(Received for publication, February 4, 1997, and in revised form, May 15, 1997)
From the Institut für Medizinische Chemie und
Biochemie, Universität Innsbruck, Fritz-Pregl-Straße 3, A-6020
Innsbruck, Austria and the § Department of Medical
Nutrition, Karolinska Institute, Huddinge University Hospital F60,
Novum, S-14186 Huddinge, Sweden
Steroid hormone receptors and Stat factors
comprise two distinct families of inducible transcription factors.
Activation of a member of each family, namely the glucocorticoid
receptor by glucocorticoids and Stat5 by prolactin, is required for the
efficient induction of the expression of milk protein genes in the
mammary epithelium. We have studied the mode of interaction between
Stat5 and the glucocorticoid receptor in the activation of -casein gene transcription. The functional role of potential half-palindromic glucocorticoid receptor-binding sites mapped previously in the promoter
region was investigated.
-Casein gene promoter chloramphenicol acetyltransferase constructs containing mutations and deletions in
these sites were tested for their responsiveness to the synergistic effect of prolactin and dexamethasone employing COS-7 cells or HC11
mammary epithelial cells. Synergism depended on promoter regions
containing intact binding sites for the glucocorticoid receptor and
Stat5. The carboxyl-terminal transactivation domains of Stat5a and
Stat5b were not required for this synergism. Our results suggest that
in lactogenic hormone response elements glucocorticoid receptor
molecules bound to nonclassical half-palindromic sites gain competence
as transcriptional activators by the interaction with Stat5 molecules
binding to vicinal sites.
The stage-specific regulation of milk protein gene expression is
controlled by the lactogenic hormones glucocorticoids and prolactin.
Both hormones have been documented to synergistically induce the
expression of these genes at the level of transcription initiation (1).
Lactogenic hormone response elements have been defined in the promoter
region and upstream enhancer regions of the genes encoding the milk
proteins S1-casein (2),
-casein (1, 3, 4), whey acidic protein
(5-7), and
-lactoglobulin (8). They all contain binding sites for
the prolactin inducible transcription factor
MGF/Stat51 (2, 8-10). A
mutation introduced into the Stat5-binding site has been shown to
destroy the response of the promoter not only to prolactin but also to
glucocorticoids (11), indicating that Stat5 is necessary for mediating
the effects of prolactin and glucocorticoids. Glucocorticoids do not
significantly change the binding activity of MGF/Stat5 to lactogenic
hormone response elements (12), ruling out the possibility that they
act synergistically with prolactin simply by enhancing the effect of
prolactin on the activation of Stat5 DNA binding.
In vitro binding studies with purified preparations of the
glucocorticoid receptor revealed the presence of multiple GR-binding sites in the lactogenic response elements of the rat -casein gene
promoter (13), the proximal mouse whey acidic protein gene promoter
(13), and the distal rat whey acidic protein gene promoter (14).
Interestingly, only sequence motifs resembling half-palindromic GR-binding sites were contained within the footprinted regions. Transactivation mediated by the glucocorticoid receptor usually requires the binding of dimeric receptor molecules to palindromic DNA-binding sites (15-17), whereas half-sites binding monomeric GR
complexes have been reported to be insufficient by themselves to confer
hormone responsiveness (17). However, monomeric GR molecules binding to
receptor half-palindromic sites were proposed to gain competence as
transcriptional activators by interacting with other transcription
factors (18).
We tested whether GR molecules binding to half-palindromic sites in the
rat -casein gene promoter can functionally interact with MGF/Stat5
in the activation of milk protein gene transcription. COS-7 cells were
employed in transient co-transfection assays using wild-type and
mutated
-casein gene promoter CAT constructs. These cells allow the
reconstitution of prolactin-dependent signaling cascades by
transfection of prolactin receptor and Stat5 expression vectors (19).
They also can be used to study the synergy between Stat5 and GR in
cotransfection experiments (20). The results presented here suggest
that GR half-palindromic sites serve to recruit GR molecules to the
-casein gene promoter and are instrumental for mediating the
synergistic effect between glucocorticoids and prolactin. Experiments
with stably transfected HC11 mammary epithelial cells, which do not
overexpress Stat5 and GR molecules, lead to the same conclusion.
Efficient transactivation by GR molecules was dependent on the
activation and binding of Stat5 to its recognition site. Interestingly,
transactivation was also possible with Stat5 molecules devoid of their
carboxyl-terminal transactivation domain.
