Phosphorylation of histone H3 at Ser10 facilitates RNA polymerase II release from promoter-proximal pausing in Drosophila

JIL-1 associates with actively transcribed regions on polytene chromosomes

A correlation exists between active transcription in Drosophila and H3S10 phosphorylation, and both events take place specifically in most interband regions (less-condensed chromatin) on salivary gland polytene chromosomes. Upon heat-shock treatment, H3 becomes phosphorylated at Ser10 at the highly transcribed heat-shock loci, and it becomes dephosphorylated at genes that were previously active and are now repressed (Nowak and Corces 2000). Since JIL-1 phosphorylates histone H3 in vitro and H3S10 phosphorylation is dramatically reduced in JIL-1-null mutants (Wang et al. 2001), we tested whether the JIL-1 kinase is required for transcription. To determine whether JIL-1 associates with actively transcribed regions, we compared its distribution on polytene chromosomes with that of both hypophosphorylated and hyperphosphorylated forms of Pol II. We performed immunofluorescence experiments using polytene chromosomes from third instar larvae, costaining with α-JIL-1 antisera as well as 8WG16, H5, or H14 antibodies directed against Pol IIa, Pol IIo^ser2, and Pol IIo^ser5, respectively. The results reflect a broad distribution of JIL-1 throughout euchromatin, localizing to most interband regions, with an extensive overlap between JIL-1 and both phosphorylated forms of Pol II (Fig. 1A,B). Although the relative levels of JIL-1 and Pol II vary from site to site, a close examination shows that JIL-1 is found at ∼95% of sites associated with either phosphorylated form of Pol II, examples of which are indicated by arrows in Figure 1, A and B; the extent of overlap can be better appreciated in close-up views of the chromosomes (Fig. 1C,F). JIL-1 also overlaps significantly with Pol IIa, albeit to a lesser extent; only ∼60% of sites containing Pol IIa also contain JIL-1 (Fig. 1D, arrows indicate sites where there is no overlap, E shows a detailed view).

JIL-1 is required for transcription of ecdysone-induced genes

To further analyze the possible role of JIL-1 in the activation of transcription, we examined the chromosomal localization of JIL-1, Pol IIo^ser2, and phosphorylated H3S10 in developmental puffs of third instar larvae polytene chromosomes. The presence of these puffs reflects high levels of transcription regulated by ecdysone hormone. It has been previously demonstrated that Pol IIo^ser2 localizes to these regions at discrete times during development (Weeks et al. 1993). Later studies also showed phosphorylated H3S10 at these puffs, suggesting that this histone modification may be required for ecdysone-induced transcription (Nowak and Corces 2000). Our results show that JIL-1 colocalizes with Pol IIo^ser2 and phosphorylated H3S10 at prominent early ecdysone-induced puffs, such as 75B and 74EF (Fig. 1F,G). The presence of JIL-1 in the ecdysone-induced puffs led us to examine whether JIL-1 was necessary for ecdysone-induced transcription. We analyzed the expression levels of E75A, one of the isoforms encoded by the early gene E75, at the late prepupae stage. At this stage, a high titer of ecdysone triggers the transition from prepupae to pupae, accompanied by a peak in the expression of the early genes (Thummel 1996). Equal amounts of total RNA isolated from wild-type and JIL-1^z2 mutant prepupae were examined by Northern analysis using a probe specific for the E75A mRNA. We observed high levels of E75A expression in wild-type prepupae and a reduction in E75A expression in JIL-1^z2 mutants (Fig. 2A), suggesting that JIL-1 is required for transcription of the ecdysone genes.

