|Home | About | Journals | Submit | Contact Us | Français|
Nucleosomes containing histone variant H2A.Z (Htz1) serve to poise quiescent genes for activation and transcriptional initiation. However, little is known about their role in transcription elongation. Here we show that dominant mutations in the elongation genes SPT5 and SPT16 suppress the hypersensitivity of htz1Δ strains to drugs that inhibit elongation, indicating that Htz1 functions at the level of transcription elongation. Direct kinetic measurements of RNA polymerase II (Pol II) movement across the 9.5-kb GAL10p-VPS13 gene revealed that the elongation rate of polymerase is 24% slower in the absence of Htz1. We provide evidence for two nonexclusive mechanisms. First, we observed that both the phospho-Ser2 levels in the elongating isoform of Pol II and the loading of Spt5 and Elongator over the GAL1 open reading frame (ORF) depend on Htz1. Second, in the absence of Htz1, the density of nucleosome occupancy is increased over the GAL10p-VPS13 ORF and the chromatin is refractory to remodeling during active transcription. These results establish a mechanistic role for Htz1 in transcription elongation and suggest that Htz1-containing nucleosomes facilitate Pol II passage by affecting the correct assembly and modification status of Pol II elongation complexes and by favoring efficient nucleosome remodeling over the gene.
Transcription by RNA polymerase II (Pol II) takes place on a compositionally and topologically complex chromatin template at multiple steps, including promoter activation, transcription initiation, elongation, and termination. A number of mechanisms facilitate the initiation of gene transcription on chromatin templates. Two of the most prominent of these are the remodeling of nucleosomes by DNA-dependent ATPase complexes and the covalent posttranslational modification of histone proteins (20, 90). At a minimum, these pathways are thought to act by establishing a permissive chromatin structure at gene promoters, allowing access of site-specific transcription factors and the subsequent recruitment of polymerase and the general factors (25, 69).
Nucleosomes also pose a barrier to transcriptional elongation, and a variety of conserved elongation factors have been found to facilitate transcription on chromatin templates (6, 54, 57). Among these are the DSIF, Spt6, FACT, Elongator, PAF, and COMPASS complexes. While the mechanisms by which these factors carry out their elongation functions are less well understood, they include the ability to influence either transcription-coupled nucleosome assembly and disassembly, as in the case of Spt6 and FACT (9, 22, 42, 74, 85), or covalent protein modification, as in the case of Elongator, Set2, and COMPASS (44, 50, 70, 102, 104).
Histone variants provide a fundamentally different mechanism for modulating chromatin through altered composition (7, 37, 73). Histone variant H2A.Z is conserved from yeast to humans, comprising roughly 10% of the total H2A in the cell. It is an essential histone in many organisms and is conditionally essential in the budding yeast Saccharomyces cerevisiae, where it is encoded by the HTZ1 (YOL012C) gene (30, 78, 92). Htz1 has roles in a variety of distinct functions, including gene expression, chromosome segregation, silencing, and blocking the spread of heterochromatin (30, 63, 65, 78, 98). The X-ray crystal structure of nucleosomes containing H2A.Z has been determined and, together with the results of biochemical experiments, argues that H2A.Z alters the physical properties of the chromatin template (7). Histone H2A.Z is itself subject to posttranslational modifications, including acetylation, ubiquitylation, and sumoylation, and these may alter its chromatin association, exchange dynamics, structure, and function (8, 32, 39, 41, 43, 66, 81, 84, 95). H2A.Z is broadly but nonuniformly distributed throughout the chromosomes and is deposited by specific remodeling complexes, such as the yeast SWR1 complex (SWR1) (3, 45, 51, 60, 67).
H2A.Z has important functions during transcription initiation (19, 30, 32, 34, 92, 99). Early experiments in budding yeast revealed that Htz1 is partially redundant with the Swi/Snf nucleosome remodeling and SAGA histone modification complexes, providing genetic evidence for a role in transcription activation. Furthermore, at several individual loci Htz1 was preferentially enriched over the promoters of inactive genes and became relatively depleted when those genes were induced (83). Subsequent genome-wide mapping studies have shown that this principle of Htz1 occupancy is generally applicable to a large fraction of regulated genes, leading to the hypothesis that H2A.Z poises genes for transcriptional initiation (2, 29, 58, 77, 99, 108). Several mechanisms have been proposed for this activity, including recruitment of transcription factors, conditioning of promoter nucleosome architecture, and facilitating nucleosome eviction (1, 30, 99, 108).
