In vivo analysis reveals that ATP-hydrolysis couples remodeling to SWI/SNF release from chromatin

To visualize (P)BAP interactions with genomic loci in interphase cells, we imaged BRM in 3rd instar larval salivary gland polytene nuclei. In agreement with early studies (Armstrong et al., 2002; Mohrmann et al., 2004), visualization of endogenous BRM on fixated polytene chromosome spreads by epi-immunofluorescence (IF) revealed binding to loci within interbands, and to puffs (Figure 1A). BRM is largely excluded from heterochromatic bands, which stain strongly with DAPI and with antibodies directed against the core histones (α-HIS). Depletion of BRM affects does not lead to gross changes in the polytene chromosome banding pattern or the binding of RNAPII (Figure 1—figure supplement 1A). Whereas confocal IF imaging of BRM in formaldehyde-fixed whole mount salivary gland nuclei confirmed its association with chromosomal interbands, it also revealed that the majority of BRM resides in the nucleoplasm (Figure 1B). Indeed, co-localization analysis of BRM and DAPI yielded a Manders coefficient of 0.05, implying that only a small proportion of BRM is chromatin engaged at any given moment. A line-scan illustrates the preferential association of BRM with decondensed chromatin and its exclusion from highly compacted bands (Figure 1C). For comparison, the majority of RNAPII associated with well-defined chromosomal loci and has a lower abundance in the nucleoplasm (Figure 1D). In conclusion, our IF results show that BRM preferably interacts with interbands, but that, surprisingly, the majority (~95%) of BRM is not bound to chromatin.

To study the dynamics of BRM’s interactions with polytene chromosomes, we generated transgenic Drosophila lines expressing fluorescent protein (either enhanced green fluorescent protein, GFP, or monomeric cherry, mCh)-tagged BRM, histone H2A and H2B. Figure 1E illustrates the expression of GFP-BRM in 3rd instar larval salivary gland nuclei. Laser scanning confocal microscopy of GFP-H2A and mCh-H2B in isolated salivary glands revealed strong co-localization, and recapitulation of the polytene banding pattern (Figure 1F). Thus, local chromatin condensation can be captured by live cell imaging of cultured salivary glands. Indeed, the intensity of the GFP-H2B signal corresponded to the highly compacted chromocenter, telomers and heterochromatic bands (black chromatin; yellow arrowhead), gray bands (red arrowhead) and white chromatin interbands (blue arrowhead; Figure 1G). Previous analysis of a range of chromatin marks suggested that approximately 5% of Drosophila genomic DNA comprises promoters and regulatory sequences, forming white chromatin. About 25% of the genome corresponds to gray chromatin, harboring active genes, whereas the remaining 70% represents black heterochromatin, comprising mainly intercalary heterochromatin and inactive genes (Eagen et al., 2015; Filion et al., 2010; Szabo et al., 2019; Kharchenko et al., 2011; Zykova et al., 2018). To determine the level of chromosomal condensation, we measured the intensity of GFP-H2B across multiple nuclei. First, we quantified the GFP-H2B intensity distribution across z-stacks of multiple individual nuclei. Next, we binned GFP-H2B intensities in three classes, corresponding to either white, gray, or black chromatin (Figure 1—figure supplement 1B). This analysis indicated that the volume of polytene chromosomes in nuclear space comprises about 5 (±1) % black, 43 (±4) % gray, and 52 (±5) % white chromatin (Figure 1H). Applying these estimates of GFP-H2B density distribution in living cells implies that chromatin in gray bands is on average sixfold more condensed than in interbands, whereas black chromatin is condensed an additional ~25-fold (up to ~150 more than open chromatin). Thus, visualization of fluorophore-tagged histones can be used to monitor chromatin condensation in living cells.

Next, we determined the distribution of GFP-BRM in live polytene nuclei. MOR (SMARCC1/2) co-immunopurifications (co-IPs) from dissected larval salivary gland extracts revealed the efficient incorporation of GFP-BRM in remodeler complexes (Figure 1—figure supplement 1C). IF of polytene chromosomes showed extensive co-localization of GFP-BRM with endogenous MOR, confirming that it binds chromatin normally as part of (P)BAP (Figure 1—figure supplement 1D). MOR has a key architectural function that is essential for the structural integrity of (P)BAP and the stability of BRM in vivo (Moshkin et al., 2007). RNAi-mediated depletion of MOR in the larval salivary gland caused a concomitant loss of GFP-BRM, indicating that it predominantly exists as part of (P)BAP (Figure 1—figure supplement 1E). We conclude that GFP-BRM is incorporated in (P)BAP and is targeted correctly. Vital imaging of salivary gland nuclei expressing mCh-H2B and GFP-BRM showed that the majority of GFP-BRM is nucleoplasmic and excluded from the nucleolus (Figure 1I). GFP-BRM associated with interband chromatin, while its level was strongly reduced within chromosome bands. We made similar observations for another (P)BAP core subunit, GFP-SNR1 (SMARCB1; Figure 1—figure supplement 1F). Pertinently, BRM and SNR1 levels within the condensed polytene chromosome bands are much lower than in the nucleoplasm. Note that loss of BRM does not affect polytene chromosome condensation (Figure 1—figure supplement 1A). These observations raise the intriguing possibility that chromosome condensation leads to (P)BAP exclusion. In contrast to BRM, GFP-Polycomb (PC) and the RNAPII subunit mCh-RBP3 each showed strong binding to specific chromosomal loci and relatively low abundance in the nucleoplasm (Figure 1J and K). In conclusion, a surprisingly small proportion of (P)BAP engages the genome at a steady state level. (P)BAP interacts with open chromatin but is largely excluded from highly condensed chromosome bands.

