 |
INTRODUCTION |
Generating ligand-regulated activators or repressors targeted to
DNA sequences in any gene of interest represents a challenging long-term goal of protein engineering. The model systems we use to
approach this objective are based on estrogen-regulated genes. The
effects of estrogen are mediated by the estrogen receptors ER
1 and ER
. ERs are
ligand-activated transcription regulators that are capable of high
affinity binding to a specific DNA sequence, termed the estrogen
response element (ERE). On binding to the ER, estrogens exert a wide
variety of biological effects, including effects on the development and
function of male and female reproductive tissues, bone remodeling, and
the cardiovascular system, and have been implicated in breast and
uterine cancer. Estrogen-regulated genes therefore represent important
therapeutic targets. If expression of estrogen-regulated genes could be
effectively suppressed, both the discovery and the elucidation of their
roles in various physiological processes would be greatly facilitated.
Selective estrogen receptor modulators (SERMs) (reviewed in Ref. 1) and
ER mutants displaying a dominant-negative phenotype (2) have been used
to suppress ER-induced transcription. However, SERMs can display
significant agonist activity in specific tissue or cell backgrounds (3, 4). Recently, a number of hER mutants displaying a dominant-negative phenotype have been described (2, 5). Although these hER mutants and
SERMs disrupt estrogen-induced transcription, they do not affect basal
transcription of estrogen-regulated genes. We therefore designed novel
hER variants for ligand-dependent repression of the
transcription of ERE-containing genes.
To create ligand-dependent repressors targeted to
ERE-containing genes, we constructed chimeras of ER and the KRAB
(Krüppel-associated box)
transcription repression domain (6-9) of the KOX1 protein (also named
ZNF10) (7, 8). The KRAB domain is a highly conserved 75-amino acid
region found in approximately one-third of the vertebrate Krüppel-like (Cys2-His2) zinc finger
proteins (6). When tethered to DNA, the KRAB domain suppresses
transcription activation mediated by a variety of transcription factors
(7, 9-12), represses transcription mediated by all three classes of
eukaryotic RNA polymerase (10-12), and functions as a repressor even
when bound at DNA sites up to 3 kilobases from the transcription
initiation site (10, 11, 13, 14).
In this study, we characterize and examine mechanistically the ability
of ER-KRAB domain chimeras to suppress transcription of synthetic genes
containing the consensus ERE or the imperfect ERE from the natural pS2
promoter (15). Although the ER-KRAB chimeras were found to exhibit
efficient estrogen- or antiestrogen-dependent repression of
promoters containing the consensus ERE in several cell and promoter
contexts, they were unable to repress transcription from the imperfect
ERE found in the pS2 promoter. To achieve repression from a promoter
containing the native pS2 ERE, we developed a novel repressor with
increased affinity for this imperfect ERE. We recently described the
use of a modified p22 challenge phage system to select mutant steroid
receptor DNA-binding domains with altered DNA binding specificity and
an enhanced affinity for EREs (16). By integrating information obtained
from those genetically selected mutant DNA-binding modules with the
ligand-regulated ER-KRAB chimeras, we produced a prototype of a new
class of targeted gene repressor. This novel ER-KRAB chimera (KERK-3M)
is a potent repressor of both basal and estrogen-induced activities of
genes containing the consensus ERE or the imperfect pS2 ERE.
| |
EXPERIMENTAL PROCEDURES |
Cloning of hER-KRAB Chimeras--
To fuse the Kox1 KRAB
domain to hER, unique NheI sites were introduced into the
hER cDNA sequence. To facilitate sequence verification after
mutagenesis, the following hER fragments from plasmid pCMV5hER were
initially subcloned into pGEM11Zf(+) (Promega, Madison, WI): 1) the
EcoRI/NotI N-terminal fragment, 2) the
NotI/HindIII fragment containing the LBD, and 3)
the HindIII/BamHI C-terminal fragment of pCMV5hER
and pCMV5hERL540Q (3, 17). QuikChange mutagenesis (Stratagene) was then
employed to introduce unique NheI sites into these plasmids,
generating the vectors pG11EnsNhe, pG11EnhNhe, pG11EbhNhe, and
pG11QbhNhe, respectively. To achieve this, the following primers were
used: for pG11EnsNhe, GCCCGCGGCCACGGACCGCTAGCAATGACCATGACCCTCCA (forward) and TGGAGGGTCATGGTCATTGCTAGCGGTCCGTGGCCGCGGGC (reverse); for
pG11EnhNhe, AAGTATGGCTATGGAGCTAGCCAAGGAGACTCGCTA (forward) and
TAGCGAGTCTCCTTGGCTAGCTCCATAGCCATACTT (reverse); and for pG11EbhNhe and
pG11QbhNhe, GAGGCAGAGGGTTTCCTGCTAGCTGCCACAGTCTGAG (forward) and
CTCAGACTGTGGCAGCTAGCAGGAAACCCTCTGCCTC (reverse).
The KOX1 cDNA (9) was a kind gift of Dr. Hans-Jürgen Thiesen
(University of Rostock, Rostock, Germany). Polymerase chain reaction
amplification by Taq DNA polymerase (Life Technology, Inc.)
generated fragments of the KOX1 (ZNF10) protein (amino acids 1-91)
containing both the KRAB A- and B-domains that could be cloned either
at the N terminus of hER and 
A/B-hER (N-KRAB) or at the C terminus
of hER and hER L540Q (C-KRAB). The following oligonucleotides were
used: N-KRAB, CAGAATTCATGGATGCTAAGTCACTAAC (forward) and
TATCTAGAAATGCAGTCTCTGAATCAG (reverse); and C-KRAB, CTTCTAGATATGGATGCTAAGTCACTAAC (forward) and
ATGGATCCTAAATGCAGTCTCTGAATCAG (reverse).