Less well understood is the relationship between H2A.Z function and transcription elongation. Recent genome mapping experiments in Drosophila melanogaster tissue culture cells have highlighted differences in H2A.Z nucleosome compositions associated with polymerase pausing, suggesting potential roles in elongation (100). Genetic screens in budding yeast have revealed that many of the strongest genetic interactions involving htz1Δ strains are with genes encoding transcription elongation factors, including dst1, rpb9, set2, spt4, and spt16, and with multiple genes encoding subunits of the PAF complex (11, 15, 49, 51, 53, 62, 98, 101). At present, however, the interpretations of these data are unclear. It is possible that the observed genetic interactions stem from the combined insults of an Htz1 initiation defect and a separate defect in elongation. That is, for genes already defective for initiation, further impeding elongation could produce synthetic lethality. Alternatively, Htz1 could have more direct roles in transcription elongation that are currently unrecognized. Here we report the results of experiments designed to explore these alternative models. In these studies we employed a combination of mutational approaches to characterize new positive interactions with genes for elongation, together with molecular approaches to measure the rates of elongation, the composition of the elongation complexes, and the nucleosome occupancy of transcribed open reading frames (ORFs). The results of these experiments indicate that Htz1 plays a positive role in the process of transcription elongation by influencing the composition of elongation complexes and facilitating nucleosome remodeling over the gene.
The yeast strains used in this work are summarized in Table 1. Yeast extract-peptone-dextrose (YPD), synthetic complete (SC), 5-fluoroorotic acid (5-FOA), and presporulation and sporulation media were made as described previously (4). 6-Azauracil (6-AU) was added to SC medium lacking uracil (SC-Ura medium) at a final concentration of 50 μg/ml to make 6-AU plates. Gene disruptions were performed by one-step gene replacement selecting for URA3 or G418 resistance. The chromosomal ELP3-HA, SPT5-HA, and RPB3-HA alleles were created by a PCR-based method that fuses the sequence derived from plasmid pFA6a-3HA-kanMX6, encoding three copies of the hemagglutinin (HA) epitope, to the 3′ ends of the ELP3, SPT5, and RPB3 open reading frames (ORFs). SPT4 deletion strains were constructed by replacing the SPT4 ORF with the natMX4 cassette, which confers resistance to the antibiotic nourseothricin.
Plasmid pRS314::KlURA3, used in recombination-mediated PCR-directed mutagenesis, contains the Kluyveromyces lactis URA3 gene cloned into the EcoRI site of pRS314 (kindly provided by M. Christman). Plasmid pMSS102 contains the ADE3 gene cloned into pRS425. Plasmid pMSS104, used to provide HTZ1 in the mutagenesis of the SPT genes, has a 1.5-kb SalI-HindIII (blunted) genomic fragment containing HTZ1 cloned into the SalI-SmaI sites of pMSS102. Plasmid pMSS107 was constructed by cloning an NsiI 4.3-kb fragment from ATCC clone 70978 containing SPT5 into the PstI site of pRS314. Plasmid pMSS110 is the BseRI gapped version of pMSS107 used to rescue the mutation in spt5-28. A 4,667-bp BamHI fragment from plasmid pCC48 (kindly provided by Fred Winston) containing SPT16 was cloned into pRS314 to construct pMSS132. A 5-kb PCR fragment amplified from spt16-18 genomic DNA and digested with AatII-BglII was cloned into pMSS132 to generate pMSS137. Plasmid pMSS115, the disruption construct for ELP3, was constructed by cloning a HincII kanamycin resistance (KanR) cassette between the HincII sites of ELP3. The disruption plasmid for DST1, pMSS86, was constructed by cloning a BglII-XhoI (blunted) fragment from a KanR cassette into the BglII-BsmI (blunted) sites of DST1.
Recombination-mediated PCR mutagenesis was carried out as described previously (80). The yeast strain MSY1786 was created to screen for mutations that are synthetic lethal with htz1Δ by colony color sectoring (10, 38, 47). This strain has a deletion of HTZ1 and carries ade2 and ade3 mutant alleles as well as a plasmid containing HTZ1, ADE3, and LEU2. Cells that carry ade2 are blocked in the purine biosynthetic pathway and accumulate a red imidazole intermediate. However, the ade3 mutation is epistatic to ade2, blocking the purine biosynthetic pathway at an upstream step, and therefore ade2 ade3 double mutants form white colonies (38, 47). In the absence of selection for the plasmid the host yeast strain gives red/white sectored colonies. Mutations that confer synthetic lethality together with htz1Δ make cells dependent on the plasmid and thus give nonsectoring red colonies.
