ABSTRACT
Contact between a transcriptional activator and one or more components of the
RNA polymerase II transcription initiation machinery is generally believed
important for activators to function. Several different molecular targets have
been suggested for direct contact by herpes simplex virus virion protein VP16,
including the general initiation factor TFIIB. In this report we have used
several strategies to critically assess this interaction between VP16 and
TFIIB. Affinity columns of VP16 bound TFIIB activity from HeLa cell extracts
and the binding was reduced by mutations in the activation domain of VP16. In
assays of direct binding, VP16 bound recombinant human TFIIB but not
Drosophila
or yeast TFIIB. Unlike binding from an extract, however, we found that the
interaction between VP16 and recombinant human TFIIB was not affected by
mutations in VP16 that reduce transactivation. Point mutations within human
TFIIB that reduce transactivation by VP16 have been shown to reduce VP16
binding, but we show here that these same mutations critically affect both the
important TBP-TFIIB interaction and the ability of TFIIB to support activator-independent basal transcription
in vitro
. Taken together our results suggest more evidence is needed to support the notion that TFIIB is a functionally important target for the activator VP16.
Transcriptional activator proteins regulate the expression of genes in
eukaryotic cells that are transcribed by RNA polymerase II. These activator
proteins often contain separable domains, one for site-specific binding of DNA and others for transcriptional activation (
1
,
2
). Activation domains function even when attached to the DNA binding domain of a
heterologous protein. They are believed to make contact with the RNA polymerase
II transcription machinery and effect the assembly and/or the activity of the
transcription initiation complex (
3
,
4
).
Within the RNA polymerase II transcription initiation complex, a plethora of
putative targets has been suggested for direct contact by the activation
domains of
cis
-binding positive acting transcription factors. The earliest studies had
suggested that the TFIID fraction was a target of several activators (
5
,
6
) and both the TATA binding protein (TBP) (
7
,
8
) and certain of its associated TAF polypeptides (
9
,
10
) have since been shown to be capable of direct binding to an increasingly large
number of proteins with diverse activation domains (
11
-
22
).
In vitro
interactions of activators with other polymerase II general initiation factors
have also been reported. A number of activators bind the initiation factor
TFIIB (
23
-
27
) and the multi-component factor TFIIH has also been implicated as a target for several
activators (
28
). Furthermore, the potential targets for activators are not limited to just the
minimal components required for basal level transcription. In yeast cells the
ADA2 protein, identified genetically as a co-activator (
29
), has been shown to bind to the acidic activation domain of VP16 (
30
). In mammalian cells another factor, PC4, a component of the USA fraction
required for efficient activated transcription
in vitro
, also appears capable of direct activator contact (
31
). While this multiplicity of potential targets for activators may be indicative
of a complex and dynamic exchange of interactions resulting in more
transcription initiation by RNA polymerase II (
4
), evidence supporting a role within cells for some of these interactions
detected
in vitro
is either entirely incomplete or is lacking. Furthermore, the concept of an
ordered multi-step pathway for the initiation of transcription by RNA polymerase II with
the potential for having several rate limiting steps in the formation of the
initiation complex accelerated by activators (
32
,
33
) has been challenged by evidence that a large multi-component RNA polymerase II holoenzyme complex may pre-exist within cells (
34
-
37
).
Work with the acidic activator VP16 from our laboratories provided the first
evidence of direct activator-TBP interactions (
7
,
8
). Our initial reports, however, seemed to be at variance with very similar
experiments suggesting that TFIIB rather than TBP was an important target for
VP16 (
23
,
38
). Because apparently different results were obtained with ostensibly similar
experiments, we have now re-examined the interactions between VP16 and the TFIIB polypeptide from
human,
Drosophila
and yeast cells. Our results suggest that if TFIIB is indeed an important
target of this particular activator, it is so only in mammalian cells. We also
show that mutations within the activation domain of VP16 or within the putative
target human TFIIB that result in reduced levels of activated transcription
either do not affect the VP16-TFIIB interaction or they alter additional important protein-protein interactions within the transcription initiation complex.
