Histone H3 Lysine 36 Methylation Targets the Isw1b Remodeling Complex to Chromatin

Abstract

Histone H3 lysine 36 methylation is a ubiquitous hallmark of productive transcription elongation. Despite the prevalence of this histone posttranslational modification, however, the downstream functions triggered by this mark are not well understood. In this study, we showed that H3K36 methylation promoted the chromatin interaction of the Isw1b chromatin-remodeling complex in Saccharomyces cerevisiae. Similar to H3K36 methylation, Isw1b was found at the mid- and 3′ regions of transcribed genes genome wide, and its presence at active genes was dependent on H3K36 methylation and the PWWP domain of the Isw1b subunit, Ioc4. Moreover, purified Isw1b preferentially interacted with recombinant nucleosomes that were methylated at lysine 36, and this interaction also required the Ioc4 PWWP domain. While H3K36 methylation has been shown to regulate the binding of numerous factors, this is the first time that it has been shown to facilitate targeting of a chromatin-remodeling complex.

INTRODUCTION

Chromatin, a nucleoprotein structure that packages DNA in all eukaryotes, serves as a barrier to numerous nuclear processes, such as transcription, replication, and DNA repair. Numerous mechanisms exist for alteration of chromatin structure, including histone posttranslational modifications, the incorporation of histone variants, and the activity of chromatin-remodeling complexes. There are many examples of these processes functioning in the same pathway, such as histone acetylation, which promotes the binding of the RSC and SWI/SNF remodeling complexes to chromatin (7, 11).

Chromatin-remodeling complexes are multisubunit complexes that use the hydrolysis of ATP to disrupt histone-DNA contacts, promoting octamer sliding and histone eviction. Isw1b, a chromatin-remodeling complex found in Saccharomyces cerevisiae, consists of three subunits: Isw1, Ioc2, and Ioc4 (31). This complex localizes to the transcribed regions of genes, where it has been implicated in transcription elongation, termination, and pre-mRNA processing (19). While there have been some insights into how the Isw1b complex functions, how it is targeted to specific regions of the genome is unknown.

Transcriptionally active genes are packaged with histones that are mono-, di-, or trimethylated on lysines 4 and 36 of histone H3. Lysine 4 methylation is catalyzed by Set1 as part of the COMPASS complex, which is targeted to transcribed genes through interaction with RNA polymerase II (RNAP II) during the early stages of transcription elongation (26). Lysine 36 methylation is generated by the histone methyltransferase Set2, which similarly interacts with elongating RNAP II, albeit at later stages of elongation. Consistent with the differential interactions of Set1 and Set2 with the various forms of elongating RNAP II, lysine 4 trimethylation (H3K4me3) is predominantly found at the 5′ ends of transcribed genes, while lysine 36 di- and trimethylation (H3K36me2 and H3K36me3) are found in the mid- and 3′ regions (26).

Histone methylation is proposed to act as a docking site for proteins that modify chromatin structure. Several protein domains have been identified as methyl-lysine binding motifs, including the chromo-, Tudor, MBT, WD40, PWWP, and PHD finger domains (17). With the exception of the PHD finger, all of these domains are members of the Tudor domain “Royal Family,” which are characterized by hydrophobic cavities made up of 2 to 4 aromatic residues that bind methylated ligands (17). In yeast, numerous proteins have been shown to specifically interact with methylated H3K4; however, only one protein, Eaf3, has been demonstrated to preferentially interact with methylated histone H3K36 (3, 10, 12, 16, 29, 34). Thus, the downstream functions triggered by H3K36 methylation are largely unexplored.

Several lines of evidence suggest that methylated H3K36 targets Isw1b to transcribed genes. First, Isw1b is localized to the 3′ end of numerous actively transcribed genes, which correlates with the positions of H3K36me2 and K36me3 on these genes (2, 19, 22). Second, Ioc4 contains a PWWP domain, which in other proteins has been reported to bind methylated H3K36 (6, 33). In this study, we confirm that, genome wide, Isw1b is found associated with regions that are enriched in methylated H3K36, and the interaction of Isw1b with chromatin is dependent on the PWWP domain of Ioc4, the H3K36 methyltransferase (Set2), and lysine 36. Moreover, purified Isw1b directly interacts with recombinant nucleosomes that are methylated at lysine 36, and this interaction is disrupted by mutation of the Ioc4 PWWP domain. These studies reveal how this chromatin-remodeling complex is targeted to transcribed genes.

