Transcription-Independent Function of Polycomb Group Protein PSC in Cell Cycle Control

See allHide authors and affiliations

Science  11 May 2012:
Vol. 336, Issue 6082, pp. 744-747
DOI: 10.1126/science.1215927


Polycomb group (PcG) proteins control development and cell proliferation through chromatin-mediated transcriptional repression. We describe a transcription-independent function for PcG protein Posterior sex combs (PSC) in regulating the destruction of cyclin B (CYC-B). A substantial portion of PSC was found outside canonical PcG complexes, instead associated with CYC-B and the anaphase-promoting complex (APC). Cell-based experiments and reconstituted reactions established that PSC and Lemming (LMG, also called APC11) associate and ubiquitylate CYC-B cooperatively, marking it for proteosomal degradation. Thus, PSC appears to mediate both developmental gene silencing and posttranslational control of mitosis. Direct regulation of cell cycle progression might be a crucial part of the PcG system’s function in development and cancer.

Polycomb group (PcG) proteins are transcriptional repressors that maintain cell-fate decisions and control cell proliferation (15). They function as part of distinct multiprotein complexes that modulate chromatin structure. The RING domain protein Posterior sex combs (PSC) is a subunit of Polycomb repressive complex 1 (PRC1) and dRING–associated factors (dRAF), which mediate monoubiquitylation of histone H2A (26). A substantial portion of PSC is part of neither PRC1 nor dRAF (6), suggesting that PSC might have additional functions. We compared the effect of depleting either Polycomb (PC), Polyhomeotic (PH), PSC, or dRING by treating S2 cells with the appropriate double-stranded RNAs (dsRNAs) (see supplementary materials and methods section and fig. S1). PC, PH, PSC, and dRING form the core of PRC1, whereas dRING, PSC, and dKDM2 are the central subunits of dRAF (26). Knock-down (KD) of PH or PSC decreased cell accumulation, whereas depletion of PC or dRING had no appreciable effects (Fig. 1A). Fluorescence-activated cell sorter (FACS) analysis indicated that cells lacking PSC primarily accumulated at the G2-M phase of the cell cycle (Fig. 1B). Loss of other PcG proteins did not give a clear cell cycle arrest. Therefore, PSC might function in cell cycle regulation, independent of PRC1 or dRAF.

Fig. 1

PSC function in mitosis. (A) S2 cell accumulation after RNAi-mediated KD of PC, PH, PSC, or dRING. (B) Cell cycle distribution of S2 cells after KD of the indicated proteins, determined by FACS. (C) Micrograph of cell cycle 11 embryos from a cross of Psch27/CyO females mated to wild-type (wt) fathers displaying chromatin bridges (yellow arrowheads). A comparable wt embryo is shown for comparison. Scale bars, 10 μm. (D) Indirect immunofluorescence of mitotic S2 cells after KD of PC, PH, PSC, or dRING. Cells were stained with antibodies against α-tubulin (red) and centromere identifier (CID) (green); DNA was visualized by 4′,6-diamidino-2-phenylindole (blue). Arrowheads indicate misalignment and missegregation. (E) Number of cells with normal mitosis was plotted as a percentage of the total number of mitotic cells. 95% confidence intervals are indicated (error bars). The asterisk marks the significant difference between mock- and PSC KD cells (P < 0.0001), as determined by a χ2 test. See also table S1.

Consistent with the G2-M arrest caused by loss of PSC, maternal effect mutations of Psc cause mitotic segregation defects in early Drosophila embryos (7). This is illustrated by the mitotic chromosome bridges, frequently detected in the progeny of Psch27 mutant mothers (Fig. 1C). Because early embryos have nonconventional checkpoint mechanisms, problems at either S phase or mitosis can lead to segregation defects (7, 8). In S2 cells, which have a conventional cell cycle, depletion of PSC caused severe mitotic defects (Fig. 1D). After depletion of PSC, ~68% of mitotic cells displayed an abnormal phenotype, whereas loss of the other PcG proteins did not affect mitosis. (Fig. 1E and table S1). The PcG system has been implicated in the regulation of cell cycle genes (912). Yet, because the integrity of PcG complexes is required for silencing (15), we suspected that PSC’s role in mitosis extends beyond transcription repression. Indeed, a portion of cellular PSC does not appear to be part of PRC1 nor dRAF (6).

