Histone Deacetylases and Phosphorylated Polymerase II C-Terminal Domain Recruit Spt6 for Cotranscriptional Histone Reassembly

Abstract

Spt6 is a multifunctional histone chaperone involved in the maintenance of chromatin structure during elongation by RNA polymerase II (Pol II). Spt6 has a tandem SH2 (tSH2) domain within its C terminus that recognizes Pol II C-terminal domain (CTD) peptides phosphorylated on Ser2, Ser5, or Try1 in vitro. Deleting the tSH2 domain, however, only has a partial effect on Spt6 occupancy in vivo, suggesting that more complex mechanisms are involved in the Spt6 recruitment. Our results show that the Ser2 kinases Bur1 and Ctk1, but not the Ser5 kinase Kin28, cooperate in recruiting Spt6, genome-wide. Interestingly, the Ser2 kinases promote the association of Spt6 in early transcribed regions and not toward the 3′ ends of genes, where phosphorylated Ser2 reaches its maximum level. In addition, our results uncover an unexpected role for histone deacetylases (Rpd3 and Hos2) in promoting Spt6 interaction with elongating Pol II. Finally, our data suggest that phosphorylation of the Pol II CTD on Tyr1 promotes the association of Spt6 with the 3′ ends of transcribed genes, independently of Ser2 phosphorylation. Collectively, our results show that a complex network of interactions, involving the Spt6 tSH2 domain, CTD phosphorylation, and histone deacetylases, coordinate the recruitment of Spt6 to transcribed genes in vivo.

INTRODUCTION

Dynamic reorganization of chromatin structure is important for the regulation of transcription by RNA polymerase II (Pol II). Several ATP-dependent chromatin remodelers, histone-modifying enzymes, and histone chaperones promote the disruption of chromatin structure to allow transcription by Pol II and to restore chromatin structure in the wake of transcription (1, 2). Spt6 is a highly conserved and multifunctional protein shown to regulate multiple steps of transcription, including initiation (3, 4) and elongation (5, 6). In addition, it is important for other DNA-dependent processes such as recombination (7, 8), mRNA export (9), and viral replication (10).

Spt6 interacts with histones H3 and H4 (11) and functions as a histone chaperone to regulate cotranscriptional nucleosome reassembly and to modulate chromatin structure, including histone modifications (11–15). Depletion of Spt6 in yeast (Saccharomyces cerevisiae) causes a widespread reduction in histone H3 primarily from the 5′ ends of transcribed genes, indicating the importance of Spt6 in maintaining histone occupancy (16). One of the effects of losing Spt6 function is altered gene expression and activation of intragenic cryptic transcription (17–19). Aberrant transcription is also associated with histone deacetylase (HDAC) mutants. Impairing the function of the HDAC Rpd3-small (Rpd3S) and Hos2-Set3 complexes leads to cryptic and antisense transcription genome-wide (20, 21). It is not clear whether Spt6 and HDACs collaborate to suppress spurious transcription.

Spt6 localization to transcribed regions strongly correlates with Pol II occupancy (15, 22). It possesses a tandem Src homology 2 domain (tSH2) that interacts with phosphorylated Pol II C-terminal domain (CTD) peptides in vitro and is required for full recruitment of Spt6 in vivo (22–26). Although these observations suggest that CTD phosphorylation is needed for Spt6 recruitment, the role of CTD phosphorylation in vivo in this process is not fully understood. Notably, while the two phosphorylated CTD residues implicated in Spt6 recruitment, Ser2 and Tyr1, peak near the 3′ ends of transcribed genes, Spt6 recruitment begins slightly downstream of the transcription start site (TSS) (22, 23). Moreover, deletion of the primary Ser2 CTD kinase CTK1 does not affect Spt6 recruitment in vivo (27). To account for these observations, it has been suggested that initial recruitment of Spt6 near TSS occurs in a Ser2 phosphorylation-independent manner. It is therefore not evident whether Ser2 or Try1 phosphorylation is important for Spt6 recruitment in vivo and whether they cooperate or function independently in this process. In order to resolve these inconsistencies in the role of CTD phosphorylation in Spt6 recruitment, we systematically analyzed its genome-wide occupancy in Pol II CTD kinase mutants.

We report here that Ser2 kinases Bur1 and Ctk1, but not the Ser5 kinase Kin28, cooperate to recruit Spt6 near the TSS. Interestingly, we show that HDACs are required to maintain high Spt6 occupancy across the transcribed region by stabilizing Spt6 interaction with phosphorylated Pol II. This function in stimulating Spt6 recruitment by HDACs in the coding regions is independent of methylation-dependent histone deacetylation. Furthermore, our data suggest that recruitment of Spt6 near 3′ regions of transcribed genes is promoted by Tyr1 phosphorylation in a manner independent of Ser2/Ser5 kinases. We therefore propose that CTD serine kinases and HDACs coordinate the recruitment of Spt6 to the transcribed region. In addition, our data show that the recruitment of Spt6 at the very 3′ end of a transcribed gene is tSH2 mediated and Ser2 phosphorylation (Ser2P) independent, possibly occurring through CTD-Tyr1 phosphorylation.

MATERIALS AND METHODS

Yeast strains.

All of the yeast strains used in the present study are listed in Table 1. The wild-type (WT) BY4741 and deletion derivatives (32), including SPT6-tet (33), were purchased from Thermo Scientific, USA. Myc-tagged strains were generated as described previously (34). Briefly, plasmid pFA6a-13Myc-His3MX6 was used as a template to PCR amplify using primers with flanking nucleotides homologous to upstream and downstream regions of the stop codon of Spt6. The last 202 amino acid residues of Spt6 were deleted by inserting 13Myc at residue 1250 of Spt6 to generate the spt6Δ202-Myc-tagged strain. The primers used are listed in Table 2.

TABLE 1.

