Proc. Nadl. Acad. Sci. USA
Vol. 91 pp. 6279-6282, July 1994
Interaction of the Drl inhibitory factor with the TATA binding
protein is disrupted by adenovirus ElA
VIRGINIA BYERS KRAUS*t, JUAN A. INOSTROZA*, KAM YEUNG*, DANNY REINBERG, AND JOSEPH R. NEVINS*
*Howard Hughes Medical Institute, Department of Genetics, and tDepartment of Medicine, Duke University Medical Center, P.O. Box 3054, Durham, NC
27710; and *Department of Biochemistry, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, Piscataway, NJ 08854
Communicated by W. K. Joklik, March 22, 1994
ovirus ElAus product activates the hsp70 promoter, depen-
dent on the TATA element and dependent on N-terminal ElA
sequences. Other experiments have Identified a factor termed
Drl that interacts with and inhibits the transcriptional activity
of the TATA-binding protein (TBP). We now find that the
ElA12s protein can disrupt the interaction of the Drl factor
with the TATA-specific TBP factor, allowing the productive
interaction ofTBP with TFHA. This ElA-mediated disruption
is dependent onN-te
l sequences thatare also essential for
the TATA-dependent trans-activation of the hsp7O promoter.
Moreover, we also find that Drl expression in transfected cells
can inhibit t iption from thehsp70promoterand that this
can be overcome bycoexpression ofthe wild-typeElA protein,
dependent on N-teil sequences. We conclude that the
activation of hsp70 trgh the TATA element may be mech-
anisicall simllar to the activation oftheE2prmoter via E2F,
in each case involg a release of a transcription factor from
an inactive complex.
Past experiments have shown that the aden-
The adenovirus ElA gene encodes multiple protein products
that activate transcription ofviral and cellular genes (1). The
CR3 domain, which is unique to the so-called E1A13S prod-
uct, is clearly responsible for the majority of transcriptional
activation of viral genes during a lytic infection. Neverthe-
less, it is also clear that the E1Aj2s product possesses
transcriptional activating function. For example, recent ex-
periments have shown that theE1Am8 protein can activate
transcription by releasing the E2F transcription factor from
inhibitory complexes containing proteins such as the retino-
blastoma gene product (2). The ability ofthe ElAr2s protein
to activate the E2F factor is dependent on protein sequences
within the CR1 and CR2 domains but independent of the
N-terminal sequences. The E1Aj2s product also activates
transcription in an E2F-independent manner, an example
being activation via the hsp70 promoter (3). In this case, the
TATA element is the critical target for activation. Moreover,
in contrast to the requirement for the CR1 and CR2 domains
for the activation ofE2F (4), it is the N terminus ofthe ElA
protein that is necessary for the TATA-dependent activation
of transcription (3).
Recent experiments have identified an activity termed Drl
that inhibits the activity ofthe TATA-specific bindingprotein
(TBP) transcription factor (5). This inhibition involves a
direct physical interaction ofDrl with TBP and, as a conse-
quence ofthis interaction, TBP is unable to form interactions
with othercomponents ofthe transcriptional machinery such
asTFIIAand/orTFIIB. Given the fact thatthe ability ofElA
to activate transcription through the E2F factor involves the
release ofE2F from inhibitory complexes with proteins such
as Rb, we have explored the possibility that ElA-mediated
activation of transcription through the TATA element in-
volves a similar event with respect to the TBP-Drl interac-
MATERIALS AND METHODS
DNA-Binding Assays. DNA-binding assays were per-
formed as described using a probe derived from the adeno-
virus major late promoter (6). ElA protein was synthesized
in a rabbit reticulocyte lysate as described (7).
