ABSTRACT
Oct-3, a member of the POU family of transcription factors, is expressed in
pluripotent cells of early mammalian embryos and in undifferentiated embryonal carcinoma cell lines. Using a variety of Oct-3 mutants, we have identified two different domains of Oct-3 which activate transcription in transfected mammalian cells. One
of these domains, located in the C-terminal part of the protein, plays a major role in transcriptional
activation when Oct-3 is bound to its cognate site, the octamer motif. An Oct-3 mutant containing a single amino acid substitution in the POU
homeodomain is unable to bind the octamer target
in vitro
, yet is still able to activate transcription in an octamer-dependent manner. We provide evidence that transactivation by this mutant
involves protein-protein interactions with the ubiquitous octamer binding factor Oct-1. This interaction requires the POU-specific domain of Oct-3 and allows recruitment of Oct-3 to the target promoter even in the absence of
Oct-3 DNA binding.
The octamer
cis
-acting transcriptional regulatory motif (ATGCAAAT) is found in enhancers
and promoters of many genes which are expressed either ubiquitously or in a
tissue-specific fashion (
1
-
7
). The octamer motif regulates gene expression by the binding of transcription
factors belonging to the POU family (
8
-
10
). These factors are characterized by the presence of the conserved 160 amino
acid POU domain, a bipartite DNA binding structure containing a 74-82 amino acid N-terminal POU-specific region (POU
S
) and a 60 amino acid POU homeodomain (POU
HD
), connected by a 15-27 amino acid linker region (
9
-
12
). The POU
HD
shares ~33% amino acid identity with the homeodomains of the Antennapedia class of
homeoproteins. Furthermore, its tertiary structure, as determined by X-ray crystallography (
13
), is very similar to the structures of other homeodomains. DNA binding by the
POU domain involves interaction of the third `recognition' helix of the POU
HD
with bases in the major groove. The POU
S
domain is required for site-specific, high affinity DNA binding and bending (
14
) and contacts DNA in the 5'-half of the octamer motif using a helix-turn-helix structure similar to that of the [lambda] and 434 repressors (
15
-
17
).
In addition to their DNA binding function, both the POU
S
and the POU
HD
can participate in protein-protein interactions with both POU proteins and other transcriptional
regulators. (
9
,
10
,
18
-
22
) Homo- and heterodimerization, mediated by the POU domain (
20
), has been demonstrated for several POU proteins binding to multiple adjacent
sites. For example, POU domain interactions promote the cooperative binding of
Pit-1 and Oct-1 to the Pit-1-responsive element of the prolactin promoter (
23
). Protein-protein interactions mediated by the POU domain are important for both
activation and repression of transcription. For example, Oct-1 is able to bind to the herpes simplex virus transactivator VP16 through
specific residues of the POU
HD
, leading to activation of the viral immediate early genes (
24
-
29
), whereas specific interaction between helix 1 and 2 of the POU
HD
of the
Drosophila
proteins I-POU and Cf1a leads to specific inhibition of transactivation of the DOPA
decarboxylase gene (
30
,
31
).
The expression of most POU proteins is regulated during embryogenesis, as
revealed by
in situ
hybridization studies, which reflects their critical roles in development and
cell type determination (
11
,
12
,
32
,
33
). Expression of one of these genes,
Oct
-3 (also termed Oct-4) is restricted to the pluripotent cells of the blastocyst during
early stages of embryonic development and it is confined to the primordial germ
cells after gastrulation (
34
-
37
).
Oct
-3 is also expressed in embryonal stem cells and teratocarcinoma cell lines
and its expression is down-regulated after induction of differentiation with retinoic acid. This
suggests a role for
Oct
-3 in maintaining the undifferentiated state of multipotent embryonic cells
(
34
,
36
-
39
).
The Oct-3 protein can activate transcription from octamer-containing promoters (
34
,
35
,
37
,
39
). By fusion of Oct-3 domains with heterologous DNA binding domains, a transactivator domain
has been mapped within the proline-rich N-terminal region (
39
,
40
). In this report, we show that in the context of octamer-mediated transcriptional activation, the strongest activation domain of
Oct-3 actually lies within the C-terminal part of the protein, while the N-terminus has a much weaker activity. Furthermore, we present evidence that protein-protein interactions mediated by the POU
S
domain with at least another POU protein, the ubiquitous factor Oct-1, can be functionally relevant for the octamer-specific transcriptional activating function of Oct-3.
