The Spt4p Subunit of Yeast DSIF Stimulates Association of the Paf1 Complex with Elongating RNA Polymerase II

Yeast strains and plasmids.All yeast strains used in this work are listed in Table 1. The wild-type parent strain BY4741 and deletion derivatives thereof were described previously (48) and purchased from Research Genetics. The presence of all reported deletion alleles was confirmed by PCR amplification or complementation of mutant phenotypes by plasmid-borne wild-type genes. Myc-tagged strains were constructed as described previously (45), as were strains carrying arg1-ΔTATA (34). To construct kin28Δ::kanMX strains harboring episomal KIN28-HA or kin28-HA-ts16, the parent strains were first transformed with URA3 plasmid pHQ1429 carrying KIN28-HA. The resulting strains were transformed with a PCR fragment containing kin28Δ::kanMX4, synthesized using DNA from the heterozygous kin28Δ::kanMX/KIN28 diploid as a template, and selected on G418 medium. Deletion of chromosomal KIN28 was indicated by the conferred sensitivity to 5-fluoro-orotic acid and confirmed by PCR analysis. pHQ1429 in the confirmed kin28::kanMX strain was replaced by LEU2 plasmids carrying KIN28-HA (pHQ1430) or kin28-HA-ts16 (pHQ1431) by plasmid shuffling.

Plasmids pHQ1429 and pHQ1430 were created by inserting the HindIII-BamHI fragment containing KIN28-HA from pGK13 (11) between the corresponding sites of YCplac33 and YCplac111, respectively. pHQ1431 was created by inserting the EcoRI fragment containing kin28-HA-ts16 from pGK33 (11) into YCplac111. pHQ1437 carries the SPT4 ORF plus 329 bp and 298 bp of 5′ and 3′ flanking sequences, respectively, synthesized by PCR using genomic DNA as the template and cloned between the EcoRI and SalI sites of YCplac33.

Biochemical methods.The chromatin immunoprecipitation (ChIP) experiments were conducted as described previously (45) using the primers and antibodies described below, except that sulfometuron induction was carried out for 30 min, the time of maximal Gcn4p binding at ARG1 (5). The ARG1 upstream activation sequence (UAS) and TATA primer pairs used for ChIP analysis were described previously (34, 45). The following additional primer pairs were used to amplify the indicated ARG1 sequences: (i) 5′ORF, 5′ TGGCTTATTCTGGTGGTTTAG 3′ (nucleotide positions +23 to +43) and 5′ ATCCACACAAACGAACTTGCA 3′ (+186 to +166); (ii) middle ORF, 5′ ACATTTCTTACG AGGCAGGTA 3′ (+557 to +577) and 5′ CTTGTTGTCGGTGTAGGTCA 3′ (+720 to +701); (iii) 3′ORF, 5′ TTCTGGGCAGATCTACAA AGA 3′ (+1091 to +1111) and 5′ AAGTCAACT CTTCACCTTTGG 3′ (+1258 to +1238); (iv) down1, 5′ CTTCATGAGAACAGAACATTG 3′ (+1489 to +1509) and 5′ TTCCTTATCAGTCAATTGAGG 3′ (+1643 to +1623); (v) down2, 5′ GTTTATTCCTCGTTGTGAGGT 3′ (+1880 to +1900) and 5′ TACCAATATCAATTAGCTCA TTG 3′ (+2040 to +2018). The primer pairs used to amplify PMA1 and PYK1 sequences are as follows: (i) PMA1 TATA, 5′ CGATTATATAAAAAGGCCAAATAT 3′ (−341 to −318) and 5′ TATCAACG AGGTTGATAGAAAAA 3′ (−171 to −193); (ii) PMA1 5′ORF, 5′ CGACGACGAAGACAGTG ATA 3′ (+168 to +187) and 5′ ATTCTTTTTCGTCAGCCATTTG 3′ (+331 to +310); (iii) PYK1 TATA, 5′ CATCAAAACGATATTCGT TGG 3′ (−235 to −215) and 5′ AACCAAAGGATGAA AAAGAATG 3′ (−80 to −101); (iv) PYK1 5′ORF, 5′ ACAAGTCTGTCATTGACAACG 3′ (+179 to +199) and 5′ CAT CGG TGG TGA AGA TCA TTT 3′ (+343 to +323). The primer pair used for an internal control amplifies a nontranscribed intergenic sequence from chromosome V: 5′ CCGATTTGTGAGATTCTTCCT 3′ and 5′ GAACAAGGTTACAAATCC TGA T 3′.

