The tightly controlled deubiquitination activity of the human SAGA complex differentially modifies distinct gene regulatory elements

Abstract

The multisubunit SAGA coactivator complex facilitates access of general transcription factors to DNA through histone acetylation mediated by GCN5. USP22 (ubiquitin-specific protease 22) was recently described as a subunit of the human SAGA complex that removes ubiquitin from monoubiquitinated histone H2B and H2A in vitro. Here we demonstrate an allosteric regulation of USP22 through multiple interactions with different domains of other subunits of the SAGA deubiquitination module (ATXN7, ATXN7L3, and ENY2). Downregulation of ATXN7L3 by short hairpin RNA (shRNA) specifically inactivated the SAGA deubiquitination activity, leading to a strong increase of global H2B ubiquitination and a moderate increase of H2A ubiquitination. Thus, SAGA is the major H2Bub deubiquitinase in human cells, and this activity cannot be fully compensated by other deubiquitinases. Here we show that the deubiquitination activity of SAGA is required for full activation of SAGA-dependent inducible genes. Interestingly, the reduction of the SAGA deubiquitination activity and the parallel increase in H2B ubiquitation at inducible target genes before activation do not induce aberrant gene expression. Our data together indicate that different dynamic equilibriums of H2B ubiquitination/deubiquitination are established at different gene regulatory elements and that H2B ubiquitination changes are necessary but not sufficient to trigger parallel activation of gene expression.

Fig. 1.

H2Aub or H2Bub are deubiquitinated by the purified human SAGA complex but not by the free USP22 enzyme. Mononucleosomes (A) or free histones (B) extracted from HeLa cells and enriched in H2Aub and H2Bub were incubated with no enzyme (CT) or with purified SAGA complex (hSAGA) or recombinant USP22 (USP22 rec). Specific purification of hSAGA complex was verified by mass spectrometry analysis (see Table S1 at http://igbmc.fr/Lang_mcb2011). The deubiquitination reaction was analyzed by Western blotting using antibodies directed against ubiquitin, H2Bub, H2Aub, H2A, and USP22 as indicated. The Coomassie blue staining shows that similar amounts of mononucleosomes or free histones were used.

Fig. 2.

The DUB module is stabilized by a network of interactions between the different subunits. (A) Sf9 cells were infected with different combinations of baculoviruses expressing Flag-USP22, HA-ATXN7(75-172), Flag-ATXN7L3(3-151), and ENY2 as indicated. Whole-cell extracts were analyzed by Western blotting and revealed with the indicated antibodies. (B) Immunoprecipitations using an anti-HA antibody followed by peptide elution were analyzed by Western blotting revealed with the indicated antibodies. ATXN7(75-172) (in bold) is harboring the HA epitope. The asterisk indicates a nonspecific band.

Fig. 3.

Identification of the domains required for the formation of a stable DUB module. (A) Schematic representation of the domains of ATXN7L3 and ATXN7 involved in the formation of the DUBm. The functional regions of the human proteins are indicated as compared to their yeast orthologues Sgf11 and Sgf73, respectively (also see Fig. S3 and S4 at http://igbmc.fr/Lang_mcb2011). The different truncated mutants of ATXN7L3 and ATXN7 that were used in immunopurification experiments are shown and their ability to form a stable/active DUBm composed of the four proteins is summarized. (B and C) Sf9 cells were infected with several combinations of baculoviruses expressing different versions of USP22, ATXN7, ATXN7L3, and ENY2 as indicated. Immunoprecipitations using an anti-HA antibody followed by peptide elution were analyzed by Western blotting revealed with the indicated antibodies. Coomassie blue staining of the purified complexes are shown in Fig. S5 at the URL given above. The subunits harboring an HA epitope (that were used to pull down the complexes) are indicated in bold.

Fig. 4.

USP22 catalytic activity is regulated by the other subunits of the DUB module. (A) The DUB module containing the first domain of ATXN7 is active. (B) The ZnF-Sgf11 of ATXN7L3 is required for the full activity of USP22. Equal amounts of the different purified complexes described for Fig. 2 and 3 were incubated with or without 5 μM the suicide substrate ubiquitin vinyl sulfone (Ub-VS). This compound can form a covalent bond only with the active form of USP22, which is revealed by a 7-kDa shift of the enzyme when analyzed by Western blotting. (C) The above-described complexes were incubated with purified mononucleosomes with or without Ub-VS. H2B monoubiquitination was assessed by Western blotting using an anti-H2Bub antibody and USP22 reactivity with Ub-VS was measured as described for panel B.

Fig. 5.

Monoubiquitinated H2A and H2B are in vivo substrates of hSAGA deubiquitination activity. (A) SAGA complexes were immunopurified from HeLa cells stably expressing a shRNA against ATXN7L3 (shATXN7L3) or a control shRNA (shCt) by using anti-GCN5 antibodies and analyzed by Western blotting with the indicated antibodies. IgG heavy chains are indicated (*). (B) Total histones were purified by acidic extractions from these cell lines. Levels of ubiquitinated histones were analyzed with specific antibodies as indicated. (C and D) H2Aub and H2Bub relative levels were quantified and normalized to H2A and H2B levels, respectively.

Fig. 6.

Regulation of H2B ubiquitination at different regions of an AR-dependent gene. (A) LNCaP cells stably expressing a shRNA against ATXN7L3 or a control shRNA were treated with or without an AR ligand (R1881). mRNA levels of an AR-dependant gene (PSA) were analyzed by qPCR of reverse transcribed RNA and normalized to GAPDH mRNA levels and are presented as mean values ± SD of three independent experiments. (B) Schematic representation of the genomic locus of the PSA gene. Black boxes represent primer sets used for ChIP analysis. (C and D) Control LNCaP cell lines and LNCaP cell lines expressing a shRNA against ATXN7L3 were incubated with (+) or without (−) ligand (R1881) for 210 min. Quantitative ChIP assays were performed for H2Bub (C) and AR (D). Precipitated DNA was quantified by real-time qPCR using primers on enhancer/AREIII, promoter/AREI/II, and coding region/exon4 of the PSA gene and an intergenic control region. The values (mean ± SD of triplicates) are expressed as percentages of the amount of immunoprecipitated DNA normalized to the respective input DNA signal (% input). These results are representative of three independent experiments. The same color code is used in all panels.

Fig. 7.

Dynamics of H2B ubiquitination at the promoter and the transcribed region of ER stress genes. (A) HeLa cells stably expressing a shRNA against ATXN7L3 or a control shRNA were treated with thapsigargin for 3 or 4 h or with DMSO as a control. mRNA levels of ER stress genes (HERPUD and ERP70) were analyzed by qPCR of reverse transcribed RNA and normalized to CycloB mRNA levels and are presented as mean values ± SD of three independent experiments. (B) Schematic representation of the genomic locus of the HERPUD and ERP70 genes. Black boxes represent primer sets used for ChIP analysis. (D to G) Control and ATXN7L3 shRNA HeLa cell lines were incubated with thapsigargin for 4 h or with vehicle only (DMSO). Quantitative ChIP assays were performed for H2Bub (D), RNA Pol II (E), SPT20 (F), and ATXN7L3 (G). Precipitated DNA was quantified by real-time qPCR using primers on ERP70 and HERPUD promoters and coding regions and on an intergenic control region. The values (mean ± SD of triplicates) are expressed as percentages of the amount of immunoprecipitated DNA normalized to the respective input DNA signal (% input). These results are representative of three independent experiments. The same color code is used in all panels.