Previous Article | Next Article ![]()
Molecular and Cellular Biology, December 2007, p. 8729-8738, Vol. 27, No. 24
0270-7306/07/$08.00+0 doi:10.1128/MCB.00846-07
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Stony Brook University, Graduate Program in Molecular and Cellular Biology, Stony Brook, New York 11794,1 Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724,2 Program of Pharmacology, Memorial Sloan Kettering Institute, New York, New York 10021,3 Stowers Institute for Medical Research, Kansas City, Missouri 64110,4 Center for Integrative Genomics, Faculty of Biology and Medicine, Génopode Building, University of Lausanne, 1015 Lausanne, Switzerland5
Received 14 May 2007/ Returned for modification 25 June 2007/ Accepted 1 October 2007
|
|
|---|
|
|
|---|
Chromatin modification at Pol II genes is accomplished by a large number of factors (47) that are recruited to chromatin through contacts with (i) histones with various modifications in their N-terminal regions, (ii) the DNA, and (iii) certain activators. Such activators can bind to chromatin templates and recruit chromatin-modifying factors, which then remodel nucleosomes or modify histones (11, 23). Successive waves of chromatin modifications are thought to allow for the regulated assembly of transcription initiation complexes, leading to active transcription.
Much less is known about the requirements for chromatin modification at Pol III-transcribed genes. Pol III, like Sp6 Pol, is capable of transcribing through a nucleosome on a mononucleosomal template, causing an intranucleosomal loop followed by transfer of the histone octamer to a different position of the same template (62, 63). This shows that Pol III can transcribe through a single nucleosome without the help of chromatin-modifying activities. However, a large number of observations suggests that the initiation step of Pol III transcription, as well as elongation through more than a single nucleosome, is influenced by chromatin-modifying activities. For example, transcription of an Alu element in vitro is profoundly repressed by packaging of the template into nucleosomes (10), and histone acetylation dramatically enhances transcription by Pol III through dinucleosomal templates as well as nucleosomal arrays (68, 69). Moreover, nucleosome depletion in vivo activates Pol III transcription at certain normally repressed Pol III promoters in the Saccharomyces cerevisiae (yeast) genome (16), and in the case of the yeast U6 snRNA gene, binding of the transcription factors IIIC and IIIB is accompanied by changes in nucleosome position (56). Such observations are consistent with the proposal that chromatin modification, either histone modification or chromatin remodeling or both, is indeed required. However, which factors are involved and what role, if any, is played by chromatin-remodeling factors acting on Pol III-transcribed genes are so far largely undocumented.
The human U6 small nuclear RNA (snRNA) promoter is a type 3 Pol III promoter, i.e., it is located in the 5' flanking sequence of the gene and contains two elements required for basal transcription in vitro, a TATA box and a proximal sequence element (PSE), as well as a distal sequence element (DSE) that enhances transcription from the basal promoter (17). The U6 promoter and all type 3 Pol III promoters are very similar to the Pol II promoters of snRNA genes, which also contain a DSE and a PSE but lack the TATA box (17).
The DSEs of both type 3 Pol III promoters and Pol II snRNA promoters generally contain an octamer sequence, which recruits Oct-1, and an SPH (for Sph1 postoctamer homology) sequence, which recruits a transcription factor called selenocysteine tRNA-activating factor (Staf) (52) or SPH binding factor (17, 45). Oct-1 activates transcription both through its POU (Pit-1, oct-1, unc86) DNA binding domain, which, thanks to a positioned nucleosome, contacts the PSE binding factor SNAPc and helps its recruitment, and through its activation domains (78). Staf was cloned first from Xenopus laevis, but two human proteins, ZNF143 and, to a lesser extent, ZNF76, are similar to Xenopus Staf, share similar DNA binding specificities, and can activate Pol II and III snRNA gene transcription (35). How the activation domains of Oct-1 and Staf activate transcription from DSE-containing promoters is unknown, but by analogy to activators of mRNA-encoding Pol II genes, it seems probable that the mechanism involves the modification of chromatin.
Among the large number of chromatin-binding proteins identified in recent years are the chromodomain-helicase-DNA binding, or CHD, proteins. These highly conserved but poorly understood proteins are characterized by two tandem chromodomains in their N terminus followed by an SNF2-like helicase domain. In mammalian cells, there are nine CHD proteins, which can be divided into three subfamilies according to sequence similarities (53). Subfamily I contains CHD1 and CHD2, subfamily II contains CHD3 to CHD5, and subfamily III contains CHD6 to CHD9. To date, the best-characterized CHDs are CHD1 and CHD3/CHD4. CHD1 and CHD2 contain a DNA binding domain with an AT hook (2). CHD1 can assemble chromatin in vitro (29, 46) and remodel nucleosomes (67). It associates with NCoR and histone deacetylases as well as with RNA splicing proteins, linking it to transcription repression and pre-mRNA splicing, respectively (65). CHD1 also associates with the transcription elongation machinery (25, 58) and is concentrated in actively transcribed areas, suggesting that it is involved in Pol II transcription activation, probably at the transcription elongation step. Indeed, the double chromodomain of human CHD1 interacts with histone H3 di- or trimethylated at lysine 4, a marker of active chromatin (13, 59).
