Abstract
Transcription of the Ig H chain germline transcripts is a prerequisite for class switching. Expression of the ε germline transcript is induced by IL-4 and requires the integrity of a composite IL-4 response element. The element is bound by the IL-4-inducible transcription factor Stat6 and one or more members of the CAAT/enhancer-binding protein (C/EBP) family, a constitutively expressed class of transcription factors. Here, we show that Stat6 and C/EBPβ cooperate to synergistically activate transcription from the ε element. The effect was most pronounced in lymphoid cells, and the activation domains of both proteins were required to achieve this synergy. Although other members of the C/EBP family are able to bind the element, very little cooperativity was seen with C/EBPα and none with C/EBPγ. In fact, C/EBPγ was able to inhibit IL-4-induced reporter activity. Stat6 and C/EBPβ bind the IL-4 response element simultaneously. The fast dissociation rate apparent when Stat6 binds this DNA element alone is slowed when C/EBPβ binds at the neighboring site. These data suggest a mechanism whereby C/EBPβ stabilizes Stat6 binding at this element, thereby increasing the likelihood that both of their activation domains will interact, possibly with other factors, to activate transcription in an IL-4-dependent manner.
Cytokine receptor binding initiates a cascade of intracellular signaling events leading to specific changes in gene expression (1). IL-4 is a multifunctional cytokine that stimulates changes in many cell types. In the case of T helper cells, IL-4 drives the commitment to the Th2 phenotype. Th2 cells produce IL-4, IL-5, IL-6, and IL-10 (2, 3), cytokines that predominantly influence humoral immune responses. IL-4 also plays a critical role in B cell isotype switching and proliferation as well as in mast cell and eosinophil activity (1, 2). In contrast, IL-12 drives the commitment to the Th1 subset, which increases cytokine secretion predominantly affecting cellular immune responses (2, 3).
Cytokines such as the interleukins and interferons have been shown to rapidly activate a signaling pathway known as the Janus kinase (Jak)3-STAT pathway. Briefly, STAT proteins are activated through tyrosine phosphorylation by receptor-associated Jak kinases following cytokine binding. The activated STAT protein dimerizes, translocates to the nucleus, and activates transcription via specific DNA response elements (4, 5, 6). Currently, there are seven known members of the STAT family that are characterized and that are activated by different cytokines. In the present study, we focused on Stat6, which is activated upon IL-4 stimulation (7, 8). The generation of Stat6-deficient mice has established the requirement of this protein in promoting polarization of T helper cells toward the Th2 phenotype. Furthermore, genes that are activated in response to IL-4, such as CD23 and MHC class II genes, cannot be induced in Stat6-deficient mice, and IgE production is profoundly impaired (9, 10).
Stat6 binding sites have been identified in several of these IL-4-responsive genes and are best characterized in the promoters that govern the expression of the Ig germline ε and γ transcripts (11, 12). IL-4-induced expression of these genes requires the integrity of the Stat6 binding site as well as an adjacent site, which is bound by C/EBP (11, 12). In the mouse germline ε promoter, the Stat6 and C/EBP sites are separated by four base pairs; in the human promoter, the two sites are directly adjacent. Previous studies have shown that either one of these composite elements (mouse or human) is able to confer IL-4-induced transcription onto a heterologous promoter (11, 13). These results indicated that Stat6 and one or more members of the C/EBP class of proteins may cooperate to activate transcription from this element by a yet undefined mechanism.
The C/EBP class of proteins consists of several members with variable and overlapping cellular expression patterns. Members of this family have been shown to play an important role in energy metabolism (C/EBPα, -δ), immune system function (C/EBPβ, -γ, -ε), and development (14, 15, 16, 17, 18, 19). C/EBPβ, like Stat6, has been shown to be expressed in lymphocytic, monocytic, and other cell types known to mediate IL-4 signaling (20, 21). However, unlike Stat6, C/EBPβ is constitutively present and is not activated upon IL-4 stimulation. Rather, numerous studies have shown that changes in expression, as well as posttranslational modification of this and other C/EBP family members, are brought about by other stimuli (22, 23).