Taken together, our results suggest that more compelling evidence is still
required to support the notion of TFIIB as a target for the activator VP16.
The bacterially expressed RNA polymerase II initiation factors yeast TBP (
39
), human TBP (
28
) and
Drosophila
TBP (
40
), human TFIIB (
41
), yeast TFIIB (
42
) and
Drosophila
TFIIB (
43
), human TFIIF (
44
) and the GAL4-VP16 derivatives (
45
) were prepared as previously described, as were protein A (pA)-VP16 (
7
) and GST-TFIIB fusion proteins (
46
). The human TFIIB mutants R185E/R193E and K198E/K200E (
46
) (kindly provided by D. Reinberg), along with wild-type human TFIIB, were expressed in BL21 (DE3) cells with a 10 histidine N-terminal tag after subcloning TFIIB cDNAs into pET19b (Novagen) and
purified on Ni
2+
-NTA columns (
15
). Human TFIIB proteins were further purified on a 0.5 ml heparin column (
16
). All purified proteins were dialyzed against affinity chromatography buffer (
7
) containing 0.1 M NaCl and stored at -70oC. Highly purified calf thymus RNA polymerase II (
47
) was kindly provided by M. Sopta. HeLa nuclear extract fractions (
48
) were kindly provided by D. Fitzpatrick and heat-inactivated nuclear extract was prepared as previously reported (
49
).
Affinity chromatography columns of pA and pA-VP16 derivatives were prepared and used as previously described (
7
).
In vitro
transcription assays using as template DNA a G-less reporter cassette driven by the adenovirus 2 major late (Ad2ML)
promoter were performed essentially as described previously (
7
,
28
).
TFIIB-TBP complex formation between yeast TBP and human TFIIB proteins was
analyzed by an electrophoretic mobility shift assay (
50
). A
32
P-labeled probe containing the TATA element of the Ad2ML promoter from -53 to +33 (a gift from B. Coulombe) was used. DNA binding reactions
were performed for 30 min at 30oC and resolved on a 4% polyacrylamide gel in Tris-glycine buffer (
50
) lacking EDTA.
35
S-Labeled GAL-VP16 proteins were synthesized by a coupled
transcription-translation procedure or by transcribing RNA from a T7 promoter
and then using this RNA for
in vitro
translation reactions in a rabbit reticulocyte lysate system (Promega).
Human TFIIB fused to GST was bound to glutathione-Sepharose beads at 0.05, 0.5 and 2 mg/ml in buffer A [20 mM Tris-HCl, 0.2 mM EDTA, 1 mM dithiothreitol (DTT), 1 mM
phenylmethylsulfonyl fluoride, 0.5 mM benzamidine hydrochloride and 28 [mu]M tosyl-phenylalanine chloromethyl ketone] containing 1 M NaCl. GST-TFIIB-coupled beads were first equilibrated three times in 10
vol buffer B (40 mM HEPES, pH 7.9, 5 mM MgCl
2
, 1 mM DTT, 0.5% NP-40 and 100 or 150 mM KCl) for 1 h. Aliquots of 20 [mu]l of these beads were incubated with the
35
S-labeled GAL4-VP16 proteins in 80 [mu]l buffer B containing 100 or 150 mM KCl for 1 h and finally
washed three times with 200 [mu]l buffer B. The beads were subsequently boiled in gel sample buffer and the
eluates were resolved by SDS-PAGE and monitored by a phosphoimager (Bio-Rad) or by autoradiography.
Initiation of transcription was performed on a synthetic template containing
Ad2ML promoter sequence with a single-stranded bubble from nucleotides -9 to +3 (
51
). The template was incubated with wild-type or mutant human TFIIB, yeast TBP, human TFIIF, calf thymus RNA
polymerase II, [
32
P]CTP and the primer dinucleotide CpA for 25 min at 23oC. RNA trimers were analyzed on an 18% polyacrylamide denaturing gel
containing 7 M urea as previously described (
51
).