Yeast strains used in this study

Strain	Parent	Genotype	Source or reference
Untagged strains
BBY 1205 (WT Ser5 kinases)	S288C	MATα ade2::hisG his3Δ200 leu2Δ0 lys2Δ0 met15Δ0 trp1Δ63 ura3Δ0	29
BBY 1207 (kin28as/bur2Δ)	S288C	MATα ade2::hisG his3Δ200 leu2Δ0 lys2Δ0 met15Δ0 trp1Δ63 ura3Δ0 kin28::kin28-L83G Δbur2::Leu2 [pSH579 ARS CEN URA3 kin28-L83G]	29
BBY 1107 (WT Ser2 kinases)	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0	28
BBY 1110 (bur1as/ctk1Δ)	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 bur1as ctk1Δ::kanMX4	28
BBY 1103 (WT HDACs)	BY 4705	MATα ade2Δ::hisG HIS3Δ200 LEU2Δ0 ura3Δ0 trp1Δ63 LYS2Δ0 met15Δ0	30
BBY 1104 (rpd3Δ/hos2Δ)	BY 4705	MATα ade2Δ::hisG HIS3Δ200 LEU2Δ0 ura3Δ0 trp1Δ63 LYS2Δ0 met15Δ0 rpd3::LEU2 hos2::LYS2	30
BBY 1213	1257	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 set1Δ::hphMX4 set2Δ::kanMX4	31
BBY 1211	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 set1Δ::hphMX4	31
BBY 1212	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 set2Δ::kanMX4	31
BBY 1101	R1158	pSPT6::kanR-tet07-TATA URA3::CMV-tTA MATa his3Δ1 leu2Δ0 met15Δ0	Thermo Scientific
BBY 1130 (WT H3)	S288C	MATa his3Δ200 leu2Δ0 lys2Δ0 trp1Δ63 ura3Δ0 met15Δ0 can1::MFA1pr-HIS3 hht1-hhf1::NatMX4 hht2-hhf2::[HHTS-HHFS]-URA3	Thermo Scientific
BBY 1140 (WT H4)	S288C	MATa his3Δ200 leu2Δ0 lys2Δ0 trp1Δ63 ura3Δ0 met15Δ0 can1::MFA1pr-HIS3 hht1-hhf1::NatMX4 hht2-hhf2::[HHTS-HHFS]-URA3	Thermo Scientific
Myc-tagged strains
BBY 1208	S288C	MATα ade2::hisG his3Δ200 leu2Δ 0 lys2Δ0 met15Δ0 trp1Δ63 ura3Δ0 SPT6-13myc::HIS3	This study
BBY 1210	S288C	MATα ade2::hisG his3Δ200 leu2Δ0 lys2Δ0 met15Δ0 trp1Δ63 ura3Δ0 kin28::kin28-L83G Δbur2::Leu2 [pSH579 ARS CEN URA3 kin28-L83G] SPT6-13myc::HIS3	This study
BBY 1209	S288C	MATα ade2::hisG his3Δ200 leu2Δ0 lys2Δ0 met15Δ0 trp1Δ63 ura3Δ0 kin28::kin28-L83G [pSH579 ARS CEN URA3 kin28-L83c]SPT6-13myc::HIS3	This study
BBY 1111	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 SPT6-13myc::HIS3	This study
BBY 1112	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 bur1as SPT6-13myc::HIS3	This study
BBY 1113	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 ctk1Δ::kanMX4 SPT6-13myc::HIS3	This study
BBY 1114	BY 4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 bur1as ctk1Δ::kanMX4 SPT6-13myc::HIS3	This study
BBY 1105	BY 4705	MATα ade2Δ::hisG HIS3Δ200 LEU2Δ0 ura3Δ0 trp1Δ63 LYS2Δ0 met15Δ0 SPT6-13myc::HIS3	This study
BBY 1106	BY 4705	MATα ade2Δ::hisG HIS3Δ200 LEU2Δ0 ura3Δ0 trp1Δ63 LYS2Δ0 met15Δ0 rpd3::LEU2 hos2::LYS2 SPT6-13myc::HIS3	This study
BBY 1218	1257	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 set1Δ::hphMX4 set2Δ::kanMX4 SPT6-13myc::HIS3	This study
BBY 1216	BY4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 set1Δ::hphMX4 SPT6-13myc::HIS3	This study
BBY 1217	BY4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 set2Δ::kanMX4 SPT6-13myc::HIS3	This study
BBY 1102	R1158	pSPT6::kanR-tet07-TATA URA3::CMV-tTA MATa his3Δ1 leu2Δ0 met15Δ0 SPT6-13myc::HIS3	This study
BBY 1131 (H3 K4, K9, K14, K18-A)	S288C	MATa his3Δ200 leu2Δ0 lys2Δ0 trp1Δ63 ura3Δ0 met15Δ0 can1::MFA1pr-HIS3 hht1-hhf1::NatMX4 hht2-hhf2::[hhts K4,9,14,18A-HHFS]-URA3 SPT6-13myc::HIS3	This study
BBY 1132 (H3 Δ1-28)	S288C	MATa his3Δ200 leu2Δ0 lys2Δ0 trp1Δ63 ura3Δ0 met15Δ0 can1::MFA1pr-HIS3 hht1-hhf1::NatMX4 hht2-hhf2::[hhts (Δ1-28)-HHFS]-URA3 SPT6-13myc::HIS3	This study
BBY 1141 (H4 Δ1-16)	S288C	MATa his3Δ200 leu2Δ0 lys2Δ0 trp1Δ63 ura3Δ0 met15Δ0 can1::MFA1pr-HIS3 hht1-hhf1::NatMX4 hht2-hhf2::[HHTS-hhfs (Δ1-16)]-URA3 SPT6-13myc::HIS3	This study
BBY 1142 (H4 Δ1-24)	S288C	MATa his3Δ200 leu2Δ0 lys2Δ0 trp1Δ63 ura3Δ0 met15Δ0 can1::MFA1pr-HIS3 hht1-hhf1::NatMX4 hht2-hhf2::[HHTS-hhfs (Δ1-24)]-URA3 SPT6-13myc::HIS3	This study
spt6Δ202-Myc-tagged strains
BBY 1219	BY4705	MATα ade2Δ::hisG HIS3Δ200 LEU2Δ0 ura3Δ0 trp1Δ63 LYS2Δ0 met15Δ0 SPT6Δ202-13myc::HIS3	This study
BBY 1220	BY4705	MATα ade2Δ::hisG HIS3Δ200 LEU2Δ0 ura3Δ0 trp1Δ63 LYS2Δ0 met15Δ0 rpd3::LEU2 hos2::LYS2 SPT6Δ202-13myc::HIS3	This study

TABLE 2.

Sequences of the primers used in this study

Primer sequence (5′–3′)	Gene and locationa
GACAAAATGAAGAAAATGCTGATGCACC	POL1 ORF (F)
TAATAACCTTGGTAAAACACCCTG	POL1 ORF (R)
GCTGAGTTTAACGGTGATTATT	TEL VI-R (F)
CCAGTCCTCATTTCCATCAAT	TEL VI-R (R)
ACGGCTCTCCAGTCATTTAT	ARG1 UAS (F)
GCAGTCATCAATCTGATCCA	ARG1 UAS (R)
TAATCTGAGCAGTTGCGAGA	ARG1 TATA (F)
ATGTTCCTTATCGCTGCACA	ARG1 TATA (R)
TGGCTTATTCTGGTGGTTTAG	ARG1 5′ORF (F)
ATCCACACAAACGAACTTGCA	ARG1 5′ORF (R)
TTCTGGGCAGATCTACAAAGA	ARG1 3′ORF (F)
AAGTCAACTCTTCACCTTTGG	ARG1 3′ORF (R)
TATACGGCCTTCCTTCCAGTT	ADH1 CORE (F)
TGTGCAGCAAAAGAAACAAGGA	ADH1 CORE (R)
CCCACGGTAAGTTGGAATACA	ADH1 5′ORF (F)
TGACCACCGACTAATGGTAGC	ADH1 5′ORF (R)
CTTACGTCGGTAACAGAGCTGA	ADH1 3′ORF (F)
ACCAACGATTTGACCCTTTTC	ADH1 3′ORF (R)
CGACGACGAAGACAGTGATA	PMA1 5′ORF (F)
ATTCTTTTTCGTCAGCCATTG	PMA1 5′ORF (R)
TGTTTTGGGTGGTTTCTACTACG	PMA1 3′ORF (F)
TTAGGTTTCCTTTTCGTGTTAGAG	PMA1 3′ORF (R)
AAATAGGCACAAACACGACC	STE11 3′ORF (F)
ATTATGTGTGCATCCAGCCA	STE11 3′ORF (R)
GCTGATGAAACCTCTGCATCTAC	FLO8 3′ORF (F)
CAACCATACCAATATTCCCAA	FLO8 3′ORF (R)
GAAACAAGAACTGGAAGATGCCGAAGAGGAGAGGAAATTGATGATGGCAGAAGCCCGTGCAAAGAGACGGATCCCCGGGTTAATTAA	SPT6Δ202-Myc tagging (F)
TTTGTAATGGTTTTAAGACGCTTCTAAAATCTAACAGTAGTAAGAATAGAATGAACAACTACCGTCGGATCCCCGGGTTAATTAA	Spt6-Myc tagging (F)
CTTACCTAAACAATGGTCAAAGTAATAATAAAATTAATAATAACAATGGACACTACATACGCATGAATTCGAGCTCGTTTAAAC	Spt6-Myc tagging (R)

^aF and R, forward and reverse primer sequences, respectively.

Growth conditions.

For chromatin immunoprecipitation, saturated overnight cultures were subcultured in 100 ml of synthetic complete (SC) medium lacking isoleucine/valine and were grown to an A₆₀₀ of between 0.5 and 0.6. The cultures were treated with 0.65 μg of sulfometuron methyl (SM)/ml for 25 min to induce Gcn4. Analog-sensitive Ser2 kinase mutants (bur1as and bur1as/ctk1Δ) were treated with 6 μM 3MB-PP1 1 h prior to the addition of SM, as described previously (28). Similarly, Ser5 kinase mutants (kin28as and kin28as/bur2Δ) were treated with 6 μM NA-PP1 for 15 min prior to Gcn4 induction by SM, as described previously (29, 35). SPT6-tet cells (Thermo Scientific) were grown in the media containing 10 μg of doxycycline to deplete Spt6 (33). For coimmunoprecipitation experiments, the cells were grown in yeast extract-peptone-dextrose (YPD) medium to an A₆₀₀ of 2 to 3.

Western blot analysis.

Western blot analysis was carried out by either using a fraction of cells collected for chromatin preparation as described above or growing cells overnight in 5 ml of synthetic complete media. Whole-cell extracts (WCE) were prepared by the trichloroacetic acid (TCA) precipitation method (35). Briefly, the cell pellets were disrupted by using glass beads in the presence of 20% TCA, and the cell extracts were collected. The beads were washed once with 5% TCA, and the extracts thus obtained were pooled with the cell extracts collected upon 20% TCA extraction. The extracts were centrifuged, and the pellet was resuspended in 50 to 100 μl of 1 M Tris buffer (pH 8.0) to neutralize the pH. An equal volume of 2× sodium dodecyl sulfate (SDS) sample buffer was added to the extracts, and the proteins were separated on 8% SDS-polyacrylamide gels and transferred to a nitrocellulose membrane. The membranes were probed with the indicated antibodies, and signals were detected by enhanced chemiluminescence (ECL; GE Healthcare, catalog no. RPN2106). The signals were quantified using Image Studio Lite (LI-COR Biosciences).