The resulting amplified products were subcloned into the pGEM-T vector
(Promega). After verifying the sequence, the N-KRAB insert was obtained
as an EcoRI/XbaI fragment and together with the
NheI/NotI fragment of plasmid pG11EnsNhe was
cloned into pCMV5hER, pCMV5hERL540Q, and pCMV5hERFS digested with
EcoRI/NotI or with the
NheI/HindIII fragment of pG11EnhNhe into
pCMV5hER, pCMV5hERL540Q, and pCMV5hERfs digested with
EcoRI/HindIII. These manipulations yielded
plasmids pCMV5KER, pCMV5KERQ, pCMV5KERFS, pCMV5K-A/B-ER, pCMV5K-A/B-ERQ, and pCMV5K-A/B-ERFS, respectively. The C-KRAB insert was obtained as an XbaI/BamHI fragment and
ligated into NheI/BamHI-digested plasmids
pG11EbhNhe and pG11QbhNhe, respectively. The resulting hER LBD-KRAB
fusions were then obtained as XbaI/BamHI fragments and cloned into similarly digested plasmids pCMV5hER, pCMV5-A/B-hER, pCMV5KER, and pCMV5K-A/B-ER. These manipulations yielded plasmids pCMV5ERK, pCMV5ERQK, pCMV5-A/B-ERK,
pCMV5-A/B-ERQK, pCMV5KERK, pCMV5KERQK, pCMV5K-A/B-ERK, and
pCMV5K-A/B-ERQK, respectively.
To establish that ERE binding is required for transcription repression
by the ER-KRAB chimera, its wild-type hER DNA-binding domain was
replaced through exchange of the respective
NotI/HindIII fragments with a mutated DNA-binding
domain. This latter DBD no longer recognizes the ERE sequence due to
the E203G, G204S, and A207V mutations in the DNA recognition helix (5,
17). To establish that a functional KRAB domain is required for
transcription repression, the previously reported E26A, E27A, and E28A
mutations (7) were introduced into the KRAB domain of the ER-KRAB
chimera with the QuikChange protocol using the following
oligonucleotides: GACTTCACCAGGGCGGCCGCGAAGCTGCTGGAC (forward) and
GTCCAGCAGCTTCGCGGCCGCCCTGGTGAAGTC (reverse).
A FLAG-GAL4-KRAB chimera was constructed to serve as a control. Dr.
C. M. Chiang (University of Illinois) provided us with a
FLAG-GAL4-VP16 fusion construct in the bacterial expression plasmid
pET11d (Novagen). We obtained the FLAG-GAL4-VP16 coding sequence by
digestion with NcoI and subsequent fill in with
Pfu polymerase followed by BamHI digestion to
liberate the insert. The gel-purified fragment was then ligated into
the mammalian expression vector pcDNA3 (Stratagene) to generate
plasmid pFGVP16. For this purpose, pcDNA3 was initially digested
with HindIII, filled in with Pfu polymerase, and
subsequently digested with BamHI. The GAL4 C terminus was
obtained in conjunction with a polylinker as a polymerase chain
reaction fragment from plasmid pM (CLONTECH),
changing the Dam methylation-sensitive BclI site into an
ApaI site in the process. The polymerase chain reaction fragment was digested with XhoI/ApaI and ligated
into similarly digested plasmid pFGVP16 to generate plasmid pFGmcs. In
our transfections, this plasmid is referred to as GAL4. The above
described N-terminal KRAB domain, obtained as an
EcoRI/BamHI fragment, was ligated into plasmid
pFGmcs, which provided the stop codon, generating the vector pFGK.
Plasmid (ERE)4-pGL3-SV40PE that we constructed served as an
indicator of repression. This plasmid is derived from plasmid pGL3-Control (Promega) and contains four consensus EREs upstream of the
SV40 promoter, which renders the plasmid estrogen-responsive. The SV40
promoter and enhancer in this plasmid constitutively drive the
expression of firefly luciferase; therefore, both activation and
repression can be studied effectively. The estrogen response elements
were obtained from plasmid (ERE)4-TATA-CAT (18), which was
digested with HindIII, blunt-ended with Pfu
polymerase, and religated to generate an NheI site. An
NheI/BglII digest was then performed to liberate
the EREs. This fragment was ligated into similarly digested vector
pGL3-Control. Another series of pGL3-Control-based reporters was
constructed containing one, two, and four EREs, respectively. To
achieve this, an extraneous BglII site was removed from the
multiple cloning site of plasmids pGL3-(ERE)1-TATA,
pGL3-(ERE)2-TATA, and pGL3-(ERE)4-TATA (16) by
HindIII/XhoI digestion and subsequent religation
after Pfu DNA polymerase-mediated fill in. Following this
treatment, the BglII/SalI backbone fragment
containing the respective number of EREs was ligated to the
BglII/SalI fragment of
BglII/PvuI/SalI-digested plasmid
pGL3-Control. To test the ER-KRAB chimeras in a non-SV40-based
promoter/enhancer context, plasmids (ERE)4-PGL3-TK and
(ERE)4-PGL3-EF1 were constructed. Plasmid pGL3-TK was
constructed by inserting the thymidine kinase promoter/enhancer as a
BglII/HindIII fragment obtained from plasmid pRL-TK (Promega) into similarly digested plasmid pGL3-Basic (Promega). Plasmid pGL3-EF-1 was constructed by inserting the elongation factor
1 promoter/enhancer obtained as a HindIII/NcoI
fragment from plasmid pEFmyc/nuc (Invitrogen) into similarly digested
plasmid pGL3-Basic. These plasmids were then made estrogen-responsive by incorporating four copies of the ERE obtained as an
NheI/BglII fragment from plasmid
pGL3-(ERE)4-TATA. To test the ability of ER-KRAB chimeras
to repress transcription from a single non-consensus ERE, a 345-base
pair SacI/SmaI fragment containing the pS2 ERE was isolated from the pS2 promoter and inserted into similarly digested
plasmid pGL3-Promoter, resulting in plasmid pGL3-pS2-SV40P. Plasmid
pGL3-(pS2 ERE)1-TATA is derived from the
pGL3-(ERE)1-TATA reporter by mutation of 2 base pairs in
the consensus ERE. The imperfect ERE created, 5'-AGGTCActgTGGCCC-3', is
the ERE in the pS2 5'-flanking region. For studies with the
FLAG-GAL4-KRAB fusions, the repression reporter plasmid
G5-pGL3-Control was constructed by inserting five
GAL4-binding sites obtained as an XhoI/BamHI fragment from plasmid pG5E1b (19) into
XhoI/BglII-digested plasmid pGL3-Control.