-Casein gene promoter constructs with mutations
in the glucocorticoid receptor half-sites were prepared by
site-directed mutagenesis using the protocol of Deng and Nickoloff
(21). The
-casein CAT promoter gene p
c(
344/
1)CAT was used as
a template (22). The selection primer for introducing the first
mutation was 5
-CCCCGGGTACAGATCTCGAATTCGT-3
. It destroys the unique
SacI and KpnI cutting sites, replacing them with
a BglII site. For a second round of mutation, the primer
5
-GGATCGATCCTCGAGTACAGATCT-3
was used. It destroys the unique
SmaI site and creates a novel XhoI site. The
primers used for introducing mutations into the GR sites (in
parentheses) were: 5
-CCTTGTTTAAGCTTCCCCAGAATT-3
(mGRc),
5
-TTTCTAATCAAGCTTACTTCTTGGA-3
(mGRd),
5
TTGGAATTAACAGACTTTTGAA3
(mGRe), and
5
-TTTCTAATCAAGCTTACTTCTTGGAATTAACAGACTTTTGAA-3
(mGRd and mGRe). For
mutations in the GR sites of footprints a and b the Quick
ChangeTM site-directed mutagenesis protocol of Stratagene
was used. The upper strands of the oligonucleotides employed were:
5
-GGCTGGGGAGAATTCTGATGACTGTTTACTAGGCTGGAG-3
(mGRa);
5
-GGCTGGAGAGAATTCCAGTTATTTGACAATTTCCTTTCC-3
(mGRa/b); 5
-CAATTTCCTTTCCTTGACGAATTCCTTCACCAGCTTCTG-3
(mGRb). The construct with the mutation of the MGF/Stat5 site and the heterologous
-casein gene thymidine kinase (tk) promoter CAT constructs employed have been
described (23). The construct pMMTV-CAT was prepared by inserting a
1.3-kilobase pair PstI/BamHI fragment, spanning
the region from
1187 to +102 of the mouse mammary tumor virus-long terminal repeat, into the 5
polylinker of pBLCAT3 (24). The constitutively active luciferase expression vector pAGLuE5, driven by
the SV40 early promoter, was provided by Dr. T. Schlake. Stat5 expression vectors were constructed by insertion of Stat5a and Stat5b
cDNA cloned from a mouse mammary gland
ZAP
library2 into pECE (25). The
carboxylterminally deleted forms of Stat5a and Stat5b were prepared
by introducing a stop codon at amino acid 738 position of Stat5a
and the corresponding position (amino acid 742) of Stat5b. The rat
glucocorticoid expression vector PSTC GR 3-795 (26) contains the
cDNA of the rat glucocorticoid receptor encoding amino acids
3-795.
COS-7 cells were propagated in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal calf serum, and 50 µg/ml gentamycin. HC11 cells were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum, 5 µg/ml insulin, 10 ng/ml epidermal growth factor, and 50 µg/ml gentamycin. Prior to hormone treatment, HC11 cells were kept for 2 days in epidermal growth factor-free medium containing 2% fetal calf serum as described (23). Hormone inductions were performed with 5 µg/ml ovine prolactin (31 units/mg, Sigma) and 0.1 µM dexamethasone (Sigma), as indicated.
Electromobility Shift AssaysExperimental procedures were
as described previously (23). Double-stranded oligonucleotides labeled
with [-32P]ATP (>6000 Ci/mmol) and annealed to the
complementary oligonucleotides as described (23) were used. The
sequence of the upper strand of the oligonucleotides employed was: SIE,
5
-GTGCATTTCCCGTAAATCTTGTCTACAATTC-3
; PRE,
5
-AGCTTAGAACACAGTGTTCTCTAGAC-3
; GRc,
5
-GCTGCCTTGTTTAATGTCCCCCAGAATTTCTTGG-3
; mGRc,
5
-GCTGCCTTGTTTAAGCTTCCCCAGAATTTCTTGG-3
; GRd+e,
5
-CTAATCATGTGGACTTCTTGGAATTAAGGGACTTTTG-3
. Electromobility
shift assays were performed on a 4% polyacrylamide gel in 0.25 × TBE electrophoresis buffer. Binding reactions were as described (27).
GR-DBD expressed in Escherichia coli was purified as
described (28) and used for binding reactions in a buffer containing 10 mM Hepes, pH 7.5, 2.5 mM MgCl2,
10% glycerol (w/v), 50 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol, 50 ng of poly(dI-dC).
