This is best illustrated in. Fig. 4 (lanes 6), which showed that active EIIaE transcription occurred in d1808-infected F9(RA+cA) cells, in the absence of detectable ...
Proc. Nati. Acad. Sci. USA Vol. 87, pp. 1782-1786, March 1990 Biochemistry
Differential activation of the E2F transcription factor by the adenovirus Ela and EIV products in F9 cells (Ela-like activity/differentiation/transactivation)
H. BOEUF, B. REIMUND, P. JANSEN-DURR, AND C. KtDINGER* Laboratoire de Gdndtique Mol6culaire des Eucaryotes du Centre National de la Recherche Scientifique, U184 de Biologie Mol6culaire et de G6nie G6ndtique de l'Institut National de la SantW et de la Recherche, Facult6 de M6decine, 11, rue Humann, 67085 Strasbourg C6dex, France
Communicated by Pierre Chambon, December 18, 1989
Expression studies of the early EIIa tranABSTRACT scription unit (EIaE) of the adenovirus EIa-deletion mutant d1312 in murine embryonal carcinoma stem cells suggested that these cells contain an activity that substitutes for the viral Ela. To further characterize this cellular EIa-like activity, we analyzed expression of the EIIaE promoter as well as the binding activity of the cognate E2F transcription factor after infection of F9 embryonal carcinoma cells and their differentiated derivatives with wild-type adenovirus, Ela (di312), or EIV (dl808) deletion mutants. We show that, in contrast to the viral Ela proteins that transactivate the EIIaE promoter in F9 cells only after differentiation, the viral EIV products activate the EIIaE promoter most efficiently in undifferentiated F9 cells. We also show that the EIV products induce a specific modification of the E2F transcription factor leading to its cooperative binding to the EIIaE promoter. Although the Ela-dependent transactivation of EIIaE in differentiated cells is also in part mediated by E2F, it does not by itself correlate with the simultaneous binding of two E2F molecules. In these cells E2F dimer binding only occurs as a secondary effect of Ela that also stimulates EIV expression. Our results suggest that Ela and- EIV act through separate pathways, inversely regulated during cell differentiation, with the so-called "Ela-like" activity contributing to this modulation.
Recently, we have observed that upon infection of these differentiated F9 cells with the wild-type (wt) adenovirus, but not with the EIa-deleted (d1312) derivative, the E2F protein undergoes modifications that lead to the cooperative binding of two E2F molecules to the EIIaE promoter (30). A very similar modification of E2F had been detected in adenovirusinfected HeLa cells too, where it also depends on Ela expression (12), suggesting that in both cell types EIIaE promoter activation was at least in part linked to this E2F modification. However, no such modification of E2F was detected in undifferentiated F9 cells in the absence of adenovirus infection (12) or early after adenovirus infection (30). It appeared therefore that the formerly defined "EMa-like" activity fails, by its own, to activate E2F. In an attempt to further characterize the mechanism of EIIaE promoter activation in undifferentiated F9 cells and to analyze the differentiation-dependent modifications of the E2F transcription factor, we have studied the expression of the adenovirus EIIaE gene and the specific DNA binding properties of E2F in F9 cells, as a function of cell differentiation. Our results suggest that transactivation of EIIaE occurs through distinct modifications of E2F induced by Ela and/or EIV, depending on the differentiation state of the cell.