Sequence analysis of spt16-18 revealed several sites at which a mixture of wild-type and mutant nucleotides were recovered, suggesting heterozygosity. Flow cytometry of its DNA content confirmed that the spt16-18 strain contained diploid DNA, suggesting that the mutated spt16-18 DNA fragment integrated into a cell that had undergone diploidization either prior to or during the transformation to become MATa/MATa SPT16/spt16-18. The MATa/MATa diploid was crossed to a MATα/MATα strain, and the resulting tetraploid strain was dissected down to haploid spores through two rounds of sporulation. Dissection of tetrads from the intermediate MATa/MATα SPT16/spt16-18 diploid produced a 2:2 ratio of live to dead spores in which the viable segregants carried wild-type SPT16. These results show that spt16-18 is nonviable on its own and recessive to wild-type SPT16 but causes either haploinsufficiency or dominant synthetic lethality in the absence of HTZ1. Haploinsufficiency was ruled out by the observation that introduction of SPT16 on a plasmid into the SPT16/spt16-18 diploid failed to restore growth and red/white sectoring in the absence of HTZ1. We PCR amplified a DNA fragment spanning the spt16-18 KlURA3 locus and integrated it by one-step gene replacement into a homozygous htz1Δ/htz1Δ SPT16/SPT16 diploid carrying wild-type HTZ1 on a plasmid. The resulting heterozygous SPT16/SPT16-18htz1 strain required the HTZ1 plasmid for viability, confirming the dominant synthetic lethality of SPT16-18htz1.
Yeast strains for the specialized miniarrays were manually selected from the systematic gene knockout collection created in the MATa BY4741 background (26) (EUROSCARF, Institute for Molecular Biosciences, Frankfurt, Germany). Deletions having negative genetic interactions with htz1Δ and set3Δ were chosen based on annotations in the Saccharomyces Genome Database (http://www.yeastgenome.org). Synthetic genetic array analysis was carried out as described previously (96), crossing htz1Δ (MSY4100), and htz1Δ swr1Δ (MSY4506) query strains against the miniarray deletion collections. The growth of the final haploid pinned colonies was recorded by digital photography, and colony sizes were determined using custom software based on the Python Imaging Library (Pythonware). Colony sizes were normalized across different plates using a set of control strains arrayed with the test sets.
Yeast whole-cell extracts and nuclei were prepared as described elsewhere (104). Approximately 30 μg of protein was used in SDS-PAGE gels. Anti-HA antibody 12CA5 (University of Virginia Hybridoma Center) was used at 1:2,000 (0.5 μg/ml) dilution. Antibodies 8WG16, H14, and H5 (Covance) were used at dilutions of 1:500, 1:10,000, and 1:1,000, respectively. The 8WG16 and 12CA5 antibodies were detected using horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG at 1:10,000 (Amersham Pharmacia Biotech). The H14 and H5 antibodies were detected using conjugated HRP-donkey anti-mouse IgM at 1:10,000 (Jackson Immunoresearch Laboratories). Rabbit anti-glucose-6-phosphate dehydrogenase (Sigma) was used at 1:5,000. This antibody was detected with HRP-conjugated donkey anti-rabbit IgG (Amersham Pharmacia Biotech) at 1:5,000 dilution.
Strains were grown to a density of 1.5 × 107 to 1.6 × 107 cells/ml in 2% raffinose media, and then galactose was added to a final concentration of 2% and cells were incubated for 1 h. Cells were fixed with 1% formaldehyde for 1 h (Elp3-HA, Spt5-HA, C-terminal domain [CTD], phospho-Ser5 [Ser5-P]) or 20 min (Ser2-P). Subsequent steps in the Elp3-HA and Spt5-HA chromatin immunoprecipitation (ChIP) experiments were carried out as described previously (83). For Pol II-CTD, Ser5-P, and Ser2-P, after quenching the formaldehyde with 250 mM glycine for 5 min, cells were lysed with glass beads in FA lysis buffer (50 mM HEPES-KOH, pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% deoxycholate [DOC]) for Pol II and Ser5-P or FA lysis buffer containing 0.1% SDS for Ser2-P. Chromatin was sheared by sonication to an average size of 300 bp. One milliliter of chromatin solution containing approximately 1 mg/ml of protein was incubated with 8WG16 antibody or H14 antibody (Covance) overnight at 4°C. For Ser2-P, 10 μl of H5 antibody (Covance) was preincubated overnight with 40 μl of magnetic beads coupled to a goat anti-IgM (Dynal), washed with phosphate-buffered saline (PBS), and then added to the chromatin solution overnight. For Ser5-P, chromatin solution containing H14 antibody was incubated with polyclonal goat anti-mouse IgM (Caltag) for 2 h at 4°C. Protein A-Sepharose beads (Pharmacia) were added to Pol II or Ser5-P samples, and samples were incubated for 2 h at room temperature. Samples were washed once for 5 min at room temperature with 1 ml of FA lysis buffer, i.e., FA buffer containing 500 mM NaCl, LiCl wash (250 mM LiCl, 1% NP-40, 1% DOC, 1 mM EDTA, 10 mM Tris-HCl, pH 8.1), and twice with Tris-EDTA (TE). After reversal of cross-linking, DNA was precipitated and analyzed by PCR under conditions ensuring that the products were linearly proportional to the amount of input template. The sequences of the primers used for GAL1 are available on request. The sequences for primers in the intergenic region of chromosome V are as in reference 46.
