(Received for publication, October 7, 1996, and in revised form, December 19, 1996)
From the Laboratory of Biochemistry and Molecular
Biology, The Rockefeller University, New York, New York 10021 and
the § Laboratory of Developmental Biology, Institute of
Molecular and Cellular Biosciences, University of Tokyo, Bunkyo-ku,
Tokyo 113, Japan
Human transcription initiation factor TFIID contains the TATA-binding protein (TBP) and several TBP-associated factors (TAFs). To investigate the structural organization and function of TFIID, we have cloned and expressed a cDNA encoding the third largest human TFIID subunit, hTAFII100. Immunoprecipitation studies demonstrate that hTAFII100 is an integral subunit that is associated with all transcriptionally-competent forms of TFIID. They further suggest that at least part of the N-terminal region lies on the surface of TFIID, while a C-terminal region containing conserved WD-40 repeats appears inaccessible. Both in vivo and in vitro assays indicate that hTAFII100 interacts strongly with the histone H4-related hTAFII80 and the histone H3-related hTAFII31, as well as a stable complex comprised of both hTAFII80 and hTAFII31. Apparently weaker interactions of hTAFII100 with TBP, hTAFII250, hTAFII28, and hTAFII20, but not hTAFII55, also have been observed. These results suggest a role for hTAFII100 in stabilizing interactions of TAFs, especially the histone-like TAFs, in TFIID. In addition, functional studies show that anti-hTAFII100 antibodies selectively inhibit basal transcription from a TATA-less initiator-containing promoter, relative to a TATA-containing promoter, suggesting a possible core promoter-specific function for hTAFII100.
TFIID is a general transcription initiation factor comprised of a small TATA-binding polypeptide (TBP)1 and a large number of TBP-associated factors (TAFs), all of which are highly conserved in evolution (reviewed in Ref. 1). One major function of TFIID is to recognize common core promoter elements and to nucleate assembly of other general factors (TFIIA, TFIIB, TFIIE, TFIIF, and TFIIH) and RNA polymerase II into a functional preinitiation complex (reviewed in Ref. 2). The primary interaction with core promoters containing TATA elements is through TBP, which is sufficient to mediate basal transcription in the absence of TAFs; in contrast, primary interactions with core promoters lacking TATA elements is thought to involve recognition of alternative core promoter elements (e.g. initiator elements) by TAFs and/or other factors that also are essential for basal transcription from these promoters (reviewed in Ref. 2).
In vitro studies in human and Drosophila systems have shown an essential role in activation both for TAFs and for other cofactors (reviewed in Refs. 1-4). Consistent with earlier studies showing that activators have qualitative and quantitative effects on TFIID binding that correlate with enhanced recruitment and function of other general factors (reviewed in Ref. 2), more recent studies have described specific activator-TAF interactions that are relevant to the overall function of the activators in vitro (reviewed in Refs. 1 and 3). More recently, and consistent with its role as a co-activator, TFIIA also has been shown to alter the topological arrangement of TAFs within a TFIID-promoter complex (5). These various studies have led to speculation that activator effects through TFIID may involve, in part, conformational changes that either stabilize TFIID binding on the promoter or facilitate interactions of other general initiation factors with the TFIID-promoter complex (reviewed in Ref. 2). Although recent studies in yeast have questioned a completely general role for TAFs in activator-mediated transcription in vivo (6, 7), the yeast TAFs are essential for viability and are likely to be necessary for activation of at least some genes and, potentially, for other events important for cell cycle progression. Possibly related to the latter point, a fraction of cellular TFIID has been shown to remain associated with chromosomes during mitosis in human cells, and mitosis-specific phosphorylation of TAFs has been shown to inactivate activator-specific functions of TFIID (8).
Of key importance for an understanding of the various TFIID functions is an appreciation of the overall structure of TFIID, including both the primary sequences and structures of individual subunits and their interactions and topological organization within TFIID. The cloning of cDNAs encoding most of the yeast, Drosophila, and human TAFs has provided relevant information, including schemes for a number of protein-protein interactions (9-13, reviewed in Ref. 1). Also of note is the documentation of three histone-like TAFs and the indication, from biochemical (10, 13-15) and structural (16) studies, of a histone-like octamer (in TFIID) comprised of the H4-like (hTAFII80/dTAFII62), H3-like (hTAFII31/dTAFII42), and H2B-like (hTAFII20/dTAFII28) TAFs.
