Similar temporal and spatial recruitment of native 19S and 20S proteasome subunits to transcriptionally active chromatin

Abstract

It has recently become clear that components of the proteasome are recruited to sites of gene transcription. Prevailing evidence suggests that the transcriptionally relevant form of the proteasome is a subcomplex of 19S base proteins, which functions as an ATP-dependent chaperone that influences transcriptional processes. Despite this notion, compelling evidence for a transcription-dedicated 19S base complex is lacking, and 20S proteasome subunits have been shown to associate with chromatin in some contexts. To gain insight into the form of the proteasome that is recruited to chromatin, we assembled a panel of highly specific antibodies that recognize native yeast proteasome subunits in chromatin immunoprecipitation assays. Using these reagents, we show that components from the three major subassemblies of the proteasome—19S lid, 19S base, and 20S core—associate with the activated GAL10 gene in yeast in a virtually indistinguishable manner. We find that proteasome subunits Rpt1, Rpt4, Rpn8, Rpn12, Pre6, and Pre10 are recruited to GAL10 rapidly upon galactose induction. These subunits associate with the entire transcribed portion of GAL10, display near-identical patterns of distribution, and dissociate from chromatin rapidly once transcription is shut down. We also find that proteasome subunits are enriched at telomeres and at genes transcribed by RNA polymerase III. Our data suggest that the transcriptionally relevant form of the proteasome is the canonical 26S complex.

The ubiquitin-proteasome system (UPS) plays a major role in cellular homeostasis by influencing the steady-state levels of proteins involved in processes such as cell cycle control, signaling, protein sorting, and apoptosis. Accumulating evidence suggests that the UPS is also involved in the regulation of gene expression, and that components of the UPS interact with chromatin and act both proteolytically and nonproteolytically to influence transcriptional processes. Perhaps one of the most interesting examples of how the UPS controls transcription centers on the proteasome, which has been linked to events in transcription ranging from control of activators and coactivators through to histone modifications, transcriptional elongation, and repression of cryptic transcription. The widespread involvement of the proteasome in transcription raises the intriguing possibility that it is involved in a majority of the critical steps in gene regulation.

Currently, a consensus has yet to emerge on the form of the proteasome that is recruited into transcriptional processes. A popular model posits that it is not the proteasome, but rather a subset of proteasomal ATPases from the 19S base complex, that interacts with chromatin during transcriptional activation. This complex, termed APIS (1), is argued to act independent of the 26S proteasome as a transcriptional chaperone that facilitates protein-protein interactions necessary for transcriptional activation. Support for APIS comes from genetic and biochemical evidence tying 19S base components to transcriptional activation domains (1, 2), and from results of chromatin immunoprecipitation (ChIP) experiments showing that 19S base proteins are recruited to activated genes separately from 20S core components (1, 3, 4). These observations, however, are countered by reports that the proteolytic activity of the proteasome is important for transcriptional activation in yeast (5, 6) and for repression of cryptic transcription in mammalian cells (7), and by the finding that 20S proteins also associate with active chromatin by ChIP, albeit with different temporal and spatial distributions than 19S proteins (1, 8). Disparate observations on the involvement of 19S versus 26S proteasomes in transcription make it difficult to arrive at a unified view of how proteasome components interact with chromatin.

Knowing which components of the proteasome are recruited to active chromatin is pivotal to resolving discrepancies in the functional importance of proteolytic versus nonproteolytic proteasome activities in transcription and to building a coherent view of proteasome function in gene expression. Because the majority of studies to date have relied on epitope-tagging of proteasome subunits to monitor their association with chromatin by ChIP, we sought to examine how native 19S and 20S proteasome subunits interact with a prototypical GAL gene in the yeast Saccharomyces cerevisiae. We assembled a panel of “ChIP-grade” antibodies that can be used to monitor the association of native 19S lid, 19S base, and 20S core proteasome components with chromatin, and used these antibodies to examine their kinetics of binding and their distribution across chromatin. Our data show that all proteasome subunits examined resolve with near-identical spatial and temporal distributions, supporting the notion that 26S proteasome subunits are recruited en masse to transcriptionally active genes.

Results

A Panel of ChIP-Grade Antibodies Against Native Proteasome Subunits.