Northern analysis was carried out as described previously (34). Western analysis of whole-cell extracts (WCEs) was conducted using WCEs prepared by trichloroacetic acid precipitation (35), and coimmunoprecipitation assays were conducted as described previously (53), using the antibodies described below.

Antibodies.The following antibodies were employed: monoclonal anti-Myc and anti-hemagglutinin (HA) (Roche), anti-Rpb3p (Neoclone), anti-phospho-Ser5 (H14; Covance), anti-Rpb1p (8WG16; Covance) and anti-histone H3 (05-499; Upstate); polyclonal anti-Gcd6p (2), anti-Spt4 and anti-Spt5 (gifts from G. Hartzog), anti-Rtf1p (gift from K. Arndt), anti-trimethyl-histone H3 (Lys4) (abcam, ab8580), and anti-dimethyl-histone H3 (Lys79) (07-366; Upstate).

The Paf1 complex is recruited by Gcn4p to the ARG1 coding region, dependent on transcription initiation.We showed previously that deletion of PAF1 produced Gcn⁻ phenotypes, indicating impaired transcriptional activation by Gcn4p under conditions of amino acid limitation where Gcn4p synthesis is induced. Using ChIP, we found that Gcn4p recruits a low level of Myc-tagged Paf1p to the promoter of its target gene, ARG1 (45). Here we found that Gcn4p recruits Myc-Paf1 at a much higher level to the ORF region of ARG1. Induction of Gcn4p by starvation for isoleucine and valine produced ∼6-fold higher levels of Myc-Paf1p cross-linked to the 5′ and 3′ ends of the ARG1 ORF than in isogenic PAF1-myc cells lacking Gcn4p (Fig. 1A and B). The ratio of Myc-Paf1p occupancy in GCN4 versus gcn4Δ cells is plotted in Fig. 1C for different locations at ARG1. The results show that Gcn4p-dependent Paf1p binding occurred at much lower levels in the UAS and TATA regions and immediately downstream of ARG1 compared to that in the ORF sequences (Fig. 1C). The distribution of Paf1p across ARG1 is similar to that described for other yeast genes (9, 10). As expected, Gcn4p showed high-level binding to the UAS and TATA regions but was detected in the ORF at levels only slightly above the nonspecific background levels of gcn4Δ cells (Fig. 1E). (Because the UAS and TATA elements are separated by <200 bp and the sheared chromatin fragments are ∼500 bp, most fragments containing the TATA element also contain the UAS, likely accounting for the apparent binding of Gcn4p to the TATA region.)

• Open in new tab
• Download powerpoint

FIG. 1.

Induction of Gcn4p increases Paf1p occupancy in the transcribed regions of ARG1. (A) Diagram depicting the ARG1 ORF, TATA element (open box), and Gcn4p binding sites in the UAS (gray boxes). Dashed lines depict the intergenic region between ARG1 and the adjacent gene, YOL057w. Names and positions (relative to the start codon) of fragments PCR amplified in ChIP analyses are given below. (B-E) ChIP analysis of Myc-Paf1p, Rpb3, and Gcn4p binding at ARG1 following induction of Gcn4p. PAF1-myc strains HQY397 (gcn4Δ), HQY389 (WT), and HQY855 (ΔTATA) were cultured in synthetic complete medium lacking Ile and Val and treated with sulfometuron for 30 min to induce Gcn4p and then were subjected to ChIP analysis with anti-Myc (panel C), anti-Rpb3p (panel D), or anti-Gcn4p (panel E) antibodies. DNA was extracted from the immunoprecipitates (IP) and input chromatin (Inp) samples and subjected to PCR in the presence of [³³P]dATP to amplify the radiolabeled fragments described in panel A and a control (C) fragment from an intergenic region on chromosome V. PCR products were resolved by polyacrylamide gel electrophoresis and visualized by autoradiography, with representative results shown in panel B. PCR products were quantified with a phosphorimager, and the ratios of ARG1 to control signals in IP samples were normalized for the corresponding ratios for Inp samples. The resulting values measured for the GCN4 strains were normalized to the corresponding values for gcn4Δ cells to yield the percent IP (GCN4/gcn4Δ) values plotted in histograms. (F) Northern analysis of gcn4Δ (249), WT (BY4741), and ΔTATA (HQY700) strains cultured under the conditions described above, probing for transcripts of Gcn4p target genes ARG1, ARG4, and SNZ1 and for Pol III transcript SCR1 analyzed as a loading control.

