sign in sign up
<<
Home , Chromatin, H3K79me2, Heterochromatin, Histone H2B, Nucleosome

Ubiquitination of H2B regulates dynamics by enhancing stability

Mahesh B. Chandrasekharan, Fu Huang, and Zu-Wen Sun1

Department of Biochemistry and Vanderbilt-Ingram Cancer Center, Vanderbilt University School of Medicine, Nashville, TN 37232

Communicated by C. David Allis, The Rockefeller University, New York, NY, July 15, 2009 (received for review February 6, 2009) The mechanism by which ubiquitination of (H2Bub1) (13, 15), but precisely how it causes this structural change has not regulates H3-K4 and -K79 and the -H2B yet been defined. In this study, the role of H2Bub1 in regulating Spt16-mediated nucleosome dynamics during transcrip- global chromatin structure was investigated. We found that tion is not fully understood. Upon investigating the effect of inducing the bulkier sumoylation at the H2B C terminus cannot H2Bub1 on chromatin structure, we find that contrary to the functionally replace ubiquitination to support H3 methylation. supposed role for H2Bub1 in opening up chromatin, it is important Thus, the SUMO or moiety, do not act as a mere for nucleosome stability. First, we show that H2Bub1 does not ‘‘wedge’’ to unfold chromatin. Using a repertoire of biochemical function as a ‘‘wedge’’ to non-specifically unfold chromatin, as and genetic analyses, we uncover an unexpected finding that replacement of ubiquitin with a bulkier SUMO molecule conju- H2Bub1 stabilizes . Collectively, our study provides gated to the C-terminal helix of H2B cannot functionally support a mechanism, wherein the transcriptional processes and -K4 and -K79 methylation. Second, using a series of biochemical chaperone-mediated chromatin dynamics are regulated by the analyses, we demonstrate that nucleosome stability is reduced or concerted action of H2Bub1 and its deubiquitination via the enhanced, when the levels of H2Bub1 are abolished or increased, stabilization and destabilization of the nucleosome, respectively. respectively. Besides elongation, we show that H2Bub1 regulates initiation by stabilizing nucleosomes positioned Results over the promoters of repressed . Collectively, our study The Engineered H2B Sumoylation Mimics the Occurrence but Not the reveals an intrinsic difference in the property of chromatin assem- Function of H2Bub1 on Chromatin. If the bulky ubiquitin (7.5 kDa) bled in the presence or absence of H2Bub1 and implicates the moiety at the H2B C terminus acts as a ‘‘wedge’’ to non- regulation of nucleosome stability as the mechanism by which specifically unfold chromatin, we reasoned that attaching a H2Bub1 modulates nucleosome dynamics and bulkier SUMO (12 kDa) might work similar to H2Bub1 in during transcription. mediating H3-K4 and -K79 methylation. To test this hypothesis, residues T122 and K123 in H2B were replaced by inserting two elongation ͉ methylation ͉ sumoylation consensus sumoylation sites (⌿KxE; ⌿, a hydrophobic residue; x, any ) to obtain an engineered H2B [H2B(2SU); Fig. A onoubiquitination of H2A (H2Aub1) and H2B 1 ]. To determine sumoylation and/or ubiquitination of H2B(2SU) by Western blotting, whole cell extracts were pre- (H2Bub1) plays important roles in regulating expres- M pared from yeast strains containing Flag epitope-tagged H2B, sion (1). In yeast, Rad6 conjugates ubiquitin to 123 (K123) H2B(2SU) or their mutant derivatives. In H2B(2SU) strain, two at the H2B C terminus (2), which in turn regulates the estab- ␣-Flag cross-reacting migrating slower than H2Bub1 lishment of H3-K4 and -K79 methylation and were seen in addition to the unmodified H2B(2SU) (Fig. 1B; (3–6). To explain this trans-histone cross-talk, it was postulated lanes 2 and 4). The faster migrating species is the ubiquitinated- that H2Bub1 might act as a ‘‘wedge’’ to non-specifically unfold H2B(2SU), as it is RAD6-dependent (Fig. 1B, lane 6). The slower chromatin for the (Set1 and Dot1) to gain migrating species might be sumoylated H2B(2SU), as it is access to their substrates or, H2Bub1 might function as a dependent on Siz1 (the SUMO E3 ligase) (Fig. 1C). Immuno- ‘‘bridge’’ to directly recruit them (6, 7). Two recent studies precipitation (IP) using ␣-Flag and Western blotting with implicate Swd2, a Set1-COMPASS complex subunit, as the link ␣-SUMO confirm that H2B(2SU) is indeed sumoylated (Fig. between H2Bub1 and H3-K4 methylation (8, 9). Swd2 seems to S1A). Sumoylation has been shown to occur at the N terminus regulate H3-K79 methylation by recruiting Dot1 (8). As viable of H2B (16), but it was not detected in H2B strain or its Swd2 mutants mainly affect H3-K4 trimethylation in vivo (9–10), derivatives after IP (Fig. S1A; lanes 2, 3, and 5), probably due to there might be other regulator(s) mediating H3-K4 mono- and its low abundance as compared to H2B(2SU)su. dimethylation. Nevertheless, the question as to whether H2Bub1 Each lysine in the inserted sumoylation sites was replaced with plays a structural (wedge) or signaling (bridge) role in modu- a leucine to test whether sumoylation occurs at one or both sites lating Swd2 or any other regulator(s) remains unanswered. (K1L and K2L; Fig. 1A). The K1L mutation prevented ubiq- Recent studies revealed that H2Bub1 regulates transcription uitination, but not sumoylation; and only ubiquitination was elongation independent of H3 methylation (11–13). In vitro detected in the H2B(2SU)-K2L strain (Fig. 1B). Thus, ubiquiti- transcription elongation experiments have shown that H2Bub1 nation and sumoylation mainly occur on K1 and K2 in promotes the function of human H2A-H2B chaperone, FACT H2B(2SU), respectively. Next, we tested whether H2B(2SU)su is (hFACT), in stimulating Pol II passage through a nucleosomal associated with chromatin. First, fractionation of yeast sphero- template by displacing an H2A-H2B dimer (12, 14). Addition- plasts to separate proteins localized to cytoplasm or nucleus ally, H2Bub1 and Spt16 (a subunit of the yeast FACT) function revealed that H2B(2SU)su partitioned with the chromatin- cooperatively in nucleosome reassembly in the wake of elongat- ing Pol II (15). Collectively, these findings imply an important role for H2Bub1 in modulating nucleosome dynamics during Author contributions: M.B.C. and Z.-W.S. designed research; M.B.C. and F.H. performed transcription. research; M.B.C., F.H., and Z.-W.S. analyzed data; and M.B.C. and Z.-W.S. wrote the paper. Since ubiquitination leads to a bulky moiety addition onto The authors declare no conflict of interest. histones, it is expected to exert a dramatic effect on chromatin 1To whom correspondence should be addressed. E-mail: [email protected]. structure and transcription. Indeed, H2Bub1 has been proposed This article contains supporting information online at www.pnas.org/cgi/content/full/ to exert its regulatory functions by affecting chromatin structure 0907862106/DCSupplemental.

