ABSTRACT
B-myb
belongs to a group of cell cycle genes whose transcription is repressed in G
0
/early G
1
through a binding site for the transcription factor E2F. Here, we show that the
B-myb
repressor element is specifically recognised by heterodimers consisting of DP-1 and E2F-1, E2F-3 or E2F-4. Surprisingly, E2F-mediated repression is dependent on a contiguous
corepressor element that resembles the CHR previously established as a
corepressor of the CDE in cell cycle genes derepressed in S/G
2
, such as
cyclin A, cdc2
and
cdc25C
. A factor binding to the
B-myb
CHR was identified in fractionated HeLa nuclear extract and found to interact
with the minor groove, as previously shown by
in vivo
footprinting for the
cyclin A
CHR. The
B-myb
and
cdc25C
CHRs are related with respect to protein binding but are functionally clearly
distinct. Our results support a model where both E2F- and CDE-mediated repression, acting at different stages in the cell cycle,
are dependent on promoter-specific CHR elements.
In mammalian cells, a specific set of cell cycle genes transcribed around the G
1
/S border is regulated by factors of the E2F/DP family (for reviews see refs
1
-
3
). The heterodimeric E2F/DP transcription factors frequently act as repressors
in G
0
/early G
1
owing to their association with pocket proteins of the pRb family. In late G
1
, the pocket proteins become hyperphosphorylated and dissociate from the complex
with E2F/DP, leading to the derepression of E2F-regulated genes. Several genes expressed in late G
1
/early S, including B-
myb
(
4
,
5
), DHFR (
6
) and
E2F-1
(
7
,
8
), have been shown to be repressed through an E2F-mediated mechanism in G
0
/G
1
and to be derepressed in late G
1
. Although a plethora of E2F, DP and pocket protein family members has been
identified, their precise role in the regulation of specific genes remains
elusive (for reviews see refs
2
,
3
).
Transcription of the
B
-
myb
gene in mouse fibroblasts greatly increases in mid-G
1
and reaches peak levels in S-phase (
4
). Structure-function analysis of the
B-myb
promoter identified an element close to the transcription start sites necessary
for cell cycle regulation (
4
). This element (CTTGGCGG) represents an E2F site, as shown by protein binding
using cell extracts and recombinant proteins (
9
). Mutation of this E2F site leads to an up-regulation of transcription in G
0
cells (
4
), indicating that the interacting protein complex acts as a repressor.
In vitro
experiments with cell extracts suggested that the G
0
complex contains the p107 pocket protein (
5
). In contrast, free E2F is found in cell extracts throughout the cell cycle,
and other higher order DNA-binding complexes are detected around S-phase entry (
5
). The function of the late G
1
/S-phase complexes, however, remains unclear, because genomic footprinting of
the
B-myb
promoter failed to show any protection of the E2F site later than mid-G
1
(
9
).
Cyclin A
,
cdc2
and
cdc25C
exemplify a group of cell cycle genes whose transcription is up-regulated later than that of
B-myb
, i.e. in S-phase (
cyclin A
,
cdc2
) and G
2
(
cdc25C
) (for a review see
3
). For all three promoters, repression of upstream activators via the `cell
cycle-dependent element' (CDE) has been established as the major regulatory
mechanism (
10
-
12
). In addition, repression of the
cyclin A
,
cdc2
and
cdc25C
promoters is also dependent on a contiguous element, termed `cell cycle genes
homology region' (CHR) (
12
). As shown by genomic footprinting, both elements are bound by the repressor
proteins in a periodic fashion, the CDE in the major groove and the CHR in the
minor groove (
12
). The nature of the proteins interacting with the CDE and CHR elements remains
at present unknown.
In the course of our studies, we noted a significant homology between the CHR in
the
cyclin A
,
cdc2
and
cdc25C
and the region immediately downstream from the E2F site in the
B-myb
promoter, raising the question as to whether E2F-mediated repression might also be dependent on a CHR-like downstream element. In this manuscript, we show that this is
indeed the case and identify a nuclear activity interacting with the
B-myb
CHR in the minor groove. We also show that this activity is related to but
distinct from the factors interacting with the CHR of
cdc25C
, indicating that both E2F- and CDE-mediated repression is dependent on promoter-specific corepressors.
NIH3T3 cells were kindly provided by R. Treisman (ICRF, London) and cultured in
Dulbecco-Vogt modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum
(FCS). HeLa cells were maintained in a medium containing DMEM and 10% calf
serum. For synchronisation in G
0
, NIH3T3 cells were maintained in serum free medium for 2 days. NIH3T3 cells
were transfected by the DEAE dextran technique and determination of luciferase
activity was performed as described (
10
). A SV40 promoter reporter construct was used to standardise the results.