We were interested in (P)BAP dynamics during developmental gene activation. Therefore, we investigated the role of (P)BAP in Ec-controlled gene transcription. We dissected salivary glands from early 3rd instar larvae and cultured them for ~1 hr, either in the presence or absence of Ec. IF of fixated polytene chromosome spreads revealed binding of the EcR to the E74 and E75 loci, harboring Ec early response genes (Figure 2A). Like the EcR, BRM is present on the E74 and E75 loci both before and after induction. Concomitant with RNAPII recruitment and puffing (Figure 2B), the amount of BRM at E74 and E75 increased substantially following Ec-induction. Next, we compared the recruitment of the BAP-specific subunit OSA (ARID1A/B) and the PBAP-specific PBRM (Figure 2C). Before induction, both OSA and PBRM were present at E74 and E75. After puffing, PBRM accumulated on the E74 and E75 loci, but we detected no appreciable binding of OSA (Figure 2D). On the E74 and E75 loci, PBRM co-localized with the EcR (Figure 2E). Thus, both BAP and PBAP bind the repressed E74 and E75 loci, but only PBAP is recruited after transcriptional activation by Ec. This observation agrees well with our earlier report that PBAP-specific mutants, but not OSA mutants, display microcephaly and leg malformations that are reminiscent of phenotypes caused by defective Ec-signaling (Chalkley et al., 2008). Collectively, these results suggest a role for PBAP in gene regulation by Ec. To test this notion, we compared the expression profiles of E74 and E75 transcripts in homozygous Bap170^kim1; Pbrm^33.2 prepupae to that of wt animals. Prepupae were collected at pupariation (t=0), and RNA was extracted at 2 hr intervals for 12 hr, and monitored by reverse transcription (RT)-qPCR (Figure 2F). Loss of BAP170 (ARID2) and PBRM disrupted the expression of the E74 and E75 genes, showing that PBAP is required for Ec-controlled gene regulation in vivo. Finally, we compared the requirement of MOR, OSA, BAP170 and PBRM for induction of E74 and E75 expression by Ec in S2 cells (Figure 2G). Again, the loss of PBAP signature subunits, but not OSA, abrogated gene induction by Ec. Collectively, these results show that PBAP is required for Ec-induced developmental gene expression.

To study Ec-regulated loci on polytene chromosomes in live salivary glands, we generated transgenic Drosophila lines expressing GFP- or mCh-tagged EcR. Confocal imaging of GFP-EcR and mCh-H2B allowed us to readily identify the E74 and E75 gene clusters in cultured salivary gland nuclei (Figure 3A). Co-expression of mCh-EcR with GFP-BRM illustrates the highly defined genomic localization of the sequence-specific transcription factor compared to the diffuse distribution of the remodeler (Figure 3B). A high-resolution zoom shows the puffing of the chromatin bound by EcR, which are flanked by condensed bands (Figure 3C). The EcR binding sites within the E74 and E75 loci have been mapped by ChIP and cover approximately 40 and 60 kb, respectively (Bernardo et al., 2014). We plotted the EcR sites on the genomic map, and indicated the simplified local chromatin state (white, gray, or black), which was derived from combining available maps of histone marks (Filion et al., 2010; Kharchenko et al., 2011; Figure 3D). Two well-defined heterochromatic bands (fragments I and III) flank the GFP-EcR-marked E75 puff (fragment II; Figure 3C and D). We measured the lengths of these fragments in 20 different nuclei, allowing us to compare observed physical distances in living cells with genomic DNA lengths (Figure 3E). The packing ratio of an 11 nm nucleosomal array is predicted to be 6.8:1. This would yield an extended fiber length for fragment I of ~18 μm. However, the observed average length of fragment I in living polytene nuclei is only 0.72 (± 0.18) μm, yielding a 25-fold compaction. Similarly, the heterochromatic band (fragment III) flanking E75 appears to be condensed ~22-fold, compared to an extended 11 nm chromatin fiber. In contrast, measurement of the GFP-EcR marked puff (fragment II) suggested a compaction of only threefold. Given that the EcR-bound enhancers and promoters form loops (Bernardo et al., 2014), and that cellular chromatin is a flexible disordered polymer, these observations are consistent with the notion that the puffed area comprises an open array of nucleosomes. When we considered the polytene chromosome as a cylinder, and included diameter measurements to calculate apparent volumes, the difference in chromatin compaction between puffs and bands became even more pronounced. For example, the chromatin in the heterochromatic fragment I appears to be ~30-fold more compacted than that in the GFP-EcR-marked puff of fragment II. Thus, vital imaging of fluorophore-tagged histones on polytene chromosomes allows measurements of relative interphase chromatin condensation at specific loci in living cells.