Transfection Assays. Hsp7O-luc was generated by inserting
a 1270-bp HindIII/Sal I DNA fragment from hsp7O-CAT
between the Sac I/Xho I sites of a luciferase reporter,
GL2-basic (Promega). ElA expression plasmids CMV-
E1Amsand CMV-ElAdl2 36have been described (3). The Drl
expression vector CMV-Drl was generated by inserting the
Drl cDNA between the Kpn I/Xba I sites of a cytomegalo-
virus expression vector, CMV4 (8). HeLa cells were trans-
fected via lipofectamine according to the manufacturer's
specifications (BRL). Cells were plated on 35-mm dishes and
received a total of 2.5 ug ofDNA and 2 pg oflipofectamine
per plate. Cells were harvested 48 hr after transfection.
Luciferase assays were performed with a kit purchased from
Plasmids. Plasmids encoding glutathione S-transferase
(GST) E1A12s fusion proteins constructs were made by PCR
using previously described E1A12S wild-type (WT) and mu-
tant cDNAs as template (4). ElA cDNAs were cloned in
frame with GST between the BamHI and EcoRI sites in
pGEX-3X. TheATG startcodonofElA occurs 3 amino acids
downstream from the factor Xa cleavage site in the fusion
proteins. The fusion protein constructs use the ElA stop
codon followed immediately by an EcoRI site. The deletion
mutants are named for the deleted residues in E1A12s (e.g.,
d12-36 lacks amino acids 2-36 of the WTElAi2s). Deletion
mutant d138-67 contains a 12-nt insertion resulting in the
substitution of4 amino acids (SSRD) for the deleted region.
The double point mutant pm928/961 results in the double
amino acid substitution, glycine for cysteine at residue 124
and lysine for glutamic acid at residue 135.
Preparation of GST Fusion Proteins. GST fusion proteins
were purified from bacterial lysates as described (9).
Protein Interaction Assays. A total of 1pi(equal to -4 pg
of eluted protein) of glutathione-Sepharose 4B bound-GST-
E1A12S WT or mutant was mixed with 24Alof a 50%6 slurry
ofglutathione-Sepharose 4B beads (Pharmacia)prewashed in
BC 100 buffer (20 mM Tris, pH 7.9/20%o glycerol/i mM
EDTA/0.2 mM phenylmethylsulfonyl fluoride/0.1 KCl).
This was added to 2.5 pg ofrecombinant Drl, along with 20
pg of bovine serum albumin and 0.15% Nonidet P-40 (final
concentration). The reaction mixtures were gently rocked at
250C for 1.5 hr. Beads were spun 30 sec at 10,000 rpm and
washed twice in 0.5 ml of BC-100 buffer, resuspended in 20
,ulof 1.3x SDS/PAGE gel loading buffer, boiled 10 min, and
Abbreviations: TBP, TATA-binding protein; GST, glutathione
S-transferase; WT, wild-type.
The publication costs ofthis article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Proc. Natl. Acad. Sci. USA 91 (1994)
run on a 15% SDS/PAGE gel. Proteins were transferred to
nitrocellulose and probed with anti-Drl (a rabbit polyclonal
antibody to recombinant Drl) (5). Detection of bands via
avidin/biotin conjugates was accomplished using the Vecta-
stain ABC kit (Vector Laboratories, catalog no. PK4001)
and the peroxidase TMB (3,3',5,5'-tetramethylbenzidine)
substrate kit (Vector Laboratories catalog no. SK4400) as
per the manufacturer's instructions.
ElA-Mediated Disruption of the TBP-Drl Interaction. Drl
was identified as a 19-kDa phosphoprotein that inhibits
TATA-dependent transcription through an interaction with
the TBP (5). Drl can remove TFIIA from TBP-TFIIA
complexes, suggesting competition between Drl and TFIIA
for the same recognition site on TBP. As a consequence of
this interaction, the assembly of a transcriptionally compe-
tent complex ofgeneral transcription factors is prevented and
transcription is inhibited. Although the interaction of TBP
with the TATA element by gel retardation assay is weak and
somewhat difficult to detect, it is possible to readily assay the
formation of a TBP-TFIIA complex that binds to the TATA
element (Fig. 1A). Addition of the adenovirus E1A125 prod-
uct, produced in a reticulocyte lysate, had no effect on the
interaction of TBP and TFIIA with the TATA element. As
shown previously (5), the addition ofDrl to the TBP-TFIIA
mixture results in a reduction or loss of the TBP-TFIIA
complex and the production of several additional complexes
that contain the Drl protein in association with TBP (Fig.