Transfections with the calcium phosphate precipitation technique, selection and propagation of stably transfected HC11 cells, and CAT assays were performed as described (27). Transient transfection of COS-7 cells were made with cultures plated on 6-well dishes. The total amount of cotransfected DNA was 20 µg/6 wells. For preparing extracts to determine CAT and luciferase activity, COS-7 cells were washed with ice-cold phosphate-buffered saline, incubated for 5 min with 1 ml of buffer containing 40 mM Tris-HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, and harvested by scraping them off the dish with a rubber policeman. The subsequent steps were performed at 0-4 °C. Two aliquots of 450 µl each of the resuspended cells were transferred to a microcentrifuge tube, centrifuged for 5 min at 1000 × g, and the supernatants were removed by aspiration. One of the aliquots was used for determination of CAT activity (27). A luciferase assay was performed with the other aliquot as described (29).
A regulatory
region in the -casein gene promoter required for the synergistic
action of glucocorticoid hormones and prolactin was mapped in HC11
mammary epithelial cells (1, 23). It comprises the MGF/Stat5 site and a
region located 5
to this site. We have investigated the role of this
5
region in mediating the synergy between glucocorticoids and
prolactin in COS-7 cells. These cells were shown to be a suitable
system to study the prolactin-dependent transactivation
activity of Stat5 (19). Cotransfections were performed with expression
plasmids for mouse Stat5a, mouse prolactin receptor, and a
-casein
(
344/
1) promoter CAT reporter gene. Increasing amounts of a
glucocorticoid receptor expression plasmid were included. The
cotransfected cells were stimulated with dexamethasone or prolactin or
a combination of both. In the absence of cotransfected glucocorticoid
receptors (Fig. 1, first 4 columns) dexamethasone did not have a significant effect on
transcription. The synergism between dexamethasone and prolactin
increased in a concentration dependent fashion by cotransfection of GR
expression vector (Fig. 1). Using 40-1000 ng of GR expression vector
the synergistic effect approached the level observed in HC11 cells for
the regulation of a stably transfected
-casein (
344/
1) gene
promoter CAT construct (23). Further cotransfection studies were
performed with 200 ng of GR expression vector.
We next investigated whether the MGF/Stat5 site alone is sufficient to
mediate the synergistic response in COS-7 cells. A 30-base pair region
comprising the MGF/Stat5 site in the -casein gene promoter or a
region extending from
344 to
82 cloned in front of a thymidine
kinase-CAT reporter gene were analyzed. The short fragment with the
MGF/Stat5 site only conferred a transcriptional response upon prolactin
stimulation to the thymidine kinase-CAT reporter (Fig.
2). However, no synergism was observed
with dexamethasone. By contrast, a marked increase in transcriptional
activation was obtained upon stimulation with dexamethasone and
prolactin, when the
-casein gene promoter fragment (
344/
82) was
used. The synergistic transcriptional activation mediated by prolactin
and dexamethasone is hence dependent on sequences 5
to the MGF/Stat5
site in COS-7 cells as has been found previously in HC11 cells
(23).
Glucocorticoid Receptor Half-palindromic Sites in the
In a previous study GR-binding sites were mapped in
the rat -casein gene promoter by in vitro footprint
analysis using purified rat liver glucocorticoid receptor (13). Five
regions were identified to be protected from digestion by DNase I (Fig.
3, footprints a-e). They
contain sequence motifs related to half-sites of classical glucocorticoid response elements. The functionality of the putative half-site contained in footprint c to bind the GR was investigated by
electromobility shift assays employing the DNA-binding domain of the
glucocorticoid receptor (GR-DBD, Fig.
4A). The oligonucleotide GRc,
spanning the footprinted region c of the
-casein gene promoter and a
corresponding oligonucleotide harboring a mutation in the putative
half-site (mGRc, mutation as shown in Fig. 3) were used as DNA probes.