16686–16691 ͉ PNAS ͉ September 29, 2009 ͉ vol. 106 ͉ no. 39 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0907862106 Downloaded by guest on October 3, 2021 H2B …VSEGTRAVTKYSSSTQA rad6 K1L K2L A 123 A H2B(2SU) H2B(2SU) H2B(2SU) H2B(2SU) …VSEGTRAV G YSSSTQA H2B(2SU) IKQE IKQE Marker MNase K1 K2 H2B(2SU)-K1L …VSEGTRAVILQEGIKQEYSSSTQA H2B(2SU)-K2L …VSEGTRAVIKQEGILQEYSSSTQA

B 6 rad

2B(2SU)/ H2B no H2Btag H2B-K123RH2B(2SU)H2B/rad6H H2B(2SU)-K1LH2B(2SU)-K2L * H2B(2SU)su B H2B-K123R ubp8 H2B(2SU)ub1 H2B H2Bub1

-Flag Marker 0 1 2.5 5 10 0 1 2.5 5 10 0 1 2.5 5 10 Time H2B(2SU) (min) H2B 15678243 C

2B(2SU)-K1L H2B no H2Btag H2B(2SU)H2B(2SU)-K1LH2B H2B(2SU)H * H2B(2SU)su Fig. 2. H2Bub1 levels can differentially affect the sensitivity of chromatin to H2B(2SU)ub1 digestion. (A) Nuclei isolated from the indicated strains -Flag H2Bub1 were treated with increasing MNase concentrations. Undigested or partially H2B(2SU) H2B digested DNA was resolved in a 1.8% agarose gel. (B) Nuclei isolated from the 1456723 indicated strains were treated with MNase (5 U/mL) for increasing incubation SIZ1 siz1 time. Arrowheads indicate the nucleosomal ladder.

H2B(2SU)-K2L D 6 H2B(2SU) H2Bub1 (17), H2B(2SU)su is present in the 5Ј ORF of consti- 4 tutively expressed genes (PMA1 and DMA2) and in an ORF-free 2 on V, but is absent from the H2B(2SU)su

