IFN-γ-activated transcriptional element (GATE)-binding factor 1 (GBF1) was identified as a transactivator that induces gene expression through GATE, a novel IFN-inducible element. Although it can induce gene expression, it is an extremely weak DNA-binding protein on its own. GATE also binds another transcription factor, C/EBP-β. Therefore, we explored whether GBF1 physically interacts with C/EBP-β to induce IFN-γ-regulated transcription. In response to IFN-γ, C/EBP-β undergoes phosphorylation at a critical ERK1/2 phosphorylation motif. Mutational inactivation of this motif and/or interference with the ERK1/2 activation prevented the IFN-γ-induced interactions between GBF1 and C/EBP-β. A 37-aa long peptide derived from the GBF1 protein can associate with C/EBP-β in an IFN-inducible manner. These results identify a converging point for two transactivators that exert their effects through a single response element. Together, our studies identify a novel regulatory mechanism that controls IFN-induced transcription.
The IFN family of cytokines suppresses the growth of infectious pathogens and neoplastic cells (1, 2, 3). They regulate a number of different physiological processes including cytokine and chemokine synthesis, mRNA translation (4), RNA and protein stability (5, 6), Ag presentation (7), nuclear trafficking (8), cell differentiation (9, 10), cell division, and apoptosis (11, 12) by modulating gene expression. The most thoroughly studied of the IFN-induced gene regulatory pathways is the JAK-STAT pathway (13, 14). In the IFN-α/IFN-β-induced pathways, JAK1 and Tyk2 induce the tyrosine phosphorylation of STAT1 and STAT2, which in association with the IFN-gene regulatory factor-9 (IRF9,3 also known as ISGF3γ or p48) protein form the multimeric transcription factor, ISGF3, for inducing the genes that contain the IFN-stimulated response element. IFN-γ induces the tyrosine phosphorylation of only STAT1, using JAK1 and JAK2. The STAT1 dimer stimulates the transcription of genes that possess the IFN-γ-activated site (GAS). IFN-induced STAT activity is rapidly down-regulated (usually within the first 1–2 h) after the initial stimulus by feedback inhibitors (15).
IFN-γ is a central regulator of innate and adaptive immunities. Its targets include the genes that mediate antimicrobial, antiviral, Ag presentation, and antitumor responses (1, 2, 3). The temporal and functional diversities in these responses and the continued activation of transcription, even at a time STATs are down-regulated by feedback mechanisms, suggest the activation of several other IFN-induced signaling pathways. The IRF9 protein is critical for regulating antiviral defenses (16) and a vast section of the IFN-induced genes (17). Down-regulation of this gene in virus-infected (18) and in transformed cells causes a severe loss of IFN-induced responses (19). Therefore, we studied the regulation of the IRF9 gene by IFN-γ. Furthermore, its induction occurs in a temporally delayed manner, in contrast to those of other IFN-stimulated genes and some IRFs (17). These studies identified a novel IFN-γ-responsive element, the IFN-γ-activated transcriptional element (GATE) (20). Our subsequent studies identified that transcription factor C/EBP-β regulates IFN-γ-induced expression of the IRF9 gene through GATE (21).
The C/EBPs are a family of six structurally similar, but functionally and genetically distinct, transcription factors. C/EBPs contain a characteristic leucine zipper and a basic region at their C termini, collectively called the bZIP domain (22, 23). The bZIP domain is essential for DNA binding, and homo- and heterodimerization among various members of this family. Among these proteins, C/EBP-β (NF-IL6, IL6-DBP, NF-M) exhibits a remarkable functional diversity (24). In addition to the genes involved in energy metabolism, it also regulates IL-6 and IL6-induced expression of the IL-1, IL-8, TNF-α, and G-CSF genes involved in acute phase responses (25). A number of defects in cytokine synthesis were reported recently in C/EBP-β−/−macrophages (26). Deletion of C/EBP-β in mice causes defects in macrophage-driven tumoricidal and bactericidal activities (27), Th1 immune responses (28, 29), female infertility (30), glucose homeostasis (31), the development and differentiation of hepatocytes (32), myelomonocytes (33), adipocytes (34), and neurons (35, 36). Given the functional diversity of C/EBP-β, it is conceivable that its association with different cellular factors in a gene context and signal-specific manner regulate its activity. Indeed, C/EBP-β can interact with transcription factors outside its family, such as NF-κB (37), pRB (38), Sp1 (39), and STAT5 (40) to regulate cellular functions.