F9, HeLa and NIH 3T3 cells were maintained in Dulbecco's modified Eagle's medium
(DMEM) supplemented with 10% fetal calf serum and L-glutamine, in a 5% CO
2
humidified atmosphere at 37oC. For transfection, cells were seeded in 100 mm tissue culture dishes, allowed to reach 1/3-1/2 confluency and transfected using the calcium phosphate precipitation method
with 20 [mu]g total DNA (2 [mu]g reporter plasmid, 0.13 [mu]g reference plasmid, 0.5-4.0 [mu]g transactivator expression plasmid and with pUC19
plasmid added to bring the total DNA to 20 [mu]g), according to standard procedures (
41
). The precipitate was removed 15-20 h after transfection by extensive washing with PBS and PBS containing
2 mM EGTA. Two days after transfection, the cells were harvested for
cytoplasmic RNA or protein extraction.
Expression vectors and mutants.
The PCG-Oct-3 plasmid was made by inserting the
Nco
I-
Bcl
I fragment of the
Oct
-3 cDNA into the PCG mammalian expression vector (
34
,
42
). Mutations in the
Oct
-3 coding sequence were generated either by site-directed mutagenesis (
43
) or by PCR and always verified by Sanger DNA sequencing (Sequenase; US Biochemical).The 5'P- plasmid had a deletion in the N-terminal, proline-rich region from Pro13 to Pro60. The 3'P- construct had a deletion in the C-terminal region from Gln287 to
Ser320. The POU construct contained the entire POU domain of Oct-3, from Glu127 to Ser272. The `frameshift' construct, which destroyed the
Oct
-3 open reading frame at 91 bp from the starting ATG, was generated by
digestion of PCG-Oct3 with
Bam
HI, blunting with Klenow polymerase and religation. The Gal4-Oct-3-5' fusion plasmid contains the 5'
Eco
RI-
Pst
I
Oct
-3 fragment inserted at the
Sma
I site of the Gal4 1-147 expression plasmid (
44
). The Gal4-E1a expression plasmid has been described previously (
45
).
Reporter and reference plasmids.
The reporter plasmid containing wild-type (B20 dpm2) and the reporter plasmid containing mutant (B20 dpm8)
octamer motifs have been previously described (
42
). In these plasmids, six copies of the synthetic SV40 B element, containing a
wild-type or mutant octamer motif, were cloned upstream of the human [beta]-globin gene. The internal reference plasmid contains four
tandem copies of the A and C SV40 enhancer elements inserted downstream of the [alpha]-globin gene (
42
). The Gal4*1-E1b and Gal4*5-E1b reporter plasmids were described previously (
45
) and contained one or five Gal4 binding sites, respectively. A RSV-[beta]-gal expression vector was used for normalization of
transfection efficiency in some experiments.
Cytoplasmic RNA was extracted from transfected cells following NP-40 lysis (
41
). Fifteen or 20 [mu]g RNA were hybridized overnight at 45oC to antisense [alpha]- and [beta]-globin riboprobes (5 * 10
5
c.p.m.), digested with an RNase A and T1 mixture and run through a denaturing
5% polyacrylamide gel (
41
). The correctly initiated [alpha]- and [beta]-globin transcripts gave rise to protected bands of 132
and 350 bp, respectively. Every experiment was performed at least in triplicate
and the results quantitatively analysed using a PhosphorImager (Molecular
Dynamics).
Total cell protein extracts were prepared from transfected cells as described (
41
). Electrophoretic mobility shift assays were carried out as described (
46
), using a gel-purified, end-labelled double-stranded oligonucleotide probe derived from the Ig-[kappa] promoter and 3 [mu]g cell extract.
For Western blot analysis, 30 [mu]g transfected cell extract were electrophoresed through a 10% SDS-polyacrylamide gel, electroblotted onto a nitrocellulose membrane, incubated with an anti-Oct-3 rabbit polyclonal antiserum raised against a
bacterially expressed protein and visualized with
125
I-labelled protein A (
41
).
The POU-specific region of
Oct
-1, generated by PCR, was cloned in-frame into the
Xma
I site of the pGex 2T vector (Pharmacia). The glutathione S-transferase fusion protein (GST-Oct-1 POU
S
) and GST alone were expressed in
Escherichia coli
according to established methods (
41
). Briefly, fresh cultures of bacteria containing either pGex 2T or the fusion
construct pGex-Oct1 POU
S
were induced for 3 h with 0.5 mM IPTG. Cells were harvested by centrifugation,
resuspended in MTPBS (150 mM NaCl, 16 mM Na
2
HPO
4
, 4 mM NaH
2
PO
4
, pH 7.3) containing 1% Triton X-100 and 1 mM PMSF and sonicated for 20 s (low energy, Branson Sonifier).
