Histone density is maintained during transcription mediated by the chromatin remodeler RSC and histone chaperone NAP1 in vitro

Abstract

ATPases and histone chaperones facilitate RNA polymerase II (pol II) elongation on chromatin. In vivo, the coordinated action of these enzymes is necessary to permit pol II passage through a nucleosome while restoring histone density afterward. We have developed a biochemical system recapitulating this basic process. Transcription through a nucleosome in vitro requires the ATPase remodels structure of chromatin (RSC) and the histone chaperone nucleosome assembly protein 1 (NAP1). In the presence of NAP1, RSC generates a hexasome. Despite the propensity of RSC to evict histones, NAP1 reprograms the reaction such that the hexasome is retained on the template during multiple rounds of transcription. This work has implications toward understanding the mechanism of pol II elongation on chromatin.

Nucleosomes pose a strong barrier to RNA polymerase II (pol II) elongation (1–3). An understanding of how this barrier is overcome will reveal principles central to all eukaryotic organisms. Genome-wide chromatin immunoprecipitation analyses reveal that histone density is inversely proportional to transcriptional activity (4). However, with the exception of the heat shock loci (5), most genes maintain limited nucleosome density in the coding region during transcription. Maintenance of nucleosome density prevents cryptic transcription, which can have potentially deleterious effects on gene expression and genomic integrity (6).

Transcription through chromatin in vivo requires ATP-dependent remodeling machines, histone modification enzymes, and histone chaperones (7). ATP-dependent remodeling enzymes such as SWI/SNF and remodels structure of chromatin (RSC) can mobilize and/or evict nucleosomes to create an unimpeded DNA template for transcription (8). These enzymes are found at the promoters of genes and within ORFs (9–11). Furthermore, SWI/SNF has been shown to travel with pol II in vivo, evicting histones on active genes (12), and RSC has been shown to directly interact with the RNA pol subunit Rpb5 (13). In mammalian cells, pol II pauses at an artificially introduced 601 positioning sequence when SWI/SNF is knocked down by RNAi (14). Histone chaperones such as ASF1, SPT6, and FACT (SPT16/POB3) also travel with pol II throughout transcription and probably assist in the removal of histones in front of pol II and redeposition behind (15–17). Indeed, one study suggests that FACT redeposits the original histones behind pol II (18).

Studies of pol II elongation on chromatin in vitro have revealed insights into the mechanism. Experiments by Studitsky and coworkers showed that pol II frequently stalls and backtracks from the nucleosome (19). Higher ionic strength, which weakens DNA–histone contacts, abrogates the barrier allowing pol II to pass (20). The reaction is stimulated by TFIIS (19). Although nucleosomal passage by pol II occurs at physiological salt concentration (i.e., 150 mM), the efficiency increases with increasing ionic strength (20). Further, one orientation of the 601 positioned nucleosome is more permissive to transcription than the other (21). Studies by Reinberg and coworkers have shown that human FACT (SPT16/SSRP1) protein promotes pol II elongation on chromatin in a system employing the general transcription factors (22). In a similar system, the bromodomain containing factors Brd2 and Brd3 facilitated elongation on acetylated chromatin substrates (23). In independent work, we found that the ATP-dependent RSC and SWI/SNF remodeling complexes from Saccharomyces cerevisiae were required for transcription of nucleosomal substrates in vitro and were stimulated by histone acetylation (24). In an effort to understand the role of histone chaperones, we purified and systematically analyzed the functions of ASF1, FACT, SPT6, and nucleosome assembly protein 1 (NAP1) during transcription of mononucleosomes in vitro. Surprisingly, among these, NAP1 was the most potent in stimulating transcription.