Fold change in right-end telomeric region of chromosome VI (Fig. 1D). 0 PMA1 DMA2 INT-V TEL-VI-R As only sumoylation was detected in H2B(2SU)-K1L, this mutant was used to determine whether sumoylation is sufficient E to mediate H3-K4 and -K79 methylation similar to H2Bub1. In ad6 r contrast to H2B(2SU) and H2B(2SU)-K2L, H3-K4 di- and 2SU)-K2L trimethylation were totally abolished in the H2B(2SU)-K1L 2B(2SU)-K1L H2B noH2B tag H2B-K123RH2B(2SU)H2B/rad6H2B(2SU)/H H2B( mutant (Fig. 1E). Likewise, H3-K79 methylation levels in - H2B(2SU)-K1L are similar to those in H2B-K123R and rad6⌬ -H3K4me2 strains (Fig. 1E). Therefore, although H2B(2SU)su is associated - with transcriptionally active and excluded from -H3 -like regions (Fig. 1D), it cannot support H3-K4 -H3K79me1 and -K79 methylation. - -H3K79me3 H2Bub1 Regulates Global Chromatin Structure. We wondered -H3 whether the bulky SUMO at the H2B(2SU) C terminus exerts 123 45678 any effect on bulk chromatin structure. To this end, we used Fig. 1. Sumoylation induced at the H2B C terminus cannot functionally micrococcal nuclease (MNase), an that preferentially replace H2Bub1 in mediating H3-K4 and -K79 Methylation. (A) Amino acid cleaves the linker region between the nucleosomes, but can also sequence of the C-terminal region of yeast histone H2B from valine (114) to nick the nucleosomal DNA (18). The nuclei of H2B(2SU)-K2L alanine (130). Two consensus sumoylation sites (bold) separated by a spacer harboring only ubiquitination exhibit an MNase cleavage pattern glycine were inserted into H2B to create the chimeric H2B [H2B(2SU)]. To ⌬ determine the sumoylation site, the two (K1 and K2) were changed to similar to H2B(2SU) (Fig. 2A). In contrast, H2B(2SU)/rad6 and leucine (underlined) to obtain mutants, H2B(2SU)-K1L and H2B(2SU)-K2L. H2B(2SU)-K1L, which contain only sumoylation, display a de- Except for the control (H2B no tag), the wild type, the chimeric H2B and their crease in the overall intensity of the nucleosomal ladder as derivatives contain a single Flag epitope at the N terminus. (B and C) Yeast compared to H2B(2SU), indicating an increased sensitivity to whole cell extracts of the indicated strains were analyzed for the occurrence MNase digestion. This result suggests that the presence of of ubiquitination (ub1) and sumoylation (su) by Western blot. Asterisk, a sumoylation affects the global chromatin structure. cross-reacting with ␣-Flag. (D) Levels of H2B(2SU)su at the indicated Alternatively, the increased sensitivity to MNase digestion ␣ ␣ genomic loci were analyzed by sequential ChIP using -Flag and -SUMO. Fold may be due to a lack of ubiquitination. Indeed, the MNase enrichment of H2B(2SU)su in the indicated regions is shown relative to the digestion profiles of the H2B-K123R nuclei treated with increas- control strain that lacks the sumoylation site [H2B(2SU)-K2L]. Error bars de- note standard error of the mean obtained from three independent experi- ing incubation time (Fig. 2B) or MNase amount (Fig. S2A) ments. (E) Histone methylation levels in whole cell extracts were assessed exhibit decreased intensity in the nucleosomal ladders as com- using the indicated H3-K4 and -K79 methylation-specific antibodies. pared to wild type (H2B). In contrast, nuclei isolated from a strain lacking Ubp8 (the H2B deubiquitinase; ubp8⌬) and con- taining elevated amounts of H2Bub1 display an increased in- enriched nuclear fraction (Fig. S1B). Second, chromatin dou- tensity of MNase-digested nucleosomal DNA as compared to ble-IP (ChDIP) using ␣-Flag and ␣-SUMO further confirmed wild type (Fig. 2B and Fig. S2A), indicating a decreased sus-

the presence and spatial occurrence of H2B(2SU) sumoylation ceptibility of the chromatin to MNase digestion. Since a nucleo- on chromatin. Quantitative PCR (qPCR) analyses show that like somal ladder is evident in the absence or elevated levels of

Chandrasekharan et al. PNAS ͉ September 29, 2009 ͉ vol. 106 ͉ no. 39 ͉ 16687 Downloaded by guest on October 3, 2021 WT H2B-K123R ubp8 WT rad6 wild type and the H2B-K123R or rad6⌬ reflect an intrinsic difference in the property of nucleosomes assembled in the PMA1 TEL-VI-R presence or absence of H2Bub1. Defects seen in the absence of H2Bub1 may be due to the loss INT-V of histone methylation. To test this possibility, we used a rad6 H2Bub1 allele (P43L), a unique mutant that affects only -Flag H2B silencing but not other functions of Rad6 (20), and severely reduces H2Bub1 but causes only a slight decrease in H3-K4 and -H3 -K79 trimethylation (Fig. S2D). Similar to mutants lacking H2Bub1 (Figs. 2 and 3), the rad6-P43L showed increased MNase 2 WT sensitivity and decreased H3 levels on chromatin (Fig. S2D). H2B-K123R 1.5 rad6 These results further confirm that H2Bub1 alone can affect ubp8 1 nucleosome structure. Global defects detected by MNase diges- tion and chromatin fractionation may appear subtle, but it can 0.5 be explained by the results from several microarray analyses

