Biosynthetic Enzyme GMP Synthetase Cooperates with Ubiquitin-Specific Protease 7 in Transcriptional Regulation of Ecdysteroid _target Genes^▿

Fly strain, genetics, and DNA constructs.

Fly stock maintenance and crosses were performed using standard procedures. Usp^Δ35 was generated by imprecise excision of the P element in P[SUPor-P]Usp7^KG06814 (29). Genomic sequence analysis revealed 24 bp of the P element remaining and loss of a part of the 5′ untranslated region (UTR) and the coding sequence for its first 23 amino acids. P[EP]Bre1^kim1 and Bre1^RNAi were purchased from GenExel (strain GE22117) and VDRC (strain 15620) (9), respectively. Gmps^RNAi was obtained from the National Institute of Genetics Fly Stock Center (strain 9242R-3). To generate UAS-Usp7 and UAS-Usp7^C250A transgenic lines, cDNAs encoding full-length USP7, USP7^C250A, GMPS^S242L, and GMPS^C95A were cloned into pUAST and verified by sequencing. P element-mediated germ line transformation was performed according to standard procedures. The Gmps^e10 and UAS-Gmps lines were a kind gift of Yong Rao (19). Gmps^K07130 has been described previously (29). EcR^V559fs and EcR^A483T were acquired from the Bloomington stock center. UAS-Usp7#J2 and UAS-Usp7^C250A#C4, both integrated on the second chromosome, were each combined with UAS-Gmps#2 (UAS-bur#2 in reference 19) and crossed with a GMR-Gal4 driver line. A stable GMR>Usp7/GMR>Gmps line (#2091) was created by recombination of UAS-Usp7 (line J1, on the third chromosome) and UAS-Gmps (UAS-bur#1 in reference 19) and crossed with a GMR-Gal4 driver line (on the second chromosome). Crosses were performed at 25°C. Details will be provided upon request.

Protein-protein interaction assays.

Recombinant glutathione S-transferase (GST) fusion proteins were expressed in Escherichia coli BL21-CodonPlus(DE3)-RIL and purified by glutathione Sepharose 4 fast flow (GE Healthcare) chromatography according to standard procedures (8). ³⁵S-labeled proteins were expressed using TNT coupled rabbit reticulocyte lysates (Promega). Prior to binding reactions, beads were blocked with 3% fetal calf serum in HEMG buffer (25 mM HEPES-KOH [pH 7.6], 0.1 mM EDTA, 12.5 mM MgCl₂, 10% glycerol) containing 1 mM dithiothreitol, 0.2 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride, 1 μg/ml leupeptin, 1 μg/ml aprotinin, 1 μg/ml pepstatin A, 150 mM KCl (making HEMG/150), and 0.1% NP-40. Binding reactions were in the same buffer but lacking serum. Following three washes with HEMG/300/0.1% NP-40 and two with HEMG/150/0.01% NP-40, bound proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by autoradiography.

Immunological procedures.

For antibody production, GMPS (amino acids 310 to 563 and 310 to 623) and BRE1 (amino acids 231 to 718) were expressed as GST fusion proteins, purified, and used for immunizations as previously described (8). Anti-USP7, anti-H1, and anti-H2B (PV57/58, raised against purified core histones and mainly recognizing H2B) have been previously published (13, 20, 29). To detect RNA polymerase (Pol) II, we used a mixture of monoclonal antibodies 8WG16, H5, and H14 (Covance). Anti-ubiquityl-histone H2B (clone 56) was from Millipore. Anti-EcR (DDA2.7 and Ag10.2), developed by C. Thummel and D. Hogness, was obtained from the Developmental Studies Hybridoma Bank (Department of Biological Sciences, National Institute of Child Health and Human Development, University of Iowa, Iowa City). Analysis of polytene chromosomes was done as described previously (21). For analysis of polytene chromosomes upon knockdown of GMPS, Gmps^RNAi was crossed with an actin driver line. To analyze H2Bub levels, 10 third-instar larvae were lysed in SDS-gel loading buffer using an Eppendorf tube pestle.