A classical palindromic GR-binding site (PRE) and a sequence devoid of
described GR-binding sites (SIE) served as controls. Very weak binding
was observed with the SIE (Fig. 4A, lanes 1 and
2). With a palindromic GRE (PRE oligo) two retarded bands
were detected (Fig. 4A, lanes 3 and 4). The two
bands represent complexes containing receptor monomers or dimers. This
is in accordance with previous reports (18, 30). With the GRc probe,
formation of a complex migrating at the position of receptor monomers
was observed (Fig. 4A, lanes 5 and 6). Binding
activity of the mutated GRc probe was reduced to the low level observed
with to the SIE probe (Fig. 4A, compare lanes 1 and 2 with 7 and 8). The
experiment confirms the assumption that the integrity of the
half-palindromic site in the GRc probe is required for the binding of
the GR. In electromobility shift assays with a DNA probe comprising the
half-sites in GRd and GRe (GRd+e), formation of complexes with two
receptor molecules was observed at high concentrations of GR-DBD (Fig. 4B, lanes 1 and 2). Electromobility shift assays
performed with oligonucleotides with targeted mutations in either Grd
or GRe palindromic half-sites revealed that the integrity of the
half-sites in GRd and GRe is important for the formation of these
complexes (data not shown). In comparison to the PRE probe with the
palindromic GR site (Fig. 4B, lanes 5-7), formation of
complexes containing two receptor molecules required significantly
higher concentration of the GR-DBD (Fig. 4B, compare
lanes 1 and 2 to lanes 5 and
6), indicating that the configuration of the two binding
sites in GRd+e does not favor co-operative binding of GR molecules to
the same extent as the classical palindromic site in PRE.
Glucocorticoid Receptor-binding Sites Are Required for Promoter Function
To determine the importance of the GR sites for promoter
function, mutations affecting the various GR half-site were introduced into the -casein (
344/
1)CAT plasmid. The mutations employed are
shown in Fig. 3 on top of the sequence. Cotransfections with the
mutated promoter constructs were performed in COS-7 cells as described
above. The ability of dexamethasone to augment the response of
prolactin was quantified by determining the ratio of the CAT activity
in cells stimulated with dexamethasone and prolactin versus
the activity in cells treated with prolactin only (Fig.
5A). Mutations in GR
half-sites a, b, c, and e led to a reduction of the ratio to 17 (mGRa),
29 (mGRa/b), 30 (mGRb), 34 (mGRc), and 42% (mGRe) of the wild-type
-casein CAT construct. The induction ratio was almost completely
abolished in constructs harboring double mutations of GRc and GRe (Fig.
5A, mutations mGRc+e and mGRc+d+e), indicating a
co-operation of these binding sites for mediating the synergism of GR
with Stat5. The effect of a mutation in GRd was not statistically
significant. This mutation also had no significant additional effect
when combined with mutations of either GRc or GRe or both (Fig.
5A, mutations mGRd, mGRc+d, mGRd+e, and
mGRc+d+e).
As shown in the upper panel of Fig. 5B for COS-7 cells, the activation of transcription by prolactin alone was not altered by the triple GR mutation, indicating that the effect is selective on the action of glucocorticoids. Mutation of the MGF/Stat5 site, however, strongly interfered with the prolactin response. In this mutant the synergistic effect of dexamethasone was also reduced. The weak but significant induction observed is most probably explained by an incomplete inhibition of Stat5 binding by the mutation. The data presented document the functional importance of the GR half-sites in GRa, GRb, GRc, and GRe for GR action. Furthermore, they show the dependence of GR on co-operation with Stat5, which also has to bind to the promoter.
To examine whether the findings obtained in Stat5 and GR overexpressing
COS-7 cells are also relevant for the situation in mouse mammary
epithelial cells expressing physiological levels of these two
transcription factors, HC11 cells were stably transfected with the same
mutated reporter constructs (Fig. 5B, lower panel). The
results obtained in HC11 cells supported the conclusions drawn from the
experiments with the COS-7 cell system. The triple mutation affecting
the three GR-binding sites c, d, and e strongly reduced the synergistic
activation of the -casein promoter by dexamethasone and prolactin.
Destroying the Stat5-binding site also inhibited prolactin- and
dexamethasone-induced transactivation.
To determine whether the GR sites in GRc and GRe are sufficient for
mediating the synergism with Stat5, constructs were prepared with 5
deletions of the
-casein promoter placed in front of a minimal
thymidine kinase promoter and analyzed in COS-7 cotransfection experiments (Fig. 5C). A (
176/
82)
-casein thymidine
kinase promoter which harbors the GRc- and GRe-binding sites failed to
confer the prolactin and dexamethasonedependent synergistic
transactivation. The responsiveness to dexamethasone and prolactin was
restored in the constructs which contain the regions GRa and GRb (Fig. 5C, p
c(
344/
82)thymidine kinase-CAT and
p
c(
282/
82)thymidine kinase-CAT)). Thus, the distal
-casein
gene promoter region with the GR-binding sites for GRa and GRb is
necessary for mediating the synergism between GR and Stat5. As shown
above (Fig. 5A), the functionality of this region is
dependent on the integrity of the proximal GR-binding sites in GRc and
GRe, implying that the synergism depends on a co-operation of distal
and proximal sites.