The subfamily II CHD3 (Mi2a) and CHD4 (Mi2b) proteins are closely related and characterized by two PHD (plant homeodomain) zinc fingers (3, 15) located N terminal of the double chromodomain. They are in complexes containing the histone deacetylases HDAC1/HDAC2 and referred to as the Mi2 (71), NuRD (nucleosome-remodeling histone deacetylase) (76, 77), and NRD (nucleosome remodeling and deacetylating) (66) complexes. These complexes can remodel chromatin to facilitate access of deacetylases and repress transcription (72). CHD4 is, however, also implicated in transcription activation because it associates with BRG1, a subunit of the SWI/SNF complex, as well as with a complex of nucleolar proteins containing UBF, MCRS1 (microspherule protein 1), and RFP (RET finger protein) implicated in activation of rRNA gene transcription by Pol I (55). Mouse CHD5 has recently been shown to correspond to a tumor suppressor controlling cell proliferation and apoptosis (4).
The subfamily III members contain three conserved motifs called CR1 through CR3, as well as a SANT domain (1), two BRK domains, and a slightly divergent AT hook DNA binding domain, all located C terminal of the helicase domain. Human chromodomain helicase DNA binding protein 8 (CHD8) copurifies with the MLL1-WDR5 complex (8), and mouse CHD8 has recently been shown to interact with CTCF through its BRK domains and to be involved in CTCF-dependent insulator function, as its down-regulation by RNA interference (RNAi) reduced the insulator activity of the H19 DMR (differentially methylated region) insulator (19).
Here, we have explored the role of human Staf (hStaf; ZNF143) in transcription from the human U6 snRNA gene. We found that hStaf activates transcription from a U6 gene template assembled into chromatin and further identified CHD8 as an hStaf-associated protein required for efficient U6 transcription in vivo, as well as for transcription of the IRF3 gene, a Pol II-transcribed gene. These results show that a function of CHD8 is in activation of transcription from type 3 Pol III promoters. This in turn suggests that transcription of Pol III genes requires chromatin modification and that at least in the case of type 3 Pol III promoters, some of the factors used are also involved in chromatin remodeling of Pol II promoters.
|
|
|---|
Chromatin assembly, MNase footprinting, and in vitro transcription from the chromatin template. The S-190 extract was prepared and the template assembled into chromatin as described previously (78). Recombinant hStaf was added to a concentration of 90 nM at the end of the chromatin assembly reaction. The quality of the chromatin template was checked by micrococcal nuclease (MNase) digestion. MNase footprinting was performed by linear PCR as described in reference 78. For transcription assays 10 µl of chromatin template with or without added hStaf was mixed with 15 µl of HeLa whole-cell extract from which endogenous hStaf had been previously depleted by incubation with anti-hStaf antibody beads in a final volume of 40 µl containing 25 mM HEPES-KOH, pH 7.9, 60 mM KCl, 10 ng/ml of poly(dG-dC)·poly(dG-dC) (GE Healthcare), 5 mM MgCl2, 1 mM EGTA, 5 mM spermidine, 4 mM dithiothreitol (DTT), 2 mM each of ATP, GTP, CTP, and UTP, and 2 ng of recombinant TATA box-binding protein. The reaction mixture was incubated at 30°C for 1 h, and RNA was analyzed as described previously (78).
ChIPs. The chromatin immunoprecipitation (ChIP) protocol was adapted from that described in reference 78. HeLa cells were fixed with 1% formaldehyde. After nucleus lysis, chromatin was sonicated for 8 min with a Bioruptor (Diagenode) with intervals of 30 seconds on and 30 seconds off. For each ChIP sample, 600 µl of chromatin and 1 µg of antibody were mixed and incubated overnight at 4°C. Six microliters of protein G magnetic beads was then added to capture DNA/protein/antibody complexes. After being extensively washed, the DNA/protein/antibody complexes were eluted twice with 150 µl of elution buffer (50 mM NaHCO3, 1% sodium dodecyl sulfate [SDS]). The supernatant was collected and incubated at 67°C for 5 h to reverse the formaldehyde cross-linking, and the nucleic acids were precipitated and analyzed by PCR. The primers used for PCR were located in the proximal (U6-4U [5'TTCTTGGGTAGTTTGCAG 3'] and U6-2L [5' GTTTCGTCCTTTCCACAAG 3']) region of the U6-1 active promoter, in the promoter region (U64-1 and U64-120 [see reference 7]) of the inactive U6-4 promoter, or in the U1 promoter region (U1-4UP [5'CACGAAGGAGTTCCCGTG3'] and U1-2L [5'CCCTGCCAGGTAAGTATG3']) as indicated in the figure legends. For the PCRs, serial dilutions were performed in test experiments to establish conditions giving rise to linear response.