Cell Maintenance, Transfection, and Reporter Gene
Assays--
HepG2 human hepatoma cells and HeLa cells were maintained
in a humidified 5% CO2-containing environment at 37 °C
in Dulbecco's minimal essential medium (Sigma) supplemented with 10%
charcoal/dextran-stripped fetal bovine serum (Atlanta Biologicals,
Inc., Atlanta, GA) 50,000 units/liter penicillin, and 50 mg/liter
streptomycin (Life Technologies, Inc.). Chinese hamster ovary (CHO)
cells were maintained in Dulbecco's modified Eagle's medium/nutrient
mixture F-12 (1:1; Sigma), 29.2 mg/liter L-glutamine
(Sigma), 5% charcoal/dextran-stripped newborn bovine serum (Atlanta
Biologicals, Inc.), 50,000 IU/liter penicillin, and 50 mg/liter
streptomycin. MCF-7 cells were maintained in Eagle's minimal essential
medium plus phenol red supplemented with 5% newborn calf serum, 50,000 IU/liter penicillin, and 50 mg/liter streptomycin. At least 2 days
prior to the experiment, cells were transferred to 1:1 Dulbecco's
modified Eagle's medium/nutrient mixture F-12, 29.2 mg/liter
L-glutamine, 5% charcoal/dextran-stripped newborn bovine
serum, 50,000 IU/liter penicillin, and 50 mg/liter streptomycin.
Transient transfections were carried out by the calcium phosphate
coprecipitation method. Briefly, cells were plated in 60-mm dishes at a
density of 4.5 × 105 cells/dish for HepG2 cells and
2.5 × 105 cells/dish for CHO cells, in 6-well plates
at 1.0 × 105 cells/well, or in 12-well plates at
5 × 104 cells/well. The next day, the medium was
replaced; and 2-6 h later, calcium phosphate crystals were added.
12-16 h later, the cells were subjected to a 3-min shock with 20%
glycerol in Tris-buffered saline, pH 7.4. The medium was replaced; and
where appropriate, hormone was added to the indicated concentrations.
The cells were harvested 48 h later for the reporter gene assay by
addition of appropriate amounts of passive lysis buffer (Promega). The
activity of the resulting extracts was determined using the dual
luciferase assay protocol (Promega) according to the manufacturer's
directions on a Monolight 2010 luminometer.
| |
RESULTS |
hER-KRAB-mediated Repression Requires Ligand, EREs, and a
Functional KRAB Domain--
To produce the KRAB-hER-KRAB (KERK)
construct (see Fig. 2B), the complete KRAB repressor domain
(containing both the KRAB A- and B-domains) was placed at both the N
and C termini of hER. The ability of KERK to repress transcription
of a reporter gene containing the SV40 promoter and enhancer (SV40PE)
and four consensus EREs was tested. This (ERE)4-pGL3-SV40PE
reporter plasmid exhibits substantial intrinsic activity, referred to
as basal transcription, which is further enhanced by ligand-activated
ER. To establish the effect of ligand on the ability of a KRAB
construct to repress transcription, transient transfections were
carried out in ER-negative HepG2 human hepatoma cells in the presence
or absence of the estrogen moxestrol, which liver cells metabolize more
slowly than 17-estradiol (20). The basal promoter activity of the
(ERE)4-pGL3-SV40PE reporter plasmid in the absence of
estrogen receptor was set at 100%. Cotransfected hER expression
plasmid elicited a moxestrol-dependent 3-4-fold induction
of luciferase activity (Fig.
1A), whereas increasing amounts of unliganded ER did not affect transcription. In the absence
of an ER ligand and at 20 ng of transfected KERK expression plasmid,
there was a modest 1.6-fold repression of transcription. However, full
repression (4.8-fold) required the presence of ligand (Fig.
1A). Since KRAB repression was largely
ligand-dependent, subsequent studies were carried out in
the presence of ligand.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 1.
Repression by ERK is DNA sequence-specific
and requires ligand and a functional KRAB domain. A,
transcription repression properties of the KERK chimera and activation
properties of hER on the (ERE)4-pGL3-SV40PE reporter
plasmid in HepG2 cells in the absence and presence of 10 nM
moxestrol (Mox). All experiments were carried out in the
presence of 10 nM moxestrol, except where noted. Luciferase
activity from the transfected reporter was determined as described
under "Experimental Procedures." The activity of the reporter
plasmid alone was normalized to 100 kilo-luciferase units. To establish
whether both sequence-specific DNA binding and a functional KRAB domain
are required for repression by the ERK chimera, the effects on
transcription from the (G)5-pGL3-SV40PE and
(ERE)4-pGL3-SV40PE reporter plasmids in HepG2 cells were
examined by cotransfection of the indicated GAL4 DBD- and hER-based
effector constructs (B and C, respectively). The
data obtained were normalized against the luciferase activity of the
indicated reporter plasmid alone. The data in A-C represent
the mean ± S.E. of at least three independent transfections.