We have reported earlier the identification of a novel IFN-γ-regulated transcription-activating factor, GATE-binding factor 1 (GBF1), which induced GATE-driven transcription (41). However, on its own, GBF1 exhibited poor DNA-binding activity. This suggested that GBF1 might interact with another DNA-binding protein to regulate GATE-dependent gene expression. Hence, we investigated the mechanism by which GBF1 activates transcription. Given our observations that C/EBP-β and GBF1 could induce transcription from the same element, it was of interest to know whether these two proteins interact with each other. Therefore, we studied IFN-inducible interactions between GBF1 and C/EBP-β. We show here that the ERK phosphorylation site of C/EBP-β is required for the IFN-γ-induced binding of GBF1. We have also identified a 37-aa motif of GBF1 that is required for its interaction with C/EBP-β. Together with these data this study identified a novel mechanism of IFN-γ-induced transcriptional control.
Materials and Methods
Recombinant murine IFN-γ was purchased from PBL Biomedical Laboratories. Mouse mAbs against flag-epitope tag and actin were obtained from Sigma-Aldrich. Mouse polyclonal Abs against GBF1 (41) and rabbit polyclonal Abs against C/EBP-β (Santa Cruz Biotechnology) and GAL4-DBD were used in these studies. The anti-ERK2 and disphosphorylated ERK (ppERK) 1/2 Abs and an Ab that detects a phospho-T189 form (pT189) of C/EBP-β were purchased from Cell Signaling Technology.
Cell culture and plasmids
Wild-type and mutant mouse embryonic fibroblasts (MEFs) were grown in DMEM with 10% FBS. The murine macrophage cell line RAW (RAW264.7) was grown in RPMI 1640 supplemented with 10% FBS (20). During IFN-γ treatment cells were grown in media containing 1% serum. Wild-type GBF1 cloned into mammalian expression vector pCMV2-flag was described earlier (41). The wild-type (41) and GBF1 mutants were PCR amplified using the primers shown in Table I and cloned into the EcoRI and BamHI sites of mammalian expression vector pCMV2-flag. All proteins expressed from this vector contained an N-terminal flag-tag. Sequence and expression confirmed mutants were used for this study. The pCMV-FA-DBD vector, which carries the DNA-binding domain (DBD) of yeast transcription factor GAL4, was purchased from Stratagene and used in some experiments. The region between aa 174 and 211 was amplified using the primers: 5′-CGGGATCCCTGGTGTCGGGGCAGC-3′ (forward) and 5′-CGCGAATTCGAGCCAGTATTTATTGCCG-3′ (reverse). GBF1-specific sequences in these primers are underlined. The other sequences at their termini correspond to the BamHI and EcoRI restriction sites, respectively. Following PCR amplification, the product was purified, digested with BamHI and EcoRI, and cloned into the similarly digested pCMV-FA-DBD vector. The resultant recombinant construct will have an N-terminal GAL4-DBD with C-terminal GBF1 sequences. The expression of the chimerical protein was detected in Western blot analyses where the blots were probed with GAL4-DBD specific Abs (Santa Cruz Biotechnology).
|Primer Sequence (5′→3′) .||Direction .||PCR Product Size (bp) .||Mutant .||Protein (aa) .|
|Primers for pCMV2-flag vector (N-terminally flag-tagged protein expression)|
|1 CGGAATTCCGGCAGAGCGAACATGGCCCCG||→||1133||Wild type||377|
|Primer Sequence (5′→3′) .||Direction .||PCR Product Size (bp) .||Mutant .||Protein (aa) .|
|Primers for pCMV2-flag vector (N-terminally flag-tagged protein expression)|
|1 CGGAATTCCGGCAGAGCGAACATGGCCCCG||→||1133||Wild type||377|
Primers 1–11 were used as forward primers in combination with primer no. 13 as the reverse primer to generate the indicated mutants. Primers 1 and 12 were used for generating CΔ150. The upstream and downstream primers have EcoRI and BamHI sites, respectively, for facilitating an in-frame cloning of the products into the EcoRI and BamHI sites of pCMV2-falg vector. In the primer sequences, italicized letters represent restriction sites for EcoRI and BamHI; bold letters represent gene-specific sequences.
C/EBP-β and its mutants were provided by P. F. Johnson (National Cancer Institute-Frederick, Frederick, MD) (23). The P4 construct was generated (20) by inserting a 74-bp element of the IRF9 gene, harboring GATE, upstream of the SV40 early promoter in pGL3 promoter vector. The CBS-Luc reporter contained three tandem copies of the consensus C/EBP-β-binding site (CBS) upstream of the SV40 early promoter.
Gene expression and immunoprecipitation (IP) analyses
Western blot analyses, transfection, β-galactosidase and luciferase assays, and SDS-PAGE, were performed as described in our earlier publications (20). In a typical transfection assay, 250 ng of the luciferase reporter was mixed with 100 ng of pCMV- β-galactosidase reporter and 100 ng of the effector plasmids and transfected using the Lipofectamine plus reagent (Invitrogen Life Technologies) per manufacturer’s suggestions. No more than 1 μg of DNA of total DNA was transfected into the cells. Where required, the total amount of transfected DNA was kept constant by including empty pcDNA3.1 vector. Luciferase activity was normalized to that of a cotransfected CMV-β-galactosidase reporter. Triplicate transfections per sample were performed to evaluate the statistical significance of the differences between various treatment groups. Experiments were repeated at least three times to obtain consistent results.