Here we document the cloning of a cDNA encoding hTAFII100, the presence of hTAFII100 within all functional TFIID molecules, novel interactions of hTAFII100 with other TFIID subunits, and a possible core promoter-specific function.
TFIID was affinity-purified from a
phosphocellulose (Whatman P11) fraction (0.85 M KCl) of
HeLa nuclear extract by use of an anti-TBP antibody (17). TFIID
subunits were resolved by SDS-PAGE and transferred to a poly(vinylidene
difluoride) membrane. The 100-kDa polypeptide was excised and digested
with endoproteinase Lys-C (18). Sequence analysis of the purified
peptides yielded the following peptide sequences: EAEEALRREAXLLEEA,
PEIEVPLDDEXEEG, and VAVEDQPDVXAXL. Two degenerate
oligonucleotides,
5-GAIGCIGA(A/G)GA(A/G)GCICTICGICGIGA(A/G)GCIIIICTICTIGAIGAIGC-3
and
5
-CCIGA(A/G)ATIGA(A/G)GTICCICTIGATGATGA(A/G)IIIGAIGAIGG-3
were
synthesized and used to screen both HeLa and Namalwa cDNA libraries
according to standard procedures (19). Three overlapping cDNA
clones were isolated. The cDNAs were sequenced entirely on both
strands by use of Sequenase (U. S. Biochemical Corp.).
The plasmids which contain cDNAs encoding hemagglutinin (HA)-hTAFII250 and FLAG-hTAFII55 have been described (20). Expression plasmids for HA- or FLAG-hTAFII80, His-TBP, FLAG-hTAFII31, FLAG-hTAFII28, and HA-hTAFII20 were constructed by polymerase chain reaction, creating, in each case, an NdeI site at the N-terminal end and an appropriate restriction enzyme site at the C-terminal end following the natural stop codon. The large number of primers used in the PCR reactions has precluded description of their exact sequences, but the information is available upon request. The PCR-generated fragments were then inserted into pFLAG(AS)-7 or HA-pGEM7 at NdeI and appropriate sites (Ref. 13).2 The cDNAs encoding epitope-tagged constructs were then inserted into pVL1393/2 (Invitrogen) as described (20). For each TFIID subunit, an individual recombinant baculovirus was generated by co-transfecting corresponding cDNA and BaculoGold linearized baculovirus DNA (PharMingen) into Sf9 insect cells with cationic liposomes (Invitrogen). Each recombinant baculovirus was further amplified by repeated infection of Sf9 cells (3-4 times). For production of recombinant TAF proteins, Sf9 cells were infected by the amplified baculovirus, harvested after 48 h, and sonicated in 0.05 volume of Buffer C (20 mM Tris, 0.2 mM EDTA, 20% glycerol, 1 mM dithiothreitol, pH 7.9) containing 100 mM KCl and 0.5 mM phenylmethylsulfonyl fluoride. After removal of insoluble debris by centrifugation, supernatants were subjected to affinity purification on anti-FLAG antibody (M2-agarose, Kodak) or anti-HA antibody (12CA5 monoclonal antibody) columns. Epitope-tagged recombinant proteins were eluted with FLAG- or HA-peptides (20) and further purified by one or two steps of ion exchange chromatography. The recombinant proteins were more than 90% pure as judged by SDS-PAGE and silver staining.
Expression of hTAFII100 and Its DerivativesA
PCR-based method was used to create an NdeI site overlapping
the initiation codon. The PCR product was inserted into pFLAG(AS)-7 and
then subcloned into the baculovirus vector pVL1393 as described (20).
