The Deubiquitylation Activity of Ubp8 Is Dependent upon Sgf11 and Its Association with the SAGA Complex

S. cerevisiae strains.

The genotypes of strains used for this study are listed in Table 1. Individual TAP-tagged strains and deletion strains were obtained from Open Biosystems. TAP-tagged strains with deletions were obtained by crossing and dissecting individual TAP-tagged strains and deletion strains. Gene deletions that were not obtained from Open Biosystems were obtained by transformation with a PCR product obtained from the pRS415 vector, which harbors the LEU2 gene. Deletions and epitope-tagged strains were tested by PCR amplification of genomic DNAs or by Western blot analysis.

TAP purification.

The purification of Ada2-TAP-containing complexes was performed as previously described in order to separate the GCN5-containing HAT complexes ADA, SLIK, and SAGA (13). Purifications of Ubp8-TAP, Sgf11-TAP, and Spt7-TAP were performed as previously described, but with several modifications (19). All steps were performed at 4°C. Briefly, 6 liters of cells were grown to approximately 2 × 10⁷ cells/ml in yeast extract-peptone-dextrose (YPD) and then washed in 10 ml of extract buffer (40 mM HEPES [pH 7.4], 350 mM NaCl, 10% glycerol, 0.1% Tween 20, 2 μg of pepstatin A/ml, 2 μg of leupeptin/ml, and 1 mM phenylmethylsulfonyl fluoride [PMSF]). Whole-cell extracts were prepared by bead beating in a bead beater (BioSpec). Cellular debris was removed by spinning the extracts at 12,000 × g for 30 min. The extracts were cleared by ultracentrifugation at 100,000 × g for 90 min. For TAP purification, 500 μl of immunoglobulin G-Sepharose (Amersham, Uppsala, Sweden) beads was incubated with clarified lysate overnight at 4°C. The beads were washed with 20 ml of extract buffer followed by 10 ml of TEV cleavage buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.1% NP-40, 0.5 mM EDTA, 10% glycerol, 1 mM dithiothreitol [DTT]) and then were resuspended in 1 ml of TEV cleavage buffer. TEV cleavage was performed by the use of 10 μl (100 U) of TEV (Gibco BRL) overnight at 4°C. TEV-cleaved products were added to 3 ml of calmodulin binding buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM magnesium acetate, 1 mM imidazole, 2 mM CaCl₂, 0.1% NP-40, 10% glycerol, 0.5 mM DTT) along with 3 μl of 1 M CaCl₂ for each milliliter of TEV elution. Calmodulin-Sepharose (Amersham) purification was performed by binding of the eluate to 500 μl of beads overnight at 4°C. Proteins bound to calmodulin beads were eluted in 0.25-ml fractions by the use of calmodulin elution buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM magnesium acetate, 1 mM imidazole, 2 mM EGTA, 0.1% NP-40, 10% glycerol, 0.5 mM DTT). TAP-purified complexes were resolved in a sodium dodecyl sulfate (SDS)-10% polyacrylamide gel and visualized by silver staining.

MudPIT.

TAP-purified protein complexes were digested with endoproteinase Lys-C and trypsin (Roche) as previously described (29, 31). Peptide mixtures were loaded on three-phase MudPIT microcapillary 100-μm-diameter columns as described previously (14). Tandem mass spectra (MS/MS) were acquired for the eluting peptides by use of a Deca-XP ion-trap mass spectrometer equipped with a nano-liquid chromatography (nano-LC) electrospray ionization source (ThermoFinnigan). SEQUEST (7) was used to match MS/MS spectra to peptides in a database containing S. cerevisiae protein sequences downloaded from the National Center for Biotechnology Information website. Spectrum-peptide matches were only retained if they had a normalized difference in cross-correlation scores (ΔC_n) of at least 0.08 and minimum cross-correlation scores (X_Corr) of 1.8 for singly, 2.5 for doubly, and 3.5 for triply charged spectra. In addition, the peptides had to be at least seven amino acids long and at least half-tryptic. DTASelect (25) was used to select and sort peptide-spectrum matches passing this set of criteria. Peptide hits from multiple runs were compared with CONTRAST software (25).

Isolation of ubiquitylated histones.