For the GAL10p-VPS13 ChIPs, strain JSY677, generously provided by J. Svejstrup, was crossed to MSY1262 and one of the RPB3 copies of the resulting diploid strain was tagged with the HA epitope. Strains MSY3183 and MSY3188, used in the ChIP experiments, were spore segregants from dissection of that diploid strain. Cells were grown to a density of 0.8 × 107 to 1 × 107 cells/ml in rich media containing 2% raffinose as the carbon source, and then galactose was added to a 2% final concentration for 2 h (time [t] = 0), followed by addition of glucose to a 4% final concentration. At 1-min time intervals following glucose addition, cells were fixed in 1% formaldehyde for 20 min and quenched with glycine at a final concentration of 125 mM. Cells were processed for ChIP as described above. Antibody 12CA5 against HA-tagged RPB3 was preincubated overnight with protein-Dynal magnetic beads, washed, and then added to chromatin extracts for an overnight incubation at 4°C. PCRs were set up using SYBR green master mix (Applied Biosystems) and carried out in a 7300 ABI system.
The rates for Pol II runoff elongation were estimated assuming a constant rate specific for each genotype. The sets of Pol II ChIP occupancy data for each strain were fit to the function S(t) = A/ [1.0 + e−(t − t1/2)/σ] + b, where A is the amplitude of the ChIP signal, t is the time after glucose addition, t1/2 is the time when the initial Pol II occupancy at a probe has decreased by one-half, σ is the decay constant defining the sharpness of Pol II passage, and b is an offset for the background ChIP signal. The inflection points of the curves at the set of ChIP probes, t1/2, are related by the function t1/2 = (p/r) + δ, where r is the elongation rate in kilobases per minute, p is the position of the ChIP probe in kilobases, and δ is a time constant for the phenotypic lag between the time of glucose addition and the start of elongation runoff. Twenty-eight data points were fit to five parameters (A, b, σ, r, and δ) separately for the HTZ1 and htz1 data using the Levenberg-Marquardt least-squares algorithm, as implemented in ScientificPython (http://dirac.cnrs-orleans.fr/plone/software/scientificpython/). The statistical variance in the fit was estimated by bootstrap analysis in which each of the fitted data points was perturbed by a randomly selected residual from the set of residuals to generate derived data sets. The least-squares fit was then calculated for each of 10,000 such data sets.
We first sought to determine whether transcription elongation was in fact impaired in the htz1Δ mutant. Previous systematic pharmacogenetic screens had revealed that htz1Δ single mutants have decreased resistance to 6-azauracil and mycophenolic acid (MPA) (17, 82). By inhibiting inosine 5′-monophosphate (IMP) dehydrogenase, both of these drugs decrease the pool of ribonucleotides in vivo, which is not well tolerated in mutants with known defects in transcription elongation (21, 35, 61, 64, 76). Thus, the hypersensitivity of the single htz1Δ mutant to these agents is potentially strong evidence for a role in elongation. However, both drugs also have other physiological targets that do not involve elongation (17, 82). Only 6-AU-sensitive cells that fail to induce the PUR5 gene, encoding IMP dehydrogenase, are actually defective in transcription elongation (88). Under permissive growth conditions PUR5 induction is only modestly affected by htz1. However, in the presence of hydroxyurea (HU), which blocks DNA replication, htz1Δ cells fail to induce PUR5 transcription, suggesting that Htz1 becomes important for PUR5 induction under conditions that make transcription-dependent chromatin remodeling necessary (55). We reasoned that if the 6-AU-sensitive phenotype of htz1Δ cells is at the level of Pol II elongation, then mutant alleles of dedicated elongation genes might either exacerbate that phenotype or, more tellingly, suppress the hypersensitivity.
SPT4 and SPT5 encode subunits of a transcription elongation factor that is conserved from yeast to humans, where it is termed DSIF, and the complex is known to play both positive and negative roles in regulating elongation (103, 107). SPT4 is not an essential gene, and a previous deletion mutant screen identified spt4 as enhancing the slow-growth phenotype of htz1Δ (15). In agreement with that result, we find that htz1Δ spt4Δ double mutants in a W303 strain background exhibit synthetic hypersensitivity to 6-AU (Fig. 1A).