ChIP experiments.

Chromatin immunoprecipitation (ChIP) experiments were performed as described previously (36). Briefly, 100 ml of cells (A₆₀₀ = 0.6) were cross-linked with 1% formaldehyde for 15 min at ambient temperature and quenched with glycine. Chromatin was isolated and fragmented by sonication (Branson 450) to an average size of 300 to 400 bp. The soluble fraction of chromatin was used for ChIP using antibodies against Myc (Roche; catalog no. 11667203001), Rpb3 (Neoclone; W0012), histone H3 (Abcam; ab1791), acetylated H3 (Millipore; 06-599), acetylated H4 (Millipore; catalog no. 06-866), phosphorylated Ser5 (Covance; MMS-134R), and phosphorylated Ser2 (Bethyl Laboratories; A300-654A). The primers used to amplify specific regions of ARG1, ADH1, and PMA1 and the internal control POL1 are listed in Table 2. SYBR green dye was added to the PCR-amplified DNA, resolved on 8% Tris-borate-EDTA gels, and the signals were quantified by using a phosphorimager. Fold enrichments (ChIP/inputs) were determined by normalizing the specific region/POL1 ChIP ratios to the ratios obtained for the input samples. For histone acetylation ChIPs, a region of the right arm of TEL VI was used as an internal control instead of POL1. The acetyl-histone/TEL VI ratio was further normalized to the H3 occupancy/POL1 in order to account for the changes in nucleosome density in the histone methyltransferase mutants. The data represent the average of ChIP performed using the chromatin prepared from at least three independent cultures, and PCRs were performed in duplicate for each ChIP sample.

ChIP-chip experiments.

ChIP experiments were performed as described above, and ChIP DNA was hybridized on tiling microarrays, custom designed by Agilent Technologies and contain a total of about 180,000 melting temperature (T_m)-adjusted 60-mer probes covering the entire yeast genome with virtually no gaps between probes. Spt6-Myc ChIP samples were hybridized in competition with control ChIP DNAs prepared from an isogenic untagged strain, except in Fig. 6, where Spt6 ChIP DNA was hybridized in competition with input DNA. Rpb3 ChIP DNA was hybridized in competition with input DNA. ChIP with microarray technology (ChIP-chip) data were normalized using the Limma Loess method, and replicates (at least duplicates) were combined as described previously (37).

FIG 6.

HDACs promote Spt6-Pol II interaction independently of histone acetylation. (A) Spt6-Myc was immunoprecipitated from whole-cell extracts prepared from WT and set1Δ/set2Δ cells, and Ser5P and Rpb3 were detected in the immunoprecipitates by Western blotting. (B) H3, acetylated H3 (H3ac), H4, and acetylated H4 (H4ac) histone N-terminal tail peptides were used in peptide pulldown experiment using whole-cell extracts prepared from Spt6-Myc- or Rli1-Myc (used as a control)-tagged strains. An untagged strain was additionally used as a negative control. (C) Spt6-Myc ChIP occupancy at the 5′ORF and 3′ORF of ADH1 and PMA1 in the histone H4 tail deletion mutants. (D) Spt6-Myc ChIP occupancy at the 5′ORF and 3′ORF of the PMA1 and ADH1 genes in the H3 tail deletion mutant (H3 Δ1-28) (left panel) and H3 K→A mutant (right panel). (E) Whole-cell extracts prepared from Myc-tagged Rco1, Rpd3, and Hos2 and an untagged strain were immunoprecipitated with anti-Myc antibodies, and Spt6 was detected using anti-spt6 antibodies. (F) Spt6-Myc was immunoprecipitated with anti-Myc antibodies from whole-cell extracts of WT and rpd3Δ/hos2Δ cells, and Ser5P, Rpb3, Rpd3, and H3 were detected by Western blotting. Rli1-Myc and untagged WT strain were used as a negative control. The specific Rpb3 and Rpd3 bands are indicated by arrows, and the IgG bands are indicated by asterisks in panels A and F.

Aggregate profiles.

Aggregate profiles were generated using the versatile aggregate profiler (38). In brief, genes were virtually cut in the middle and the first half aligned on the TSS, while the second half was aligned on the poly(A) (pA) site. The TSS and pA sites were deduced using the untranslated region (UTR) sizes as determined by Xu et al. (39). Genes for which the 5′ and 3′ UTRs have not been previously determined were therefore not included in these analyses. The aligned data were averaged over 50-bp bins (10 bins upstream from the TSS, 40 bins downstream from the TSS, and 10 bins downstream of the pA). The genes included in each specific analysis are described in the figure legends.

The normalized log₂ enrichment values in the coding region were averaged to obtained open reading frame (ORF) enrichment for Spt6 and Rpb3. The top 25% of genes (n = 1,580) bound by Spt6 were selected and analyzed. Similarly, Rpb3 ORF values were determined and used to plot Spt6 occupancy based on the average Pol II occupancy observed in the isogenic WT strain.

Coimmunoprecipitation experiments.

Coimmunoprecipitation experiments were performed as described previously (35). Cells were grown in YPD to an A₆₀₀ of 2.0, and the whole-cell extracts (0.5 to 1 mg) were lysed in lysis buffer [50 mM Tris-HCl (pH 7.5), 50 mM HEPES-KOH (pH 7.9), 10 mM MgSO₄, 100 mM (NH₄)₂SO₄, 12.5 mM potassium acetate, 0.01% NP-40, 20% glycerol, 1 μg/ml pepstatin A, 100 mM phenylmethylsulfonyl fluoride [PMSF], 1 μg/ml leupeptin]. Whole-cell extracts thus prepared were incubated with magnetic bead-conjugated anti-Myc antibodies in 1× MTB buffer (4× MTB buffer is 200 mM HEPES-KOH [pH 7.9], 800 mM potassium acetate, 54 mM magnesium acetate, 40% glycerol, 0.04% NP-40, 400 mM PMSF, 4 μg/ml pepstatin, 4 μg/ml leupeptin) for 2 h and washed three times with wash buffer (50 mM Tris-HCl [pH 8.0], 0.1% NP-40, 150 mM NaCl, 10% glycerol, 1 mM PMSF, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and phosphatase inhibitors [Sigma; catalog no. P0044]). The immunoprecipitates were processed for Western blot analysis. The experiments were performed using at least three independent cultures.

Peptide pulldown assay.

WCEs (0.5 to 1 mg) were incubated with 5 μg of biotinylated histone tail peptides purchased from Anaspec, USA (H3, catalog no. 61702; acetylated H3, catalog no. 65277; H4, catalog no. 65242, and acetylated H4, catalog no. 65248) overnight at 4°C. Streptavidin beads were preincubated with phosphate-buffered saline–bovine serum albumin (5%) for 1 h, washed once with 500 μl of high-salt wash buffer (25 mM Tris-HCl [pH 8.0], 1 M NaCl, 1 mM dithiothreitol [DTT], 5% glycerol, 003% NP-40), and washed twice with 500 μl of binding buffer (25 mM Tris-HCl [pH 8.0], 500 mM NaCl, 1 mM DTT, 5% glycerol, 0.03% NP-40). The mixture of peptide and WCEs was added to the pretreated streptavidin beads, incubated for 3 h, and washed with wash buffer containing 150 mM NaCl to collect peptide-protein complexes. The presence of Spt6 in the pulldown was detected by Western blotting. These experiments were repeated three times with reproducible results.

RESULTS

Ser2 kinases recruit Spt6 to coding regions.

To investigate the mechanisms of Spt6 recruitment, we used ARG1, a Gcn4 target gene, which is induced under amino acid starvation conditions. ARG1 was induced by treating cells with sulfometuron methyl (SM), an inhibitor of isoleucine-valine biosynthesis, and the Pol II (Rpb3) and Myc-tagged Spt6 (Spt6-Myc) occupancies were determined by chromatin immunoprecipitation (ChIP). In agreement with previous studies (15, 22), SM treatment increased Rpb3 (Pol II subunit) binding to both promoter and coding regions of the ARG1 gene and increased Spt6-Myc binding primarily to the ARG1 coding region (data not shown).