Full-length recombinant hTAFII100 was expressed in Sf9 cells as described above. For expression of histidine-tag and GST
fusion proteins, the cDNA sequences encoding N-terminal (1-455) and C-terminal (456-750) portions of hTAFII100 were
amplified by PCR (creating in each case an NdeI site at the
N terminus and a BamHI site following a stop codon at the C
terminus). PCR-generated fragments were then inserted into bacterial
expression vectors at NdeI and BamHI sites (for
the histidine-tag, 6HisT-pRSET was used; for the GST fusion,
pGEX-2TL(+) was used (13)). His-tagged hTAFII100(1-455) and His-tagged
hTAFII100(456-750) were produced in
Escherichia coli as described (13). The recombinant proteins were purified by chromatography on Ni-NTA-agarose (Qiagen, Chatsworth, CA). GST fusion proteins were generated in E. coli and
purified on glutathione-Sepharose columns (Pharmacia Biotech Inc.). A
histidine-tagged full-length hTAFII100
(His-hTAFII100) cDNA construct was similar to the
FLAG-tagged construct, except for the use of the 6HisT-pRSET vector in
place of pFLAG(AS)-7. His-hTAFII100 was generated in the
Promega TNT reticulocyte lysate system and labeled with
[35S]methionine according to the manufacturer's
instructions. After purification on a mini Ni-NTA-agarose column (20 µl), protein was stored at 80 °C until use.
His-tagged hTAFII100(1-455)
and His-tagged hTAFII100(456-750) were purified as above
and subjected to preparative SDS-PAGE. Gel slices containing the
corresponding His-tagged proteins were crushed and used for
immunization of two rabbits. The anti-hTAFII100 antibodies
were then affinity-purified on columns containing either GST-hTAFII100(1-455) or
GST-hTAFII100(456-800) proteins. For immunodepletion (Fig.
3) and immunoprecipitation (Fig. 4) experiments, 1-2 mg amounts of
antibodies were further purified by binding to 1 ml of protein
A-Sepharose (Pharmacia) and then cross-linked with dimethylpimelimidate
as described (21). For the immunodepletion studies, 5 ml of nuclear
extract was cycled through 2 ml of cross-linked anti-hTAFII100-protein A-Sepharose (2 mg/ml antibody) 5 times and the flow-through was used for Western blotting analysis and transcription assay. For analysis of direct inhibitory effects of
antibodies on transcription (Fig. 5), anti-hTAFII100
antibody was antigen-affinity-purified (as above), bound to and eluted from Protein A-Sepharose, and dialyzed against Buffer C containing 100 mM KCl. Different amounts of antibody, as indicated in Fig. 5, were then incubated with 4-µl aliquots of nuclear extract at room
temperature for 30 min. The extracts were employed directly, without
antibody removal, in transcription assays.
In Vitro Transcription
Assays with the G-less cassette
templates were carried out essentially as described (22). Apart from
the normal buffer, substrates, and salts, reactions (25 µl) contained
100 ng of pG5HMC2AT and pML53 supercoiled plasmid templates, 4-6
µl of HeLa nuclear extract and 30-60 ng of GAL4-VP16 or
GAL4-p65(NF
B) as indicated. GAL4-VP16 and GAL4-p65 consist of amino
acids 1-94 of the GAL4 DNA-binding domain fused, respectively, to a
C-terminal portion (amino acids 412-490) of VP16 and a C-terminal
portion (amino acids 416-550) of the p65 subunit of
NF
B.2 Reactions were incubated at 30 °C for 60 min
and the 32P-labeled RNA products were resolved by 6% PAGE
with 8 M urea and visualized by autoradiography. Primer
extension assays for Hsp70 and TdT transcription in nuclear extracts
were carried out as described (23). The human 70-kDa heat shock protein
(Hsp70) and murine terminal deoxynucleotidyltransferase (TdT) core
promoter templates were, respectively, plasmids pHsp70((
33/+99-CAT)
and pTdT(
41/+59) (24).
For each TFIID
subunit, an individual recombinant baculovirus was generated as
described above. Sf9 cells were co-infected by recombinant
baculoviruses expressing FLAG-hTAFII100 and a second TFIID
subunit and harvested after 48 h. The cell pellets were sonicated
in 0.05 volume of Buffer C containing 100 mM KCl and 0.5 mM phenylmethylsulfonyl fluoride. After removal of
insoluble debris by centrifugation, supernatants (cell lysates) were
stored at 80 °C until used. For co-immunoprecipitation, in a
250-µl reaction, 200 µl of cell lysate were mixed with 10 µl of
M2 beads (Kodak) or protein A-Sepharose cross-linked with
anti-hTAFII100 antibodies, and the salt concentration was
adjusted to 300 mM KCl and 0.2% Nonidet P-40. Binding
assays were carried out at 4 °C for 2 h. The beads were washed
5 times with 1 ml of incubation buffer and boiled in 40 µl of SDS
sample buffer. The eluates were analyzed by Western blot with specific
antibodies against TAFs. In control experiments, cell lysates from
singly infected cells lacking FLAG-hTAFII100 were used for
co-immunoprecipitation.
