone modifications is the acetylation of Histone 4 on Lysine 16 (H4K16ac). In organisms from yeast to humans, H4K16ac is broadly associated with active genes, and the Histone Acetyl Transferase (HAT) that writes the modification is required for viability [18–21]. In Drosophila, H4K16ac is highly enriched on the X chromosomes of males [22, 23], and the responsible HAT, Males Absent on First (Mof), is required for male viability . While Mof is associated with some genes in both males and females , Mof is greatly enriched on the male X chromosome due to targeting by the male-specific-lethal (MSL) complex. MSL is composed of proteins (Mle, Msl1, Msl2, Msl3, and Mof) and two non-coding RNAs encoded on the X (RoX1 and RoX2) . It is thought that the greatly increased H4K16ac levels act to increase X chromosome expression in males, although it is also possible that X chromosome enrichment depletes autosomes of H4K16ac [1, 25]. In either model, X chromosome and autosome expression are equilibrated to restore transcription balance.
Another chromatin modifying enzyme, Jil1, is also enriched on the X chromosome of males [26–28]. This kinase mediates phosphorylation of Histone 3 at serine 10 (H3S10ph). Jil1 is required for full dosage compensation and associates with the MSL complex [27, 29]. H3S10ph is implicated in both chromosome condensation during mitosis and transcriptional activation during interphase, suggesting that Jil1 has more general roles in addition to dosage compensation. Another mark associated with active transcription, dimethylation of histone H3 at lysine 4 (H3K4me2)  is general, and thus likely to be MSL complex independent.
We have performed chromatin-shearing experiments showing that X chromatin differs from autosomal chromatin in both males and females. Additionally, the histone marks associated with X chromosome dosage compensation in males are modestly enriched on female X chromosomes. These data indicate that X chromatin is distinct even in the absence of dosage compensation. We suggest that the pattern in females is a tolerated neutral side-effect of the evolution of X chromosome dosage compensation in males.
We took advantage of differential shearing to probe chromatin structure by deep DNA sequencing (DNA-Seq) . Specifically, we sheared cross-linked chromatin, size selected for short (200 bp) fragments, performed deep sequencing, and aligned the DNA reads to the reference genome. Because these sequencing reactions generated reads from the ends of size-selected fragments, increased mapped read density occurs in regions of preferential shearing.
We then compared the coverage profiles of X chromosome and autosome gene model features inter se to determine if the structure of the X chromosome differed from the autosomes. Indeed, the average coverage for X chromosome genes was clearly higher than autosome arm genes at exons, transcription start sites, and transcription termination sites in both males and females (Figure 2A-H). The 4th chromosome always showed greatly reduced sequence coverage. Permutation testing clearly suggests that the X chromosome shearing at exons, transcription start sites, and termination sites is significantly greater than the pattern genome-wide in both males (not shown) and females (Figure 2I-L). These data indicate that X chromosome genes are susceptible to shearing, probably as result of a more open structure.
As expected, we found a strong enrichment for H4K16ac on the male X chromosome relative to male autosomes (Figure 3B, E). The average H4K16ac ChIP ratio (ChIP enriched DNA/input) on male X chromosome genes was 1.39 fold higher than the value of all autosome gene. These data show that we are able to easily score the high levels of H4K16ac present on the male X chromosomes in adult flies. Interestingly, we observed a modest (1.06 fold) but significant enrichment for H4K16ac on female X chromosomes relative to all the autosomes (Figure 3B, H). In both males and females, the ratio of H4K16ac ChIP/input was similar for all major autosome arms (2L, 2R, 3L, 3R) indicating that there are no differences in H4K16ac among the major autosome arms within either sex. These data indicate that not only is X chromatin different from autosomal chromatin in females, but that the important dosage compensation mark H4K16ac is generally enriched on X chromosomes.
We observed a modest, but significant 1.09 fold enrichment for the H3S10ph mark on the male X chromosome relative to the autosomes (Figure 3C, F). We also observed a 1.02 fold enrichment of H3S10ph on the female X chromosome, but this difference was only modestly significant (Figure 3C, I). As observed for the other two histone marks, we found a significant enrichment for H3K4me2 on the X chromosome in males (Figure 3D, G). But again, we also found significant enrichment for H3K4me2 on the female X chromosome (Figure 3D, J). Thus, the patterns of histone mark accumulation on the male X chromosome appear to be shadowed by similar distributions on the female X chromosome.