In the current study, we investigate the functional and physical interaction between Stat6 and several C/EBP proteins. We show that Stat6 and C/EBPβ activate transcription synergistically from the germline ε IL-4-responsive element. The activation domains of both proteins are essential for this functional synergy. Moreover, Stat6 and C/EBPβ interact with DNA such that the dissociation rate of Stat6 is stabilized when C/EBPβ is bound at the adjacent site. These results help to explain the mechanism by which Stat6 and C/EBPβ cooperate to synergistically activate transcription from the Ig germline ε promoter.
Materials and Methods
Cell culture and transient transfections
Human embryonic kidney 293 cells and HepG2 cells were grown in DMEM (Mediatech, Herndon, VA) containing 10% FCS (Mediatech Herndon, VA). BJAB cells were grown in RPMI medium containing 10% FCS, 1 mM l-glutamine, and 10 μM β-mercaptoethanol. To measure luciferase activity, HEK293 and HepG2 cells were transfected in six-well plates using calcium phosphate coprecipitation. The amount of DNA used in individual transfections is given in the legend of each figure. The medium was changed 15 h posttransfection, and after 42 h the cells were either induced with 10 ng/ml human rIL-4 (R&D Systems, Minneapolis, MN) for 6 h or left untreated. A control plasmid expressing β-galactosidase under the control of the cytomegalovirus promoter was cotransfected to determine the transfection efficiency. Luciferase and β-galactosidase activity were assayed using the luciferase and β-galactosidase assay systems (Promega, Madison, WI). Stable BJAB cell lines overexpressing human Stat6, C/EBPα, and C/EBPβ were generated as described by Tewari and Dixit (24). Resistant clones were selected in 3 mg/ml G418 (Life Technologies, Gaithersburg, MD), and positive clones overexpressing the recombinant proteins were identified by Western analysis using anti-Stat6, anti-C/EBPα, or anti-C/EBPβ Abs. The cells were then transiently transfected by electroporation using 30 μg of the IL-4-inducible reporter plasmid and 10 μg of the β-galactosidase-expressing control plasmid as described (24). After 24 h, cells were stimulated with 10 ng/ml IL-4 (R&D Systems) for 6 h or left untreated. Luciferase and β-galactosidase activity were assayed using the corresponding assay systems (Promega).
Expression constructs
Mammalian and baculovirus expression constructs encoding Stat6 and Stat6(Ad−) have been described previously (13, 25). Constructs expressing the C/EBP isoforms were generated as follows: DNA fragments encoding C/EBPα, C/EBPβ, and C/EBPγ were obtained using the PCR and primers carrying EcoRI sites. The DNA fragments were subcloned into the EcoRI site of pcDNA3. The orientation and integrity of the clones were determined by DNA sequence analysis. The following primer sequences were used to subclone the C-terminal bzip (basic leucine zipper; DNA-binding domain) region of human C/EBPβ (designated C/EBPβ(Ad−) in the text) into the pcDNA3 expression vector: 5′ primer, GCA GAC GAA TTC GCC ACC ATG GTC AAG AGC AAG GCC AAG AAG; 3′ primer, ACG AGC GAA TTC CTA TCA TCA GCA GTG GCC GGA. The reporter constructs TPU474 (wild type) and TPU475 (C/EBP site mutated) have been described previously (13).
Purification of C/EBPβ, Stat6, and Jak1
C/EBPβ was partially purified as follows. The protein was transiently overexpressed in HEK293 cells and nuclear extract was prepared (7, 13). The extract was passed over a heparin sulfate column that had been equilibrated with nuclear extract buffer (7). Bound proteins were eluted with buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) containing 50 mM, 200 mM, 500 mM, and 1000 mM of NaCl. Fractions were assayed for activity using electrophoretic mobility shift assay (EMSA). The 200 mM NaCl fraction contained the highest level of C/EBPβ activity and was generally used in subsequent experiments. Recombinant histidine-tagged Stat6 and Jak1 were expressed in High Five cells (Invitrogen, San Diego, CA) using the baculovirus expression system (PharMingen, San Diego, CA). The proteins were purified using nickel chromatography (Qiagen, Valencia, CA) as described previously (25).