IP analyses were conducted as described elsewhere (42). Briefly, 150 μg of total cellular lysate from each transfectant was incubated overnight with the desired Ab at 4°C and then with protein A-agarose for an additional 90 min. The products were washed extensively and separated on a 10% SDS-PAGE. They were then Western blotted onto polyvinylidene difluoride membranes. These blots were probed with appropriate first and second Abs. The second Abs were labeled with HRP. The blots were developed using a commercially available Super signal ECL kits (Pierce).
Chromatin IP (ChIP) assays
These assays were performed as described earlier (43). Briefly, cells (1 × 108) were stimulated with IFN-γ for 8 h, and chromatin was cross-linked using paraformaldehyde. Nuclei were isolated and chromatin was sheared into ∼1–2-kb fragments using a Bronson Sonicator fitted with a microtip probe. After removing the debris, soluble chromatin was subjected to IP with either C/EBP-β- or GBF1-specific Abs. After extensively washing the immunoprecipitated products, cross-links were reversed and the DNA was extracted using phenol-chloroform. The resultant DNA was used for 32 cycles of PCR with the following primers specific for mouse IRF9 promoter: 5′-AAGGTGCTACTGCTGACTGAGG-3′ and 5′-AAGGGCGGACGTGAAGAAATGG-3′ (20). PCR with these primers yields a 443-bp product. DNA extracted from an aliquot of the initial soluble chromatin was used for input control. Control IP reactions performed with nonspecific IgG did not yield this PCR product (data not shown).
In vitro translation
In vitro transcription and translation of the GBF1 and C/EBP-β cDNAs (21, 41) cloned in the pGEM-7Zf vector was conducted using commercially available kits (Promega) as suggested by the manufacturer. The products were translated in presence of [35S]methionine (Amersham). Because the molecular weights of GBF1 and C/EBP-β are very close (38 and 35 kDa, respectively), and are difficult to resolve on the gels, we have used only one of the proteins as labeled product in the IP reactions. The second protein in these reactions was generated after translation of the RNA in the presence of unlabeled methionine. Equal volumes of the cold and radiolabeled proteins were incubated for 1 h at 37°C and IP was performed with Ab against the unlabeled protein. The IP products were separated on a 10% SDS-PAGE, dried and fluorography was performed to detect the bands.
GBF1-modulates IFN-γ-regulated gene expression
To demonstrate that GBF1 can promote IFN-γ-induced gene expression, we transfected an empty expression vector pCMV-flag2 or the same vector expressing the wild-type GBF1 cDNA along with P4-Luc, an IFN-responsive reporter that contained GATE into untransformed MEFs. Following this, cells were exposed to IFN-γ for 12 h and luciferase activity was determined. The luciferase data were normalized to those of internal transfection control, β-galactosidase, and presented in Fig. 1,A. While IFN-γ induced transcription of this reporter, it was robustly induced further in presence of GBF1. The stimulation of transcription was due to the expression of GBF1 protein (Fig. 1, B and C). We have also shown earlier that the endogenous IRF9 gene and promoters carrying up to 1 kb of the murine IRF9 promoter are also similarly induced in the presence of GBF1 (41). Interestingly, GBF1 had no effect on STAT1-dependent, GAS-driven, IFN-γ-inducible transcription, indicating its specificity (41).
To further demonstrate its transactivating effects, we generated two mutant forms of GBF1 using PCR. The resultant inserts were expressed as N-terminally flag-tagged proteins by cloning them into the mammalian expression vector pCMV2-flag. These two mutants, named CΔ150 and NΔ226, lacked the C-terminal and N-terminal sequences, respectively. The mutants were sequenced and their expression was verified following transfection into MEFs. Both mutants expressed to a comparable extent and their mobilities on SDS-PAGE are in conformation with their sequence coordinates (Fig. 1 D).
The mutants were coexpressed with P4-Luc, and their ability to stimulate GATE-driven gene expression was quantified and compared with that of wild-type GBF1. As shown in Fig. 1 F, coexpression of wild-type GBF1 strongly stimulated transcription. The CΔ150 mutant drove some significant amount of IFN-induced gene expression, although several-fold lower than the wild-type. In contrast, the NΔ226 failed to stimulate the reporter. Thus, the C-terminal 150 aa are insufficient to stimulate IFN-γ-induced gene expression through GATE.