Dosage compensation may be not being required for genes that should be differentially regulated in the female and male soma. In support of this idea, we also observed lower H4K16ac among X chromosome genes showing male-biased expression in the soma (average ChIP/input ratio = 0.178 for genes with male-biased expression vs. 0.351 for genes with non-biased expression, P < 10-2, Bonferroni corrected KS test). These data are also consistent with the idea that genes with male-biased expression are poorly expressed on the X chromosome due to limited dosage compensation . Interestingly, the same pattern is observed for genes with female-biased expression. We observed lower H4K16ac among genes showing female-biased expression in females (average ChIP/input ratio = -0.053 for genes with female-biased expression vs. 0.060 for genes with non-biased expression, P < 10-2, Bonferroni corrected KS test). These data suggest that sex differentially regulated genes are not H4K16ac regulated.
Our data indicate that the dosage compensation marks associated with X chromosomes in males are also enriched on X chromosomes more generally. This is part of a growing body of data showing that the X chromosome is distinct from the autosomes. For example, previous studies show that Jil1 mutants cause accumulation of the H4K9me2 mark and the transcriptional negative regulator HP1 on the X chromosome of both males and females . Additionally, 9 of 29 chromatin associated proteins surveyed in female Drosophila Kc167 cells are enriched on the X chromosome . Finally, both Mof and H4K16ac enrichment has been observed on the X chromosomes of female Kc167 cells [24, 37], although female results served as controls in those manuscripts and the enrichment in females was not highlighted. Collectively, these data indicate that X chromatin differences from autosomes are not restricted to male-specific dosage compensation. We have previously observed a modest elevation of gene expression from the female X chromosome relative to autosomes in D. melanogaster  and this same modest (1.11 to 1.24 fold) but highly significant elevation of female X chromosome expression is consistently observed in six other Drosophila species , suggesting that this modest elevation of expression on X chromosome is independent of microarray platform and species. We suggest that specialized X chromatin contributes to elevated X chromosome gene expression in females in addition to males, albeit to a much more limited extent.
While modest over expression of X chromosome genes in females may be of little evolutionary consequence, one can envision that more extensive over expression would be detrimental. For example, large duplications are poorly tolerated in Drosophila and  trisomy is poorly tolerated in humans . If the X chromosome tends to be in a favourable state for transcription in both males and females, this could ultimately lead to counter measures in females to prevent functional tetrasomy. This might be what has occurred in mammals, where one X chromosome is inactivated in females. There is recent evidence that the active X chromosomes of mammals show increased expression [4, 42, 43] raising the possibility that X inactivation is a consequence of X chromosome dosage compensation in males. The main dosage compensation counter measure in Drosophila females is inhibition of MSL complex formation by Sxl protein . However, if the observed over expression of X chromosomes in Drosophila females  was more extreme, then inactivating an X chromosome in females to counter X chromosome dosage compensation in males would be logical. This may be how such a mechanism evolved in mammals.
In this study we examined the general chromatin structure and multiple histone marks in both Drosophila males and females. We found that X chromatin structure is different from autosome structure independent of sex. Histone marks associated with dosage compensation or active transcription that are highly enriched on the male X chromosome are also slightly enriched on the female X chromosome. These findings suggest that the distinct female X chromatin structure is an evolutionary consequence of dosage compensation in males.