Our analysis of (P)BAP in polytene nuclei showed that only a fraction of (P)BAP engages chromatin at a given moment. This raises the question of remodeler turnover on chromatin and the role of ATP-hydrolysis in this process. To address this issue, we first generated S2 cell lines that express either GFP-BRM or a catalytically inactive ATP-binding mutant, GFP-BRM-K804R (Elfring et al., 1998). Western blotting established that GFP-BRM and GFP-BRM-K804R were expressed at comparable levels, which were well below that of endogenous BRM (Figure 4A). IPs with antibodies directed against GFP followed by mass spectrometry showed that both GFP-BRM and GFP-BRM-K804R associated with the full complement of BRM complex subunits (Figure 4B). Interestingly, we identified a homolog of the mammalian GBAF-specific subunit GLTSCR1/L in the GFP-BRM IPs, suggesting that Drosophila contains a corresponding complex. Both GFP-BRM and GFP-BRM-K804R were nuclear and excluded from the nucleolus (Figure 4C). Although they were expressed at comparable levels, GFP-BRM and GFP-BRM-K804R displayed notably different patterns of distribution. Although wild type (wt) GFP-BRM was distributed evenly throughout the nucleoplasm, GFP-BRM-K804R displayed a punctate pattern. Quantification of local variance of GFP intensity confirmed the clustering of GFP-BRM-K804R (Figure 4D). Recently, liquid-liquid phase separation driven by weak hydrophobic interactions has been implicated in localized enrichment and condensation of proteins (McSwiggen et al., 2019; Plys and Kingston, 2018). Poly-alcohols, such as 1,6-hexanediol (1,6-Hex) that disrupt weak hydrophobic interactions, have become popular as a diagnostic tool to identify phase separation in biological processes. However, the addition of 1,6-hexanediol did not affect the punctate pattern of GFP-BRM-K804R, indicating that phase separation does not explain the clustering of GFP-BRM-K804R (Figure 4C).

To gain insight in the dynamic behavior of GFP-BRM and GFP-BRM-K804R, we compared their fluorescence recovery after photobleaching (FRAP) profiles after bleaching a 16-pixel wide strip across the center of the nucleus (Figure 4E). The turnover of GFP-BRM-K804R was approximately two-times slower than that of GFP-BRM, with a time of half recovery (t_1/2) of ~ 1.7 (± 1.3) seconds (s) compared to ~ 0.7 (± 0.4) s for wt BRM. Cancer-associated mutations in the ATP-binding cleft of human SMARCA4 caused a comparable ~ 2-fold reduction in FRAP recovery kinetics in diploid cells (Hodges et al., 2018). These observations suggest a role for ATP binding and hydrolysis in remodeler dynamics. The recovery curves did not identify a substantial immobile fraction for either GFP-BRM-K804R or GFP-BRM. Next, we used chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) assays to compare the binding of GFP-BRM and GFP-BRM-K804R to selective (P)BAP target loci that were identified earlier (Moshkin et al., 2012). The catalytically inactive BRM-K804R yielded two- to fivefold stronger ChIP signals at a variety of functional binding sites than wt BRM (Figure 4F). The increased chromatin binding of the BRM ATP-binding mutant to functional target loci is reminiscent of earlier observations for the yeast Iswi2 remodeler (Gelbart et al., 2005). Collectively, these results indicate that the loss of ATP-binding causes chromatin retention and clustering of BRM-K804R.

Vital imaging of BRM in S2 cells revealed a fast exchange dynamic, suggestive of transient probing of the genome. However, it is impossible to visualize transcription factor interactions with genomic loci in regular interphase cells. To overcome this limitation, we imaged BRM in salivary gland polytene nuclei. First, we used fluorescence loss in photobleaching (FLIP)-FRAP to determine the in vivo dynamics of GFP-BRM in salivary gland polytene nuclei. Following photobleaching of half a nucleus, we observed a rapid recovery that was largely complete after ~2 min (Figure 5A and B). The increase in fluorescence in the bleached part of the nucleus was accompanied by a comparable decrease in the unbleached part. We noted no difference between BRM in the nucleoplasm or associated with chromatin (Figure 5A). In contrast to GFP-BRM, mCh-H2B showed no recovery within a timeframe of minutes (min; Figure 5C and D). Thus, (P)BAP can move freely through the nucleus, whereas the turnover of bulk histones is limited. FRAP analysis of GFP-BRM bands on chromosomes also showed a rapid and complete recovery of fluorescence (Figure 5E and F). Kinetic analysis revealed a recovery t_1/2 of 3.1 (±1.1) s and no substantial immobile fraction. Similar FRAP kinetics were observed for a low-expressing GFP-BRM Drosophila line (Figure 5—figure supplement 1A–C). Moreover, FRAP analysis of GFP-SNR1 (SMARCB1) gave comparable results (Figure 5G and Figure 5—figure supplement 1D–F). Finally, co-localization with RNAPII did not affect the FRAP kinetics of GFP-BRM substantially (Figure 5—figure supplement 1G,H). The rapid exchange and absence of an immobile fraction shows that the interactions of (P)BAP with chromatin are highly dynamic, with retention times of just a few seconds. Consequently, (P)BAP kinetics are dominated by free diffusion in the nucleoplasm.