Gene expression analyses.

For prepupal analysis, the Usp7^kim1, Usp7^KG06814, and Gmps^K07130 mutant lines were rebalanced with green fluorescent protein (GFP)-marked balancer chromosomes. GFP-negative mutant prepupae and wild-type (WT) prepupae were collected at 2-h intervals from the moment of pupariation (t = 0) for 12 h. RNA was extracted using TRIzol. S2 cells were cultured and treated with double-stranded RNA (dsRNA) as described previously (33). After 3 days, dsRNA was reapplied, and after 5 days, cells were harvested and RNA was extracted using TRIzol. Knockdowns were performed in biological triplicates. Knockdown efficiency was monitored in parallel by immunoblotting. 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^−ΔΔC_T method (17). CG11874 was used as an internal control mRNA.

The RT-qPCR primers used were as follows: CG11874, 5′-AGTGTTGCTCTGCCTAAGTGG-3′ and 5′-CGGATGATGGTGCGGATTGG-3′; GAPDH, 5′-TGCTGGAGCCGAGTATGTGG-3′ and 5′-GCCGAGATGATGACCTTCTTGG-3′; ImpL3, 5′-ATCGGCAGCGGCACCAAC-3′ and 5′-CGGCAATGTTCACTCCAGACC-3′; ImpE2, 5′-ACTTCCTGGGCGGCAATCG-3′ and 5′-CTACGGCTGGCTTCTCTGTGG-3′; Ftz-F1, 5′-ATGGAAGGCGAACGAAGGATAC-3′ and 5′-AGTTGGTGGTAGTAGTGATGATGC-3′; E74A, 5′-GATGGTCGTCTTGTTGGAGGTC-3′ and 5′-TGCGGGTTGTTCGGATTGC-3′; E75A, 5′-CCTTTCATTGACTAACTGCCACTC-3′ and 5′-CGAAACGAAACGAACGGAACG-3′; E75C, 5′-CGGCTCGGAAGTTTGTGGTTAG-3′ and 5′-GCTGATGCTGCTGCTGATGC-3′; CG1381, 5′-ACACCCGAACAGGCGAGAATCC-3′ and 5′-TCATCATCGTCGTCGTTGTCATCC-3′.

GMPS and USP7 interact genetically.

Previously, we presented biochemical evidence revealing that parts of GMPS and USP7 associate in a highly stable complex, mediating epigenetic gene silencing (29). Here, we set out to explore their cooperation in vivo during developmental gene control. To facilitate the genetic analysis of USP7, which is encoded by an essential X-chromosomal gene, we created new alleles by mobilizing the P element in the Usp7^KG06814 mutant strain. One of the derived alleles, Usp7^Δ35, turned out to be a hemizygous viable hypomorph. Usp7^Δ35 lacks part of the USP7 5′ UTR and the coding sequence for its first 23 amino acids and expresses USP7 at a reduced level. Hemizygous Usp7^Δ35 males, but not revertants, generated by precise P element excision, displayed a “held-out wing” phenotype (Fig. 1A to C). This phenotype is convenient for genetic interaction assays. Indeed, we found that the “held-out wing” phenotype was strongly enhanced when Usp7^Δ35 was combined with either a heterozygous Gmps^e10 or Gmps^K07130 allele, which by itself did not affect wing positioning (Fig. 1D to G). These results showed that USP7 and GMPS interact not only biochemically but also genetically.

