ABSTRACT
The adenovirus E1A-243R protein has the ability to force a resting cell into uncontrolled
proliferation by modulating the activity of key targets in cell cycle control.
Most of these regulatory mechanisms are dependent on activities mapping to
conserved region 1 (CR1) and the non-conserved N-terminal region of E1A. We have previously shown that CR1 functions
as a very potent transactivator when it is tethered to a promoter through a heterologous DNA binding domain. However, artificial DNA binding was not sufficient to convert full-length E1A-243R to a transactivator. Thus, an additional function(s) of the E1A-243R protein modulates the effect of CR1 in transcription
regulation. Here we demonstrate that a 44 amino acid region at the extreme C-terminus of E1A inhibited transactivation by a Gal4-CR1 fusion protein. Inhibition correlated with binding of the
nuclear 48 kDa C-terminal binding protein (CtBP), which has been implicated in E1A-mediated suppression of the metastazing potential of tumour cells.
This might suggest that CtBP binding can regulate E1A-mediated transformation by modulating CR1-dependent control of transcription.
The adenovirus E1A gene encodes two major proteins, of 243 and 289 amino acids
respectively (E1A-243R and E1A-289R), which are identical except for an internal deletion of 46
amino acids in E1A-243R (
1
). This deletion almost exactly removes one of the three conserved regions (CR3)
(
2
) and, with this, the major transcription activating function of E1A. Remaining
in E1A-243R are CR1 and CR2, which, together with the N-terminal domain of E1A, are required for the induction of cellular transformation (
3
,
4
). The underlying mechanism for transformation appears to be the ability of E1A to modulate gene expression through the
interaction with key proteins that regulate the cell cycle (reviewed in
5
).
CR1 and CR2 mediate the association between E1A-243R and a family of negative regulators of transcription, such as the
retinoblastoma tumour suppressor (p105-Rb) and its related proteins p107 and p130 (
6
). The interaction between E1A and p105-Rb or p107 dissociate inactive E2F-p105-Rb and E2F-p107 transcription factor complexes (reviewed in
7
,
8
). Originally, E2F was shown to activate transcription of the adenovirus E2 gene
(
9
), but has since gained status as one of the key regulators in cell cycle
progression, due to its control of genes essential for DNA replication (
10
-
12
).
Binding of the cellular protein p300 to CR1 and the N-terminus of E1A (
13
,
14
) also correlates with the E1A effects on cell cycle control and partially
overlap with the p105-Rb binding effects (
3
,
15
,
16
). Although the exact function of the p300 protein remains unclear, it displays properties characteristic of transcription co-activator proteins (
17
) and shows significant similarity to CBP, the transcriptional co-activator of CREB (
18
).
Far from all transcription regulatory activities by E1A-243R can be explained by a mere interaction between E1A and p105-Rb and/or p300.
In vitro
transcriptional repression by E1A has been explained by a direct association
between amino acids 1-80 of E1A and TBP, the TATA box binding protein of the general
transcription factor TFIID (
19
).
In vivo
the N-terminus and part of CR1 (
20
) has been shown to activate TATA-dependent transcription (
21
), possibly through the dissociation of TBP from the transcriptional inhibitor
Dr1 (
22
). Similarly, by dissociating ATF-CREB from YY1, the N-terminus and CR1 have been reported to relieve YY1 inhibition of
the mouse c-
fos
promoter (
23
). Dependent on the composition of the AP1 dimer (
24
), AP1-responsive genes are either repressed (
25
) or activated (
26
) through CR1-dependent mechanisms.
We have previously shown that a Gal4-CR1 fusion protein activates transcription from a reporter containing
Gal4 binding sites (
27
). This transcription regulatory mechanism is apparently independent of binding
to either the p105-Rb or p300 proteins (
27
). Thus, it is possible that CR1 has a general role in regulating gene activity,
but depending on the target, other regions in the E1A proteins may be involved
in interactions with additional cellular factors.