JIL-1 is required for transcription of the heat-shock genes

Histone H3 becomes phosphorylated at Ser10 concomitantly with transcription activation of the heat-shock genes. We therefore asked whether JIL-1 is involved in the transcription activation of the heat-shock genes by examining whether JIL-1 localizes to the heat-shock puffs of polytene chromosomes upon thermal induction. Using α-JIL-1 antisera, we performed immunofluorescence microscopy on polytene chromosomes from wild-type third instar larvae salivary glands subjected to 25 min of heat shock. JIL-1 localization at the heat-shock loci was revealed by a robust staining at several heat-shock puffs, most prominently at the hsp70 genes (chromosomal subdivisions 87A/C) and at the 93D and 95D puffs (Fig. 3A). However, unlike phosphorylated H3S10, JIL-1 can still be detected at the previously transcribed genes that are repressed after heat shock. To confirm that JIL-1 recruitment is transcription dependent, we examined whether JIL-1 is present at the hsp70 genes under non-heat-shock conditions. Drosophila has six nearly identical genes that encode Hsp70, the major heat-shock protein. Two are located at division 87A and four are located at 87C on the right arm of chromosome 3 (Gong and Golic 2004). Anti-JIL-1 staining combined with in situ hybridization to a portion of the hsp70 transcription unit shows no colocalization between JIL-1 and the hsp70 genes at the 87C locus in the absence of heat shock (Fig. 3B). Nevertheless, JIL-1 is present at very low levels at the 87A locus. This could be explained by the fact that the hsp70 genes at 87A are surrounded by actively transcribed genes that cannot be distinguished from hsp70 at the level of resolution provided by polytene chromosomes. This is supported by the presence of low levels of Pol IIo^ser2 at 87A but not 87C under non-heat-shock conditions (Fig. 3C). These results suggest that JIL-1 is recruited to the heat-shock loci in a transcription-dependent manner.

To confirm that JIL-1 is required for the phosphorylation of H3S10 at the hsp70 loci upon heat shock, we tested whether this modification is JIL-1 dependent. Antibodies to phosphorylated H3S10 were used for immunofluorescence microscopy combined with in situ hybridization on polytene chromosomes. Before heat shock in wild-type chromosomes, phosphorylated H3S10 cannot be detected at the 87C locus while low levels are observed at 87A (Fig. 3D), consistent with the localization of both JIL-1 and Pol IIo^ser2. After heat shock, an intense signal can be detected at both loci in wild-type (Fig. 3E) but is severely reduced in chromosomes from JIL-1-null mutants (Fig. 3F). To further confirm this conclusion, we examined the expression of the hsp70 genes in wild-type and JIL-1^z2 mutants before and after heat shock by Northern analysis. A robust increase in the levels of hsp70 mRNA is detected in wild-type larvae after heat shock, while no hsp70 transcripts are detected in JIL-1-null mutants under the same conditions (Fig. 2B). These results suggest that JIL-1-mediated H3 phosphorylation is required for transcription of the hsp70 genes.

JIL-1 is not required for the recruitment of the HSF transcription factor

Having established that JIL-1 is required for transcription of the hsp70 genes, we sought to place H3S10 phosphorylation in the cascade of events leading from transcription factor recruitment to transcription elongation. It is conceivable that the overall disruption of chromosome structure in JIL-1 mutants (Wang et al. 2001) prevents the binding of the heat-shock transcription factor HSF to its response element upon gene induction. We therefore examined the recruitment of HSF to polytene chromosomes after heat shock in wild-type and JIL-1^z2 mutant larvae. When inactive, HSF is diffusely distributed throughout the chromosomes at very low levels, including at the heat-shock genes (Fig. 4A). After 25 min of heat shock, HSF redistributes to discrete sites, most prominently at the heat-shock loci, as can be visualized by immunofluorescence microscopy of polytene chromosomes using anti-HSF antibodies (Fig. 4A; Westwood et al. 1991). Mutation of JIL-1 has no effect on this relocalization, as HSF can be observed to bind polytene chromosomes of JIL-1^z2 mutants despite their poorly organized chromatin (Fig. 4B). This binding appears to be very specific, as suggested by the strong signal at the 87A and 87C loci and by the lack of nonspecific binding before heat shock (Fig. 4B). These results indicate that the disrupted chromatin organization of chromosomes in JIL-1 mutant larvae does not preclude transcription factor binding, suggesting that JIL-1 acts to promote a transcription step downstream from transcription factor recruitment.