NAP1 is a homodimeric histone chaperone that binds to the histone folds of H3-H4 tetramers and H2A-H2B dimers and also interacts with the N-terminal tails of the H3-H4 tetramer (25, 26). The precise histone docking site on NAP1 is unknown but the crystal structure reveals an acidic surface (26). This surface may facilitate histone binding by neutralizing the basic charge of histones, as observed with other chaperones. NAP1 promotes nucleosome assembly by preventing nonnucleosomal histone–DNA interactions (27) and is found in the ORFs and promoters of S. cerevisiae and Schizosaccharomyces pombe genes by ChIP (28, 29). Deletion of NAP1 in S. cerevisiae affects expression of approximately 10% of the genome (30) and increases the H2A-H2B dimer density across genes (27). Some biochemical studies have supported the idea that NAP1 may play a role in transcription (31, 32). In vivo, NAP1 is recruited to sites of active transcription and functions in context with the TREX complex component YRA1, which is linked to mRNA biogenesis (28).

Previous studies had shown that NAP1 promotes the ability of RSC to evict the histone octamer from DNA (33). In our attempts to reproduce this phenomenon, we discovered that nanomolar concentrations of NAP1 and RSC promoted the loss of one H2A-H2B dimer, generating a hexasome. Under these conditions, NAP1 stimulated RSC-dependent pol II transcription of a nucleosomal template. Remarkably, under conditions where RSC would normally transfer the octamer to another DNA molecule, NAP1 promoted retention of the hexasome on the original DNA template. Hence, NAP1 and RSC coordinate to promote passage of pol II through a nucleosome, while maintaining partial nucleosome integrity.

Results

RSC-Dependent Pol II Elongation is Stimulated by Nucleosome Eviction.

Our method for establishing pol II elongation complexes on chromatin involves the use of a single-stranded C tail attached to a DNA molecule bearing the 601 positioning sequence (Fig. S1A). The C tail serves as a binding site for pol II. Upon addition of nucleoside triphosphates, pol II transcribes into the double-stranded DNA containing the nucleosome assembled from recombinant octamers. The inclusion of RNase H prevents the formation of long DNA–RNA hybrids and allows pol II to establish a transcription bubble (34). The tailed-template approach was used to obtain the first crystal structures pol II elongation complexes (35, 36). The proteins used in this study included four chaperones (ASF1, NAP1, FACT, and SPT6), two ATP-dependent remodeling machines (SWI/SNF and RSC), and pol II. All were purified to near homogeneity (Fig. S1B). Pol II binds well to the template in both the free and nucleosomal forms as shown by EMSA (Fig. S1C).

Our previous work demonstrated that RSC facilitates elongation by pol II on nucleosomal templates in an ATP-dependent manner, both enhancing the overall levels of RNA synthesis and decreasing the pausing at the nucleosome (24). Although RSC alone stimulates transcription on the nucleosome, the strongest effects are dependent on the presence of unlabeled supercoiled plasmid DNA containing the 601 sequence (termed acceptor DNA) in the reaction. At 2 nM RSC, a fivefold increase in full-length transcript was observed with the enzyme alone, but this increased to a 152-fold stimulation upon the addition of acceptor DNA (Fig. 1A). The fold-stimulation varied only twofold in numerous experiments and correlated with small differences in the amount of free DNA remaining in different nucleosome preparations.

Fig. 1.

RSC-dependent elongation requires a histone acceptor. (A) In vitro transcription with RSC in the presence or absence of acceptor DNA. Reactions containing 1 ng (0.3 nM) of C-tailed nucleosome template, 0.9 nM pol II, and 0, 0.2, or 2 nM RSC were incubated with ATP in the presence or absence of 10 ng pGEM3Z601R for 60 min. Nucleoside triphosphates containing ³²P-CTP were added for 15 min and the products were resolved on a 7 M urea 10% acrylamide gel. A PhosphorImage is shown. FL indicates the full-length transcript and arrows point to the nucleosomal arrests. Quantitation of the full-length product is indicated below each lane. (B) Nucleosome remodeling/eviction reactions with RSC in the presence or absence of acceptor DNA. Reactions containing 0.3 nM of ³²P-labeled mononucleosome template and 0, 0.6, 2, or 6 nM RSC were incubated with 2 mM ATP in the presence or absence of 10 ng pGEM3Z601R for 60 min, poly(dI:dC) was added, and the products were resolved by 4.5% native PAGE. An autoradiograph is shown. Quantitation of free DNA is shown below the gel. (C) In vitro transcription assay screening various histone chaperones. Transcription reactions with 0.9 nM pol II contained, from left to right, no addition, 2 nM RSC, and 2 nM RSC with either 42 nM NAP1, 53 nM SPT6, 49 nM FACT, 10 ng pGEM3Z601R, or 44 nM ASF1. The mean amount of FL RNA from two independent experiments is shown below each lane. A bar graph display of the data is shown in Fig. S3B.

