Chromatin regulates origin activity in Drosophila ...

letters to nature ..............................................................

Chromatin regulates origin activity in Drosophila follicle cells Bhagwan D. Aggarwal & Brian R. Calvi Department of Genetics, University of Pennsylvania School of Medicine, 415 Curie Blvd, Philadelphia, Pennsylvania 19104, USA .............................................................................................................................................................................

It is widely believed that DNA replication in multicellular animals (metazoa) begins at specific origins to which a prereplicative complex (pre-RC) binds1. Nevertheless, a consensus sequence for origins has yet to be identified in metazoa. Origin identity can change during development, suggesting that there are epigenetic influences. A notable example of developmental specificity occurs in Drosophila, where somatic follicle cells of the ovary transition from genomic replication to exclusive rereplication at origins that control amplification of the eggshell (chorion) protein genes2. Here we show that chromatin acetylation is critical for this developmental transition in origin specificity. We find that histones at the active origins are hyperacetylated, coincident with binding of the origin recognition complex (ORC). Mutation of the histone deacetylase (HDAC) Rpd3 induced genome-wide hyperacetylation, genomic replication and a redistribution of the origin-binding protein ORC2 in amplification-stage cells, independent of effects on transcription. Tethering Rpd3 or Polycomb proteins to the origin decreased its activity, whereas tethering the Chameau acetyltransferase increased origin activity. These results suggest that nucleosome acetylation and other epigenetic changes are important modulators of origin activity in metazoa. To explore the relationship between chromatin modification and metazoan origin activity, we examined the origins that control developmental amplification of the Drosophila chorion genes. During stage 10A of oogenesis, somatic follicle cells undergo a developmental transition from genomic replication to continuous re-replication from origins at the chorion loci on the X and 3rd chromosome (hereafter referred to as X and 3rd chorion), as well as at two other recently identified loci2–5. Beginning at stage 10B of oogenesis, this continuous re-replication can be seen as four subnuclear foci by 5-bromodeoxyuridine (BrdU), fluorescence in situ hybridization (FISH), or antibody labelling for replication proteins2,6–8. The regulation of chorion origins resembles normal cell cycle control in that they assemble a pre-RC, which is activated by S-phase kinases9. The 3rd chorion locus amplifies to highest copy number, and the two sequences that are most important for this are termed amplification controlling element 3 (ACE3) and Ori-b, which bind to ORC in vitro and in vivo10 (Fig. 1a). These sequences can direct amplification when inserted at ectopic chromosomal sites, but the level of amplification is extremely sensitive to genomic position11, which can be buffered by chromatin insulators12. This is consistent with the notion that the chromatin neighbourhood in which the origin resides can have a major influence on its activity. To evaluate whether chromatin is modified at chorion origins, we immunolabelled Drosophila ovaries with antibodies against polyacetylated histone H4 (AcH4). We detected four prominent nuclear foci in amplification-stage follicle cells, with a lower level of staining throughout the nucleus (Fig. 1b). Several observations suggested that these four hyperacetylation foci correspond to origins at amplification loci during active initiation. First, the four hyperacetylation foci were coincident with amplification foci detected by an antibody to ORC2 (Fig. 1b–d). Second, at the 3rd chorion locus both ORC2 and AcH4 were localized to the origin during the period of active initiations from early stage 10B to late stage 11, but were not seen during the majority of stages 12–14 (a 372