To determine whether Paf1C recruitment by Gcn4p is mediated by Pol II, we examined the effect of impairing PIC formation on Paf1p occupancy. In agreement with previous results, deletion of the ARG1 TATA element lowered Pol II (i.e., Rpb3p) occupancy in the promoter and ORF regions (Fig. 1D) and produced commensurate reductions in the steady-state level of ARG1 mRNA without affecting levels of mRNAs induced from other Gcn4p target genes (Fig. 1F). The ΔTATA mutation also decreased the size of ARG1 transcripts, most likely reflecting aberrant initiation downstream of the normal start site (Fig. 1F). Importantly, the ΔTATA mutation greatly reduced Paf1p occupancy at ARG1 (Fig. 1C), suggesting that Paf1C recruitment is dependent on PIC assembly and transcription. The small amount of shorter ARG1 transcripts produced in the ΔTATA strain (Fig. 1F) probably explains the residual occupancy of Pol II and Paf1p in the ARG1 ORF in the ΔTATA mutant (Fig. 1C and D).

The ratio of Paf1p to Pol II occupancy is lower in the TATA region and downstream from the ORF (Down1) than in the 5′ and 3′ ORF sequences at ARG1 (cf. Fig. 1C and D and also WT results in Fig. 2A and B). The relatively low Paf1p occupancy downstream of the ORF is consistent with the finding that Paf1C is released from Pol II at the polyadenylation site (10). The comparatively low Paf1p occupancy in the TATA region suggests that Paf1C associates with Pol II only after the transition from PIC to elongation complex, as discussed further below.

• Open in new tab
• Download powerpoint

FIG. 2.

Recruitment of Paf1p, but not Pol II, by Gcn4p requires the Paf1C subunits Cdc73p and Rtf1p. PAF1-myc (panel A) and RPB3-myc (panel B) strains containing no mutations (WT) or the indicated deletions were subjected to ChIP analysis using anti-Myc antibodies as described in the legend to Fig. 1. PAF1-myc gcn4Δ strain HQY397 and RPB3-myc gcn4Δ strain HQY422 were analyzed in each experiment as a negative control. Down1, region downstream of the ORF, as shown in Fig. 1A.

To confirm that Paf1p is recruited in the context of intact Paf1C, we took advantage of the fact that deletions of Paf1C subunits Cdc73p and Rtf1p dissociate Paf1p from the TEF1 gene in vivo (25). Consistent with this, binding of Paf1p at ARG1 was greatly reduced in cdc73Δ and rtf1Δ cells but unaffected by deletion of Leo1p (Fig. 2A). Pol II occupancy was unaffected in the cdc73Δ and rtf1Δ strains (Fig. 2B), suggesting that high-level recruitment of Paf1p is not required for a WT level of elongating Pol II in the ARG1 ORF (see also Fig. 6B). These last results fit with our finding that cdc73Δ and rtf1Δ mutants do not exhibit Gcn⁻ phenotypes (45) and with previous data indicating that cdc73Δ and rtf1Δ did not reduce Pol II occupancy at CLN1, TEL1, or TEF1 (25) nor decrease the rate or processivity of elongation during transcription of a Gal4p-induced gene (23).

Evidence that Paf1C interacts primarily with elongating Pol II.Although we detected a low level of Paf1p binding at the TATA region (Fig. 1C and 2A), it was possible that this signal derives from chromatin fragments that contain both promoter and 5′-ORF sequences and that Paf1C associates with Pol II only after it clears the promoter. This would explain the relatively low Paf1p:Rpb3p ratio in the TATA region versus the ORF sequences noted above. To explore this possibility, we examined the effect on Paf1p recruitment of impairing the kinase Kin28p that phosphorylates Ser5 of the CTD and stimulates promoter clearance.