Identification of hStaf-associated proteins. The GST pull-down experiment was carried out according to the procedure described in reference 27. Briefly, 200 µg of GST fusion protein-coupled glutathione beads prepared as described above were equilibrated with buffer D200 (50 mM HEPES-KOH, pH 7.9, 200 mM KCl, 20% glycerol, 0.5 mM EDTA, 5 mM MgCl2, 1 mM DTT). Nuclear HeLa cell extracts (6) from 1 x 109 HeLa cells were diluted 1:2 with D200 buffer and incubated with glutathione beads coupled to fusion proteins for 3 h at 4°C. After the incubation, the beads were washed extensively with K200 buffer (D200 buffer without glycerol) before elution with 2 column volumes of T300 buffer (50 mM HEPES-KOH, pH 7.9, 300 mM KCl, 0.5 mM EDTA, 5 mM MgCl2, 1 mM DTT, 0.05% Tween 20) followed by 2 column volumes of T600 buffer (50 mM HEPES-KOH, pH 7.9, 600 mM KCl, 0.5 mM EDTA, 5 mM MgCl2, 1 mM DTT, 0.05% Tween 20). The eluted proteins were trichloroacetic acid (TCA) precipitated for MudPIT analysis. TCA-precipitated proteins were denatured with urea, reduced, alkylated, and digested with endoproteinase Lys-C (Roche) followed by modified trypsin (Roche) as described in reference 73. Peptide mixtures were pressure loaded onto 100 µm fused silica microcapillary columns packed with 5 µm C18 reverse phase (Aqua; Phenomenex), strong cation-exchange particles (Partisphere SCX; Whatman), and reverse phase (30). Loaded microcapillary columns were placed in-line with a Quaternary Agilent 1100 series high-performance liquid chromatography pump and a Deca-XP ion trap mass spectrometer equipped with a nano-liquid chromatography electrospray ionization source (ThermoFinnigan). Fully automated six-step MudPIT runs were carried out, as described in reference 14. Tandem mass spectra were searched with SEQUEST (9) against a database of 60,238 amino acid sequences, consisting of 34,180 human proteins (NCBI 2006-09-05 release), 177 usual contaminants (such as human keratins, immunoglobulin G, and proteolytic enzymes), 126 epitope-tagged proteins, and, to estimate false-discovery rates, 30,119 randomized amino acid sequences derived from each nonredundant protein entry. Spectrum/peptide matches were retained only if they had a normalized difference in cross-correlation scores of at least 0.08, and minimum cross-correlation score of 1.8 for singly charged spectra, 2.5 for doubly charged spectra, and 3.5 for triply charged spectra. In addition, peptides had to be fully tryptic and at least seven amino acids long. Peptide hits from multiple runs were compared using CONTRAST (64) and contrast-report (14). A further criterion for consideration was that proteins had to be detected by at least two such peptides in all combined runs. Under this set of criteria, no false-positive "shuffled" proteins were detected (false-discovery rate = 0). Spectral counts were normalized as described in reference 80 to calculate normalized spectral abundance factors, parameters used to estimate relative protein levels.
Co-IPs. Antibodies against hStaf and CHD8 were cross-linked to protein G agarose beads (Roche) with dimethyl pimelimidate (Sigma). The antibody against hStaf (antibody 19164) was raised against peptide RIASRIQQGETPGLDD, which is present in ZNF143 but not in ZNF76. The antibodies against CHD8 were raised against peptide DSLTDDSFNQVTQDPIEE (antibody 19224) or peptide SQGYDSSERDFSLIDDPM (antibody 19225). For coimmunoprecipitations (co-IPs), HeLa nuclear extracts (200 µg protein) were incubated with 100 µl of protein G beads with cross-linked antibody for 3 h at 4°C. The beads were washed extensively with K300 buffer, and the immunoprecipitated proteins were released by the addition of 100 µl of 2x Laemmli buffer and boiling for 5 min. The eluted proteins were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and analyzed by immunoblotting.
Reporter cell line construction. To establish the IMR90-C2.5 clonal cell line expressing a U6 promoter-directed unstable RNA, IMR-90Tert cells were transfected by the calcium phosphate method with 5 µg of pU6/RA.2+U6end-Dsred, a derivative of pDsRed-Express-DR (Clontech) with an insert consisting of the human U6 promoter followed by a piece of ß-globin gene cloned in the reverse orientation as in pU6/Hae/RA.2 (28), followed by the 3' end of the U6 gene and 3' flanking sequences. The cells were split 48 h later and kept under G418 selection (500 µg/ml) for 21 days. Individual clones were expanded and tested for expression of the U6 construct.