To determine the role of Pol II CTD phosphorylation in recruiting Spt6, we first analyzed its occupancy in the Ser2 CTD kinase mutants, ctk1Δ, bur1as (as, analog sensitive), and bur1as/ctk1Δ by ChIP at the ARG1 gene. The kinase activity of bur1as is rapidly inhibited by treating growing cells with the ATP analog 3MB-PP1 prior to induction by SM (28). As expected, the Ser2P levels were higher in the ARG1 3′ part of the coding region (here called the 3′ORF) than in the 5′ part of the coding region (here called the 5′ORF) in WT cells (Fig. 1A) and were substantially reduced across the ARG1 coding region in the ctk1Δ and bur1as/ctk1Δ mutants. This is consistent with the fact that Ctk1 is responsible for the majority of Ser2 phosphorylation in vivo (40, 41). The bur1as mutant treated with 3MB-PP1 reduced Ser2P only in the 5′ORF. This last result is in agreement with the role of Bur1 in phosphorylating Ser2 at promoter-proximal regions (28, 42). In contrast to the large reduction in Ser2P in the ctk1Δ mutant, Spt6 binding was only modestly reduced at the ARG1 ORF in this mutant (Fig. 1B). Similarly, a reduction in Spt6 binding was also observed in the bur1as mutant in the ARG1 5′ORF and, interestingly, greater reductions in the ARG1 5′ORF and 3′ORF were observed in the bur1as/ctk1Δ mutant. We, however, noted that total Pol II occupancy over the ARG1 coding region was reduced in ctk1Δ and bur1as/ctk1Δ cells (Fig. 1C). To account for this effect, we calculated the Spt6-Myc/Rpb3 ratio in WT and kinase mutants and found that the Spt6-Myc/Rpb3 ratios were similar in the WT and ctk1Δ cells, indicating that Ctk1 is dispensable for Spt6 recruitment (Fig. 1D), an observation in agreement with a previous report (27). A greater reduction in the Spt6-Myc/Rpb3 ratio at both the 5′ORF and the 3′ORF of the ARG1 gene was observed in the bur1as/ctk1Δ mutant compared to the bur1as mutant (Fig. 1D), indicating that both Bur1 and Ctk1 contribute to the full recruitment of Spt6 in vivo. Similarly, at the ADH1 gene, a greater Spt6 binding defect was observed in the double mutant than in the respective single mutants (Fig. 1E).

FIG 1.

Ser2 kinases promote Spt6 recruitment to coding regions. (A to E) WT, ctk1Δ, bur1as, and bur1as/ctk1Δ cells expressing Myc-tagged Spt6 were treated with a 6 μM concentration of ATP analog (3MB-PP1) to inhibit Bur1 kinase activity for 1 h prior to the induction of Gcn4 target genes by SM for additional 25 min. Chromatin extracts were subjected to ChIP using antibodies against serine2 phosphorylated Pol II (Ser2P), Myc, or Rpb3. ChIP occupancy of Ser2P (A), Spt6-Myc (B), and total Pol II (Rpb3) (C) at the promoter (TATA), as well as the 5′ (5′ORF) and the 3′ parts (3′ORF) of the ARG1 coding region, is shown. (D) Spt6-Myc/Rpb3 ratio at the ARG1 gene. (E) Spt6-Myc occupancy at the core promoter (TATA), 5′ (5′ORF), and 3′ (3′ORF) parts of the ADH1 gene. The signal for ARG1 and ADH1 ChIP DNA was normalized to their respective input DNA (ChIP/Input) and further normalized against the POL1 ChIP/input ratios. ChIPs were conducted using at least three independent cultures, and the error bars represent the standard errors of the mean (SEM). (F and G) Western blots showing the levels of Myc-tagged Spt6 (F) and untagged Spt6 (G) in WT and Ser2 kinase mutant strains. Proteins from the indicated strains were extracted using the TCA method after a 1-h treatment with 3MB-PP1. Gcd6 was used as a loading control. The band intensities were quantified by using Image Studio Lite (LI-COR Biosciences), and the average data from three independent cultures and two different loadings from each sample were plotted and are shown below the blots. Error bars indicate the SEM.

A recent study suggested that the deletion of CTK1 greatly reduces Spt6 cellular levels (43). Although ctk1Δ did not reduce Spt6 occupancy in the coding regions in our experiments (Fig. 1B and 1D) (27), we nonetheless determined Spt6 levels in the Ser2 kinases mutants. The Spt6 protein levels were largely unaffected in ctk1Δ, bur1as, or bur1as/ctk1Δ cells (Fig. 1F). Similarly, no reduction in Spt6 levels was observed in chromatin samples used for ChIP analysis or in cells grown in YPD medium (data not shown). Nearly equal levels of Spt6 were also observed in untagged strains, ruling out the possibility that introducing the Myc tag stabilized the protein in the ctk1Δ mutant (Fig. 1G). The discrepancy between our results regarding the levels of Spt6 in the ctk1Δ mutant and those obtained by a previous study (43) might indicate the differences in whole-cell extract preparation and preservation of protein stability. Altogether, our results (Fig. 1A to E) suggest that Ser2 phosphorylation by both Bur1 and Ctk1 contribute to Spt6 recruitment in the coding regions.

Ser2 kinases promote Spt6 recruitment to the 5′ coding regions genome-wide.

In order to look at the possible contribution of Ser2 phosphorylation on Spt6 recruitment at a global scale, we made use of bur1as/ctk1Δ cells. The bur1as/ctk1Δ cells were treated with 3MB-PP1 to inactivate the Bur1 kinase activity, and Rpb3 and Spt6-Myc immunoprecipitated DNA were hybridized on 4x180K Agilent arrays (see Materials and Methods) to determine their genome-wide occupancy.

In agreement with previous studies (15, 22), the average Spt6 occupancy in the coding regions very strongly correlated with Rpb3 occupancy in WT cells (Fig. 2A) (r = 0.91, Pearson correlation; see also Table S1 in the supplemental material). To examine whether Spt6 occupancy is affected in bur1as/ctk1Δ cells, we selected the top 25% of the Spt6-occupied genes, based on the average Spt6 occupancy over each transcribed region (n = 1,580; see Table S1 in the supplemental material) and plotted the average Rpb3 and Spt6 log₂ values for these genes. Although the Rpb3 occupancy in the mutant strongly correlated (Fig. 2B) (r = 0.85, Pearson) with that of WT cells at the top 25% of Spt6 occupied genes, only a moderate correlation (r = 0.57) was observed between Spt6 occupancy in the mutants and Rpb3 in WT cells. This suggests that Ser2 kinases are important for recruiting Spt6 (Fig. 2B). More importantly, in the bur1as/ctk1Δ mutant, Spt6 binding was primarily reduced at the 5′ ends of genes, with only minor defects in the 3′ORF (Fig. 2C). This was somewhat surprising given that Ser2 phosphorylation is primarily detected nearing the 3′ORF. This result suggests that phosphorylation of few (perhaps specific) CTD repeats by Bur1 and Ctk1 near the 5′ORF, rather than bulk levels of Ser2 phosphorylation, is important for the efficient recruitment of Spt6.

FIG 2.

Ser2 kinases promote Spt6 recruitment to the 5′ORFs. (A) Log₂ enrichment ratios for Spt6 and Rpb3 in WT cells were averaged over the transcribed regions, and Spt6 values were plotted against Rpb3 values for each gene (>6,000 genes). (B) Scatter plot of the average log₂ enrichment ratio of Rpb3 and Spt6 in the bur1as/ctk1Δ mutant against the average Rpb3 log₂ enrichment ratio in WT cells for the top 25% Spt6-occupied genes (n = 1,580). The Pearson correlation coefficients (r) are indicated. (C) The log₂ enrichment ratio of Rpb3 and Spt6 across metagenes made of the top 25% Spt6-occupied genes in WT and bur1as/ctk1Δ cells is shown on the left. The change in the Spt6 log₂ enrichment ratio (mutant − WT) is shown on the right. The box at the bottom of the plots represents the transcribed region. The transcription start site (TSS) and transcription end site (TES) are marked. (D) Rpb3 (left) and Spt6 (right) log₂ enrichment ratio in WT and bur1as/ctk1Δ cells along groups of Spt6-occupied genes with different Rpb3 average log₂ enrichment ratios (>2, high; between 1 and 2, medium; <1, low). (E) Rpb3 (left) and Spt6 (right) log₂ enrichment ratio in WT and bur1as/ctk1Δ cells along genes with Rpb3 average log₂ enrichment ratios of <0.5. The number of genes in each group is indicated.

To further determine whether highly transcribed genes are more reliant on Ser2 kinases for recruiting Spt6, we analyzed these genes based on Rpb3 occupancy (see Table S1 in the supplemental material). Unlike the Rpb3 defect that uniformly extends along the entire ORF (Fig. 2D, left panel), reduction in Spt6 occupancy was more pronounced near the 5′ end (Fig. 2D, right panel) of the genes, irrespective of Rpb3 occupancy. The role of Ser2P in recruiting Spt6 was even more evident for genes containing <0.5 log₂ Rpb3 occupancy (Fig. 2E; see Table S1 in the supplemental material), indicating that Ser2 kinases are essential for recruiting Spt6 at the 5′ORFs of genes transcribed at low levels (low-transcribed genes) as well.