We first measured the occupancy of Rpb1p hyperphosphorylated on Ser5 (Ser5P) across the ARG1 gene and found that Ser5P occupancy in WT cells was highest in the TATA region and declined from 5′ to 3′ across the ORF (Fig. 3A), in accordance with previous findings (13). In kin28-ts16 cells at 37°C, by contrast, nearly background levels of Ser5P were found at all locations tested at ARG1 (Fig. 3A). Western analysis of WCEs showed that the cellular level of Ser5P was reduced but not abolished in kin28-ts16 cells at 37°C and that the level of Kin28p itself was decreased by the ts16 mutation (Fig. 3B). We observed reduced Pol II occupancy in the 5′ and 3′ regions of the ARG1 ORF in kin28-ts16 cells at 37°C but no reduction in Pol II occupancy of the TATA region (Fig. 3C), consistent with a defect in promoter clearance. As expected from the WT level of Pol II binding to the promoter, there was no reduction in Gcn4p occupancy of the UAS in the kin28-ts16 mutant (Fig. 3D). Together, the results in Fig. 3A to D indicate that inactivation of the kin28-ts16 product had the expected effects at ARG1 of reducing the level of Ser5P with attendant reductions in promoter clearance by Pol II. The fact that occupancy in the ORF was reduced less for Rpb3p than for Ser5P by kin28-ts16 suggests that promoter clearance at ARG1 can be achieved, albeit inefficiently, by Pol II molecules hypophosphorylated on Ser5. Similar results were reported previously for this (27) and other kin28 alleles (13, 21). The presence of hypophosphorylated Pol II in the ARG1 ORF in kin28-ts16 cells could be attributed to the residual Ser5 phosphorylation (Fig. 3B) produced by either the kin28-ts16 product or Srb10p (21) that is below the detection limit of the ChIP assay or to lack of an absolute requirement for Ser5 phosphorylation in promoter clearance (22, 38).

• Open in new tab
• Download powerpoint

FIG. 3.

Optimal Paf1p recruitment to ARG1 requires Kin28p-dependent phosphorylation of the Pol II CTD on Ser5. (A-E) Isogenic PAF1-myc strains HQY963 (KIN28 [WT]), HQY964 (kin28-ts16), and HQY965 (gcn4Δ) were grown at 25°C to an optical density at 600 of ∼0.6, transferred to 37°C for 30 min, and treated with sulfometuron for an additional 30 min at 37°C. (A) ChIP analysis was conducted as described in the legend to Fig. 1 using H14 antibodies against phosphorylated Ser5. (B) Western analysis of WCEs from the strains in panel A using H14 antibodies and antibodies against Rpb3p, Myc (to detect Myc-Paf1p), HA (to detect HA-Kin28p), and Gcd6p. (C-E) ChIP analysis was conducted as described in the legend to Fig. 1 for the strains in panel A using Rpb3p antibodies (panel C), Gcn4p antibodies (panel D), or Myc antibodies (panel E). (F) ChIP analysis was conducted on PAF1-Myc strains HQY397 (gcn4Δ), HQY997 (WT), and HQY998 (kin28-as), which were cultured in synthetic complete medium lacking Ile and Val and treated with NA-PP1 at 12 μM for 12 min and then with sulfometuron for 30 min. (G) The ratio percent immunoprecipitation (GCN4/gcn4Δ) values for Myc-Paf1p in panel E were normalized to those for Rpb3p in panel C.

Importantly, kin28-ts16 at 37°C reduced to nearly background levels Paf1p occupancy at both TATA and ORF regions of ARG1 (Fig. 3E). The reductions in Paf1p occupancy produced by kin28-ts16 (Fig. 3E) were more extensive than the corresponding reductions in total Pol II occupancy (Fig. 3C), leading to a reduced Paf1p:Rpb3p ratio in kin28-ts16 versus WT cells at all three locations tested (Fig. 3G). This comparison suggests that Paf1C binds poorly to elongating Pol II molecules that escaped the promoter but are hypophosphorylated on Ser5. We reached the same conclusion after examining the kin28-as strain containing a mutant form of Kin28p that can be inactivated by the inhibitor NA-PP1 (21). As shown in the left panel of Fig. 3F, treatment of kin28-as cells with NA-PP1 reduced the Gcn4p-dependent occupancy of Ser5P in the ARG1 promoter and 5′ ORF sequences by ∼70%. Whereas inhibition of the kin28-as product decreased Rpb3p occupancy of these sequences by only ∼20% (middle panel), it decreased Myc-Paf1p binding to the TATA and 5′ ORF sequences by ∼50%. The fact that Paf1p occupancy was not reduced as much as Ser5P occupancy may indicate that Paf1C can still associate with Pol II molecules phosphorylated on a fraction of the CTD which might persist in NA-PP1-inhibited kin28-as cells.