SPT5 is an essential gene, and therefore we targeted recombination-mediated PCR mutagenesis directly to the chro mosomal copy of SPT5 (80). Mutagenized cells were then screened for synthetic lethality, as revealed by a nonsectoring red colony morphology (see Materials and Methods). This strategy identified two alleles, spt5-24 and spt5-28, that conferred a severe slow-growth phenotype in combination with htz1Δ and a strong nonsectoring red phenotype (Fig. 1B). Both of the single mutants also displayed a severe cold-sensitive growth defect (Fig. 1C), a phenotype shared with other spt5 alleles characterized previously (35). Using a series of SPT5 plasmids, linearized to create gaps across the gene, we localized the mutation in spt5-28 to a 409-bp BseRI restriction fragment toward the 3′ end of the open reading frame (ORF) (see Materials and Methods). DNA sequence analysis of this segment revealed a single nucleotide substitution that changed the codon for residue 861 from leucine to an amber stop codon. This allele truncates the final 200-amino-acid C-terminal regulatory domain (Fig. 1D), which contains 15 copies of a 6-amino-acid repeat sequence (93).
Next, we screened for SPT5 alleles that suppress htz1Δ strain 6-AU hypersensitivity. This screen identified two alleles, SPT5-15 and SPT5-18 (Fig. 2). The SPT5-15 allele has mutations changing Asn920 to Asp and Ser925 to Ala, while SPT5-18 carries mutations changing Gln342 to Arg, Tyr725 to Cys, and Asp767 to Glu (Fig. 2D). In addition to suppressing the 6-AU sensitivity of htz1Δ cells, these alleles were also able to suppress the 6-AU sensitivity of the dst1 deletion mutant and partially suppress the severe defect of the htz1Δ dst1Δ double mutant (shown for SPT5-18 in Fig. 2A). This phenotype is not a bypass of 6-AU sensitivity since both SPT5-15 and SPT5-18 improve the growth of HTZ1 DST1 wild-type cells on drug plates, evident at the limit dilutions shown in Fig. 2A. The suppression phenotype is dominant, as demonstrated by the continued suppression of the 6-AU sensitivity in the htz1Δ/htz1Δ SPT5/SPT5-18 diploid strain (Fig. 2B). Identical results were obtained for SPT5/SPT5-15 cells (data not shown). Furthermore, introducing an episomal wild-type copy of SPT5 on a plasmid into either htz1Δ SPT5-15 or htz1Δ SPT5-18 haploid strains does not revert the suppression phenotype (Fig. 2C).
The FACT complex promotes transcription elongation on chromatin templates (23). In budding yeast, subunits of the FACT complex are encoded by SPT16, POB3, and NHP6A (13, 24). Previous work identified spt16-11 as synthetic lethal with htz1Δ (11), making SPT16 a good candidate for further analysis. We carried out a mutational screen of SPT16, similar to that for SPT5, and identified one synthetic lethal mutant, spt16-18, with unusual properties (Fig. 3A). Analysis revealed that this mutant strain had undergone spontaneous diploidization to MATa/MATa SPT16/spt16-18 (see Materials and Methods). Interestingly, spt16-18 alone is recessive, but in combination with htz1Δ it becomes dominant synthetic lethal. We will refer to the spt16-18 allele in this context as SPT16-18htz1. Using a primer pair in which the reverse primer was specific for the KlURA3 sequence, we amplified the spt16-18 locus and determined the DNA sequence of the mutant allele. From this analysis, spt16-18 is predicted to code for a truncated Spt16 protein due to the introduction of a stop codon in place of the Arg 670 codon. This truncation deletes the DUF1747 domain, which has homology to Rtt106 and other histone chaperone proteins, and partially deletes the conserved SPT16 motif. In addition, Spt16-18 is predicted to have two conservative amino acid substitutions: V120G and K648R (Fig. 3B). We were also able to identify spt16 alleles, such as SPT16-61 and SPT16-183, which suppress the 6-AU sensitivity of htz1Δ cells. Both mutant alleles are also dominant suppressors, as transformation of the haploid strains with a wild-type SPT16 plasmid fails to revert the 6-AU resistance (Fig. 3C).
To evaluate whether Htz1 might function indirectly in elongation by activating the expression of elongation genes, we examined the 56 genes in budding yeast with the Gene Ontology annotation of “RNA elongation from RNA polymerase II promoter” (GO:0006368) and compared their expression profiles in two independent wild-type HTZ1 and htz1Δ data sets (65, 67). For comparison, the expression profiles of these same elongation genes were also examined in the data sets for deletions of four genes encoding specialized sequence-specific transcription factors that do not regulate elongation genes: YAP1, ARR1, CAD1, and RPN4 (36). As shown in Fig. 4, the range of gene expression ratios for htz1Δ cells is comparable to those for the control transcriptional activators.