Exon 1 of E1A alone is capable of transforming cells in cooperation with an
activated
ras
oncogene (
4
,
28
,
29
). In fact, the presence of exon 2 appears to negatively modulate
in vitro
transformation, tumourigenesis and metastasis (
30
-
32
), effects which, in part, appear to correlate with the binding of a cellular
protein (CtBP) to the C-terminus of E1A (
33
). Together with CR1, exon 2 sequences may also control metastasis by repressing
metalloprotease gene expression (
34
), a function which is also dependent on CR1 (
25
). Thus, in transformation-related processes exon 2 of E1A has the potential to modulate exon 1 activities.
This work has focused on CR1-dependent activation of transcription and demonstrates that
transactivation by Gal4-CR1 was inhibited by the C-terminus of E1A-243R. Transcription inhibition was found to correlate with
the presence of the binding site for CtBP. These findings open the possibility
that CtBP regulates E1A-mediated transformation by modulating the E1A-dependent regulation of transcription.
G1E1BCAT has been described (
35
).
Gal4-RBoff and Gal4-MetBoff (
27
) express amino acids 28-90 and 1-90 of the adenovirus E1A-243R protein respectively fused to the DNA binding domain of
the yeast Gal4 transcription factor Gal4(1-147). In the different Gal4-CR1 fusion proteins the following amino acids of E1A-243R are expressed: Gal4-243R, 1-243; Gal4-MetDroff, 1-199; Gal4-MetXoff, 1-177; Gal4-MetDoff, 1-146;
Gal4-MetD^Dr, 1-146+200-243; Gal4-MetB^Dr, 1-91+ 200-243; Gal4-RB^Dr, 28-91+200-243; Gal4-ctE1A, 200-243. In Gal4-MetB^DrINV,
a DNA fragment corresponding to amino acids 200-243 was inserted in the opposite orientation, creating an extended open
reading frame of 46 unrelated amino acids following amino acid 91 of E1A. Gal4-MetB^Dr[Delta]225-238 and Gal4-RB^Dr[Delta]225-238 are derivatives of Gal4-MetB^Dr and Gal4-RB^Dr respectively that
specifically lack amino acids 225-238.
In MetBoff-E2 the activation domain of the bovine papilloma virus transactivator E2
in pCMVE2 wt ori (
36
) was replaced by amino acids 1-90 of the adenovirus E1A-243R protein.
pML00512S and pML00512S[Delta]CR1 have been described (
37
) and express cDNAs for the E1A-243R protein and a derivative lacking amino acids 38-65 respectively. pML00512S[Delta]225-238 and pML00512S[Delta]CR1[Delta]225-238 were constructed by deletion
of amino acids 225-238 in pML00512S and pML00512S[Delta]CR1 respectively.
Gal4-c-Jun (p178) has been described (
38
). Gal4-c-Jun-ctE1A and Gal4-c-Jun-INVctE1A were created by inserting a DNA fragment encoding amino acids 200-243 of E1A-243R in both orientations
into Gal4-c-Jun. The amino acid extension in Gal4-c-Jun-INVctE1A is identical to that created in Gal4- MetB^DrINV.
GST-MetBoff, GST-MetB^Dr and GST-MetB^Dr[Delta]225-238 are derivatives of the pGEX vectors
(Pharmacia) and express fusion proteins between the glutathione S-transferase (GST) protein and E1A moieties of corresponding Gal4-CR1 fusion proteins. GST-ctE1A expresses amino acids 200-243 of E1A-243R.
pcDNA3-CtBP has been described (
33
).
HeLa cells were maintained in DMEM supplemented with 10% NCS, 100 U/ml
penicillin, 100 [mu]g/ml streptomycin and 2 mM L-glutamine. Transfections were done in 60 mm Petri dishes by the
calcium phosphate co-precipitation technique essentially as described (
39
) using the amounts of purified plasmids indicated in the figure legends. The
total amount of transfected DNA was equalized with salmon sperm DNA. Cells were harvested at ~48 h post-transfection and cell extracts prepared by freeze-thawing three times in 0.25 M Tris-HCl, pH 7.5. CAT assays were performed essentially as described (
40
). The results were quantitated using the ImageQuant computer program on a PhosphorImager (Molecular Dynamics).