levels on chromatin 0

Fold change in histone H2B H3 undertaken to detect changes in -wide transcript levels, which have shown that only a subset of genes is affected in the Fig. 3. Changes in H2Bub1 levels affect the amount of histones on chroma- absence of H2Bub1 (21, 22) or due to excess H2Bub1 (21, 23). tin. Nuclei isolated from the indicated strains were lysed in a hypotonic solution to obtain chromatin. Equal amounts of chromatin were resolved by 15% SDS/PAGE. Histone H2B and H3 levels were assessed by Western blots. H2Bub1 Affects the Histone H2B Levels on Constitutive and Repressed Both the initial and normalized amounts of chromatin were measured by Genes. To reaffirm the affects of H2Bub1 on histone occupancy quantitative PCR using two-fold serially diluted (triangle) genomic DNA tem- on chromatin, we evaluated the changes in H2B occupancy on plate isolated from the chromatin. Primers for the indicated regions (as in Fig. both constitutively expressed and repressed genes using ChIP 1D) were used in multiplex PCR. Graph depicts the fold change in H2B and H3 assay. Since H2Bub1 is important for the nucleosome reassembly amounts on chromatin in the mutants relative to wild type. Error bars denote function of Spt16 during transcription elongation (15), we ex- standard error of the mean obtained by measuring the histone amount in the amined the H2B occupancy on chromatin. H2B levels are mutant at each dilution normalized to the respective amount in the wild-type increased only in the ORF but not in the promoters of highly chromatin (set as 1). expressed PMA1 and moderately expressed FUN12 genes in ubp8⌬ as compared to wild type (Fig. 4A). This data shows that H2Bub1 at any incubation time or MNase amount tested (Fig. 2B H2Bub1 increases histone occupancy, probably by stabilizing the and Fig. S2A), these are not the digestion profiles typically nucleosomes and regulates chromatin dynamics during transcrip- observed upon changes in bulk chromatin structure (15, 18). tion elongation. Surprisingly, there is no apparent difference in Hence, it would be erroneous to conclude that, globally, a H2B levels between the wild type and the H2B-K123R mutant complete nucleosomal displacement or their increased accumu- (Fig. 4A) across the and ORF regions. lation including the linker regions occurs in the absence or In addition to positively regulating transcription, H2Bub1 is also involved in gene repression. In the absence of H2Bub1 presence of excess H2Bub1, respectively. While the digestion ⌬ profiles are atypical, they were reproducibly observed in inde- (rad6 or H2B-K123R), ARG1 shows derepressed transcription in rich media (non-inducing condition) with a concomitant pendent strains. Importantly, they reveal that the loss or gain of increase in the binding of Spt15 to the promoter (24). This result H2Bub1 does affect the global chromatin structure in a con- suggests that H2Bub1 might repress ARG1 transcription by trasting manner. Collectively, these results suggest that contrary stabilizing the nucleosome positioned on the TATA box to to the supposed role for H2Bub1 in opening up the chromatin, exclude Spt15 binding. Indeed, our ChIP analysis revealed that the bulky ubiquitin moiety on H2B might lead to a compact and the H2B levels on ARG1 TATA region are enhanced when stable chromatin conformation. H2Bub1 levels are increased (ubp8⌬, ubp10⌬,orubp8⌬ ubp10⌬) (Fig. 4C). This correlates well with the reported reduction in its H2Bub1 Affects the Histone Levels on Chromatin. Differences in transcript levels in ubp10⌬ and ubp8⌬ ubp10⌬ (23). Similar MNase digestion may be due to changes in nucleosome content results were obtained upon examining the levels of well- on chromatin or due to an altered nucleosome conformation characterized positioned nucleosomes (Fig. 4B) that encompass resulting from changes in histone-DNA contacts. To test these either the TATA boxes or upstream activating sequences of possibilities, we examined the effect of H2Bub1 on global histone PHO5, GAL10, and ADH2 promoters (Fig. 4C). For PHO5, the levels on chromatin following fractionation (19) of isolated observed increase in H2B levels correlates well with reported nuclei. Briefly, nuclei isolated from the wild type and mutants reduction in its basal expression in ubp8⌬, ubp10⌬, and ubp8⌬ were gently lysed using a hypotonic solution to obtain chromatin. ubp10⌬ under non-inducing conditions (21, 23). Similar to Equal amounts of chromatin, as confirmed by qPCR (Fig. 3 and constitutively expressed genes (Fig. 4A), no change in H2B levels Fig. S2B), were subjected to Western analysis. Similar to the was detected in the absence of H2Bub1 (H2B-K123R) (Fig. 4C). MNase digestion results (Fig. 2), we observed a contrasting Next, we performed ChDIP assay to determine the presence of change in the levels of H2B and H3 on chromatin. Histone levels H2Bub1 employing ␣-Flag and an antibody that recognizes were increased in ubp8⌬, but they were reduced in H2B-K123R ubiquitinated proteins. qPCR analysis detected low levels of and rad6⌬ as compared to wild type (Fig. 3). Since the amount H2Bub1 on the repressed promoters under non-inducing con- of Spt15 (TBP) on chromatin appears to be similar in all strains dition in the wild type as compared to the control (H2B-K123R) (Fig. S2C), the observed changes in histone levels cannot be due (Fig. 4D). Furthermore, the levels of H2Bub1 on all promoters to any discrepancies in chromatin quantitation. Therefore, our are significantly increased in strains containing excess H2Bub1 fractionation result suggests that H2Bub1 might regulate histone (ubp8⌬, ubp10⌬, and ubp8⌬ ubp10⌬). Collectively, the ChDIP occupancy on chromatin. Similar to H2B-K123R and rad6⌬, data demonstrate that H2Bub1 occurs on repressed promoters H2Bub1 was not seen in chromatin obtained from the wild type and suggest that it can restrict transcription by stabilizing nu- (Fig. 3), suggesting that it is lost due to the action of deubiq- cleosomes positioned at the Spt15 or binding sites. uitinases during fractionation. Nevertheless, the difference in the Given the increase in H2B occupancy in deubiquitinase- amount of histones detected in the fractionated-chromatin of deficient strains (Fig. 4A–C), we also tested whether increased

16688 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0907862106 Chandrasekharan et al. Downloaded by guest on October 3, 2021 on nucleosome stability. To retain and detect H2Bub1, we used A 2.5 PMA1 FUN12 WT 2 H2B-K123R the ubp8⌬ ubp10⌬ strain that harbors higher levels of H2Bub1 ubp8 1.5 than ubp8⌬ (23, 25). Further, to measure the degree of salt- 1 dependent nucleosome disruption, we evaluated the solubility of 0.5 H3, a component of the tetramer core within the nucleosome.