For in vitro immunoprecipitation experiments, 40 ng of FLAG-hTAFII100 was immobilized on 10 µl of M2-agarose or protein A-Sepharose cross-linked with anti-hTAFII100 antibodies (for hTAFII250, FLAG-hTAFII55, and FLAG-hTAFII28). Binding reactions contained 100 µg of bovine serum albumin, 100 ng of recombinant TAFs (about 90% purity from Sf9 cells or E. coli), 300 mM KCl, and 0.2% Nonidet P-40 in a 200-µl volume. After incubating at 4 °C for 1 h, the beads were washed five times with 1 ml of incubation buffer and the proteins were eluted and analyzed by Western blotting.
The human TFIID complex was purified from a
HeLa cell nuclear extract on a TBP antibody column (17). Two of the
hTAFII100-derived peptide sequences, EAEEALRREAXLLEEA and
PEIEVPLDDEXEEG, were used to design degenerate oligonucleotides to
screen human cDNA libraries. Three overlapping clones were
assembled and sequenced. Sequence analysis revealed an open reading
frame which encodes an 800-amino acid polypeptide (Fig.
1A) with a predicted mass of 89.0 kDa. This
open reading frame contains all three of the peptide sequences derived
from the natural hTAFII100 (Fig. 1A). When a
rapid amplification of cDNA ends method was employed to obtain a
longer 5 cDNA sequence, all transcripts stopped at an identical
position 14 nucleotides ahead of the first ATG (data not shown).
To verify that the polypeptide encoded by the cloned cDNA is an
intrinsic human TFIID subunit, antibodies were generated with bacterially expressed N-terminal (1-455) and C-terminal (456-750) portions of this protein. Western blot analysis with antigen
affinity-purified antibodies (Fig. 2A)
confirmed the presence of a 100-kDa immunoreactive band in both HeLa
nuclear extract (lane 1) and immunopurified TFIID
(lane 2). Furthermore, the cDNA was subcloned into
pBluescript vector (Stratagene) and subjected to in vitro
transcription and translation in the presence of
[35S]methionine. The size of the 35S-labeled
protein (Fig. 2B, lane 2) was indistinguishable from that of
the 100-kDa subunit of TFIID detected by Western blot (Fig. 2B,
lane 1). A control, antisense construct does not produce this
100-kDa protein (data not shown). When a cDNA encoding the corresponding FLAG-tagged protein (with an extra 18 amino acids) was
expressed in Sf9 cells, a correspondingly larger protein was detected
both by the purified anti-hTAFII100 antibody (Fig.
2A, lane 3) and by anti-FLAG antibody (data not shown).
Thus, although the longest cDNA does not contain a stop codon
preceding the open reading frame, these data indicate that it encodes a
full-length human TAFII100. This conclusion is further
supported by comparison with a rat TAFII100 cDNA, which
encodes a protein whose N-terminal region is nearly identical with that
of human TAFII100.3 The
hTAFII100 subunit in anti-TBP antibody-purified human TFIID (Ref. 17 and this paper) is equivalent to the hTAFII95
subunit in anti-FLAG antibody-purified human fTFIID (22), which
hereafter will be designated hTAFII100 in accordance with
more general convention (12, 17).
A GenBank search indicated that hTAFII100 shares highly
conserved amino acid sequences with Drosophila
TAFII80 (dTAFII80) and yeast
TAFII90 (yTAFII90), particularly in the
C-terminal regions of these proteins (25-27). Fig. 1B shows
an alignment of hTAFII100, dTAFII80, and
yTAFII90 according to the highly conserved WD-40 repeat
consensus sequences (28) that were previously reported in
dTAFII80 (25, 26) and yTAFII90 (27). Recent
data from the crystal structure of a G protein dimer (29, 30)
indicate that the WD-40 repeats form a compact
-propeller structure,
leading to speculation that the C terminus of hTAFII100 (as
well as that of dTAFII80 and yTAFII90) may form
a similar compact structure.