To complement these loss-of-function analyses with an in vivo ectopic expression assay, we employed the GAL4/UAS system (6). We used the glass multimer reporter (GMR) to drive GAL4 expression in the developing eye, resulting in overexpression of USP7 or GMPS under UAS control. Concomitant ectopic expression of USP7 and GMPS (GMR>Usp7/GMR>Gmps) caused clear defects in eye development characterized by disorganized or missing ommatidia, loss of bristles, and black necrotic patches (Fig. 1I). When either only USP7 (Fig. 1J) or only GMPS (Fig. 1K) was expressed ectopically, we observed no eye defects. Thus, the phenotype was strictly dependent on the overexpression of both GMPS and USP7. Next we wondered whether the enzymatic activities of USP7 or GMPS were required for the distorted eye phenotype. The phenotype of ectopic coexpression of USP7 with either a GMPS mutant defective in ATP hydrolysis (GMPS^S242L) or a GMPS mutant defective in glutamine hydrolysis (GMPS^C95A) was similar to that resulting from the coexpression of WT GMPS (Fig. 1L and M). Thus, in agreement with our earlier in vitro results (29), GMPS's enzymatic activity is not required for stimulation of USP7. In contrast, concomitant overexpression of GMPS and USP7^C250A, a deubiquitylation-defective substitution mutant, no longer affected eye development (Fig. 1N). As is the case for the WT proteins, ectopic overexpression of only GMPS^C95A, GMPS^S242L, or USP7^C250A had no effect on eye development. (Fig. 1O, P, and Q). We conclude that GMPS and USP7 cooperation is critically dependent on USP7's deubiquitylating activity but not on de novo GMP synthesis by GMPS.

GMPS/USP7 counteracts BRE1 by deubiquitylation of histone H2B.

Our earlier in vitro reconstitution experiments suggested that histone H2Bub is a _target for deubiquitylation by GMPS/USP7 (29). Because BRE1 is the E3 ubiquitin ligase that catalyzes H2B ubiquitylation, we wondered whether reduced BRE1 levels would modulate the GMPS/USP7 overexpression phenotype. To test this hypothesis, we combined a heterozygous Bre1^kim1 mutant allele with GMR-driven ectopic expression of GMPS and USP7. Note that the recombined GMR>Gmps/GMR>Usp7 line used in these assays has a milder eye phenotype (Fig. 2A) than the one used above (Fig. 1I; see Materials and Methods for details). The heterozygous Bre1^kim1 allele, when combined with GMPS/USP7 overexpression, led to strongly (∼50%) reduced viability, whereas the eye phenotype in surviving adults is strongly enhanced (Fig. 2B). The combination of Bre1^kim1 with overexpression of only USP7 or only GMPS had no effect on eye development, again emphasizing the dependence on cooperation between GMPS and USP7 (Fig. 2C and D). Development of Bre1^kim1 heterozygotes was normal (Fig. 2E). When we used the GMR-GAL4 driver for ectopic GMPS and USP7 expression and to express an interfering RNA _targeting the Bre1 mRNA, we also observed an enhanced phenotype (Fig. 2F). Importantly, the GMR>Bre1^RNAi line by itself displayed no defective eye development (Fig. 2G). When GMR>Bre1^RNAi was combined with either GMR>Usp7 or GMR>Gmps alone, we observed no phenotype (data not shown). Collectively, these results strongly support the notion that GMPS and USP7 cooperate in vivo to counteract the activity of ubiquitin ligase BRE1.

The ability of GMPS/USP7 to deubiquitylate H2Bub in vitro (29) and the strong genetic interaction between GMPS/USP7 and BRE1 in vivo strongly support the relevance of histone H2Bub as a substrate. To provide further evidence for this notion, we determined H2Bub levels in Gmps^K07130 and Usp7^KG06814 mutants by immunoblot analysis of protein extracts made from third-instar larvae. Homozygous Usp7^KG06814 and Gmps^K07130 larvae have strongly reduced levels of USP7 and GMPS, respectively (Fig. 3A). We observed a clear increase in bulk H2Bub levels in homozygous Usp7^KG06814 larvae compared to WT or heterozygous larvae (Fig. 3B, compare lanes 5 and 6 with lanes 1 to 4). Note that to show that we are within a quantitative blotting range, we loaded two amounts of protein for each set. In support of the regulatory role of GMPS in the catalytic activity of USP7, we also observed an increase in H2Bub levels in homozygous Gmps^K07130 larvae (Fig. 3B, compare lanes 9 and 10 with lanes 1 and 2 and lanes 7 and 8). Collectively, our earlier results (29) and those presented here demonstrate that GMPS/USP7 deubiquitylates histone H2Bub.