MATERIALS AND METHODS Cells, Virus, and Extracts. The mouse F9(EC) stem cells were grown in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum. EC cells were induced to
The characterization of the cellular transcription factors implicated in the coordinate regulation of adenovirus gene expression (1) is an essential step toward the understanding of the mechanisms of transcriptional control in eukaryotic cells. It is now well documented that the viral EIa proteins transinduce adenovirus genes through different cellular transcription factors (2). Proteins such as E2F, AP1, E4F, TFIID, and TFIIIC have been identified as targets for Ela-mediated transinduction of the adenovirus EIla early (EIIaE) (3, 4), EIII (5), EIV (6), EOb (7), and VA RNA gene (8) promoters,
differentiate in the presence of 0.1 /iM retinoic acid and 1 mM N6,02'-dibutyryladenosine 3',5'-cyclic monophosphate (Sigma) [F9(RA+cA)], essentially as described (12, 13). Infections with the wt adenovirus type 5 (Ad5), the EIadefective d1312 derivative (14), or the EIV-defective d1808 derivative of adenovirus 2 (15) were carried out at 100 plaque-forming units per cell and cells were harvested 20 hr after infection. Whole cell extracts were prepared by using the freeze-thawing method (0.4 M KCI buffer) of Kumar and Chambon (16). The gel-shift and dimethyl sulfate-interference experiments were performed under probe-excess conditions as described (12, 17). Recombinant Plasmids and Transfections. The LSWT EIIaE recombinant contains the EIIaE sequences extending between -250 and +719 (with respect to the EIIaE major start site), inserted in front of the rabbit 8-globin sequences (-9 to +1700), which provide splice and poly(A) sites (18). The Xba I linker substitution (LS series) and internal deletion mutants (A series) of the EIIaE promoter used in this study were derived from the parental LSWT recombinant, the
respectively. A cellular function, akin to the viral EIa activity, has been described in undifferentiated mouse embryonal carcinoma (EC) cells (9, 10). In these cells, early adenovirus promoters, like the EIIaE promoter, are active in the absence of viral Ela products, whereas after differentiation their expression becomes dependent on Ela. The murine equivalent of E2F, a factor previously found in HeLa cells and implicated in the EIa-mediated activation of the ETIaE promoter, has been detected in F9 cells as well. It has been reported that E2F binding activity in F9 cell extracts was abolished (11) or significantly reduced (10, 12) after differentiation, suggesting that the activity of this factor was modulated by the EIa-like activity. In addition, Reichel et al. (11) found that detectable amounts of E2F could be restored in differentiated F9 cells by adenovirus infection.
Abbreviations: Ad5, adenovirus type 5; dl, deletion; m.u., map units; wt, wild-type; EC, embryonal carcinoma; F9(EC), F9 EC cells; F9(RA+cA), F9 cells induced to differentiate in the presence of 0.1 ,uM retinoic acid and 1 mM N6,02'-dibutyryladenosine 3',5'-cyclic
The publication costs of this 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.
monophosphate. *To whom reprint requests should be addressed.
1782
Biochemistry: Boeuf et al. numbers after LS and A referring to the position of the nucleotides relative to the EIIaE major start site (+1), between which an Xba I linker has been inserted (18, 19). The pEIASV recombinant, comprising the entire adenovirus Ela transcription unit but with a Simian virus 40-derived poly(A) site, and pEIA-, an EIA-defective recombinant retaining only the Ela region upstream to position + 129, have also been described (18). The pEIV construct comprises AdS sequences extending from 100 map units (m.u.) to the HindIII site at 91 m.u. (20). Cells were transfected by calcium phosphate coprecipitation (18) with 10 t&g of the EIIaE recombinants and 5 ,ug of either pEIA- or pEIASV or 7.5 pug of pEIV or M13mp8 double-stranded carrier DNA. The precipitate was left on the cells for 16-20 hr before the medium was changed, and the incubation was prolonged for another 20-24 hr before cytoplasmic RNA extraction (20). Undifferentiated and differentiated F9 cells were equally susceptible to DNA transfection as determined after transfection of a Simian virus 40 promoter-lacZ fusion recombinant, by staining the transfected cells for p-galactosidase activity. Transfection efficiencies were routinely controlled by this method on separate plates processed in parallel. EIIaE-specific transcripts were analyzed by quantitative S1 nuclease mapping (18). The levels of the endogenous glyceraldehyde-3-phosphate dehydrogenase (G3PDH) transcripts have been determined in parallel, using a G3PDH cDNA probe (21). The amount of G3PDH RNA remained essentially unchanged in the F9 cells upon their differentiation (unpublished observation). In the present study only experiments where these G3PDH signals were of nearly equal intensity in the different mouse cell lines were taken into consideration. All transfection experiments were repeated at least three times with independent plasmid DNA preparations.