The
B-myb
constructs spanned the region from -301 to +100 relative to the major transcription start site of the mouse
B-myb
gene (
4
). The
cdc25C
constructs spanned the region from -290 to +121 (
10
). The promoter fragments were generated by PCR with compatible ends for cloning
into the pXP2 luciferase vector (
13
). Mutations were introduced by PCR strategies as previously described (
12
). All PCR-amplified fragments were verified by DNA sequencing using the
dideoxynucleotide chain-termination method (
14
) using Sequenase (USB). Ambiguous sequences and GC-rich stretches were verified by `cycle sequencing' using
Tth
polymerase (Pharmacia).
Nuclear extract (4 [mu]g) or MonoQ fractions (0.5 [mu]g) were incubated in 12 [mu]l of a buffer containing 50 mM Tris-HCl (pH 8.0), 10% v/v glycerol, 0.2 mM EDTA, 1 mM DTT, 0.8% sodium
deoxycholate, and 1 [mu]g poly(dA/dT) for 10 min. NP-40 was added to a final concentration of 1.5% and incubation was
continued for another 20 min.
32
P-labelled probe (0.2 pmol) was added and the reaction mixture was incubated
for another 20 min. All reactions were performed on ice. Probes were labelled
by filling-in 5' overhanging ends of 4-7 bases. Samples were run on 4% non-denaturing polyacrylamide gels in 0.5* TBE at 4oC and 10 V/cm. Gels were exposed to X-ray films and quantitatively
evaluated using a Molecular Dynamics PhosporImager. The following double-stranded probes were used:
B-myb
: 5'-GGCGCCGACGCA
CTTGGCGG
GAGAT
AGGAAGTTCTGTG
, E2F site and CHR underlined. Mutations are indicated in the corresponding
figures.
Cyclin A
: 5'-TCAATAG
CDE and CHR underlined.
Cdc25C
: 5'-ACTGGGC
CDE and CHR underlined.
The following antibodies were used: E2F-1 (Santa Cruz SC-251X), E2F-1/C (Santa Cruz SC-193X), E2F-2 (Santa Cruz SC-632X), E2F-3 (Santa Cruz SC-879X), E2F-4 (Santa Cruz SC-512X; also kindly
provided by R. Bernards, Amsterdam), E2F-5 (Santa Cruz SC-999X; also kindly provided by N. La Thangue, Glasgow), DP-1 (obtained from N. La Thangue), DP-2 (Santa Cruz SC-830X), DP-3 (kindly provided by N. La Thangue). DP-2 and DP-3 antibodies are directed
against homologous proteins (DP-2 is the human homologue of mouse DP-3;
15
,
16
).
Nuclear extracts were prepared from HeLa suspension cultures as described
including protease inhibitors leupeptin, pepstatin A and aprotinin but omitting
the dialysis step at the end of the procedure (
17
). Extract was diluted 10-fold with buffer A [50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 5% (v/v) glycerol, 0.5 mM DTT], centrifuged
10 min in a TLA 45 rotor at 45 000 r.p.m. and 4oC. The sample was loaded on a 1 ml Mono Q column equilibrated with buffer A
and run with a flow rate of 1 ml/min at room temperature. Protein was eluted
with a gradient of up to 1 M KCl in buffer A and monitored at 280 nm. Fractions
of 1 ml were collected as soon as proteins appeared in the eluate.
For
in vitro
DMS footprinting the coding strand oligonucleotide was end-labelled, purified and annealed to the non-coding strand. Binding reactions were carried out as described
above. Two microliters of 2% DMS was added, and the methylation reaction was
stopped 3 min later by adding 2 [mu]l of 60 mM [beta]-mercaptoethanol. The samples were run on a 4% gel and
transferred to ion-exchange paper. Both the shifted and the unshifted (free probe) bands were cut
out, rinsed with TE buffer and eluted with TE buffer containing 1.5 M NaCl at
65oC. The eluted DNA was extracted with chloroform, precipitated and dissolved
in water. Equal radioactive amounts of free probe and shifted complex were
cleaved with 10% piperidine at 95oC for 30 min. The DNA was precipitated and loaded on a 15% denaturing
acrylamide gel.
The alignment of the proximal B-
myb
promoter with other cell cycle-regulated genes showed that the sequence 5 nt downstream of the E2F site
(TAGGAA) closely resembles the CHR in
cyclin A
,
cdc2
and
cdc25C
(T/CTTGAA) found 5 nt downstream of the CDE in the latter genes (Fig.