Whole cell extracts were prepared, as described below, from HeLa cells
transfected with plasmids encoding different Gal4 fusion proteins. One tenth of
the extract prepared from one 60 mm Petri dish was used to monitor binding to
0.25 ng of a
32
P-end-labelled 45 bp fragment containing one Gal4 binding site from G1E1BCAT (
35
). Binding was at room temperature for 30 min in binding buffer [20 mM Tris, pH 7.5, 75 mM KCl, 1 mM DTT, 0.25 mg/ml BSA,
0.056 mg/ml poly(dI[middot]dC), 12% glycerol] before loading on a 4.25% native polyacrylamide gel.
Electrophoresis buffer was 0.5* TBE (44 mM Tris, pH 7.5, 44 mM boric acid, 1.7 mM EDTA, pH 8.0) and the gel was run at 200 V for 3 h.
The commercially available coupled
in vitro
transcription/translation wheat germ extract system from Promega (TNTtm
SP6) was used according to the manufacturer's instructions. Full-length CtBP was synthesized directly from plasmid pcDNA3-CtBP (
33
).
HeLa cells were labelled with 2 mCi/ml [
32
P]orthophosphate for 4 h, washed extensively with PBS and disrupted in WCE
buffer (25 mM HEPES, pH 7.6, 300 mM NaCl, 1.5 mM MgCl
2
, 0.2 mM EDTA, 0.1% Triton X-100, 0.5 mM DTT) on ice. The following protease inhibitors were added
during lysis: leupeptin (10 [mu]g/ml), pepstatin (10 [mu]g/ml), aprotinin (1 [mu]g/ml) and PMSF (0.5 mM). Cell debris was removed by centrifugation.
GST fusion proteins were produced in
Escherichia coli
and bound to glutathione-agarose beads (
41
). Protein concentrations were estimated on a Coomassie stained SDS-polyacrylamide gel. Approximately equal amounts of GST fusion proteins
were mixed with either [
35
S]methionine-labelled
in vitro
translated proteins or crude protein extracts from half of a 60 mm Petri dish of
32
P-labelled HeLa cells. The binding reaction was in WCE[Delta] buffer (25 mM HEPES, pH 7.6, 75 mM NaCl, 1.5 mM MgCl
2
, 0.2 mM EDTA, 0.025% Triton X-100, 0.5 mM DTT and the aforementioned amounts of protease inhibitors)
with rotation at 4oC for 2 h. Beads were washed four times in HEPES binding buffer (20 mM
HEPES, pH 7.6, 50 mM NaCl, 2.5 mM MgCl
2
, 0.1 mM EDTA, 0.05% Triton X-100) and bound proteins were separated on a polyacrylamide gel at 200 V
for 6 h and visualized by autoradiography.
We have previously shown that a fusion between the yeast Gal4(1-147) DNA binding domain and essentially CR1 of E1A-243R (Gal4-MetBoff and Gal4-RBoff; Fig.
1
) strongly activates transcription from a CAT reporter plasmid (G1E1BCAT;
35
) driven by a synthetic promoter containing one Gal4 binding site upstream of
the adenovirus E1B TATA element (Fig.
1
;
27
). In the absence of Gal4 binding sites, the Gal4-CR1 fusion protein repressed SV40 enhancer-dependent transcription (
27
). We have previously postulated that both activities might be related to the
role of CR1 as a transcriptional repressor domain in its natural context of the
E1A-243R protein (
27
). This implied a simple model where CR1 could activate transcription when
tethered to a promoter, but repress transcription, possibly through squelching of essential component(s) of the transcription machinery, in the absence of promoter binding. However, promoter binding was not
sufficient to convert a Gal4 fusion expressing the full-length E1A-243R protein (Gal4-243R) to an activator of transcription (Fig.