Proc. Natl. Acad. Sci. USA 87 (1990) -36
A
.1-23 -331
i
EllaE
-48-55 -68 -82
I-HIr-H
h-H
Ti E2F E2F ATF
+40
Si probe
-40
*I
retardation-probes
1xE2F F-I 2xE2F E2F
competitors
i*X -*-4
B
E2Fm
F9(RA+cA)
F9(EC) E2F binding activity
wt wt dI312 d1312 E2F - + -l:-- E2Fm +._ _,-_~_---_ -._I_ -
-
(Probe:ixE2F) cl- *
to
W
1 2 3 4 5 6 wt d1312 + E2F E2Fm --i-:-- +:+
7 8 9 10 11 12 wt : d1312
IZ.-..
-
C3-11U
(Probe:2xE2F)
C2_ ----
13 14 15 16 17 18
19 20 21 22
c EllaE expression
wt :d1312
RESULTS Maximal EIIaE Promoter Activity in Adenovirus-Infected F9 Cells Correlates with the Simultaneous Binding of Two E2F Molecules. Undifferentiated F9 cells [F9(EC)] and F9 cells that had been induced to differentiate in the presence of retinoic acid and N6,02'-dibutyryladenosine 3',5'-cyclic monophosphate [F9(RA+cA)] were infected with wt AdS or the Ela-deleted derivative, d1312. In agreement with previous results (30), at 20 hr after infection, EIIaE transcripts accumulated in F9(EC) cells in the absence of the viral Ela products (dl312 infection) to levels similar to those reached in their presence (wt infection), as revealed by quantitative S1 nuclease analysis (Fig. 1C). By contrast, in the differentiated F9(RA+cA) cells, EIIaE-specific RNA was detected most readily after wt infection. These experiments, together with earlier protein expression studies (9), suggest that the undifferentiated cells provide a function that substitutes for the viral EIa products in a d1312 infection and this was referred to as the Ela-like activity. In a previous study (12), we have shown that the Elatriggered activation of the EIIaE promoter in HeLa cells was correlated with a modification of the E2F transcription factor that resulted in its cooperative binding to both E2F sites of the EIIaE promoter. No such complexes (referred to hereafter as the dimeric E2F complexes) could be detected in uninfected F9(EC) cells (12) or early after infection of these cells (30), indicating that the so-called Ela-like activity, present in these cells, was not by itself sufficient to induce the relevant modification of E2F. Here we examined whether a component produced during the viral infection was required, in addition to the cellular Ela-like activity, to induce this modification, thereby leading to the formation of dimeric E2F complexes. To this end, we performed band-shift assays using synthetic oligonucleotide probes comprising either one E2F binding site (1 x E2F, Fig.
1783
-
EllaE1 [ 1
wt
'dl312
[3 3
4
FIG. 1. The presence of ternary E2F complexes is correlated with EIIaE expression in infected F9 cells. (A) The structure of the EIIaE promoter region with critical promoter elements (open boxes with coordinates relative to the major start site +1) is depicted and the corresponding factors are indicated. The probe used for S1 nuclease mapping is the single-stranded synthetic oligonucleotide corresponding to the transcribed strand of this promoter between positions -40 and +40, labeled at +40. Probes used for gel-retardation experiments are synthetic double-stranded oligonucleotides (12) spanning one [1xE2F, (-31) ACTAGTTTCGCGCGCTTTCT (-50)] or two [2xE2F, (-35) GTTTCGCGCCCTTTCTCAAATTTAAGCGCGAAAA (-68)] E2F binding sites. The competitor oligonucleotides used in the gel-shift assays are indicated below and correspond to the unlabeled 2 x E2F probe oligonucleotide (E2F) and to an oligonucleotide [E2Fm, (-35) GTTl ACTCAGATAACTCAAATTTAAGIACIAQAA (-68)] with sequences altered (underlined) at both E2F binding sites (X). (B) The 5' end-labeled 1 xE2F (lanes 1-12) or 2xE2F (lanes 13-22) probes were used in standard gel-shift assays, with 10 Ag of either wt or d1312-infected F9(EC) (lanes 1-6 and 13-18) or F9(RA+cA) (lanes 7-12 and 19-22) cell extracts in the presence of 2 1Lg of polyd(A-T) as nonspecific competitor. Where indicated, the extracts were preincubated with the unlabeled E2F or E2Fm competitor oligonucleotides, at molar ratios of 100, relative to the probe. C1, C2, and C3 refer to retarded complexes (the unbound probe, always present in excess over the retarded probe, is not shown). (C) Total cellular RNA from wt or d1312-infected cells, isolated at 20 hr after infection was analyzed by S1 nuclease mapping using a synthetic oligonucleotide (-40/+40) as probe. The specific transcripts (initiated at the major EIIaE start site, EIIaE1) are bracketed.