GMPS and USP7 are required for ecdysteroid signaling in vivo.

To compare their genome-wide distribution, we used indirect immunofluorescence to determine the binding of GMPS and USP7 on third-instar larval salivary gland polytene chromosomes (Fig. 4A). USP7 and GMPS colocalize on many of their chromosomal binding sites, but each also occupies unique loci. Thus, although USP7 and GMPS can form a tight complex, they can also bind DNA independently. In agreement with this notion, immunodepletion experiments revealed that portions of USP7 and GMPS are not associated with each other (data not shown). Overall, USP7 does not colocalize with RNA Pol II, indicative of a role in gene silencing rather than activation (Fig. 4B). This is consistent with its genetic function as an enhancer of Polycomb silencing (29) and its biochemical activity: removal of the active H2Bub mark. Analysis of their binding pattern revealed that both GMPS and USP7 associate with the E74 and E75 loci harboring ecdysone early-response genes (7, 24) (Fig. 4C). We note that we only observed GMPS and USP7 colocalization in these regions prior to their induction (see below). RNA interference (RNAi)-mediated knockdown of GMPS does not affect USP7 binding to polytene chromosomes (Fig. 4D). These results suggest that the recruitment of USP7 to chromatin is independent of GMPS.

The binding of USP7 and GMPS to the E74 and E75 loci suggested a role in the transcriptional regulation of these gene clusters. Consistent with this notion, strong USP7 and GMPS mutants die at the prepupal or pupal stage, when ecdysone signaling is critical. Furthermore, Gmps and Usp7 mutant pupae frequently lack anterior spiracles, one of the hallmarks of defective ecdysteroid signaling during pupariation (data not shown). To test the role of USP7 and GMPS in ecdysone signaling, we compared the expression profiles of E74A and E75A in homozygous Usp7^KG06814, Usp7^kim1, and Gmps^K07130 prepupae to that in WT animals (Fig. 5A). Prepupae were collected at pupariation (t = 0), and RNA was extracted at 2-h intervals for 12 h and monitored by RT-qPCR. To facilitate comparison, E74A and E75A expression levels in Gmps or Usp7 mutants were also plotted relative to those in WT animals (Fig. 5B). In WT prepupae, E74A and E75A expression reflects the level of ecdysone signaling. Following a peak at pupariation, transcription ceases until, after ∼10 h, another ecdysone pulse strongly activates E74A and E75A transcription. Compared to WT animals, Usp7^kim1 prepupae overexpress E74A and E75A at early and late time points. In Gmps^K07130 and Usp7^KG06814 mutants, there is a striking derepression from t = 4 to 8 h. In contrast to Usp7^kim1 mutants, which harbor a weaker allele than Usp7^KG06814 and are homozygous viable, the development of Gmps^K07130 and Usp7^KG06814 mutants arrests prior to pupation and E74A and E75A expression ceases after t = 10 h. In summary, GMPS and USP7 bind ecdysone _target loci and are required for their normal regulation, most likely as transcriptional corepressors.

Binding dynamics at ecdysone _target loci.