Protein knock-down with RNAi and RT-PCR. RNAi oligonucleotides were transfected into IMR90-C2.5 cells with Hi-Perfect transfection reagent (QIAGEN). The cells were collected 24 h after the last of three rounds of transfections performed at 24-h intervals. Total RNA was isolated with miRNeasy mini kit (QIAGEN). One microgram of total RNA was used for first-strand DNA synthesis with TaqMan reverse transcription (RT) reagents (Applied Biosystems), and the resulting cDNA was used as the starting material for PCRs. For the PCRs, we first diluted the PCR templates 1:10, 1:50, and 1:250 to determine, for each primer pair, the highest template concentration that still gave a signal in the linear range. These concentrations were used for subsequent experiments. The intensity of the PCR bands was measured with an Alpha imager (Alphainnotech).
CHD8 interaction with histone peptides. The protocol of CHD8 interaction with histone peptides was adapted from reference 43. hStaf-associated proteins were eluted with 600 mM KCl as described above, dialyzed against buffer D150T (10% glycerol, 20 mM HEPES-KOH, pH 7.9, 150 mM KCl, 0.5 mM EDTA, 0.05% Tween 20), and incubated with 2 µg of different biotinylated histone H3 peptides for 2 h at 4°C. The peptides were purchased either from Upstate Biotechnology or from the Tufts University peptide synthesis facility. At the end of the incubation, 20 µl of streptavidin beads was added to each sample, and the sample was incubated at 4°C for 2 h. The beads were washed and the bound proteins eluted by Laemmli buffer and boiling. The samples were analyzed by SDS-PAGE and immunoblotting with anti-CHD8 and anti-hIno80 antibodies. Antibodies for hStaf, CHD8, and hIno80 were produced by Custom Hybridoma.
|
|
|---|
![]() View larger version (35K): [in a new window] |
FIG. 1. hStaf activates U6 transcription from a chromatin template. (A) ChIP of hStaf on the U6-1 promoter. Rapidly growing HeLa cells were treated with formaldehyde, and cross-linked chromatin was extracted, sonicated, and used as starting material for IPs with the antibodies indicated above the lanes or with no antibody (No Ab). The upper panel shows a titration of input material analyzed by PCR with test primers (T) U6-4U and U6-2L specific for the U6-1 promoter. The lower panel shows immunoprecipitated material analyzed by PCR with test primers U6-4U and U6-2L as described above or control primers (C) U64-1 and U64-120 hybridizing to an inactive U6-4 gene. (B) Either naked (N; lane 1) or chromatinized (C; lanes 2 to 7) U6 template was incubated with hStaf and/or apyrase as indicated above the lanes, treated with micrococcal nuclease, and then analyzed by linear PCR as described previously (73). The location of the SPH sequence is indicated. Arrowheads mark the differences in the digestion patterns of chromatinized and naked DNA. Numbers to the left of the lanes indicate sizes of DNA markers in nucleotides. (C) hStaf activates U6 transcription from a chromatin template. A U6 template with wild-type (lanes 1, 2, 5, and 6) or mutated (lanes 3, 4, 7, and 8) SPH motif (SPH motif TTCCCATGATTCCTTCAT mutated to TTaaaATGATTCCTTCAT [lowercase italic letters indicate mutations]) was either assembled (lanes 1 to 4) or not assembled (lanes 5 to 8) into chromatin with (lanes 2, 4, 6, and 8) or without (lanes 1, 3, 5, and 7) hStaf and incubated with an hStaf-depleted HeLa cell extract for transcription. The signal corresponding to U6 promoter-directed transcription (U6), as well as an internal control (IC) corresponding to a radiolabeled G-less RNA fragment added at the beginning of the transcription reaction to control for RNA handling, is indicated. Lanes 1 through 4 are from one gel, and lanes 5 through 8 are from another.
|
We then tested the transcriptional activity of such a template. The addition of recombinant hStaf to the preassembled chromatin template before the start of the transcription assay activated U6 transcription (Fig. 1C, lanes 1 and 2), and this was dependent on an intact SPH sequence (compare lanes 2 and 4; note that lanes 1, 2, 3, and 4 are from the same gel). As observed before (45), the addition of purified recombinant hStaf to a HeLa cell extract also activated U6 transcription from a naked DNA template in an SPH sequence-dependent manner, although in our hands, this activation was rather weak (lanes 5 to 8). Thus, like many transcription factors, including chimeric proteins containing the VP16 activation domain (see, for example, reference 37), hStaf can activate transcription from both naked (45) and chromatinized template. Together, the results indicate that hStaf is capable of binding to its target site in a chromatinized U6 template and may activate transcription at least in part by recruiting histone-modifying or chromatin-remodeling factors.