The
two highly related Stat5 molecules Stat5a and Stat5b are expressed in
the mammary epithelium. We wanted to examine whether they behave
similarly in synergizing with the GR. Cotransfection experiments in
COS-7 cells were performed as above. As shown in Fig.
6, no significant differences were
observed in transactivation of the -casein gene promoter upon
induction with dexamethasone and prolactin. We also tested
carboxyl-terminally deleted forms of Stat5a and Stat5b, lacking the
region with the major transcriptional activation domain (31).
Surprisingly, the deleted forms were as efficient as the full-length
Stat5 isoforms in their synergy with the GR. Thus, the transactivation
domain in the carboxyl-terminal region of Stat5 is not required for the
establishment of the synergism with the glucocorticoid receptor.
The steroid receptor superfamily and Stat molecules represent two archetypal families of signaling molecules which evolved to meet the requirement of cells to respond differentially to diverse extracellular stimuli. The specificity of the response is controlled at two levels (15, 32): At the first level, there is a selective activation of steroid receptors or Stat factors by different extracellular signals. This is achieved by the specific binding of the steroid receptors to their ligands and the selective recruitment of Stat factors by their activating receptors. The second level of specificity is brought about by the recognition of distinct DNA-binding sites by the activated steroid receptors and Stat factors.
Integration of different signaling pathways is a further means to increase the versatility of the response to extracellular stimuli (33-35). In this study we have investigated how integration of the signaling pathways triggered by the two pleiotropic hormones prolactin and glucocorticoids leads to the specific activation of milk protein gene transcription.
When this article was in preparation, Stöcklin et al.
(20) reported a direct interaction between Stat5 and the GR
overexpressed in COS-7 cells. A model was proposed where the GR acts as
a co-activator of Stat5 in a mode which is independent of a GRE. The
results presented in our study do not support this model, since they
provide clear evidence that GR and Stat5 molecules, activated by the
two hormones, interact in a lactogenic hormone response element
dependent fashion. DNA-binding sites for both the GR and Stat5 were
demonstrated to be essential in mediating their synergism on -casein
gene induction. However, protein-protein interactions between GR and Stat5, as the one observed by Stöcklin et al. (20),
might also be important for a productive functional interaction of GR
and Stat5 molecules, recruited to the lactogenic hormone response elements of milk protein gene promoters. In fact, a role for both DNA-template dependent and independent interactions of the GR or Stat5
with other transcription factors has been demonstrated frequently. GR
homodimers bound to palindromic GR consensus sites were described to
interact with several unrelated transcription factors bound to vicinal
sites (15, 33, 36). In addition, direct DNA-template independent
interactions of the GR with the transcription factors NF
B (37),
NF-IL6 (38), and AP-1 (reviewed in Ref. 39) have been reported.
Protein-protein interactions of GR monomers and AP-1 were suggested to
be involved in the repression of transcription factor AP-1 activity
(40). Interactions involving both DNA binding and protein-protein
contacts are the hallmark of composite glucocorticoid response elements
(41), which mediate a positive or negative effect of glucocorticoids on
transcription depending on the type of the cooperating transcription
factor.
For Stat factors, DNA template-dependent synergy with
members of the Ets family and C/EBP has been described. In the
c-fos gene promoter, Stat1 and Stat3 cooperate with ternary
complex factors (42); in the Fc receptor gene, Stat1 cooperates with PU.1 (43); for the induction of the interleukin-4 gene vicinal sites of
Stat6 and C/EBP have been found to be important (44); activation of the
lipopolysaccharide-binding protein required Stat3, C/EBP, and AP-1
sites (45). Examples of direct binding of Stats to other proteins
involved in transcription are the interaction of the Stat3
splice
variant with c-jun (46); the cooperative interactions of dimeric Stat
molecules with themselves, mediated by the NH2-terminal
domain of Stats (47); and the binding of the carboxyl-terminal domain
of Stat2 to the p300/CBP transcriptional adaptor (48).
GR molecules usually utilize palindromic sites for high affinity DNA
binding. They are composed of hexameric half-sites with the consensus
5-TGTTCT-3
spaced by 3 nucleotides (15). The GR-binding sites
characterized in the lactogenic hormone response elements were
non-classical sites. They represent half-palindromic sites (13, 14).
