Involvement of STAT-1 and Ets Family Members in Interferon-γ Induction of CD40 Transcription in Microglia/Macrophages*

Cluster of differentiation (CD)-40 is a cell surface receptor belonging to the tumor necrosis factor receptor family that plays a critical role in the regulation of immune responses. We have previously shown that the cytokine interferon (IFN)-γ induces CD40 expression in microglia. Herein, we have elucidated the molecular mechanisms underlying IFN-γ induction of CD40 gene expression in microglia/macrophages. IFN-γ up-regulates CD40 expression at the transcriptional level, and this regulation involves the STAT-1α transcription factor. Microglia from STAT-1α-deficient mice were refractive to IFN-γ induction of CD40 expression, illustrating the importance of STAT-1α in this response. Functional analysis of the CD40 promoter indicates that two gamma activated sequence elements as well as two Ets elements are involved in IFN-γ induction of CD40 promoter activity. STAT-1α binds to the gamma activated sequence elements, whereas PU.1 and/or Spi-B bind to the Ets elements. The expression of PU.1 and Spi-B, in conjuction with STAT-1α activation, correlates with IFN-γ inducibility of CD40 expression. Collectively, our data demonstrate the involvement of STAT-1α, PU.1, and Spi-B in IFN-γ induction of CD40 gene expression in cells of the macrophage lineage.

CD40 1 is a 50-kDa type I phosphoprotein member of the tumor necrosis factor-receptor superfamily (for review see Refs. 1 and 2). CD40 is expressed by a wide variety of cells such as B-cells, macrophages, dendritic cells, keratinocytes, endothelial cells, thymic epithelial cells, fibroblasts, and tumor cells (for a review see Ref. 3). The interaction between CD40 and its cognate ligand, CD40L (CD154), is critical for a productive immune response. X-Linked hyper IgM syndrome individuals have defects in CD154-CD40 interactions between their T-cells and antigen-presenting B-cells, exhibit elevated levels of IgM with the virtual absence of other antibody isotypes, and are extremely susceptible to bacterial, viral, and opportunistic infections (4). CD154-CD40 interactions promote B-cell growth, differentiation, and immunoglobulin class switching (5). As well, up-regulation of various costimulatory molecules (intercellular adhesion molecule-1 (ICAM-1), a vascular cell adhesion molecule-1, E-selectin, leukocyte functional antigen-3, B7.1, B7.2, and CD40) occurs upon CD40-CD154 contact, as does production of numerous cytokines and chemokines (IL-1, IL-6, IL-8, IL-10, IL-12, tumor necrosis factor-␣, and macrophagederived inflammatory protein-1␣) (for review see Ref. 1). The production of IL-12 is thought to be particularly important for promoting T-cell maturation toward the Th1 pathway (6).
CD40 has been implicated in participating in many human diseases, particularly autoimmune diseases (for review see Ref. 7). Interaction of CD40-CD154 is necessary for the initiation of insulitis and diabetes in non-obese diabetic mice (8). Aberrant expression of CD40 and CD154 has been described in rheumatoid arthritis (9), multiple sclerosis (10), and other diseases that involve a hyperactive immune system (11,12). Because CD40 is functionally critical and nonredundant for the activation of immune responses, blocking the interaction between CD40-CD154 with anti-CD154 or CD40-Ig has been shown to be beneficial in animal models of autoimmune diseases (13)(14)(15). These findings collectively illustrate the importance of CD40-CD154 interactions for homeostasis of immune responses.
Microglia/macrophages constitutively express CD40 at a low level, which is enhanced by IFN-␥ (16,17). IFN-␥ also induces or enhances CD40 expression on other cell types such as thymic epithelial cells, smooth muscle cells, and endothelial cells (for review see Ref. 3). The binding of IFN-␥ to its surface receptor activates the receptor-associated tyrosine kinases JAK1 and JAK2. JAKs tyrosine phosphorylate and activate the latent cytoplasmic transcription factor STAT-1␣, which then dimerizes, translocates into the nucleus, and binds to the gamma activated sequence (GAS) element of IFN-␥ responsive genes, resulting in gene activation (for review see Ref. 18). IFN-␥ inducible genes such as class II transactivator and interferon regulatory factor (IRF)-1 cannot be induced by IFN-␥ in STAT-1␣-deficient mice, illustrating the critical role of STAT-1␣ in IFN-␥-induced gene expression (19). STAT-1␣ has been shown to cooperate with other transcription factors such as upstream stimulatory factor-1, Sp1, and PU.1 (Spi-1) to induce gene transcription, as well as to confer cell-specific gene expression (20,21).