The RNAPII subunit GFP-RBP3 displayed a very different kinetic profile. In agreement with earlier studies (Yao et al., 2007), photobleaching of chromatin associated GFP-RPB3 resulted in an initial rapid increase in fluorescence, followed by a second, more prolonged linear phase of recovery which lasted for several min (Figure 5G). The rapid recovery at the start probably reflects dynamic binding to the promoter, whereas the linear phase corresponds to elongating RNAPII. The turnover of bulk histone H2B is more than two orders of magnitude slower than that of (P)BAP. FRAP analysis of GFP-H2B in polytene chromosomes revealed that it takes over 40 min to obtain a ~20% recovery (Figure 5H). GFP-H2B recovery in polytene bands versus interbands appeared to be proportional, consistent with the notion that both comprise the same basic nucleosomal fiber at different degrees of condensation (Ou et al., 2017).

To determine the turnover of BRM at highly transcribed loci that depend on BRM for expression, we activated E74 and E75 by adding Ec to isolated salivary glands. Ec induces prominent polytene chromosome puffs, which are a manifestation of very high levels of gene transcription (Figure 6A,B). The addition of Ec leads to increased RNAPII occupation of the E74 and E75 loci (monitored by GFP-RPB2), which peaks ~35 min after hormone addition (Figure 6C,D and Figure 6—figure supplement 1A). FRAP analysis of GFP-RPB2 at the peak of RNAPII accumulation (30 min), revealed only a partial (~60%) recovery that ceased after ~5 min (Figure 6E). The absence of a full recovery of RNAPII at the height of E74 and E75 expression and puffing might reflect either stalling or recycling. However, similar behavior of RNAPII on fully induced heat-shock loci was shown to be caused by local recycling (Yao et al., 2007). FRAP analysis of mCh-EcR revealed a recovery t_1/2 of 3.0 (± 1.1) s, which is comparable to BRM. In contrast to BRM, however, a small but consistent fraction of about 0.14 (± 0.07) of mCh-EcR appeared to be immobile (Figure 6F). Histones at the E74 and E75 puffs did not exchange notably (Figure 6—figure supplement 1B). Finally, we compared the targeting of the BAP-specific D4 (DPF1-3) subunit and PBAP-specific BRD7 by live cell imaging. In agreement with our results for endogenous BAP versus PBAP on fixated polytene spreads (Figure 2D), GFP-BRD7 co-localized with mCh-EcR on the E74 and E75 loci, whereas GFP-D4 did not (Figure 6G, Figure 6—figure supplement 1C,D). FRAP analysis of GFP-BRD7 revealed fast kinetics on the active E74 and E75 loci, which was comparable to that of BRM (Figure 6H). Likewise, GFP-D4 it displayed a fast turnover on BAP target loci (Figure 6—figure supplement 1E). In summary, our FRAP results suggest that at the peak of induction, a substantial portion of RNAPII is recycled at the E74 and E75 puffs. However, PBAP, which is essential for expression of E74 and E75, exchanges rapidly without a substantial immobile fraction. We conclude that (P)BAP acts through a continuous and rapid probing of the genome. By comparison, the bulk turnover of histones is limited and slow.

To investigate the role of ATP hydrolysis in (P)BAP-chromatin interactions, we permeabilized salivary glands with digitonin, using a method developed to study Ec-induced puff formation (Myohara and Okada, 1987). Digitonin-treated glands are permeable to high-molecular-weight molecules but dependent upon the addition of ribonucleoside triphosphates, retain the ability to form puffs in response to Ec. Treatment of cultured salivary glands with 0.01% digitonin resulted in a complete loss of GFP fused to a nuclear localization signal (NLS-GFP), whereas mCh-H2B remained bound to the chromosomes (Figure 7A). In striking contrast, chromatin-binding of GFP-BRM increased dramatically following permeabilization. Moreover, free GFP-BRM was no longer detectable in the nucleoplasm. Conversely, chromatin association of GFP-PC and mCh-RBP3 was substantially reduced in permeabilized glands. Although BRM normally only interacts transiently with chromatin and is mainly nucleoplasmatic, heat maps of GFP-BRM illustrate that permeabilization resulted in strong accumulation of GFP-BRM onto chromatin (Figure 7B). Chromatin-bound GFP-BRM in permeabilized cells was essentially immobile, as determined by FRAP (Figure 7C). Following permeabilization, GFP-BRM remained bound for more than an hour. Tellingly, upon the addition of ATP (2 mM), GFP-BRM was immediately released from chromatin and diffused out of the nuclei (Figure 7D). We used this assay to assess the nucleotide requirements for GFP-BRM release (Figure 7E). First, ATP could not be replaced by ADP. Second, efficient release of GFP-BRM required ATP concentrations above 1 mM. Third, neither the slowly hydrolyzed analogue ATP-γ-S nor the non-hydrolysable ATP ground state mimetic ADP-AlF₄^– could replace ATP. Thus, the release of GFP-BRM from polytene chromosomes requires hydrolysable ATP at concentrations greater than 1 mM. Note that ATP concentrations in living cells normally do not drop to a level below 1 mM, when it would impede remodeler dynamics. To monitor the effect on endogenous (P)BAP, we repeated the permeabilization in either the presence- or absence of ATP, followed by fixation and IF of MOR (SMARCC1/2). Similar to our vital imaging results, we found that MOR accumulated on chromatin in permeabilized glands but dissociated and diffused out of the nucleus upon addition of ATP (Figure 7F). Thus, endogenous (P)BAP and GFP-BRM show similar behavior. Taken together, these results show that ATP hydrolysis is required for (P)BAP release, but not for binding to chromatin.