While this manuscript was in preparation, Dubrovskaya et al. (12) also reported the molecular cloning of human TAFII100. Compared to their sequence, the hTAFII100 cDNA sequence reported here contains several nucleotide changes. Our sequence encodes an alanine rather than a threonine at position 126, a phenylalanine rather than a leucine at position 300, a proline rather than a serine at position 455, an alanine rather than a valine at position 469, and an additional threonine at position 778. Currently we do not know the reason for these differences, although the phenylalanine at position 300 is conserved in yTAFII90 and the additional threonine at position 778 provides better sequence alignment of hTAFII100 with both dTAFII80 and yTAFII90 (containing serine and threonine, respectively, at this position).
hTAFII100 Is Associated with Transcriptionally Active Forms of TFIID Containing Other TAFsSince hTAFII100 has not yet been implicated in any TFIID function, and to verify whether it is in fact in an active TFIID complex, we next determined whether affinity-purified antibodies raised against the N- and C-terminal regions of hTAFII100 can inhibit transcription in vitro. Initially, using strong TATA-containing promoters, we failed to observe inhibition of transcription by direct addition of purified antibodies to reactions containing either nuclear extract or highly purified RNA polymerase II and general initiation factors (data not shown, see also below). We then used affinity-purified antibodies to deplete TFIID from nuclear extracts by cycling the extracts through an antibody column. A Western blot analysis of the treated extracts revealed that more than 90% of the various type II TAFs (including hTAFII100) were depleted by antibody specific for the N-terminal portion of hTAFII100, whereas none were depleted by a control antibody column (Fig. 3A, lane 3 versus lane 1 and data not shown). In contrast, less than 50% of the total TBP was removed (Fig. 3A), consistent with the presence of TBP in complexes utilized for transcription by RNA polymerases I and III (31). Moreover, immunoprecipitates obtained with the N-terminal specific antibody were shown to contain TBP and all type II TAFs (Fig. 3D, and data not shown).
When tested in a transcription assay with templates containing strong core promoters (Fig. 3B), the nuclear extract immunodepleted with the N-terminal-specific antibody was deficient in both basal (lower arrow) and activator (GAL4-VP16)-dependent (upper arrow) transcription (lanes 7-9) relative to the control extract (lanes 1-3). When purified TFIID was added back to the hTAFII100-depleted nuclear extract, both transcription activities (basal and activator-dependent) were restored (Fig. 3C, lanes 3-8 versus lanes 1-2). In contrast, antibody specific for the C-terminal portion of hTAFII100 failed to deplete TFIID subunits from nuclear extract (Fig. 3A, lane 2 versus lane 1) and had no effect on either the basal or the activator-dependent functions of TFIID (Fig. 3B, lanes 4-6 versus lanes 1-3). Since the antibodies to the N- and C-terminal portion of hTAFII100 show comparable reactions with hTAFII100 in immunoblots (Fig. 2), this could indicate that the C-terminal portion of hTAFII100 (containing the WD-40 repeats) is buried inside the TFIID complex. Taken together, these data indicate that hTAFII100 is an integral (tightly associated) component of essentially all transcriptionally active forms of TFIID, and that at least part of the N-terminal half of hTAFII100 is exposed on the surface of the TFIID complex and potentially capable of interacting with other factors.
hTAFII100 Interactions with Other Subunits of TFIIDWe next investigated possible interactions of hTAFII100 with TBP and other TAFs both in vivo and in vitro. To examine intracellular (in vivo) interactions, individual TFIID subunits were expressed either alone or together with a FLAG epitope-tagged hTAFII100 (FLAG-hTAFII100) in Sf9 cells. Derived cell lysates were immunoprecipitated either with anti-FLAG antibodies (for hTAFII80, hTAFII31, hTAFII20, and TBP) or with anti-hTAFII100 antibodies (for HA-hTAFII250, FLAG-hTAFII55, and FLAG-hTAFII28). As revealed by immunoblot analysis, comparable levels of other TFIID subunits were expressed in the absence (Fig. 4A, lane 1) and presence (lane 3) of co-expressed hTAFII100. When hTAFII100 was co-expressed, anti-hTAFII100 antibodies co-immunoprecipitated significant levels of TBP, hTAFII31, hTAFII80, and hTAFII250, lower but readily detectable levels of hTAFII28 and hTAFII20, and no detectable hTAFII55 (Fig. 4A, lane 4). In control assays, with extracts from cells lacking hTAFII100, no significant amounts of other TFIID subunits were co-precipitated (Fig. 4A, lane 2). In these analyses, recombinant virus inputs were adjusted so that hTAFII100 (when expressed) was limiting (expressed at a cellular level comparable to that of HeLa cells), while the co-expressed TFIID subunits were expressed at very high levels (over 10-50-fold greater). Hence, it was expected, even for strong interactions, that only a small fraction of the co-expressed interacting TFIID subunits would be co-immunoprecipitated with hTAFII100.