Enh, enhancer.
|
|
The sequence specificity of repression was shown by the inability of an
hER-KRAB (ERK) chimera (shown in Fig.
2, A and B) to
repress transcription from the five GAL4-binding sites in the G5-pGL3-SV40PE reporter (Fig. 1B) and by the
inability of GAL4-KRAB to repress transcription from the four EREs in
the (ERE)4-pGL3-SV40PE reporter (Fig. 1C). The
reporters were functional since GAL4-KRAB repressed transcription by
>90% from the G5-pGL3-SV40PE reporter (Fig.
1B), whereas hER activated basal transcription by 3.8-fold and ERK repressed transcription by 4.5-fold on the
(ERE)4-pGL3-SV40PE reporter (Fig. 1C). The issue
of DNA binding specificity was also addressed by introducing mutations
into the DNA recognition helix of the hER DBD that shift the
specificity from the ERE to the glucocorticoid response element and
thereby prevent binding to the ERE (5, 17). This chimera (ERKmutDBD) no
longer repressed transcription on either of the reporter plasmids (Fig.
1, B and C). As expected, introducing the
mutations E26A, E27A, and E28A into the KRAB domain (7) of ERK
(ERKmutKRAB) abolished repression (Fig. 1C).
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 2.
Influence of AF1 and AF2 on repression
properties of hER-KRAB chimeras. A, the KRAB domain was
fused in frame at either the N or C terminus and at both termini of
full-length wild-type hER, at the N terminus of several hER
mutants in which the ligand-independent activation function (AF1) was
removed through deletion of the A/B-domain (A/B) or in which the
ligand-dependent activation function (AF2) was ablated by
point mutations L540Q (Q) and S554fs (FS), or a
combination of these two classes of mutations. In the constructs, the
DBD is indicated as a shaded box, and the AF2 mutations in
the LBD are indicated as Q (L540Q) and FS
(S554fs), respectively. Ablation of AF1 activity, achieved through
deletion of the first 178 amino acids of hER, is indicated as
A/B. The KRAB repressor domain is indicated as a black
box. B, increasing amounts (5, 20, or 40 ng) of the
expression plasmids encoding the hER-KRAB chimeras were transfected
into HepG2 cells using the (ERE)4-pGL3-SV40PE plasmid as a
reporter. A vertical line in the ER LBD indicates the L540Q
point mutation (Q), whereas the striped box
extending the LBD C terminus represents the additional amino acid
sequence introduced by the S554fs frameshift mutation (FS).
The data obtained were normalized against the luciferase activity of
the reporter plasmid alone, which was set at 100%. The data in
B represent the mean ± S.E. of at least three
independent transfections.
|
|
Influence of Ligand and Estrogen Receptor AF1 and AF2 Mutations on
KRAB Repression--
Although the mechanism of transcription
repression by the KRAB domain is not fully understood, KRAB has been
shown to interact with the human corepressors TIF1 and TIF1 (also
isolated as KAP-1) and their murine homologue KRIP-1 (13, 21, 22).
Interestingly, TIF1 (23), but not TIF1, is thought to act as a
coactivator of steroid receptor-mediated transcription activation by
interacting with the AF2 region of ligand-occupied steroid receptors.
The interactions of TIF1 with the KRAB domain and with the AF2
region of steroid receptors take place via two distinct interaction
domains found within the TIF1 protein and might interfere with the
ability of the KRAB domain to function as a repressor in the presence of AF2. It was therefore of interest to examine whether presenting the
KRAB domain in different ways in the context of estrogen receptor chimeras would favor a functional interaction of KRAB and its corepressors, thereby enabling the KRAB domain to operate more effectively as a transcription repressor.
To analyze the effect of position and the influence of the ER
activation domains on KRAB repression, the KRAB domain was fused in
frame at either the N or C terminus and at both ends of hER (Fig. 2,
A and B). To prevent interaction with steroid
receptor coactivators, we also employed a number of hER mutants in
which AF1 and/or AF2 activity was ablated. Since the ligand-independent activation function AF1 is spread through much of the A/B-domain of
hER (24, 25), AF1 ablation was achieved by deleting the entire
A/B-domain (amino acids 1-178, indicated as A/B). Removal of the
ligand-dependent activation function AF2 was achieved
through introduction of either of two point mutations in the
ligand-binding domain, L540Q and S554fs (Q and FS, respectively) (Fig.
2A). These mutations confer a dominant-negative phenotype on
hER (2), which might further potentiate transcription repression by
the KRAB domain.
The ability of the ER-KRAB chimeras to repress transcription was
determined by cotransfecting the (ERE)4-pGL3-SV40PE
reporter plasmid and increasing amounts (5, 20, or 40 ng) of the
expression plasmid encoding each KRAB chimera into HepG2 cells in the
presence of 10 nM moxestrol (Fig. 2B). Even at
the lowest amount transfected, all of the chimeras achieved at least
45% repression, and most achieved >55% repression. The differences
in repression among the various constructs were modest. All of the
ER-KRAB chimeras are therefore effective transcription repressors.
Surprisingly, ablation of AF1 and/or AF2 activity had little or no
effect on the extent of KRAB repression. For example, at 40 ng of
transfected expression plasmid, the AF2-containing chimera KER
repressed transcription by 75%. Ablation of AF2 by the L540Q mutation
in the KERQ chimera or by the S554fs mutation in the KERFS chimera (2)
had little effect on the magnitude of transcription repression.
Deletion of AF1 modestly enhanced repression only when the KRAB domain was present at the C terminus of the protein. At 40 ng of transfected expression plasmid, the ERK and A/B-ERK constructs repressed transcription by 78 and 88%, respectively. The KERK, KERQK and A/B-ERK constructs were the most effective, with each repressing transcription by 87-88%. Since these differences were negligible, we
elected to use the KERK repressor in subsequent experiments.