The supernatants were incubated for 10 min at 4oC with glutathione-Sepharose 4B resin (Pharmacia) on a rotating wheel. After three
washes with MTPBS, the resin, bearing approximately equal amounts of either GST
or GST-Oct1 POU
S
(>90% pure as determined by Coomassie staining on an SDS-polyacrylamide gel), was incubated for 1 h at 4oC on a rotating wheel with
in vitro
translated,
35
S-labelled Oct-3 protein in 200 [mu]l HND binding buffer (20 mM HEPES, pH 7.4, 50 mM NaCl, 0.1% NP-40, 5 mM DTT, 10 mg/ml BSA) in the presence of 200 [mu]g/ml ethidium bromide. The Oct-3 protein was generated by
in vitro
transcription, using T7 RNA polymerase, and subsequent
in vitro
translation, using rabbit reticulocyte lysates, according to the manufacturer's
procedure (Promega). After four washes of 5 min each at 4oC with MTPBS containing 0.1% NP-40, the resin was resuspended in SDS-PAGE sample buffer and electrophoresed through a 10% SDS-PAGE gel. The gel was treated for fluorography with
Entensify*A and Entensify*B solutions (DuPont), dried and exposed to Kodak X-AR films at -70oC with an intensifying screen.
To characterize the domains of the Oct-3 protein required for octamer-dependent transcriptional activation, expression constructs for full-length Oct-3 and various deletion mutants of Oct-3 were co-transfected into HeLa cells with a reporter
plasmid containing six copies of the octamer motif upstream of the [beta]-globin promoter/gene. The transcriptional activity of the various
Oct-3 mutants was analysed using an RNase protection assay of mRNA from
transfected cells. As shown in Figure
1
A, removal of the N-terminal, proline-rich region (5'P-) had no significant effect on the transactivation
activity of the protein (Fig.
1
A, lanes 5 and 6) when compared with the activity of wild-type Oct-3 (Fig.
1
A, lanes 3 and 4). Conversely, deletion of the C-terminal serine/threonine/proline-rich region (3'P-) caused a dramatic reduction in the transcriptional
activity (Fig.
1
A, lanes 7 and 8). In the same assay, the POU domain alone (Fig.
1
A, lanes 9 and 10) was totally inactive, as was a control frameshift mutant of
Oct-3 (Fig.
1
A, lanes 1 and 2) truncated after the first 30 amino acids. These results
indicate that most of the transcriptional stimulatory activity of Oct-3 on an octamer-containing target is located in the C-terminal part of the protein.
To test whether DNA binding by Oct-3 was required for its transactivation of an octamer-containing target, we inactivated Oct-3 DNA binding by changing a single amino acid (Val235 -> Pro) in the recognition helix of the POU
HD
(V-P construct). A similar mutation was previously shown to abolish the
binding of Pit-1 to its target sequence (
47
). As expected, the mutant failed to bind the octamer DNA probe (Fig.
2
A, lane 8), although the protein was efficiently produced in transfected cells
(Fig.
2
B, lane 6). Surprisingly, co-transfection of the V-P mutant with the octamer reporter plasmid resulted in a significant
activation of transcription (Fig.
2
C, lane 7), ~50% of that observed with the wild-type Oct-3 protein (Fig.
2
C, compare lanes 4 and 7). The activity of the V-P mutant was nevertheless dependent on the presence of an intact octamer
motif in the reporter construct, since both the wild-type Oct-3 and the V-P mutant failed to activate expression of a construct
containing six copies of a mutated octamer element (Fig.
2
C, lanes 5 and 8) or a promoter lacking an octamer motif (Fig.
2
C, lanes 6 and 9). None of these reporter plasmids were transactivated by the
frameshift construct (Fig.
2
C, lanes 1-3).
The above findings suggested the existence of additional factors in the
transfected cells which were able to direct octamer-dependent transcriptional activity from an Oct-3 mutant unable to bind the octamer sequence.
The above transactivation assays were performed in HeLa cells, which contain Oct-1 as the sole octamer binding protein. Oct-1, by itself, is unable to transactivate the octamer reporter
construct used in these assays (Fig.
4
C, lane 1;
42
). However, we considered the possibility that Oct-1 might interact with the transfected Oct-3 and target it to the reporter construct even in the absence of Oct-3 DNA binding activity. To provide
in vitro
evidence for protein-protein interactions between Oct-1 and Oct-3, the POU
S
domain of Oct-1 was expressed as a fusion protein with GST in bacteria. The fusion
protein was adsorbed to a glutathione-Sepharose resin and then incubated with
in vitro
translated, radiolabelled Oct-3 protein. The reactions were performed in the presence of ethidium
bromide, to exclude potential artifacts related to non-specific binding of the proteins to nucleic acids in the binding reactions
(
9
,
48
). The proteins retained by the GST-Oct-1 POU
S
resin were analyzed by SDS-PAGE. Figure
3
shows that ~30% of the labelled Oct-3 specifically associated with the fusion protein, while no binding
was observed to the GST control protein (Fig.