To test whether the indicated hTAFII100 interactions are direct, in vitro binding assays were performed with highly purified recombinant TBP and hTAFs. In these assays roughly equivalent amounts of recombinant proteins were incubated prior to immunoprecipitation. As shown in Fig. 4B, significant amounts of TBP, hTAFII80, hTAFII28, and hTAFII31, but no hTAFII250, TAFII55, or hTAFII20, were detected in the anti-hTAFII100 immunoprecipitates. These data indicate that hTAFII100 can interact directly with TBP, hTAFII80, hTAFII31, and hTAFII28 (see also below) and are consistent with the intracellular interaction data described above. The failure to see significant interactions of hTAFII100 with hTAFII250 and hTAFII20, as well as the apparently weak interaction between hTAFII100 and hTAFII31, contrasts with the results of the intracellular interaction assays. Although this could reflect indirect interactions of these proteins in the intracellular assays, it could also reflect limiting concentrations of properly folded TAFs or interference by the anti-hTAFII100 antibodies used in the in vitro interaction assays. We have therefore utilized two alternative assays to discern direct interactions.
The first alternative assay analyzed direct binding of purified in vitro translated hTAFII100 (35S-labeled) to purified hTAFII31, purified hTAFII80, and a purified in vivo assembled complex of hTAFII31 and hTAFII80. The purity of these latter components is shown in Fig. 4C. In the interaction assay (Fig. 4D) hTAFII100 showed strong hTAFII31 (lane 3) and hTAFII80 (lane 4) interactions, which paralleled those observed in the intracellular assays, as well as interactions with the stable preformed hTAFII31-hTAFII80 complex (lane 5). The second assay involved reconstitution of purified recombinant proteins and direct visualization by SDS-PAGE and silver staining, and revealed the formation of a stable TBP·hTAFII100·hTAFII250 complex whose composition is shown in Fig. 4E. This result is consistent with the previously documented in vivo interactions of hTAFII100 with TBP and hTAFII250.
Except for the case of hTAFII55, our present observations on hTAFII100 interactions are in agreement with those reported by Dubrovskaya et al. (12). However, we report novel direct interactions of hTAFII100 with the histone H4-like hTAFII80, the histone H3-like hTAFII31, and the naturally occurring complex of these two TAFs. Interactions with hTAFII80 and hTAFII31 were not examined by Dubrovskaya et al. (12) and appear in our analyses to represent the strongest hTAFII100 interactions.
Anti-hTAFII100 Antibodies Selectively Inhibit TATA-less Core Promoter FunctionAlthough not essential for basal
transcription from conventional TATA-containing promoters, TAFs are
absolutely required, along with other cofactors, for basal
transcription from certain initiator-containing promoters that lack
TATA elements (23, 24). Specific TAF requirements for these functions
have not been described. However, a position-specific
photocross-linking study has localized hTAFII100 near
positions 9/
8 in a TFIID-Ad2ML promoter complex (5), leading to
speculation that hTAFII100 might play a role in core
promoter recognition at or near initiator (Inr) elements. To explore
this possibility, as well as possible promoter-specific functions, we
compared the effects of affinity-purified anti-hTAFII100(1-455) antibody on basal and activated
transcription from strong TATA-containing promoters (pG5HMC2AT and
pML
53, also employed in the analysis of Fig. 3) and on basal
transcription from the TATA-less Inr-containing TdT promoter. Because
basal transcription from the TdT core promoter is much weaker (about 50-fold lower) than that from the TATA-containing AdML promoter, an
Inr-less TATA-containing promoter (human Hsp70) of comparable strength
was used as a control. As shown in Fig. 5A,
high concentrations of anti-hTAFII100(1-455) antibody
showed only marginal inhibitory effects on basal and
GAL4-VP16-activated transcription from the pML
53 and pG5HMC2AT
templates. In contrast, similar amounts of anti-hTAFII100(1-455) antibody dramatically decreased
transcription from the TATA-less Inr-containing TdT template, while
showing no significant effect on the Inr-less TATA-containing Hsp70
promoter (Fig. 5, lanes 1-4). As further evidence of
specificity, comparable amounts of a control anti-HA antibody had no
effect on transcription from either the TdT or the Hsp70 template (Fig.