KRAB-mediated Repression Is Not Blocked by Trichostatin A--
It
has been proposed that KRAB repression is mediated through recruitment
of the corepressors TIF1 and TIF1. These proteins contain RBCC
(RING finger-B boxes-coiled
coil), PHD finger, and bromodomain interaction domains.
Since these domains are also found in complexes implicated in
chromatin-mediated transcription repression, it has been suggested that
KRAB may act by modifying chromatin to achieve a repressive state (21,
26). Many chromatin modifiers recruit histone deacetylases or contain
intrinsic histone deacetylase activity. The histone deacetylase
inhibitor trichostatin A has been widely used to identify chromatin
events based on histone deacetylation (27, 28). Addition of 0.25 or 1 µM trichostatin A had no effect on the ability of the
KERK or GAL4-KRAB chimeras to repress transcription from several
reporter genes (Fig. 3). Although
trichostatin A failed to affect KRAB repression, it is functional in
HepG2 cells, as judged by its ability to strongly potentiate
moxestrol/ER-mediated transcription of a stably integrated vitellogenin
promoter in HepG2 cells.2
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 3.
Trichostatin A does not influence repression
by KRAB chimeras. To establish whether trichostatin A
(TsA) could relieve KRAB-mediated repression, we
cotransfected reporter plasmids (ERE)4-pGL3-SV40PE and
G5-pGL3-SV40PE and the indicated hER- and GAL4 DBD-based
effector constructs, respectively, in the absence (open
bars) or presence (0.25 µM, cross-hatched
bars; 1 µM, filled bars) of trichostatin
A. Moxestrol (10 nM) was present when hER or KERK was used.
Where appropriate, trichostatin was added 24 h prior to harvest of
the HepG2 cells. The data represent the mean ± S.E. of at least
three independent transfections.
|
|
Effect of Cell Line, Promoter, and Ligand on ER-KRAB
Repression--
We wanted to determine whether KRAB repression was
equally effective in different cell lines on strong and weak promoters and whether the KRAB chimera could repress transcription in the presence of wild-type ER or ER (29, 30). To examine the effect of
promoter strength on KRAB repression, repression was evaluated in
reporter genes containing the relatively weak thymidine kinase
promoter, the strong SV40 promoter/enhancer (SV40PE), and the extremely
powerful elongation factor 1 promoter. Repression in the presence of
endogenous ER was determined by cotransfecting plasmids encoding
wild-type ER or ER into the cells along with the KERK expression
plasmid. Even though we used three times more hER expression plasmid
than hER expression plasmid, in agreement with earlier studies (29,
31), hER was significantly less effective in activating
transcription than hER (Fig. 4,
A and D; 3.3-fold versus 15-fold in
Fig. 4A; note that the ordinate of A
is set on a logarithmic scale).
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 4.
KERK effectively represses transcription on
several estrogen-responsive promoters in HepG2 and CHO cells in the
presence and absence of hER. Repression was assessed in the
presence of 10 nM moxestrol in the presence and absence of
the indicated amounts of cotransfected hER or hER expression
plasmids using reporter plasmids (ERE)4-pGL3-TK,
(ERE)4-pGL3-SV40PE, and (ERE)4-pGL3-EF-1 in
HepG2 cells (A-C, respectively) and plasmids
(ERE)4-pGL3-TK, (ERE)4-pGL3-SV40PE, and
pGL3-EREVIT in CHO cells (D-F, respectively). The
transfections and luciferase assays were carried out as described under
"Experimental Procedures." The data represent the mean ± S.E.
of at least three independent transfections normalized to the activity
of the indicated reporter plasmid alone, which was set equal to
100%.
|
|
There was an inverse correlation between promoter strength and the
additional contribution to promoter activity due to hER-activated transcription. hER increased transcription 15-, 2.6-, and 0.9-fold on the (ERE)4-pGL3-TK, (ERE)4-pGL3-SV40, and
(ERE)4-pGL3-EF-1 reporter plasmids, respectively.
However, on all promoters, in both HepG2 cells (Fig. 4,
A-C) and CHO cells (Fig. 4, D-F), increasing amounts of transfected KERK repressed all, or nearly all, of the hER- or hER-induced activity and most of the basal promoter activity. In the absence of hER, KERK repressed up to 82-92% of basal
promoter activity on these reporter plasmids. When transfected at a
3-fold excess relative to hER, KERK repressed thymidine kinase
promoter activity to 45% of the basal thymidine kinase promoter
activity, which is a 33-fold reduction from the hER-induced level of
transcription (Fig. 4A).
In CHO cells, we tested the thymidine kinase, SV40, and
Xenopus vitellogenin B1 promoters using the
(ERE)4-pGL3-TK, (ERE)4-pGL3-SV40PE, and
pGL3-EREVIT reporter plasmids, respectively. These experiments suggested an interesting difference between transcription activation and repression. The EREVIT promoter contained only one consensus ERE,
two functional imperfect EREs, and one nonfunctional imperfect ERE
(32). The other test promoters contained four consensus EREs. In CHO
cells, hER activated transcription more powerfully from the EREVIT
promoter than from the other test promoters (3.4-fold versus
1.7-1.9-fold). However, transcription repression by the KRAB chimera
was more closely correlated with the number of consensus EREs, and
repression was somewhat more effective with the
(ERE)4-pGL3-TK and (ERE)4-pGL3-SV40PE reporters
than with the pGL3-EREVIT reporter. At a 1:1 ratio of transfected KERK
and hER, repression was clearly dominant, as activity was reduced
3.3-3.6-fold relative to the activity in the presence of hER alone
(Fig. 4, D-F). Similar results were obtained when
repression by KERK from the (ERE)4-pGL3-TK and
(ERE)4-pGL3-SV40PE reporter genes was evaluated in the
ER-negative breast cancer cell line MDA-MB231 and in HeLa cells (data
not shown).