1
). Thus, it appeared as if other regions present in E1A-243R had the potential to control CR1-dependent transcriptional activation.
The C-terminus of E1A has previously been shown to bind a 48 kDa cellular phosphoprotein called the C-terminal binding protein (CtBP) (
32
). Binding of CtBP correlates with the capacity of the C-terminus of E1A to reduce tumourigenesis and metastasis of cells co-transformed by E1A and
ras
(
33
), activities which are intimately associated with CR1. Amino acids 225-238 are required for binding of CtBP to exon 2 of E1A (
33
) and we show that the last 44 amino acids of E1A, expressed as a GST fusion
(GST-ctE1A), was sufficient to bind a phosphoprotein of ~48 kDa from HeLa cell whole cell extracts (Fig.
2
A). Furthermore,
in vitro
translated CtBP was able to bind specifically to GST-MetB^Dr and GST-ctE1A. CtBP binding by GST, GST-MetBoff and GST-MetB^Dr[Delta]225-238 was calculated to be <5% of the observed binding to GST-MetB^Dr (Fig.
2
B).
The target for the Gal4-CR1 activator is not known. However, activation by Gal4-CR1 was precluded in the presence of wild-type E1A-243R, suggesting that CR1 transactivation is mediated
by a soluble factor which binds Gal4-CR1 as well as E1A-243R (
27
). With this in mind, it seemed possible that the CtBP binding domain affected
the interaction between CR1 and its mediator. Therefore, it was of interest to
determine whether the inhibitory effect could be observed if the CtBP binding
domain and CR1 were present on different molecules.
The G1E1BCAT reporter (
35
) contains a strong, cryptic binding site for the bovine papilloma virus E2
transactivator (Sollerbrant and Svensson, unpublished results). This allowed a
fusion protein expressing amino acids 1-90 of E1A linked to the E2 DNA binding domain (MetBoff-E2) to activate transcription from G1E1BCAT (Fig.
5
). Co-transfection of the reporter with increasing concentrations of either Gal4
(1-147) or a Gal4 fusion expressing amino acids 200-243 of E1A-243R (Gal4-ctE1A) resulted in more or less comparable levels of
transactivation by MetBoff-E2 (Fig.
5
). These results suggested that Gal4-ctE1A (in
trans
) did not cause a specific inhibition of MetBoff-E2 transactivation. The same results were observed when G1E1BCAT
transcription was induced by a c-Jun-E2 fusion protein (
38
; data not shown). These results indicated that the inhibitory effect of the C-terminus of E1A on CR1-dependent transactivation required the presence of the CtBP binding
domain on the same molecule as CR1.
The mechanism by which CtBP modulates transformation by E1A is still largely
unknown (
33
). The correlation between the CtBP binding domain and inhibition of the
transcriptional activation by Gal4-CR1 suggested that the CtBP binding domain might function as a silencer of transcriptional activation. Therefore, it was of interest to determine whether transcriptional inhibition by
the CtBP binding domain was restricted to CR1-dependent transactivation. For these analyses a DNA fragment encoding the
last 44 amino acids of E1A-243R was fused, in both orientations, to a previously described Gal4 fusion protein construct expressing the transcription activating domain of human c-Jun (Gal4-c-Jun;
38
). As shown in Figure
6
, Gal4-c-Jun displays a potent transactivating potential when measured on the reporter containing a Gal4 binding site. However, in contrast to the CR1 activator,
the activating domain of c-Jun was not significantly affected by the presence of the C-terminus of E1A (Gal4-c-Jun-ctE1A and Gal4-c-Jun-INVctE1A; Fig.