5B, lanes 5-7). Altogether, these results raise the
interesting possibility that hTAFII100 may be involved in
core promoter-specific effects, possibly relating to initiator
function.
The TAF subunits of TFIID have been implicated as targets for both gene-specific activators (1, 3) and global repression mechanisms during mitosis (8), and as modulators (negative and positive) of TFIID binding to diverse core promoter elements (2, 3). Although not generally essential for basal transcription directed by TATA elements in core promoters, TAFs are essential for the function of other core promoter elements (initiator, downstream promoter element) either alone or in conjunction with TATA elements (reviewed in Ref. 2). Related to these functions, a number of studies have described interactions of TFIID subunits with each other (1, 9-16), with specific activators and general initiation factors (1-3), and with specific core promoter sequences (1-3, 32, 33), in addition to regulatory (post-translational) modifications of TFIID subunits (8). To investigate these questions, as part of our ongoing efforts to understand the assembly, structure, and function of human TFIID, we report here the cloning of a cognate cDNA and initial characterization of the 100-kDa subunit (hTAFII100) of human TFIID.
The cloned cDNA-encoded polypeptide described here appears to represent the natural hTAFII100 on the basis of amino acid sequences, size, and immunological cross-reactivity. That it is a bona fide TAF is suggested by sequence relationships to Drosophila (dTAFII80, Refs. 25 and 26) and yeast (yTAFII90, Ref. 27) TAFs and by functional tests. Of particular significance is the demonstration that hTAFII100 is a component of all transcriptionally active forms of TFIID in nuclear extracts. To date, however, no specific function of TFIID in basal or activator-enhanced transcription has been reported.
One of the striking features of hTAFII100 and its
Drosophila and yeast homologues is the presence of perfect
WD-40 repeats (6 in hTAFII100) in the C-terminal half of
the molecule. The WD-40 repeats were originally found in the
-subunit of the G protein (34), a heterotrimeric complex that
transduces signals from transmembrane receptors to other second
messenger-generating effectors, but is now evident in a various of
regulatory proteins (28). A role for the WD-40 repeats in facilitating
protein-protein interaction is evident from mutational studies of yeast
transcription factors (35), and from structural studies showing that
WD-40 repeats in the G protein
subunit form a compact
-propeller
structure and interact directly with the G protein
subunit (29,
30). Although it was shown that WD-40 repeats are dispensable for
dTAFII80 interactions with two Drosophila TFIID
subunits (TBP and dTAFII110, Ref. 24) and for
co-immunoprecipitation of hTAFII100 with TFIID (12), some
of our observations are consistent with the possibility of
hTAFII100 interactions with other TAFs through the WD40
repeats. First, unlike antibodies directed against the N-terminal half of hTAFII100, antibodies directed toward the C-terminal
portion containing the WD-40 repeats do not immunoprecipitate TFIID.
Second, the C-terminal half of hTAFII100 shows direct
interactions with other TFIID subunits.4
Hence, it is likely that at least part of the N-terminal half of
hTAFII100 is on the surface of TFIID while the C-terminal
WD-40 repeat regions may, at least in part, be buried within TFIID. Although Deubrovskava et al. (12) reported that a monoclonal antibody recognizing a C-terminal epitope could interact with TFIID and
modestly reduce basal transcription (and RAP30 interactions), this
epitope is outside the WD-40 repeat region.
While there have been no reports of hTAFII100 interactions with any activators, in contrast to what has been observed for many other TAFs (1, 3), in vivo and in vitro protein-protein interaction assays have suggested interactions that are potentially important for TFIID assembly and stability. In agreement with the recent report of Dubrovskaya et al. (12), we have detected physical interactions of hTAFII100 with TBP, hTAFII250, hTAFII28, and hTAFII20. However, we also have observed novel and considerably stronger interactions of hTAFII100 with hTAFII80 and hTAFII31, and with a complex of these proteins. Of significance in this regard are previous observations that hTAFII80 and hTAFII31 (and their Drosophila counterparts) have regions that are homologous to histones H4 and H3, respectively, and form a stable structure similar to the H3-H4 heterotetramer in the nucleosome (15, 16). In addition, hTAFII20 is related in sequence to histone H2B, shows interactions with hTAFII80 that parallel the histone H2B-H4 interactions, forms a dimer which mirrors the H2B-H2A interactions, and is present in a molar ratio of about four in TFIID (13, 15).