To complement the ATP-depletion experiments in permeabilized cells, we analyzed the dynamics of the ATP-binding deficient GFP-BRM-K804R in polytene salivary glands. BRM-K804 acts as a dominant negative mutant in developing Drosophila. Depending on its level of expression, BRM-K804 causes homeotic transformations or lethality (Elfring et al., 1998). IF of polytene chromosome spreads revealed co-localization of GFP-BRM-K804R with endogenous MOR (SMARCC1/2), indicating largely normal targeting of this mutant (Figure 8A). The chromosomes in GFP-BRM-K804R expressing glands, however, were considerably thinner than in wt glands, and had an increased tendency to break during preparation. These observations suggest that the expression of GFP-BRM-K804R leads to DNA under-replication and fragile polytene chromosomes. Vital imaging of intact salivary glands revealed that, in contrast to wt BRM, the majority of GFP-BRM-K804R is bound to chromatin (Figure 8B). Moreover, FRAP analysis showed that GFP-BRM-K804R has no substantial turnover (Figure 8C). Permeabilization with digitonin, either in the presence or absence of ATP, did not affect the chromosome-association of GFP-BRM-K804R (Figure 8D and E). IF of SNR1 (SMARCB1) showed that the expression of GFP-BRM-K804R forces binding of endogenous (P)BAP to polytene chromosomes (Figure 8F). The clustered binding of GFP-BRM-K804R prompted us to consider the potential role of phase separation. The addition of 1,6-hexanediol, however, affected neither the diffuse distribution of GFP-BRM nor the clustered binding to interband chromatin of GFP-BRM-K804R (Figure 8G). As reported (Strom et al., 2017), 1,6-hexanediol did reduce the accumulation of mCh-HP1 on the chromocenter. The insensitivity to 1,6-hexanediol and the lack of mobility revealed by FRAP suggest that phase separation does not play a major role in the clustering of GFP-BRM-K804R. Rather, we propose that inadequate release from chromatin drives the local accumulation of BRM-K804R (and that of wt BRM in the absence of ATP; see Figure 7). We note that the effect of the K804R mutation on BRM mobility is more pronounced in polytene nuclei than in S2 cells, which are near tetraploid. Most likely, the high local density of target loci in polytene chromosomes amplifies the effect of BRM-K804R retention on chromatin.

When co-expressed in the larval salivary glands, mCh-BRM-K804R altered the intra nuclear distribution of GFP-BRM dramatically. Instead of being ~95% nucleoplasmic, GFP-BRM now co-localizes with mCh-BRM-K804R on the chromosomes (compare Figures 1I and 8H). Moreover, FRAP analysis showed that in the presence of mCh-BRM-K804R, GFP-BRM also became immobile (Figure 8I). Note that remodeler complexes contain a single ATPase subunit and do not multimerize in solution (Jungblut et al., 2020; Sundaramoorthy and Owen-Hughes, 2020). Nevertheless, BRM-K804R captures wt (P)BAP onto polytene chromosomes, possibly via a chromatin-mediated interaction. These results raise the intriguing possibility that remodeling involves direct cooperation, or a hand-off mechanism between different BRM complexes. Finally, we studied the impact of BRM-K804R on histone turnover. We used FRAP to determine histone exchange in salivary glands that co-expressed mCh-H2B with either GFP-BRM or GFP-BRM-K804R (Figure 8J). The fluorescence recovery of mCh-H2B overlapping with GFP-BRM-K804R (7 (± 3) %) was significantly slower than in the presence of GFP-BRM (18 (± 3) %; Figure 8K). These results suggest that chromatin remodeling and remodeler release are intrinsically coupled.

Full-length cDNAs encoding BRM, SNR1, D4, BRD7, EcR, RBP2, RBP3, H2A, and H2B were cloned into pENTR/TEV/D-TOPO by TOPO cloning according to the manufacturer’s protocol, resulting in a series of pENTR vectors, which was used to generate derivative constructs. BRM-K804R was constructed via PCR-based site-directed mutagenesis of pENTR-Brm using standard procedures. To generate vectors that express eGFP- or mCherry (mCh)- fusion proteins, the appropriate coding sequences were recombined into the destination vectors pTGW or pTChW using LR ClonaseII (Invitrogen) according to the manufacturer’s protocol. pTGW was obtained from the Drosophila Genomics Resource Center (DGRC). The pTChW vector was constructed from pTGW by replacing the GFP coding sequence with mCh using the NcoI and AgeI restriction sites. The constructs for expression of GFP-BRM or GFP-BRM-K804R under the control of the metallothionein (Mt) promoter in S2 cells were generated by LR Clonase II-mediated recombination of BRM and BRM-K804R into pMGW. pMGW is a modified version of pAGW (DGRC) in which the actin promoter has been replaced by the Mt promoter. The Mt promoter was PCR-amplified from pMT/V5-HisA (Invitrogen) and ligated into pAGW using BglII and EcoRV restriction sites. The integrity of all constructs we generated was verified by DNA sequencing of the entire coding sequence and flanking regions.