Previous studies have shown that RSC transfers the octamer from one DNA template to another (37). We performed chromatin remodeling reactions with RSC, in the presence and absence of acceptor DNA, on nucleosomes assembled with a ³²P-labeled 601 DNA sequence. After the remodeling reaction, RSC was competed from remodeled nucleosomes with poly(dI:dC) and the nucleosomal products were separated on native gels. The data in Fig. 1B show that in the presence of a supercoiled acceptor DNA molecule, RSC initially remodels the nucleosome. Higher levels of RSC lead to eviction of the octamer, generating free ³²P-labeled DNA. At 6 nM RSC, approximately 70% of the octamer is evicted. In contrast, in reactions lacking acceptor DNA, most of the octamer is retained on the template in the remodeled state. The data suggest that the effect of RSC on transcription in vitro is associated with its ability to evict the octamer from the template. However, an important caveat is that RSC generated a product migrating faster than the remodeled nucleosome in reactions bearing acceptor DNA. We will comment on the nature of this product below because it was also associated with the action of RSC and NAP1.

NAP1 Allows RSC-Dependent Elongation Without Nucleosome Eviction.

We hypothesized, based on biochemical and genetic data, that histone chaperones would substitute for acceptor DNA and permit efficient RSC-dependent elongation. Chaperones known to be involved in transcription were purified (Fig. S1B) and tested in the elongation reaction. All of the proteins displayed some nucleosome assembly activity, the hallmark of chaperones, and FACT and NAP1 displayed the most potent effects (Fig. S2).

We next screened these histone chaperones in our in vitro transcription assay (Fig. 1C and Fig. S3A). Of the chaperones tested, NAP1 was most efficient in the ability to promote pol II elongation in the absence of acceptor DNA. NAP1 displayed a 69-fold stimulation of full-length RNA compared to the 71-fold average stimulation by acceptor DNA. ASF1 and SPT6 also stimulated slightly, but FACT did not. This result was surprising because studies by Reinberg and coworkers had shown that human FACT stimulates pol II elongation through nucleosomes (22). Formosa et al. have suggested that NHP6 facilitates the function of yeast FACT by substituting for the HMG domain of mammalian SSRP1 (38). However, addition of NHP6A to yeast FACT suppressed rather than stimulated transcription and strongly inhibited the ability of RSC to evict histones both in the presence and absence of FACT (Fig. S4). This result is consistent with experiments that suggest NHP6 functions to stabilize nucleosomes in vivo (39). It is also consistent with the observation that mutations of yeast FACT lead to the cryptic transcription phenotype (16). Cryptic transcription is hypothesized to be due to an inability to properly assemble nucleosomes in transcribed regions (3).

To rule out the possibly that NAP1 was stimulating transcription independent of RSC, we performed a reciprocal titration of RSC and NAP1 in our transcription assay (Fig. 2A). We found that stimulation by NAP1 requires RSC and that both proteins display dose-dependent effects. A time course of the reaction suggests that the template undergoes continuous transcription for up to 45 min (Fig. 2B and Fig. S3B). These data implicate the combinatorial action of RSC and NAP1 in optimal transcription. Stimulation of transcription was also observed with NAP1 and SWI/SNF, a member of the same class of chromatin remodelers as RSC (Fig. S5).

Fig. 2.