LA) or both of the adjacent E2F recognition sites (2xE2F, Fig. LA) present in the ETMaE promoter. The lx E2F probe generated a single retarded band (Cl) after incubation with
Proc. Natl. Acad. Sci. USA 87 (1990)
Biochemistry: Boeuf et al.
1784
extracts from undifferentiated and differentiated F9 cells, whether infected with the wt or d1312 virus (Fig. 1B, lanes 1-12). That this band corresponded to an E2F-specific complex was demonstrated by competition experiments. Preincubation of the extracts with a 100-fold molar excess of the E2F competitor oligonucleotide (see Fig. UA) prevented the formation of band C1 (Fig. 1B, lanes 2, 5, 8, and 12). By contrast, the same amount of an oligonucleotide altered in both E2F binding sites (E2Fm, see Fig. 1A) had essentially no effect (Fig. 1B, lanes 3, 6, 9, and 11). The apparent DNA binding activity of E2F, as reflected by the intensity of band C1, was essentially the same in each of the extracts tested in this particular experiment (compare lanes 1, 4, 7, and 10). However, depending on the extract preparation, E2F DNA binding activity could be up to 3-fold higher in extracts from undifferentiated F9 cells, under conditions where similar amounts of the Spl transcription factor were extracted from undifferentiated and differentiated F9 cells (B.R., unpublished observation). However, in no case was the E2F binding activity significantly affected by the viral EIa products. When the 2xE2F probe was used with the same infected F9 cell extracts (Fig. 1B, lanes 13-22), the following patterns were obtained. Only one major E2F-specific complex (C2) was observed in the presence of d1312-infected F9(RA+cA) cell extracts (lanes 21 and 22), whereas an additional E2Fspecific complex (C3) was seen with the other extracts (lanes 13-20), irrespective of the presence of the viral Ela products. The precise binding sites of C2 and C3 were localized on the 2x E2F probe by dimethyl sulfate-interference mapping. The results of a typical experiment are shown in Fig. 2, where the C2 and C3 complexes generated by wt and d1312-infected F9(EC) cell extracts have been mapped. In complexes C2 from wt and dl312 extracts (lanes 5 and 3), the nucleotides between positions -41 and -45 were slightly undermethylated, indicating that the proximal (relative to the EIIaE start site) E2F binding site was predominantly occupied in these complexes. By contrast, the clear undermethylation of residues between -36 and -48 (and most clearly between -39 and -44) and at positions -61 and -63 in complex C3 (lanes 2 and 4) reflected simultaneous interactions with both E2F binding sites, in agreement with the slower migration rate of these complexes (Fig. 1B, lanes 13-18). Interference patterns F9(EC) FIG. 2.
dl31 2
wt
-
C3 C2 C3 C
-
-41
'-45 ..
:?of
4
.f;" 'I"
1w,
gm .;,
.4ww " "I
Rk",
-61
-
-
Dimethyl sulfate-inter-
ference mapping of complexes C2 and C3 formed in the presence of infected F9(EC) cell extracts. Bands corresponding to complexes C2 (lanes 3 and 5) and C3 (lanes 2 and 4) and unbound probe (lane 1) were excised from a preparative retardation gel run with wt (lanes 1-3) or d1312-infected (lanes 4 and 5) F9(EC) cell extracts and a dimethyl sulfatetreated 2xE2F probe, 5' endlabeled on the nontranscribed strand. After purification, the corresponding DNA was cleaved at methylated residues. Residues whose methylation interfered with complex formation are marked by open bars or circles (solid lines and dashed lines span strong and weak interferences, respectively).