Identification of hStaf-associating proteins. Xenopus Staf can be divided into three major regions: a transcription activation domain, a zinc finger domain, and a C-terminal domain with no known function (51). The activation domain contains a region required for activation of snRNA-type genes and another region composed of four imperfect tandem repeats required for the activation of mRNA-encoding Pol II genes (51). The zinc finger domain contains seven zinc fingers of the C2-H2 type, different sets of which can be used to bind to different DNA targets (49). These regions are conserved in human Staf, as illustrated in Fig. 2A.
![]() View larger version (56K): [in a new window] |
FIG. 2. MudPIT analysis of hStaf-associated proteins. (A) hStaf constructs used in the GST pull-down experiments. a.a, amino acid. (B) Proteins associated with full-length hStaf and hStaf C or with hStaf C only that were identified in the MudPIT analysis. Proteins identified only in the hStaf C pull-down experiment are marked with an asterisk. Entries for proteins known or suspected to be involved in chromatin modification are shaded blue. The proteins are arranged by the number of peptides identified and then by sequence coverage.
|
C) or N-terminal (hStaf
N) sequences, all fused to GST (Fig. 2A). We used Nef, a human immunodeficiency virus type 1 protein localized in the cytoplasm, fused to GST, as a negative control. The GST fusion proteins expressed in E. coli and coupled to glutathione agarose beads were incubated with HeLa nuclear extracts, and proteins retained on the beads were sequentially eluted with 300 and 600 mM KCl and precipitated with TCA for MudPIT analysis.
A total of 289 nonredundant proteins were identified by at least two peptide hits. After elimination of the proteins also pulled down by GST-Nef and the usual contaminants, 230 remained. Since the N-terminal domain of hStaf contains the transcription activation domains (Fig. 2A), we further eliminated 190 proteins that were pulled down by hStaf
N, which lacks these N-terminal sequences, as well as the proteins pulled down by full-length hStaf but not by hStaf
C. Twenty-five proteins that were pulled down either by both full-length hStaf and hStaf
C (13 proteins) or just by hStaf
C (12 proteins) remained. The merged list of these 25 proteins, which are likely to associate with the hStaf sequences uniquely present in hStaf
C, i.e., sequences located N terminal of the zinc finger domain (Fig. 2A), is shown in Fig. 2B, with the proteins identified only in the hStaf
C pull-down experiment indicated. Many of these proteins are involved or likely to be involved in chromatin modification (see Discussion). The protein with the most peptide hits was CHD8, with 23 peptide hits and a 13.3% sequence coverage in the hStaf
C analysis, and we therefore focused on it as a possible hStaf binding partner and transcription activator.
CHD8 primary structure. CHD8 belongs to subfamily III of CHD proteins. As shown in Fig. 3A, it contains two chromodomains, a helicase domain, three regions conserved in CHD6 through CHD9, a SANT domain, and two BRK domains. The peptide hits from our MudPIT results were matched to an old version of the CHD8 sequence, XP_370738, which was later replaced with a newer version, NP_065971. The older version contained an N-terminal 279-amino-acid extension (Fig. 3B, red) in place of the first two amino acids (methionine and lysine) of the newer version. Our data are consistent with a CHD8 protein matching the older sequence. Indeed, some of the CHD8 peptides identified by MudPIT were matched to the sequence unique to the large form, and one peptide spanned the junction of the large and small forms of CHD8 (Fig. 3B). Moreover, we raised two antibodies directed against peptides in the N- and C-terminal regions of CHD8 (Fig. 3B), which both recognize the same major band greater than 250 kDa in nuclear extracts (data not shown). Finally, BLAST searches of the NCBI protein database identified mouse (XP_619244) and rat (XP_573762) CHD8 homologs, as well as a partial CHD8 sequence from a chimpanzee (XP_509818). These three proteins contain N-terminal sequences that are 94, 95, and 99% identical, respectively, to the N-terminal extension of the human protein. These data strongly suggest that the large form of CHD8, XP_37038, exists in cells and was pulled down by GST-hStaf. They do not, however, exclude the possibility that the short form also associates with hStaf.