In vitro activation of Stat6
Purified Stat6 was activated in vitro using purified Jak1 kinase. Both proteins were expressed in High Five cells. The kinase reaction was conducted as follows: 0.5 μg Jak1 and 1 μg Stat6 were incubated in 10 mM HEPES pH 7.4, 50 mM NaCl, 5 mM MgCl2, 5 mM MnCl2, 50 μM ATP, and 0.1 mM Na3VO4 for 30 min at room temperature. One microliter of the reaction was used for EMSA.
DNA binding assays, EMSA, and “trap” experiments
Labeling of DNA probes, DNA binding, and EMSAs were performed as previously described (13). Trap experiments were done as follows: Stat6 and C/EBPβ, either alone or in combination, were incubated with 0.1 pmol of labeled probe containing both the Stat6 and C/EBP binding site. The binding reaction was allowed to reach equilibrium to be reached for 2 min. Then, 20 pmol of unlabled DNA (trap) identical in sequence to the probe DNA was added to the reaction mix. Aliquots were taken at various times and loaded onto a continuously running native gel.
Results
Stat6 and C/EBPβ synergistically activate transcription in an IL-4-dependent manner
Previous studies have shown that the minimal IL-4 responsive element in the mouse and human ε promoters is a composite element of a Stat6 and a C/EBP binding site (11). This element, when fused to a truncated promoter, is able to promote transcription in an IL-4-dependent manner (11, 13). These observations indicate that Stat6 and one or more member of the C/EBP class of proteins cooperate to activate transcription from this regulatory element. To study the nature of this cooperativity, we transiently overexpressed Stat6, C/EBPα, and C/EBPβ in HEK293 cells in the presence of an IL-4-inducible reporter construct carrying four copies of the composite IL-4 response element derived from the human germline ε promoter (13). Previously, we showed that HEK293 cells do not express Stat6 (13). However, they do contain all other components of the IL-4 signaling pathway, such as the IL-4R and Jak kinases, which are essential for IL-4-induced gene activation. Thus, as Figure 1,A shows, no IL-4-dependent activation was observed in the absence of exogenous Stat6 (Fig. 1,A). Overexpression of Stat6 resulted in a 5-fold increase of IL-4-induced luciferase activity (13, Fig. 1, A and B). Overexpression of C/EBPα or C/EBPβ elevated the basal level of transcription but did not confer IL-4 responsiveness (Fig. 1,A). However, the combination of Stat6 and C/EBPβ resulted in a dramatic increase in IL-4-induced transcription (11-fold). This increase was significantly greater than the sum of each protein’s individual ability to drive transcription from this reporter construct. Thus, Stat6 and C/EBPβ functionally synergize to activate transcription from the human ε IL-4 responsive element. Overexpression of Stat6 and C/EBPα also showed an increase in basal and IL-4-induced transcription (Fig. 1,A). However, the fold induction in response to IL-4 was identical to the one observed in the absence of C/EBPα (Fig. 1 B). Hence, the IL-4-dependent synergistic effect seen with C/EBPβ is far more pronounced than that seen with C/EBPα.
To further substantiate the functional synergy between Stat6 and C/EBPβ, we explored the cooperativity between the two proteins in a lymphoid cell line. BJAB cells, a human B cell line, have been shown to express Stat6, and IL-4-responsive genes are properly regulated (unpublished data). Stable cell lines that overexpress Stat6, C/EBPα, or C/EBPβ were prepared and then transiently transfected with the IL-4-inducible reporter construct. The parental BJAB cell line showed a 3-fold increase in luciferase activity following IL-4 treatment (Fig. 1,B). A 10-fold increase was observed in cells that overexpressed Stat6. As observed in HEK293 cells, overexpression of C/EBPα gave rise to elevated basal as well as IL-4-induced transcription (data not shown), resulting in no increase in IL-4 inducibility when compared with the parental cell line (Fig. 1 B). In contrast, BJAB cells that overexpressed C/EBPβ did not show an elevated basal level compared with the parental line. However, a 75-fold increase in luciferase activity was seen in response to IL-4 treatment. This specific increase in reporter activity upon IL-4 treatment strongly implies a cooperative interaction between Stat6 and C/EBPβ to synergistically drive transcription from the ε IL-4 response element. This finding is further supported by using reporter constructs that carried mutations in the C/EBP binding site. In HEK293, BJAB (data not shown), or HepG2 cells (see below), IL-4-induced promoter activity was completely abolished even in the presence of overexpressed C/EBPβ when the mutant reporter was used.