To generate transgenic fly lines, the appropriate constructs were injected in Drosophila yw embryos or w¹¹¹⁸ and balanced transformants were isolated by BestGene Inc (Chino Hills, USA) and further analyzed in our lab. GFP-PC flies were a gift from Renato Paro (Dietzel et al., 1999) and the mCh-HP1 line was a gift from Jean-Michael Gilbert and Francois Karch (University of Geneva). UAS-NLS-NES^P12-GFP (stock number 70330) and Sgs3-GAL4 (salivary gland driver, stock number 6870) fly lines were obtained from the Bloomington Drosophila Stock Center (https://bdsc.indiana.edu↗). The polybromo33.2 and bap170^kim1 mutants have been described (Chalkley et al., 2008). All stocks were maintained on standard corn medium at 18°C. All crosses and experiments were carried out at room temperature. A list of Drosophila lines is provided in Supplementary file 1.

Schneider 2 cells (obtainded from ATCC; ATCC Cat# CRL-1963, RRID:CVCL_Z232↗) were maintained in Schneiders Drosophila medium containing 10% fetal calf serum at 25°C. S2 cells were transfected with cellfectin II reagent (Thermo Fisher Scientific) according to the manufacturer’s protocol. Cells tested negative for mycoplasma contamination. To enable drug selection, pCoBlast (Invitrogen) was cotransfected with either pMGW-Brm or pMGW-Brm-K804R in a ratio of 1:20. After 24 hr, transfected cells were selected with 25 µg/ml blasticidin. Following the establishment of polyclonal stable lines, cells were cultured in the presence of 5 µg/ml blasticidin.

Schneider 2 (S2) cells were maintained in Schneider’s Drosophila medium containing 10% fetal calf serum at 25°C. GFP-BRM and GFP-BRM-K804R expression in S2 cells was induced for three days by addition of CuSO4 (500 µM final concentration). Cells were fixed with 1% formaldehyde for 10 min. Fixation was stopped with 100 mM glycine. Next, cells were washed twice with PBS and lysed in 0.75% SDS, 10 mM EDTA, 50 mM Tris pH 8.0, supplemented with protease inhibitors (1 µg/ml pepstatin A, 1 µg/ml leupeptin, 1 µg/ml aprotinin, 0.2 mM AEBSF), which were present in all subsequent buffers until elution. The volume of the lysates was measured and SDS concentration was adjusted to 0.6%. Cross-linked chromatin was sheared to 200–300 bp of DNA length in a Bioruptor UCD-200 sonicator. Chromatin concentration was measured in a NanoDrop spectrophotometer (Thermo Fisher Scientific) and equal amounts were used as input for each ChIP. Chromatin was diluted 10-fold with 1% Triton X-100, 2 mM EDTA pH 8.0, 150 mM NaCl and 20 mM Tris pH 8.0, cleared by centrifugation and incubated overnight at 4°C with either 1.5 µl anti-GFP (Abcam #290) or mock antibodies. Next morning, 10 µl preblocked Protein A Sepharose (GE Healthcare) was added and incubated for 2 hr, followed by four washes with wash buffer (20 mM Tris-HCl pH8.0, 2 mM EDTA pH 8.0, 0.1% SDS, 1%Triton X-100) containing 150 mM NaCl and 1x with wash buffer containing 500 mM NaCl. For DNA elution, samples were incubated in 0.1 M NaHCO₃, 1% SDS, and 0.5 mg/ml proteinase K for 2 hr at 37°C and overnight at 65° C. DNA was subsequently purified by phenol/chloroform extraction, followed by ethanol precipitation. Immunopurified chromatin samples were analyzed by real-time PCR, using the comparative Ct method (Mutskov and Felsenfeld, 2004). The results of three independent biological replicates were averaged and the standard deviation was determined. For prepupal analysis, polybromo33.2 and bap170^kim1 mutants mutant lines were rebalanced with GFP-marked balancer chromosomes. GFP-negative mutant prepupae and wt prepupae were collected at 2 hr intervals from the moment of pupariation (t = 0) for 12 hr. RNA was extracted using TRIzol. Knockdown of (P)BAP subunits in S2 cells was performed as described (Chalkley et al., 2008; Moshkin et al., 2007; Moshkin et al., 2012). Twenty-four hr after treatment with dsRNA, 20-hydroxyecdysone was added to a final concentration of 20 μM. Cells were harvested 24 hr later and RNA was extracted using TRIzol. Knockdowns were performed biological triplicates. RNA levels in prepupae or S2 cells were analyzed by first-strand cDNA synthesis with Superscript II reverse transcriptase (Invitrogen) and subsequent quantitative PCR (qPCR) with SYBR green I using a MyiQ single-color real-time PCR detection system (Bio-Rad). Analysis of the reverse transcription (RT)-qPCR data was performed using the 2^−ΔΔCT method (Livak and Schmittgen, 2001). CG11874 was used as an internal control mRNA. A list of primers used is provided in Supplementary file 2.