Fold change in H2B Fold change in 0 Initial histone extraction using a broad-range stepwise gradi- PP5’ M 3’ 5’ M 3’ ent salt concentration (0.2–2.0 M NaCl) showed a dramatic TATAA ARG1 TATAA PHO5 GAL10 TATAA TATAA ADH2 alteration to the nucleosome stability between 0.5 and 0.6 M, as B UAS UAS UAS UAS ORF evident from the increased histone solubility (Fig. S4). This data agrees with the reported dissociation of yeast nucleosome at 0.5 M NaCl (26) and it restricted the range of salt concentration to 3 WT C H2B-K123R be used in the solubility assay. To test whether increased H2Bub1 ubp8 2 stabilizes nucleosomes by reducing its solubility, equal amounts ubp10 ⌬ ⌬ ubp8 ubp10 of nuclei isolated from wild type and ubp8 ubp10 were 1 extracted with solutions of increasing salt concentrations (475– 615 mM). H3 in ubp8⌬ ubp10⌬ resists salt extraction and is less Fold change in H2B Fold change in 0 ARG1 PHO5 GAL10 ADH2 soluble as compared to wild type at lower salt concentrations (475–515 mM) (Fig. 5A). To test whether absence of H2Bub1 24 D 18 decreases nucleosome stability, equal amounts of nuclei isolated 12~ ~ ~ ~ ~ from wild type, H2B-K123R, and bre1⌬ (lacking the E3 ligase for ~ ~ ~ ~ ~ H2Bub1) were subjected to salt extraction (200–400 mM). In 4 contrast to the histone solubility in ubp8⌬ ubp10⌬, H3 is readily 3 soluble at very low salt concentration (200–300 mM) in H2B- 2 K123R and bre1⌬ relative to wild type (Fig. 5B). Since the relative soluble H3 amounts at high salt concentrations are Fold change in H2Bub1 Fold change in 1 0 similar between wild type and mutants (Fig. 5A and B; Input), ARG1 PHO5 GAL10 ADH2 they serve as a control to show that equal amounts of nuclei were Fig. 4. Increase in H2Bub1 augments H2B amounts on constitutive and used in the assay. Next, low salt solubility of H2B was assessed. repressed genes. (A) Levels of H2B at the promoter and ORF regions of PMA1 Similar to H3, H2B is readily soluble at very low salt concen- and FUN12 were analyzed by ChIP assay using ␣-Flag. Fold enrichment of H2B tration (200–300 mM) in H2B-K123R and bre1⌬ relative to wild levels on the promoter (P) or ORF regions [proximal (5Ј), middle (M), distal (3Ј)] type (Fig. 5C). Further, given its outward occurrence in the in the mutants are shown relative to the wild type. (B) Ovals with solid line, nucleosome and its propensity to undergo constant eviction, experimentally-verified positioned nucleosomes; Oval with the dotted line, a H2B appears to be more soluble than H3 in the absence of speculative positioned nucleosome; rectangle, upstream activating sequence H2Bub1 (Fig. 5B and C). The differential sensitivity to salt- (UAS); Arrow, transcription start site; TATAA, the TATA element; and the thick dependent nucleosome disruption in the absence or elevated solid line, region amplified by PCR following ChIP. (C) H2B levels at the promoters were analyzed as in (A). Fold change in H2B levels in the mutants levels of H2Bub1 provides strong evidence that H2Bub1 affects is shown relative to the wild type. (D) H2Bub1 levels were analyzed by the stable association of histones with chromatin. sequential ChIP using ␣-Flag and an antibody that recognizes ubiquitinated Since H2Bub1 modulates the overall chromatin structure by proteins. Fold enrichment in H2Bub1 levels in the indicated strains is shown affecting nucleosome stability (Figs. 2–5), it is conceivable that relative to H2B-K123R. Error bars denote standard error of the mean obtained the changes in H2Bub1 levels might affect cell growth by from at least three independent experiments. hindering transcription. While loss of H2Bub1 (H2B-K123R) causes modest slow growth as compared to wild type, deletion of both H2B deubiquitinases (ubp8⌬ ubp10⌬) leads to a severe H2Bub1 levels led to increase in H3 occupancy. No change in H3 growth defect (Fig. 5D). Importantly, the slow growth defect of levels was observed on both constitutive and repressed genes ubp8⌬ ubp10⌬ can be overcome upon eliminating H2Bub1 (Fig. S3). This data shows that H2Bub1 affects the H2A-H2B through the mutation of the site of ubiquitination (H2B-K123R). dimer but not the core H3-H4 tetramer, and further suggests that This data strongly suggests that the increased nucleosome sta- H2Bub1 might stabilize the nucleosome by preventing transcrip- bility due to elevated amounts of H2Bub1 impedes cell growth. tion-dependent and -independent H2A-H2B eviction. The ChIP result shows that the lack of H2Bub1 does not affect Discussion histone occupancy on chromatin (Fig. 4A–C). Thus, it seems to contradict the decrease in global histone levels observed by H2Bub1 Does Not Function as a Mere ‘‘Wedge’’ for H3 Methylation. chromatin fractionation (Fig. 3). One possibility is that changes While rejecting the untested ‘‘Wedge’’ model, our result shows in histone amounts in the absence of H2Bub1 were not detected that only ubiquitin, but not SUMO, moiety can support H3 as only a subset of genes was analyzed by ChIP assay. Another methylation and implicates the ‘‘Bridge’’ model in which likely explanation is that the lack of H2Bub1 weakens the H2Bub1 might directly recruit Swd2 to regulate Set1 and Dot1 histone-DNA (or histone-histone) interactions and causes nu- activities. However, since Swd2 or other COMPASS components cleosome instability. Consequently, histones are easily extract- do not contain any of the known ubiquitin-binding domains, an able and lost during the process of nuclei isolation and chromatin alternative model is that the stable association of Swd2 or other fractionation (Fig. 3), and the chromatin becomes susceptible to regulator(s) might depend on a specific chromatin structure increased MNase digestion (Fig. 2). However, formaldehyde adopted following H2B ubiquitination. Supporting this model, cross-linking stabilizes these weakened interactions and prevents H2Bub1 enhances the function of Spt16 and its stable association their detection by ChIP assay. Collectively, the results (Figs. 2–4) with chromatin during elongation (15). In humans, H2Aub1 strongly suggest that H2Bub1 might modulate nucleosome sta- inhibits elongation by preventing the recruitment of FACT to a bility on chromatin during various phases of transcription. subset of genes (27). These findings suggest that it is not the conjugated ubiquitin itself, but the H2Bub1- or H2Aub1- H2Bub1 Is Important for Nucleosome Stability and Regulates Cell modulated chromatin structure that affects the stable association Growth. Next, we performed a histone solubility assay using of FACT. Based on our finding that H2Bub1 affects chromatin CELL BIOLOGY increasing salt concentrations to examine the effect of H2Bub1 structure by increasing nucleosome stability, we propose that the