We also assessed the negative impact of the SWR1 complex on elongation phenotypes in the absence of its normal Htz1 substrates (32, 68). The deletion of SWR1 is partially able to suppress the sensitivity of htz1Δ cells to 6-AU and MPA (Fig. 5 A). The swr1Δ single mutant is hypersensitive to both drugs, consistent with a loss of Htz1 deposition, and the phenotype of the double htz1Δ swr1Δ mutant is similar to that of the swr1Δ mutant. Thus, the loss of Htz1 histone causes a major impairment of these drug-dependent pathways, and SWR1 contributes an additional partially negative impact. Deletion of SWR1 is also able to partially suppress the synthetic growth defect of htz1Δ dst1Δ cells, but, interestingly, swr1 has little to no effect on htz1Δ spt4Δ cells (Fig. 5B). Similarly, swr1 cannot suppress the synthetic lethality of htz1Δ spt5-24 and can only weakly suppress htz1Δ spt5-28 (Fig. 5C). These results indicate that pathways involving Spt4 and Spt5 are primarily sensitive to the loss of Htz1 itself and not the function of SWR1 in the absence of Htz1.
An attractive model for how promoter-localized Htz1 nucleosomes might function in downstream Pol II elongation is through the assembly and maturation of the productive elongation complex. To test this prediction, we assayed Pol II and other elongation factors at the inducible GAL1 gene by ChIP assays. We first examined the distribution of RNA Pol II at GAL1 using an antibody (8WG16) directed against the unmodified C-terminal domain (CTD) of Rpb1. As shown in Fig. 6A, the association of RNA Pol II with the GAL1 ORF recognized by this antibody depends on transcriptional activation by galactose. However, the levels and patterns of this association were similar in wild-type and htz1Δ cells.
The phosphorylation of CTD Ser2 and Ser5 provides two well-characterized markers for different stages of Pol II elongation (14, 46). As shown in Fig. 6B, polymerase with Ser5-P was associated with the GAL1 gene only following galactose induction and it was preferentially localized to the 5′ end of the gene as expected. Furthermore, this pattern of association was found in both wild-type and htz1Δ strains. The results of Ser2-P assays were significantly different. In wild-type cells, CTD Ser2-P was preferentially cross-linked over the body of the GAL1 gene, as expected, and this association required galactose induction. However, in htz1Δ cells the relative level of CTD Ser2-P was severely reduced (Fig. 6C). The global levels of unphosphorylated CTD, Ser2-P, and Ser5-P in whole-cell extracts were not significantly affected by deletion of HTZ1, while deletion of CTK1 results in a global loss of Ser2-P as expected (Fig. 6D). These results indicate that Htz1 is important for the normal distribution of CTD Ser2-P in elongating Pol II at GAL1 and that this function acts at the level of the local chromatin environment.
The htz1Δ-dependent defect in CTD Ser2-P predicted that there should be corresponding defects in associated elongation factors. To test this prediction, we assayed wild-type and htz1Δ cells for the loading of Spt5, which associates with Pol II independent of CTD phosphorylation (35, 52, 59, 75, 89). Spt5 protein was readily detected by ChIP assays over the body of the GAL1 gene following galactose induction, in agreement with previous studies (75). Surprisingly, in the absence of Htz1 we observed an increase in the ChIP signal for Spt5, particularly in the 5′-proximal region of the ORF (Fig. 7 A). Next, we tested the loading of Elongator, a multifunctional protein complex originally purified by its preferential association with the Ser2-P hypermodified form of elongating Pol II and known to physically associate with nascent RNA transcripts during elongation (27, 33, 72, 91). The Elp3 Elongator subunit was also detected at the GAL1 ORF following induction in wild-type cells (Fig. 7B). The ChIP signals for Elp3 were well above control levels in each of six independent experiments using different protocols and yeast strains, and Elp3 cross-linking for wild-type cells was both time and induction dependent. However, unlike that of Spt5, the association of Elp3 with GAL1 was severely impaired in cells with HTZ1 deleted (Fig. 7B). Western blot analysis showed that equivalent levels of Spt5 and Elp3 proteins were expressed in HTZ1 and htz1Δ cells (Fig. 7C). Taken together, these results suggest that the establishment or maintenance of a normal Pol II elongation complex, or both, are facilitated by chromatin containing histone variant Htz1.
To directly measure the dependence of Pol II elongation rates on HTZ1, we assayed Pol II by kinetic quantitative chromatin immunoprecipitation (ChIP) (64) across the 9.5-kb VPS13 ORF driven by the inducible GAL10 promoter (GAL10p-VPS13) (Fig. 8) (48). Cultures of HTZ1 and htz1Δ cells were grown in raffinose and then induced with galactose for 2 h. Under conditions of active transcription on galactose, the relative distribution patterns of polymerase occupancy from 5′ to 3′ across the ORF were similar for both strains, suggesting similar processivities of Pol II; however, the ChIP signal for Pol II in htz1Δ cells was approximately 70% that in HTZ1 wild-type cells throughout (Fig. 8B), which may indicate decreased polymerase loading on the gene in the absence of Htz1 (1).