MATERIALS AND METHODS

Yeast strains and plasmids.

All S. cerevisiae strains used in this study are isogenic to S288C and are listed in Table 1. Yeast culture and genetic manipulations were performed using standard protocols (1). Genomic deletions were verified by PCR, and whole-cell extracts were generated as previously described (13).

Table 1.

Yeast strains used in this study

Strain	Mating type	Genotype
YIJ86	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 LEU2::IOC4HA6
YIJ87	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 set1::HISMX6 LEU2::IOC4HA6
YIJ88	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 set2::KANMX6 LEU2::IOC4HA6
YVM146	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 SAS3HA6::TRP
YVM147	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 SAS3HA6::TRP yng1ΔPHD::KANMX6
YVM157	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 SAS3HA6::TRP set1::HISMX6
YVM243	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 (hht1-hhf1)::LEU (hht2-hhf2)::KAN IOC4HA6::HISMX6 pHHF2.hht2K36A
YVM244	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 (hht1-hhf1)::LEU (hht2-hhf2)::KAN IOC4HA6::HISMX6 pHHF2.hht2K36R
YVM245	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 (hht1-hhf1)::LEU (hht2-hhf2)::KAN IOC4HA6::HISMX6 pHHF2.HHT2
YVM268	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ioc4::KANMX6 LEU2::IOC4HA6
YVM269	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ioc4::KANMX6 LEU2::ioc4F18AHA6
YVM270	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ioc4::KANMX6 LEU2::ioc4W21AHA6
YVM271	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ioc4::KANMX6 LEU2::ioc4F114AHA6
YVM272	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 IOC4TAP::TRP
YVM273	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ISW1TAP::TRP ioc4::KANMX6 LEU2::IOC4HA6
YVM275	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ISW1TAP::TRP ioc4::KANMX6 LEU2::ioc4F18AHA6
YVM276	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ISW1TAP::TRP ioc4::KANMX6 LEU2::ioc4W21AHA6
YVM277	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 ISW1TAP::TRP ioc4::KANMX6 LEU2::ioc4F114AHA6
YVM291	mata	his3Δ200 leu2Δ1 lys2-128d ura3-52 trp1Δ63 IOC4TAP::TRP ioc2::HISMX6

ChIP-qPCR and ChIP-on-chip.

Chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR) and ChIP-on-chip analyses were performed as previously described (20, 25). To plot the average Ioc4HA profile, we used an approach similar to a previously published analysis (22, 25). All open reading frames (ORFs) (derived from the SGD database [www.yeastgenome.org]) were aligned according to the location of the translation initiation and termination sites. Each ORF was divided into 40 bins of equal length, and the average enrichment values were calculated for each bin. The 500 bp upstream and downstream of the coding start and end, respectively, were assigned to 20 bins, and the average enrichment value for each bin was plotted as the 5′ and 3′ intergenic regions. The CHROMATRA-T plot for Ioc4 ChIP-on-chip was generated as described previously (8).

Purification of Isw1b with or without associated histones.

Isw1b, together with associated histones, was purified from a strain expressing tandem affinity purification (TAP)-tagged Ioc4 as previously described (14), with the exception that after preparation, extracts were clarified by centrifugation at 12,000 × g for 5 min at 4°C. The posttranslational modification status of the Isw1b-associated histones was analyzed using antibodies specific for H3K4me3 (ab1012; Abcam), H3K36me3 (07-549; Upstate), and H3 (produced by Genscript to peptide CKDIKLARRLRGERS). Isw1b without associated histones was purified using calmodulin affinity chromatography from either the Isw1TAP or Ioc4TAP strain. Extracts prepared in EX350 buffer (50 mM HEPES, pH 7.5, 350 mM NaCl, 0.1% NP-40, 10% glycerol) with 2 mM CaCl₂ were incubated with calmodulin affinity resin (Stratagene) overnight at 4°C. The resin was washed two times with 10 bed volumes of EX350 buffer with CaCl₂ and one time in EX350 lacking CaCl₂ prior to elution in 10 bed volumes of EX350 with 2 mM EGTA.

Preparation of immobilized methyl-lysine mimic mononucleosomes.