1B). The various TBP-Drl complexes appear to result from
the multimerization of Drl (5). Addition of the EMA protein
abolished the complexes containing the Drl protein, restor-
ing the TBP-TFIIA complex as the major component. Thus,
itwould appearthatthe EMAproteincan disrupt theTBP-Dr1
interaction but not the TBP-TFIIA complex.
A direct assay for the interaction ofTBP and Drl reveals
theformation ofatleasttwo distinctDNA-proteincomplexes
(Fig. 1C). Addition ofa Drl-specific antibody to the binding
assay demonstrates the presence ofDrl in these complexes.
Once again, addition of the EMA protein, in the form of a
GST-E1A fusion protein, disrupts the TBP-Drl complexes
(Fig. 1D). The slowest migrating TBP-Drl complex was
readily disrupted, whereas the faster migrating complex
required the addition of larger amounts of ElA. In contrast,
addition ofa control GST protein only slightly diminished the
slower migrating complex, whereas the same quantity ofEMA
resulted in the total loss of both complexes. It would thus
appearthatEMA does indeed possess the ability to disrupt the
interaction ofTBP with the inhibitory Drl activity.
The N Terminus of EMA Is Essential for Disruption of
TBP-Drl and for Binding to Drl. Previous experiments have
shown that sequences at the N terminus of the EMA protein
are critical for TATA-dependent activation of transcription,
whereas sequences within the CR2 domain are not essential
(3). To define the relationship ofthe ElA-mediated disruption
of the TBP-Dr1 complex with the ElA-mediated TATA-
dependent transcription activation, a set of EMA mutants
previously assayed for transactivation activity (Fig. 2A) was
assayed for TBP-Drl dissociation. Whereas the WT EMA
protein, theElAur73-120mutant, and theElApml,,8/6/l mutant
could efficiently disrupt the TBP-Dr1 complex, theElA,=.36
mutant was inactive and theElA&u,3s7mutant was impaired
(Fig. 2A). The ElAd,3867 mutant begins to disrupt the
TBP-Dr complex when added in increasing quantities, while
the ElA-36 mutant totally lacks TBP-Dr dissociating ac-
These results closely coincide with the TATA-dependent
transactivation phenotype ofthese mutants (3) (summarized in
Fig. 2A). The ElA regions required for TBP-Drl complex
_,o _so-TBP ,_
interaction ofTBP with TFIIA is not altered by ElA. DNA-binding
assays used a 3' end-labeled TATA-containing fragment from the
adenovirus major late promoter (Ad-MLP), recombinant yeast TBP
(yTBP) 20 ng), and TFIIA (0.12 Mg) prepared from HeLa cells (6).
Reticulocyte lysate programed with E1A12s (0.5 Il) or an equivalent
amount ofbrome mosaic virus (BMV) programed control lysatewas
added to the binding reaction, incubated 30 min at 250C, and then
loaded on the gel. (B) The interaction of TBP with Drl is disrupted
by ElA, leaving the TBP-TFIIA complex. Complexes were formed
on the Ad-MLP as described in A using yTBP, TFIIA, and Drl (25
ng) (HeLa phenyl Superose fraction) (5). Reticulocyte lysate pro-
gramed with EMAU2S or BMV control was added as in A. (C)
Formation of a TBP-Drl complex. Formation of a TBP-Drl com-
plex, using recombinant human TBP (-10 ng) and Drl (-25 ng)
(HeLa Superdex S200 column fraction), was assayed using the
Ad-MLP. The presence of the Drl protein in the complexes was
demonstrated by the addition ofan anti-Drl antibody (0.2 Mg) (5).An
equal amount of anti-large T antibody (T Ab-2, Oncogene Science)
was added in the control lane. (D) The interaction of TBP with Drl
is disrupted by ElA. TBP-Drl complexes were formed as in C.