Flies were grown at 25°C on standard cornmeal media. Wild type flies (y 1 w 67 c ) for chromatin immunoprecipitation were aged for 5-7 days post-eclosion, sexed and flash frozen. About 0.5 g of sex-sorted adult flies was cross-linked with 1.8% formaldehyde in cross-linking solution (50 mM pH 8.0 HEPES buffer, 1 mM EDTA, 0.5 mM EGTA, and 100 mM NaCl) for 20 minutes at room temperature on a shaker. We confirmed that this procedure cross-linked the DNA by extracting DNA from treated and untreated flies without reversing the cross-linking and performing gel migration retardation experiments. The cross-linked flies were incubated in PBS supplemented with 125 mM Glycine and 0.01% Triton for 5 minutes and then washed twice with PBS and 0.01% Triton. The flies were homogenized and the cells were collected from the pellet after centrifuge at 7500 rpm for 5 minutes at 4°C. The collected cells from adult flies were disrupted in cell lysis buffer (5 mM pH8.0 PIPES buffer, 85 mM Potassium chloride, 0.5% Nonidet P40 and protease inhibitors) for 10 minutes and then resuspended in nuclei lysis buffer (50 mM pH8.1 Tris HCl, 10 mM EDTA, 1% SDS and protease inhibitors) for 20 minutes at 4°C. The nuclear extract was sheared to 200-1000 bp by sonication for 8 minutes on ice (pulsed 8 times for 30 seconds with 30 second intervals) using a Misonix Sonicator 3000 (Misonix, Inc. Farmingdale, NY). After sonication, cell debris was removed by centrifugation and 500 ul of chromatin solution was used for one immunoprecipitation reaction. 5 ul anti-H4K16ac (Upstate, 07-329), 5 ul anti-H3S10ph (Upstate, 05-817) or 3 ul anti-H4K3me2 (Upstate, 07-030) was incubated with the chromatin for 2 hours and then was bound to protein A agarose beads at 4°C overnight. The beads were washed three times with 0.1% SDS, 1% Trition, 2 mM EDTA, 20 mM ph8.0 Tris, 150 mM NaCl; three times with 0.1% SDS, 1% Trition, 2 mM EDTA, 20 mM ph8.0 Tris, 500 mM NaCl; and twice with 10 mM pH8.1 Tris, 1 mM EDTA, 0.25 M LiCl, 1% NP40, 1% sodium deoxycholate. The immunoprecipitated DNA was eluted from the beads in 0.1 M NaHCO3 and 1% SDS. Formaldehyde cross-links were reversed by incubation at 65°C overnight. DNA was purified by phenol-chloroform extraction and ethanol precipitation.
DNA amplification was performed using a Ligation-mediated PCR (LM-PCR) protocol  from FlyChip http://www.flychip.org.uk/protocols/chip/lm_pcr.php. 600 ng of amplified DNA (ChIP enriched DNA or input DNA) were then labeled using the Cy3- or Cy5-labeled random nonamers (Trilink Biosciences, San Diego, USA) with Klenow enzymes. The labeled DNA was purified and resuspended in hybridization buffer as described . For hybridizations we used FlyGEM arrays , that are spotted arrays with PCR amplicons biased to the 3' ends of annotated genes, which is a suitable platform given that H4K16ac is found to be present along the whole length of target genes with a 3' bias  and given our results on shearing patterns at gene models. Labeling with Cy3 and Cy5 random primers was as described [4, 46]. There is no dye effect using this end-labeling method . Preliminary ChIP-chip experiments on S2 cells showed no dye effects (not shown). All labeling, hybridization and scanning was done under low ozone conditions. We used Cy5 for ChIP and Cy3 for input in all experiments. Four biological replicates (ChIP vs. input) were performed for H4K16ac in males and three biological replicates were performed for H4K16ac in females. Two biological replicates were performed for H3S10ph and H3K4me2 experiments. Arrays were scanned on an Axon GenePix 4000B (Molecular Devices Corporation, Sunnyvale, CA) and signal for each array elements were extracted with GenePix v.5.1 image acquisition software (Molecular Devices Corporation).
300 ng of DNA derived from sheared chromatin of female and male adult flies was prepared as outlined for ChIP input controls above. This was used for solexa library preparation using the genomic DNA sample preparation kit (Illumina, San Diego, CA). We size selected libraries (~200 bp) by excision of the appropriate region following agarose gel electrophoresis. We determined library concentration on a Nanodrop spectrophotometer (NanoDrop products, Wilmington, DE) and hybridized 4 pM of adaptor-ligated DNA to the flow cell. DNA clusters were generated using the Illumina cluster station, followed by 36 cycles of sequencing on the Illumina Genome Analyzer. Image analysis and base calling were performed using a manufacturer-provided computational pipeline (version 0.3) including the Firecrest and Bustard applications and sequence reads were then aligned with the Drosophila melanogaster assembly (BDGP Release 5, dm3 ) using Bowtie (v 0.10.0) . We used only uniquely mapped reads with no more than two mismatches.