We obtained independent evidence that Paf1C is associated primarily with elongating Pol II phosphorylated on Ser5 by coimmunoprecipitation analysis of Pol II-Paf1C complexes in WCEs. Myc-Paf1p was immunoprecipitated with anti-Myc antibodies, and immune complexes were probed with H14 antibodies specific for Ser5P (employed above in ChIP analysis) or with 8WG16 antibodies that react primarily with hypophosphorylated Rpb1p. (Although it appears that 8WG16 antibodies recognize Rpb1p that is Ser5 phosphorylated in a subset of heptapeptide repeats of the CTD, this species has been observed only in ctk1Δ cells lacking Ser2P (28).) As shown in Fig. 4, a much larger proportion of hyperphosphorylated versus hypophosphorylated Rpb1p coimmunoprecipitated with Myc-Paf1p (Fig. 4A, lanes 1 and 2), suggesting that Paf1C is preferentially associated with Ser5P. The association of Myc-Paf1p with Ser5P is specific, because it was nearly abolished by deletion of the Paf1C subunit Cdc73p (lanes 4 to 6).

• Open in new tab
• Download powerpoint

FIG. 4.

Paf1p preferentially coimmunoprecipitates with Ser5-phosphorylated Pol II dependent on Cdc73p and Spt4p. (A) WCEs from PAF1-myc strains HQY389 (WT), HQY889 (cdc73Δ), and HQY905 (spt4Δ) were immunoprecipitated (IP) with antibodies against Myc, and immune complexes were subjected to Western analysis using antibodies against Myc, Ser5P (H14), hypophosphorylated Rpb1p (8WG16), Spt5p, and Spt4p. (B) WCEs of SPT4 strain HQY993 and SPT4-myc strain HQY994, both containing PAF1-HA, were immunoprecipitated with Myc antibodies and analyzed as in panel A except using HA antibodies to detect HA-Paf1p. I, 5% of input WCE; P, 50% of immunoprecipitated pellet; S, 5% of supernatant. (C) WCEs from SPT4-myc strains HQY987 (WT) and HQY988 (kin28-ts) were immunoprecipitated with Myc antibodies and subjected to Western analysis using antibodies against Myc, Spt5p, and Rtf1p.

Recruitment of Paf1p and Pol II-Paf1C interaction requires elongation factor Spt4p.Our finding that Paf1C is preferentially associated with Ser5P but shows low occupancy of the TATA region where Ser5P occupancy peaks suggests that another factor associated exclusively with elongating Pol II is required for Paf1C recruitment. Hence, we measured Gcn4p-dependent recruitment of Paf1p in isogenic PAF1-myc strains carrying deletions of known Pol II-associated elongation factors (42). These include (i) Set2p and Dot1p, involved in histone H3 methylation; (ii) Rad6p and Bre1p, required for histone H2B ubiquitylation and H3 methylation; (iii) nucleosome-remodeling factors Chd1p and Isw1p; (iv) subunits of the THO complex (Mft1p, Thp2p, and Tex1p); (v) the Ser2 CTD kinase Ctk1p; (vi) Spt4p, a subunit of yeast DSIF; and (vii) Dst1p/TFIIS. Of the 12 deletion mutants examined, only spt4Δ produced a strong reduction in Paf1p occupancy (Fig. 5A and B). This defect was complemented by plasmid-borne WT SPT4 (see Fig. 5C, left panel), which also complements the known (6) 6-azauracil sensitivity of this spt4Δ strain (data not shown). Importantly, spt4Δ does not reduce the steady-state level of Myc-Paf1p (Fig. 5D), nor does it decrease the amount of Rtf1p that coimmunoprecipitates with Myc-Paf1p (Fig. 5F). Thus, Spt4p is important for recruitment of Paf1C to ARG1, not to maintain the cellular level of intact Paf1C.

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG.5.