Half-sites binding monomeric GR molecules have been found to be unable
to confer glucocorticoid-inducible transcription on a heterologous
promoter by themselves (17). However, in complex glucocorticoid
response elements, half-palindromic sites have been mapped which are
essential in mediating the effect of glucocorticoid hormones (17, 18).
Functional half-palindromic binding sites were also described for the
estrogen receptor in the far upstream estrogen response element of the
ovalbumin gene (49). Mutational analysis of the three proximal GR
half-sites in the
-casein gene promoter (Fig. 5A)
revealed the functional importance of the half-palindromic sites in
GRa, GRb, GRc, and GRe. In the lactogenic hormone response element the
utilization of half-sites by the GR appears to be one important means
for ensuring the stage-specific activation of milk protein expression: here, glucocorticoids are unable to induce transcription via the GR
half-sites alone, but require the synergy with Stat5 activated by
prolactin. Although the unusual spacing and orientation of GR
half-sites in lactogenic hormone response elements would not allow a
favorable steric alignment of GRs like the one observed with classical
GREs, direct interactions between receptors bound to several sites
might be important. In addition, or alternatively, interactions of the
GR with other factors binding to neighboring sites could stabilize the
binding of the GR to half-palindromes. For the
-casein gene a
cooperative binding between C/EBP and GR could be important, since
several binding sites for C/EBP were mapped vicinal to GR
half-palindromic sites (23). In the case of the whey acidic protein
gene promoter, CTF/NF-1 could be involved (10, 14).
Two cellular systems were employed in the present study. COS-7 cells
allowed to assess directly the role of Stat5 in transient cotransfection experiments. However, a drawback of this system is that
the cells are not of mammary epithelial origin and the expression
levels of Stat5 and GR are very high, giving rise to the question
whether the results obtained are relevant for the in vivo
regulation of milk protein genes by lactogenic hormones. We thus also
used HC11 mammary epithelial cells which can be induced by
glucocorticoids and prolactin to express the endogenous -casein gene. It was possible to demonstrate a functional role of GR half-sites in mediating the response of lactogenic hormones in this cell line
(Fig. 5B), suggesting that the mechanism of interaction
between GR and Stat5 is the same in HC11 and COS-7 cells and thus of
general relevance for the regulation of milk protein gene
expression.
In HC11 cells the action of lactogenic hormones on milk protein
expression is slow (50). An indirect effect of glucocorticoids on
transcription, mediated by the activation or repression of a gene
regulated by the GR has been described to account for the slow response
(22). Our data presented here implicate that, in addition, the GR
exerts a direct effect on the lactogenic hormone response element.
Regulation of gene expression by direct and indirect mechanisms has
also been reported to account for the action of glucocorticoids on the
2-uteroglobulin (51) and the
-amylase 2 (18) genes, which both
contain functional GR half-sites similar as the lactogenic hormone
response elements.
Stat5a and Stat5b, which differ significantly in their carboxyl terminus, were equally potent in their cooperativity with the GR (Fig. 6). Interestingly, Stat5 devoid of the carboxyl terminus was as efficient in mediating the synergism with the GR as full-length Stat5. The deleted region contains the major transactivation domain of Stat5 important for the response to prolactin alone (31). Thus, the finding implies that either the transcriptional activation in the synergistic response is predominantly mediated by the GR, or that a cryptic transactivation domain of Stat5 is unmasked by the interaction with the GR. Stat5 molecules lacking the carboxyl-terminal transactivation domain have been described as dominant negative (31). They efficiently inhibited the activation of prolactin-dependent genes, presumably because they retained the ability to bind DNA and thereby were able to compete with transcriptionally active, full-length Stat5 molecules. Since, as shown in Fig. 6, carboxyl-terminally deleted Stat5s are not impaired in their ability to synergize with the GR to activate transcription, these molecules should allow the specific repression of genes regulated by prolactin only, without affecting genes synergistically regulated by both dexamethasone and prolactin. Further mutation and deletion experiments will reveal the domains of Stat5 required for a functional interaction with GR. It will be interesting to see whether other steroid receptors can also interact with Stat5 and whether other members of the Stat family are able to cooperate with the glucocorticoid receptor.
We thank S. Philipp, N. Greier, and C. Soratroi for excellent technical assistance, Dr. R. K. Ball for the mouse prolactin receptor expression vector, Dr. T. Schlake for the luciferase expression vector, Dr. S. Rusconi for providing the rat glucocorticoid receptor expression vector, and Dr. H. Klocker and Dr. A. Helmberg for critical reading of the manuscript.