It was surprising to see that Spt6 binding around the transcription end site (TES) was largely unaffected in the Ser2 double kinase mutant (Fig. 2C to E), despite the loss from the 5′ORF. This observation suggests that Spt6 recruitment toward the 3′ end of the gene is carried out by other mechanisms, whereas Ser2 kinases are required for promoting Spt6 recruitment to the 5′ ends of genes. Tyr1-CTD phosphorylation, which has the greatest affinity to the Spt6 tSH2 domain in vitro (23) and peaks at the 3′ ends of genes, may facilitate Spt6 recruitment toward the end of the ORF in the absence of Ser2P, since Tyr1P levels are not affected by the known Pol II CTD kinases (23).

Spt6 occupancy is not impaired in Ser5 kinase mutants.

Given that the Spt6 tSH2 domain strongly binds to Ser5-phosphorylated (Ser5P) but not to unmodified CTD peptides and that a stronger binding is observed with doubly phosphorylated CTD peptides (Ser2P/Ser5P or Tyr1P/Ser5P) (23–25), we examined whether Ser5 phosphorylation alone, or in combination with Ser2P, plays a role in recruiting Spt6. To this end, we measured Spt6 binding in kin28as and kin28as/bur2Δ mutants. As expected, treating these mutants with the ATP analog NA-PP1 substantially reduced Ser5P across the ARG1, ADH1, and PMA1 genes (Fig. 3A and data not shown). In contrast, Spt6 and Rpb3 occupancies at ARG1 were similar to that of WT in these mutants (Fig. 3B and C), indicating that Ser5P by Kin28 is dispensable for Spt6 recruitment. A small reduction in Spt6 binding at ADH1 and a substantial reduction in the PMA1 gene (Fig. 3D and F) could be attributed to the reduced Pol II in these mutants (Fig. 3E and G). As such, the Spt6-Myc/Rpb3 ratios in the kinase mutants were largely similar to the WT, except for a slight reduction in the ratio at the 3′ORF of PMA1 (Fig. 3H). Collectively, these data suggest that Ser5 phosphorylation by Kin28 is not significantly required for recruiting Spt6 to these genes.

FIG 3.

Ser5 kinases are dispensable for Spt6 recruitment. Cells were treated with 6 μM NA-PP1 to inactivate Kin28 and processed for ChIP. The occupancy of Ser5P, Spt6-Myc, and Rpb3 (Pol II) was determined for WT, kin28as, and kin28as/bur2Δ cells. The occupancy of Ser5P (A), of Spt6-Myc (B), and of Rpb3 (C) is shown for the ARG1 gene. (D and E) ChIP occupancy of Spt6-Myc (D) and of Rpb3 (E) across the ADH1 gene. (F and G) ChIP occupancy of Spt6-Myc (F) and of Rpb3 (G) at the 5′ and 3′ ends of PMA1. (H) The Spt6-Myc occupancies at the 5′ and 3′ ORFs were normalized to the Rpb3 occupancies at these regions, and the Spt6-Myc/Rpb3 ratios are shown for ARG1, ADH1, and PMA1. (I) Metagene analysis showing Rpb3 and Spt6 log₂ enrichment ratios in WT and kin28as/bur2Δ for the top 25% of the Spt6-occupied genes (n = 1,580; selected based on the average Spt6 binding in the coding region of each gene). The box at the bottom of the plots represents the transcribed region. The TSS and TES are marked. (J) Log₂ enrichment ratios for Rpb3 (left) and Spt6 (right) on genes binned based on their average Rpb3 occupancy in the coding region are indicated (>2, high; between 1 and 2, medium; <1, low). The number of the genes in each group is also indicated.

To evaluate the role of Ser5 phosphorylation genome-wide, we examined Spt6-Myc occupancy in the kin28as/bur2Δ mutant by using ChIP-chip experiments. We selected this mutant because deleting BUR2 in the kin28as background elicits greater reductions in Ser5 phosphorylation than kin28as alone (29). To determine whether Spt6 occupancy is affected in kin28as/bur2Δ cells, we calculated the average Spt6 occupancy over each transcribed region, selected the top 25% of the Spt6-occupied genes (n = 1,580) (see Table S2 in the supplemental material), and plotted the average Rpb3 and Spt6 log₂ values for these genes. Rpb3 occupancy in the mutant closely followed that of WT until the TSS and then showed a sharp decline in the ORF of the metagene (Fig. 3I). As previously observed (29, 44–46), the sharp reduction in Rpb3 occupancy downstream of the TSS in various Kin28 mutants is more consistent with a promoter clearance defect than with a Pol II processivity defect, as the latter would have produced a gradual reduction in Rpb3 binding. Similarly, Spt6 occupancy was reduced in the ORF (Fig. 3I), which can be attributed to the Pol II binding defect. Overall, these data are consistent with the gene-specific ChIP results (Fig. 3B, D, and F) and suggest that the kin28as/bur2Δ mutant has a minor, if any, effect on Spt6 occupancy. Further analyses of Spt6 binding at genes binned based on their Rpb3 occupancy (high, medium, and low) (see Table S2 in the supplemental material) showed that Spt6 binding defect mirrors the reduction in Pol II (Fig. 3J). Altogether, our data show that Ser5 phosphorylation indirectly regulates Spt6 binding by modulating Pol II occupancy in vivo.

Spt6 is required for maintaining histone occupancy in coding regions.

To investigate whether Spt6 regulates histone density in the coding regions, we measured histone occupancy at the ARG1 ORF in Spt6-depleted cells. We utilized a strain in which SPT6 expression is driven by a tetracycline-titratable promoter (SPT6-tet) (33) and is repressed in the presence of doxycycline (Dox). Dox-treated SPT6-tet cells revealed a substantial reduction in total Spt6 levels (Fig. 4A). Depletion of Spt6 (Dox-treated) led to a greater reduction in H3 occupancy in the ARG1 3′ORF under inducing conditions (Fig. 4B, compare induced No Dox and Dox), a finding consistent with the role of Spt6 in regulating histone occupancy. This reduction, in part, stems from the failure of histones to fully recover upon repression to the levels observed prior to ARG1 induction (Fig. 4B, compare repressed No Dox and Dox) in Spt6-depleted cells. This is unlikely due to sustained transcription of ARG1 in Spt6-depleted cells because Rpb3 binding was reduced to a noninducing level upon repression (Fig. 4C). Overall, our data indicate that Spt6 is dispensable for regulating Pol II occupancy at the ARG1 gene. It is, however, needed to maintain normal cotranscriptional histone occupancy in the coding region. Interestingly, the reduction in H3 occupancy in the ARG1 ORF resembled that of strains lacking histone deacetylases (35, 47), suggesting that HDACs function with Spt6 to maintain appropriate nucleosome occupancy. This is consistent with the fact that mutations in Spt6, Rpd3S, and Hos2-Set3 complexes result in spurious transcription from many genes (17, 18, 21, 48).

FIG 4.

HDACs promote Spt6 recruitment to the coding regions. (A) SPT6-tet cells carrying a Myc-tagged Spt6 were grown in SC plus 10 μg of doxycycline (Dox)/ml overnight and subcultured in the presence of 10 μg of Dox/ml the next day. Cells were treated with SM to induce Gcn4 targeted genes and processed for ChIP as described in the legend of Fig. 1. A fraction of the cells (10 ml) was collected prior to the cross-linking step, and TCA-precipitated whole-cell extracts were used to determine Spt6-Myc levels by Western blotting with anti-Myc antibodies. “Dox” and “No Dox” represent cells treated and not treated with doxycycline, respectively. Gdc6 was used as a loading control. (B and C) Histone H3 (B) and Rpb3(C) occupancy at the upstream activating sequence (UAS), the promoter (TATA), and the coding region (3′ORF) of the ARG1 gene with (Dox) or without (No Dox) doxycycline treatment under noninduced, induced, and repressed conditions. A 10× isoleucine-valine solution was added to the induced cells for 5 min to repress ARG1 transcription. (D) Spt6 occupancy in WT and rpd3Δ/hos2Δ cells at the ARG1, ADH1, and PMA1 genes. (E) Spt6-Myc/Rpb3 ratio along the ARG1, ADH1, and PMA1 genes. (F) Scatter plot of Rpb3 and Spt6 log₂ enrichment ratio in rpd3Δ/hos2Δ cells against the Rpb3 log₂ enrichment in WT cells for the top25% Spt6-occupied genes. The Pearson correlation coefficient (r) is indicated in parentheses. (G) Rpb3 (left) and Spt6 (right) log₂ enrichment ratio in WT and rpd3Δ/hos2Δ cells on groups of genes binned based on their Rpb3 average log₂ enrichment levels. The box at the bottom of the plots represents the transcribed region. The TSS and TES are marked. (H) Rpb3 (left) and Spt6 (right) average log₂ enrichment ratio in WT and rpd3Δ/hos2Δ cells along genes with gene lengths of >2 kb or between 1 and 2 kb. (I) Rpb3 (left) and Spt6 (right) log₂ enrichment ratio in WT and rpd3Δ/hos2Δ cells at 418 genes showing a >0.5 reduction in the Spt6 log₂ binding ratio.