Activation domains of both Stat6 and C/EBP β are required for synergistic activation
Previously we showed that a mutant version of Stat6 lacking the C-terminal 186 amino acids is unable to activate transcription, although this deletion does not affect IL-4-dependent phosphorylation or DNA binding affinity. These and other experiments mapped the transcriptional activation domain to the C-terminal 186 amino acids of Stat6 (13). Now we wanted to determine whether the activation domain of C/EBP β was also an essential domain contributing to the ability of these two proteins to drive transcription from the ε IL-4-responsive element. Figure 2 shows the results obtained with transiently transfected HEK293 cells. In these experiments, we expressed the truncated versions of Stat6 and C/EBPβ lacking the transcription activation domains (Stat6(Ad−) and C/EBPβ(Ad−)), either alone or in combination with the full length form of the other protein. No IL-4-induced transcription was observed when either one of the two truncated proteins (Stat6(Ad−) or C/EBPβ(Ad−)) was overexpressed. Furthermore, the combination of wild-type Stat6 and C/EBPβ(Ad−) or Stat6(Ad−) and wild-type C/EBPβ resulted in a transcription readout identical to the single, full length partner alone. The results obtained with both full length proteins are shown for comparison. These data clearly show that the activation domains of both Stat6 and C/EBPβ are absolutely essential to synergistically activate transcription from this IL-4-responsive element.
Dominant negative effects of Stat6(Ad−), C/EBPβ(Ad−), and C/EBPγ on IL-4-induced transcription
Previously, we showed that Stat6 lacking the activation domain (Stat6(Ad−)) is able to bind DNA and behaves as a potent dominant negative when transiently expressed in HepG2 cells that contain endogenous Stat6 and C/EBPβ (13). Based on the observation that C/EBPβ(Ad−) is unable to cooperate with Stat6 to activate transcription (Fig. 2), we were interested to see whether expression of this protein would also have a dominant negative effect on IL-4-induced transcription in HepG2 cells. C/EBPβ(Ad−) is able to bind DNA as well as the full length protein (data not shown). As Figure 3 shows, strong activation was seen in response to IL-4 treatment when the reporter construct was transiently transfected into these cells, even in the absence of any added activator. Consistent with previous data (13), a reporter carrying a mutation in the C/EBP binding site could not be activated in response to IL-4 stimulation.
Overexpression of C/EBPβ in the presence of the wild-type reporter resulted in a significant increase of both basal and IL-4-induced promoter activity. These results are consistent with the observations made in HEK 293 cells. In contrast, no IL-4-specific increase was seen with the mutant reporter in the presence of exogenous C/EBPβ, although a slight elevation in basal expression was observed. These data clearly show that the increase in IL-4-induced activation mediated by overexpressed C/EBPβ is absolutely dependent on the integrity of the C/EBP binding site adjacent to the Stat6 site of this IL-4-responsive element.
Overexpression of either Stat6(Ad−) or C/EBPβ(Ad−) completely suppressed IL-4-induced activation of the wild-type reporter construct. These experiments clearly support the observations that both proteins, Stat6 and C/EBPβ, have to be present to drive IL-4-induced transcription and that the integrity of both activation domains is required for this function.
C/EBPγ is a ubiquitously expressed member of the C/EBP family of proteins (26). It lacks an activation domain, and it has been implicated in suppression of transcription at genes known to be activated by other members of the C/EBP class of proteins, including C/EBPβ (18). C/EBPγ is able to bind to the site of the IL-4 response element of the ε promoter (data not shown). Hence, we were interested to see what effect C/EBPγ may have on IL-4-induced promoter activity. Figure 3,A shows that overexpression of C/EBPγ also suppressed IL-4-induced transcription, identical to the dominant negative effects seen with Stat6(Ad−) or C/EBPβ(Ad−). To further investigate the inhibitory effects of C/EBPγ on IL-4-induced transcription from this reporter element, we overexpressed increasing amounts of C/EBPβ in the presence or absence of C/EBPγ in HepG2 cells. Figure 3 B shows that C/EBPγ suppresses both the endogenous IL-4 response as well as the elevated IL-4 response mediated by overexpressed C/EBPβ. This observation raises the possibility that the balance between C/EBPβ and C/EBPγ may determine the extent of germline ε transcription in B cells following IL-4 treatment.