GFP-BRM and GFP-BRM-K804R expressing S2 cells were grown on glass coverslips and experiments were performed in Schneider Drosophila medium containing 10% FCS. Intact salivary glands expressing eGFP or mCh-tagged proteins were excised in PBS, rinsed and transferred to an incubation chamber containing Grace’s insect medium diluted 5: one with water (Yao et al., 2008). Confocal images (1024 x 1024 pixels) of the nuclei were acquired at room temperature using a Leica TCS SP5 confocal microscope (Argon 488 nm and HeNe 594 nm laser lines), an 63x HCX PL APO CS 1.4 NA oil-immersion objective and Leica LAS-AF acquisition software. Images were processed and analysed using Fiji/ImageJ software (https://imagej.net↗). FRAP experiments were performed on a Leica TCS SP5 confocal microscope using Leica LAS-AF data acquisition software and a 40x HCX PL APO CS 1.3 NA (salivary glands) or a 63x HCX PL APO CS 1.4 NA (Schneider 2 cells) oil-immersion objective. For Schneider 2 cells, a strip of 128 x 16 pixels across the nucleus was bleached by 100% laser power of the Ar 488 nm laser line. Pre- and post-bleaching images (128 x 128 pixels, total of 20 and 200 images, respectively) were collected with 70 ms intervals and the average pixel intensity of both the bleached area and the whole nucleus was determined. Fluorescence recovery curves were full-scale normalized, double exponential fitted and the half-time of fluorescence recovery (t_½) and mobile fraction were determined using the EasyFRAP online FRAP analysis tool (https://easyfrap.vmnet.upatras.gr↗; Koulouras et al., 2018). Half-nucleus FLIP-FRAP and (dual color) single-band FRAP experiments on salivary glands nuclei were performed in Grace’s insect medium diluted 5:1 with water at room temperature. Images (512 x 512 pixels) were acquired using the Ar 488 nm (eEGF) and/or HeNe 594 nm (mCh) laser lines with an interval of 648 ms. Photobleaching was obtained by 100% laser power of the Ar 488 nm (eGFP) and/or 100% laser power of the DPSS 561 nm (mCh). For half nucleus FLIP-FRAP, the average pixel intensity of both the unbleached and bleached areas were determined, corrected for background fluorescence and expressed as percentage of the average pre-bleaching value. For single band FRAP, the average pixel intensity of the bleached area and of the whole nucleus were determined for a series of 10 pre-bleach and 250 post-bleach images and corrected for background fluorescence. Recovery curves were double exponential fitted and the half-time of fluorescence recovery (t_½) and mobile fraction were determined using EasyFRAP online. The intensity of GFP-H2B was measured across 10 nuclei, nine confocal slices per nucleus. Brightness was auto-adjusted using Fuji/imageJ software (https://imagej.net↗). GFP-H2B pixel intensities were binned into four classes, corresponding to background, white, gray, and black chromatin, using the following tresholds: background 0–82; Interband/white chromatin, 83–148; (Low-intensity bands/gray chromatin 149–224) and (high-intensity/black chromatin >224). For raw data see Figure 1—source data 1.

Freshly isolated salivary glands were transferred to a glass multiwall plate and permeabilized essentially as described previously (Myohara and Okada, 1987). Permeabilization was started by replacing the medium with MTB1 buffer (15 mM KH₂PO₄; 50 mM KCl, 15 mM NaCl, 7.5 mM MgCl₂, 1% PEG 6000, pH = 7.0, KOH) containing 0.01% digitonin. After 20 min of permeabilization and three subsequent washes with MTB1 buffer, the glands were transferred to a glass Sigma cloning ring containing MTB1 buffer mounted on the coverslip of the incubation chamber. Confocal imaging was performed in MTB1 medium as described above. For immunocytochemistry, the permeabilized glands were washed with MTB1 buffer (three times), incubated for 5 min in MTB1 either in the absence or presence of ATP (10 mM), and subsequently fixed and processed for immunocytochemistry as described.

Immunolocalization of proteins on Drosophila salivary gland polytene chromosome spreads was performed essentially as described previously (Mohrmann et al., 2004), using the indicated primary antibodies. Slides were mounted in mounting medium containing 4’,6’-diamidino-2-phenylindole (DAPI) counter stain (Vector Laboratories). For preparations of polytene chromosomes in the presence of ecdysone hormone, dissected salivary glands were incubated for 1 hr in Grace’s medium (11605–045 Invitrogen), diluted 5:1 with H₂O, containing 20 μM 20-hydroxyecdysone. Endogenous proteins in intact, dissected salivary glands were immunolocalized by incubating the glands in MTB1 buffer (15 mM K₂HPO₄, 50 mM KCl, 15 mM NaCl, 7.5 mM MgCl2, 1% PEG6000, pH7.0 with KOH) in the presence or absence of digitonin 0.01% or 10 mM ATP. Glands were then washed three times with MTB1 buffer before fixation with 3.7% formaldehyde in TBST (50 mM Tris-HCl pH8, 150 mM NaCl, 0.1% Triton X-100) at room temperature for 10 min. Preparations were blocked with TBST containing 5% FCS. Next, they were incubated with primary antibodies, diluted in TBST containing 1% BSA, overnight at 4°C. Following multiple washes with TBST buffer, they were incubated with secondary Alexa Fluor antibodies, washed again and finally mounted in Vectorshield mounting medium (Vector Laboratories) containing DAPI. A list of antibodies used is provided in Supplementary file 3.