![]() View larger version (47K): [in a new window] |
FIG. 3. Structure of human CHD8. (A) Diagram of human CHD8. (B) Amino acid sequence of CHD8 according to the XP_370738 entry. The amino acids marked in red are the additional amino acids missing in the current human CHD8 protein sequence, NP_065971. The underlined sequences correspond to the peptide matches identified by MudPIT analysis. In some cases, the sequence underlined corresponds to several peptides of different ionization states or to two or more adjacent peptides. The two peptide sequences in uppercase letters were used to produce anti-CHD8 antibodies.
|
The CHD8 protein coimmunoprecipitates with hStaf and is localized on the U6 and U1 promoters in vivo. To confirm the interaction between CHD8 and hStaf, we tested whether the two proteins coimmunoprecipitated. As shown in Fig. 4A, hStaf was easily detected in anti-CHD8 immunoprecipitates, and CHD8 was detected, albeit weakly, in anti-hStaf immunoprecipitates (lanes 3 and 6), suggesting that CHD8 and hStaf interact inside cells. However, since hStaf is involved in transcription activation of several Pol II and Pol III genes (34, 50), it was possible that hStaf interacted with CHD8 but did not recruit CHD8 to the U6 promoter. We therefore tested whether CHD8 is localized on the U6 promoter by ChIP as described above. We also tested the Pol II snRNA promoter U1, as hStaf can also activate Pol II snRNA promoters (35). As shown in Fig. 4B, CHD8 was detected on the U6-1 promoter but not on the promoter of the inactive U6-4 gene (compare lanes 9 and 10). Moreover, both hStaf and CHD8 were detected on the Pol II U1 snRNA promoter, as shown in Fig. 4C (lanes 9 to 12). The antibody used in the ChIP specifically recognizes the large isoform of CHD8, indicating that this isoform resides on the U6 and U1 promoters. This does not exclude the possibility that the small isoform is also present.
![]() View larger version (59K): [in a new window] |
FIG. 4. CHD8 interacts with hStaf and is localized on the U6 promoter. T, test primers; C, control primers; Ab, antibody. (A) Co-IP of hStaf and CHD8. A HeLa cell nuclear extract (NE) was used as the starting material for nondenaturing IPs with antibodies indicated above the lanes or, as a control, protein G beads. The immunoprecipitates were fractionated by SDS-PAGE, transferred to a membrane, and probed by immunoblotting with antibodies directed against hStaf (left panel) or CHD8 (right panel). The percentages below the lanes indicate the fractions of the total samples loaded in the lane. (B) CHD8 is localized on the U6-1 promoter. ChIPs were performed as described for the lower panel of Fig. 1A, with the antibodies indicated above the lanes. (C) CHD8 and hStaf are localized on the Pol II snRNA U1 promoter. ChIPs were performed with the antibodies indicated above the lanes as described for the lower panel of Fig. 1A except that the precipitated material was analyzed by PCR with primers hybridizing to the human U1 promoter (see Materials and Methods for the primer sequences).
|
![]() View larger version (64K): [in a new window] |
FIG. 5. CHD8 interacts with histone H3 peptides. NE, nuclear HeLa cell extract. (A) CHD8 interacts specifically with an unmodified histone H3 N-terminal peptide and a histone H3 N-terminal peptide dimethylated at lysine 4. CHD8 partially purified over the GST-hStaf affinity column was incubated with biotinylated histone H3 peptides carrying various modifications, as indicated above the lanes. The protein-peptide complexes were recovered with streptavidin beads, fractionated by SDS-PAGE, transferred to a membrane, and probed by immunoblotting with an anti-CHD8 antibody as indicated to the right of the panels. For a control, the membrane was also probed with an anti-hIno80 antibody as indicated to the right of the panels. (B) CHD8 preferentially interacts with a histone H3 peptide dimethylated at lysine 4. The experiment was performed as described for panel A but with the H3 peptides indicated above the lanes. (C) CHD8 preferentially interacts with histone H3 peptides di- and trimethylated at lysine 4. The experiment was performed as described for panel A but with the H3 peptides indicated above the lanes. In panels A, B, and C, the H3 peptides corresponded to the first 21 amino acids of histone H3 unmodified (UM), dimethylated at lysine 4 (K4M2), acetylated at lysine 9 (K9Ac), phosphorylated at serine 10 (S10P), dimethylated at lysine 9 (K9M2), monomethylated at lysine 4 (K4M1), trimethylated at lysine 4 (K4M3), or corresponded to amino acids 21 to 44 of histone H3 dimethylated at lysine 36 (K36M2). (D) The interactions between CHD8 and the H3 K4M3 peptide are not prevented by treatment with RNase A. The experiment was performed as described for panel A, but the protein samples were left untreated or treated with water (Mock) or RNase A before incubation with biotinylated H3 K4M3 peptide. (E) The U6-1 promoter region contains histone H3 di- and trimethylated at lysine 4. ChIPs were performed as described for the lower panel of Fig. 1A. Ab, antibody; T, test primers; C, control primers.
|
Since CHD8 localizes to the U6-1 promoter and since it associates with peptides corresponding to the N terminus of histone H3 di- and trimethylated on lysine 4, we checked whether histone H3 carrying these modifications is enriched in the vicinity of the U6 promoter. Indeed, as shown in Fig. 5E, ChIPs with antibodies specific for histone H3 di- and trimethylated on lysine 4 revealed the presence of these modified H3 histones in the U6 promoter region. These results suggest that Pol III genes, like Pol II genes, can be modified by methylation of histone H3 on lysine 4, and they are consistent with the U6 promoter being a CHD8 target.