1
). This observation raised the possibility that an element similar to the CHR
might also be involved in B-
myb
regulation and thus play a role in E2F-mediated transcriptional repression. We therefore investigated the role of
the putative
B-myb
CHR in a functional assay by analysing the effect of a CHR mutation on the
repression of the
B-myb
promoter in quiescent NIH3T3 cells. As shown in Figure
2
, mutation of the CHR led to an ~10-fold increased activity in G
0
cells, and thus had an even stronger effect than the mutation of the E2F site (~6-fold increase in G
0
) previously described to be involved in
B-myb
repression (
4
). Both mutations had only small effects in normally cycling cells (~2-fold; data not shown). These data demonstrate that the E2F site and a
downstream located element resembling a CHR cooperate in the repression of the
B-myb
promoter in quiescent cells.
In order to investigate the role of the CHR in further detail, we first asked
which proteins interact with the
B-myb
E2F site and whether such interactions might be dependent on the presence of an
intact CHR. The data in Figure
3
show that four specific complexes could be identified with HeLa nuclear extract
(see labelling at the left margin). The formation of these complexes was
totally abolished by a mutation in the E2F site, but not affected by the CHR
mutation. All complexes contained DP-1 as shown by the complete supershift caused by a DP-1 specific antibody, while DP-2 (the human homologue of mouse DP-3;
15
,
16
) could not be detected in any of the complexes. The slowest migrating complex
was specifically extinct by an antibody against E2F-4 while the slightly faster migrating band represented two complexes
containing E2F-1 or E2F-3. This is indicated by the fact that both the E2F-1 and the E2F-3 specific antibody alone led only to a partial
extinction of this band, while the combination of both led to complete
extinction. Furthermore, an antibody against the C-terminus of E2F-1 but cross-reacting with E2F-3 ([alpha]-E2F-1/C) also completely abrogated
formation of this band. A fourth minor complex of faster mobility was
identified, but none of the antibodies directed against the five known E2F
family members affected this complex. Since this complex was extinct by the [alpha]-E2F-1/C antibody, it is likely that it also contains an E2F
protein, either a novel family member or an unidentified variant of the known
E2F proteins. Taken together, our results indicate that the
B-myb
promoter E2F site interacts mainly with E2F-1/DP-1, E2F-3/DP-1 and E2F-4/DP-1 complexes, and that the formation of these
complexes occurs independently of the CHR.
We next sought to obtain direct experimental evidence that the
B-myb
CHR indeed represents a protein binding site, and to investigate whether the
adjacent E2F site might play a role in such interactions. To address this
question we attempted to identify a
B-myb
CHR-binding activity in nuclear or whole cell extracts from different cell
lines, but all attempts invariably failed (data not shown; see also Fig.
3
). It is a well known fact that certain transcription factors are detectable by
EMSA only after enrichment or partial purification from nuclear extracts, which
prompted us to analyse fractions of HeLa nuclear extract obtained after MONO-Q FPLC. This attempt proved successful: fractions 8 and 9 (see Materials
and Methods) contained an activity that bound to the
B-myb
promoter probe in a CHR-dependent, but E2F site-independent manner (Fig.
4
; three left-most lanes). In addition,
in vitro
methylation protection footprinting of this activity showed a clear protection
of two adenine residues within the
B-myb
CHR (Fig.
5
), and hypermethylation of a third adenine located immediately downstream. This
altered reactivity of the N3 position in adenine residues clearly indicates
minor groove protein interaction. This observation is therefore in perfect
agreement with previous
in vivo
experiments demonstrating minor groove protection of the
cyclin A
CHR (
12
). In contrast, as expected, no protection of the E2F site was observed.
We also tested the potential interaction of the binding activity identified
above with the
cdc25C
and
cyclin A
promoter by using appropriate promoter fragments at different molar ratios as
competitors. The results presented in Figure
3
clearly show that the highest affinity was seen with the
B-myb
probe.
cdc25C
and
cyclin A
were also able to compete, but only at higher concentrations, which is clearly
seen at a molar ratio of probe over competitor of 1:20. In contrast, an
unrelated oligonucleotide competitor (NIP;
18
) had no effect on complex formation.
The results of the competition experiment described above suggest that the
factor interacting with the
B-myb
CHR may be different from those binding to the
cdc25C
and
cyclin A
promoters. To test this hypothesis by a functional approach we constructed a
cdc25C promoter molecule harbouring the
B-myb
CHR (Cdc25C-BmybCHR; Fig.
6
). This construct showed an ~8-fold increase in activity when tested in quiescent NIH3T3 cells as
compared with the wild-type cdc25C promoter (Fig.