Chandrasekharan et al. PNAS ͉ September 29, 2009 ͉ vol. 106 ͉ no. 39 ͉ 16689 Downloaded by guest on October 3, 2021 WT ubp8 ubp10 slow growth, likely by impeding ongoing transcription, which is A 1 530 475 505 515 545 590 NaCl (mM) 490 560 575 Input relieved by the removal of H2Bub1 (Fig. 5D). These observations 0.8 raise the question, how does H2Bub1 function in stabilizing Wild type -H3 0.6 0.4 nucleosomes while enhancing Pol II elongation? Both biochem- ubp8 ubp10 -H3 ical and structural studies of the yeast nucleosome have shown Soluble H3 0.2 0 that it is intrinsically less stable than those from other 475 515 545 (26, 28). Also, H2A-H2B is constantly evicted from and reas- 200 mM 300 mM sembled into chromatin in the absence of both DNA replication B 6 and transcription (29). Thus, the newly reassembled nucleo-

NaCl (M) Input 0.2 0.3 0.2 0.3 Input 4 somes immediately adjacent to the transcribing Pol II might be Wild type -H3 Wild type inherently highly unstable, exacerbated by the torsional stress. 2 We propose that H2Bub1 initially stabilizes the nucleosomes to bre1 -H3 H2B-K123R Soluble H3 0 K123R bre1 enhance Spt16 binding to chromatin in front of the polymerase. WT Deubiquitination by Ubp8 might then occur to destabilize the 200 mM 300 mM nucleosomes and enhance the disassembly activity of Spt16 to C 16 allow Pol II progression. Subsequently, during Spt16-mediated

NaCl (M) Input 0.4 0.3 0.2 12 Wild type nucleosome reassembly in the wake of Pol II transcription, 8 H2Bub1 stabilizes the newly deposited nucleosomes. Our model H2B-K123R Flag-H2B 4 provides a mechanistic explanation to not only account for a bre1 Soluble H2B 0 reduction of chromatin-bound Spt16 in the absence of H2ub1, WT K123R bre1 but also for a decrease in H2Bub1 levels resulting from an

WT inefficient nucleosome reassembly by a defective Spt16, as shown D by Fleming and coworkers (15).

H2Bub1 Affects the Stability of Positioned Nucleosomes. While ubp8 ubp10 H2B-K123R H2Bub1 has been shown to be closely associated with transcrip- tion elongation (11, 12), the observed presence of H2Bub1 on repressed promoters suggests a role in regulating transcription initiation (Fig. 4D). Since H2A-H2B is constantly evicted from ubp8 ubp10 H2B-K123R and reassembled into the chromatin of repressed genes even in the absence of DNA replication (29), this probably provides a Fig. 5. H2Bub1 levels affect both the sensitivity of chromatin to salt- window-of-opportunity for Spt15 or activators to bind to their dependent disruption and cell growth. (A) Equal amounts of nuclei isolated cognate sites, allowing basal transcription. Our findings suggest from the indicated strains were resuspended in solutions of increasing NaCl that H2Bub1 enforces gene repression at the transcriptional concentration. Following incubation and centrifugation, amount of H3 in the initiation phase under non-inducing condition by preventing the supernatant (soluble fraction) was analyzed by Western blotting. As the soluble H3 amount in the wild type and the mutant remain the same at high H2A-H2B eviction and stabilizing the nucleosomes positioned NaCl concentrations (Ն 590 mM), the amount of soluble H3 at various salt over the regulatory regions to prevent factor binding. Impor- concentrations (475, 515, or 545 mM) was normalized to the amount obtained tantly, the observed changes in nucleosome stability upon alter- after 615 mM NaCl wash (Input). Graph depicts the average soluble H3 amount ing the H2Bub1 amounts correlate well with its reported effect obtained from two independent experiments. (B) Soluble H3 obtained by on the basal transcription. While ARG1 is derepressed in H2B- resuspension of nuclei in solutions containing low NaCl concentration was K123R (21, 24), its transcript levels are reduced in ubp10⌬ and analyzed as in (A). The amount of soluble H3 obtained following low salt wash ubp8⌬ ubp10⌬ strains (23). Also, the basal expression of PHO5 was normalized to the amount obtained from 2.0 M NaCl wash (Input). Fold is reduced due to excess H2Bub1 (21, 23). change in soluble H3 amounts in the mutants were calculated relative to the In contrast to the repressed genes, histone levels on the normalized soluble H3 amount obtained by washing the wild-type nuclei with 200 mM NaCl (set as 1). Graph depicts the average fold change in soluble H3 promoter regions of the constitutively expressed genes (PMA1 amounts obtained from two independent experiments. (C) Low salt solubility and FUN12) remain unchanged regardless of the status of of H2B was determined as described for H3. Fold change in soluble H2B H2Bub1 (Fig. 4A). This can be attributed to the general deple- amounts from one representative experiment is shown. (D) To assess cell tion of nucleosomes in the promoters of active genes that occurs growth, the indicated yeast strains were grown on rich media at 30 °C. to maintain high levels of expression (30), or alternatively, H2Bub1 does not regulate nucleosome stability and occupancy in this region. The observed differences between the constitu- stabilized nucleosome provides a secure platform for the stable tively expressed and repressed genes suggest distinct roles for association of not only Spt16, but also the methyltransferases, H2Bub1 during various phases of transcription (initiation or their subunits or any other regulator(s) on chromatin. This leads elongation). to a complete assembly and/or stimulation of the activity of Set1-COMPASS and Dot1. How Does H2Bub1 Modulate Nucleosome Stability? Structure studies of ubiquitin have shown that all its charged residues (one-third Role of H2Bub1 in Regulating Nucleosome Dynamics during Transcrip- of the molecule) are on the surface and might constitute tion Elongation. H2Bub1 has been proposed to affect transcrip- potential interacting regions (31). Thus, these charged residues tion elongation and nuclear architecture by regulating the Spt16- of ubiquitin in H2Bub1 might promote its direct contact with the mediated chromatin dynamics (12, 13, 15). Our observation that DNA and/or other core histones. Indeed, in vitro reconstituted only H2B but not H3 levels are increased over the transcribed polynucleosomal fiber using H2Aub1 shows poor solubility as regions of constitutively expressed genes upon increase in compared to those containing only the unmodified H2A, sug- H2Bub1 levels, puts forth the possibility that H2Bub1 stabilizes gesting that the ubiquitin moiety might stabilize the intra- and the nucleosome by retaining H2A-H2B during elongation. This, inter-chromatin array contacts by promoting H2Aub1-DNA in turn would suggest that changes in nucleosome stability due and/or -histone interactions (32). Alternatively, ubiquitin might to altered H2Bub1 levels might affect transcription and cell indirectly affect the histone-DNA and histone-histone interac- growth. Indeed, increased nucleosome stability causes severe tions, particularly, the contact between H2A-H2B dimer and the