Immunoprecipitations and immunoblotting experiments were performed using standard methods (Chalkley and Verrijzer, 2004; Chalkley et al., 2008; Harlow and Lane, 1998). All procedures were on ice or at 4°C. S2 cells expressing either GFP-BRM or GFP-BRM-K804R were harvested 3 days after the addition of CuSO4 to a final concentration of 500 µM. Following a freeze-thaw step, whole cell extracts were prepared in HEMG/150: 25 mM HEPES-KOH pH7.6, 0.1 mM EDTA, 12.5 mM MgCl2, 10% glycerol, containing 150 mM KCl, 0.5% NP40 and a cocktail of protease inhibitors: 1 µM pepstatin, 1 µM aprotinin, 1 µM leupeptin, 0.2 mM 4-(2-Aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF). After rotation for 20–30 min at 4°C, the extracts were sheared through an insulin syringe. Extracts were then diluted with HEMG/150 to obtain a final NP40 concentration of 0.1%, and then clarified by centrifugation. For each immunoprecipitation, 20–25 μl of Chromotek GFP-trapA beads were used for 1–4 mg total protein. Extracts were incubated with beads for 1 hr at 4°C. The beads were then washed extensively with the following buffers: HEMG/100/0.1% NP40; HEMG/600/0.1% NP40; HEMG/100/0.01% NP40 and finally with HEMG/100 without NP40. Co-immunoprecipitation reactions of proteins from Drosophila salivary glands were carried out as follows: Drosophila salivary glands were dissected and collected in PBS. Following collection, PBS was removed and replaced with urea extraction buffer (HEMG/150 mM KCl, containing 0.1% NP40, 1M urea and 1 mM DTT) at a ratio of ~1 μl urea extraction buffer per pair of glands. Collected glands were snap frozen in liquid nitrogen. Once sufficient glands were collected, samples were pooled and next crushed with an eppie-pestle, followed by shearing by passage through an insulin needle. Extracts were then clarified by centrifugation. For immunoprecipitation assays, the equivalent of ~300 dissected salivary glands was incubated with 100 µg of anti-MOR antibodies cross-linked to Protein A-Sepharose beads (or control beads coated with pre-immune serum) for 2–3 hr at 4°C. Beads were washed sequentially with 2x HEMG/500/0.1% NP40 and 1x HEMG/100/0.01% NP40; all buffers contained protease inhibitors. Bound proteins were eluted by boiling for 1 min in SDS-loading buffer. Proteins were resolved by SDS-PAGE followed by immunoblotting. Each lane on the IP-Western analysis represents the equivalent of ~30 animals.

For mass spectrometric analysis, proteins were on-bead subjected to reduction with dithiothreitol, alkylation with iodoacetamide and digested with trypsin (sequencing grade; Promega). Nanoflow liquid chromatography tandem mass spectrometry (nLC-MS/MS) was performed on an EASY-nLC coupled to an Orbitrap Fusion Tribid mass spectrometer (Thermo), operating in positive mode. Peptides were separated on a ReproSil-C18 reversed-phase column (Dr Maisch; 15 cm × 50 μm) using a linear gradient of 0–80% acetonitrile (in 0.1% formic acid) during 90 min at a rate of 200 nl/min. The elution was directly sprayed into the electrospray ionization (ESI) source of the mass spectrometer. Spectra were acquired in continuum mode; fragmentation of the peptides was performed in data-dependent mode by HCD. Raw mass spectrometry data were analyzed with the MaxQuant software suite (Cox et al., 2009; version 1.6.7.0) with the additional options ‘LFQ’ and ‘iBAQ’ selected. A false discovery rate of 0.01 for proteins and peptides and a minimum peptide length of 7 amino acids were set. The Andromeda search engine was used to search the MS/MS spectra against the Uniprot database (taxonomy: Drosophila melanogaster, release January 2019) concatenated with the reversed versions of all sequences. A maximum of two missed cleavages was allowed. The peptide tolerance was set to 10 ppm and the fragment ion tolerance was set to 0.6 Da for HCD spectra. The enzyme specificity was set to trypsin and cysteine carbamidomethylation was set as a fixed modification, while STY phosphorylation was set as a variable modification. Both the PSM and protein FDR were set to 0.01. In case the identified peptides of two proteins were the same or the identified peptides of one protein included all peptides of another protein, these proteins were combined by MaxQuant and reported as one protein group. Before further statistical analysis, known contaminants and reverse hits were removed. To determine the relative abundances of proteins within a sample, the iBAQ intensities were compared.

Statistical parameters including the value of n, % recovery, t_1/2 and mobile fraction (mean +/- SD) are indicated in the figures or figure legends. Data is judged to be statistically significant when p < 0.05 by two-tailed Student’s t test and is indicated by an asterisk in the figures.