To identify interaction partners, we used three distinct, affinity-purified antibodies to isolate PSC from whole-cell extracts of 0- to 12-hour-old Drosophila embryos. Mass spectrometric analysis revealed that, in addition to PRC1 and dRAF subunits, cyclin B (CYC-B), cell division cycle 2 (CDC2, also called cyclin-dependent protein kinase 1), and key subunits of the anaphase-promoting complex (APC) associated with PSC (Fig. 2A and table S2). Although PSC was present in PC, dRING, and PH purifications, CYC-B and the APC were absent. The APC is a multisubunit E3 ubiquitin ligase that is pivotal to cell cycle regulation (1323). CYC-B ubiquitylation by the APC, marking it for destruction by the proteasome, is required for completion of anaphase and cytokinesis (1623). We confirmed the selective association of PSC with CYC-B and APC by a series of immunoprecipitations (IPs) combined with protein immunoblotting. IPs of PcG proteins (Fig. 2B) showed that only PSC associated with CYC-B; CDC2; and the APC subunits MORULA (MR, also called APC2), CDC23 (APC8), and LEMMING (LMG, also called APC11). Reverse IPs further established the unique association of PSC with CYC-B (Fig. 2C) and the APC (fig. S2). LMG is a small 85–amino acid protein, comprising mainly a RING domain, that is essential for the ubiquitin ligase activity of the APC (2023). Many RING domain proteins are E3 ubiquitin ligases and frequently function as homo- or heterodimers (20). For example, PSC and its mammalian homolog BMI1 bind dRING or RING1B, respectively, and stimulate histone H2A ubiquitylation (2, 6, 2426). We found that the RING domain of PSC was necessary and sufficient to bind LMG, whereas its C-terminal region bound CYC-B (fig. S3). Thus, PSC appears to associate with LMG and CYC-B directly (Fig. 2D).

Fig. 2

PSC interacts biochemically and genetically with CYC-B. (A) Interaction heatmap, based on mascot scores, depicting associated factors identified by mass spectrometry after immunopurification of PSC, PC, dRING, and PH. See table S2 for details. (B) Co-IPs of PcG proteins. Associated proteins were detected by immunoblotting with antibodies against the indicated proteins. Input represents 10% of the binding reactions. (C) CDC2 and CYC-B co-IPs. (D) Cartoon summarizing the proteomics results. MR, Morula. (E) Scanning electron micrographs of fly eyes in which ectopic CYC-B expression was driven by the GMR enhancer. CYC-B overexpression was combined with RNAi-mediated depletion of PSC or PC. Genotypes are indicated.

To complement these biochemical experiments with a genetic-interaction assay, we employed the GAL4-UAS system in Drosophila (27). We used the glass multimer reporter (GMR) to drive ectopic CYC-B expression (GMR>CYC-B) in the developing eye (Fig. 2E). Ectopic CYC-B caused a mild rough-eye phenotype, characterized by disorganized ommatidia and loss of bristles. Concomitant expression of dsRNA directed against Psc mRNA [GMR>CYC-B; GMR>PSCRNAi (RNAi, RNA interference)] enhanced the GMR>CYC-B phenotype, consistent with the notion that PSC is a negative regulator of CYC-B. In contrast, expression of dsRNA directed against Pc had no effect on the CYC-B overexpression phenotype. Alone, neither reduction of PSC levels nor PC depletion had an appreciable effect on eye development. Thus, PSC interacts both genetically and biochemically with CYC-B.

To test whether PSC regulates abundance of CYC-B in vivo, we used the Patched (Ptc) driver to direct the expression of dsRNA directed against Psc mRNA in a central band across the wing imaginal disc of third instar larvae. Immunostaining of CYC-B (red) and PSC (green) revealed a strong increase in CYC-B, precisely in the area of the disc where PSC was depleted (Fig. 3, A and B). CYC-B abundance was also reported to increase in cellular clones that lack both Psc and Su(z)2, but not in Pc or dRing mutant clones (12). The effect of PSC on CYC-B was transcription-independent because expression of cyc-B mRNA was not affected by PSC depletion (Fig. 3C and fig. S4). Likewise, loss of PSC or LMG in S2 cells caused accumulation of CYC-B, which was even greater when both factors were depleted (fig. S5A). However, the abundance of cyc-B mRNA in S2 cells was not affected by depletion of PSC or LMG (fig. S5B). Thus, CYC-B accumulation appears to be caused by a transcription-independent mechanism, possibly involving PSC-directed ubiquitylation.

Fig. 3

Regulation of CYC-B ubiquitylation and degradation by PSC. Micrographs of wing imaginal discs from either wt or ptc-Gal4; tub-Gal80ts; UAS-PSCRNAi (PTC>PSCRNAi) third instar larvae. The ptc driver directs expression of dsRNA directed against Psc. (A) Immunostaining of wt disc with antibodies against PSC (green) or CYC-B (red). (B) Staining of disc from PTC>PSCRNAi larvae after heatshock. Arrowheads indicate areas where PSC is knocked down. (C) In situ hybridization to detect Psc and cyc-B mRNAs in discs from either PTC>PSCRNAi. See fig. S4 for wt larvae. (D) Immunoblotting of CYC-B immunoprecipitated from extracts of S2 cells depleted for PSC, LMG, or both proteins in the presence of proteasome inhibitors (fig. S6), using antibodies against ubiquitin (α-Ub) or CYC-B(α-CYC-B). (E) Quantification of mitotic phenotypes of S2 cells in which PSC, LMG, or PC were overexpressed (OE) as indicated. Asterisks mark a significant difference with mock; error bars denote 95% confidence intervals. See legend to Fig. 1E, fig. S7, and table S3.