A portion of the human CD40 5Ј-flanking region had been isolated (22); however, a functional analysis on this 758-nucleotide fragment has never been performed. It is a TATA-less promoter that contains three potential GAS elements at Ϫ521, Ϫ483, and Ϫ129 bp. Other potentially relevant cis-regulatory elements in the human CD40 5Ј-flanking region are nuclear factor-B, Ets family, activator protein-1, c-Myc, and Sp1 (23). These cis-acting elements may account for differential cytokine modulation of CD40 expression in various cell types.
In this study, we demonstrate that IFN-␥ can induce the macrophage cell line RAW264.7 to express CD40 in a similar fashion to that of microglia, the resident macrophage cells of the brain (17). We further show that IFN-␥ up-regulates CD40 expression transcriptionally and that this regulation involves the STAT-1␣ transcription factor. Analysis of the human CD40 promoter indicates that the two GAS sites at Ϫ521 (distal GAS (dGAS)) and Ϫ483 (medial GAS (mGAS)) are important for IFN-␥ induction of CD40 transcription. In addition to the two GAS elements, two Ets family member binding sites located at Ϫ553 (etsA) and Ϫ447 (etsB) also contribute to IFN-␥ induction of human CD40 promoter activity. Combinations of site-directed mutagenesis and electrophoretic mobility shift assay (EMSA) studies suggest that STAT-1␣ binding to the mGAS element is critical for IFN-␥ induction of CD40 expression in microglia/macrophages. Furthermore, the binding of PU.1 and/or Spi-B to the etsA element and PU.1 to the etsB element may confer cell-specific expression of IFN-␥-induced CD40 in microglia/macrophages. The results from our study are the first to delineate the molecular basis of cytokine regulation of CD40 expression in any cell type.
CD40 Promoter Constructs-The 5Ј-flanking sequence of the human CD40 gene was isolated by PCR. Primers were designed according to the published sequence deposited in GenBank TM by Rudert et al. (22). The sequence for the sense primer was 5Ј-GATCTGCCCGCCTCGGCC-3Ј, and for antisense it was 5Ј-GGCGAGGTGAGACCAGGC-3Ј. PCR was performed with the Taq PCR Core Kit (Qiagen, Santa Clarita, CA) according to the manufacturer's instructions with 2.5 units of Taq DNA polymerase, 0.1 M of each primer, and 600 ng of genomic DNA from the human macrophage cell line THP-1. The PCR amplification protocol consisted of an initial 1-min melting step at 94°C, followed by 30 cycles with 40 s melting at 92°C, 40 s annealing at 60°C, and 1 min 30 s extension at 75°C, except for the last cycle, which contained a 5-min extension step. The resulting 758-bp fragment was gel-purified and ligated into the linearized pCRII vector (Invitrogen, Carlsbad, CA). The complete sequence of the insert was obtained by automatic sequencing, which was performed by the University of Alabama at Birmingham Center for AIDS Research Molecular Biology Core Facility (see Fig. 7). The 758-bp insert was released from pCRII by digestion with the restriction enzyme EcoRI and gel-purified, and the restriction ends were blunted with the Klenow fragment of DNA polymerase I according to the manufacturer (Promega, Madison, WI). The blunt-ended fragment was ligated into the SmaI site of the pGL3-Basic vector, which contains the gene for luciferase as the reporter. The designated name for this construct is hCD40p0.7. 5Ј-Deletion constructs were prepared by cutting hCD40p0.7 once 5Ј of the promoter insertion site with KpnI. Next, the restriction enzymes AccB7I, NsiI, ApoI, PvuII, Bsu36I, NheI, SmaI, and BanII, which each cut once inside the promoter, were used to remove the unwanted fragments, and the resulting vectors containing various lengths of the 5Ј-end-deleted promoter were religated, generating the constructs hCD40p⌬1-hCD40p⌬8 (see Fig. 2). For the internal deletion constructs 3⌬1 and 3⌬2, fragments were removed from hCD40p0.7 by digestion with SmaI (Ϫ155) in combination with PvuII (Ϫ456) or Bsu36I (Ϫ411); the deleted promoters were religated, generating hCD40p3⌬1 and hCD40p3⌬2, respectively (see Fig. 5A). The insertion construct Ins5N was generated by inserting 5 bp (5Ј-AGCTT-3Ј) at Ϫ542 of the hCD40p0.7 construct, whereas for Ins10N, 5 bp (5Ј-CTAGA-3Ј) were inserted into the Ins5N construct at Ϫ543 (see Fig.  6B). The site-directed mutation and insertion constructs were generated on the hCD40p0.7 plasmid backbone using the QuikChange sitedirected mutagenesis kit (Stratagene, La Jolla, CA) following the manufacturer's instructions and were confirmed by sequencing.