Transcription initiation takes place at the hsp70 genes in JIL-1 mutants, but elongation is impaired

The heat-shock genes accumulate a hypophosphorylated form of promoter-paused Pol II at a location 20–40 bp downstream from their transcription start sites prior to induction. Immediately after induction and recruitment of HSF, levels of Pol IIo^ser5 rapidly increase, especially at the promoter-proximal region of the gene, indicating promoter clearance. Soon thereafter, Pol IIo^ser2 can be detected at the transcription start site and throughout the reading frame, marking transcription elongation (Boehm et al. 2003). In order to determine the stage of the transcription process at which JIL-1 acts, we first examined the localization of Pol IIo^ser5 at the hsp70 loci of wild-type and JIL-1^z2 mutants using antibody H14, which is specific for this form of Pol II. Anti-HSF antibodies were also used to mark the heat-shock puffs and as an internal control of signal level. Pol IIo^ser5 localizes to the 87A and 87C loci of wild-type chromosomes after 25 min of heat-shock treatment (Fig. 4C). Similarly, JIL-1^z2 mutants display high levels of Pol IIo^ser5, suggesting that transcription initiation is not affected and that elongation might be defective in these mutants (Fig. 4D). To test this possibility, we carried out immunofluorescence analysis using the H5 antibody, which recognizes Pol IIo^ser2. In contrast to HSF and Pol IIo^ser5, levels of Pol IIo^ser2 are dramatically reduced in JIL^z2 mutants (Fig. 4F) as compared with wild type (Fig. 4E).

JIL-1 is required for P-TEFb recruitment during promoter-proximal pausing

The results so far suggest a requirement for H3S10 phosphorylation by JIL-1 during transcription elongation of the hsp70 genes. The low levels of Pol IIo^ser2 observed at 87A/C could be explained by two scenarios. It is possible that H3S10 phosphorylation is necessary for the recruitment of the kinase responsible for phosphorylation of the CTD at Ser2. If this is the case, the polymerase should not become phosphorylated in JIL-1-null mutants and would not be released from its pause. A second possibility is that the chromatin structure in JIL-1 mutants interferes with elongation. In this case, the polymerase could become phosphorylated, but disengages shortly after being released.

The P-TEFb complex, a heterodimer of the CDK9 kinase and cyclin T (CycT), is responsible for the phosphorylation of the CTD at Ser2. This complex is rapidly recruited upon heat shock and acts to release Pol II from promoter-proximal pausing by antagonizing the effects of the pausing factors DSIF and NELF. Immunofluorescence analysis using anti-CycT antibodies has previously detected high levels of P-TEFb at 87A/C after heat shock (Lis et al. 2000). To obtain a better understanding of the role of JIL-1 and to narrow down the window during which it acts, we performed similar analyses and compared levels of CycT at 87A/C in wild-type and JIL-1^z2 mutant larvae. Antibodies against Pol IIo^ser5 were used to mark the puffs and as an internal control for signal intensity. A dramatic decrease in CycT binding at heat-shock genes was observed in JIL-1 mutants, as can be seen by comparing Figure 5, B and A. Quantitative RT–PCR analyses revealed normal transcript levels of CycT in these larvae as compared with wild type, suggesting that this observation is not due to indirect effects on CycT expression (data not shown). We then performed the same comparative analyses using antibodies against SPT5, a component of the pausing factor DSIF. Unlike the case with P-TEFb, we observed no changes in SPT5 binding in JIL-1^z2 mutants (Fig. 5D) when compared with wild type (Fig. 5C). These data suggest that JIL-1 acts prior to P-TEFb recruitment.