3
), and thus had a very similar effect as the replacement of the CHR or the CDE
with an irrelevant sequence (data not shown;
10
,
12
). In contrast, the increase in luciferase activity in normally growing cells
was only ~2-fold (data not shown), indicating selective deregulation of cell
cycle-regulated transcription in G
0
cells. Based on these results we conclude that the
cdc25C
and
B-myb
CHRs are functionally not equivalent.
Even though the function of E2F as a transcriptional activator is now well
established, there is a growing body of evidence pointing to a crucial role for
E2F complexes in cell cycle-regulated transcriptional repression (for a review see
3
). It is generally believed that E2F-mediated repression is a consequence of the association of E2F/DP
heterodimers with pocket proteins (pRB, p107, p130) (
19
-
28
). This association not only blocks the activation function of E2F but also
converts it to an active DNA-bound repressor. The pocket protein component is thought to establish
physical contacts with other transcription factors bound to the promoter to be
repressed, such as upstream activators, thereby blocking their function in
establishing an active transcription complex. At least for pRB there is
experimental evidence supporting this hypothesis (
29
,
30
), although other mechanisms may also apply (
31
). While it is clear from studies with artificial promoters that E2F-binding suffices to activate transcription, nothing is known about the
sequences or elements required for transcriptional repression.
In the present study, we have used the
B-myb
promoter to address this question. This investigation was fostered by our
observation that the region immediately downstream of the E2F site resembles a
similarly located element previously shown to play an essential role in the CDE-mediated repression of
cyclin A
,
cdc2
and
cdc25C
(Fig.
1
). As shown by the mutational analysis in Figure
2
, our notion that this element, referred to as
B-myb
CHR, is functionally relevant was fully confirmed. Destruction of the
B-myb
CHR leads to deregulation in quiescent cells, as does the mutation of the E2F
site itself. The
B-myb
CHR thus represents the first element identified to date that synergises with
E2F in the establishment of transcriptional repression.
Protein binding studies showed that both the E2F site and the
B-myb
CHR are able to bind specific nuclear proteins, and that these interactions
occur in a mutually independent fashion. Thus, E2F-1/DP-1, E2F-3/DP-1 and E2F-4/DP-1 major groove complexes are formed with the
B-myb
E2F site in the absence of an intact CHR, and a nuclear factor recognising the
CHR in the minor groove was not dependent on the E2F site for DNA-binding. It thus appears that the cooperation of the two elements must
occur at a level other than DNA-binding. It is possible that both factors synergise in the establishment
of appropriate contacts with other promoter-bound transcription factors, perhaps by inducing a favourable DNA topology
(as is often seen with minor groove-binding proteins; see e.g.,
32
). The answer to this question certainly has to await the purification, cloning
and functional analysis of the CHR-interacting factor(s). The identification of such a factor in the present
study, as shown in the experiments in Figures
4
and
5
, clearly represents an important step in this direction.
The competition data in Figure
4
, taken together with the functional analysis in Figure
6
, suggest that different factors interact with the CHRs in
B-myb
and
cdc25C
. In agreement with these results is the observation that the
cdc25C
CDE shows no interaction with E2F or DP family members, neither in nuclear
extracts nor with recombinant proteins (N. Liu and K. Engeland, unpublished
observation), while the
B-myb
E2F site does (
9
; Fig.
3
). These data clearly suggest the formation of promoter-specific repressor complexes of E2F and CDF with different CHR-binding activities, and that it is the precise composition of these
complexes that determines the timing of expression. This hypothesis is
supported by our observation that the exchange of the region encompassing the
E2F site and CHR in
B-myb
with the CDE-CHR module from
cdc25C
leads to a late induction of the
B-myb
promoter, similar to that of the wild-type cdc25C promoter (J. Zwicker and F.C. Lucibello, unpublished
observation). Once CDF has been identified and its cDNA cloned, the questions
relating to the mechanisms involved in the formation of promoter-specific complexes and their function in cell cycle-regulated repression can be addressed in detail, and these studies
can be expected to unravel new mechanisms orchestrating the periodic expression
of genes.
We are grateful to Dr R. Watson for a
B-myb
promoter plasmid and useful discussions, to Drs N. La Tangue and R. Bernards
for antibodies, to Dr R. Lührmann and F. Seifart for spinner cultures of HeLa cells and to Dr M.
Krause for synthesis of oligonucleotides. This work was supported by the DFG,
the BMBF and the Dr Mildred Scheel Stiftung. J.Z. was supported by a fellowship
from the Boehringer Ingelheim Fonds.
REFERENCES
Return