Yeast histone octamers with methyl-lysine mimics incorporated at residue 36 of histone H3 were prepared as described previously (27). The Lytechinus variegatus 5S ribosomal DNA (rDNA) sequence was amplified from an EcoRI-digested p207-12 plasmid (28) using primers AATTCCAACGAATAACTTCCAGGG and GCGGTATTCCCAGGCGGTC, the latter of which carried a 5′ biotin tag. The biotinylated DNA (1.8 μg) was added to a tube containing 3 μg of NAP1 (Millipore) and 1.2 μg of histone octamers in 10 mM HEPES, pH 7.6, 50 mM KCl, 5 mM MgCl₂, 0.5 mM EGTA, 0.1 mM EDTA, 10% glycerol, 0.1 mg/ml bovine serum albumin (BSA) to a total volume of 50 μl. Reconstitution was allowed to proceed for 6 h at 30°C prior to immobilizing the reconstituted chromatin on streptavidin-Sepharose resin (GE Healthcare Life Sciences). To assay Isw1b-nucleosome interactions, purified Isw1b was rotated with immobilized nucleosomes in 50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% NP-40 with 3 μg of chicken mononucleosomes (a generous gift of Juan Ausio) overnight at 4°C. Beads were washed 3× with 50 bed volumes of binding buffer, and the proteins were eluted by boiling in SDS-PAGE buffer.

RESULTS

Isw1b was found at the midregions and 3′ ends of transcriptionally active genes.

Isw1, Ioc2, and Ioc4 were shown to localize to the 3′ ends of a small set of transcribed candidate genes, including MET16, PMA1, IMD2, and PKA1 (19). To determine whether 3′ ORF localization is a general characteristic of Isw1b, we performed a ChIP-on-chip experiment. Protein-DNA complexes containing a 6× hemagglutinin (HA)-tagged version of Ioc4 were immunoprecipitated from a wild-type strain, and the associated DNA was labeled and used to probe Affymetrix Gene Chip Yeast Genome 2.0 arrays. Enriched regions were detected as previously described by comparing signal intensities of the ChIP to input DNA (25). The results showed that Ioc4 was depleted from the promoter and 5′ end of genes, enriched at the mid-coding region, and tapered off in the 3′ intergenic region (Fig. 1A). ChIP-on-chip performed using an untagged strain did not show a similar pattern, indicating that this profile was specific to Ioc4 (data not shown).

Fig 1.

Isw1b was enriched at the midregions and 3′ ends of transcriptionally active genes. (A) ChIP-on-chip analysis was performed to analyze localization of Ioc4HA genome wide. All open reading frames, independent of length, were split into 40 bins (see Materials and Methods). The average Ioc4HA model-based analysis of tiling arrays (MAT) scores for each bin were calculated and plotted. The gray bar represents the coding region, and the dashed lines represent the translation start and end sites. The 5′ and 3′ intergenic regions (IGRs) represent 500 bp up- and downstream of the translation start and end sites, respectively. B) CHROMATRA-T plot for Ioc4 ChIP-on-chip. MAT scores were binned into segments of 150 bp, and the average enrichment value for each bin is color coded and plotted as a heat map. Genes were ordered by gene length and grouped into five classes according to their number of transcripts per hour (9). (C) Genes enriched for Ioc4 were more frequently transcribed than all other genes. The all-genes column shows all yeast genes binned based on transcriptional frequency (9), and the Ioc4-enriched column shows genes with at least 50% of all probes over the open reading frame having an enrichment score above a threshold of 1.5.

To examine the correlation between Ioc4 levels and the transcription rates of genes, the CHROMATRA (for chromatin mapping across transcripts) visualization tool was used (8). This tool allows visualization of ChIP-on-chip enrichment scores for all genes in yeast as a heat map after binning first by transcription frequency and second by gene length. Figure 1B shows that in addition to reiterating the midregion and 3′ localization of Ioc4 across many genes, the results showed that Ioc4 exhibited increased abundance on the most highly transcribed genes. To further demonstrate this observation, a set of genes most enriched for Ioc4 was identified. A gene was termed Ioc4 enriched if at least 50% of all probes over the open reading frame had an enrichment score above a threshold of 1.5. Based on the published transcriptional frequencies of all yeast genes (9, 21), Ioc4-enriched genes were transcribed at higher average rates than all other genes (P = 0.0001) (Fig. 1C).

Isw1b colocalized with H3K36 methylation.