time when no new initiations occur but forks continue to migrate)6,10,13 (Supplementary Fig. S1). However, AcH4 did persist slightly later than ORC2 into early stage 12 (a difference of less than one hour in developmental time). Third, hyperacetylation at the 3rd chorion locus corresponded closely to ORC2 labelling at the origin, but did not overlap with labelling for Double-parked protein (Dup), which co-localizes with migrating replication forks. This further suggested that AcH4 labelling does not represent acetylation on newly deposited nucleosomes behind forks (Fig. 1e–g and Supplementary Fig. S2)13. Although amplification increased the prominence of AcH4 foci, their intensity could not be accounted for by DNA copy number alone (Supplementary Figs S2 and S3). Similar results were obtained with antibodies against poly-acetylated histone H3 and histone H4 acetylated on lysine 8 (Supplementary Fig. S4). To delimit histone modification near the origin with higher resolution, we performed chromatin immunoprecipitation (ChIP) with an AcH4 antibody on genomic DNA from amplification-stage egg chambers. This indicated that ACE3 and Ori-b were both enriched approximately 25-fold in the AcH4 precipitation, relative to a non-amplified control locus, whereas sequences 10–50 kilobases (kb) proximal and distal to the 3rd chorion locus were not enriched (Fig. 2). This suggests that nucleosomes are hyperacetylated at ACE3 and Ori-b, the two ORC binding sites critical for amplification.

Figure 1 Histone hyperacetylation and ORC2 co-localize at chorion origins. a, Organization of the 3rd chromosome chorion locus. Arrows represent the four chorion genes, grey boxes represent the five regions that contribute to amplification. The two most important, ACE3 and Ori-b are binding sites for ORC. Sal I sites (S) define the 3.8-kb fragment used in subsequent experiments. b–d, Labelling of stage-10B follicle cells with poly-acetylated histone H4 (AcH4) antibody (green; b), ORC2 (red; c), and merge (yellow; d). The two brightest foci represent the chorion locus on the 3rd and X chromosome. e, Representation of ORC (red) binding to ACE3/Ori-b on amplified chorion DNA fibres based on previous reports6,10,13. f, High magnification showing co-localization of ORC2 (red) and AcH4 (green) at the 3rd chorion origin in stage 11. g, AcH4 (green) does not co-localize with Dup protein (red) which labels replication forks that migrate bi-directionally outward from the 3rd chorion origin13. Scale bars represent 10 mm (b–d) and 3 mm (f, g).

©2004 Nature Publishing Group

NATURE | VOL 430 | 15 JULY 2004 | www.nature.com/nature

letters to nature To determine whether acetylation regulates origin activity, we examined loss-of-function mutations in the HDAC gene Rpd3. In yeast, Rpd3 mutants advance the time at which late origins fire, a relatively subtle effect compared with the effect on transcription14. Severe mutations in Drosophila Rpd3 are homozygous lethal15. Therefore, we used the FLP/FRT recombination system to create clones of follicle cells homozygous for the strong Rpd3 m5–5 allele (J. Simon and M. O’ Connor, personal communication). To avoid pleiotropic effects that result from prolonged depletion of Rpd3 protein, we generated small clones (1–10 cells) late in oogenesis. Labelling with a Rpd3 antibody confirmed that cells within these clones had reduced levels of Rpd3 protein (Supplementary Fig. S5). Consistent with a defect in an HDAC, labelling with AcH4 and other antibodies indicated that amplification-stage Rpd3-mutant follicle cells had a 3 to 4 fold increase in hyperacetylation throughout the nucleus (Fig. 3a, b and Supplementary Fig. S6). Approximately 85% (n ¼ 100) of small clones with global acetylation also had altered replication, with BrdU incorporation throughout the nucleus instead of just at the amplification loci as seen in neighbouring wild-type cells (Fig. 3c, d and Supplementary Fig. S7). To test whether BrdU labelling in Rpd3-mutant cells could represent a change in origin usage, we examined the distribution of ORC2. Instead of the normal focal staining at chorion loci, many Rpd3-mutant cells with hyperacetylation also had a low level of

Figure 2 ChIP analysis indicates that nucleosomes are hyperacetylated at the 3rd chromosome chorion origin. ChIP from stage-10B and stage-11 egg chambers using anti-polyacetylated H4 (AcH4). a, Fivefold dilutions of input or pellet template DNA were subjected to PCR using primers to 3rd chorion ORC binding sites (ACE-3 and Ori-b; see Fig. 1a), sequences 10 kb and 50 kb proximal and distal to the origin, and a non-amplified control at 6C (control). PCR products were separated on agarose gels and stained with ethidium bromide. b, Quantification of PCR products. The ratio of pellet to input was normalized to the pellet to input ratio for the control. The bar graph represents the average and standard deviations of normalized enrichment for two independent precipitations and multiple PCR reactions. NATURE | VOL 430 | 15 JULY 2004 | www.nature.com/nature