These results have led to the proposal that these histone-like TAFs form a nucleosomal-like "octamer" within TFIID during assembly or function (15, 16). Our present results, showing strong interactions of hTAFII100 with hTAFII80 and hTAFII31, and modest interactions with hTAFII20, raise the possibility that hTAFII100 may help stabilize interactions of the histone-like TAFs, perhaps within an octamer structure. This notion is particularly appealing, since it has proved difficult to isolate a stable complex of all three histone-like TAFs in the expected stoichiometry5 and since it is known that the core histone octamer structure is unstable at physiological salt concentrations in the absence of DNA (36). Moreover, recent studies in yeast have identified mutated forms of yTAFII90 (the yeast homologue of hTAFII100) as suppressors of a mutant form of yTAFII60 (the yeast homologue of hTAFII80),5 providing strong evidence for in vivo interactions between these TAFs. Also consistent with more stable interactions of hTAFII100 with hTAFII80 and hTAFII31 than with other TAFs, systematic TFIID disruption studies with anti-TBP- and anti-hTAFII20-immobilized TFIID showed concomitant dissociation of hTAFII100, hTAFII80, hTAFII31, and hTAFII250 (13). Similarly, TFIID assembly studies have indicated that hTAFII100 binds more strongly to an hTAFII80·hTAFII31 complex relative to an hTAFII250·TBP complex (Fig. 4 and data not shown). Other studies have shown that hTAFII100, but not the histone-like TAFs (with the possible exception of hTAFII31), are efficiently cross-linked to a region just upstream of the transcription initiation site in a TFIID-Ad2ML promoter complex (5). This leaves open the possibility that the histone-like TAFs are positioned for previously described interactions (some activator induced) of TFIID with downstream sequences (reviewed in Refs. 2 and 5). Alternatively, they may be buried within the TFIID complex and play an integral role in its topological organization and function.
Another novel feature of the current study is the finding that anti-hTAFII100 antibodies selectively inhibit the function of an Inr-containing TATA-less promoter (TdT), when compared to a TATA-containing promoter of comparable strength (Hsp70). These results are consistent with the unique TAF requirement for basal transcription by Inr-containing TATA-less promoters (23, 24) and with the above-mentioned localization of hTAFII100 near the initiation site when TFIID is bound to the TATA- and Inr-containing AdML promoter (5). These observations raise the interesting possibility that hTAFII100 may have core promoter-specific functions, being relatively more important for the function of Inr-containing TATA-less promoters. However, present uncertainty about the factors, whether TAFs or other essential cofactors (2, 24), that directly contact such promoters limits consideration of possible mechanisms for hTAFII100 function (for example, whether it recognizes DNA directly, acts as a bridging factor to an Inr-bound protein, or is critical for a certain TFIID conformation). Nonetheless, our studies indicate a fundamental difference in the topology and/or function of the corresponding TFIID-promoter complex that makes it uniquely sensitive to the presence of anti-hTAFII100 antibody. Further studies with reconstituted TFIID specifically lacking hTAFII100 or containing mutated forms of hTAFII100 should answer these questions.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U80191[GenBank].
We are grateful to C.-M. Chiang for anti-hTAFII55 antibody, A. Hoffmann for bacterial expression vectors (6HisT-pRSET and pGEX-2TL(+)) and anti-hTAFII20 antibody, S. Stevens for anti-hTAFII135 antibody, and X. L. Shi for recombinant GAL4-VP16 protein. We also thank other members of the Roeder laboratory, especially H. Xiao and Z. X. Wang, for helpful suggestions, and Y. Nakatani (National Institutes of Health) for communicating results prior to publication. Peptide sequencing was performed by the Protein Sequencing Facility of The Rockefeller University.
Tanese et al. (Tanese, N., Saluja, D., Vassallo, M. F., Chen, J.-L., and Admon, A. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 13611-13616) have recently reported the cloning of a cDNA encoding hTAFII100, and their deduced protein sequence matches exactly that reported here.