Increased levels of Ioc4 at the midregions and 3′ ends of highly transcribed genes is reminiscent of H3K36 methylation, the abundance of which also increases with transcription rate (22). To provide additional support for Isw1b interacting with regions containing high levels of H3K36 methylation, we purified Isw1b and examined whether the copurifying histones were enriched in H3K36me3. Isw1b, purified from an Ioc4TAP strain, was subjected to immunoblot analysis using antibodies specific for H3K36me3 or the carboxyl terminus of histone H3. Since these two antibodies were generated in different species, we were able to simultaneously analyze the signals from the two antibodies on the same blot and overlay the signals to determine the level of methylation relative to total histone H3. Figure 2A shows that while the H3 (green) and H3K36me3 (red) signals in the input whole-cell extracts were equal (i.e., yellow when overlaid), the histones associated with purified Isw1b were red when overlaid, indicating enrichment in H3K36me3 in these histones. In contrast, H3K4me3 was not enriched in the Isw1b-associated histones (i.e., they were still yellow when overlaid) (Fig. 2B). Quantification of these data from four biological replicates demonstrated that these differences were statistically significant (P = 0.0029) (Fig. 2C). Collectively, the results in Fig. 1 and 2 indicated that Isw1b preferentially interacts with regions containing high levels of H3K36-methylated histones.

Fig 2.

Isw1b preferentially interacted with H3K36-methylated chromatin. (A and B) Isw1b was purified from a whole-cell extract (WCE) of a strain expressing TAP-tagged Ioc4 (+), and the associated histones were subjected to immunoblot analysis. An untagged strain (−) was used as a negative control. Primary antibodies against histone H3 and H3K36me3 or H3K4me3 were used as indicated. Signals were generated using IRDye-labeled secondary antibodies and overlaid. (C) Immunoblots shown in A and B were repeated in four independent experiments and the results quantified. Error bars indicate means ± standard errors of the means (SEM). An unpaired t test comparing the relative enrichment of H3K4me3 and H3K36me3 in the Ioc4-associated histones revealed a P value of 0.0029. (D) Isw1b was purified from IOC2 and ioc2Δ strains expressing TAP-tagged Ioc4 (+), and the associated histones were subjected to immunoblot analysis for histone H3. An untagged strain (−) was used as a negative control.

To determine whether Ioc4 was binding histones on its own or in the context of Isw1b, we tested whether another subunit of Isw1b was required for histone binding. Ioc2 is a unique subunit of Isw1b, and deletion of IOC2 disrupts the interaction of Ioc4 with Isw1 (31). Deletion of IOC2 resulted in decreased levels of histones copurifying with Ioc4 (Fig. 2D), indicating that Ioc4 was binding histones in the context of Isw1b and that the remaining subunits of Isw1b played a role in the histone binding activity of the complex. This is not unexpected, as many methyl-histone binding modules require accessory complex subunits to bind chromatin (5, 15, 24).

H3K36 methylation was required for Isw1b-chromatin interaction.

To determine whether Isw1b was dependent on H3K36 methylation for chromatin interaction, we performed ChIP-qPCR for HA-tagged Ioc4. Previously, Ioc4 was shown to be present at the actively transcribed MET16 gene, with the greatest enrichment of Ioc4 at the 3′ end of the gene (Fig. 3A; also see reference 18). Figure 3B shows that association of Ioc4 with chromatin was compromised in a set2Δ mutant, despite only minor changes in total cellular Ioc4 levels (Fig. 3C). In contrast, disruption of the histone H3 lysine 4 methyltransferase, Set1, resulted in an unexplained increase in Ioc4 at MET16. It was unlikely that the reduced levels of Ioc4 at the 3′ end of MET16 in the SET2 mutant were due to a MET16 transcriptional defect, since the levels of the Met16 protein were not decreased (Fig. 3C).

Fig 3.

SET2, which encodes the sole H3K36 methyltransferase, was required for Ioc4 association with chromatin. (A) The levels of Ioc4HA at the 5′ and 3′ ends of the active MET16 gene were measured relative to input by ChIP-qPCR. IgG represents negative-control ChIPs with rabbit IgG antibody in lieu of HA antibody. (B) Ioc4HA ChIP in the indicated strains at the 3′ end of an active MET16 gene. Results are shown relative to the wild type (WT), which was set to 1. (C) Quantitative immunoblots for levels of TAP-tagged Met16 or HA-tagged Ioc4 in the indicated strains, with the WT set to 1. For all experiments, error bars indicate the means ± SEM from three independent experiments.