Increasing concentrations (1.5-12 Mg) of aGST-E1A12sfusion pro-
tein or GST control protein (12 Mg) were added.
The TBP-Drl interaction is disrupted by ElA. (A) The
disruption sharply contrast with the regions required for
disruption of the E2F complexes (4). For instance, the N-ter-
minal mutantElAda-36is as active as the WT ElA protein in
disrupting E2Fcomplexes, whereas theElAp1,m/%61mutant is
inactive (4). TheElAd,067mutant is impaired for E2F com-
plex dissociation as it is for disruption of the TBP-Dr1
complex, thus demonstrating an overlapping requirement for
CR1 sequences working both with the N terminus as well as
with CR2. Given the fact that the mutants that are inactive in
one assay are active in the other assay provides strong
evidence that the failure of the ElAd,=36 mutant protein to
disrupt the TBP-Drl complex is not a consequence of an
altogether inactive protein.
The finding that the EMA protein could disrupt the TBP-
Drl complex suggested the possibility of a direct interaction
Biochemistry:Kraus et al.
Proc. Natl. Acad. Sci. USA 91 (1994)
3 7.5 1.5
3 7.5 1.53
absorbed onto glutathione-Sepharose 4B beads and then
incubatedwithapreparationofrecombinantDrl protein. The
beads were washed and boiled for 10 min to release associ-
ated proteins, and then the released proteins were analyzed
in an SDS/acrylamide gel. Following transfer ofthe proteins
to nitrocellulose, the Drl proteinwas detected with a specific
antibody. As shown in Fig. 2B, the Drl protein did indeed
interact with the WT ElA protein, whereas there was no
interaction with the control GST protein. Of most impor-
tance, theE1Ada-36 mutant, which was defective for disrup-
tion of the TBP-Drl complex as well as TATA-dependent
transcription activation, was also defective for interaction
with Drl. The ElAu8.67 mutant, which is impaired for
TBP-Drl complex disruption but which at higher concentra-
tions displays some ability to disrupt, can interact with Drl
underthe conditions ofthis assay. TheElAR73120mutantand
theElAp,8/96lmutant were capable ofinteracting with the
Drl protein just as they were capable of disrupting the
TBP-Drl complex. This is in contrast to the fact that the
ElApaa/961mutant is unable to bind to the Rb protein. We
thus conclude that the ability ofEMA to disrupt the TBP-Drl
complex, which coincides with the activation of TATA-
interact with Drl.
Drl-Mediated Inhibition of hsp7O Promoter Activity Is Re-
versed by ElA. The experiments utilizing purified TBP and
Drl, demonstrating an ability ofEMA to disrupt the TBP-Drl
interaction, imply that Drl would inhibit TATA-dependent
transcription activation and that EMA should reverse the
inhibition. As adirect test ofthis expectation, we assayed the
ability of a cotransfected Drl-expressing plasmid to affect
expression ofa luciferase gene under the control ofthe hsp70
promoter, a promoter that we have previously shown is
activated by EMA in a TATA-dependent manner (3, 10).
Although the HeLa cells utilized for these assays do express
the human papilloma virus E7 product, which has functional
properties related to adenovirus EMA in terms ofRb associ-
ation and E2F activation, our previous experiments have
nevertheless shown that the activity ofthe hsp70 promoter is
not altered by the E7 product (11). As can be seen in Fig. 3,
Drl expression results in a 10-fold inhibition of hsp70 pro-
moter activity. Cotransfection of a plasmid expressing the
adenovirus E1A12S gene product reverses this inhibition in a
FiG. 2. EMA N-terminal sequences are essential for disruption of
the TBP-Drl complex. (A) Disruption of the TBP-Drl complex by
EMA mutants. Shown at the top is a schematic depiction ofthe EMA
mutants employed for TBP-Drl disruption assays and their activity
in activation of TATA-dependent transcription using the hsp70
promoter. The ability ofeachEMA mutant to trans-activate the hsp70
promoter in comparison to the WT ElA protein is indicated (3).