To investigate the role of PSC in CYC-B ubiquitylation, we immunopurified CYC-B from cells that were depleted of PSC or LMG and treated with proteasome inhibitors (Fig. 3D and fig S6). Immunoblotting revealed that the loss of either PSC or LMG caused decreased levels of polyubiquitylated CYC-B (Ub–CYC-B). Almost no Ub–CYC-B was detectable in cells lacking both PSC and LMG. To test whether failed CYC-B destruction could explain the mitotic defects after the loss of PSC, we overexpressed CYC-B in S2 cells. After ectopic expression of CYC-B, ~70% of mitotic cells displayed a variety of defects (Fig. 3E, fig. S7, and table S3). Concomitant overexpression of either PSC or LMG almost completely reversed the CYC-B misexpression phenotype. In contrast, extra PC had no effect. Collectively, these results suggest that PSC-mediated CYC-B ubiquitylation is crucial for normal mitosis.

We used purified PSC, LMG, and CYC-B (fig. S8) in a reconstituted ubiquitylation system, which was dependent on E1 and E2 enzymes, to test the ability of PSC to act as a ubiquitin E3 ligase for CYC-B. Approximately equimolar amounts of either PSC or LMG could direct CYC-B ubiquitylation (Fig. 4A). But together, PSC and LMG generated higher levels of Ub–CYC-B. Determination of the CYC-B ubiquitylation rate revealed a more-than-additive effect of combining PSC and LMG, indicating that they function cooperatively (Fig. 4B and fig. S9). A substitution mutation replacing a signature cysteine residue of the RING consensus with an alanine [PSC-C287A (C287A: Cys287→Ala287)] abrogated PSC’s ability to ubiquitylate CYC-B (Fig. 4C). PSC-C287A blocked ubiquitylation of CYC-B by LMG, suggesting that it acts as a dominant negative. Indeed, the C287A mutation did not affect PSC binding to LMG or CYC-B (fig. S10). In contrast to ectopic PSC, expression of PSC-C287A caused severe mitotic defects in S2 cells (Fig. 4D and fig. S11). This mitotic phenotype was relieved by concomitant overexpression of LMG, suggesting that extra LMG squelches the dominant-negative PSC. Whereas ectopic expression of either PSC or LMG in S2 cells did not affect mitosis, overexpression of both PSC and LMG caused mitotic defects. These results suggest that PSC and LMG cooperate in the ubiquitylation of CYC-B, marking it for destruction by the proteasome.

Fig. 4

Cooperation between PSC and LMG during CYC-B ubiquitylation. (A) Reconstituted CYC-B ubiquitylation by PSC and LMG, detected by immunoblotting with α-Ub or α-CYC-B. CYC-B was incubated with a buffer control or in the presence of increasing amounts (~20 or 80 nM) of PSC alone or in addition to ~20 nM LMG (+), ~20 nM or 80 nM of LMG alone, or in addition to ~20 nM PSC (+). For purified proteins, see fig. S8. (B) Quantification of CYC-B ubiquitylation by PSC, LMG, or PSC and LMG (~20 nM each) by lumiimager fluorometry of gels (fig. S9). Ub–CYC-B was plotted as a function of the reaction time. a.u., arbitrary units. (C) Effect of PSC-C287A on CYC-B ubiquitylation. Analysis as described above. (D) Quantification of mitotic phenotypes of S2 cells in which PSC, PSC-C287A, and LMG were overexpressed as indicated. See legend to Fig. 1E, fig. S11, and table S3.

Regulated protein destruction is fundamental to cell cycle progression. The work reported here shows that, in addition to transcriptional repression, PSC cooperates with LMG in the APC to direct CYC-B degradation. During mitosis, PSC (and its mammalian homologs) and key PRC1 subunits PH and PC dissociate from the chromatin, making a transcriptional function at that time unlikely (28, 29). Like PSC, other chromatin regulators may also target proteins that are neither involved in chromatin dynamics nor transcription.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

Tables S1 to S3


References and Notes

  1. Acknowledgments: We thank S. Bray, J. Svejstrup, and Y. Moshkin for valuable discussions. This work was supported in part by Nederlandse Organisatie voor Wetenschappelijk Onderzoek Chemical Sciences Veni grant 700.57.406 (to A.M.-S.).

Stay Connected to Science

Navigate This Article