The GAL10 promoter was then repressed by the addition of glucose to the media, and Pol II occupancy was assayed by ChIP of its HA-Rpb3 subunit at 1-min intervals following repression using four probe locations across the ORF. These assays were repeated for five separate experiments, starting with independent cultures, and the ChIP signals were determined by quantitative PCR from the average of two technical replicates for each experiment. When transcription was repressed in HTZ1 wild-type cells, we observed a well-behaved Pol II depletion that progressed sequentially across the ORF as the last wave of elongating polymerase completed transcription of the gene (Fig. 8C). To estimate the rate of elongation, we modeled the Pol II runoff as a shock wave with a phenotypic lag, defined by a sigmoidal function, and determined the midpoint of the inflection at each location along the ORF (see Materials and Methods). The runoff of Pol II began at the 5′ ORF probe within approximately 1.8 min of glucose addition and then propagated across the ORF at 54 ± 3 bp/s (Fig. 8D and E). This estimate of elongation rate is consistent with previous estimates of the in vivo rate, which range from 20 to 100 bp/s (5, 16, 64, 71, 87, 97).
When transcription of GAL10p-VPS13 was repressed in htz1Δ mutant cells, we also observed a pattern of elongation runoff similar to that of the HTZ1 wild type. The runoff of elongating Pol II began at the 5′ ORF probe approximately 2.4 min after glucose addition. However, it took longer for the last wave of polymerase to clear the downstream 8.3-kb probe in htz1Δ cells than it did in HTZ1 wild-type cells. The elongation rate in htz1Δ cells is approximately 41 ± 2 bp/s. Thus, the average elongation rate for Pol II on GAL10p-VPS13 is approximately 24% slower in the htz1Δ mutant than in wild-type cells.
Pol II transcription from the GAL10 promoter normally displaces nucleosomes across the open reading frame, causing a decrease in bulk nucleosome occupancy as measured by ChIP (48, 86). We reasoned that in htz1Δ cells the replacement of Htz1-containing nucleosomes by canonical H2A nucleosomes might impede Pol II elongation because of their more stable occupancy (18, 108). Indeed, Htz1-containing nucleosomes have been detected over the normal VPS13 coding sequence in previous assays involving genome-wide ChIP followed by high-throughput sequencing (ChIP-seq) (2) (Fig. 9 A). Therefore, we first measured Htz1 occupancy across the GAL10p-VPS13 open reading frame by ChIP, using the occupancy of histone H3 to normalize for nucleosome density. The maps of total and Htz1-containing nucleosomes at the normal GAL10 promoter and over the wild-type VPS13 gene are diagrammed in Fig. 9A (2). Consistent with these maps and previous reports (29, 58, 83, 108), we found that the fraction of nucleosomes containing Htz1 is enriched in the promoter region relative to the open reading frame under glucose repression (Fig. 9B, black bars). Total nucleosome density decreases over the promoter when transcription is activated by galactose (Fig. 9C), and the fraction of the remaining nucleosomes that contain Htz1 decreases as well (Fig. 9B). Nevertheless, Htz1-containing nucleosomes were also readily detected above background (Fig. 9B, dashed line) across the VPS13 coding sequence under both activating and repressing conditions. Surprisingly, the fraction of nucleosomes containing Htz1 actually increased over the 3′ third of the gene in the presence of active transcription.
Finally, we measured the transcription-dependent decrease in nucleosome density in HTZ1 and htz1Δ strains (Fig. 9C). As expected from previous studies (48, 86), in HTZ1 cells H3 nucleosome density is reduced by approximately 50% over the VPS13 ORF and by 90% over the GAL10 promoter when transcription is activated with galactose (Fig. 9C, black bars). Remarkably, this transcription-dependent decrease in nucleosome occupancy is largely absent in htz1Δ cells, over both the promoter and open reading frame (Fig. 9C, gray bars). Taken together, these data are consistent with models in which Htz1-containing nucleosomes serve to facilitate the remodeling of chromatin during Pol II transcription elongation.
Histone H2A.Z is strongly implicated in the activation of transcription initiation (19, 30, 32, 34, 92, 99). The results of the experiments reported here argue that H2A.Z also influences one or more steps in transcription elongation. First, htz1Δ produces strong synthetic growth defects in combination with mutations in genes for the transcription elongation factors Spt5 and Spt16. Second, htz1Δ single mutants are hypersensitive to elongation inhibitors, and our genetic results suggest that the mechanism of this sensitivity is at the level of transcription elongation. The mutations in SPT5 and SPT16 that suppress 6-AU sensitivity create dominant alleles, consistent with active suppression of an htz1Δ-dependent elongation defect. Third, direct biochemical measurements at GAL10p-VPS13 show that htz1Δ reduces the rate of RNA Pol II elongation runoff by approximately 24% compared to that for the wild-type HTZ1 strain. Fourth, transcription-dependent nucleosome remodeling over the GAL10p-VPS13 ORF requires Htz1. Together, these results indicate that budding yeast cells experience actual defects in Pol II transcription elongation in the absence of histone H2A.Z nucleosomes.
