Requirement of RSF and FACT for Transcription of Chromatin Templates in Vitro

See allHide authors and affiliations

Science  04 Dec 1998:
Vol. 282, Issue 5395, pp. 1900-1904
DOI: 10.1126/science.282.5395.1900


Transcription of naked DNA in vitro requires the general transcription factors and RNA polymerase II. However, this minimal set of factors is not sufficient for transcription when the DNA template is packaged into chromatin. Here, a factor that facilitates activator-dependent transcription initiation on chromatin templates was purified. This factor, remodeling and spacing factor (RSF), has adenosine triphosphate–dependent nucleosome-remodeling and spacing activities. Polymerases that initiate transcription with RSF can only extend their transcripts in the presence of FACT (facilitates chromatin transcription). Thus, the minimal factor requirements for activator-dependent transcription on chromatin templates in vitro have been defined.

Intense biochemical efforts have resulted in the purification of a minimal set of factors necessary for transcription of class II genes in vitro. This minimal transcription system consisting of the general transcription factors (GTFs) TFIIB, TFIID, TFIIE, TFIIF, and TFIIH and RNA polymerase II (RNAPII) was established with assays that reconstituted accurate transcription from class II promoters on naked DNA templates (1, 2). In vivo, the DNA template is organized by histones into chromatin. The minimal RNAPII transcription system cannot transcribe DNA that is packaged into chromatin (3). However, transcription can be reconstituted on chromatin templates with nuclear extracts in the presence of an activator, suggesting that accessory factors are present in crude nuclear extracts that facilitate RNAPII transcription from chromatin templates (4).

Different adenosine triphosphate (ATP)–dependent nucleosome-remodeling complexes have been isolated. The NURF (nucleosome-remodeling factor) and ACF (ATP-utilizing chromatin assembly and remodeling factor) complexes were purified from Drosophila nuclear extracts and have been shown to facilitate activator-dependent transcription on chromatin templates (5, 6). Several other ATP-dependent chromatin remodeling enzyme complexes have been purified from different organisms: SWI/SNF from yeast, human, andDrosophila; RSC (remodels the structure of chromatin) from yeast; and CHRAC (chromatin accessibility complex) fromDrosophila (7). Biochemical studies with the human and yeast SWI/SNF complexes have suggested that it may participate in transcription by assisting factor binding (8–10) and transcription elongation (11) in the context of chromatin.

All previous studies that analyzed the role of remodeling enzymes in transcription of chromatin-assembled templates used crude transcription systems. We set out to identify factors that would allow a highly defined reconstituted transcription system to transcribe purified chromatin templates. To identify factors that facilitate transcription initiation on chromatin templates, we used an abortive initiation assay (12), in which the RNAPII transcription machinery is given only the nucleotides necessary to form the first phosphodiester bond (13). This assay scores for the ability of the RNAPII transcription machinery to access promoter elements and catalyze the synthesis of the first phosphodiester bond. Initiation of transcription from chromatin templates required an activity present in a crude fraction derived from HeLa cell nuclei (Fig. 1A, lane 4). This transcription initiation was specific, because no dinucleotide was formed when RNAPII and GTFs were omitted (Fig. 1A, lane 3).

Figure 1

(A) Abortive transcription initiation with purified chromatin templates, GAL41-94-VP16, and the reconstituted transcription system is dependent on a crude fraction (DEAE-52 fraction). Reactions were reconstituted on chromatin templates with additions as indicated. The GTFs and RNAPII used were highly purified as depicted in Fig. 3. “ApC” denotes the dinucleotide product formed. (B) Nucleosome-remodeling assay. Various combinations of GAL41-94-VP16, ATP, and the crude fraction (DEAE-52 fraction) were incubated with purified chromatin, followed by limited Mnase digestion and Southern blot analysis with a promoter probe (left) or distal probe (right). In this as in all other experiments that use Mnase, the reaction samples were split and digested with two different concentrations of Mnase, as indicated by the arrow on the top of the panel. The location of the promoter and distal probes in relation to the GAL4 DNA-binding sites and TATA box is shown below. (C) Coomassie blue staining of a polyacrylamide-SDS gel loaded with fractions derived from the last step of RSF purification (Superose 12 column). Protein molecular mass markers were loaded on the first lane, and their masses are expressed in kilodaltons on the left side of the figure. IP denotes the input sample loaded onto the Superose 12 gel filtration column. MW, molecular weights. (D) Abortive initiation assay on chromatin templates with fractions from the Superose 12 column. The reaction loaded in the first lane was reconstituted with the Superose 12 column input, although TFIID was omitted. (E) Nucleosome-remodeling assay with fractions from the Superose 12 column, ATP, and GAL41-94-VP16. Promoter proximal nucleosome-remodeling activity is shown in the top panel. Periodic nucleosome spacing observed with the distal probe is shown in the bottom panel.