To confirm that Ioc4 binding was dependent on H3K36 methylation, Ioc4 was HA tagged in strains expressing H3K36R or H3K36A mutants as the sole form of histone H3. Figure 4A shows that mutation of H3 lysine 36 resulted in loss of Ioc4 from the 3′ end of MET16, despite a significant increase in Ioc4 levels in cell extracts (Fig. 4B). Although these results confirm the importance of H3K36 for the interaction of Ioc4 with MET16, it is unknown why mutation of histone H3 results in such a significant increase in total Ioc4 levels. One explanation is that disruption of Ioc4-chromatin interaction renders Ioc4 more soluble to our protein extraction protocol. This is unlikely, however, since other mutations that disrupt Ioc4-chromatin interaction do not have the same effect (Fig. 3 and 4).

Fig 4.

Histone H3K36 was required for Ioc4 association with chromatin. (A) The levels of Ioc4HA at the 3′ end of the active MET16 gene were measured relative to input by ChIP-qPCR in the indicated strains. Results are shown relative to the WT, which was set to 1. (B) Quantitative immunoblots for TAP-tagged Met16 or HA-tagged Ioc4 in the indicated strains. (C) The levels of Sas3HA at the 5′ end of the COX10 gene were measured relative to input by ChIP-qPCR in the indicated strains. Results are shown relative to the WT. (D) Purified Isw1b was incubated with immobilized mononucleosomes generated from recombinant histones with the indicated methyl-lysine analogs at H3K36. Samples were immunoblotted for TAP-tagged Ioc4 or histone H3 as indicated. Input represents 10% of the Isw1b added to the binding assays. For panels A to C, error bars indicate the means ± SEM from three independent experiments.

Another interesting observation is that mutation of H3K36 has a much more severe effect on the ChIP of Ioc4 than mutation of SET2. We find this to be a common theme in our research and believe that it reflects the fact that although methyl-histone binding domains show a preference for methylated histones, they still have affinity for the unmethylated sites. In contrast, mutations to the methylated lysine or the methyl-histone binding domain have a much more severe effect on chromatin interaction. An example of this is shown in Fig. 4C. The PHD finger of Yng1, a component of the NuA3 histone acetyltransferase (HAT) complex, is required for the interaction of NuA3 with H3K4-methylated histones. While deletion of SET1, which encodes the H3K4 methyltransferase, compromises the interaction of the NuA3 subunit, Sas3, with the COX10 gene, deletion of the PHD finger has a much more severe effect (Fig. 4C).

To further confirm the preference of Isw1b for H3K36-methylated histones, we asked whether Isw1b could directly interact with lysine 36-methylated histone H3 in vitro. To this end, we purified Isw1b from yeast and examined the binding of this complex to modified mononucleosomes immobilized on beads. Figure 4D shows that while Isw1b was still able to interact with unmodified nucleosomes as suggested above, the chemical modification of H3 to generate functional di- or trimethyl-lysine analogs at lysine 36 enhanced the binding of Isw1b to nucleosomes. Collectively these results supported our hypothesis that Isw1b interaction with chromatin was dependent on lysine 36 methylation.

The Ioc4 PWWP domain was required for Isw1b-chromatin interaction.

Recent work investigating the histone binding properties of the human protein BRPF1, a subunit of the MOZ/MORF HAT complexes, revealed that this protein specifically interacts with H3K36-methylated histone tails via its PWWP domain (32). To investigate whether the PWWP domain of Ioc4 could serve a similar function, we performed a CLUSTALW2 alignment of the region surrounding the PWWP motifs of BRPF1, Ioc4, and Pdp1, another PWWP domain-containing protein with methyl-lysine binding ability (33). In BRPF1, the tri-methyl group of H3K36me3 fits into a pocket formed by Y1096, Y1099, and F1147, and Fig. 5A shows analogous hydrophobic amino acids in the PWWP domain of Ioc4 (F18, W21, and F114, indicated by arrows and shading) (32). To determine whether these residues were required for the interaction of Ioc4 with chromatin, we generated strains expressing HA-tagged Ioc4 with alanine substitutions at these sites. After confirming that these mutations did not affect the interaction of Ioc4 with Isw1 (Fig. 5B), we subjected the strains to ChIP-qPCR for the HA tag. Figure 5C shows that Ioc4 levels are significantly reduced at the 3′ end of MET16 in the PWWP domain mutants, indicating that this domain is important for the interaction of Ioc4 with chromatin.

Fig 5.