JIL-1 is required for transcription of most Drosophila genes

JIL-1 localizes to all interbands on polytene chromosomes and colocalizes extensively with RNA polymerase throughout the genome, suggesting that JIL-1 may play a role in the transcription of the vast majority of Drosophila genes (Fig. 1A). The results from the previous experiments suggest that JIL-1 is required for proper elongation but not initiation of the hsp70 genes. To examine whether this role is conserved throughout the genome, we performed Western analyses of protein extracts from wild-type and JIL-1^z2 mutant larvae and compared total levels of the two phosphorylated forms of Pol II. Total Pol IIo^ser2 levels are significantly reduced in JIL-1^z2 mutants as compared with wild type (Fig. 6A). In contrast, total Pol IIo^ser5 levels are maintained in these mutants at wild-type levels (Fig. 6B). Furthermore, levels of trimethylated H3K4—a marker of a chromatin environment that promotes transcription initiation—are also maintained in JIL-1^z2 mutants (Fig. 6C). Taken together, these results suggest a general role for JIL-1 in early elongation of transcription.

Histone phosphorylation requires the Brahma chromatin remodeling complex

To more precisely place JIL-1-mediated H3S10 phosphorylation into currently understood steps of the transcription process, we examined the presence of phosphorylated H3S10 in mutants that abrogate transcription at two distinct steps by disrupting key components of chromatin remodeling complexes. The brahma (brm) gene encodes an ATPase that is part of the SWI/SNF remodeling complex. When a dominant-negative form of BRM (UASbrm^K804R) is expressed in salivary glands, levels of Pol IIo^ser2 and Pol IIa are dramatically reduced, suggesting that chromatin remodeling by the BRM complex is required at an early step in transcription initiation (Armstrong et al. 2002). A synergistic genetic interaction between JIL-1^z2 and mutations in the trx-G genes brm and trithorax (trx) has been reported previously (Zhang et al. 2003), suggesting that JIL-1 functions in the same pathway. We compared phosphorylated H3S10 levels on polytene chromosomes derived from wild-type larvae (Fig. 7A) with those derived from salivary glands that express brm^K804R (Fig. 7B). The loss of normal BRM function dramatically reduces levels of phosphorylated H3S10 while levels of Su(Hw), a chromatin-associated protein used as a control, are maintained. Western blot analysis performed on protein samples derived from wild-type versus brm^K804R salivary glands using anti-H3pS10 antibodies further shows that overall levels of phosphorylated H3S10 are reduced in brm mutants (Fig. 6D). These results are consistent with a role for JIL-1-mediated H3S10 phosphorylation at a later step in transcription and suggest that the modification is dependent on chromatin remodeling by the BRM complex.

Histone H3S10 phosphorylation is not affected in kismet mutants

Kismet-L (KIS-L) is another transcriptional regulator that exhibits trx-G phenotypes. It contains an ATPase domain that is highly related to that of BRM and has been proposed to act as the catalytic subunit of a novel chromatin remodeling complex required downstream from BRM. Disruption of the kismet gene (kis^k13416) leads to reduction in the elongating form of Pol II but not that of the initiating form, suggesting that KIS-L is required in an early step in transcription elongation (Srinivasan et al. 2005). To further characterize the step at which JIL-1 acts during transcription elongation, we examined the levels of phosphorylated H3S10 in kis^k13416 mutants. Unlike the brm^K804R mutant, kis^k13416 had no effect on H3S10 phosphorylation as visualized by immunofluorescence microscopy (Fig. 7C) when compared with wild type (Fig. 7A). These results were confirmed by Western blot analysis (Fig. 6E). These observations suggest either that H3S10 phosphorylation by JIL-1 acts upstream of KIS function or that the two chromatin regulators function in independent pathways.