Deletion of SPT4 impairs Paf1p association with ARG1, PMA1, and PYK1 without decreasing cellular levels of Paf1C or Ser5P. (A-B) PAF1-myc strains containing the indicated deletions were subjected to ChIP analysis as described in Fig. 1, using anti-Myc antibodies. PAF1-myc gcn4Δ strain HQY397 was analyzed as a negative control. (C) PAF1-myc strains HQY389 (WT), HQY905 (spt4Δ) containing an empty vector, and HQY905 containing SPT4 plasmid pHQ1437 (spt4Δ+pSPT4) were subjected to ChIP analysis as described in the legend to Fig. 1 using Myc antibodies (left panel) or Rpb3p antibodies (right panel). HQY397 was analyzed as a negative control. (D) Western analysis of WCEs from strains HQY389 (WT) and HQY905 (spt4Δ) conducted as described in the legend to Fig. 3B. (E) PAF1-myc strains HQY389 (WT), HQY905 (spt4Δ), and HQY397 (gcn4Δ) were subjected to ChIP analysis using H14 antibodies. (F) WCEs of strains BY4741 (PAF1), HQY389 (WT), and HQY905 (spt4Δ) were immunoprecipitated with Myc antibodies, and immune complexes were probed with antibodies against Myc or Rtf1p. Inp, adjacent lanes contain 1.25%, 2.5%, and 5% of input WCEs; P, 50% of pellet; S, 5% of supernatant. (G) ChIP analysis of Myc-Paf1p occupancy at the promoter and 5′ ORF of PMA1 and PYK1 measured as described in the legend to Fig. 1, using Myc antibodies and strains described in the legend to panel E, except that the results were not normalized to those observed with gcn4Δ cells.

It was possible that spt4Δ impairs Paf1C recruitment indirectly by reducing Pol II occupancy at ARG1. However, we found that spt4Δ did not reduce Rpb3p occupancy at any position in the ARG1 ORF (Fig. 5C, right panel). It was shown recently that spt4Δ decreases the processivity of elongating Pol II on a Gal4p-induced gene in vivo; however, a substantial reduction in Pol II occupancy was evident only at a distance of 8 kb from the promoter (23). Thus, a detectable decrease in Pol II occupancy in spt4Δ cells would not be predicted for the ∼1.25-kb ARG1 ORF, as we observed. We found that spt4Δ also does not reduce the cellular level of Ser5P (Fig. 5D) nor the amount of Ser5P associated with the ARG1 ORF (Fig. 5E), ruling out an indirect role for Spt4p in Paf1C recruitment at the level of Ser5 phosphorylation. We conclude that Spt4p is required for association of Paf1C with elongating Pol II phosphorylated at WT levels on Ser5.

We obtained evidence that Spt4p plays a general role in promoting Pol II-Paf1C interaction by showing that coimmunoprecipitation of Ser5P with Myc-Paf1p was greatly reduced by spt4Δ (Fig. 4A, cf. lanes 1 to 3 and 7 to 9 in second row). Spt4p is likely to function directly in linking Paf1C to elongating Pol II because Spt4p is associated with both of these factors in vivo. Thus, we found that Spt4p and its partner, Spt5p, both coimmunoprecipitated from WCEs with Myc-Paf1p in a manner reduced by disruption of Paf1C by cdc73Δ (Fig. 4A, lanes 4 to 6). In addition, both HA-tagged Paf1p and Ser5P coimmunoprecipitated with Myc-Spt4p (Fig. 4B). These findings agree with previous results indicating that Spt5p is physically associated in vivo with Pol II (6) and Paf1C (43). Importantly, we found that coimmunoprecipitation of Spt5p with Myc-Paf1p was impaired by spt4Δ (Fig. 4A, cf. lanes 1 to 3 and 7 to 9), suggesting that Spt4p enhances interaction of Spt5p with Paf1C. Finally, by ChIP experiments we found that the low-level Paf1p occupancy in the TATA region and the high-level Paf1p occupancy in the ORF at both PMA1 and PYK1 were strongly diminished by spt4Δ (Fig. 5G). Note that gcn4Δ had no effect on the Paf1p occupancy at PMA1 and PYK1 because these are not Gcn4p target genes. These results support a global role for Spt4p in Paf1C recruitment to the ORF that is not limited to activator Gcn4p or amino acid starvation conditions.