To evaluate the ability of a KRAB chimera to repress transcription in
cells containing high levels of endogenous ER, we tested the
effectiveness of the KERK chimera in ER-positive MCF-7 human breast
cancer cells (Fig. 5). The ability of
SERMs to act as KERK ligands to potentiate KRAB repression was also
tested. SERMs, which are mixed agonists/antagonists such as
4-hydroxytamoxifen (OHT), prevent the ER ligand-binding domain from
adopting the conformation required for interaction with
AF2-dependent coactivators (33), but do not interfere with
DNA binding. "Pure" antiestrogens such as ICI 182,780 and RU 58,668 are thought to alter cytoplasmic-nuclear shuttling of hER and to
increase receptor degradation (34-36) and might be expected to impair
the ability of ER-KRAB chimeras to repress transcription. To facilitate
comparisons of the ability of the different ligands to induce
repression, we set luciferase activity in the absence of transfected
KERK equal to 100% for each ligand. Repression was not affected by the
type of ligand used. Transcription was repressed by 75-89% in the
presence of 17-estradiol, OHT, or ICI 182,780. Surprisingly,
repression was most effective when ICI 182,780 was present, indicating
that KERK·ICI 182,780 complexes are not rapidly degraded and
translocate into the nucleus and bind to ERE-containing DNA. OHT and
ICI 182,780 also elicited efficient repression as KERK ligands in the
estrogen receptor-negative HepG2 cell line (data not shown), indicating that repression was not due to the SERMs interfering with hER-mediated transcription activation.
View larger version (34K):
[in this window]
[in a new window]
|
Fig. 5.
Antiestrogens induce repression by KERK in
MCF-7 human breast cancer cells. Repression was assessed on the
(ERE)4-pGL3-SV40PE reporter plasmid in the presence of
17-estradiol (10 nM), OHT (10 nM), or ICI
182,780 (10 nM). Transfections and luciferase assays were
carried out as described under "Experimental Procedures." Since the
different effects of agonists and antagonists on the growth of MCF-7
cells influenced the activity of the internal standard, to facilitate
comparisons, the data obtained for each individual treatment group were
normalized against the luciferase activity of the reporter plasmid
alone in the absence of transfected chimera, which was set at 100%.
The average luciferase units for each treatment were as follows: no
ligand, 170,000; 17-estradiol, 39,000; ICI 182,780, 60,000; and OHT,
198,000. The data represent the mean ± S.E. of at least three
independent transfections.
|
|
Effect of the Number of EREs and ERE Binding Affinity on
Transcription Repression--
Virtually all studies employing KRAB
repressors have utilized conditions favorable to repression in which
the KRAB chimera binds to synthetic constructs containing multiple
copies of a perfect DNA-binding site (7-12, 14, 37, 38). Since KERK repressed expression from the EREVIT promoter (which contains a single
consensus ERE and three additional non-consensus EREs) less effectively
than it repressed promoters containing four consensus EREs (Fig. 4,
D-F), it was of interest to establish the minimum number of
consensus EREs required for repression. We therefore constructed
SV40-based reporter genes containing one, two, and four EREs and
examined the ability of transfected KERK to repress their transcription
(Fig. 6A). Repression was
similar for the reporter genes containing two or four EREs and reached
a plateau at 87%. Although repression from the reporter gene
containing a single ERE was dose-dependent, the inability
to reduce promoter activity below ~30% of basal activity, even at
high levels of transfected KERK, was troubling (Fig. 6B). We
therefore set out to enhance the potential of KERK to repress
transcription.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 6.
Repression from a single consensus ERE is
increased when ERE binding by the chimera is enhanced.
Transfections and luciferase assays were carried out as described under
"Experimental Procedures." In all cases, the data obtained were
normalized against the luciferase activity of the indicated reporter
plasmid alone, which was set at 100%. A, transcription
repression by the KERK chimera was assessed in HepG2 cells on
pGL3-SV40PE reporter plasmids containing the indicated number of EREs
(REs). B, repression by KERK and by a mutant KERK
possessing increased DNA binding (KERK-3M) was assessed in HepG2 cells
on the (ERE)1-pGL3-SV40PE plasmid. Note that the data for
the KERK chimera are also shown in A. Because of its
enhanced effectiveness as a repressor, 16 ng was the highest level of
KERK-3M tested. The data represent the mean ± S.E. of at least
three independent transfections.
|
|
Through the use of a modified form of the bacteriophage p22 challenge
phage selection system (39), our laboratory recently identified
progesterone receptor DNA-binding domain mutations that changed the DNA
binding specificity from the glucocorticoid response
element/progesterone receptor element to the ERE and that resulted in
enhanced binding to the consensus ERE and to the imperfect ERE in the
pS2 gene (16). One of the progesterone receptor DBD mutants we
isolated, DBD5, exhibited >10-fold higher affinity than the wild-type
ER DBD for the consensus and pS2 EREs. We reasoned that enhancing the
ability of KERK to bind to the ERE might potentiate its transcription
repression properties. Therefore, the corresponding three mutations
(E203W, Q214A, and H216G) from the progesterone receptor DBD5 mutant
were introduced into the DNA-binding domain of KERK, resulting in
KERK-3M. We compared the ability of KERK-3M and KERK to repress
transcription from the promoter containing a single ERE. KERK-3M was a
more potent repressor than KERK. Almost 2-fold less transfected KERK-3M was required to reach a given level of repression, and the extent of
repression by KERK-3M increased progressively at all of the amounts
tested (Fig. 6B).