NAP1-dependent elongation requires RSC, but not nucleosome eviction. (A) In vitro transcription reactions were performed with 0, 0.2 , 0.7, or 2 nM RSC and 14 nM (+) or 43 nM (+++) NAP1. Quantitation of full-length RNA is shown below the gel. (B) Time course of in vitro transcription with 2 nM pol II, 2 nM RSC, and 43 nM NAP1. Quantitation of full-length (FL) RNA from three independent experiments is indicated below the gel. A bar graph of the quantitation is displayed in Fig. S3B. (C) Nucleosome remodeling assay with 0.6 nM RSC and 1.3 or 13 nM NAP1. No acceptor DNA was added. The question mark (?) indicates a NAP1-dependent product. (D) A nucleosome eviction assay with NAP1 was performed with 10 ng acceptor DNA and 0.6–17 nM RSC. NAP1, 4.8–130 nM, was titrated into reactions containing 17 nM RSC. Quantitation of free DNA is shown below the gel.

To further study the mechanism by which NAP1 stimulates elongation, we employed a nucleosome remodeling assay. We found that NAP1 does not facilitate RSC-dependent nucleosome eviction in the absence of acceptor DNA (Fig. 2C). However, NAP1 with RSC promotes formation of a faster-migrating band in the gel (labeled “?”), representing a unique remodeled species or a partially disassembled nucleosome. These results were in contrast to a previous study that showed NAP1 was sufficient for nucleosome eviction by RSC (33). The discrepancy may be due to our use of lower concentrations of NAP1 (43 nM as opposed to 2.4 μM) and recombinant Xenopus laevis histones rather than rat liver histones.

Because NAP1 is a nucleosome assembly protein, we hypothesized that histones evicted by RSC might be reassembled back onto the template, explaining the lack of nucleosome eviction. To examine this possibility, we tested whether NAP1 could facilitate RSC-dependent octamer transfer to acceptor DNA. If histones were fully evicted and then reassembled onto DNA, they should be redeposited onto the supercoiled acceptor DNA, which is in significant excess over the template. Surprisingly, NAP1 blocked RSC-dependent octamer transfer and instead produced the faster-migrating species described above (Fig. 2D).

NAP1 and RSC Generate a Hexasome.

To determine the composition of the remodeled intermediate, we developed a sensitive assay to quantitate the relative amounts of H2A-H2B dimers and H3-H4 tetramers present at the subnanogram amounts in our assay. We fused protein kinase A (PKA) phosphorylation sites onto the N terminus of H3 and the C terminus of H2A (Fig. S6A). We assembled these separately into octamers and then into nucleosomes. In the presence of PKA and γ³²P-ATP, the PKA-tagged histones within the context of the nucleosome could be labeled to high specific activity enabling facile detection on gels. The labeling was specific as measured by two criteria (Fig. S6 B and C). First, only the PKA-tagged histone within the nucleosome was ³²P-labeled, as shown in the SDS gel of Fig. S6B, even though all four histones were present in the octamer. Second, only nucleosomes bearing a PKA-tagged histone were labeled as shown in the native gel of Fig. S6C. The presence of the PKA tag, either phosphorylated or unphosphorylated, did not affect the remodeling reaction or the distribution of remodeled products (Fig. S6D).

