Developmental control of gene copy number by repression of replication initiation and fork progression

Abstract

Precise DNA replication is crucial for genome maintenance, yet this process has been inherently difficult to study on a genome-wide level in untransformed differentiated metazoan cells. To determine how metazoan DNA replication can be repressed, we examined regions selectively under-replicated in Drosophila polytene salivary glands, and found they are transcriptionally silent and enriched for the repressive H3K27me3 mark. In the first genome-wide analysis of binding of the origin recognition complex (ORC) in a differentiated metazoan tissue, we find that ORC binding is dramatically reduced within these large domains, suggesting reduced initiation as one mechanism leading to under-replication. Inhibition of replication fork progression by the chromatin protein SUUR is an additional repression mechanism to reduce copy number. Although repressive histone marks are removed when SUUR is mutated and copy number restored, neither transcription nor ORC binding is reinstated. Tethering of the SUUR protein to a specific site is insufficient to block replication, however. These results establish that developmental control of DNA replication, at both the initiation and elongation stages, is a mechanism to change gene copy number during differentiation.

Figure 1.

Euchromatic regions are repressed for replication early in development. (A) aCGH of the proximal euchromatic half of chromosome 2L of Drosophila Oregon-R (OrR) third-instar salivary glands indicates the presence of euchromatic under-replicated regions. The genome profile shows the log₂ ratio of copy number in the third-instar salivary gland genome relative to that of diploid embryonic DNA. Under-replicated loci are indicated by vertical light green bars. OrR flies are the wild-type strain used throughout this study. For this and other figures, UCSC Genome Browser (build dm3) was used (http://genome.ucsc.edu) (Rhead et al. 2010; Fujita et al. 2011). (B) Under-replication begins with the first endocycle in the embryonic salivary gland. Levels of DNA at four under-replicated loci relative to a fully replicated control locus were measured by qPCR in embryonic and first-, second-, and third-instar salivary glands. The diploid whole-embryo sample shows a 1:1 ratio between any two DNA loci. Error bars, SD of three biological replicates. By a one-sample t-test, the embryonic salivary gland sample is significantly different from the diploid embryo sample, P = 0.012. (C) Model of replication fork structure at under-replicated region.

Figure 2.

Under-replicated regions are restrictive for ORC2 and RNA Pol II binding and are marked by weak binding of H3K27me3. (A) aCGH of proximal half of chromosome 2L, as described in Figure 1A. (B–E) RNA-seq, RNA Pol II ChIP-chip, H3K27me3 ChIP-chip, and ORC2 ChIP-seq peaks in Oregon-R (OrR) third-instar salivary glands, respectively. Total Illumina RNA sequencing reads are plotted; RNA Pol II ChIP was normalized to IgG control ChIP, and H3K27me3 ChIP data to input DNA. ORC2 ChIP-seq peaks were called by SPP compared with input sequencing lane; peaks for these data are organized downward by the UCSC Genome Browser.

Figure 3.

orc2 mutants have reduced ploidy levels and show increased copy number of most under-replicated regions, as well as disorganized replication. (A) Ploidy levels of orc2 mutants and heterozygous sibling controls were quantified using DAPI microdensitometry. The intensity of DAPI staining was measured relative to diploid cells to calculate their ploidy. Each symbol represents the DAPI signal intensity from a single nucleus in two experimental replicates, and the bar is the mean of the measured nuclear values. (B) aCGH was performed comparing DNA of orc2^k43/Df(3R)Exel6288 and orc2^k43/Df(3R)Exel6171 third-instar salivary glands with that of diploid embryo. Oregon-R (OrR) aCGH is shown for comparison. The baseline of each aCGH is based on the average of all points on the array, so that the baseline of the orc mutant aCGHs is fourfold below that of the OrR aCGH shown. Plotted is the mean probe intensity. The asterisk marks examples of disordered but experimentally consistent overreplication, whereas the ^ symbol indicates disorganized but reproducible under-replication.

Figure 4.

Comparison of ORC binding sites in salivary gland cells (SG) to those in Kc, S2, and Bg3 Drosophila cell lines. (A) Origin binding sites were compared between the different cell types (cultured cell lines or salivary gland) indicated on the left. (Red boxes) Origins unique to a single cell type; (orange) origins shared by two cell types; (green) origins shared by three cell types; (purple) origins shared by all four cell types studied. Bars in blue above the boxes show the number of origins shared by the cell types indicated by the color-coded boxes below each bar. Less than 1 kb proximity of ChIP peaks was required for binding site conservation. The number of ORC binding sites differs slightly from that reported by Eaton et al. (2011) because our data merge some ORC binding sites into broader peaks. (B) Rectangular Venn diagram comparing the relationship between ORC binding and TSSs in salivary glands and Kc cell culture, with rectangles drawn to scale. After identifying the TSS nearest to each ORC site, the percentile rank of the corresponding transcript (in RPKM) was determined from the salivary gland and Kc cell RNA-seq data. From the difference in percentile ranks (DPR) between the salivary gland and Kc cell, each transcript was classified as salivary gland specific (SG >> Kc, for DPR > 40, black), higher in salivary gland or no difference (for DPR between 1 and 40, blue), Kc specific or no difference (for DPR ≤ 0, yellow). The percentile of transcripts not expressed in the salivary gland are in white. (C) ORC binding across the under-replicated regions in the salivary gland and the corresponding regions in Kc cells. Each of the 34 domains is displayed vertically, in the same order for ease of comparison. Each domain that is under-replicated in the salivary gland was divided into 100 windows, and each flanking region (the length of half the corresponding under-replicated region) was divided into 50 windows. Blue indicates the presence of an ORC binding site in the window.