To date, a clear picture has yet to emerge on how proteasome subunits interact with chromatin. We sought to resolve this issue by comparing the binding kinetics and distribution of 19S and 20S proteins at activated GAL genes. During the course of this work, we learned that many currently available antibodies against proteasome subunits do not work well in ChIP, and that the standard manipulation of epitope-tagging these proteins can induce transcriptional artifacts. For example, addition of a hemagglutinin (HA) tag to the carboxy-terminus of Rpt4 inhibits galactose-induced transcription of GAL10 (Fig. 1A), to a level comparable to that of the sug1-25 mutation in Rpt6 (9). To avoid epitope-tagging, therefore, we developed a panel of monoclonal antibodies (mAbs) against yeast proteasomes that give robust signals by ChIP. Mice were immunized with purified 19S and 20S proteasomes (10), hybridomas generated, and resulting clones screened for production of mAbs that yielded specific signals at the activated GAL10 open reading frame (ORF) by ChIP. This analysis resulted in five unique mAbs—against Rpn1, Rpt1, Rpt4, Pre6, and Pre10—that give clear ChIP signals at GAL10, and a number of others that work well in Western blots (WB; Table S1). With the exception of the α-Rpn1 mAb, we mapped critical epitopes recognized by each antibody (Table S1), and found that deletions within these epitopes result in a specific loss in reactivity in WB (Fig. 1B) and immunoprecipitation experiments (Fig. S1). We also compared ChIP signals for each of these four mAbs at the constitutively active PMA1 gene, in either wild-type yeast or yeast deleted for the respective epitope (Fig. 1C). In wild-type cells, we found that each mAb recovered between 0.02 and 0.06% of the equivalent of input DNA, and that disruption of specific epitopes reduces this signal approximately fivefold, to a level equivalent to that of an IgG control. We conclude that our mAbs against Rpt1, Rpt4, Pre6, and Pre10 specifically recognize their target proteins in a variety of assays.

Fig. 1.

Generation of ChIP-grade mAbs against proteasome subunits. (A) Epitope-tagging Rpt4 perturbs GAL10 induction. Wild type (WT; Sc507), RPT4-HA (yWS01), and sug1-25 (Sc660) were grown in raffinose-containing media, induced with 2% galactose for 30 min, and RNA isolated for analysis of GAL10 transcription by Q-PCR. GAL10 levels are expressed relative to a 25S RNA control. (B) Characterization of mAbs. WB was performed with the indicated mAbs on extracts from cells in which the respective epitopes for each mAb had been deleted. The Rpn2-2 mAb was a loading control. (C) Specificity of mAbs in ChIP. ChIP analysis was performed on the epitope-delete strains shown in (B), together with their congenic wild-type (WT) control strains. Chromatin was precipitated with the indicated mAbs, and coprecipitating DNA corresponding to the constitutively active PMA1 gene was measured by Q-PCR using a primer set centered on position + 633 in the PMA1 ORF. Data are presented as the overall efficiency of ChIP (i.e., percentage of input DNA recovered).

Our screen failed to produce mAbs capable of recognizing components of the 19S lid complex, either by ChIP or WB. To remedy this deficit, we surveyed a number of antibodies against 19S lid proteins and identified two polyclonal antibodies (pAbs), developed by the Finley laboratory against Rpn8 and Rpn12 (11), that give specific ChIP signals at the active GAL10 ORF. As these antibodies are raised against recombinant Rpn8 and Rpn12 proteins, we were unable to map specific epitopes, but we confirmed that these pAbs selectively recognize Rpn8 and Rpn12 in WB (Fig. S2), and note that they give patterns of behavior identical to our validated mAbs in ChIP assays. We therefore used these two pAbs, together with our four mAbs, to study how native subunits of the proteasome interact with chromatin.

19S and 20S Proteasome Subunits Are Recruited to an Activated Gene with Similar Kinetics.

Previous studies reported that components of the 19S base are recruited early to GAL genes after induction (3), whereas 20S core proteins accumulate significantly later (4), and 19S lid proteins fail to associate (3). To determine if these patterns of association are also displayed by native proteasome subunits, we used our panel of antibodies to perform ChIP assays at the activated GAL10 gene (Fig. 2).