3
, lanes 2 and 3).
To provide
in vivo
functional evidence of an interaction between Oct-3 and Oct-1, we tested whether the activity of the Oct-3 V-P mutant protein could be blocked by co-expression of either the Oct-1 POU
S
domain or the Oct-3 POU
S
domain. As shown in Figure
4
A, expression of the isolated Oct-3 POU
S
domain in transfected HeLa cells completely abolished the ability of the Oct-3 V-P protein to activate transcription from the octamer reporter
construct. A similar blocking effect was observed when the Oct-1 POU
S
domain was expressed, but not when the Oct-3 frameshift mutant was tested (Fig.
4
B, lanes 3 and 4). The ability of an isolated Oct-1 and Oct-3 POU
S
domain to interfere with the transcriptional activity of the Oct-3 V-P protein was consistent with the hypothesis that the Oct-3 V-P protein and the endogenous Oct-1 protein interacted through their respective POU
S
domains.
These experiments suggested that the Oct-3 POU
s
domain was necessary for targeting of the Oct-3 V-P protein to the reporter construct. To provide further support for
this notion, we constructed a variant form of Oct-3 V-P in which the POU
S
domain was deleted, but the POU
HD
was retained. This protein (Oct-3 V-P POU
S
-) was unable to activate transcription through the octamer motif (Fig.
4
C, lane 3), again pointing to a critical role for the Oct-3 POU
S
domain in this phenomenon.
We were next interested in whether the Oct-3 POU
S
domain was sufficient for this targeting. To test this notion, we constructed a
plasmid which expressed a fusion protein between the Oct-3 POU
S
domain and the acidic transcriptional transactivator domain of the herpes
simplex virus protein VP16 (Oct-3 POU
S
A.A.). Transfection of HeLa cells with this construct together with the octamer
reporter plasmid revealed that the Oct-3 POU
S
A.A. chimeric protein was able to activate transcription (Fig.
5
, lane 1) at a level comparable with, if not higher than the transactivation
observed for wild-type Oct-3 (Fig.
5
, compare lanes 1 and 3). Furthermore, the transactivation by Oct-3 POU
S
A.A. was octamer specific, since no activity was observed when using a reporter
plasmid that contained mutant octamer motifs (Fig.
5
, lane 2). Transfection of the Oct-3 POU
S
domain alone did not transactivate the reporter gene (data not shown). Since
the POU
HD
is required for high affinity binding of the POU domain to the octamer motif (
9
-
11
,
15
), the Oct-3 POU
S
A.A. protein would not be expected to bind alone to the reporter plasmid.
Rather, the most likely interpretation of this experiment is that the Oct-3 POU
S
domain of the Oct-3 POU
S
A.A. fusion protein interacted with endogenous Oct-1 in the transfected cells, which was able to tether the fusion protein to
the octamer reporter plasmid.
Oct-3 has been shown to be a potent transcription factor when tested on octamer-containing target sequences (
34
-
37
,
49
). Previous analyses of Oct-3 functional domains outside the POU region suggested that the N-terminal region of the protein, which is rich in proline residues,
is responsible for its transcriptional activating function (
39
,
40
). This conclusion was drawn from experiments in which the Oct-3 proline-rich region was fused to a heterologous DNA binding domain derived
from the c-Jun protein and the resulting chimera was able to activate transcription
of a reporter construct containing c-Jun binding sites. We show here that the same N-terminal region of Oct-3 fused to the Gal4 DNA binding domain is able to activate
transcription from a reporter gene driven by multimerized Gal4 binding sites,
albeit at a low level when compared with the activator domain of the E1a
protein. These results confirm the potential activating function of the proline-rich domain of Oct-3 when assayed in the context of a heterologous DNA binding domain.
Interestingly, however, we obtained a substantially different result when we
analysed the transcriptional activity of Oct-3 when bound to an octamer motif through its own POU DNA binding domain.
In this assay, the proline-rich N-terminal domain was not required for strong transcriptional
activation by Oct-3, whereas a deletion of the C-terminal proline/serine/threonine-rich region severely reduced transcriptional activation. This
domain was previously reported to be transcriptionally inactive when fused to a
heterologous DNA binding domain (
40
). Thus, by analysing the domains of Oct-3 within their natural protein context, we have revealed an unanticipated
second activation domain which is responsible for most of the transcription
activity of the wild-type Oct-3 protein bound to an octamer DNA motif.
We wish to thank W.Herr, M.Tanaka and T.Kadesh for kindly providing expression
and reporter plasmids. We also thank F.Mavilio and V.Zappavigna for very
helpful discussions.
REFERENCES
Return