Cooperation between C/EBP-β and GBF1 in regulating gene expression
In light of our previous observations that both GBF1 and C/EBP-β could up-regulate gene expression through GATE in response to IFN-γ, we first determined whether C/EBP-β and GBF1 synergistically induced GATE-driven transcriptional response. Because the levels of endogenous C/EBP-β are induced by IFN-γ (21), it would have been difficult to distinguish whether the enhancement of GATE-driven gene expression was due to an increase in the levels of C/EBP-β protein or its association with GBF1. Therefore, to avoid the contribution of newly induced endogenous C/EBP-β to gene induction, we conducted these experiments in C/EBP-β−/− MEFs. In addition, these cells express a low level of GBF1 (data not shown), which permits one to readily determine the interactions between C/EBP-β and GBF1. We first measured the IFN-γ-induced expression of P4-Luc in the presence and absence of these proteins (Fig. 2,A). In the presence of empty expression vector no significant induction of the luciferase activity occurred. However, while the transfection of C/EBP-β alone could strongly induce transcription, GBF1 alone could only stimulate a marginal induction. More importantly, when GBF1 and C/EBP-β were present together, a significantly stronger induction of the reporter gene was observed. These differences were not due to differential expression of the transfected genes. Western blot analyses clearly showed that both of these proteins expressed to a comparable extent in the lysates (Figs. 2, B and C). These data show that GBF1-driven GATE-dependent transcription is C/EBP-β-dependent.
Because the above experiment uses overexpression as a means for assessing the cooperative gene regulatory effects of C/EBP-β and GBF1, we used an antisense GBF1 construct to determine whether it could down-regulate C/EBP-induced gene expression through GATE in response to IFN-γ. We first tested whether this construct could down-regulate the expression of GBF1. MEFs were transfected with an empty vector or antisense GBF1 expression vector. Cellular lysates were prepared and Western blotted using Abs against the GBF1 protein. Coexpression of the antisense GBF1 construct depressed ∼80% of GBF1 protein expression when compared with the vector control (Fig. 2,D). These lysates were also blotted for monitoring the expression of C/EBP-β and actin. Expression of antisense GBF1 did not affect either of these proteins (Fig. 2, E and F). These data indicate the specificity of the antisense GBF1 construct.
In the next experiment, we transfected MEFs with GATE-Luc in the presence or absence of antisense GBF1 constructs. Following this, cells were exposed to IFN-γ and luciferase activity was determined. The empty expression vector had no effect on IFN-inducible expression of GATE-Luc. In contrast, the antisense expression vector blocked gene expression, significantly (Fig. 2,G). These data suggest that GBF1 is important for the optimal induction of GATE-driven transcription. Similar results were obtained when this experiment was repeated with murine RAW macrophage cells (data not shown). To demonstrate the specificity of this effect, we used another reporter gene, CBS-Luc, whose expression is driven by consensus CBSs (Fig. 2 H). This reporter gene expresses at high levels in the absence of any stimulus. The antisense GBF1 construct had no effect on this reporter.
IFN-γ promotes physical interactions between GBF1 and C/EBP-β
Although the above results show a functional collaboration between C/EBP-β and GBF1, they do not show whether such interaction is physical. To investigate this issue we cotransfected GBF1 and C/EBP-β into C/EBP-β−/− cells and analyzed their interactions using IP experiments. These experiments were performed in the absence and presence of IFN-γ. Abs specific for C/EBP-β were used for IP and the products were probed with flag-GBF1-specific Abs following a Western blot transfer (Fig. 3,A). As expected, no protein was seen in mock-transfected cells. In the absence of C/EBP-β, no GBF1 protein was detected in these blots, although an ample amount of GBF1 protein is expressed in the cells (Fig. 3,B). In contrast, in the presence of C/EBP-β, GBF1 was readily coimmunoprecipitated in the reactions. Although a low basal level of interaction was seen in the untreated cells, it was further augmented by IFN-γ. All the transfectants had a comparable expression of the C/EBP-β protein (Fig. 3 C).
Because the above experiments show an interaction between transfected gene products, it is important to ensure whether endogenous GBF1 and C/EBP-β proteins interact with each other in normal cells. For demonstrating this, MEFs and RAW264.7 macrophage cell lines were treated with IFN-γ and lysates were immunoprecipitated with C/EBP-β-specific Abs. The products were subjected to Western blot analyses with GBF1-specific Abs. As shown Fig. 3,D, the endogenous GBF1 and C/EBP-β proteins interact with each other and this interaction is further enhanced by IFN-γ treatment. Consistent with our earlier report, IFN induced the endogenous C/EBP-β protein in both these cell types (Fig. 3,E). GBF1 levels were comparable at this time point in both cell types (Fig. 3,F). Although this experiment does not show whether IFN-induced enhancement of interactions between GBF1 was due to an increase in the protein level or posttranslational modification(s), the experiment shown in Fig. 3 A clearly suggests that IFN-induced posttranslational modification(s) could promote these interactions.