Fig. 2.

Rapid recruitment of 19S and 20S proteasome subunits to the activated GAL10 ORF. Wild-type yeast (BY4742) were grown in media containing raffinose, induced with 2% galactose, and formaldehyde cross-linking performed at 0, 5, 10, 20, 30, 40, 50, 60, 90, and 120 min after galactose induction. Chromatin was immunoprecipitated with antibodies against Rpb3 (A), histone H3 (B), Rpt1 (C), Rpt4 (D), Rpn8 (E), Rpn12 (F), Pre6 (G), and Pre10 (H), and coprecipitating DNA from the GAL10 ORF (position + 886) measured by Q-PCR. ChIP data are expressed relative to the signal observed at an intergenic region on chromosome V.

Yeast were grown under noninducing (raffinose) conditions, galactose added, and cross-linked chromatin prepared at multiple time points after induction. Samples were immunoprecipitated with the individual antiproteasome antibodies—as well as α-Rpb3 (pol II) and α-histone H3 controls—and coprecipitating DNA measured by Q-PCR with a set of primers centered at + 886 on the GAL10 ORF (a region corresponding to the peak of proteasome subunit binding). As expected, induction with galactose results in rapid recruitment of Rpb3 (Fig. 2A) and eviction of histone H3 (Fig. 2B) from the GAL10 ORF. For proteasome subunits, we observed a pattern of binding that was reminiscent of the inverse of histone H3 eviction. All six proteasome subunits (Fig. 2 C–H) loaded rapidly onto the GAL10 ORF, each reaching a plateau at approximately 30 min after addition of galactose. Importantly, when expressed as a percentage of maximal ChIP signal for each antibody, the binding curves for the proteasome subunits are very similar (Fig. S3), with the midpoints for each curve ranging from 16 min (± 6 min) for Pre6 to 21 min (± 2 min) for Rpt1. There is thus no significant temporal difference in the recruitment of any of the proteasome subunits examined. We conclude that components of the 19S lid, 19S base, and 20S core of the proteasome are recruited to the activated GAL10 gene with near-identical kinetics.

19S and 20S Proteasome Subunits Show Similar Patterns of Spatial Distribution at Active Chromatin.

In addition to noting significant temporal differences in the recruitment of 19S and 20S proteins to activated GAL genes, previous reports concluded that there is a significant difference in the distribution of these proteins on active chromatin (3, 4). To determine if native proteasome subunits are differentially distributed, we performed ChIP under two conditions—noninducing and 60 min after induction with galactose—and surveyed subunit distribution across GAL10 (Fig. 3).

Fig. 3.

Similar spatial distribution of 19S and 20S proteasome subunits across the GAL10 locus. Wild-type yeast (BY4742) were grown overnight in media containing raffinose as the sole carbon source. Half the culture was kept in raffinose media (blue lines) and half transferred to galactose media (red lines) for 60 min prior to cross-linking. Chromatin was immunoprecipitated with antibodies against Rpt1 (A), Rpt4 (B), Rpn8 (C), Rpn12 (D), Pre6 (E), and Pre10 (F). Coprecipitating DNAs were then quantified by Q-PCR with primer sets corresponding to positions −448, −157, −25, + 250, + 474, + 886, + 1397, + 1850, + 2094, + 2360, and + 2633 of the GAL10 gene. ChIP data are expressed relative to the signal observed at an intergenic region on chromosome V. (G) Similar distribution of 19S lid, 19S base, and 20S core proteasome subunits across GAL10. Data from (B), (D), and (E) are presented on a single graph for comparison, with a cartoon of the GAL10 locus shown to scale underneath. Green box: UAS. Red box: TATA element. Blue box: ORF. Orange box: polyadenylation signal.

Under noninducing conditions (Fig. 3 A–F; blue lines) proteasome subunits bound at levels comparable to that observed at an intergenic region of chromosome V. After 60 min induction with galactose, however (Fig. 3 A–F; red lines), all six proteasome subunits could be detected throughout the GAL10 ORF. Contrary to previous reports (3, 4), we did not observe enrichment of 19S base proteins at the upstream activating sequence (UAS; Fig. 3 A–B), nor did we observe a 3′ bias in the distribution of 20S proteins (Fig. 3 E–F). Instead, we found little if any recruitment of 19S proteins to the UAS, low levels of binding at core promoter and nontranscribed 3′ sequences, and a broad peak of enrichment across the gene that peaked within the middle (+ 886) of the transcribed GAL10 ORF. Importantly, the patterns of binding for each of the proteasome subunits were nearly identical (Fig. 3G), demonstrating that there are no significant differences in the spatial resolution of proteasome subunits across active GAL10 chromatin.