The PWWP domain of Ioc4 was required for the interaction of Isw1b with chromatin. (A) ClustalW2 multiple-sequence alignments of the PWWP motif of Ioc4 with other known methyl-histone binding PWWP motifs from S. cerevisiae (S.c.), S. pombe (S.p.), and human (H.s.). Residues predicted to form the methyl-lysine binding pocket are highlighted in gray. Arrows indicate residues subjected to site-directed mutagenesis. Asterisks indicates positions that have a single, fully conserved residue, a colon indicates conservation between groups of strongly similar properties, and a period indicates conservation between groups of weakly similar properties. (B) Whole-cell extracts (WCE) and TAP-purified Isw1b from strains expressing Isw1TAP (+) and wild-type or mutant versions of HA-tagged Ioc4 were immunoblotted for levels of Isw1 and Ioc4. A untagged strain (−) was used as a negative control. (C) ChIP for Ioc4 at the 3′ end of the active MET16 gene in strains expressing wild-type or mutant versions of Ioc4HA was measured relative to input by qPCR. Results are shown relative to a WT strain, which was set to 1. Error bars indicate the means ± SEM from three independent experiments. (D) Isw1b was TAP purified from strains expressing Isw1TAP and either wild-type (WT) or mutant (Ioc4W21A) HA-tagged Ioc4. The purified complex was incubated with immobilized mononucleosomes generated from recombinant histones with a trimethyl-lysine analog at H3K36. Samples were immunoblotted for HA-tagged Ioc4 or histone H3 as indicated. Input represents 10% of the Isw1b added to the binding assay.

As a final confirmation of the importance of the Ioc4 PWWP domain for chromatin interaction, we TAP purified Isw1b from a strain expressing Ioc4 with a mutation in the PWWP domain (W21A) and tested the ability of this complex to bind recombinant, methylated nucleosomes in vitro. Figure 5D shows that while Isw1b containing wild-type Ioc4 is able to bind to methylated chromatin, mutation of W21 disrupts this interaction. Taken together with the H3K36 methylation dependence of Isw1b-chromatin interaction, these results support our hypothesis that Isw1b localization to the mid- and 3′ regions of transcribed genes is mediated in part via the interaction of the PWWP domain of Ioc4 with K36-methylated histone H3.

DISCUSSION

In this study, we demonstrated that the Isw1b complex (Isw1/Ioc2/Ioc4) primarily localizes to the midregions and 3′ ends of transcriptionally active genes, where it is targeted, in part, via the interaction of the Ioc4 PWWP domain with methylated histone H3 lysine 36. These results are consistent with several previous studies. First, it was shown that while Isw1a (Isw1/Ioc3) and Isw1b complexes have equivalent ATPase activities, Isw1a exhibits stronger nucleosome spacing and sliding activities in vitro (31). As proposed in that study and confirmed in this work, this is likely because Isw1b shows a preference for posttranslationally modified nucleosomes. Second, deletion of SET2, which encodes the sole yeast H3K36 histone methyltransferase, was shown to compromise the association of Isw1 with the 3′ end of the MET16 gene (18). However, Isw1 exists as a monomer as well as a component of the Isw1a and Isw1b complexes (31); thus, it was not known which form of Isw1 is dependent on H3K36 methylation until our work. Interestingly, the same study demonstrated that both Isw1 and H3K36me3 appear transiently at the 5′ region of MET16 during an early, regulatory phase of transcription (18). Therefore, an intriguing possibility is that H3K36 methylation also targets Isw1b to alternate loci during other phases of transcription.

Our work suggests that Isw1b and H3K36 methylation function in the same pathway, which may reveal novel information about the function of Isw1b. H3K36 methylation left in the wake of elongating RNAP II promotes the deacetylation of histones by the Rpd3S complex and, in doing so, restores chromatin structure following transcription (3, 10, 12). Loss of this pathway results in increased acetylation in the coding regions of genes and transcription initiation from cryptic intragenic promoters. Therefore, an intriguing hypothesis is that Isw1b is targeted in a fashion similar to that of Rpd3S to maintain a repressive chromatin environment that is refractory to transcription initiation in the midregions and 3′ ends of transcribed genes. The Isw1a complex similarly has been shown to block transcription initiation at promoters through the specific positioning of promoter-proximal nucleosomes (19). Consistent with this hypothesis, several studies have shown that loss of ISW1 results in cryptic initiation at multiple genes (4, 23, 30). We, however, have been unable to detect cryptic initiation at the LYS2 or FLO8 gene in IOC4 mutants (data not shown), so whether the aberrant initiation observed in ISW1 mutants is due to loss of Isw1b is unknown. Thus, although we have been able to show a role for H3K36 methylation in targeting the Isw1b complex to transcribed regions, the function of this complex, once recruited, awaits further exploration.