The relative levels of ubiquitylated histone H2B (ubH2B) in different strain backgrounds (YKH045, YKH046, YKH047, YKL142, and YKL143) were tested by use of an N-terminally FLAG-tagged histone H2B to aid in purification and detection with an anti-FLAG-horseradish peroxidase (HRP) antibody (Sigma) as described previously (20). Purifications of the ubH2B substrate from strains YKH047 and YKL142 for the deubiquitylation assays described below were also performed as described above.

Deubiquitylation assay.

The FLAG-tagged H2B substrate (containing ubH2B and unmodified H2B) was obtained as described previously (20). Between 250 and 500 ng of this substrate was incubated at 30°C for 60 min in DUB buffer (10 mM Tris-HCl [pH 8.0], 1 mM DTT, 1 μM PMSF, 1 μg of aprotinin and pepstatin A/ml) with 1% TAP-purified fractions. As a control, the substrate was also incubated in DUB buffer to which only calmodulin elution buffer was added. The reaction was stopped by freezing in liquid nitrogen, followed by boiling in 1 volume of 2× SDS sample buffer for 5 min and electrophoresis in an SDS-18% polyacrylamide gel. Gels were transferred to polyvinylidene membranes (Immobilon), and Western blot analysis was performed with anti-FLAG-HRP (to detect ubH2B and H2B) and anti-hemagglutinin (HA)-HRP (to detect ubiquitin) antibodies.

Experiments involving reconstitution of the in vitro deubiquitylation activity were performed as described above, except that equal amounts of complexes were mixed 1:1 or 1:1:1 for 1 h at 30°C prior to the deubiquitylation reaction.

HAT assays.

HAT assays were performed essentially as described previously (6), with the following changes: purified ubH2B and unmodified H2B were used as substrates for HAT reactions, which were performed with either wild-type (WT) SAGA or SAGA lacking Ubp8 and Sgf11 (for assays involving the effect of preacetylating the histone template on subsequent deubiquitylation). The HAT reactions were carried out in a buffer containing 10 mM Tris-HCl (pH 8.0), 1 mM DTT, and 1 mM PMSF.

RNA analyses.

Saturated cultures were grown in YPD and diluted 1/100 in YPD or minimal medium. The cells were grown at 30°C to an A₆₀₀ of ∼1.0 (∼10⁷ cells/ml in a 10-ml total volume), and RNAs were extracted by the use of Trizol reagent (Invitrogen). For reverse transcription-PCR (RT-PCR) analysis, 2.5 μg of total RNA was used. The reverse transcriptase reaction was carried out in a 50°C water bath for 1 h with a random hexamer primer and Superscript III (Invitrogen), followed by incubation at 95°C for 5 min to inactivate the enzyme. Control RT reactions were carried out in the absence of reverse transcriptase. Five microliters of the RT reaction mixture was subsequently used in a 25-μl PCR. The ARG1 primers (5′-GTTGGGTACCTCTTTGGCAA-3′ and 5′-GCCCAGAATGATGACGTTACCC-3′) (18) generated a 700-bp product, and the ACT1 primers (5′-GTGAACGATAGATGGACCAC-3′ and 5′-TTGGGTTTGGAATCTGCCGG-3′) generated a 300-bp product.

Ubp8 is only associated with SAGA and SLIK, and this association requires Sgf11.

In order to understand how Ubp8 functions as a part of SAGA, we first needed to determine whether Ubp8 was associated with any other complexes in the cell. To this end, we tagged Ubp8 with TAP and subsequently isolated protein complexes associated with Ubp8 and showed that Ubp8 copurifies with HAT activity (Fig. 1A; also data not shown). The purification of Ubp8 looked similar to purifications of SAGA (Fig. 1A, compare lanes 2 and 3). In order to confirm this, we performed mass spectrometry analysis (MudPIT) of the purification product and confirmed that Ubp8 is only associated with proteins that were previously identified in the SAGA, SLIK, and ADA HAT complexes (Fig. 1C) (22). We also used a modified TAP purification of Ubp8 that allowed us to separate the various Gcn5-containing HAT complexes and confirmed that Ubp8 is part of both the SAGA and SLIK complexes, but not the ADA complex (5; also data not shown).