The broad binding of all six proteasome subunits to the GAL10 ORF under inducing conditions suggested to us that proteasome subunits may distribute similarly to pol II. To test this idea, we compared distribution patterns of proteasome subunits with that of the core pol II subunit Rpb3, as well as patterns of serine 5 (S5) and serine 2 (S2) phosphorylation within the carboxy-terminal domain (CTD) of Rpb1. This analysis (Fig. S4) revealed that proteasome distribution patterns most closely resembled that of the S5-phosphorylated CTD, which is generally considered to be a hallmark of initiated pol II. A similar pattern of proteasome subunit distribution was observed at the constitutively active PMA1 gene (Fig. S5). We conclude that Rpt1, Rpt4, Rpn8, Rpn12, Pre6, and Pre10 associate with active chromatin in a virtually indistinguishable manner and that they track with the initiated form of pol II.

Association of 19S and 20S Proteasome Components Depends on Transcription.

To ask if association of proteasome subunits with chromatin depends on transcription, we glucose-inhibited GAL10 after induction, and performed a time course ChIP with representative lid, base, and core components (Fig. 4A). Under these conditions, proteasome subunits behave similarly to Rpb3, and dissociate rapidly after treatment of cells with glucose, consistent with the idea that ongoing transcription is required for retention of proteasome subunits on chromatin.

Fig. 4.

Rapid loss of proteasome subunit association with GAL10 upon cessation of transcription. (A) Loss of proteasome subunit (PS) association upon glucose-induced shutdown of GAL10. BY4742 yeast were induced with galactose, glucose added, and formaldehyde cross-linking performed at the 0, 2, 5, 10, 20, and 40 min time points. Chromatin was immunoprecipitated with antibodies against Rpb3 (orange line), Rpt1 (red line), Rpn8 (green line), and Pre10 (blue line), and coprecipitating DNA from the GAL10 ORF (position + 886) measured by Q-PCR. (B–D) Chemical-genetic inhibition of transcription induces a rapid loss of proteasome subunits from an active GAL10 gene. kin28-as yeast were grown overnight in galactose-containing media and treated with 30 μM NA-PP1 for either 0 (blue line), 20 (red line), or 120 (green line) seconds before ChIP was performed with antibodies against Rpb3 (B), Rpt1 (C), or Pre10 (D). Coprecipitating GAL10 DNAs were quantified by Q-PCR using the primer sets described in Fig. 2.

To further challenge this idea, we inhibited GAL10 transcription using a second method that involves chemical-genetic inhibition of the basal factor kinase Kin28 (12). Low concentrations (6 μM) of the small molecule inhibitor 1-napthyl-PP1 (NA-PP1) have been shown to partially inhibit transcription initiation in the presence of an engineered altered-specificity Kin28 protein (12), and we have found that higher concentrations of NA-PP1 (30 μM) result in a rapid and comprehensive inhibition of transcription initiation in vivo (Fig. 4B). We therefore activated GAL10 transcription, treated cells with NA-PP1, and performed ChIP. As expected, addition of NA-PP1 resulted in a rapid clearance of Rpb3 from GAL10, with a pronounced bias towards 5′ sequences (Fig. 4B); 20 s after NA-PP1 treatment (red line), Rpb3 levels at the 5′ end of GAL10 were reduced to near basal levels, whereas those at the 3′ end of the gene were less affected. This pattern likely reflects the fact that initiation of transcription is inhibited by NA-PP1, while preengaged pol II complexes continue to elongate. Consistent with this idea, Rpb3 was cleared from the entire GAL10 gene 120 s after NA-PP1 treatment (green line). Like Rpb3, association of both Rpt1 (Fig. 4C) and Pre10 (Fig. 4D) was inhibited by NA-PP1 in a time-dependent manner, indicating that proteasome subunit association with GAL10 is linked to ongoing transcription. Interestingly, however, the time-dependent loss of both proteins from GAL10 occurred evenly across the locus—without a 5′ bias—indicating that although proteasome subunit association requires ongoing transcription, it is not strictly linked to the passage of pol II across the template.