Shown below are assays for the formation ofthe TBP-Drl complex
and the ability ofGST-ElA12sWT or mutant to disrupt the complex.
Assays were as described in the legend to Fig. 1 C and D. GST-
EMA12s fusion proteins were added in increasing amount from 1.5 pg
to 7.5 pg as depicted at bottom. GST control protein was added at
the highest quantity (7.5 fg) to the last lane. (B) Direct interaction of
EMA with Drl, dependent on N-terminal EMA sequence. GST fusion
proteins, either WT E1A12S protein or the indicated mutant EMA
proteins (d12-26, d138-67, d173-120, pm928/%1), each bound to
glutathione-Sepharose 4B, were incubated with 2.5 pg of recombi-
nant DrL. As a control, the same amount ofDrl was incubated with
GST bound to Sepharose. After washing and elution, the proteins
were analyzed in an SDS/acrylamide gel and blotted to nitrocellu-
lose. Drl (indicated by arrow) was detected through the use of a
specific antibody (see text). An amount ofrecombinant Drl equal to
that used for the binding assays (2.5 fg) was run in an adjacent lane.
ofEMA with Drl. To address this possibility, we assayed for
the ability of a WT GST-E1A12S fusion protein and ElA
mutants to interact with Drl. The various ElA proteins were
by adenovirus ElA. HeLa cells were transfected with a hsp70-
luciferase plasmid (1.5 pg) alone, together with the CMV-Drl
plasmid (0.5 pig), or with the Drl plasmid and with increasing
amounts of either the CMV-E1A12s or theCMV-E1Ada-36plasmid
(0.1, 0.2, and 0.3jug).Extracts were prepared 48 hr aftertransfection
and assayed for luciferase activity. The luciferase activity of cells
transfected with hsp70-luc reporter plasmid alone was set as 100%.
Drl-mediated inhibition ofthe hsp7opromoteris reversed
Biochemistry:Kraus et al.
Proc. Nat!. Acad. Sci. USA 91 (1994)
Role of ElA in TBP-dependent transcription activation.
dose-dependent manner. Moreover, cotransfection of the
mutant fails to reverse the repression. We thus
conclude that the ability of ElA to disrupt a TBP-Drl
complex as seen in in vitro assays coincides with the ability
of ElA to reverse the Drl-mediated inhibition of transcrip-
tion in an in vivo context.
The E1A12s product has been shown to liberate the E2F
transcription factor from various multicomponent com-
plexes, activatingthe transcriptional potentialofE2F (2). The
ElA-mediated TATA-dependent activation ofthe hsp70 pro-
moter appears to be a similar event, coinciding with the
release ofTBPfrom a complex with DrW. Although we do not
yet know the basis for the previously observed TATA
specificity evident in the ElA activation (3, 10), we suspect
it may relate to the presence ofdistinctTFIID complexes that
may recognize TATA elements differentially (12, 13). Our
results also demonstrate that the N-terminai domain ofElA,
which is essential for TATA-dependehit transcription activa-
tion, is also required for TBP-Drl "disruption. Previous
experiments have demonstrated the interaction of a 300-kDa
polypeptide with the N-terminal ElA sequences (14-16).
Although an involvement of p300 in TATA-dependent tran-
scription activation has not beendemonstrated in previous
assays, itremains possible thatp300couldbeinvolved,either
through an association ofDrl or through an interaction with
At least part of the activity provided by the adenovirus
ElA CR3 sequence, unique to theElAD3S product, appears
to involve the TATA element (17), and recent evidence
shows that the ElAO3S product can interact with the TBP
transcription factor (18, 19). Based on these observations, we
speculate that the net result ofthe targeting of the TBP-Drl
complex may depend on the nature of the ElA protein
involved. As schematically depicted in Fig. 4, we suggest that
through an interaction ofthe ElA N-terminal sequences with
the Drl factor, there is a displacement -of TBP from the
TBP-Dr1 inhibitory complex. This would then leave Drl
associated with ElA or, alternatively, this interaction may
dissociate to leave the free components. Ifthe EMA135 protein
targets the complex, the CR3 sequences may then contact the
TBP molecule. If this interaction were more stable than the
interaction with Drl, one could then envision a stable com-
plex remaining involving ElA and TBP. In either case, there
would be a displacement of an inhibitory activity from TBP,
but in the case of theE1AI3M protein, an activating region
would be left in association with TBP.