We previously showed that the deletion of Sgf11 results in the preferential loss of Ubp8 from the SAGA complex, so we tested the effects of deleting Sgf11 from the Ubp8-TAP strain. Since Ubp8 copurified with SAGA components and its association with SAGA depends on Sgf11, we were not surprised to find that upon deletion of Sgf11, we were only able to isolate the Ubp8 protein (Fig. 1B, lane 2, and C). These purified complexes were further used in the deubiquitylation experiments described below.

Sgf11 association with SAGA/SLIK requires Ubp8.

Since Ubp8 requires Sgf11 for its association with SAGA, we tested whether there was a reciprocal requirement. Using a TAP-tagged Ada2 strain, we purified both SAGA and SLIK from a strain lacking Ubp8 (Fig. 2A, lanes 5 and 6). It was previously shown that the loss of Ubp8 does not affect HAT activity and therefore that Ubp8 is not required for the integrity of the SAGA or SLIK HAT complex (10; also data not shown). Through protein gel analysis, we could not determine if any components, including Sgf11, were missing from the SAGA or SLIK complexes. Therefore, we went on to analyze these purified SAGA and SLIK complexes by MudPIT and found that in the absence of Ubp8, the only other subunit lost was Sgf11 (Fig. 2B). Therefore, Ubp8 and Sgf11 are reciprocally required for their stable associations with SAGA and SLIK.

In previous work, Powell et al. purified complexes that associate with Sgf11 through TAP purification (17). They found that most of the subunits that purify with Sgf11 are part of the SAGA complex, but their mass spectrometry analysis also showed proteins that are independent of SAGA and that copurified with Sgf11, in particular, TATA-binding protein (17). In our purification of Sgf11-TAP, only proteins associated with SAGA and SLIK were identified (13). In order to resolve this difference, we set out to identify proteins that were associated with Sgf11 independently of Ubp8 by deleting the UBP8 gene from the Sgf11-TAP strain. In contrast to the stability of Ubp8-TAP in the absence of Sgf11, Sgf11-TAP was unstable in the absence of Ubp8 (data not shown). Possible explanations for this apparent instability are discussed below.

Deletion of Sgf11 results in an increase in the overall level of H2B ubiquitylation.

Our data indicate that Ubp8 and Sgf11 function together in SAGA as a subcomplex that carries out the deubiquitylation of histone H2B. However, the exact role that Sgf11 plays in this process is not clear. Previous data demonstrated that the overall cellular histone H2B ubiquitylation levels are increased in the absence of Ubp8. Moreover, this increase was apparent in a yeast strain carrying a mutation in the catalytic domain of Ubp8, indicating that Ubp8 alone is responsible for the deubiquitylation of H2B in vivo. To determine if this is indeed the case, we deleted Sgf11 from a yeast strain harboring FLAG-tagged H2B and HA-tagged ubiquitin that was previously used to assay H2B ubiquitylation levels. We found that the deletion of Sgf11 increased the overall H2B ubiquitylation levels to the same extent observed for a Ubp8 deletion (Fig. 3A, compare lanes 1 and 3). Therefore, both Ubp8 and Sgf11 are important for maintaining wild-type H2B ubiquitin levels in the cell.

Ubp8 is necessary, but not sufficient, for deubiquitylation of H2B in vitro.

Since we were able to purify Ubp8 away from SAGA through the deletion of Sgf11, we wanted to determine if Ubp8 alone was able to deubiquitylate H2B in vitro. We found that although Ubp8 is necessary for deubiquitylation (10), it is not able to carry out this reaction on its own (Fig. 3B, lane 7). This finding is intriguing, as it suggests that it is not only the protein but also the context in which the protein is located that allows it to perform its function. We also found that SAGA purified in the absence of Sgf11 was unable to deubiquitylate H2B in vitro since it also lacked Ubp8p (Fig. 3B, lane 6). Additionally, both Ubp8-TAP and Sgf11-Tap purified complexes were able to deubiquitylate H2B (Fig. 3B, lanes 3 and 4). However, we did find that Ubp8-TAP purified complexes were less efficient than Sgf11-TAP or Ada2-TAP purified complexes in deubiquitylation. This was probably due to the presence of the TAP tag on Ubp8 causing some partial inhibition of activity.

GCN5 is not required for H2B deubiquitylation in vitro or in vivo.