To determine the relative amounts of H2A-H2B dimer and H3-H4 tetramer in the nucleosomes remodeled by RSC and NAP1, we performed remodeling assays on nucleosomes in which the DNA, H3, or H2A were ³²P-labeled (*DNA, *H3, *H2A). We quantitated the NAP1-dependent remodeling product using ImageQuant TL. The results show that the remodeled species generated by NAP1 is still present in reactions in which DNA, H3, or H2A are separately labeled with ³²P (Fig. 3A). We noted, however, that in the reactions containing ³²P-labeled H2A, the remodeled species generated by NAP1 appeared lighter relative to the untreated nucleosome or nucleosome remodeled by RSC alone. To quantitate the remodeled species generated by NAP1, we divided the intensity of that band by the intensity of the untreated nucleosome for ³²P-labeled DNA, which normalizes the amount of nucleosome converted to the NAP1 remodeled species. We then obtained the same ratios for the reactions in which H3 or H2A were ³²P-labeled. We then divided those ratios by the ratio obtained with ³²P-labeled DNA because, in principle, if all of the histones are retained in the NAP1-generated remodeled species, then the resulting ratio should be one. Indeed in reactions containing ³²P-labeled H3, the ratio was very close to one (Fig. 3B; *H3/*DNA). However, in reactions containing ³²P-labeled H2A, the ratio was approximately 0.5 (Fig. 3B; *H2A/*DNA). In contrast, when we quantitated the nucleosome remodeled with RSC alone, the ratio was approximately one irrespective of whether H3 or H2A was ³²P-labeled. We argue that the species generated by RSC and NAP1 lacks half of the amount of H2A (and probably H2B), and is therefore a hexasome. To further validate our quantitation methodology, we repeated the reactions with ³²P-labeled histones assembled onto Cy5 labeled DNA templates (Fig. 3C). This technique allowed us to internally measure the DNA amounts within the ³²P-labeled nucleosomes using fluorescence, rather than comparing them to a separate reaction. Not surprisingly, the results were essentially identical. The ratios obtained with *H3/*DNA and *H2A/*DNA were equivalent to those obtained with *H3/Cy5-DNA and *H2A/Cy5-DNA, respectively (Fig. 3B). Similar results were obtained using a mononucleosome containing the natural 5S positioning sequence (Fig. S7 A and B), indicating that this effect is not specific to the 601 sequence.

Fig. 3.

NAP1 stimulates RSC-dependent hexasome formation. (A) Remodeling assay with NAP1 using nucleosomal substrates containing labeled DNA (*DNA), H3 (*H3), or H2A (*H2A) where indicated. Reactions contained 2 nM RSC, 43 nM NAP1, and 10 ng acceptor DNA where indicated. The hexasome migrates below the remodeled nucleosome (“Hex”). Because of differences in the labeling efficiency, a darker exposure of H3 is shown. (B) Quantitation of the histone to DNA ratios of the remodeled nucleosome and hexasome from A and C. The means from three independent experiments are displayed with the standard deviations as error bars. The ratios were calculated by dividing the intensity for each remodeled nucleosome or hexasome by the intensity of untreated nucleosome (Rem. Nuc/Nuc or Hex/Nuc). The value obtained for either *H3 or *H2A was then normalized to the value obtained for DNA as measured by Cy5 or ³²P label. P values were calculated using a two-tailed Student’s t test. (C) Remodeling assay as in A using Cy5 labeled DNA with ³²P-H3 (*H3) or ³²P-H2A (*H2A). Left panels measure fluorescence intensities of Cy5 DNA template, whereas right panels are PhosphorImages of *H3 or *H2A. As in A, a darker exposure of H3 is shown. (D) Remodeling assay with 0.3, 1, 4, or 16 nM H2A-H2B dimer added to reactions containing 2 nM RSC and 43 nM NAP1.

We performed the remodeling reactions with RSC and NAP1 in the presence of excess H2A-H2B dimer. This approach was previously employed independently by the Kornberg and Reinberg laboratories to study the composition of partially disassembled nucleosomes (33, 40). We found that increasing amounts of the H2A-H2B dimer converts the NAP1 generated remodeled species into a remodeled nucleosome (Fig. 3D). These data strengthen our argument that the remodeled species is a hexasome. Note that our inclusion of (poly)dI:dC after the reaction does not influence the appearance of the products, but is necessary to remove RSC so we can observe the hexasome (Fig. S7 C and D). Also, at high concentration of *H2A, we do observe the previously reported complex of NAP1 with H2A-H2B (25), but it migrates below the hexasome in our gels (Fig. S8) and is only weakly visible in reactions containing 0.3 nM nucleosome.

Pol II Forms Active Elongation Complexes on the Hexasome.