Finally, it is also worth mentioning that these findings
demonstrate an activity of the ElA protein that correlates
with the transforming capacity of ElA, defined by the
requirement for the N-terminal domain. If the events ofDrW
inactivation and transformation are indeedrelated, this might
suggest a role for Drl in regulating cell proliferation.
We thank Kaye Culler and Teresa Jenkins for help with the
manuscript and Nancy-Hora for expert technical assistane
work was supported by the Howird Hughes Medical Institute
(J.R.N.), grants from the National Institutes of Health and the
American Cancer Society (to D.R.), and a National Institutes of
Health Physician Scientist Award (CA01358 to V.B.K.). D.R. was
the recipientofaFacultyResearchAwardfromtheAmerican Cancer
Moran, E. & Mathews, M. B. (1987) Cell 48, 177-178.
Nevins, J. R. (1992) Science 258, 424-429.
Kraus, V. B., Moran, E. & Nevins, J. R. (1992)Mol. Cell. Biol.
Raychaudhuri, P., Bagchi, S., Devoto, S. H., Kraus, V. B.,
Moran, E. & Nevins, J. R. (1991) Genes Dev. 5, 1200-1211.
Inostroza, J. S., Mermelstein, F. H., Ha, I., Lane, W. S. &
Reinberg, D. (1992) Cell 70, 477-489.
Maldonado, E., Ha, I., Cortes, P., Weis, L. & Reinberg, D.
(1990) Mol. Cell. Biol. 10, 6335-6347.
Bagchi, S., Rauchaudhuri, P. & Nevins, J. R. (1990) Cell 62,
Andersson, S., Davis, D. L., Dahlback, H., Jornvall, H. &
Russell, D. W. (1989) J. Biol. Chem. 264, 8222-8229.
Taylor, D. A., Kraus, V. B., Schwarz, J. K., Olson, E. N. &
Kraus, W. E. (1993) Mol, Cell. Biol. 13, 4714-4727.
Simon, M. C., Fisch, T. M., Benecke, B. J., Nevins, J. R. &
Heintz, N. (1988) Cell 52, 723-729.
Phelps, W. C., Bagchi, S., Barnes, J. A., Raychaudhuri, P.,
Kraus, V. B., Munger, K., Howley, P. M. & Nevins, J. R.
(1991) J. Virol. 65, 6922-6930.
Timmers, H. T. & Sharp, P. A. (1991) Genes Dev. 5, 1946-
Brou, C., Chandhoury, S., Davidson,
Egly, J.-M., Torm, L. & Chambon, P. (1993) EMBO J. 12,
Stein, R. W., Corrigan, M., Yaciuk, P., Whelan, J. & Moran,
E. (1990) J. Virol. 64, 4421-4427.
Whyte, P., Ruley, H. E. & Harlow, E. (1988) J. Virol. 62,
Egan, C., Jelsma, T. N., Howe, J. A., Bayley, S. T., Fergu-
son, B. & Branton, P. E. (1988) Mol. Cell. Biol. 8, 3955-3959.
Wu, L., Rosser, D. S. E., Schmidt, M. L. & Berk, A. (1987)
Nature (London) 326, 512-515.
Horikoshi, N., Maguire, K., Kralli, A., Maldonaldo, E., Rein-
berg, D. & Weinmann, R. (1991) Proc. Natl. Acad. Sci. USA
Lee, W. S., Kao, C. C., Bryant, G. O., Liu, X. & Berk, A. J.
(1991) Cell 67, 365-376.
I., Lutz, Y., Wu, J.,
Biochemistry:Kraus et al.