6
). Addition of E1A sequences to Gal4-c-Jun did not affect accumulation of the fusion proteins, but
specifically allowed a GST-c-Jun-ctE1A fusion to bind
in vitro
translated CtBP and Gal4-c-Jun-ctE1A to be immunoprecipitated by an E1A-specific antibody (data not shown). We therefore
conclude that the CtBP binding region of the E1A-243R C-terminus is not a general repressor domain.
Wild-type E1A-243R is able to both activate and repress transcription
in vivo
(reviewed in
5
). The transcriptional repression has been observed on a variety of promoters, often controlling tissue-specific or differentiation-specific genes (for examples see
25
,
29
,
37
,
4
2
-
4
7
). Transactivation, on the other hand, is described for a small subset of genes
and is generally weak (
20
,
23
,
4
8
-
5
1
). Although the exact regions of E1A required for the different regulatory
mechanisms varies, both activation and repression often require CR1. It was
therefore interesting to find that the CR1 domain fused to the heterologous
yeast Gal4 DNA binding domain activated transcription from reporter constructs
carrying Gal4 binding sites (
27
). Thus, transactivation possibly occurs through the ability of Gal4-CR1 to recruit an essential factor of the transcription machinery to a
promoter. Similarly, in the absence of promoter localization, transrepression
by wild-type E1A-243R might be explained by squelching of the same cellular factor by
CR1. Surprisingly, a Gal4 fusion expressing full-length E1A-243R was unable to activate transcription in the same assay (Fig.
1
). We here demonstrate that the inability of Gal4-243R to activate transcription correlates with the presence of sequences
in the C-terminal exon of E1A.
In our search for the region within exon 2 which inhibited CR1-mediated transcription activation we identified a minimal genetic element
consisting of amino acids 225-238 of E1A-243R (Fig.
3
). These amino acids correspond to the binding site for CtBP (
33
). Inhibition of Gal4-CR1-dependent transactivation was alleviated in the presence of an
excess of a competitor able to bind CtBP, implying that the inhibition was
mediated by binding of a diffusible factor to the CtBP binding domain of E1A (Fig.
3
). Similar models for inhibition of transactivation (quenching) have been proposed, for example, for the Gal80 repression of Gal4 (
5
2
,
5
3
), MDM2 repression of p53 (
5
4
) and PHO80 repression of PHO4 (
5
5
). In these examples, binding of the repressor protein constitutes an elegant
way of regulating transcription factor activity.
Transcriptional repression can also be observed as a general ability of a
repressor protein to interfere with assembly of the basal transcription
complex. One example is the Dr1 protein, which by binding TBP excludes TFIIA
and/or TFIIB from entering the basal transcription complex (
5
6
). A Gal4 fusion protein expressing only the CtBP binding domain (Gal4-ctE1A) was, however, unable to specifically inhibit transcription induced
by a MetBoff-E2 fusion protein (Fig.
5
) or the SV40 enhancer (data not shown). The same results were obtained
irrespective of the presence of Gal4-responsive elements in the reporter constructs (Fig.
5
; data not shown). Moreover, the inhibition was only observed when the CtBP
binding domain was present on the same molecule as the CR1 activating domain
(Figs
1
and
5
). Collectively, these data suggest that tethering of the CtBP binding domain to
a promoter does not result in a general repression of transcription. Since the
CtBP binding domain failed to inhibit transactivation when attached to an
unrelated Gal4-c-Jun fusion protein (Fig.
6
), inhibition appeared specific for CR1. Based on these data, we propose that
the CtBP binding domain in E1A prevents Gal4-CR1 from interacting with its biological target or alternatively modifies
CR1 or its target protein(s).
We are grateful to Drs J.Baichwal, M.Green, J.Lillie and G.Westin for kind gifts
of plasmids. We also thank Drs G.Akusjärvi, M.Bondesson and M.Mannervik for fruitful discussions and for critically reading the manuscript and A.Richnau for excellent technical assistance. This work was supported by grants
from the Swedish Cancer Society and the Swedish Medical Research Council. G.C.
is supported by a grant (CA-31719) from the National Cancer Institute.
REFERENCES
Return