16690 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0907862106 Chandrasekharan et al. Downloaded by guest on October 3, 2021 H3-H4 tetramer. For instance, H2Aub1 facilitates the binding of Y131 [MAT a, (hta1-htb1)⌬::LEU2, (hta2-htb2)⌬, leu2–2,-112, ura3–1, trp1–1, linker to reconstituted nucleosomes (33); and its his3–11,-15, ade2–1, can1–100, ssd1, HTA-HTB1 (2␮, URA3)] (provided by M.A. absence releases H1 from chromatin (34). Collectively, all these Osley). Detailed genotypes are listed in Table S1. findings support the view that histone ubiquitination plays an important role in controlling chromatin structure by regulating Chromatin Fractionation and Salt Solubility Assay. Pure nuclei were isolated using a Percoll gradient (see SI Text for details). To obtain chromatin, the pellet nucleosome stability. Importantly, in vitro reconstituted nucleo- following nuclei lysis was resuspended in a buffer by brief sonication. Nor- some containing H2Bub1 shows significantly slower rate of malization of the soluble chromatin amounts was based on DNA quantity, as DNase I digestion compared to those containing only H2B (35), evaluated by spectrophotometry and qPCR. correlating well with our finding that H2Bub1 stabilizes the Equal amounts of nuclei were washed in an extraction buffer containing nucleosome in vivo. increasing NaCl concentration. Following incubation, the nuclei were centri- In summary, by establishing that H2Bub1 enhances nucleo- fuged to obtain soluble fractions. Equal volumes of soluble fractions were some stability, we have defined the molecular bases for the used in Western blotting after concentration or desalting. chromatin dynamics cooperatively regulated by H2Bub1 and Spt16 (15) and for the configuration of ‘‘closed’’ chromatin ACKNOWLEDGMENTS. We thank M. Grunstein (University of California, Los Angeles), B. Strahl (University of North Carolina School of Medicine, Chapel structure in the H2Bub1-enriched, repressed proto-oncogenes Hill, NC), and A. Weil (Vanderbilt University, Nashville, TN) for the antibodies, regulated by RNF20 (the human homolog of yeast Bre1) (36). and D. Gottschling (Fred Hutchinson Cancer Research Center, Seattle), M. A. Osley (University of New Mexico, Albuquerque, NM), A. Weil, and F. Winston Materials and Methods (Harvard Medical School, Boston) for the yeast strains. We thank S. Briggs, S. Hiebert, and B. Strahl for their critical reading of the manuscript, and S. Details of all of the procedures are described in SI Text. Bhaskara, J. Layer, and J. Wang for suggestions and assistance. This research was supported by Vanderbilt-Ingram Cancer Center, The Kleberg Foundation, Yeast Strains and . Yeast strains containing Flag-H2B or its derivative National Cancer Institute SPORE-Breast Cancer (5P0CA098131) and National used in this study were obtained following -shuffle from parental Institutes of Health (RO1CA109355).