To determine whether these interactions occur directly, we generated the C/EBP-β and GBF1proteins using in vitro translation. We translated both these proteins separately with and without 35S labeling. The proteins were then mixed and immunoprecipitated with either GBF1- or C/EBP-β-specific Abs. In the first set, unlabeled GBF1 protein was incubated with 35S-labeled C/EBP-β protein for 1 h 37°C. These reactions were immunoprecipitated with GBF1 Abs and products were separated on SDS-PAGE. The gel was dried and subjected to fluorography to detect the presence of labeled C/EBP-β protein. As shown in Fig. 3,G, the C/EBP-β protein coimmunoprecipitated with GBF1 protein. Similarly, we conducted a converse experiment in which unlabeled C/EBP-β was mixed with 35S-lableled GBF1 and incubated for 1 h. These reactions were immunoprecipitated with C/EBP-β-specific Abs and the products were analyzed after fluorography. Indeed, GBF1 coimmunoprecipitated with C/EBP-β (Fig. 3 H). In the control reactions no such coimmunoprecipitation of either C/EBP-β or GBF1 was observed. It should be noted that these interactions are very weak because it required a 5-day exposure to detect the bands.
To further demonstrate a functional collaboration between GBF1 and C/EBP-β, we performed ChIP assays, which can detect protein interactions at the endogenous IRF9 promoter, using wild-type (C/EBP-β+/+) and C/EBP-β−/− MEFs (Fig. 3,I). Cells were stimulated for 8 h with IFN-γ, chromatin was cross-linked, soluble chromatin was prepared and immunoprecipitated with Abs specific for C/EBP-β and GBF1. DNA present in the immunoprecipitated products was used as a template for PCR amplification with primers specific for the IRF9 promoter. The products were resolved on an agarose gel, stained with ethidium bromide to detect a 443-bp band corresponding to the IRF9 promoter. As shown in Fig. 3 I, these primers detected a weak basal level of the IRF9 promoter, when immunoprecipitated with C/EBP-β, in the untreated C/EBP-β+/+ cells. IFN-γ potently stimulated the binding of C/EBP-β to the IRF9 promoter, as revealed by a stronger PCR product. Such differential binding of C/EBP-β with the IRF9 promoter in the untreated and IFN-γ-treated cells was not due to different amounts of chromatin input into these reactions, since the IRF9 promoter was amplified equivalently in the controls (see the bottom panels of the figure). The GBF1-specific Abs barely immunoprecipitated the IRF9 promoter from untreated cells. However, IFN-γ treatment strongly promoted the association of GBF1 with the IRF9 promoter as revealed by a strong PCR product. No such IP of the IRF9 promoter was observed in the control IP reactions performed with nonspecific Abs (data not shown). Thus, IFN-γ stimulates the recruitment of GBF1 to the IRF9 promoter. More importantly, such an inducible recruitment of GBF1 to the IRF9 promoter was dependent on C/EBP-β, because GBF1 was not found in a complex with the IRF9 promoter in C/EBP-β−/− MEFs, despite stimulation with IFN-γ. As expected, no C/EBP-β was found at the promoter in these mutant cells. Together, these observations show an IFN-γ-driven cooperative binding of GBF1 and C/EBP-β to the IRF9 promoter and, hence, a strong transcriptional activation.
IFN-induced interactions between C/EBP-β and GBF1 are sensitive to the inhibitors of ERK1/2 activation
The above studies suggest that IFN-γ-induced signals promote interactions between GBF1 and C/EBP-β. Indeed, our previous studies have shown that IFN-induced GATE-driven transcription is sensitive to the inhibitors of MAPK, ERK1/2, activation (44). Therefore, we first examined whether inhibition of ERK1/2 activation disrupts the IFN-induced associations between C/EBP-β and GBF1. C/EBP-β−/− MEFs were transfected with C/EBP-β and GBF1 expression vectors and treated with IFN-γ in the presence and absence of U0126, a known inhibitor of ERK1/2 activation (45). Cell extracts were immunoprecipitated with C/EBP-β-specific Abs and Western blotted with GBF1-specific Abs. As seen in other experiments a basal interaction between C/EBP-β and GBF1 was further enhanced strongly by IFN-γ treatment (Fig. 4,A). Such inducible association between these proteins was suppressed by U0126, despite a comparable expression of C/EBP-β and GBF1 proteins between various treatments (Fig. 4, B and C). To ensure that this effect was indeed due to an inhibition of ERK1/2 activation, the same lysates were Western blotted and probed with Abs specific for ppERK, which detect the activated diphosphorylated forms of ERK1 and ERK2 (Fig. 4,D). As expected, U0126 inhibited ERK1/2 activation. The difference in the ERK activity was not due to a depletion of total ERK levels in the presence of U0126 (Fig. 4 E).