Previous work indicated that the deletion of Ubp8 together with GCN5 resulted in severe defects in the transcription of both GAL1 and ADH2 (10). Instead of examining the role that Ubp8 or Sgf11 plays in SAGA, we set out to see if GCN5 and, more specifically, acetylation by SAGA, plays a role in deubiquitylation. We first tested to see if the deletion of Sgf11 alone or in combination with GCN5 had similar growth defects as those seen for the UBP8 GCN5 double deletion. We found this to be the case. The synthetic phenotype upon the deletion of Sgf11 and Gcn5 may result from the loss of the two enzymatic activities from the SAGA complex. However, it may also arise if the double deletion is more disruptive to the complex than either single deletion alone. These possibilities are currently being investigated. The deletion of Sgf11 in combination with Ubp8 had no growth defects on galactose, indicating that these two proteins function together and partially redundantly with GCN5 to regulate GAL1 (Fig. 4A). We next examined if a deletion of GCN5 resulted in a change in the overall H2B ubiquitylation level in vivo. We found that unlike the deletion of Sgf11 and Ubp8, the deletion of GCN5 allowed wild-type levels of H2B ubiquitylation, the first indication that these two SAGA activities are independent (Fig. 4B, compare lanes 1 and 2). We then checked to see if SAGA purified from a strain lacking GCN5 or lacking the GCN5 bromodomain was able to deubiquitylate H2B in vitro. We found that both GCN5 mutant complexes were able to deubiquitylate H2B, similar to wild-type SAGA (Fig. 4C). We next tested whether the deubiquitylation of H2B was affected by acetylation. First, we looked at the effect of acetylating purified FLAG-H2B or FLAG-ubH2B with SAGA lacking Sgf11 (to separate acetylation from deubiquitylation) and, surprisingly, found that preacetylating purified FLAG-H2B or FLAG-ubH2B slightly inhibited subsequent deubiquitylation by wild-type SAGA (Fig. 5A and B). This was a specific effect of acetylation and not a steric effect of the SAGA added to the initial HAT reaction, as the presence of this complex in the absence of acetyl-coenzyme A (CoA) did not inhibit the reaction (Fig. 5A, compare lanes 1, 3, 5, 7, 9, and 12). We also examined the effect of the addition of acetyl-CoA on deubiquitylation and found that the two enzymatic activities of SAGA were able to function normally when occurring at the same time (Fig. 5C and D).

Sgf11 and Ubp8 are required for the expression and/or repression of ARG1.

Since SAGA and ubiquitylation function in the activation and repression of ARG1, we examined the effect of increased cellular H2B ubiquitylation levels on ARG1 transcription (18, 27). Turner et al. demonstrated that ARG1 is derepressed by a RAD6 deletion and by a K123R mutation in H2B, indicating that the ubiquitylation of H2B is required for the repression of ARG1 (27). They also showed that GCN5 is required for the activation and repression of ARG1. Since deubiquitylation has also been implicated in gene activation, we performed an RT-PCR analysis of ARG1 transcription in both Ubp8 and Sgf11 deletion strains under repressing conditions (Fig. 6A and B). We found that under repressing conditions (growth in YPD), the deletion of either Ubp8 or Sgf11 resulted in a decrease in ARG1 expression to levels below wild-type expression; however, the deletion of Sgf11 had less of an effect on ARG1 expression than the deletion of Ubp8 (fivefold versus twofold) (Fig. 6B). We observed only a modest twofold increase in ARG expression in the K123R H2B mutation due to the fact that our strain had a high level of ARG1 expression even under repressive conditions (Fig. 6A, lane 1). However, the high level of ARG1 allowed us to see the repression caused by both Ubp8 and Sgf11 deletion. These two findings indicate that Ubp8 and Sgf11 antagonize RAD6 and BRE1 to create a balance of histone ubiquitylation that allows the proper transcription of ARG1. Additionally, we looked at the effects of these same mutations on ARG1 activation (Fig. 6C). We found that the K123R mutation as well as the Ubp8 and Sgf11 deletions resulted in an approximately twofold decrease in ARG1 expression (Fig. 6C). However, as previously shown, the deletion of GCN5 resulted in a larger decrease in the amount of ARG1 transcripts. Therefore, H2B ubiquitylation seems to have less of an effect on ARG1 activation than it does on ARG1 repression.