Both Stat6 and C/EBPβ bind the human ε promoter IL-4-responsive element
It has often been shown that adjacently bound transcription factors that functionally cooperate to activate transcription also physically associate to some extent, either in the presence or absence of their cognate DNA elements (27, 28). In an attempt to determine the mechanism underlying the functional synergy observed between Stat6 and C/EBPβ, we explored whether the two proteins facilitate each other in their interaction with the ε IL-4-responsive element. Human Stat6 overexpressed in insect cells does not bind DNA, indicating that it does not become tyrosine phosphorylated in these cells (data not shown). Therefore, the protein was purified to homogeneity and activated in vitro using recombinant Jak1, which also had been expressed and purified from insect cells. Human C/EBPβ was overexpressed in HEK293 cells and partially purified. The proteins were then incubated with radiolabeled DNA probes carrying either both protein binding sites or mutations within either one of the two sites (Fig. 4). Both proteins, Stat6 and C/EBPβ, bound the wild-type sequence independently, resulting in the formation of complex A (mediated by C/EBPβ) and complex B (mediated by Stat6). When the two proteins were mixed in the presence of the wild-type probe, an additional, more slowly migrating complex C was observed. Using Abs directed against either Stat6 or C/EBPβ we could show that both proteins are present in complex C. These experiments were done in the presence of excess probe. Hence, the relative amount of material present in complex C suggests that some degree of DNA binding cooperativity exists for these two proteins.
The lanes (DNA Probe II) in Figure 4 show that a mutation in the C/EBP binding site eliminates C/EBPβ binding without affecting Stat6 binding. Similarly, Stat6 binding is abolished if the probe carries a mutation in the Stat6 binding site, but C/EBPβ binding is unaffected (DNA Probe III). Formation of complex C is not seen with either one of the mutant probes.
In an attempt to show a physical association between these two proteins in the absence of DNA, we performed coimmunoprecipitation experiments with cellular extracts that contained Stat6 and C/EBPβ. These experiments did not show a physical association between the two proteins in either the presence or absence of IL-4 treatment (data not shown).
The presence of C/EBPβ reduces the fast dissociation rate of Stat6 bound to the ε IL-4-responsive element
We next investigated whether the dissociation rate between Stat6 and the ε IL-4-responsive element would be effected when C/EBPβ is bound to the adjacent site. A slower dissociation rate in the double occupancy situation would be indicative of a cooperative interaction between these two proteins while bound to DNA and might contribute to the functional synergy seen in our reporter assays. Band shift experiments, like those shown in Figure 4, are equilibrium measurements of the proteins affinity for its DNA binding site. To determine the dissociation rate of Stat6 and C/EBPβ when bound to DNA, we investigated the stability of the protein/DNA complexes, as shown in Figure 5. Stat6 and C/EBPβ were incubated either separately or together with the labeled ε probe. After a brief incubation period to allow the protein-DNA binding equilibrium to be reached with the labeled probe, an excess of unlabeled oligonucleotides (trap) carrying the same DNA sequence as the probe was added to the reaction mix. The time-dependent disappearance of the radiolabeled protein-DNA complex reflects the dissociation rate of the individual protein and the bound DNA probe (29). This can be visualized because subsequent rebinding events are far more likely to occur to the unlabeled DNA fragments, which are in vast excess over the labeled probe. Several repeats of this experiment showed that in the absence of C/EBPβ, Stat6 rapidly dissociates from its cognate DNA sequence; after 0.5 min, the majority of Stat6 was released from the radiolabeled oligonucleotide (Fig. 5, left panel). In contrast, C/EBPβ has a much slower dissociation rate. Even after 12 min, some C/EBPβ still remained bound to the radiolabeled oligonucleotide (right panel). When both proteins bind the same DNA probe (state of double occupancy), the dissociation rate of Stat6 from complex C is reduced. Complex C is still visible after 2 min (middle panel). The increased stability of Stat6 in the double occupancy complex suggests some physical contact between Stat6 and C/EBPβ or a change in the conformation of the DNA that stabilizes the binding of Stat6 when C/EBPβ is bound at the adjacent site. Interestingly, the increased residence time of Stat6 in the double occupancy complex was independent of the integrity of the activation domain; e.g., truncated proteins (Stat6 or C/EBPβ) lacking the activation domain also resulted in more stable protein/DNA complexes (data not shown).