The first experiments identifying targets amongst the RNA polymerase II
initiation factors contacted by the potent acidic activator VP16 implicated
both TFIID and TFIIB as potential targets. In one set of experiments, TFIID
activity was depleted by passage of HeLa cell nuclear extracts competent for
in vitro
transcription over columns of immobilized VP16 (
7
); in another TFIIB activity was depleted (
23
). The ability of these affinity columns of VP16 to quantitatively remove any
particular initiation factor may depend upon salt and ligand concentrations,
upon the bed volume of the columns and also upon the relative levels of
particular initiation factor activities, which undoubtedly vary amongst
different extract preparations. Rather than assess whether initiator factors
are depleted by chromatography of an extract over VP16, we have now monitored
the ability of wild-type and mutant VP16 columns to retain a portion of either TFIIB or TFIID
activity. TFIIB activity in column eluates was assessed in a reconstituted
in vitro
transcription system using partially fractionated HeLa cell components. TFIID
activity was assessed using heat-treated (TFIID-deficient) HeLa nuclear extracts (
49
). As shown in Figure
1
(lanes 1-3), these assay systems show a near complete dependence upon the addition
of either TFIIB or TFIID (TBP). A comparison of lanes 4 and 7 indicated that
the eluate from the wild-type pA-VP16 column, but not the control pA column, contained both TFIIB
and TFIID activities. Consistent with results reported earlier by Lin and Green
for TFIIB (
23
) and Ingles
et al
. (
8
) for the TBP component of yeast TFIID, the ability of VP16 to bind TFIIB or TBP
was reduced by mutations in the activation domain of VP16. Truncation of the
activation domain to amino acid 456 reduces transactivation by VP16 to ~50% when assessed in murine cells with the herpes virus ICP4 promoter (
52
) and to an undetectable level when measuring GAL4-VP16 activation with a reporter having only a single GAL4 binding site (
53
). Within the context of a full-length activation domain, substitution of Phe with Pro at position 442
moderately reduces transactivation by VP16 (
54
). When combined with the truncation to position 456, this FP442 mutation
completely inactivates transactivation by VP16 (
52
,
53
). As lanes 5, 6 and 8 of Figure
1
indicate, the ability of these mutant forms of pA-VP16 to retain either TFIIB or TFIID activity was very similar.
Truncation to position 456 markedly affected binding of both factors, as did
the Phe -> Pro mutation at 442 in full-length VP16. In the context of truncated VP16, the FP442 mutation
completely prevented retention of either factor. This correlation between
transactivation activity
in vivo
and initiation factor binding
in vitro
argues that these interactions could be important during the activation of
transcription. These experiments do not, however, address whether these
activator-initiation factor interactions are direct or are mediated by one or more
of the components present in the HeLa cell extracts loaded on these columns.
To examine whether the retention of human TFIIB by immobilized VP16 was the
result of a direct interaction between VP16 and TFIIB, human TFIIB was produced
in and purified from
Escherichia coli
cells (
41
). For these affinity chromatography experiments we used truncated derivatives
of the activation domain of VP16; the effect of mutations at position 442 on
transcription being more marked in this context. When the concentration of
ligand VP16 on the columns was >2.0 mg/ml a significant portion of the
recombinant human TFIIB was retained by the columns. The 0.1 M NaCl wash
fractions eluted some of the TFIIB, but the majority of applied TFIIB was only
eluted by the 0.6 M NaCl step (E in Fig.
2
A). While the mutations in VP16 that reduced transactivation did markedly affect
the recovery of TFIIB when unfractionated HeLa cell extracts were applied to
the columns (Fig.