HDACs promote Spt6 recruitment.

To test whether HDACs promote Spt6 occupancy in the ORFs of transcribed genes, we examined the recruitment of Spt6-Myc in the HDAC rpd3Δ/hos2Δ mutant. This mutant revealed a small (∼25%) reduction in Spt6-Myc binding at the ARG1 3′ORF (Fig. 4D). Interestingly, the rpd3Δ/hos2Δ mutation elicited greater reductions (2- to 2.5-fold) in Spt6 binding to the coding regions of constitutively expressed ADH1 and PMA1 genes (Fig. 4D). Since we observed a small Rpb3 binding defect at these genes, we calculated the Spt6-Myc/Rpb3 ratios and found that they were reduced in the coding regions of the ARG1, ADH1, and PMA1 genes (Fig. 4E). The rpd3Δ/hos2Δ double mutant produced greater defects in Spt6 binding and H3 occupancy (35) than the single mutants (data not shown). Collectively, these data suggest that both Rpd3 and Hos2 are needed for full recruitment of Spt6 to the coding regions.

To further define the role of these HDACs in Spt6 recruitment, we determined the genome-wide occupancy of Spt6 and Rpb3 in WT and rpd3Δ/hos2Δ cells by ChIP-chip analysis (see Table S3 in the supplemental material). While Rpb3 showed a strong correlation between WT and the double mutant for the top 25% Spt6-occupied genes, Spt6 in the mutant only modestly correlated with Rpb3 in WT cells (Fig. 4F). This suggests a role for these HDACs in Spt6 recruitment. Spt6-bound genes were grouped into high-, medium-, and low-transcribed genes based on the average Rpb3 occupancy in the ORF. Although the rpd3Δ/hos2Δ mutant produced a small reduction in Rpb3 binding across the metagene of the high and medium groups, the low-transcribed group was largely unaffected (Fig. 4G, left panel). In contrast to the uniform reduction in Rpb3 across the ORF, the reduction in Spt6 binding in the rpd3Δ/hos2Δ mutant was mostly localized to the middle of the transcribed region (Fig. 4G, right panel). This was even more apparent in the low-transcribed group, which has the WT levels of Rpb3. This suggests that HDACs potentially regulate Spt6 occupancy in the midtranscribed ORFs after initial recruitment to the 5′ end. To determine whether the role of HDACs in regulating Spt6 occupancy depends on gene length, we plotted Spt6 occupancy at genes with medium (1 to 2 kb, n = 452) and long (>2 kb, n = 116) ORFs (see Table S3 in the supplemental material). Compared to a slight reduction in Rpb3, Spt6 occupancy in the rpd3Δ/hos2Δ mutant was greatly diminished from the middle of coding regions for both groups of genes (Fig. 4H). This observation provides further support to the idea that HDACs maintain Spt6 association in the coding region after initial recruitment. Furthermore, we identified 418 genes in which Spt6 was reduced (≥0.5 log₂) in the HDAC mutant (see Table S3 in the supplemental material) and plotted the Rpb3 and Spt6 occupancies. Although these genes showed an overall Rpb3 binding defect, the Spt6 binding was diminished to nearly background levels in the coding region (Fig. 4I). Interestingly, even at these genes, a distinct Spt6 binding peak can be observed near the TES, suggesting that Spt6 is recruited near the transcription end sites in a manner independent of HDACs or Ser2 kinases. In summary, our data indicate that HDACs Rpd3 and Hos2 regulate Spt6 occupancy in coding regions. Unlike Ser2 kinases, Rpd3 and Hos2 appear to function in sustaining high levels of Spt6 occupancy in the mid-transcribed regions following the recruitment to the 5′ proximal region.

HDACs promote Spt6 interaction with phosphorylated Pol II CTD.

Since Spt6 interacts with phosphorylated Pol II, an impaired Spt6-Pol II interaction could account for the reduced Spt6 association in the mid-transcribed regions. To test this, we immunoprecipitated Spt6-Myc and examined for the presence of phosphorylated Ser2 and Ser5 in the immunoprecipitates. Consistent with the proposed interaction of Spt6 with elongating Pol II, Ser2P and Ser5P coimmunoprecipitated with Spt6-Myc from the WT but not with the Myc-tagged Rli1 (Rli1-Myc), a translation factor used as a control (Fig. 5A). Interestingly, phosphorylated forms of Pol II were greatly diminished in Spt6-Myc immunoprecipitates from the rpd3Δ/hos2Δ mutant despite having WT levels of Ser2P and Ser5P (Fig. 5A, Ser2P and Ser5P inputs). These results suggest that Rpd3 and/or Hos2 promote association of Spt6 with the phospho-Pol II CTD.

FIG 5.

Histone methyltransferases are not required for Spt6 recruitment. (A) Whole-cell extracts prepared from Spt6-Myc tagged WT and rpd3Δ/hos2Δ strains were immunoprecipitated with anti-Myc antibodies, and the immunoprecipitates were then subjected to Western blot analysis with antibodies against Myc, Ser2P, and Ser5P. Rli1-Myc was used as a control. (B and C) Spt6-Myc (B) and Rpb3 (C) occupancy in the 5′ (5′ORF) and 3′ (3′ORF) part of the ADH1 and PMA1 coding regions in the indicated histone methyltransferase mutants. (D and E) Spt6-Myc (D) and Rpb3 (E) enrichment at the ARG1 gene in WT and set1Δ/set2Δ cells. (F to H) The ChIP occupancies of acetylated H3 and H4 were normalized to the H3 occupancies. The levels of H3 acetylation/H3 (F) and H4 acetylation/H3 (G) at the 3′ORF of the STE11 and FLO8 genes and at the 5′ORF and 3′ORF of PMA1 (H) are shown.

The simplest explanation for both the diminished coimmunoprecipitation of phosphorylated Pol II with Spt6 (Fig. 5A) and the reduced Spt6 binding in the coding regions in the rpd3Δ/hos2Δ mutant (Fig. 4G to I) is that these HDACs facilitate Spt6 association with the transcribed regions by stabilizing its interaction with phosphorylated Pol II. Given that both Rpd3S and Hos2-Set3 complexes deacetylate ORF nucleosomes (47–49), it is possible that hyperacetylation of the ORF nucleosomes in rpd3Δ/hos2Δ suppresses Spt6 binding. To test this, we examined Spt6 occupancy in set1Δ, set2Δ and set1Δ/set2Δ mutants. Set1 and Set2 methylate H3K4 and H3K36, respectively, and these methylated residues are required for Hos2-Set3C and Rpd3S binding and subsequently for deacetylation of the ORF nucleosomes (35, 47, 48). Unlike reduced Spt6 binding seen in rpd3Δ/hos2Δ (Fig. 4D), the binding in the ADH1 and PMA1 coding regions was not affected in set1Δ cells and was only slightly reduced in the set2Δ and set1Δ/set2Δ mutants (Fig. 5B). A small reduction in binding was also seen at the ARG1 ORF in the set1Δ/set2Δ mutant (Fig. 5D). Rpb3 occupancy in the single mutants was similar to that observed in the WT (Fig. 5C and E) and was slightly reduced in the set1Δ/set2Δ double mutant. Overall, histone methyltransferase mutants exhibited only a small reduction in Spt6 binding. To confirm that acetylation is in fact increased in these histone methyltransferase mutants under our experimental conditions, we analyzed both H3 and H4 acetylation in these mutants. We found that both H3 aand H4 acetylation increased in the 3′ORFs of STE11 and FLO8 in the mutants (Fig. 5F and G). The increase in H3 and H4 acetylation in the set1Δ/set2Δ mutant was largely similar to that observed in the set2Δ single mutant, except for the acetylated H3 at the STE11 3′ORF, where it appears that both Set1 and Set2 cooperate to regulate histone acetylation. Likewise, the histone acetylation was also increased by ∼1.5- to 2-fold in the coding regions of PMA1 in the set1Δ/set2Δ mutant (Fig. 5H). An increase in H3 acetylation, though to a lesser degree, was also observed at the ADH1 ORF in these mutants (data not shown). Our results showing increased histone acetylation in the methyltransferase mutants are consistent with previous studies (20, 47, 48) and indicate that impaired Spt6 binding in the rpd3Δ/hos2Δ mutant cannot be accounted for simply by altered histone acetylation in this mutant. Furthermore, a greater reduction in Spt6 occupancy in rpd3Δ/hos2Δ cells than that observed in set1Δ/set2Δ cells suggests that HDACs regulate Spt6 association with Pol II before deacetylation of nucleosomes. To test this, we immunoprecipitated Spt6-Myc from the WT and set1Δ/set2Δ cells and probed for Ser5P and total Pol II (Rpb3). Both Rpb3 and Ser5P were efficiently coimmunoprecipitated from the WT and set1Δ/set2Δ extracts (Fig. 6A), indicating that Set1 and Set2 (or methylation-dependent deacetylation) are dispensable for Spt6-Pol II interaction.