19S and 20S Function Is Required for Full Activation of GAL Gene Transcription.

The association of multiple 19S and 20S subunits with an activated GAL10 gene raises the issue of whether this association is functionally relevant. Mutations in 19S proteins impair GAL gene induction (Fig. 1A), but whether 20S function is required for this process is controversial. Recently, we learned that the proteasome inhibitor MG132 is only partially effective in yeast because the tryptic- and caspase-like sites of the proteasome compensate for chemical inhibition of the chymotryptic site by MG132 (6). We therefore used a chemical-genetic strategy in which MG132 is combined with inactivating point mutations in the tryptic- (PUP1) and caspase- (PRE3) like sites to effect rapid and comprehensive inhibition of proteasome function (6, 13) (Fig. 5). Consistent with the resistance of wild-type yeast to inhibition of the chymotryptic site, addition of MG132 has only a modest effect on GAL10 induction in PUP1PRE3 cells (Fig. 5A). In contrast, addition of MG132 to pup1pre3 cells results in a significant inhibition of GAL10 induction, at both the level of RNA synthesis (Fig. 5A) and pol II recruitment to the GAL10 ORF (Fig. 5B). The level of inhibition is similar to that seen with the sug1-25 mutation (Fig. 1A), demonstrating that both 19S and 20S proteasome function are required for full GAL gene activation.

Fig. 5.

Proteolytic activity of the proteasome is required for full GAL gene activation. (A) Chemical-genetic inhibition of the proteasome impairs induction of GAL10 mRNA. Wild-type (PUP1PRE3; WCG4) yeast, or yeast with inactivating mutations in the tryptic- and caspase-like sites of the proteasome (pup1pre3; YUS5) were grown in raffinose-containing media, treated with either DMSO or 50 μM MG132 for 45 min, and either kept in the presence of raffinose (white bars) or induced with galactose (black bars) for another 30 min. RNA was then harvested, and Q-PCR used to measure levels of GAL10 RNA, relative to a 25S RNA control. (B) MG132 treatment attenuates recruitment of pol II to the GAL10 ORF. Yeast strain YUS5 was treated as described in (A), except that ChIP was used to monitor association of Rpb3 with the GAL10 ORF (+ 886). ChIP signals are presented relative to the signal observed at an intergenic region on chromosome V.

19S and 20S Proteasome Subunits Are Found at Telomeres and Genes Transcribed by RNA Polymerase III.

Finally, we used our antibodies to ask whether proteasome subunits are located at regions of the genome other than those transcribed by pol II. We found no evidence of proteasome association at centromeres, origins of replication, or the ribosomal DNA loci. We did, however, find two additional sites of proteasome subunit enrichment on chromatin (Fig. 6). First, we found that lid, base, and core proteasome subunits are enriched at the right arm of chromosome six (VIR), specifically at sequences immediately adjacent to the telomere on that chromosome (Fig. 6A). The level of enrichment of proteasome subunits on subtelomeric chromatin is equivalent to that observed at active genes, implying that telomeres may be a major site of proteasome localization to chromatin. Second, we found that 19S and 20S proteins are enriched at two genes transcribed by RNA polymerase III (pol III); the noncoding RNA gene SCR1 and the tRNA gene tw(CAA)M (Fig. 6B). Interestingly, as we saw for pol II genes, proteasome association with pol III genes depends on transcription, as inhibition of pol III activity with transient MMS treatment (14) resulted in a reduction in proteasome subunit occupancy at both SCR1 and tw(CAA)M, to a level equivalent to that observed with the core pol III subunit Rpc34 (Fig. 6B). Association of proteasome subunits with telomeres, in contrast, was not affected by MMS treatment. We conclude that 19S and 20S proteasome subunits associate not only with pol II-transcribed genes, but also with genes transcribed by RNA polymerase III and with telomeres, indicating a broad involvement of the proteasome in chromatin-centric events.