CHD8 is involved in U6 transcription. The interaction between CHD8 and hStaf as well as the localization of CHD8 on an active, but not an inactive, human U6 promoter suggests that CHD8 may be involved in U6 transcription in vivo. To address this possibility, we knocked down CHD8 in an IMR90/hTert U6 reporter cell line (IMR90-C2.5) containing an integrated hU6/Hae/RA.2 construct derivative in which the human U6-1 promoter directs the synthesis of unstable transcripts (28; see Materials and Methods). The reporter line was used because it is difficult to detect transcriptional changes by measuring the cellular U6 snRNA content. The reasons for this are not clear but could be due to the existence of different populations of U6 snRNA (for example, assembled or not assembled with snRNPs) with different stabilities. The reporter cell line was transfected with two different anti-CHD8 double-stranded silencing RNA oligonucleotides (small interfering RNAs [siRNAs]) and CHD8 mRNA and protein levels were measured with RT-PCR and Western blot analysis, respectively. RNA levels derived from the U6 reporter were tested by RT-PCR. As shown in Fig. 6A, both CHD8 protein and mRNA levels went down significantly after anti-CHD8 siRNA transfection (oligonucleotides C8-1 and C8-2) but not after control siRNA transfection (oligonucleotide C) (see the three upper panels). hStaf levels were unaffected (data not shown). In independent CHD8 knock-down experiments, we observed decreases in U6 reporter transcription ranging from 4-fold (Fig. 6A, panel U6r-a) to 0.6-fold (panel U6r-b). However, even though the decrease in the U6 transcript was moderate, it was significant (n = 6, P = 0.0022 [Wilcoxon rank sum test]; see the legend to Fig. 6C). Thus, CHD8 contributes to efficient transcription from the human U6 promoter in vivo. The modest effect of CHD8 knock-down on ongoing U6 transcription is, perhaps, not unexpected, as CHD8 may be part of an extensive network of chromatin modifying proteins at the U6 promoter, which may have partially redundant functions.
![]() View larger version (59K): [in a new window] |
FIG. 6. CHD8 is required for expression of the U6 reporter. (A) The IMR90-C2.5 reporter cell line was either not transfected (NT) or transfected with the siRNAs indicated above the lanes. After three rounds of transfection, the CHD8 and hIno80 contents were analyzed by immunoblotting (two upper panels) and the CHD8 mRNA, U6 reporter transcript, and ß-actin mRNA contents were analyzed by RT-PCR (four lower panels). U6r-a and U6r-b, expression from the U6 reporter plasmid, experiments a and b, respectively. (B) CHD8 is involved in transcription of the IRF3 gene. The IRF3 and ß-actin mRNA contents were analyzed by RT-PCR. The numbers below the lanes indicate the intensity of the RT-PCR signal normalized to the ß-actin signal. NT, cells not transfected. (C) The relative amount (nontransfected cells [NT] = 100) of U6 reporter transcript or IRF3 mRNA from control oligonucleotide (C)- or C8-2 oligonucleotide-transfected cells. The averages from six (for U6 reporter) or four (for IRF3 mRNA) independent RNAi experiments were plotted and the standard deviations (error bars) are shown. Wilcoxon rank sum tests were used to test the significance of the RNAi knock-down effect. For the U6 reporter (n = 6), the P value for C8-2-transfected cells in comparison to nontransfected cells is 0.0022 and the P value for the control oligonucleotide-transfected cells in comparison to nontransfected cells is 1. For the IRF3 mRNA (n = 4), the P value for C8-2-transfected cells in comparison to nontransfected cells is 0.0286 and the P value for the control oligonucleotide-transfected cells in comparison to nontransfected cells is 0.31. (D) CHD8 is localized on the IRF3 promoter. ChIPs were performed as described for the lower panel of Fig. 1A except that the PCR was performed with test primers specific for the IRF3 promoter (T) or control (C) primers hybridizing to an inactive U6-4 gene. Ab, antibody.
|
If CHD8 is indeed involved in IRF3 expression, one should be able to detect it at the IRF3 promoter. We performed a ChIP with anti-CHD8 and other antibodies and used a primer set specific for the IRF3 promoter to amplify the immunoprecipitated DNA fragments. As shown in Fig. 6D, the ChIP results indicated the presence of TFIIB, but not the Pol III transcription factor Brf2, on the IRF3 promoter (lanes labeled T) but not on the 5' flanking region of the inactive U6-4 gene (lanes labeled C). Importantly, hStaf as well as CHD8 was also specifically detected on the IRF3 promoter (lanes 9 to 12). The CHD8 signal was weak but comparable to that of the positive TFIIB control (lane 5). These results show that hStaf and CHD8 reside on a Pol II mRNA promoter.