To further examine whether hyperacetylation inhibits Spt6 binding, we performed histone N-terminal tail peptide pulldown assays. Both unmodified and acetylated (ac) H3 and H4 peptides precipitated similar amounts of Spt6-Myc from whole-cell extracts prepared from the Spt6-Myc tagged strain (Fig. 6B). Myc signals were not detected in the pulldown experiments using Rli1-Myc or untagged WT strains (Fig. 6B). These results show that Spt6 interaction with H3 and H4 N-terminal tails do not strongly depend on acetylation. In further support of this observation, deleting 1 to 16 residues of H4 (Δ1-16 H4) or a larger deletion (Δ1-24 H4), which removes all the acetylable residues within the H4 tail (K5, K8, K12, and K16), did not significantly reduce Spt6 occupancy in the PMA1 or ADH1 ORFs (Fig. 6C). Similarly, deleting H3 N-terminal tail, which eliminates acetylable lysines within the H3 tail (H3 K9, K14, K18, K23, and K27) also failed to reduce Spt6 occupancy in the ORFs of ADH1 and PMA1 (Fig. 6D, left). Consistent with this, the H3K→A mutant (H3 K4, K9, K14, and K18 substituted to alanines) displayed WT levels of Spt6-Myc occupancy at the PMA1 ORF and elicited only a minor reduction in Spt6-Myc binding at the ADH1 5′ORF (Fig. 6D, right). These last results are in agreement with a previous study (50) showing that histone H3 tail is not required for Spt6 recruitment at the RNR3 gene. These results imply that both H3 and H4 N-terminal tails are dispensable for recruiting Spt6 to the transcribed regions of these genes and that HDACs maintain Spt6 association to the transcribed region by promoting Spt6-Pol II interactions.

Considering that phosphorylated Pol II recruits HDACs Rpd3S and Hos2 (35, 51), as well as Spt6, to the transcribed regions (Fig. 1 to 4), we hypothesized that Spt6-HDAC interactions promote the association of Spt6 with transcribed genes. To this end, we immunoprecipitated Myc-tagged Rpd3, Rco1 (a subunit specific to Rpd3S), and Hos2 and examined the coimmunoprecipitation of Spt6 by Western blot analysis. Spt6 coimmunoprecipitated with Rco1 and, to a lesser extent, with Rpd3 but not with Hos2 or the untagged WT used as a control (Fig. 6E). This suggests that Rpd3-containing complexes are largely responsible for promoting Spt6-Pol II interactions. However, the single rpd3Δ mutant failed to impair Spt6-Pol II interaction (data not shown), suggesting that both Rpd3 and Hos2 are promoting this interaction. These results are in agreement with the observation that the rpd3Δ/hos2Δ mutant produced a greater reduction in Spt6 occupancy at ADH1 and PMA1 ORFs than the respective single mutants (data not shown).

We considered the possibility that the lack of Spt6 interaction with the elongating Pol II also diminishes its interaction with histones. To test this, we immunoprecipitated Spt6-Myc and examined the coimmunoprecipitation of histone H3. Consistent with our earlier observations (Fig. 5A and 6A, and E), coimmunoprecipitation of Ser5P, Rpb3, and Rpd3 was seen with Spt6-Myc in the WT cells but not in rpd3Δ/hos2Δ cells (Fig. 6F). Similarly, H3 was observed in immunoprecipitates from the WT but not from the HDAC mutant, indicating that the Spt6 interaction with Pol II is needed for efficient interaction with histones. Altogether, our results support a model in which initial recruitment of Spt6 by Ser2 CTD kinases near the TSS is maintained by HDACs across the transcribed region and that interactions between HDACs and Spt6 contribute to this process.

HDACs and CTD kinases coordinate Spt6 recruitment.

Considering that rpd3Δ/hos2Δ impairs Spt6-Pol II interactions (Fig. 5A and 6F) and reduces Spt6 occupancy downstream of TSS genome-wide (Fig. 4G to I), it is possible that HDACs are needed to maintain Spt6 binding after CTD-dependent recruitment at the 5′ORF. Alternatively, HDACs could independently recruit Spt6 to the mid-transcribed regions. To distinguish between these two possibilities, we deleted the tSH2 domain (spt6Δ202) in WT and rpd3Δ/hos2Δ cells and examined the occupancy of this Spt6 mutant (Fig. 7A). Since the tSH2 domain interacts with the phosphorylated CTD (24–26), deleting this domain should diminish CTD-dependent Spt6 recruitment. In agreement with a previous study (22), deleting the tSH2 greatly reduced Spt6 but not Pol II (Rpb3) occupancy in the ARG1, ADH1, and PMA1 coding regions (Fig. 7B and C). However, the rpd3Δ/hos2Δ mutation did not cause any further reduction in Spt6Δ202-Myc occupancy at these genes, supporting the idea that HDACs stabilize Spt6 association rather than independently recruit Spt6 in coding regions. This is further supported by ChIP-chip experiments showing that the Spt6Δ202-Myc binding profile in WT cells was almost identical to that of rpd3Δ/hos2Δ cells at the top 25% Spt6 occupied genes, as well as at low-expressed genes (Fig. 7D and E; see Table S4 in the supplemental material). Comparing the Spt6-Myc binding profile to that of Spt6Δ202-Myc revealed a significantly reduced Spt6 binding peak around the TSS. The Spt6-Myc binding around the TES, which was largely unaffected in bur1as/ctk1Δ and in rpd3Δ/hos2Δ mutants, was strikingly reduced in the spt6Δ202 mutant, indicating that the tSH2 domain is essential for recruiting Spt6 near the TSS, as well as near the TES. The fact that tSH2 recognizes both Ser2/Ser5 and Tyr1 phosphorylated CTD (23–25), combined with the fact that tSH2 deletion, but not inactivation, of Ser2 kinases (Fig. 2C to E) reduced Spt6 binding near the TES, strongly suggests that phosphorylated Tyr1 is needed for recruiting Spt6 in this region. Because the kinase responsible for phosphorylating Tyr1 is not yet identified in S. cerevisiae, we were unable to directly determine whether this phosphorylation is needed for Spt6 recruitment in vivo.

FIG 7.

CTD kinases and HDACs coordinate Spt6 recruitment. (A) Schematic diagram of Spt6 with tSH2 showing the amino acids deleted to obtain the spt6Δ202 mutant. (B) ChIP occupancy of Spt6-Myc in WT, spt6Δ202, rpd3Δ/hos2Δ, and spt6Δ202/rpd3Δ/hos2Δ cells at the ARG1, ADH1, and PMA1 genes. (C) Rpb3 occupancy in WT and spt6Δ202 cells at the ARG1, ADH1, and PMA1 genes. (D) Average log₂ enrichment of Spt6 at the top 25% Spt6-occupied genes in WT, spt6Δ202, and spt6Δ202/rpd3Δ/hos2Δ cells. The box at the bottom of the plots represents the transcribed region. The TSS and TES are marked. (E) Average log₂ enrichment of Spt6 at low-transcribed genes affected in the rpd3Δ/hos2Δ mutant in WT, spt6Δ202, and spt6Δ202/rpd3Δ/hos2Δ cells.

DISCUSSION

In this study, we investigated the mechanisms by which Spt6 is recruited to the coding regions of transcribed genes. We report that both Ser2 CTD kinases and HDACs coordinate Spt6 association to different regions of the transcribed genes. Our results revealed that the Pol II CTD phosphorylated at Ser2 is important for recruiting Spt6 to the 5′ORF. This CTD-dependent recruitment is, surprisingly, insufficient to maintain the high levels of Spt6 binding that is seen on transcribed regions. We provide evidence for an unexpected role of HDACs in sustaining high occupancy of Spt6 in transcribed ORFs by promoting its interaction with the phosphorylated Pol II CTD. Furthermore, our data also indicate that Spt6 recruitment to the 3′ ends of transcribed genes requires the Spt6 tSH2 domain but not Ser2/Ser5 CTD phosphorylation. This requirement of tSH2 domain for Spt6 recruitment at 3′ORFs suggests a role for Tyr1 phosphorylated Pol II CTD.