We used indirect immunofluorescence of polytene chromosomes to monitor GMPS and USP7 binding to ecdysone _target loci prior to or after hormone signaling (Fig. 6). An ecdysone pulse at the end of the third-instar larval stage causes activation of the E74 and E75 gene clusters, hallmarked by chromosome “puffing.” EcR and USP7 both bind E74 and E75 prior to induction (Fig. 6A). However, following the ecdysone pulse, additional EcR is recruited but USP7 disappears, as expected of a corepressor (Fig. 6B). GMPS was present at the uninduced E74 and E75 loci (Fig. 6C) but, in contrast to USP7, persisted following activation (Fig. 6D). The dynamics of USP7 binding is independent of GMPS, because it is unaffected by RNAi-mediated depletion of GMPS (Fig. 6E and F). Our genetic analysis indicated that GMPS/USP7 acts antagonistically to BRE1 in vivo (Fig. 2). Therefore, we compared the binding dynamics of BRE1 with those of USP7 and RNA Pol II. In agreement with a low level of transcription prior to ecdysone induction, we detected both BRE1 and RNA Pol II at E74 and E75 (Fig. 6G and I). However, following activation, BRE1 and RNA Pol II levels were dramatically increased, whereas USP7 was removed (Fig. 6H and J). In conclusion, our analysis of binding to ecdysone _target loci supports a role for GMPS/USP7 as a transcriptional corepressor. Moreover, in accordance with their genetic antagonism, and opposing biochemical activities, there is an exchange of the ubiquitin protease USP7 for the ubiquitin ligase BRE1 concomitant with RNA Pol II recruitment.

GMPS/USP7 acts as an EcR corepressor.

To obtain additional evidence for the role of GMPS/USP7 as an EcR corepressor, we performed genetic interaction assays utilizing two distinct EcR mutants. EcR^V559fs is a loss-of-function mutation that has an activation defect (2). In contrast, EcR^A483T can still respond to hormone induction but its interaction with the corepressor SMRTER is abrogated, causing enhanced transcription in the absence of hormone (27). When EcR^V559fs was combined with GMPS/USP7 ectopic expression in the eye, we observed a clear enhancement of the developmental defect (Fig. 7A to C). In striking contrast, EcR^A483T suppressed the GMPS/USP7 overexpression eye phenotype (Fig. 7D). By themselves, the EcR^V559fs and EcR^A483T mutations result in normal eyes (not shown). Thus, defective activation by EcR enhances the GMPS/USP7 ectopic expression phenotype, whereas loss of repression causes suppression. A plausible molecular explanation is that the failure of EcR^A483T to bind the SMRTER corepressor compensates for the overexpression of GMPS/USP7 by lowering the overall EcR-directed transcriptional repression. Alternatively, EcR^A483T might be defective for GMPS/USP7 recruitment, possibly through SMRTER. Suggestively, SMRTER was detected by mass spectrometry in our USP7 purifications from Drosophila embryos (J. A. van der Knaap, unpublished results). Whatever the detailed molecular mechanism, these genetic interactions provide further in vivo support for the notion that GMPS/USP7 acts as an EcR corepressor.

To test for direct binding, we expressed and purified GST-tagged GMPS or USP7, which was immobilized on glutathione-Sepharose beads and incubated with radiolabeled EcR. We found that EcR bound efficiently to USP7 but not to GMPS (Fig. 7E). Additional binding studies revealed that USP7 bound the EcR ligand binding domain (LBD) but not the DNA binding domain (DBD). We observed very weak binding of GMPS to the LBD, but considering the lack of binding to full-length EcR, this was most likely background. We conclude that GMPS/USP7 interacts not only genetically with the EcR but also physically, through binding of USP7 to the LBD.

Finally, we decided to test the effect of RNAi-mediated depletion of USP7 or GMPS on a selection of ecdysone-regulated genes in S2 tissue culture cells. Whereas USP7 was efficiently depleted when _targeted by RNAi, only a modest reduction in GMPS levels could be achieved (Fig. 7F). Nevertheless, RT-qPCR revealed clear derepression of the ecdysone-regulated genes ImpE2, ImpL3, E75C, and E74A (1, 32) following USP7 or GMPS depletion. In contrast, reduced USP7 or GMPS levels did not affect the transcription of βFtz-F1, gapdh, or CG1381. We conclude that GMPS/USP7 acts as a gene-selective transcriptional corepressor of ecdysone-inducible genes.