1
), these mutations in VP16 had no effect on the binding of purified recombinant
human TFIIB. The data shown in Figure
2
A indicates that the VP16 derivative [Delta]456-FP442, which is completely defective in transactivation and failed
to interact with TBP (
8
), bound human TFIIB just as well as did the matrix made with the corresponding
wild-type but truncated derivative, a result similar to that reported by
Goodrich
et al
. (
10
). The effect of other missense mutations at the 442 position is similar.
Changes from Phe to Ala or Ser, which markedly compromised TBP binding (
8
) and the transactivation potential of VP16 (
52
), had no effect on the binding of human TFIIB (data not shown, but see Fig.
3
). In the context of a full-length activation domain, the mutations at position 442 were also without
effect on the binding of recombinant human TFIIB (data not shown).
As the mutations in VP16 at position 442 had no effect on the ability of
immobilized pA-VP16 to directly bind human TFIIB, we decided to examine these
interactions another way. Since a direct interaction between human TFIIB and
the activation domain of VP16 has also been detected when the TFIIB protein
rather than VP16 is immobilized (
46
), we purified a similar GST-human TFIIB chimeric polypeptide and bound it to glutathione-Sepharose beads at several ligand concentrations. The GST-TFIIB-containing beads were then incubated with
35
S-labeled GAL4-VP16 derivatives, washed and finally TFIIB-bound GAL4-VP16 was eluted by boiling in gel sample buffer and analyzed by SDS-PAGE. A ligand concentration of 0.5 mg/ml GST-TFIIB was required to obtain significant
GAL4-VP16 binding. With this ligand concentration ~30% of the input GAL4-VP16 [Delta]456 was bound by GST-TFIIB. We then tested the effects of various
mutations in VP16 at amino acid 442. As shown in Figure
3
, the Phe to Pro or Ser changes, which completely inactivate the transactivation
potential of VP16 (
52
), had no effect whatsoever on the binding of GAL4-VP16 to immobilized human TFIIB. The conservative FY442 change, which
permits a moderate level of transactivation (
52
), also had no effect on binding of GAL4-VP16. Increasing the salt concentration to 0.15 M reduced the binding of
these GAL4-VP16 derivatives to GST-TFIIB, but still no effect of the mutations in VP16 was evident
(data not shown). The lack of effects of these mutations in VP16 on binding to
immobilized human TFIIB is consistent with the data presented in Figure
2
A using immobilized VP16.
A VP16-interacting region of human TFIIB has been shown by analysis of deletion
clones of human TFIIB to map between amino acids 178 and 201 (
46
). This same region of TFIIB has, however, also been shown in several studies to
be important for interaction with TBP (
55
-
57
). Missense mutations in this region of TFIIB have been shown to affect binding
to VP16 (
46
). We have now carefully analyzed the effects of two different double point
mutations in human TFIIB used in these earlier studies. As reported by Roberts
et al
. (
46
), we too found that recombinant human TFIIB proteins containing either the
R185E/R193E or K189E/K200E mutation were defective in binding to VP16 (data not
shown). However, since other charge-change mutations in this region of the
Drosophila
TFIIB protein compromised the interaction between TBP and TFIIB (
56
) and crystallographic studies of a TFIIB-TBP-DNA complex indicate that this region of TFIIB contacts TBP (
57
), we also assessed the ability of these particular mutant forms of human TFIIB
to interact with TBP. As assessed by formation of a TFIIB-TBP complex on Ad2ML promoter DNA (Fig.
4
), addition of as much as 100 ng of each of these mutant forms of TFIIB failed
to make a stable complex with TBP. In contrast, formation of a readily detected
TFIIB-TBP complex on DNA was possible with as little as 25 ng of wild-type TFIIB.