Discussion
Activation of the ε promoter in B lymphocytes triggers the recombination event leading to class switching and expression of the IgE isotype. The ε promoter is activated in response to multiple signals, the most crucial being IL-4 (11). The IL-4 response element has previously been characterized and shown to contain binding sites for both Stat6 and C/EBP (11, 13). In this study, we explored possible mechanisms underlying the functional synergy seen between Stat6 and C/EBP in driving IL-4-induced expression of the germline ε sterile transcript. We investigated whether Stat6 would cooperate with different members of the C/EBP family. Previously, we showed that HEK293 cells lack Stat6 (13). Furthermore, only low amounts of C/EBP proteins can be detected in these cells using Western blots or gel shift experiments (data not shown). These data are substantiated by the observation that very low luciferase activity was obtained with a C/EBP-dependent reporter construct in the absence of C/EBP overexpression. Hence, transfection studies in HEK293 cells allowed us to determine the relative contribution of each protein (Stat6 or C/EBP) to activate transcription from the IL-4-dependent reporter. These experiments clearly show that overexpression of Stat6 alone is sufficient to drive a low level of IL-4-induced expression. In contrast, overexpression of C/EBPα or C/EBPβ elevates basal expression, but does not trigger IL-4-dependent transcription. However, both C/EBP isoforms, when expressed in the presence of Stat6, greatly increase the overall luciferase activity in response to IL-4 treatment, but C/EBPβ is significantly more effective than C/EBPα. The increase in IL-4-induced transcription seen when both Stat6 and C/EBPβ were overexpressed is far greater than the sum of each protein’s individual ability to drive transcription from this reporter. This superadditive effect is a defining feature of transcriptional synergy (27, 28). These results were most dramatic with C/EBPβ in the human B cell line BJAB, suggesting that C/EBPβ is limiting in these cells and that C/EBPβ rather than C/EBPα cooperates with Stat6 to drive IL-4-induced germline ε expression in B cells.
The functional synergy between Stat6 and C/EBPβ requires the activation domains of both proteins. Deletion of either the Stat6 or C/EBPβ activation domain results in a protein that can still bind to the element, but can no longer partner with the other protein to synergistically activate transcription. These data were further substantiated by overexpression studies in HepG2 cells. HepG2 cells express both Stat6 and C/EBP and are able to activate our IL-4-inducible reporter (13). Overexpression of either protein lacking an activation domain suppressed promoter activity, arguing for an almost equivalent role of C/EBPβ and Stat6 in driving transcription from the ε IL-4-responsive element. A dominant negative effect was also observed with C/EBPγ, a naturally occurring C/EBP isoform that lacks an activation domain. These results are particularly interesting as this C/EBP family member has been implicated as playing a negative regulatory role in antagonizing the function of other C/EBP proteins (18, 20). Hence, the balance between C/EBPβ and C/EBPγ may determine the amount of sterile transcript produced in response to IL-4 stimulation.
While no naturally occurring splice variant of Stat6 lacking the C-terminal activation domain has been yet characterized, such naturally occurring splice variants have been observed for Stats 1, 3, 4, and 5 (30, 31, 32). Studies with these STAT proteins and additional cooperating factors highlight similarities and differences when compared with our observations with Stat6 and C/EBPβ. The naturally occurring splice variant of Stat3, designated Stat3β, is missing most of the C-terminal activation domain. Yet, both the full length form of Stat3 (Stat3α) and Stat3β can cooperate with c-Jun to activate transcription from an IL-6-responsive element taken from the rat α2-macroglobulin gene (30). Similarly, both the full length form of Stat4 (Stat4α) and a splice variant lacking the C-terminal activation domain (Stat4β) can cooperate with c-Jun to activate transcription from an IL-12-responsive element taken from the IRF1 promoter (X. Xu and T. Hoey, unpublished observations). In both these examples, activation function appears to be provided by c-Jun, which binds at an adjacent site in each element (29, 30). Yet at other STAT-activated promoter elements, the activation domains of these proteins are clearly required (31, 32). In the case of IFN-α-induced gene expression, activation is mediated by the multiprotein complex IFN-stimulated gene factor-3, which is composed of Stat1, Stat2, and the nuclear protein p48. Here, both the full length Stat1 protein (Stat1α) or the variant (Stat1β) lacking the activation domain can participate in the complex, because all transactivation depends on Stat2 (33), whereas in the case of IFN-γ-induced gene expression, only the full length form of Stat1 is capable of activating transcription (31). Hence, there appear to be multiple mechanisms that allow STAT proteins to cooperate with other transcription factors. Specific interactions are probably dictated by the individual promoter context (31).