|
|
|---|
If type 3 Pol III promoters recruit at least some of the chromatin-modifying activities used by Pol II genes, they may use different chromatin-remodeling activities than type 1 and 2 Pol III promoters. Indeed, the WICH complex, composed of the Williams syndrome transcription factor WSTF, SNF2h, and other factors, was recently shown to associate with the type 1 5S and the mixed type 2 and 3 7SL genes and to be involved in their transcription, as down-regulation of WSTF by RNAi resulted in reduced 5S and 7SL transcription (5). However, no function of WICH in transcription from type 3 promoters could be demonstrated. How WICH is recruited to 5S rRNA and 7SL promoters is unknown, but a key player may well turn out to be the TFIIIC complex. This complex, which is recruited by type 1 and 2 promoters but not by type 3 promoters, is essential for chromatin remodeling of the yeast U6 promoter (which, unlike the human U6 promoter, is a type 2 Pol III promoter with a TATA box) (56, 57), suggesting that the complex recruits chromatin-remodeling activities. Moreover, two of the human TFIIIC subunits display histone acetyltransferase activity (18, 26). Thus, TFIIIC may be central to chromatin modification at type 1 and 2 Pol III promoters, both modifying histone tails and recruiting chromatin-remodeling activities.
The mechanism by which hStaf and CHD8 activate transcription from the U6 promoter and how the action of hStaf combines with that of Oct-1 remain to be elucidated. In previous experiments, we showed that the Oct-1 POU domain could activate U6 transcription from a chromatin template when added during chromatin assembly (78). hStaf is capable of activating U6 transcription even when added after chromatin assembly, suggesting that it might bind to the U6 promoter before Oct-1. Its interaction with CHD8 suggests that the mechanism by which it activates U6 transcription involves recruitment of CHD8, which in turn leads to chromatin remodeling. However, it is also possible that CHD8 binds DNA before hStaf. Indeed, we showed that, like CHD1, CHD8 binds preferentially to histone H3 carrying a di- or trimethyl group at lysine 4 and that the U6 promoter region is enriched in histone H3 carrying these modifications. It is conceivable that in vivo, these modifications occur before CHD8 binding and then allow binding of CHD8 through contacts with lysine 4-methylated histone H3 and perhaps through direct contacts with the DNA through the CHD8 putative C-terminal DNA binding domain. DNA-bound CHD8 could then help recruit hStaf to its DNA binding site by protein-protein interaction.
The recent discovery that CHD8 associates with CTCF (19) together with our observation that CHD8 can be localized around the U6 promoter region open the possibility that CTCF is bound to insulator sites at or close to the U6 promoter. The U6-1 gene resides in a relatively barren region in the human genome, with very few expressed sequence tags encoded in the upstream 215 kb (http://genome.ucsc.edu/cgi-bin/hgGateway). In the downstream region, the first known gene is located 5 kb away and corresponds to the human homolog of mouse Corl1, a transcription corepressor expressed only in the brain and testis (32). The U6-1 gene may, therefore, require an active insulator to separate itself from its probably largely transcription-silent environment. Although no exact matches were identified, several sequences diverging from the human and mouse ß-globin and H19 CTCF binding sites (12, 20) at only a few positions can be found around the U6-1 gene.
Our analysis of hStaf-associated proteins has revealed several proteins known, or likely, to be involved in chromatin modification. For example, HCF-1 is a heterodimeric protein generated from its 2,035-amino-acid precursor by site-specific proteolysis, whose N-terminal subunit associates with the Sin3 histone deacetylase and the Set1/Ash2 histone methyltransferase complexes (75). It might bring the histone methyltransferases needed to give rise to the di- and trimethylated histone H3 present around the U6 promoter. Brd2 and Brd3 are bromodomain proteins known to interact with acetylated histones and, at least in the case of Brd2, to activate transcription (60). As well, hIno80, nuclear factor related to kappa B binding protein, and RUVB-like 1 protein are members of the hIno80 complex (22), whose yeast counterpart is involved in transcription activation as well as in other processes, such as DNA repair (33, 54, 70). Although we cannot exclude the possibility that these factors are involved only in Pol II transcription of mRNA genes activated by hStaf, they are all potential chromatin-modifying factors involved in Pol III transcription from the U6 promoter.
This work was funded by the Howard Hughes Medical Institute, by NIH grant GM38810, and by SNSF grant 3100A0-109941/1.
Published ahead of print on 15 October 2007. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»