To obtain a snapshot of pol II in the act of transcribing the hexasomal template, we captured the elongation complexes on native polyacrylamide gels. In this experiment, complexes were detected in separate reactions containing ³²P-labeled *DNA, *H3, and *H2A (Fig. 4A). The pol II complexes with nucleosome or DNA migrate with slower but unique mobilities versus the nucleosome or free DNA alone (Fig. S1B). The complexes of pol II on the remodeled nucleosome or hexasome migrate between the pol II:DNA and pol II:Nuc complexes (see blowup of Fig. 4A in Fig. S9A). Importantly, the amount of pol II:DNA under NAP1-dependent transcription conditions did not increase when compared to RSC alone but did increase in reactions lacking NAP1 but containing RSC and acceptor DNA (Fig. 4A). This observation supports the idea that pol II is transcribing a hexasomal template in the presence of RSC and NAP1. We do not understand the precise mechanics by which pol II passes through a hexasome but it may be similar to the mechanism proposed by Studitsky and coworkers (41).

Fig. 4.

Hexasome is maintained during transcription. (A) In vitro transcription reactions with C-tailed nucleosomal templates ³²P-labeled on DNA, *H3, or *H2A. The reactions contained unlabeled nucleoside triphosphates (NTPs), with 2 nM pol II, 2 nM RSC, 43 nM NAP1, and/or 10 ng acceptor DNA, where indicated. After addition of poly(dI:dC), the products were resolved by 4.5% native PAGE. A PhosphorImage of the gel is shown. As in Fig. 3, a darker exposure of H3 is shown. (B) In vitro transcription time course (5, 15, or 45 min) as in Fig. 2B with ³²P-labeled nucleosomal templates (*DNA, *H3, or *H2A) (Txn time, transcription time). (C) Quantitation of A. The histone to DNA ratios for the pol II:Nuc band (which migrates slightly faster when remodeled by RSC) was quantified for each condition. Calculations were performed as in Fig. 3C. P values were calculated as in Fig. 3B. (D) Quantitation of B. Histone to DNA ratios for the pol II:Hex band were calculated as in Fig. 3B. P values were calculated using an ANOVA test.

The time course experiment in Fig. 4B shows that pol II remains bound to the hexasomal template throughout an extended time frame during which RNA accumulates linearly (Fig. 2B and Fig. S3B). Note that, in Figs. 4A and 4B, reactions containing RSC and NAP1 displayed a decrease in H2A signal. Quantitation of that signal by comparing the pol II:Hex to pol II:Nuc ratio of labeled histone with that of labeled DNA, revealed that the hexasomes contain approximately the same amount of H3 as DNA but approximately half the amount of H2A (Fig. 4 C and D).

To further strengthen this argument, we added a C-tail oligonucleotide to compete pol II off of the hexasomal templates during transcription. In principle, once pol II has transcribed the template, fallen off, and become bound by the C-tail oligonucleotide, it should release the hexasome, which should allow us to determine whether the hexasome remains intact over an extended time course. Indeed, at both short (5 min) and long (45 min) times, the hexasome is released (Fig. 5). This result argues that the hexasome is maintained during continuous transcription by pol II in the presence of RSC and NAP1. We also note that the pol II:Hex complexes become more resistant to C-tail oligonucleotide competition over time, which may reflect the accumulation of arrested pol II elongation complexes trapped in an inactive state.

Fig. 5.

Hexasomes are continually present. In vitro transcription reactions were performed as in Fig. 4B using ³²P-labeled DNA templates. After transcription for the indicated times, C-tail competitor was added and the products were separated by 4.5% native PAGE. A PhosphorImage is shown.

It is plausible that the hexasome generates a tetrasome intermediate during transcription. To address this idea, we assembled tetrasomes using purified H3-H4 tetramers and examined transcription in the presence of combinations of RSC, NAP1, and acceptor DNA (Fig. S9B). The tetrasome blocked full-length transcription under all conditions indicating that H2A-H2B dimers within the hexasome are required for RSC and NAP1-stimulated elongation. These data diminish the likelihood of a tetrasomal intermediate during transcription. Our work is consistent with a previous study, which reported that tetrasomes, like intact nucleosomes, are refractory to transcription elongation (42).