Fig. 6.

Proteasome subunits associate with subtelomeric DNA and with genes transcribed by RNA polymerase III. (A) Proteasome subunits are enriched at telomere VIR. ChIP was performed (BY4742 strain) with antibodies against Rpt1 (red line), Pre10 (green line) and Rpn8 (orange line), and coprecipitating DNAs corresponding to the right arm of chromosome VI measured by Q-PCR. Primer sets used were centered at positions 11017, 4381, 2962, 995, and 113 bp, relative to the VIR telomere. The top cartoon shows a representation of this region of the chromosome, indicating the IRC7 locus and the telomeric repeats (red oval). ChIP data are expressed relative to the signal observed at an intergenic region on chromosome V. (B) Proteasome subunits associate with pol III-transcribed genes in an activity-dependent manner. Yeast (BY4742) were treated with MMS to inhibit pol III transcription, and ChIP performed with proteasome subunit (PS) antibodies against Rpt4 or Pre6, or with antibodies against the pol III subunit Rpc34. Coprecipitating DNAs were quantified by Q-PCR using primers corresponding to the pol III target genes SCR1 and tw(CAA)M, and telomere VIR. ChIP data are expressed relative to the signal observed at an intergenic region on chromosome V.

Discussion

It has recently become clear that the ubiquitin–proteasome system is intimately involved in transcription. One of the most controversial issues in this area, however, relates to the form of the proteasome that participates in gene regulation. We have attempted to resolve this issue by assembling a panel of ChIP-grade antibodies that recognize native proteasomal proteins, eliminating the need for epitope-tagging. The antibody panel described here allowed us to monitor the chromatin association profiles of six native proteasome subunits, two from each of the three major proteasome subcomplexes.

The major conclusion from this work is that all six subunits of the proteasome we monitored are recruited to an activated GAL gene in a virtually indistinguishable manner. We observe no temporal differences in the rates at which the subunits either associate with GAL10 upon activation (Fig. S3), or dissociate once transcription is inhibited (Fig. 4). We also note the substantial physical overlap of these subunits across the GAL10 (Fig. 3G) and PMA1 (Fig. S5) genes, as well as at telomere VIR (Fig. 6A) and pol III-transcribed loci (Fig. 6B). The extensive overlap we observe in both space and time makes it unlikely that proteasome components are recruited independently, and is more consistent with the idea that the form of the proteasome that is recruited to transcriptionally active chromatin is the canonical 26S complex.

The concept that the entire 26S proteasome is recruited to sites of transcription makes it possible to rationalize previous observations on the contribution of 19S and 20S proteasome subunit activities in transcription. As 26S proteasomes have the ability to both destroy proteins and to nonproteolytically dissociate protein complexes (15), there is no conceptual need to segregate proteasome subcomplexes for its various proteolytic and nonproteolytic capabilities to function. We propose that the 26S proteasome acts as a kind of “Swiss Army knife” in transcription, representing an integrated set of biochemical functions that are recruited as a whole but are sampled individually depending on the stage or challenge to gene expression. In some cases, nonproteolytic chaperone functions of the proteasome, mediated by the 19S base proteins, predominate, whereas in others proteolysis plays a rate-limiting role. By recruiting a multifunctional protein machine to chromatin, transcriptional processes have access to a collection of activities that could both reversibly (nonproteolytic) and irreversibly (proteolytic) drive transcriptional processes forward.

How are proteasomes recruited to chromatin? Interaction with Gal4 has been proposed as a mechanism of 19S base recruitment to chromatin (1), although one of the surprising results of our study is that little if any proteasomes can be detected at the GAL1/10 UAS (Fig. 3G), suggesting that direct contact with activators is not a primary mechanism of recruitment. We reported that ubiquitylation of histone H2B is required for 19S recruitment to GAL genes (16), and others have proposed that interaction with the histone H4 tail is required for this process (17). Given that interaction of proteasomes with chromatin is coupled to transcription, however, any mutation that disrupts transcription will also likely disrupt proteasome recruitment to chromatin, making it difficult to assign primary versus secondary effects based solely on loss-of-function experiments. We suggest that a number of distinct mechanisms bring proteasomes to sites on chromatin, as we observe proteasome enrichment at both pol II- and pol III-transcribed genes, and transcriptionally silent subtelomeric DNA. If the canonical 26S proteasome is recruited to chromatin, one possibility is that it interacts with chromatin in response to the presence of ubiquitylated substrates, analogous to the conventional view of how proteasomes are recruited into other biological processes.