We calculated average read coverage depth at each base per Kb per million sequenced base pairs to normalize between samples with slightly different total sequence depth. The average coverage in non-overlapping 1 kb windows were calculated on each chromosome arm for the distribution of sequence coverage in the genome. We also calculated the average coverage for each annotated  exon, intron and +/- 3000 bp around the transcription start sites and transcription termination sites in the genome. To test whether the sequenced reads coverage over non-overlapping 1 kb windows or different gene features on individual chromosome arms were significantly different from the whole genome, we performed two sample Kolmogorov-Smirnov tests (KS tests) between random sampled coverage values from non-overlapping 1 kb windows or gene features on individual chromosome arms and the same number of random sampled coverage values from the whole genome. The coverage for gene features on male X chromosome was multiplied by 2 to correct for the different number of X chromosome in male. The sample size for random sampling was 1000 for all chromosome arms except for the 4th chromosome (sample size = 200) and the permutation procedure was repeated 100 times. The distribution of P values from the 100 permutation KS tests were used to determine whether the genes features on one particular chromosome arms were significant different from the gene features in the genome. For visualization purpose, we calculated the average coverage along different gene features in 100 overlapping windows with widow size as 1/10 of the whole region for individual genes and then plotted the moving averaged coverage along different gene features for genes on the X chromosome and different autosome arms separately.
All microarray data were processed and analyzed in R/Bioconductor  package limma. For the ChIP-chip studies, the log2 ratio between two channels for individual arrays were median-centered and then all the arrays were normalized using quantile normalization based on the input channel across arrays while assuring that the input channel has the same empirical distribution among arrays. Background corrections were then applied to exclude array elements with intensities less than the average intensities of control elements (designed against non- Drosophila DNA) in both channels for downstream analysis. The average intensities of the control elements were also subtracted from the intensities of every array element on two channels separately. Duplicated array elements were merged by calculating the average ratio from duplicated probes. The consistency between biological replicates was checked by the density scatter plot of the input and ChIP DNA intensities among all replicates (see Additional file 1, 2 for H4K16ac ChIP replicates in male adult files, Additional file 3, 4 for H4K16ac ChIP replicates in female adult files and Additional file 5 for other histone marks). The raw and normalized data were inspected to make sure that normalization was appropriate (see Additional file 6).
The comparisons of the distribution of ChIP/input ratios or ChIP input intensities among different chromosome arms were performed using two sample Kolmogorov-Smirnov tests (KS tests) and the P values were adjusted using the Bonferroni correction for multiple-comparison. The maximum adjusted P value was truncated at 1.0. KS tests and the Bonferroni corrections were applied to ChIP/input ratios from individual replicates and averaged ratios to test the significance of X chromosome in different replicates. Unless otherwise noted, all ChIP/input ratios used in the figures were averaged across all the biological replicates. Spearman's Rank Correlation was used to test the relationship between H4K16ac level and all other histone modifications in males and females. To test the correlation among different histone marks, we also used the ChIP/input ratio ranks in each ChIP sample for K-means clustering analysis with 2 nodes using the Euclidean similarity metric. X chromosome genes and autosome genes were clustered separately and the order of ChIP samples was fixed as indicated in the figure in both clustering analysis. Data were clustered and visualized using Cluster3.0/Tree-View. The expression data for sex-sorted whole adults and for gonadectomized male and female carcasses were from GEO accession number GSE6640 and GSE442. The expression values used for the correlation between marks and expression were averaged expression values from all replicates in each dataset. For the correlation between expression and different histone marks, genes with ChIP/input ratio percentile rank >90% were used as genes with enriched marks and genes with ChIP/input ratio percentile rank <10% were used as genes with depleted marks to assure similar sample size for different histone marks. The list of genes with male-, female- or non-biased expression in adult flies were got from a previous study  and the list of genes with sex-biased expression in gonadectomized carcasses were produced from the data set GSE442 using the same analysis approach .
All DNA-Seq and ChIP-chip data and full platform description are available at GEO under accession number GSE15593.
We thank the NIDDK genomics core laboratory for solexa sequencing. We also thank Shirley Liu, David Sturgill, and John Malone for discussions and preliminary analysis of DNA-Seq data. The authors are grateful to members of Oliver lab, Brian Charlesworth and the dosage compensation community for many helpful comments. This research was supported by the Intramural Research Program of the NIH, NIDDK.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.