This is the first example of Stat6 functionally synergizing with another transcription factor to activate transcription from an IL-4-responsive element. In contrast, C/EBPβ has been shown to functionally synergize with a large number of transcription factors, including those playing important roles in immune cell function, such as NF-κB, AML1, Myb, and PU.1 (34, 35, 36, 37). In several of these examples, C/EBPβ has been shown to associate physically in some way with the other cooperating factor. These and other examples of transcription factors physically associating to mediate transcriptional synergy, as well as the proximity between the Stat6 and C/EBP binding sites that is conserved in both the mouse and human ε promoter, prompted us to search for a physical association between C/EBPβ and Stat6.
Our DNA binding studies showed that both proteins are able to bind the ε promoter element independently. However, based on mobility shift experiments conducted in the presence of excess DNA, simultaneous binding of both proteins appeared to be favored over single occupancy, suggesting some degree of facilitated DNA binding. To further investigate the possibility for a direct interaction between these two proteins, we measured the dissociation rates of Stat6 and C/EBPβ when bound to this DNA element. These experiments showed that in the double occupancy state, where both Stat6 and C/EBP are bound to the probe, the dissociation rate of Stat6 is slower than when it binds the probe alone. This increased stability of Stat6 binding under double occupancy conditions must be the result of some physical contact with the adjacently bound C/EBPβ, or possibly it is the result of a change in the conformation of the DNA that stabilizes the binding of Stat6 when C/EBPβ binds the adjacent site. Although we and others (11) have clearly shown that most isoforms of C/EBP are able to interact with the IL-4 response element of the ε promoter, it remains to be determined whether C/EBPβ binds to this site in vivo.
The requirement for the activation domains of both Stat6 and C/EBPβ secures the IL-4 responsiveness of the germline ε promoter. Yet, it is unclear how these activation domains interact to synergistically activate transcription. Stat6(Ad−) also exhibited a reduced dissociation rate from this element when C/EBPβ was bound (data not shown). Similarly, C/EBPβ(Ad−) also stabilized Stat6 binding, indicating that this effect is not due to the activation domains. Thus, while the change in dissociation rate may contribute to the synergistic effect seen in transcriptional activation, it is clearly not the main component. A more likely mechanism could be that both activation domains interact to adopt a unique conformation that allows them to cooperate with the basal transcription machinery more effectively than they can separately. Another possibility is that the two activation domains independently interact with other proteins in the basal machinery to achieve the same result. Alternatively, both of these mechanisms could be operative simultaneously. Either way, the decrease in dissociation rate of Stat6 in the presence of C/EBPβ is likely to increase the chances that the two activation domains will interact (either with themselves, or with additional factors) to activate transcription. Furthermore, the superadditive effect on transcription activation ensures the strong induction of gene expression at this immunologically important locus.
Acknowledgements
We thank Dr. Andrew Henderson for the human C/EBPγ clone and Dr. Steven McKnight for the human C/EBPα and C/EBPβ clones and Abs directed against different C/EBP isoforms. We also thank Keith Williamson for DNA sequencing, Carla Daniel for providing purified Stat6 and Jak1 proteins and for helpful discussions, and Drs. Cao Zhaodan, Greg Peterson, and Tim Hoey for critical comments on the manuscript.
Footnotes
Abbreviations used in this paper: Jak, Janus kinase; (Ad−), lacking the activation domain; C/EBP, CAAT/enhancer-binding protein; EMSA, electrophoretic mobility shift assay; HEK293, human embryonic kidney 293.