Spt6 relies on Ser2 phosphorylation for recruitment to the coding regions.

Although it was known that Ser2 phosphorylation enhances Spt6 tSH2 domain interaction with the Pol II CTD in vitro (25, 26, 52), it was not clear whether this interaction promotes Spt6 recruitment in vivo for several reasons. First, Ser2 phosphorylation peaks 1 kb after the TSS (29), whereas Spt6 is recruited near the 5′ORF. Second, deleting Ctk1, the primary Ser2 kinase, did not reduce Spt6 recruitment to the coding regions (Fig. 1B, D, and E) (27). Finally, deleting the tSH2 domain substantially reduced but did not completely abolish Spt6 binding across the ORF (Fig. 7D) (22). All of these observations lead to a suggestion that Spt6 recruitment to the 5′ORF occurs in a phospho CTD-independent manner (22, 52). Our results, however, clearly show that Spt6 recruitment to the coding regions is promoted redundantly by both Ser2 kinases (Bur1 and Ctk1) (Fig. 1). Unexpectedly, these kinases promote recruitment of Spt6 at the 5′ proximal regions (Fig. 2), whereas Ser2 phosphorylation is generally low in that region. These results are consistent with the role of Bur1 and Ctk1 in phosphorylating Ser2 near the 5′ORFs (28, 42) and is analogous to the role of Ser2 phosphorylation in recruiting the Paf1 complex (53), which is also recruited early during the transition from initiating to elongating. Our study makes two important observations regarding the role of Ser2 kinases in Spt6 recruitment. First, although bur1as reduced Spt6 recruitment in the 5′ORFs of ARG1 and ADH1, a greater reduction of Spt6 occupancy in the bur1as/ctk1Δ mutant indicates the importance of Ctk1 kinase activity for full recruitment during early transcription. Second, unlike the high levels of Ser2P required for recruiting termination/mRNA processing factors near 3′ ends of transcribed genes (27), the low levels of Ser2P at 5′ORF are sufficient to efficiently recruit Spt6. These results are further supported by our ChIP-chip results showing maximum Spt6 binding defects at the 5′ORFs in the bur1as/ctk1Δ double mutant (Fig. 2).

In addition to phosphorylating the Pol II CTD on Ser2, Bur1 also phosphorylates the Spt5 C-terminal region (CTR), which consists of 15 hexapeptide repeats (53–55). Considering this, along with the fact that Spt6 interacts with Spt5 (54), it is possible that Bur1 contributes in Spt6 recruitment through Spt5-CTR phosphorylation. However, an earlier study (54) showed that deleting Spt5-CTR did not impair Spt5-Spt6 interaction and, more recently, it was reported that Spt5-CTR did not significantly reduce Spt6 binding in vivo (56). Although the role of phospho-CTR in Spt6 recruitment cannot be ruled out, these studies, along with our results, strongly suggest that Bur1, together with Ctk1, promote Spt6 recruitment in a phosphorylated Ser2-dependent manner.

Role of HDACs in Spt6 recruitment.

In addition to CTD phosphorylation, we provide evidence that HDACs (Rpd3 and Hos2) regulate Spt6 occupancy in the coding regions. The rpd3Δ/hos2Δ double mutant reduced Spt6 binding in the ORFs (Fig. 4D and E), and the genome-wide experiments revealed that the maximum defect in Spt6 binding occurs in the middle of transcribed regions (Fig. 4G to I). These regions are enriched for H3K4 and H3K36 dimethylated histones, which are targeted by Hos2-Set3C and Rpd3S complexes for deacetylation (47–49, 51). It is therefore interesting to see that the maximum defect in Spt6 binding is observed in the region of transcribed ORFs where these HDACs deacetylate nucleosomes. Although this suggests that hyperacetylated nucleosomes suppress the association of Spt6 with coding regions in the HDAC mutant, our data suggest a new role for HDACs in regulating Spt6 occupancy apart from deacetylating ORF nucleosomes. In support of this, Spt6 occupancy (Fig. 5B and D), as well as the Spt6-Pol II interaction (Fig. 6A), are retained in the set1Δ/set2Δ mutant. Since this mutant increases histone acetylation (Fig. 5F to H) (35, 47–49), it appears that hyperacetylated nucleosomes are not sufficient to prevent Spt6 occupancy in rpd3Δ/hos2Δ cells. Our observation that rpd3Δ/hos2Δ diminishes Spt6-Pol II interactions (Fig. 5A and 6F) indicates that HDACs promote Spt6 binding to Pol II and thereby reduce Spt6 occupancy in ORFs. Loss of Spt6-Pol II interactions in the rpd3Δ/hos2Δ mutant is not totally unexpected, considering the fact that both HDACs and Spt6 bind to the phosphorylated Pol II (Fig. 5A) (35). HDACs can thereby promote this interaction. Efficient coimmunoprecipitation of Spt6 with both Rpd3 and Rco1 (Fig. 6E) provides support for these HDACs enhancing Spt6 occupancy in the coding regions. We propose that CTD-bound HDACs (Rpd3 and Hos2) strengthen Spt6 and phospho-Pol II interactions, allowing Spt6 to migrate farther into coding regions. However, this interaction is labile in the absence of HDACs, and results in reduced Spt6 occupancy in TSS-distal regions. This is supported by our results showing that deleting HDACs did not further reduce binding of Spt6 lacking the tSH2 domain (Fig. 7D and E). Whether a physical interaction between Spt6 and HDACs is sufficient to prevent premature dissociation of Spt6 from the transcribed regions or dynamic acetylation of a nonhistone protein(s), perhaps Spt6 itself, plays a role in enhancing Spt6 association with the transcribed regions remains to be elucidated.

Since Rpd3S and Hos2-Set3 are recruited to the phosphorylated Pol II CTD in a manner dependent on Ser5 phosphorylation by Kin28 (35, 51), it was surprising to see that the kin28as/bur2Δ mutation did not reduce Spt6 binding. However, drastic reductions in Pol II occupancy, particularly in the middle of coding regions in the kin28as/bur2Δ mutant, made it difficult to assess a Spt6 binding defect due to the impaired HDAC recruitment in this mutant.

Minimal coimmunoprecipitation of H3 with Spt6 in the rpd3Δ/hos2Δ mutant (Fig. 6F) is most likely the result of reduced Spt6 binding in the coding region rather than an inability of Spt6 to recognize or bind histones. This is supported by the fact that both unmodified and acetylated H3 and H4 N-terminal tail peptides pulled down similar amounts of Spt6 from whole-cell extracts (Fig. 6B). However, the reduced histone occupancy in the coding regions in the HDAC mutant (35) might have also contributed in inefficient coimmunoprecipitation of histone H3 with Spt6 in this strain.

Our results also demonstrate the importance of the tSH2 domain in recruiting Spt6 at both ends of the transcription units (Fig. 7D and E). The tSH2 domain relies on Ser2 phosphorylation for recruiting Spt6 near early transcribed regions since Spt6 occupancy is greatly reduced in the 5′ORFs in the bur1as/ctk1Δ mutant. However, the inability of the Ser2 kinase mutants to dampen Spt6 recruitment near the 3′ ends indicates that tSH2-mediated Spt6 recruitment toward the end is carried out in a Ser2P-independent manner. Given that the tSH2 binds with greater affinity to the CTD peptides phosphorylated at Tyr1 than at Ser2 or Ser5 and that the Tyr1P peak coincides with the Spt6 binding peak at the 3′ ends (23, 24), our results strongly suggest that Spt6 recruitment toward the 3′ end is mediated through Tyr1P-tSH2 domain interactions.

In conclusion, our study provides in vivo evidence for the importance of Ser2 in recruiting Spt6 to the 5′ ends of transcribed genes. As Pol II elongates farther away from the TSS, HDACs prevent premature dissociation of Spt6 from the elongating Pol II. Increasing levels of Tyr1 phosphorylation toward the 3′ end then stabilizes Spt6-Pol II CTD interaction via the tSH2 domain. A recent study showed preferential loss of histones from the 5′ ends of the transcribed genes in Spt6-depleted cells (16). We speculate that Spt6 recruitment by Ser2 phosphorylation is important for regulating histone occupancy near the 5′ ends. Spt6 interaction with HDACs would facilitate association with the mid-ORF region, which may contribute in efficient reassembly of histones displaced in the wake of Pol II elongation, and in preventing cryptic transcription which tends to originate within intragenic regions (15, 19). Finally, Tyr1P-mediated recruitment of Spt6 at the 3′ ends of genes might help in coordinating transcription termination and mRNA processing (57, 58).