Discussion

ATP-dependent remodeling enzymes, histone chaperones, and chromatin modifying enzymes work in concert to allow pol II to transcribe nucleosomal DNA in vivo (7). Our efforts to reproduce the basic enzymatic requirements in vitro led to the discovery that NAP1 and RSC coordinate to allow pol II transcription through a nucleosome while maintaining a hexasome on the template. Although RSC alone displays a propensity to evict the octamer, the addition of NAP1 prevents RSC-mediated eviction but not remodeling. The formation of the hexasome seems likely to be causal for pol II transcription, as previous biochemical analyses, where transcription was stimulated by elevated salt concentrations, revealed hexasomes. Apparently, the elevated ionic strength, in conjunction with the translocase activity of pol II, promotes release of an H2A-H2B dimer during transcription (20). Hexasomes constructed with the H2A-H2B dimer at either the proximal or the distal position, relative to the oncoming pol II, revealed that the proximal H2A-H2B dimer is critical for promoting reassembly of the nucleosome behind pol II via the “0-loop intermediate” (41). In this model, the proximal H2A-H2B dimer nucleates reassembly of the nucleosome by interacting with DNA trailing the transcribing pol II. Moreover, human FACT generates a hexasome in a transcription system that does not require ATP-dependent remodeling proteins (40). Taken together, our study and others collectively point to the hexasome as an intermediate that allows the nucleosome to remain partially assembled during transcription.

We were surprised that other chaperones tested did not efficiently substitute for NAP1. SPT6 and ASF1 slightly stimulated transcription, whereas FACT had almost no effect. We note that the process of elongation in vivo involves coordinated histone and pol II modifications. In addition to histone acetylation, the pol II is phosphorylated at serines 2, 5, and 7 by Cdk7 (KIN28), BUR1, and CTK1 kinases that coordinate RNA processing with transcription at the beginning and end of the gene (43). Further, chromatin is methylated at H3K4 by COMPASS near the start of a gene and at H3K36 by SET2 and at H3K79 by DOT1 within the coding region. Therefore, a large number of coordinated modifications are necessary for proper gene control. Hence, chaperones may function only at specific steps and their action may be difficult to recapitulate without the other proteins and modifications involved.

Nevertheless, NAP1 faithfully recapitulated some chaperone-mediated processes. For example, NAP1 reprogrammed RSC to allow remodeling while simultaneously preventing nucleosome eviction, similar to the phenomenon observed by Strubin and coworkers (18), where SPT16 (FACT) apparently redeposits the original histones evicted during pol II elongation. In that study, H3 and H4 were evicted and replaced by new H3-H4 tetramers in SPT16 mutants. Our data are also consistent with the notion that ATP-dependent remodeling enzymes and chaperones function in concert. Schwabish and Struhl have shown that SWI/SNF travels in coding regions with pol II and that SWI2 mutant strains suppress cryptic transcription phenotypes caused by mutations in SPT16 (12).

Finally, the deletion of NAP1 was shown to increase the density of histone H2A and H2B in the promoter and coding regions of GAL genes in vivo (27). It was proposed that this effect was due to NAP1 preventing nonnucleosomal H2A-H2B dimer interactions with DNA. However, the data are also consistent with a model in which NAP1 facilitates hexasome formation. A loss of NAP1 would result in H2A-H2B dimer enrichment due to failure of converting nucleosomes into hexasomes. We favor a model where RSC can destabilize one H2A-H2B dimer, which NAP1 then removes.

Materials and Methods

See SI Materials and Methods for the details of experimental methods for purification of yeast proteins, preparation of PKA-tagged histones, chromatin assembly and template preparation, in vitro transcription assays, nucleosome eviction and remodeling assays, and EMSA of pol II.