Drosophila stocks

Stocks were maintained in standard medium at 23°C. Oregon R larvae were used for wild-type controls in all experiments. The JIL-1^z2 stock was a gift from Dr. K. Johansen (Iowa State University, Ames, IA). The brm^K804R and kis^k13416 stocks were a gift from Dr. J. Tamkun (University of California at Santa Cruz, Santa Cruz, CA).

Induction of the heat-shock response and analysis of polytene chromosomes

Salivary gland polytene chromosome squashes were prepared from wandering third instar larvae maintained at 23°C. For heat-shock experiments, third instar wild-type and JIL-1^z2 mutant larvae were subjected to heat-shock treatment as described previously (Nowak et al. 2003). Salivary glands were dissected in 0.7% NaCl and fixed for 2 min in 45% acetic acid and 1.85% formaldehyde. Fixed salivary glands were subsequently squashed in 45% acetic acid on subbed slides. The slides were frozen in liquid nitrogen and stored dry at −70°C. For immunostaining of H3S10 phosphorylation, salivary glands were fixed for 2 min in 15% acetic acid and 1.85% formaldehyde. Slides were incubated overnight at 4°C in antibody dilution buffer (PBS, 0.1% Triton X-100, 1% BSA) containing primary antibodies at concentrations of 1:30 chicken α-JIL-1 (gift from Dr. K. Johansen), 1:20 rabbit α-H3pS10 (Upstate Biotechnology), 1:30 mouse α-Pol IIo^ser2 (H5, Covance), 1:30 mouse α-Pol IIo^ser5 (H14, Covance), 1:50 rabbit α-HSF (gift from Dr. C. Wu, National Cancer Institute, Bethesda, MD), 1:50 rabbit α-SPT5 (gift from Dr. F. Winston, Harvard Medical School, Boston, MA), 1:20 rabbit α-CycT (gift from Dr. D. Price, University of Iowa, Iowa City, IA; affinity-purified as described in Lis et al. 2000), and 1:150 rat α-Su(Hw). Following incubation, slides were washed three times in PBS and 0.1% Triton X-100 and incubated for 1 h at 37°C in the appropriate secondary antibody (Jackson ImmunoResearch Laboratories) diluted 1:200 in antibody dilution buffer. Slides were washed three times as described above, stained with 0.5 μg/mL 4′,6-diaminidine-2-phenylindole (DAPI), and mounted in Vectashield mounting medium (Vector Laboratories) for viewing.

Preparation of Drosophila protein extracts and Western analysis

Protein extracts from third instar larvae were prepared by lysing the anterior tissues, containing salivary glands, brains, and imaginal discs, in sodium dodecyl sulfate (SDS) sample buffer (0.175 M Tris-HCl, 5% SDS, 15% glycerol, 0.0005% bromophenol blue, 4.95 mM β-mercaptoethanol) using a Dounce homogenizer. Following homogenization, protein samples were immediately run on SDS–polyacrylamide gels and transferred to polyvinylidene difluoride (PVDF) membranes for immunodetection. Membranes were incubated overnight at 4°C in antibody dilution buffer (PBS, 0.05% Tween, 5% milk) containing primary antibodies at concentrations of 1:100 mouse α-DDA2.7 EcR common (Developmental Studies Hybridoma Bank), 1:1000 mouse α-Pol IIo^ser2 (H5, Covance), 1:500 mouse α-Pol IIo^ser5 (H14, Covance), 1:1000 rabbit α-H3pS10 (Upstate Biotechnology), 1:1000 rabbit α-Lys4 trimethylated histone H3 (Upstate Biotechnology), 1:1000 rabbit α-histone H2A (Upstate Biotechnology), and 1:1000 mouse α-lamin (Risau et al. 1981). After incubation, the membranes were washed twice with PBS and 0.25% Tween, incubated for 1 h at room temperature in the appropriate HRP secondary antibody (Jackson ImmunoResearch Laboratories), and washed twice with PBS and 0.25% Tween. Antibody signal was visualized using chemiluminescence detection methods (SuperSignal West Pico kit, Pierce).