ERK-dependent interactions between C/EBP-β and GBF1 were further confirmed in another manner. Our recent studies (46) have shown that IFN-γ-stimulated ERK activation and inducible expression of GATE-driven reporters are completely inhibited in cells lacking the gene for MEK kinase 1 (MEKK1). Therefore, we next determined whether IFN-γ-induced association between C/EBP-β and GBF1 occurred in cells lacking the MEKK1 gene. Isogenic MEKK1+/+ and MEKK1−/− MEFs were stimulated with IFN-γ and the cellular lysates were immunoprecipitated with C/EBP-β-specific Abs. The products were Western blotted with Abs specific for GBF1. While GBF1 coimmunoprecipitated with c/EBP-β, IFN-γ markedly stimulated it further in the wild-type cells (Fig. 4,E). No such IFN-γ-induced association between C/EBP-β and GBF1 was observed in the MEKK1−/− MEFs. This difference was not due to differential expression of GBF1 and C/EBP-β proteins in these two cell types (Fig. 4, F and G). We have also shown that IFN-induced ERK1/2 activation was defective in MEKK1−/− MEFs (Fig. 4, H and I). Thus, activation of ERK1/2 is critical for promoting the IFN-induced associations between GBF1 and C/EBP-β.
An intact GTPS motif in the regulatory domain 2 of C/EBP-β is required for an interaction with GBF1
Our earlier report identified a critical role for an ERK consensus phosphorylation site, present in the RD2 domain of C/EBP-β, in driving IFN-γ-induced transcription through GATE. C/EBP-β mutants lacking this domain inhibited IFN-γ-induced transcription through GATE (44). Because IFN-γ promoted associations between GBF1 and C/EBP-β leading to transcriptional activation, we next examined whether this motif was necessary for promoting the physical interaction of the two proteins. For this purpose C/EBP-β−/− cells were transfected separately with expression vectors carrying wild-type C/EBP-β, and its mutants, Mut 1 (184SSSS187 to 184AAAA187) and Mut 2 (188GTPS191 to 188GAAA191). Fig. 5,A shows a diagram and amino acid sequence changes introduced into these mutants. Following cotransfection of these mutants with a GBF1 expression vector, these cells were treated for 2 h and the lysates were immunoprecipitated with C/EBP-β-specific Abs. Western analysis was performed with flag-GBF1-specific Abs. As shown in Fig. 5,B, Mut2 was unable to associate with GBF1, despite IFN-γ treatment. In contrast, mutant1, which lacked the adjacent serine residues, was able to bind to GBF1, similar to the wild type. These differences in interactions cannot be attributed to differential expression of the C/EBP mutants and GBF1 (Fig. 5, C and D). Thus, the GTPS motif is necessary for promoting interactions between C/EBP-β and GBF1.
To confirm that phosphorylation occurred at the GTPS motif, we probed the Western blots with an Ab that specifically detects the phosphorylated form of C/EBP-β. This Ab only detects the phospho-T189 residue of the protein. As expected, this Ab did not detect any protein in vector-transfected cellular lysates (Fig. 5,A). However, in wild-type C/EBP-β and in mut 1 transfected cells this Ab detected the phosphorylated protein. Although a baseline phosphorylation can be seen in unstimulated cells, it was significantly induced by IFN-γ. More importantly, no such IFN-induced phosphorylation of Mut 2 was detected. The phosphorylation-defective Mut2 did not bind to GBF1 (Fig. 5 A). Thus, phosphorylation at T189 appears to couple C/EBP-β and GBF1.
GBF1 domains required for interaction with C/EBP-β
To identify the regions of GBF1 required for its association with C/EBP-β, we initially used the GBF1mutants CΔ150 and NΔ226 (see Fig. 6,A for a schematic diagram). These mutants were expressed as N-terminally flag-tagged proteins for detecting their expression. The mutants were cotransfected into C/EBP-β−/− MEFs with wild-type C/EBP-β and their interactions were studied using the coimmunoprecipitation strategies. Although the CΔ150 mutant bound to C/EBP-β, the NΔ226 mutant did not (Fig. 6,B). No differences in the expression of either the GBF1 mutants or the C/EBP-β protein were noticed under these conditions (Fig. 6, C and D). These data clearly indicate that N terminus of GBF1 contains a CID.
To further localize the CID, we have generated a number of truncated mutants lacking various lengths of the N terminus (Fig. 6,A). These mutants contained deletions ranging from 16 to 277 aa from the N terminus. Mutants NΔ16 to NΔ179 bound to C/EBP-β in an IFN-dependent manner. Deletion mutants lacking the first 179 aa bound to C/EBP-β normally (Fig. 6,E). Mutant NΔ211, NΔ247, and NΔ277 aa failed to bind C/EBP-β. These differential interactions could not be attributed to a variable expression of the GBF1 mutants and C/EBP-β (Fig. 6, F and G). Altogether these data indicate a region between aa 179 and 211 contains a CID.