punctate ORC2 labelling throughout the nucleus, with smaller clones most frequently having redistributed ORC2 (Fig. 3e–j and Supplementary Fig. S8). Among Rpd3-mutant clones comprised of five or fewer cells, 20% (n ¼ 41) had at least one large nucleus, and measurement of total 4,6-diamidino-2-phenylindole (DAPI) fluorescence indicated that they contained approximately twofold more DNA than neighbouring non-mutant cells (Fig. 3f, h, j, asterisk and data not shown). This suggests that they had undergone an extra genome reduplication rather than a developmental delay in genomic replication that normally occurs before stage 10B. In most cells that incorporated BrdU, however, we could not detect a significant increase in DNA content, suggesting that ongoing replication had not resulted in a full reduplication of the genome. Similar cell–cell variation has been noted in a number of other mutants that alter replication in follicle cells16,17. Treatment of egg chambers in vitro with the HDAC inhibitors sodium butyrate or trichostatin A (TSA) also resulted in hyperacetylation and extra genomic BrdU incorporation in stage-10B follicle cells (Fig. 4a, b, g and data not shown). This included a notable increase in BrdU incorporation within the heterochromatic chromocentre, which is under-replicated in most endocycles18 (Fig. 4b). The increased BrdU incorporation into this nuclear compartment, which is known to differ in histone acetylation, suggests that HDAC inhibitors, and mutation of Rpd3, alter replication specificity by changing acetylation status. In the cells with hyperacetylation, it is possible that the altered replication specificity is due to an increase in replication-protein expression, rather than a direct effect on chromatin at the origin19. Measurement of the total fluorescence intensity of redistributed ORC2 in Rpd3-mutant cells, however, indicated that it was not significantly increased (Supplementary Fig. S8 and data not shown). To answer this question, we asked if the transcription inhibitor a-amanitin would block the extra replication that results from inhibition of HDACs with sodium butyrate16. Although a-amanitin pre-treatment strongly inhibited heat induction of an hsp70:Myctagged reporter in controls, it did not have a significant effect on the extra genomic replication seen in sodium-butyrate-treated egg chambers (Fig. 4a–g). These results suggest that the extra genomic replication after treatment with HDAC inhibitors and in Rpd3mutant cells is not mediated solely by increased transcription of replication-protein genes. To test further whether chromatin modification at the origin can have an impact on replication activity locally, we established a twopart system in which chromatin-modifying proteins are tethered to the 3rd chorion origin in vivo (Fig. 5a). We transformed flies with an amplification reporter, Tether Target 1 (TT1), which contains five copies of GAL4-binding sites adjacent to the minimal 3rd chorion origin (which contains the ORC binding sites ACE3 and Ori-b; Figs 1a and 5a). Other strains were transformed with Rpd3 fused to the GAL4 DNA binding domain (GAL4DBD), and expressed under control of the heat-inducible hsp70 promoter (hsp70:GAL4DBD:Rpd3) (Fig. 5a and data not shown). In females containing both transgenes, we measured the ratio of copy number for TT1 versus the endogenous 3rd chorion locus within each follicle cell nucleus by FISH, using an origin probe2. Heat induction of hsp70:GAL4DBD:Rpd3 expression reduced TT1 amplification to 59% of controls (P , 0.0001; Fig. 5d, e, j). Expression of GAL4DBD alone, or heat-treatment alone, had no specific effect on TT1 amplification (P < 0.6; Fig. 5b, c, j). Moreover, hsp70:GAL4DBD:Rpd3 expression had no effect on S6.9-5, a 3rd chorion P element that lacks GAL4-binding sites (data not shown). In some cells, expression of hsp70:GAL4DBD:Rpd3 reduced TT1 amplification to below the level of detection by FISH. Therefore, we also used the fraction of nuclei with detectable TT1 and endogenous 3rd chorion signal as a measure of relative origin strength. This gave similar results, with hsp70:GAL4DBD:Rpd3 reducing detection to 70% of controls (Fig. 5j). Finally, quantitative