The experiments described in this report were undertaken because we felt it was
important to more carefully assess the different conclusions reached with the
rather similar experimental protocols used by Stringer
et al
. (
7
) and Lin and Green (
23
) in the identification of targets of the acidic activation domain of the herpes
simplex transactivator VP16. We now have shown, in agreement with the initial
report of Lin and Green (
23
), that TFIID and TFIIB activities can both be retained by affinity columns of
immobilized VP16. The retention of both these initiation factors was markedly
reduced by several mutations in the activation domain of VP16 that reduce its
transactivation potential. Whether TFIID or TFIIB is quantitatively depleted by
passage over VP16 may simply be a reflection of the particular chromatographic
conditions (e.g. ligand and salt concentrations and column volumes) and the
relative concentration of each factor in different extract preparations. The
binding of these initiation factors to immobilized VP16 need not be direct,
however. Documented factor-factor interactions (
58
) and the recent reports of the existence of polymerase II holoenzyme complexes
containing many of the polymerase II initiation factors (
34
-
37
) raise the possibility that certain activator-initation factor interactions could indeed be indirect.
To examine this issue, different polymerase II initiation factors have been
either highly purified or expressed as recombinant proteins. Our experiments
with recombinant human TFIIB largely confirm the observations of Lin
et al
. (
38
). Immobilized VP16 bound TFIIB and, as reported by Roberts
et al
. (
46
), immobilized TFIIB can bind the chimeric activator GAL4-VP16, albeit only at ligand concentrations considerably higher than the
0.05 mg/ml reportedly used by these authors. In one crucial aspect, however,
our results differ in a substantive way. While Lin
et al
., using glutathione-Sepharose-bound GST-VP16 matrices, did see reduced binding of recombinant human
TFIIB to a single VP16 mutant (
38
), we found, under a variety of salt and ligand concentrations with both pA
derivatives (Fig.
2
A) and GST derivatives (data not shown), that not only this mutation but others
in the activation domain of VP16 at the critical Phe442 residue did not alter
the strength of the direct interaction of VP16 with recombinant TFIIB. Our
results are similar to those reported by Goodrich
et al
. (
10
) for the FP442 mutation of VP16 when they examined the interaction between
recombinant human TFIIB and glutathione-Sepharose-bound GST-VP16 derivatives. In this respect we also note that
mutations at several other positions within the C-terminal region of the activation domain of VP16 which contribute to
inactivation of the transactivation potential of VP16 were reported by Walker
et al
. (
53
) to be without effect on the binding of human TFIIB present in a nuclear
extract to covalently coupled GST-VP16-Sepharose matrices. The lack of effect of these VP16 mutations on
the direct interaction with TFIIB contrasts with the marked effects of these
same mutations on the binding of VP16 to recombinant TBP (
8
) and TFIIH (
28
). A correlation between the effects of mutations within a second acidic
activation domain, that of the yeast activator GAL4 and binding to yeast TBP
has also been reported (
59
). GAL4, like VP16 (see Fig.
2
A), was reported not to bind yeast TFIIB (
59
).
Correlations between transactivation potential of wild-type and mutant transactivation domains and the strength of interaction
between an activator and its targets can help establish the biological
relevance of activator-initiation factor contact. In addition, mutations within the putative
target(s) within the polymerase II initiation complex may, if their effects are
limited to just the activator-factor interaction, be used to establish the importance of contacts of
activators with their putative targets. Such mutations within both TBP (
60
) and TFIIB (
46
) have been described. These mutations reportedly diminish interaction with the
activator VP16 and compromise VP16-activated, but not activator-independent, basal transcription
in vitro
. Close examination of the published data detailing the selective effect of
several missense mutations in human TFIIB on activated transcription reveals,
however, that the mutant forms of human TFIIB may not have been equivalent to
wild-type TFIIB in supporting basal transcription (
46
). In particular, it seems that with 5 ng of wild-type TFIIB transcription was maximized and that with comparable amounts of
the mutant K189E/K200E form of TFIIB equivalent levels of transcription were
not attained. The data shown in Figure
5
in this report explores this finding in greater detail. While additions of as
little as 25 ng of wild-type TFIIB supported a maximal rate of transcript initiation in our
reconstituted system, additions in the range 100-300 ng of the same two mutant forms of TFIIB used by Roberts
et al
. (
46
) were needed to approach wild-type levels of transcript initiation in our system. As studies from
several laboratories had already indicated, the region of TFIIB believed to
interact with VP16 is also important for interaction of TFIIB with TBP (
55
-
57
). We have now shown that these same double point mutations in human TFIIB are
defective in binding to a TBP-DNA complex, a result that could account for the effects of these
mutations on basal transcription and which may also explain the crippled
response to the activator VP16.