We next analyzed the crude fraction for ATP-dependent nucleosome-remodeling activity. The chromatin remodeling assay used (14) was adapted from Tsukiyama et al.(15). For this assay, periodically spaced chromatin templates that were assembled in Drosophila S-190 extract, Sarkosyl-treated, and purified by gel filtration were used (16). The templates contained five GAL4-binding sites upstream of the adenovirus major late promoter. This chromatin was incubated with various combinations of ATP, GAL4-VP16, and the crude fraction. The chromatin was then digested with Micrococcal nuclease (Mnase), and the products were analyzed by Southern (DNA) blotting with probes corresponding to the promoter region (promoter probe) or a region of the plasmid 1000 base pairs away from the promoter (distal probe). Remodeling was indicated by a loss of periodic nucleosome spacing. Promoter proximal nucleosome remodeling was observed with the crude fraction in an ATP- and activator-dependent manner (Fig. 1B). Furthermore, this remodeling occurred only in the vicinity of the promoter because reprobing with a distal probe revealed periodically spaced nucleosomes.

We used both the abortive initiation assay and the nucleosome-remodeling assay to purify the factor that facilitates activator-dependent transcription initiation on chromatin templates (17). Both activities coeluted through four chromatographic steps. The active fractions from the final step of purification (Superose 12 gel filtration column) contained two polypeptides, a 325-kD polypeptide (p325) and a 135-kD polypeptide (p135), that eluted with a native molecular mass of 400 to 500 kD, suggesting that they form a heterodimer (Fig. 1C). A similar size was estimated for the complex when it was subjected to sucrose gradient sedimentation (18). As shown in Fig. 1, D and E, the transcription initiation activity and the nucleosome-remodeling activity coeluted during the last step of purification. The nucleosome-remodeling activity of the purified factor is ATP-dependent (Fig. 2A) and specific for the promoter region because reprobing with a distal probe revealed regular periodic nucleosome spacing. We also found that the transcription initiation activity of RSF is nucleosome specific, because it does not stimulate transcription initiation reconstituted on naked DNA templates (18).

Figure 2

The purified factor exhibits ATP-dependent nucleosome remodeling and spacing activities. (A) The remodeling activity of RSF is ATP-dependent. Purified chromatin templates and GAL41-94-VP16 were incubated with various combinations of RSF and ATP, as indicated on the figure. The reactions were then subjected to partial digestion with Mnase and Southern blot hybridization with the promoter probe (lanes 1 to 6) or distal probe (lanes 7 to 12). (B) RSF exhibits ATP-dependent nucleosome spacing activity. Irregularly spaced chromatin was incubated with ATP alone (lanes 1 and 2), RSF alone (lanes 3 and 4), or RSF and ATP (lanes 5 and 6) followed by limited Mnase digestion and Southern blot analysis. (C) Silver staining of a polyacrylamide-SDS gel containing a portion of RSF derived from the last step of purification (Superose 12). Peptide sequencing by microcapillary HPLC–ion trap mass spectrometry identified the 135-kD polypeptide as human SNF2h. (D) Schematic comparison of the identity between Drosophila ISWI, human SNF2L, and human SNF2h; a.a., amino acid.

We found that addition of the remodeling factor and ATP substantially improved the regularity of bulk nucleosome spacing. Therefore, we tested the remodeling factor in a nucleosome-spacing assay (19, 20) that measures the conversion of irregularly spaced chromatin into nucleosome arrays with regular periodic spacing. The substrate for this assay was chromatin formed in S-190 assembly extract with an excess of core histones in the absence of ATP, Sarkosyl-treated and purified by gel filtration. As shown inFig. 2B, the remodeling factor in the presence of ATP converted irregularly spaced chromatin (lanes 1 to 4) into arrays of nucleosomes with periodic spacing (lanes 5 and 6). The addition of ATP alone to the irregularly spaced chromatin did not alter the nucleosome spacing (lanes 1 and 2). Because the remodeling factor also exhibits nucleosome-spacing activity, we named it RSF for remodeling and spacing factor. The nucleosome-spacing activity of RSF was similar to that reported for the Drosophila factors ACF (6) and CHRAC (20).

SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining of purified RSF (Superose 12 column, fraction 16) revealed that it is composed of two subunits, a 325-kD polypeptide and a 135-kD polypeptide (Fig. 2C). We identified the subunits of RSF by sequencing peptides generated by in-gel tryptic digestion using microcapillary high-performance liquid chromatography (HPLC) directly coupled to an ion trap mass spectrometer (21). Thirty-four peptide sequences revealed that the 135-kD subunit is hSNF2h, a human homolog of Drosophila ISWI. hSNF2h is a newly identified member of the SNF2 family (22). On the basis of computer comparisons, hSNF2L is a human ISWI homolog (23, 24) having 74.9% identity toDrosophila ISWI (Fig. 2D). The hSNF2L and hSNF2h proteins share 85.9% identity (Fig. 2D), suggesting that they have evolved from a gene duplication event. Fifteen tryptic peptide sequences revealed that there are no cDNAs that correspond to the 325-kD polypeptide deposited in the GenBank database. The 325-kD subunit may be susceptible to degradation because the sequences of peptides derived from a few minor polypeptides of about 180 kD revealed that they matched identical expressed sequence tags as the 325-kD polypeptide.