Lastly, to demonstrate the minimal CID of GBF1, we generated a fusion peptide, in which we fused the region between aa 176–211 of GBF1 to the DBD of yeast transcription factor GAL4. Constructs carrying the DBD and chimerical peptide genes were coexpressed with full-length C/EBP-β in C/EBP-β−/− MEFs and IP was performed with c/EBP-β-specific Abs (Fig. 6,H). The products were Western blotted and probed with GAL4-specific Abs. As expected, the DBD alone did not bind to full-length C/EBP-β. However, the fusion peptide bearing GBF1 readily interacted. IFN-γ, weakly but significantly, enhanced the binding. It should be noted, however, that this interaction was very weak compared with the binding of full-length GBF1. It could only be seen after a prolonged exposure of the blots (20 min) under the conditions of ECL detection. The binding of full-length protein can be readily detected with <2 min exposure (Fig. 6 B). These weaker interactions could be a result of the chimerical nature of the peptide and/or its incomplete folding. Nonetheless, these data identify a potential region for C/EBP-β and GBF1 interactions. Further deletions were not tolerated in this region, indicating it as the minimal motif of GBF1 required for binding to C/EBP-β.
GBF1 was identified as a factor that regulated transcription from the IFN-γ-induced enhancer element GATE (41). Although it potently activated IFN-γ-induced expression of endogenous IRF9 and the luciferase reporters driven by the IRF9 promoter and the minimal GATE, rGBF1, on its own, formed no detectable complexes with GATE. Interestingly, it was able to form a specific complex with GATE, when expressed in the mammalian cells and stimulated with IFN-γ (41). These results suggested that it might interact with another cellular protein to bind to the target element. A similar property was observed for many transcription factors belonging to the IRF family (47). Although these proteins do not bind to DNA on their own in vitro but can effectively activate gene expression by binding to DNA in vivo. These include IRF3, IRF5, IRF7, and IRF8. These proteins may associate with other cellular factors to regulate gene expression. For example, IRF8 can associate with IRF1, IRF2, Pu.1, and Ets related factors (48). Such interactions not only expand the transcriptional repertoire induced by a single transcription factor but also contribute to empirical regulation of specific genes.
We have previously shown that C/EBP-β also binds to GATE and activates transcription. Therefore, it is reasonable to presume prima facie that an interaction between GBF1 and C/EBP-β can potently stimulate gene expression. If this were the situation, suppression of GBF1 expression should down-regulate GATE-driven transcription and GBF1 overexpression should increase the level of this transcription. Consistent with this notion, the antisense GBF1 construct significantly inhibited GATE- driven gene expression despite the normal expression of C/EBP-β protein (Fig. 2,G), and conversely, overexpression promoted gene expression (Fig. 1,A). Interestingly, antisense GBF1 did not inhibit gene expression driven by the consensus CBS (Fig. 2,H). This result suggests that enhancer elements also play an important role in regulating GBF1 effects. Indeed, our previous studies (41) have shown that GBF1 did not alter gene expression driven by GAS and IFN-stimulated response element, the other IFN-responsive elements. We have conducted these studies in normal MEFs to demonstrate such effects are not a result of oncogenic transformation of cells. A stronger transcriptional activation by GBF1 requires both N and C termini, although some minimal transcription can be driven by the NΔ266 (Fig. 1,B). Furthermore, although the CΔ150 mutant bound to C/EBP-β, like the wild-type GBF1 protein (Fig. 6), it did not transactivate gene expression as well as the wild-type GBF1 (Fig. 1 F). These observations suggest that the region between aa 151 and 277 plays an important role in regulating transactivation. The significance of these results will be investigated in future studies.