Quantitative Analysis of CD40 Protein Expression by Immunofluorescence Flow Cytometry-Cells were plated at 2 ϫ 10 5 cells/well into 12-well plates (Costar, Cambridge, MA), and duplicate wells were either untreated or treated with IFN-␥ (75 units/ml) for 40 -48 h as described previously (17). The murine cells (RAW264.7, EOC13, primary microglia, A20, and NIH 3T3) were scraped, incubated with 100 l of 2.4G2 hybridoma supernatant (which contains rat anti-mouse Fc␥R antibody) supplemented with 10% normal mouse serum for 30 min at 4°C, washed, incubated with 10 g/ml of anti-CD40 antibody for 30 min at 4°C, washed, incubated with 10 g/ml of biotinylated anti-rat IgG 2a , washed, incubated with 10 g/ml of phosphatidylethanolamine-conjugated strepavidin, washed, and then were fixed in a final volume of 200 l of 1% paraformaldehyde. Other cells (THP-1, primary human monocytes and astrocytes, U937, CRT, U373-MG, U251-MG, HeLa, HT 1080, Jurkat, Molt 48, CEM T4, and H9) were blocked with 20% normal rabbit serum and then stained with fluorescein isothiocyanate-conjugated anti-human CD40 antibody for 30 min at 4°C, washed, and fixed as above. The cells were analyzed on the FACScan (Becton Dickinson, Mountain View, CA). Negative controls were incubated with isotypematched antibody. Ten thousand cells were analyzed for each sample.
Nuclear Extracts and EMSA-Cells were incubated with medium or IFN-␥ (75 units/ml) for 30 min, and nuclear extracts were prepared as described previously (17). EMSA was performed with 5 g of nuclear extract in a total volume of 15 l of binding buffer (50 mM NaCl, 1 mM MgCl 2, 0.1 mM EDTA, 4% glycerol, 0.5 mM dithiothreitol, 4 mM Tris-Cl (pH 7.5), 1 g of polydeoxyinosinic-deoxycytidyl acid, and 20,000 cpm 32 P-labeled oligonucleotide probe), and incubated on ice for 15 min. Bound and free DNA were then resolved by electrophoresis through a 6% polyacrylamide gel in 0.5ϫ Tris-borate-EDTA buffer at 250 V for 1 h. For supershift analysis, 1 g of indicated antibody was added, or for competition analysis, a 100-fold molar excess of the indicated cold oligonucleotides was added to the nuclear extracts and incubated on ice for 30 min followed by an additional incubation for 15 min with the labeled probe.
Transient Transfection and Analysis-One microgram of the hCD40 promoter constructs (both CD40 deletion and mutant constructs) were cotransfected with 0.2 g of the pCMV-␤-galactosidase construct into 4 ϫ 10 5 RAW264.7 cells in 6-well plates using the LipofectAMINE Plus method according to the manufacturer's directions. pGL3-Basic was used as a negative (background) control in all experiments. After 3 h of transfection, cells were allowed to recover for 6 h prior to treatment with IFN-␥ (75 units/ml) for 12 h, which we have previously determined to be optimal for IFN-␥-induced activation of the hCD40p0.7 construct (data not shown). Cells were washed with phosphate-buffered saline and lysed with 250 l of lysis buffer (25 mM trisphosphate (pH 7.8), 2 mM dithiothreitol, 2 mM diaminocyclohexane tetraacetic acid, 10% glycerol, and 1% Triton X-100). Extracts were assayed in triplicate for luciferase activity in a total volume of 130 l (30 l of cell extract, 20 mM Tricine, 0.1 mM EDTA, 1 mM MgCO 3 , 2.67 mM MgSO 4 , 33.3 mM dithi-othreitol, 0.27 mM coenzyme A, 0.47 mM luciferin, and 0.53 mM ATP), and light intensity was measured using a luminometer (Promega, Madison, WI). Luciferase activity was integrated over a 10-s time period. Extracts were also assayed in triplicate for ␤-galactosidase enzyme activity as described previously (31). The luciferase activity of each sample was normalized to ␤-galactosidase activity to yield relative luciferase activity. Fold induction was calculated as the ratio of relative luciferase activity between IFN-␥ and medium-treated samples that were transfected with the same construct. For comparison between different constructs, percent of wild type was calculated as the ratio of fold induction of mutated constructs to that of the full-length promoter, which was set at 100%.