1. Weake VM, Workman JL (2008) Histone ubiquitination: Triggering gene activity. Mol 20. Freiberg G, Mesecar AD, Huang H, Hong JY, Liebman SW (2000) Characterization of Cell 29:653–663. novel rad6/ubc2 ubiquitin-conjugating enzyme mutants in yeast. Curr Genet 37:221– 2. Robzyk K, Recht J, Osley MA (2000) Rad6-dependent ubiquitination of histone H2B in 233. yeast. Science 287:501–504. 21. Mutiu AI, Hoke SM, Genereaux J, Liang G, Brandl CJ (2007) The role of histone 3. Briggs SD, et al. (2002) Gene silencing: Trans-histone regulatory pathway in chromatin. ubiquitylation and deubiquitylation in as determined by the analysis 418:498. of an HTB1(K123R) strain. Mol Genet Genomics 277:491–506. 4. Dover J, et al. (2002) Methylation of histone H3 by COMPASS requires ubiquitination 22. Zhang X, et al. (2005) Transcriptional regulation by Lge1p requires a function inde- of histone H2B by Rad6. J Biol Chem 277:28368–28371. pendent of its role in histone H2B ubiquitination. J Biol Chem 280:2759–2770. 5. Ng HH, Xu RM, Zhang Y, Struhl K (2002) Ubiquitination of histone H2B by Rad6 is 23. Gardner RG, Nelson ZW, Gottschling DE (2005) Ubp10/Dot4p regulates the persistence required for efficient Dot1-mediated methylation of histone H3 lysine 79. J Biol Chem of ubiquitinated histone H2B: Distinct roles in telomeric silencing and general chro- 277:34655–34657. matin. Mol Cell Biol 25:6123–6639. 6. Sun ZW, Allis CD (2002) Ubiquitination of histone H2B regulates H3 methylation and 24. Turner SD, et al. (2002) The E2 ubiquitin conjugase Rad6 is required for the ArgR/Mcm1 gene silencing in yeast. Nature 418:104–108. repression of ARG1 transcription. Mol Cell Biol 22:4011–4019. 7. Henry KW, Berger LB (2002) Trans-tail histone modifications: Wedge and bridge? 25. Emre NC, et al. (2005) Maintenance of low histone ubiquitylation by Ubp10 correlates Nature Struc 9:565–566. with telomere-proximal Sir2 association and gene silencing. Mol Cell 17:585–594. 8. Lee JS, et al. (2007) Histone crosstalk between H2B monoubiquitination and H3 26. Pin˜eiro M, Puerta C, Palacia´n E (1991) Yeast nucleosomal particles: Structural and methylation mediated by COMPASS. Cell 131:1084–1096. transcriptional properties. Biochemistry 30:5805–5810. 9. Vitaliano-Prunier A, et al. (2008) Ubiquitylation of the COMPASS component Swd2 27. Zhou W, et al. (2008) Histone H2A monoubiquitination represses transcription by links H2B ubiquitylation to H3K4 trimethylation. Nat Cell Biol 10:1365–1371. inhibiting RNA polymerase II transcriptional elongation. Mol Cell 29:69–80. 10. Nedea E, et al. (2008) The Glc7 phosphatase subunit of the cleavage and polyadenyl- 28. White CL, Suto RK, Luger K (2001) Structure of the yeast nucleosome core particle ation factor is essential for transcription termination on snoRNA genes. Mol Cell reveals fundamental changes in internucleosome interactions. EMBO J 20:5207–5218. 29:577–587. 29. Jamai A, Imoberdorf RM, Strubin M (2007) Continuous histone H2B and transcription- 11. Xiao T, et al. (2005) Histone H2B ubiquitylation is associated with elongating RNA dependent histone H3 exchange in yeast cells outside of replication. Mol Cell 25:345– polymerase II. Mol Cell Biol 25:637–651. 355. 12. Pavri R, et al. (2006) Histone H2B monoubiquitination functions cooperatively with 30. Giresi PG, Gupta M, Lieb JD (2006) Regulation of nucleosome stability as a of FACT to regulate elongation by RNA polymerase II. Cell 125:703–717. chromatin function. Curr Opin Genet Dev 16:171–176. 13. Tanny JC, Erdjument-Bromage H, Tempst P, Allis CD (2007) Ubiquitylation of histone 31. Vijay-Kumar S, et al. (1987) Comparison of the three-dimensional structures of human, H2B controls RNA polymerase II transcription elongation independently of histone H3 yeast, and oat ubiquitin. J Biol Chem 262:6396–6399. methylation. Genes Dev 21:835–847. 32. Jason LJ, Moore SC, Ausio J, Lindsey G (2001) Magnesium-dependent association and 14. Belotserkovskaya R, et al. (2003) FACT facilitates transcription-dependent nucleosome folding of oligonucleosomes reconstituted with ubiquitinated H2A. J Biol Chem alteration. Science 301:1090–1093. 276:14597–14601. 15. Fleming AB, Kao CF, Hillyer C, Pikaart M, Osley MA (2008) H2B ubiquitylation plays a 33. Jason LJ, Finn RM, Lindsey G, Ausio´ J (2005) Histone H2A ubiquitination does not role in nucleosome dynamics during transcription elongation. Mol Cell 31:57–66. preclude histone H1 binding, but it facilitates its association with the nucleosome. J Biol 16. Nathan D, et al. (2006) Histone sumoylation is a negative regulator in Saccharomyces Chem 280:4975–4982. cerevisiae and shows dynamic interplay with positive-acting histone modifications. 34. Zhu P, et al. (2007) A histone H2A deubiquitinase complex coordinating histone Genes Dev 20:966–976. and H1 dissociation in transcriptional regulation. Mol Cell 27:609–621. 17. Kao CF, et al. (2004) Rad6 plays a role in transcriptional activation through ubiquity- 35. Davies N, Lindsey GG (1994) Histone H2B (and H2A) ubiquitination allows normal lation of histone H2B. Genes Dev 18:184–195. and core particle reconstitution. Biochim Biophys Acta 1218:187–193. 18. Thoma F (1996) Mapping of nucleosome positions. Methods Enzymol 274:197–214. 36. Shema E, et al. (2008) The histone H2B-specific ubiquitin ligase RNF20/hBRE1 acts as a 19. Liang C, Stillman B (1997) Persistent initiation of DNA replication and chromatin-bound putative tumor suppressor through selective regulation of gene expression. Genes Dev MCM proteins during the in mutants. Genes Dev 11:3375–3386. 22:2664–2676. CELL BIOLOGY

Chandrasekharan et al. PNAS ͉ September 29, 2009 ͉ vol. 106 ͉ no. 39 ͉ 16691 Downloaded by guest on October 3, 2021