Recruitment of Spt4p to the ARG1 ORF does not require Paf1C.Our finding above that a much larger proportion of Ser5P versus unphosphorylated Rpb1p coimmunoprecipitated with Myc-Spt4p (Fig. 4B) is consistent with the idea that Spt4p is specifically associated with elongating Pol II. Supporting this interpretation, Myc-Spt4p cross-linked at higher levels to the coding region of ARG1 (both 5′ and 3′ ORF sequences) than to the TATA region (Fig. 6A, WT columns), as described above for Paf1p (Fig. 1C and 2A). In contrast to Paf1p, however, Spt4p maintains high-level occupancy in the region downstream of the ARG1 ORF (Fig. 6A, WT, Down1), as observed for Pol II itself (Fig. 6B). These findings fit with previous results on several other yeast genes (10). We found that recruitment of Spt4p to the ARG1 ORF is strongly impaired by deletion of the TATA box (Fig. 6A), just as observed for Pol II (Rpb3p) (Fig. 6B). In addition, Spt4p occupancy at ARG1 was reduced by kin28-ts16 (Fig. 6D) but to a much smaller degree than was Ser5P (Fig. 6C). In fact, Spt4p occupancy was reduced to the same extent observed for Pol II (Rpb3p), so that the Spt4p:Rpb3p ratio at ARG1 was virtually unchanged by inactivation of kin28-ts16 (Fig. 6E). Thus, it appears that Spt4p can interact with elongating Pol II that is hypophosphorylated on Ser5 of Rpb1p, a conclusion reached previously for Spt5p from coimmunoprecipitation experiments (19). Note that Spt4p differs from Paf1p, which showed little association with the ARG1 ORF in the kin28-ts16 mutant (Fig. 3E), suggesting that Paf1C is more highly dependent than Spt4p on Ser5 phosphorylation for association with elongating Pol II.

• Open in new tab
• Download powerpoint

FIG. 6.

Spt4p recruitment to ARG1 is independent of Paf1C and Ser5 CTD phosphorylation. (A-B) SPT4-myc strains containing deletions of RTF1, CDC73, or the ARG1 TATA were subjected to ChIP analysis using Myc (A) or Rpb3p (B) antibodies, with strain HQY975 (gcn4D) analyzed as a negative control. (C and D) SPT4-myc strains containing KIN28 (WT) (HQY987) or kin28-ts16 (HQY988) were subjected to ChIP analysis using H14 (C) or Myc (D) antibodies, analyzing HQY989 (gcn4D) as negative control. (E) The ratio percent immunoprecipitation (GCN4/gcn4Δ) values for Myc-Spt4p in panel D were normalized to the ratio percent immunoprecipitation (GCN4/gcn4Δ) values for Rpb3p in Fig. 3C. (F) Hypothetical model summarizing the role of Spt4p/Spt5p in bridging interaction between Paf1C and elongating Pol II phosphorylated on Ser5. Paf1C is shown interacting directly with Pol II based on the observed Cdc73p-Pol II interaction (39). Spt4p/Spt5p are shown bridging Paf1C and Pol II based on our findings on the role of Spt4p in promoting Paf1C-Pol II interaction. Paf1C is shown interacting with the Ser5-phosphorylated Pol II to reflect our finding of preferential association of Paf1p with Ser5-hyperphosphorylated versus hypophosphorylated Rpb1p. Paf1C may interact directly with the phosphorylated CTD (30), but this is conjectural, as is the suggestion that Spt4p and Spt5p each interact directly with both Paf1C and Pol II. Bur1p/Bur2p is shown stimulating the association of Paf1C with Spt4p/Spt5p and Ser5-phosphorylated Pol II based on our findings shown in Fig. 7A and previously published results (17).

The ability of Spt4p, but not Paf1p, to interact with hypophosphorylated Pol II at ARG1, plus the fact that Spt4p but not Paf1p is present at high levels downstream of the ARG1 ORF, implies that Spt4p association with Pol II is independent of Paf1C. Mueller et al. reported that Spt5p occupancy in the CLN1 ORF was reduced by ∼50% in paf1Δ and rtf1Δ mutants (25). However, we found that Spt4p occupancy in the ARG1 ORF was unaffected by disruption of Paf1C in the rtf1Δ and cdc73Δ mutants (Fig. 6A). Thus, while strong Paf1C recruitment to the ARG1 ORF requires Kin28p and Spt4p, recruitment of Spt4p to the ARG1 ORF is independent of intact Paf1C and is reduced by impairing Kin28p only to the extent that Pol II occupancy in the ORF is diminished. The simplest explanation for our findings is that Spt4p (likely in association with Spt5p) interacts with elongating Pol II and mediates recruitment of Paf1C in a manner stimulated by CTD phosphorylation (Fig. 6F). A final piece of evidence supporting a stimulatory role for Ser5 phosphorylation in Paf1C recruitment is provided by the data in Fig. 4C showing that inactivation of kin28-ts16 eliminates coimmunoprecipitation of the Paf1C subunit Rtf1p with Myc-Spt4p without affecting interaction of Myc-Spt4p with Spt5p.