Figure 5.

Loss of SuUR function restores genome replication without restoring ORC binding or transcription, but does lead to depletion of the repressive H3K27me3 mark. (A,B) aCGH of proximal half of chromosome 2L in OrR (A) and SuUR mutant (B) third-instar salivary glands, as described in Figure 1A. (C) ChIP-seq peaks of ORC in SuUR mutant third-instar salivary glands, called as described in Figure 2. (D) The ChIP-seq peaks of ORC from wild type for comparison. (E) Number of reads from Illumina RNA sequencing of SuUR mutant third-instar salivary glands. (F) RNA Pol II ChIP of SuUR mutant third-instar salivary glands, normalized to IgG control ChIP. (G) H3K27me3 ChIP-chip in SuUR mutant third-instar salivary glands, normalized to input DNA.

Figure 6.

Quantitative comparisons of ORC binding, H3K27me3 enrichment, and RNA expression levels between Oregon-R (OrR) and SuUR mutant in under-replicated regions and the remainder of the genome. Note that in the SuUR mutant these regions are no longer under-replicated, but we refer to them as UR. (A) Comparison of number of ORC binding sites per 100 kb of genome, as assayed by ChIP-seq, shows that in both OrR and SuUR mutants the under-replicated regions have many fewer ORC binding sites than the remainder of the genome. (B) Boxplots showing quantile-normalized H3K27me3 ChIP-chip data, comparing level of H3K27me3 in under-replicated regions with the rest of the genome, in OrR and SuUR mutant salivary glands. OrR UR was significantly enriched for H3K27me3 over the SuUR mutant UR with a P-value < 1 × 10⁻¹⁵ (unpaired Wilcoxon test). (C) Boxplots showing quantile-normalized RNA sequencing data (figure from one replicate; second is indistinguishable) comparing transcriptional levels in under-replicated regions with the rest of the genome in OrR and SuUR mutant salivary glands. Mutation of SuUR does not restore transcription in UR regions. In both strains, the transcription levels in the under-replicated regions were significantly lower than the remainder of the genome (P < 1 × 10⁻¹⁵, Wilcoxon test). The difference between the under-replicated regions in the two strains was not statistically significant (P = 0.3234, Wilcoxon test).

Figure 7.

SuUR mutant follicle cell amplicons exhibit enhanced replication fork progression. (A) aCGH of Oregon-R (OrR) and SuUR mutant FACS sorted 16C follicle cell nuclei at cytological locations encompassing Drosophila amplicon in follicle cells DAFC-66D and DAFC-7F. (B) SuUR mutant forks travel farther in the same developmental time window at DAFC-66D. SuUR mutant replication fork progression occurs between stage 11 and 13 in egg chamber development and is not due to alterations in egg chamber development timing. Quantitative PCR with primers 20 kb apart along the DAFC-66D amplicon was performed on hand-sorted egg chambers, relative to an unamplified control locus. The distances are from the peak of amplification (0 point). The experiment was performed in triplicate.

Figure 8.

Tethering experiments indicate that tethering of a single SUUR is not sufficient to induce under-replication in salivary glands or to impede fork movement in a follicle cell amplicon. (A) Insertion constructed for tethering experiments, SUUR fused in frame to the DNA binding domain (DBD) of GAL4, with an hsp70 promoter. (B) Complementation assays showing that GAL4DBD-SUUR is capable of functioning as wild-type SUUR protein, restoring under-replication in SuUR mutant larvae. Copy number is relative to a control fully replicated locus. (C) Tethering of GAL4DBD-SUUR at a fully replicated locus is not sufficient to induce under-replication. (Inset) GAL4DBD-SUUR was bound to the UAS as analyzed by ChIP analysis relative to a negative control locus (pol α, where GAL4 is not bound). (D) Tethering of GAL4DBD-SUUR to the DAFC-62D amplicon in follicle cells does not affect amplification levels. The left panel shows the DNA copy number across the amplicon with GAL4DBD-SUUR tethered at the position designated by the asterisk. The right panel shows the sibling controls without the UAS binding site subjected to the same heat shock induction. The inset panel shows that GAL4DBD-SUUR was bound to the UAS, as determined by ChIP analysis.