The chromosomal high-affinity binding sites for the Drosophila dosage compensation complex

Abstract

Dosage compensation in male Drosophila relies on the X chromosome-specific recruitment of a chromatin-modifying machinery, the dosage compensation complex (DCC). The principles that assure selective targeting of the DCC are unknown. According to a prevalent model, X chromosome targeting is initiated by recruitment of the DCC core components, MSL1 and MSL2, to a limited number of so-called "high-affinity sites" (HAS). Only very few such sites are known at the DNA sequence level, which has precluded the definition of DCC targeting principles. Combining RNA interference against DCC subunits, limited crosslinking, and chromatin immunoprecipitation coupled to probing high-resolution DNA microarrays, we identified a set of 131 HAS for MSL1 and MSL2 and confirmed their properties by various means. The HAS sites are distributed all over the X chromosome and are functionally important, since the extent of dosage compensation of a given gene and its proximity to a HAS are positively correlated. The sites are mainly located on non-coding parts of genes and predominantly map to regions that are devoid of nucleosomes. In contrast, the bulk of DCC binding is in coding regions and is marked by histone H3K36 methylation. Within the HAS, repetitive DNA sequences mainly based on GA and CA dinucleotides are enriched. Interestingly, DCC subcomplexes bind a small number of autosomal locations with similar features.

Figure 1. Confirmation of RNAi target depletion by western blotting.

Effect of (A) MOF RNAi on MOF and (B) MLE and MSL3 RNAi on corresponding protein levels. Relative amounts of target protein after interference normalized to α-tubulin are indicated. C) Effect of RNAi on MSL1 protein levels. Relative amounts of MSL1 are indicated.

Figure 2. Differential crosslinking alters the binding patterns and improves mapping of high-affinity sites.

A) Genome browser snapshot with gene spans and gene models. MSL1 and MSL2 profiles after low or high formaldehyde (FA) crosslinking are depicted as the log2 of the mean enrichment ratio (IP/Input) of at least 2 replicate experiments. The Tao-1 gene contains a confirmed HAS (DBF 9B), which is indicated by the green box below the profiles. B) Absolute changes in numbers of probes significantly bound by MSL2 after differential crosslinking. C) Corresponding relative changes according to functional context. D) Changes in MSL2 signals on MSL2 target probes grouped according to genomic context when crosslinking under low FA conditions compared to high FA crosslinking.

Figure 3. Changes in MSL1 binding after MOF RNAi.

A) Genome browser snapshot of a representative region with MSL1 profiles after control (top) and MOF (bottom) RNAi. B) Changes of the relative distribution of significant MSL1 binding. C) Changes in MSL1 signal after MOF RNAi as compared to the signals after GST RNAi. Probes are grouped according to their functional annotation.

Figure 4. The new set of 131 X-chromosomal sites contain all established high-affinity sites of the Drosophila DCC.

A) rox2 and B) the site at 18D displayed with MSL1 wild type and residual profiles after RNAi and/or low formaldehyde crosslinking. Histograms of the length distribution (C) and distance distribution (D) between high-affinity sites (histograms were calculated on log-transformed values).

Figure 5. High-affinity sites in SL2 cells frequently map to MSL2 signals in Sxb1-2C females by Immuno-FISH on polytene chromosomes.

Each row corresponds to one locus and is labeled with the name of the closest gene and the cytological position. (+) on the right indicates that FISH probe signal and MSL2 colocalized robustly; in the case of (−) no colocalization could be observed. No colocalization is expected with the control FISH probes Or2a and dpr8.

Figure 6. Common features of X-chromosomal high-affinity sites.

A) A high affinity site (HAS, red box) within the Suv4-20 gene. Displayed are wild type profiles for MSL1, MSL2 and MOF. Regions that are significantly depleted of histone H3 are indicated by blue boxes. B) Location of 51 high-affinity sites with respect to functional genomic context. “Proximal” indicates location within 3 kb up- or downstream of annotated genes. C) Cumulative H3 profile of 100 high-affinity sites. The red line indicates the mean signals along a region from −2 kb to +2 kb from the center of each high affinity site. The green cloud resembles the kernel density plot of all signals contributing to the mean. A darker color indicates higher density. D) Top motif found enriched in high-affinity sites and 1.5-fold over-represented on the X chromosome.

Figure 7. Autosomal binding sites of the DCC.

A) Autosomal high affinity site. B) Autosomal site bound by MSL1, MSL2 and MOF. C) Venn diagram showing the colocalization of MSL1, MSL2 and MOF on autosomal binding sites. D) Boxplot comparing histone H3 signal on probes located on autosomal binding sites (AS) and probes located elsewhere on autosomes. E) Autocorrelation (ACF) of MSL1 binding to the X chromosome and (F) to chromosome 3R. Correlation was performed on smoothed profiles with 500 base spacing.

Figure 8. Compensation of X-linked genes is dependent on distance from high-affinity sites (HAS).

Based on microarray expression profiling studies the relation of mean fold changes (log2) in gene expression after RNAi and the genes' distance from the closest HAS is depicted. A distance of 0 indicates that the HAS is directly within the gene. Effects of MSL2 (A), MOF (B) and MSL3 (C) knockdown are displayed as boxplots on gene groups of varying distance. For clarity, extreme outliers have been omitted from the panels. Genes located more than 3 kb away from the next HAS respond significantly less to MSL2 RNAi (p-value 0.00087; two-sided t-test). On the contrary, upon MOF and MSL3 RNAi these genes are stronger affected (p-value 3.115e-05 and 5.804e-08, respectively).