Next, we used a reconstituted transcription system to determine whether the ability of RSF to direct transcription initiation using a reconstituted transcription system was dependent on an activator. The reconstituted transcription system used was composed of the highly purified recombinant GTFs TFIIB, TFIIE, and TFIIF and the affinity-purified multisubunit mammalian factors TFIID, TFIIH, and RNAPII (Fig. 3). In addition, the system also contained the highly purified recombinant factor GAL4-VP16 and the coactivators, PC4 and TFIIA (Fig. 3), that are minimally required for activation (25). The chromatin used for transcription assays was assembled in Drosophila S-190 extract, Sarkosyl-treated, and purified by gel filtration. SDS-PAGE followed by silver staining of the purified chromatin templates revealed that they are greater than 95% pure (Fig. 3). Transcription initiation on chromatin templates requires both RSF and an activator (Fig. 4A, lane 7). GAL4-VP16 did not promote initiation in the absence of RSF (lane 4). Also, the addition of RSF in the absence of GAL4-VP16 (lane 5) or in the presence of GAL41-94 (lane 6), which lacks an activation domain, did not result in initiation. Only the combination of RSF and GAL4-VP16 resulted in strong initiation (lane 7). This initiation was RNAPII-specific because no initiation product was detected in an identical reaction where GTFs were omitted (lane 8).

Figure 3

Silver staining of polyacrylamide-SDS gels containing the factors used to reconstitute transcription on chromatin templates. The labels on the right side of each panel denote the factor or its subunits.

Figure 4

Activator-dependent reconstituted transcription on chromatin templates minimally requires GTFs, RSF, and FACT. (A) Initiation of transcription with the reconstituted system on chromatin templates requires both an activator and RSF. The presence or absence of RSF and the addition of GAL41-94-VP16 or GAL41-94 are indicated on the figure. GTFs were omitted in lane 8. Lane 1 is transcription initiation reconstituted on naked DNA. (B) Productive full-length transcription on chromatin templates requires RSF and FACT. Transcription was reconstituted under the same conditions used for initiation assays, except that the reactions were provided with a nucleotide mixture sufficient to generate a 390-nt transcript. RSF, FACT, and GAL41-94-VP16 or GAL41-94 were added as indicated in the figure. GTFs were omitted from the reaction loaded in lane 8. Lane 1 is full-length transcription reconstituted on naked DNA.

Next, we determined whether these initiated polymerases could elongate their transcripts. As shown in Fig. 4B (lane 9), the reconstituted system was not capable of full-length productive transcription in the presence of RSF and an activator. Under identical conditions, transcription initiation was observed (Fig. 4A, lane 7). We have previously demonstrated that the failure of the reconstituted transcription system to elongate its transcripts on remodeled templates is due to nucleosome-induced stalling (26). Productive transcription through nucleosomes requires a factor termed FACT (facilitates chromatin transcription) (26). FACT is a heterodimeric factor of 140- and 80-kD polypeptides (Fig. 3), whose subunits have recently been identified (27). The addition of FACT in the presence of RSF and an activator facilitated the production of a 390-nucleotide (nt), full-length transcript (Fig. 4B, lane 7). However, FACT did not promote transcription in the absence of RSF, even when an activator was present (lane 4). Consistent with the results from our initiation assays, strong transcription was not observed in the absence of an activator (lanes 2 and 5) or in the presence of the GAL41-94 DNA-binding domain alone (lanes 3 and 6). The activator-dependent transcription observed in the presence of RSF and FACT was RNAPII-specific, because no product was detected when GTFs were omitted (Fig. 4B, lane 8).

This report describes a system that is capable of reconstituting both the initiation and elongation phases of transcription from chromatin templates in vitro. Previous studies that reconstituted transcription from chromatin templates used crude systems with many unidentified proteins (3). RSF is capable of directing activator-dependent transcription initiation on chromatin templates. We believe that RSF facilitates transcription initiation by remodeling nucleosomes in the promoter region to allow the formation of preinitiation complexes. However, the activities of RSF are not sufficient to allow polymerases to extend their transcripts. Elongation through nucleosomes requires FACT, a chromatin-specific elongation factor (26). Although we did not use theDrosophila NURF complex in our assays, we expect it will substitute for RSF. Also, because the activities of theDrosophila ACF complex are similar to those of RSF, it may also facilitate transcription initiation in our system. We do not believe that the SWI/SNF complex or the Drosophila CHRAC complex will substitute for RSF in our assays because a recent study found that these complexes could not substitute for NURF in an in vitro transcription assay (5).

In this study, we have defined the minimal factor requirements for activator-dependent transcription on chromatin templates. It is likely that there are other activities present in HeLa cells that will function in reconstituted chromatin transcription assays. Additional coactivators or mediator type complexes may be required for high levels of transcriptional activation. Also, histone acetyltransferase complexes may stimulate transcription from chromatin templates. The use of chromatin transcription assays will provide a more natural environment to study transcription factors in vitro.

  • * To whom correspondence should be addressed: E-mail: reinbdf{at}


Stay Connected to Science

Navigate This Article