KERK-3M, but Not KERK, Effectively Represses Transcription from a
Promoter Containing the Imperfect pS2 ERE--
Although the above
studies demonstrate that KERK and KERK-3M are able to repress
transcription from a single consensus ERE, most estrogen-regulated
genes contain imperfect EREs. To test repression from an ERE in a
native gene, we elected to use a fragment from the estrogen-inducible
pS2 gene that contains the single imperfect ERE
(5'-AGGTCActgTGGCCC-3') responsible for the strong estrogen
induction of pS2 gene expression. Although pS2 is a clinical and
prognostic marker for hormone-responsive breast cancer (40), the
function of pS2 and its role in breast cancer development and
progression remain poorly understood.
In vitro DNA binding and in vivo transactivation
by wild-type ER and by the ER DBD are both substantially reduced when
the non-consensus pS2 ERE is present rather than the consensus ERE (41). Since binding of the ER to an imperfect ERE is difficult to study
directly in intact cells, as a test of pS2 ERE-ER interaction, we
tested ER-mediated transactivation from a single pS2 ERE. We inserted
the three up-binding mutations used in the KERK-3M repressor (E203W,
Q214A, and H216G) into the DBD of wild-type hER (hER-3M) and
assessed the ability of the resulting hER-3M to activate transcription from the pS2 ERE. Relative to wild-type hER, 10 or 50 ng of transfected hER-3M increased transactivation from the pS2 promoter by 2.5- and
1.9-fold, respectively (n = 6; data not shown). This
supports the view that these mutations enhance in vivo
binding of the ER to the pS2 ERE.
To evaluate whether KERK and KERK-3M could repress transcription from
an imperfect ERE in a native promoter context, we constructed a
pS2-based reporter gene using the 345-nucleotide fragment from the pS2
promoter that contains the pS2 ERE (15). KERK only weakly repressed
moxestrol/hER-induced transcription of the pS2-based reporter and was
unable to repress basal transcription of the reporter (Fig.
7A). In striking contrast, the
KERK-3M chimera effectively repressed all of the moxestrol/hER-induced
transcription and elicited a strong dose-dependent
repression of basal promoter activity (Fig. 7B). These data
indicate that use of a genetically selected set of up-binding mutations
strongly potentiates the ability of ER-KRAB chimeras to repress
transcription from a naturally occurring imperfect ERE.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 7.
Transcription repression from a promoter
containing the pS2 ERE. The pGL3-pS2-SV40P reporter gene was
transfected into HepG2 cells in the presence or absence of
cotransfected hER and 10 nM moxestrol. The activity of the
reporter gene in the absence of any transfected repressor or hER was
set at 100%. A, repression by the indicated amounts of
transfected KERK expression plasmid; B, repression by the
KERK-3M plasmid. Since this experiment was carried out with cells
plated in smaller wells than in the study in Fig. 6, 1 ng of
transfected KRAB chimera expression plasmid in this study corresponds
to ~2.5 ng of transfected DNA in the study shown in Fig. 6. The data
represent the mean ± S.E. of at least three independent
transfections.
|
|
| |
DISCUSSION |
ER-KRAB Chimeras Containing ER Activation Domains Repress
Transcription--
In a study of repression of the human
immunodeficiency virus type 1 long terminal repeat, dominant-negative
Tat mutants linked to KRAB were far more effective repressors than
Tat-KRAB chimeras retaining an active Tat transactivation domain (42).
In a similar way, ER activation domains could interfere with KRAB
repressor activity since the putative KRAB corepressor TIF1 acts as
a coactivator on interaction with the AF2 domain of ligand-occupied ER
(23). Deleting or mutating one or both ER transactivation domains did not enhance repression of transcription, indicating that the KRAB domain is dominant over the ER transactivation domains and can overcome
the activity of any ER coactivators still able to bind the ER-KRAB chimeras.
KERK Represses Transcription when Wild-type ER Is Present--
If
the ER-KRAB chimeras and wild-type ER have similar affinities for the
ERE, it seemed plausible that wild-type ER could compete effectively
for binding to the EREs in our reporter genes and might block the
ability of the KRAB chimeras to repress transcription. Consistent with
our finding that the KRAB domain is dominant over the AF1 and AF2
domains, we found that KERK effectively represses transcription in the
presence of either hER or hER in several cell and promoter
contexts (Fig. 4).
Not only can KERK repress transcription in the presence of ER, it also
represses transcription of the powerful (ERE)4-pGL3-EF-1 reporter, whose expression is not up-regulated by the ER. In the progression of breast cancers to an estrogen-independent phenotype in
which antiestrogens no longer limit their growth, it has been suggested
that genes that were initially estrogen-regulated become constitutively
active (43-45). The (ERE)4-pGL3-EF-1 construct serves
as a prototype for this class of genes. KERK effectively suppresses the
high level of basal transcription from this promoter (Fig.
4C).
An ER Ligand Is Required for Repression--
The role of ligand in
ERE binding by the ER has been controversial (reviewed in Ref. 46).
Although most studies support the view that liganded ER binds with
higher affinity to the ERE than unliganded ER, variable levels of ERE
binding by unliganded ER have been reported using promoter interference
assays (46-48). We observed a minimum level of repression with
unliganded KERK (Fig. 1A). The presence of ER ligands that
are either agonists or antagonists strongly potentiated repression by
KERK. Since ER-KRAB chimeras in which the KRAB domain was linked to
either the N or C terminus had equal potency (Fig. 2B), the
presence of the large KRAB repressor domain linked to the C terminus of the ER does not appear to limit the access of ligand to the binding pocket.