The specific mutations in SPT5 and SPT16 reported here fit well with their known functions and domains. The recessive synthetic lethal allele, spt5-28, expresses a truncation of Spt5 that is missing the C-terminal repeat (CTR) domain. In human cells this region is phosphorylated by P-TEFb and is critical for the elongation activation role of DSIF (106). In yeast, the BUR kinase is important for the phosphorylation of Spt5 in vivo and can also phosphorylate the Spt5 CTR in vitro (109). It has been suggested that the CTR may serve as a platform to recruit other elongation factors such as the PAF complex (106, 109). The dominant suppressor SPT5-15 encodes a sequence with two mutations immediately upstream of the CTR, one of which introduces an acidic charge that theoretically could serve as a phosphomimetic. The second dominant suppressor, SPT5-18, also encodes a sequence with two amino acid substitutions. One of these, Q342R, falls in the NGN domain, which serves as the binding interface between Spt5 and Spt4 (31). We suggest that the Q342R substitution might further stabilize the NGN domain and activate the positive elongation function of the Spt4-Spt5 complex.
The SPT16-18htz1 allele is particularly intriguing. Aside from encoding two conservative amino acid substitutions, its main feature is that it encodes a truncation that deletes 357 residues from the C-terminal end of the protein. The truncation in Spt16-18 deletes a conserved acidic C-terminal domain that is shared with histone chaperones such as Rtt106. Deletion of this domain from Spt16 destroys the ability of reconstituted FACT to bind mononucleosomes and stimulate Pol II transcription on chromatin templates in vitro (9).
While the phenotypic properties of SPT5-15 and SPT5-18 suggest functional interactions with htz1Δ during Pol II elongation, it is important to note that SPT genes have been linked to transcription initiation and that, in particular, FACT has been shown to have roles in establishing transcription initiation complexes as well as modulating promoter accessibility (11, 12, 79, 94, 105).
There are at least two models for how Htz1 might positively regulate Pol II transcription elongation. First, we suggest that Htz1 may facilitate Pol II elongation by helping to establish the complete assembly, or correct modification state, of the elongation complex. Acting within promoter-proximal chromatin, the positional cues established by Htz1-containing nucleosomes (2, 29, 77) or direct factor recruitment (1, 28, 56) could provide space, time, or cofactors needed to assemble a fully functional elongation complex. In support of this model we have detected physical alterations in the elongation complexes formed in htz1Δ mutants, including increased association of Spt5 and decreases in both the phosphorylation of CTD Ser2 and the loading of Elongator. This model also fits well with the preferential occupancy of Htz1 within promoter chromatin.
A second model for the function of Htz1 in elongation is that it serves to modulate the properties of the nucleosomes encountered by Pol II as it traverses the open reading frame. Although Htz1 is generally concentrated at promoters genome wide, it is not entirely absent from ORFs and can actually be enriched within the ORFs of individual genes (2, 32, 39, 100). Acetylated Htz1-K14ac has also been observed over the ORF and positively correlates with the level of transcription (66). The altered stability of H2A.Z nucleosomes could then facilitate Pol II elongation just as it is thought to help remodel promoter chromatin for initiation (39, 40, 60, 100, 108). Consistent with this model, we found that the transcription-dependent decrease in nucleosome occupancy is strongly attenuated in htz1Δ cells. These results are in agreement with genome-wide studies which found that the global turnover of histone H3 is decreased in htz1Δ cells, especially in regions with the highest turnover (18). This dynamic cycling of nucleosomes could be driven by the exchange of Htz1 into canonical nucleosomes through the action of the SWR1 complex (60), and this activity provides an attractive explanation for the partial role for SWR1 in specific elongation phenotypes of htz1Δ that we observed.
These models are not mutually exclusive and could be interrelated. For example, it is formally possible that the increased nucleosome density in htz1Δ cells causes the change in the composition of the Pol II elongation complex. Alternatively, the defective elongation complex could be less efficient in remodeling nucleosomes for eviction, resulting in a higher nucleosome occupancy in ChIP assays. Further studies focused on the physical and temporal relationships between Htz1 and elongation cofactors and modifications will help to discriminate among alternate models. Nevertheless, the results of the genetic and molecular experiments reported here support the hypothesis that Htz1-containing chromatin has an important direct mechanistic role in facilitating Pol II transcription elongation.
We thank Jesper Svejstrup for reagents and Carl Wu, Stefan Bekiranov, David Auble, and members of the Smith lab for helpful discussions.
This work was supported by an ASCB MAC Visiting Professorship Award to M.S.S., a Robert R. Wagner Fellowship to M.H., and grants GM28920 and GM60444 from the National Institutes of Health to M.M.S.
Published ahead of print on 28 February 2011.