How then is transcription activated and just which are the important
interactions between transcriptional activators and the polymerase II
machinery? Several lines of evidence (reviewed in
61
) suggest that an important aspect of activator function is to promote assembly
of the polymerase II transcription initiation complex at a promoter. In
particular, some experiments have shown that the VP16 activation domain can
facilitate the recruitment of TFIIB into the assembling pre-initiation complex (
23
,
32
,
60
). It is unclear whether this recruitment of TFIIB is a consequence of the
interaction of VP16 with TFIIB or with TBP (
32
,
60
). Although a multi-step assembly pathway, with the opportunity of having one or more rate
limiting steps in assembly be accelerated by activators, was for many years an
attractive model (
32
,
33
), it now appears that much if not all of the polymerase II initiation machinery
could exist as a pre-assembled holoenzyme (
34
-
37
). Contact with one or more components of such a polymerase II holoenzyme may
help bring polymerase II to the promoter in a single step. Recent experiments
with yeast cells bearing a mutation within one component of the holoenzyme,
GAL11, argue persuasively for this view (
62
). Other studies have suggested that, for certain promoters, activators can
function by promoting recruitment of TBP to the promoter (
63
-
65
). In this respect it is interesting to note that TBP was initially reported to
co-purify with the polymerase II holoenzyme (
34
). Although subsequent preparations of the yeast polymerase II holoenzyme did
not contain TBP, it is not yet clear whether the absence of either TBP, or
certain other of the polymerase II initiation factors, reflects the
in vivo
situation or is a consequence of the purification procedure. Since both TFIIB
and TFIIH are present within at least one preparation of yeast holoenzyme (
35
), the contacts between the activator studied in this report, VP16, and either
of these initiation factors could lead to holoenzyme recruitment at promoters.
Our data, however, suggests that if VP16 does indeed contact the factor TFIIB
within cells, then this contact differs from the contact between VP16 and
either TBP (
8
) or TFIIH (
28
) in being insensitive to mutations in VP16 that compromise transactivation.
Since mutations within TFIIB that reduce the transactivation response to VP16
also critically affect other important functions of this initiation factor, it
may be premature at this time to conclude that direct contact of activators
with TFIIB is an important aspect of the transactivation process. It should be
noted, however, that the
in vivo
relevance of direct activator-TBP contact has also been questioned recently. The ability of TBP to
interact with activation domains
in vitro
was not required for TBP to support activated transcription
in vivo
(
66
). If contact between an activator and virtually any subunit of holoenzyme can
lead to recruitment of RNA polymerase II to the promoter and transcriptional
activation (
62
), there may indeed be multiple contacts between a strong activator and the
transcription apparatus and no one particular contact may be essential for
activation. It seems that if the molecular details of the mechanism of
transactivation are going to be securely established, then perhaps the
application of new and quite different experimental approaches may be required.
We thank D. Reinberg, S. Wampler and M. Hampsey for TFIIB DNAs and Lina
Demirjian for technical assistance. This work was supported by grants to CJI
and JG from the Medical Research Council of Canada and from the National Cancer
Institute of Canada with funds from the Canadian Cancer Society. AE and GP are
recipients of a Medical Research Council of Canada Studentship and Post-doctoral Fellowship respectively. JG is an International Research Scholar
of the Howard Hughes Medical Institute.
REFERENCES
Return