The mechanisms by which pure antiestrogens such as ICI 182,780 interfere with ER-mediated transcription have been the subject of
considerable interest (49). The ER occupied by pure antiestrogens is
thought to be largely localized in the cytoplasm (34, 35, 50), where it
is rapidly destroyed (34, 35), depleting cellular ER. Although ICI
182,780-occupied receptor binds DNA in vitro with slowed
kinetics (51), in vivo, at least part of the receptor population retains the ability to bind to the ERE (48). Since KERK
displayed a similar dose-dependent repression curve when liganded by 17-estradiol, OHT, or ICI 182,780, our data suggest that
even ICI 182,780 induces KERK binding to the ERE. The putative KRAB
corepressor TIF1 may potentiate nuclear localization of ICI
182,780-occupied KERK. In a study using an ER mutant missing the
nuclear localization signal, the ER coactivator TIF1 allowed ligand-dependent nuclear localization (23).
The Histone Deacetylase Inhibitor Trichostatin A Does Not Interfere
with Repression by the KRAB Domain--
One possible explanation for
the ability of the KRAB domain to repress transcription is that it
recruits a corepressor complex containing histone deacetylase activity.
Since the histone deacetylase inhibitor trichostatin A (27, 28) had not
previously been used in conjunction with the KRAB repressor, we
examined its ability to interfere with repression by the KRAB domain.
Trichostatin A did not affect repression by two KRAB chimeras on
several promoters. Under these conditions, which employ transient
transfections, KRAB repression uses a pathway independent of histone
deacetylation. One possible explanation for these data is that the
maintenance of a repressed chromatin state by the KRAB domain involves
the heterochromatin-enriched factors HP1a, MOD1, and MOD2, which
reportedly interact with KRAB corepressors TIF1 and TIF1 (13, 52,
53). These factors may prevent histone acetylases involved in the
relief of repression from gaining access to their substrates.
Binding to a Single ERE Is Sufficient for KRAB Repression--
Our
studies show that a GAL4-KRAB chimera and an ER-KRAB chimera each
exhibit DNA sequence-specific repression and that changing the DNA
binding specificity of an ER-KRAB chimera abolishes KRAB repression in
ERE-containing genes (Fig. 1). This corroborates earlier findings that
tethering the KRAB domain to DNA is required for repression (7-12, 14,
37, 38) and demonstrates that our ER-KRAB chimeras are targeted to EREs.
We find that a single ERE is sufficient for KRAB-mediated repression.
After completion of our work, a meeting report described effective
repression by a different type of steroid receptor-based KRAB repressor
(54). After completion of this paper, successful repression of
ERE-containing promoters by ER-NCoR fusions was reported (55). Our data
indicate that different rules apply for transcription activation and
repression. Although cell type and promoter context play a critical
role in the induction of transcription by the ER, the level of KRAB
repressor occupancy of the ERE appears to be the overriding factor in
repression. In addition, our data demonstrate that it is the presence
of the ERE, rather than the capacity for estrogen induction, that
determines the potential for repression of a gene by an ER-KRAB
chimera. Consistent with these conclusions is our finding that the
extent of repression was similar from the thymidine kinase, SV40, and elongation factor 1 promoters containing the same number of EREs, whereas induction by the ER varied from 15-fold to 0 on these promoters.
Interestingly, although synergism between ER bound at different EREs
can mask diminished binding (56) when the ER is activating transcription, this is not true for KRAB-mediated repression. Two EREs
were clearly more effective in enabling repression by KERK than a
single ERE, but there was no further increase in repression in going
from two to four EREs (Fig. 6A). This contrasts with hER-mediated transcription activation in the same cell line, where strong synergistic effects were seen in comparisons of activity on
reporter genes containing one, two, and four EREs (18, 56). Additional
support for the idea that tight binding to a response element is
important for KRAB repression comes from studies with the promoter
fragment containing the pS2 ERE. The ER binds to the pS2 ERE with a
lower affinity than to the consensus ERE (15, 41). Despite this
diminished binding, hER achieved a 3-fold transcription activation on
the pS2 ERE. In striking contrast, KERK was unable to suppress basal
promoter activity when bound to the same pS2 ERE. The ability of KERK
to partially suppress ER-mediated induction of the reporter containing
the pS2 ERE may stem from the ability of KERK to act as a
dominant-negative mutant interfering with the binding of wild-type ER,
without exerting active repression. In contrast, KERK-3M achieved
effective dose-dependent transcription repression of the
pS2 ERE. This suggests that high affinity binding to the imperfect ERE,
resulting in the continued presence of the ER-KRAB chimera on the
promoter, is critical for repression.
Combining Genetic Selection with ER-KRAB Chimeras Provides a Novel
Approach to Targeting Genes for Repression--
Most studies of gene
targeting use multiple rounds of phage display to select mutant
DNA-binding domains with affinity for a DNA target (57, 58). The
resulting proteins do not provide for ligand-regulated activation or
repression. Our surprising finding that binding of ER-KRAB chimeras to
the ERE can be modulated either by estrogens or by the widely used
SERMs OHT and ICI 182,780 makes ligand-dependent modulation
of gene activity feasible using these chimeras. The ability to use the
pure antiestrogen ICI 182,780 to activate ER-KRAB repressors enhances
their long-term potential for use as gene repressors in breast cancer
cells and in other systems in which use of ER agonists would be inappropriate.
We recently described a genetic selection using a modified form of the
bacteriophage p22 challenge phage selection system, which requires only
a single selection cycle (16). To repress transcription from the
imperfect pS2 ERE, it proved essential to modify the KRAB repressor
using information from our recent genetic selection for DBDs with
altered and enhanced ERE binding (16). To produce the KERK-3M
repressor, we combined information from our genetic selections
performed using steroid receptor DNA-binding domains with the KERK
chimera, whose ability to repress transcription can easily be modulated
using ER ligands. The KERK-3M repressor provides a model for a novel
class of gene-targeting protein that combines the ease of use of a
ligand-regulated steroid receptor with specificity and affinity gained
through large-scale genetic selection. The unique characteristics of
these hER-KRAB chimeras make them powerful new tools for the functional
analysis of ER-regulated genes.
§,
, and
TOP
ABSTRACT
INTRODUCTION