The relevant functions and targets of the proteasome in transcription largely remain to be determined. Enrichment of proteasome subunits with the GAL10 and PMA1 ORFs (Fig. 3 and Fig. S5), and correlation with the initiated form of pol II (Figs. S4 and S5), indicates that its primary function in transcription may be exerted within transcribed gene sequences, although initiation of transcription is also clearly proteasome-dependent (Figs. 1 and 5). Further work will be needed to parse out proteasome function at distinct phases in transcription and to learn how proteolytic versus nonproteolytic activities of the 26S complex are employed in this process.

Materials and Methods

Generation and Characterization of mAbs.

Yeast 19S and 20S proteasomes were purified (10) and used to immunize mice for production of mAbs. Supernatants from approximately 100 positive clones were screened by WB and ChIP, comparing reactivity at the induced versus inactive GAL10 gene. Proteins targeted by each mAb were identified by screening a panel of yeast carrying epitope-tagged proteasome subunits and monitoring shifts in molecular weight (in WB) compared to untagged strains. Targets were confirmed, and epitopes mapped, by measuring reactivity to in vitro translated proteasome subunits and deletion mutants thereof. Antibodies characterized here are listed in Table S1. Other antibodies used in this study were (i) α-Rpb3 (W0012; Neoclone), (ii) α-Rpn8 and α-Rpn12 [ref. 11; from D. Finley, Harvard Medical School, Boston], (iii) α-H3 (Ab1791; Abcam), (iv) α-S2- and α-S5-CTD (H5 and H14, respectively; Covance), and (v) α-Rpc34 (rabbit polyclonal; from S. Hahn, Fred Hutchinson Cancer Research Center, Seattle).

Yeast Strains.

Yeast strains are described in Table S2. Unless noted, the majority of ChIP assays were performed in the wild-type BY4742 background. Epitopes recognized by the Rpt1-1 and Rpt4-1 mAbs were deleted in BY4742 by gene replacement with mutant versions of RPT1 and RPT4 in which the critical epitopes had been replaced with a single HA-epitope tag. The epitope recognized by mAb Pre6-1 was deleted by appropriately mutating PRE6 sequences in the plasmid p424-PRE6-T7 (18) and using plasmid shuffling to replace the wild-type PRE6 gene in yeast MHY1600 (18). Yeast carrying a deletion in the Pre10-1 epitope were a gift from M. Hochstrasser (Yale University, New Haven, CT).

Induction of GAL Genes and Inhibitors.

To induce GAL genes, cells were grown overnight in rich media containing 2% raffinose, 2% galactose added, and cells harvested at the indicated times for either RNA or ChIP analysis. Proteasome inhibitor treatment was performed as described (13). Chemical-genetic inhibition of Kin28 was performed using the strain kin28-AS (12) and treatment with 30 μM NA-PP1 (Tocris). To achieve DNA damage-induced transcriptional inhibition of pol III-transcribed genes (14), cells were treated with 0.04% MMS (Sigma) for 30 min prior to ChIP.

RNA Analysis.

RNA was extracted with hot phenol and reverse-transcribed with random hexamers using M-MLV reverse transcriptase (ABI Biosystems). Transcripts were quantified by Q-PCR with gene-specific primers. Primer sequences available on request.

Chromatin Immunoprecipitation (ChIP).

ChIP assays were carried out as described (19) with three modifications. First, cells were cross-linked for 40 min in 1% formaldehyde in the presence of 1× PBS. Second, cells were lysed, and chromatin sheared, in the presence of 0.5% SDS, which was diluted to 0.1% before immunoprecipitation. Finally, for ChIP assays using mouse antibodies, 10 μg of rabbit anti-mouse IgG (61-6500; Invitrogen) was added to the immunoprecipitation reactions 30 min before addition of Protein A agarose beads for immune complex recovery. Precipitated DNAs were quantified by Q-PCR with locus-specific primers. Primer sequences available on request.