A direct interaction between C/EBP-β and GBF1 appears to account for the transcriptional collaboration between GBF1 and C/EBP-β. IFN-γ treatment seems to enhance these interactions. This suggestion was further supported by ChIP assays (Fig. 3,I). GBF1 was recruited to the IRF9 promoter in an IFN-γ-dependent manner only in the presence of C/EBP-β. Our previous studies have shown that C/EBP-β levels are induced by IFN-γ treatment. Studies using the wild-type cells do not distinguish whether the enhanced interactions between GBF1 and C/EBP-β are a result of increased C/EBP-β levels or due to a posttranslational modification. However, experiments using the C/EBP-β−/− MEFs clearly show that enhancement of interactions between GBF1 and C/EBP-β is due to a signal-induced posttranslational modification(s). This observation was confirmed by several independent experiments. Pharmacologic (using U0126) and genetic blockade (in MEKK1−/− cells) of ERK1/2 activation ablated IFN-γ-induced association of GBF1 with C/EBP-β. Such inhibitory effects were not at least due to a decrease in the physical levels of GBF1 to C/EBP-β proteins (Fig. 4). Furthermore, C/EBP-β mutants lacking the ERK1/2 phosphorylation site, 188GTPS191, failed to associate with GBF1 in the presence of IFN-γ. These results suggest that ERK1/2-induced modifications at the GTPS motif are critical for promoting IFN-induced associations between GBF1 and C/EBP-β. Consistent with this notion, we have shown that IFN-γ-induced phosphorylation at the T189 residue of the 188GTPS191 motif occurred normally with wild-type and mut1 forms of C/EBP-β, but not with mut2 (Fig. 5). Previous studies have shown that this site can be directly modified by activated ERKs (49). These results support our hypothesis that phosphorylation serves as a coupling signal between C/EBP-β and GBF1 proteins. In fact, our earlier studies showed that C/EBP-β mutants lacking this site fail to activate GATE-driven transcription. The basal binding seen between C/EBP-β and GBF1 in the absence of IFN-γ treatment suggests a low affinity interaction of GBF1 with a different site on the C/EBP-β protein. It can also be due to induction of phosphorylation by other serum-derived growth factors. IFN-γ-induced phosphorylation at the GTPS motif may induce a conformational change in the C/EBP-β protein and opens up another higher affinity site for GBF1 interaction. This probably is a reason for a weaker interaction seen between the in vitro translated proteins (Fig. 3, G and H). Our deletion analyses have identified a 37-aa domain, VSGQPLEEIITYYPAMKAVNDQGKEVTEFGNK YWLMLNE, located between residues 174–211 of GBF1 possess a potential CID. This region has several leucine, isoleucine, and valine residues, which may form a potential α-helix, that permit the interaction with C/EBP-β. This same region is very well-conserved (99%) between mouse, human, and monkey GBF1 sequences. Although it is possible that the MAPK signaling cascades also modify GBF1, we have not found any homologous sites to those of MAPK phosphorylation sites in this protein. In particular, although some serine residues are present in the CID of GBF1, none of them exhibit a homology to the sites phosphorylated by known proline-directed kinases, such as the MAPKs.
It is known that C/EBP-β interacts with a number of other transcription factors, such as NF-κB (37), pRB (38), Sp1 (39), and STAT5 (40). Such interactions expand not only the transcriptional repertoire of C/EBP-β but also diversify the physiological roles of C/EBP-β. We suggest that interaction with GBF1 is one part of its complex mechanisms of action. Indeed, the deletion of the C/EBP-β gene resulted in a number of complex phenotypes, such as a loss of macrophage-driven tumoricidal and bactericidal activities (27), Th1 immune responses (29), female fertility (30), glucose homeostasis (31), defects in the development and differentiation of hepatocytes (32), myelomonocytes (33), adipocytes (34), and neurons (35, 36). The C/EBP-β−/− mice also develop lymphoproliferative disorders (28). In skin keratinocytes C/EBP-β cooperates with activated ras to transform cells (50). It is likely that GBF1 may contribute to one or more of these diverse phenotypes, which can be realized only after deletion of this gene in mice.
Interestingly, GBF1 protein is also present in the cytoplasm and perinuclear membrane. A recent study showed that its human sequence homologue possesses PGE2 synthase activity (51). Based on the above study, it is tempting to speculate that such activity may be necessary for a transcriptional control, although we have not experimentally determined it in the case of mouse GBF1. This issue is being investigated in a separate study. However, a recent study shows that PGE2 activates IL-8 gene expression through the activation of AP1 and C/EBP-β (52). An important question is why would an enzyme like this participate in transcriptional activation? It is interesting to note from recent studies that a number of “nonclassical factors” participate in transcriptional regulation. For example, the steroid receptors recruit an RNA molecule (53) and proteasome subunits and ubiquitin-conjugating enzymes (54) into transcription complexes. More recently, a role for actin in regulating IFN-induced transcription has been shown (55, 56). An arginine biosynthetic enzyme has been recently implicated in direct transcriptional regulation of cellular genes (57). A protein with pseudouridine synthase activity has also been shown to function as a transcriptional regulator (58). In summary, our studies identified a novel IFN-γ-induced transcriptional-activating mechanism, where ligand-induced signals modify the phosphorylation status of C/EBP-β which then associates with GBF1 with a greater affinity to drive gene transcription. Future studies should reveal the further control points involved in this regulation.
We thank Peter Johnson for providing the C/EBP-β mutants used in this study.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Cancer Institute Grants CA78282 and CA105005 (to D.V.K.).
Abbreviations used in this paper: IRF, IFN-gene regulatory factor; GATE, IFN-γ-activated transcriptional element; GBF1, GATE-binding factor 1; MEF, mouse embryo fibroblast; IP, immunoprecipitation; ChIP, chromatin IP; DBD, DNA-binding domain; CBS, C/EBP-binding site; CID, C/EBP-β-interacting domain; MEKK, MEK kinase; GAS, IFN-γ-activated site; ppERK, diphosphorylated ERK.