Deletion of BUR2 reduces association of Spt4p, Spt5p, and Pol II with Paf1p.It was reported recently that deletion of BUR2, encoding the cyclin partner of cyclin-dependent protein kinase Bur1p, impaired Paf1C occupancy of the PMA1 ORF without reducing Pol II occupancy of this gene (17). We made a similar observation at the ARG1 gene, where bur2Δ nearly eliminated Paf1C recruitment by Gcn4p but had little impact on Pol II occupancy (Fig. 7A, compare right and left panels). Given our findings on the importance of Spt4p in Paf1C recruitment, we asked whether BUR2 deletion would reduce association of Spt4p with Paf1C. As shown in Fig. 7B, bur2Δ essentially eliminated coimmunoprecipitation of Ser5P with Myc-Paf1p, providing additional evidence that Bur2p is required for Paf1C recruitment by Pol II (17). Interestingly, bur2Δ also greatly reduced association of Spt4p with Myc-Paf1p and substantially diminished Spt5p-Paf1p interaction (Fig. 7B). Thus, Bur2p may promote Paf1C-Pol II association at least partly by enhancing interaction of Spt4p/Spt5p with Paf1C (Fig. 6F).

• Open in new tab
• Download powerpoint

FIG. 7.

Deletion of SPT4 reduces global histone H3-K4 methylation. (A) Deletion of BUR2 impairs Paf1p association with ARG1 without affecting Pol II occupancy. PAF1-myc strains HQY389 (WT) and HQY1001 (bur2Δ) were subjected to ChIP analysis as described in the legend to Fig. 1 using Myc (left panel) or Rpb3 (right panel) antibodies, with HQY397 (gcn4Δ) analyzed as a negative control. (B) Bur2p promotes association of Paf1C with Spt4p, Spt5p, and Ser5-phosphorylated Pol II. WCEs from PAF1-myc strains HQY389 (WT) and HQY1001 (bur2Δ) were immunoprecipitated with Myc antibodies, and immune complexes were probed with Myc antibodies, H14 antibodies, 8WG16 antibodies, and antibodies against Spt5p and Spt4p. (C) spt4Δ reduces H3-K4 trimethylation at ARG1. PAF1-myc strains HQY389 (WT) and HQY905 (spt4Δ) were subjected to ChIP analysis as described in the legend to Fig. 1 using antibodies against H3-K4-Me3. The results were not normalized to those observed in gcn4Δ cells. (D) spt4Δ reduces H3-K4-Me3 in bulk histones. Increasing amounts of WCEs prepared from independent cultures of HQY389 (WT) and HQY905 (spt4Δ) were subjected to Western analysis using antibodies against histone H3-K4-Me3, H3-K79-Me2, and total H3. (E) Densities of bands in lanes 2, 5, 8, and 11 of panel C were calculated with a scanner and NIH image software (version 1.63), and the values for H3-K4-Me3 and H3-K79-Me2 in the duplicate cultures were averaged and normalized to the corresponding averages for total H3. IP, immunoprecipitation.

Deletion of SPT4 reduces histone H3-K4 trimethylation.Mutants lacking Rtf1p or Paf1p are defective in transcription-coupled histone H3 methylation, including trimethylation of Lys4 (H3-K4-Me3) by Set1p and dimethylation of Lys79 (H3-K79) by Dot1p (14, 27). We obtained two pieces of evidence that disrupting Paf1C-Pol II interaction by spt4Δ is sufficient to reduce H3-K4-Me3 at ARG1 and many other loci. First, ChIP analysis using antibodies specific for H3-K4-Me3 revealed a significant reduction in this modification at the 5′ end of ARG1 in spt4Δ cells (Fig. 7C). Next, we quantified the levels of H3-K4-Me3 in bulk histones by Western analysis. Interestingly, spt4Δ lowers the ratio of H3-K4-Me3 to total H3 by ∼70% but does not significantly reduce the corresponding ratio for H3-K79-Me2 (Fig. 7D and E). A selective reduction in H3-K4-Me3 was unexpected; however, Laribee et al. (17) observed a similar phenotype in bur2Δ cells. Our findings suggest that the reduction in Paf1C-Pol II association in spt4Δ cells is sufficient to reduce H3-K4-Me3 formation but not to impair dimethylation of H3-K79.