373

letters to nature Southern blotting of stage-13 egg chambers suggested that hsp70– GAL4DBD–Rpd3 reduces TT1 amplification to 75% of controls (Fig. 5j). Similar results were obtained with another TT1 inserted at a different genomic location (data not shown). This suggests that Rpd3 can inhibit origin activity locally. To evaluate the contribution of HDAC activity to repression of the origin, we tethered an Rpd3 protein mutated in an active-site histidine that is essential for HDAC activity in other organisms20. This mutant, GAL4DBD:Rpd3H137A, retained partial repression activity and reduced TT1 amplification to 66% (P , 0.0001), and reduced detection to 89%. This is consistent with results from yeast and mammalian cells, where catalytically inactive GAL4DBD:Rpd3 fusions can partially repress transcription in vivo20. Similarly to the results at promoters, it is likely that Rpd3H137A can recruit other repressing proteins to the origin and, therefore, does not distinguish the contribution of deacetylation to origin repression.

Therefore, we tethered HAT1, a product of the chameau gene and the putative fly orthologue of human HBO1, to TT121,22. HBO1 has been shown to associate with the Drosophila and human preRC proteins MCM2 and ORC1 (refs 21, 23). Expression of hsp70:GAL4DBD:HAT1 increased the ratio of TT1/3rd to 153% (P , 0.0001), increased the fraction of nuclei with detectable TT1 to 126% and increased the copy number on Southern blots to 165% of controls (Fig. 5f, g, j). Similar results were obtained with a TT1 inserted at a different genomic location (data not shown). This suggests that increased acetylation stimulates origin activity. We also asked whether a repressive chromatin environment mediated by the silencing protein Polycomb (Pc) diminishes origin activity. Polycomb is part of a multi-protein complex that associates with HDACs, and is required for maintenance of chromatin domains that repress transcription at homeotic and other loci24. Tethering Pc to TT1 reduced the TT1/3rd ratio to 66%

Figure 3 Rpd3 loss-of-function clones show hyperacetylation, increased replication and altered ORC2 distribution. Homozygous mutant Rpd3 m5-5-follicle-cell clones were generated using the FLP/FRT technique. a, c, e, g, i, Low magnification of stage-10B egg chambers. b, d, f, h, j, Higher magnification images of clones designated by arrows. a, b, Mutant clones were identified in stage-10B egg chambers by the absence of green fluorescent protein (GFP) fluorescence (green), and labelling for AcH4 indicated that most Rpd3 m5-5 mutant cells within small clones have nucleus-wide hyperacetylation (red). Rare regional hyperacetylation can be seen for one cell in the lower part of the clone in (b). c, d, Many cells in small clones also had inappropriate nucleus-wide incorporation of

BrdU, instead of the focal staining at chorion loci seen in neighbouring wild-type cells (see Supplementary Fig. S7). e–j, Orc2, AcH4 double labelling. Acetylation (red; e, f), ORC2 (green; g, h), merge (yellow; i, j). ORC2 was distributed throughout the nucleus in 51% of clones comprised of ten or fewer cells (n ¼ 242 clones; see Supplementary Fig. S8). Note that outside the clones the green colour is from ORC2 and GFP. The asterisk in f, h and j indicates a nucleus that contains twice the DNA content of its neighbours, as measured by total DAPI fluorescence. Scale bars represent 20 mm (a, c, e, g, i) and 10 mm (b, d, f, h, j).

374


NATURE | VOL 430 | 15 JULY 2004 | www.nature.com/nature

letters to nature model awaits further characterization of additional origins in Drosophila. Like Rpd3, the Drosophila retinoblastoma-family protein, Rbf1, represses the activity of chorion and genomic origins in follicle

Amplification reporter

a

GAL4 fusion protein

TT1 UAS ACE3 Ori-β

w+

>

S

S

Control

Heat-shock

c

d

e

Control

Heat-shock

f

g

h

i

Gal4–Pc

Gal4DBD

b

Gal4–HAT1