Symbols on the left refer to complex C3; those on the right refer to complex C2. Coordinates are given with respect to the EIIaE
-63
1
2
3
4
major start
site
(+1).
very similar to those shown in Fig. 2 were also observed with the corresponding complexes (Fig. 1B, lanes 19-22) produced with d1312 and/or wt-infected F9(RA+cA) cell extracts (not
shown). Altogether, these results suggested that, as in HeLa cells (12), the murine E2F factor must undergo a specific modification(s) in order to simultaneously saturate both E2F binding sites of the EIIaE promoter and thereby induce its transcriptional activity. In differentiated F9 cells this modification(s), as monitored by the level of C3 complex, appeared to be dependent on the presence of the viral EIa products (Fig. 1B, compare lanes 19 and 21). This was clearly not the case in undifferentiated F9 cells, where similar amounts of C3 complex were found after d1312 or wt infection (compare lanes 13 and 16). Furthermore, our previous ob-
servations that no dimeric E2F complexes could be detected by this assay with extracts from uninfected F9(EC) cells (12) or early after infection of these cells (30) indicated that the components of undifferentiated cells did not by themselves induce the effect but that viral compounds other than Ela (since produced upon d1312 infection) were required in ad-
dition. The Adenovirus EIV Transcription Unit Transactivates the EIIaE Promoter in F9 Cells. Cotransfection experiments in HeLa cells, with plasmids bearing the EIIaE, EIV, or EIa transcription units (20) and more recently infection experiments with adenovirus deletion mutants (22, 23), have shown that EIIaE promoter activity can be stimulated by EIV gene products as efficiently as by the Ela proteins. These observations raise the interesting possibility that the products of the viral EIV transcription unit might mediate the increase of EIIaE promoter activity in F9(EC) cells at 20 hr after infection. Therefore, it was important to establish that, in contrast to EIa, the EIV proteins were capable of transinducing EIIaE in undifferentiated F9 cells. This was assayed in cotransfection experiments, in which recombinant plasmids encoding the Ad5 EIa (pEIASV) or EIV (pEIV) transcription units were cotransfected with recombinants bearing the wt (LSWT), linker substitution (LS derivatives), or deletion mutants (A derivative) of the EIIaE promoter. The results of a typical experiment in F9(EC) cells are shown in Fig. 3A. As already reported (30), the viral Ela products were not able to transactivate the EIIaE promoter in these cells (Fig. 3A, compare lanes 1 and 2). However, a 5-fold induction of EIIaE promoter activity could be obtained by cotransfecting the LSWT recombinant with the EIV-encoding plasmid (pEIV) (Fig. 3A, compare lanes 1 and 4 and lanes 3 and 7). This effect was strictly dependent upon both E2F recognition sites, since the LS-4839 and A-7152 recombinants, specifically mutated in these sites, were not transinduced by pEIV (Fig. 3A, compare lanes 6 and 9 and data not shown). Mutation in the ATF binding site (A-9170) did not impair EIV transinduction (Fig. 3A, compare lanes 5 and 8). Although responsiveness to viral Ela was recovered upon differentiation of the F9 cells (30), we examined whether the EIV products retained their transactivating capacity in differentiated F9 cells. As shown in Fig. 3B, the viral Ela products as well as the EIV products did activate about 10-fold the EIIaE promoter in the differentiated cells (Fig. 3B, lanes 1-3). Cotransfection of pEIASV, together with pEIV, did not further increase EIIaE expression under our present transfection conditions (not shown). Transcriptional analysis of selected EIIaE promoter mutants (Fig. 3B, lanes 4-7) revealed that in these cells the EIV-mediated stimulation mainly involved the E2F recognition sites, as in undifferentiated cells. Similarly, the ATF recognition site was not it clearly was implicated in this effect (not shown), whereas required for Ela-mediated activation (30). Taken together, these data suggest that the viral compound responsible for EIIaE promoter activation in F9(EC) cells, and hence for the
Biochemistry: Boeuf et al. B
F9(EC)
A
Proc. Natl. Acad. Sci. USA 87 (1990)
+pEIA7 +pEIASV
+pEIV
F9(RA+cA)
F9(EC)
+pEIA- +pEIASV +pEIV
A
E2F binding
activity
1785
F9IRA+cA)
_
_
'a x
'a a
O
C3-
CD
-J
-j
-j
Co)
co)
C
-j
-i
C2
-j
1
i2 .2 3
4 5 6
EIlaE1 [ B
3
2
-pEIV F-rHN J Co c n 1i