Preparation of Drosophila RNA and Northern analysis

Total RNA was isolated from 20 wild-type or JIL-1^z2 third instar larvae and prepupae using RNA-Bee (Tel-Test, Inc.) according to the manufacturer’s instructions. RNA samples (20 μg per lane) were denatured in MOPS/formaldehyde/formamide buffer and separated by electrophoresis on a 1% formaldehyde/agarose gel at low voltage. Following electrophoresis, RNA was transferred to a Nytran supercharge membrane (Scheleicher and Schuell) by capillary blotting overnight in 10× SSC. Resolved RNAs were cross-linked to the membrane by UV irradiation and incubated in hybridization solution (Ultrahyb, Ambion) for 1 h at 45°C before addition of the probe. Radiolabeled DNA probes were prepared using the Prime-A-Gene random priming kit (Promega). Labeled probes were purified using MicroSpin G-50 columns (Amersham), added to the hybridization mixture, and incubated overnight at 45°C. The blots were then washed using solutions containing decreasing concentrations of SDS and SSC (1× SSC is 0.15 M NaCl, 0.015M sodium citrate). Transcripts were visualized by autoradiography or PhosphorImager. All probes were generated from amplified genomic Drosophila DNA. A probe specific for hsp70 mRNA was generated using the following pair of primers: Hsp70 IIIA (5′-TGGTGCTGAC CAAGATGAAG-3′) and Hsp70 IIIB (5′-TAGTCTGCTTGCAC GGAATG-3′). A probe specific for the E75A mRNA isoform was generated using the following pair of primers: E75A-1 (5′-CATTGACTAACTGCCACTCGCA-3′) and E75A-2 (5′-CAA CACTGCAGTGGGACCATCG-3′). A probe against rp49 mRNA was synthesized using primers rp49 I (5′-CCCAAGGGTATC GACAACAG-3′) and rp49 II (5′-ATGGTGCTGCTATCCC AATC-3′). Membranes used to analyze expression of the E75A and hsp70 genes were stripped in 1% SDS boiling solution for 30 min and rehybridized with rp49 probe as a loading control.

DNA in situ hybridization

Digoxigenin-labeled DNA probes were prepared using the Prime-A-Gene random priming kit using a mixture of deoxyribonucleotides containing DIG-11-dUTP (Roche). The reactions were incubated overnight at room temperature. The template used to generate the probe was synthesized as described above. Labeled probes were precipitated and stored in hybridization buffer (4× SSC, 1× Denhardt’s, 0.4 mg/mL sonicated salmon sperm, 50% formamide) until ready for use. Salivary glands were dissected in 0.7% NaCl, transferred for 30 sec to 45% acetic acid, and then fixed for 2 min in 1 vol lactic acid/2 vol water/3 vol acetic acid. Fixed salivary glands were squashed as described above, and the slides were left overnight at room temperature. Slides were then frozen in liquid nitrogen, the coverslips were removed, and the slides were dehydrated in 70% ethanol for 5 min and incubated twice 5 min in 95% ethanol. The slides were stored at room temperature until ready for use. Slides were incubated in 2× SSC for 60 min at 65°C, dehydrated through an ethanol series, air-dried, and denatured for 2 min in 70 mM NaOH. Following denaturation, the slides were dehydrated again as described above and air-dried at room temperature prior to hybridization. For hybridization, boiled probes were added to the dried, denatured polytene chromosome squashes and covered with a coverslip. The slides and coverslips were sealed with rubber cement and incubated overnight at 37°C in a humidified chamber. Following hybridization, coverslips were removed and the slides were washed twice for 10 min at 37°C in 2× SSC, twice for 5 min at room temperature in 1× PBS, 2 min in PBS/0.1% Triton X-100, and 5 min in 1× PBS. After washing, immunocytochemical treatment was carried out as described above.