Nuclear Run-on Assay-Fifty million RAW264.7 or EOC13 cells were treated with medium or IFN-␥ (75 units/ml) for 2 or 6 h, respectively. Cells were harvested by scraping and washed twice in ice-cold phosphate-buffered saline. The cells were then resuspended in 4 ml of lysis buffer (10 mM Tris, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 , 0.5% Nonidet P-40), spun down at 500 ϫ g for 5 min, and aspirated, and then the procedure was repeated. The nuclei were then resuspended in 200 l of storage buffer (50 mM Tris, pH 8. , or murine primary microglia were incubated with medium alone or with murine IFN-␥ for the indicated times, and then total RNA was prepared as described under "Experimental Procedures." 10 -20 g of total RNA was subjected to ribonuclease protection assay for the expression of CD40 and GAPDH mRNA. IFN-␥-induced CD40 mRNA expresssion was enhanced by ϳ38-, ϳ43-, and ϳ35-fold above the untreated samples in RAW264.7, EOC13, and primary microglia, respectively. Data shown are representative of three experiments. C, IFN-␥ enhances CD40 gene transcription. RAW264.7 or EOC13 cells were incubated without or with murine IFN-␥ (100 units/ml) for 2 or 6 h, respectively. Nuclei were subject to nuclear run on assay. Labeled RNA was then hybridized to membrane-anchored plasmid pGEM-4Z as the vector control DNA, murine CD40 cDNA, murine IRF-1 cDNA as a positive control DNA, and murine GAPDH cDNA as reference. IFN-␥ enhanced CD40 transcription ϳ15-fold for both cell types. Data shown are representative of two experiments. D, IFN-␥ induction of CD40 requires STAT-1␣. Murine primary microglia isolated from wild type (left panel) or STAT-1␣-deficient CD1 mice (right panel) were incubated with medium alone or with murine IFN-␥ (100 units/ml) for 48 h. Surface CD40 protein expression was assessed by FACS analysis. Data shown are representative of three experiments.
FIG. 2. IFN-␥-induced activity of human CD40 promoter 5-deletion constructs. Deletion constructs of the human CD40 promoter or empty vector pGL3-Basic were co-transfected with the reference plasmid pCMV-␤-galactosidase into RAW264.7 cells. The transfected cells were treated with medium or murine IFN-␥ (75 units/ml) for 12 h and then analyzed for luciferase and ␤-galactosidase activity as described under "Experimental Procedures." Fold induction was calculated as the ratio of relative luciferase activity of IFN-␥-treated samples to that of medium-treated samples (see "Experimental Procedures"). Data shown are the mean Ϯ S.E. of five experiments. Some cis-acting elements, as suggested by the MatInspect program, are depicted on the full-length promoter. Numbers above each potential cis-regulatory element indicate the relative position with respect to the transcription initiation site.
Tris, pH 7.4, 0.5 M NaCl, 50 mM MgCl 2 , 2 mM CaCl 2 , 50 g of DNase) was then added and incubated at 30°C for 10 min, then 200 l of protein digestion buffer (0.5 M Tris, pH 7.4, 0.125 M EDTA, 5% SDS, 100 g of proteinase K) was added and incubated at 42°C for 30 min. The nuclei were lysed by guanidine thiocyanate, and then labeled RNA was extracted by phenol/chloroform extraction and ethanol precipitated. The radioactivity of each sample was adjusted with hybridization buffer (50 mM Na 2 PO 4 , pH 6.5, 50% formamide, 1% SDS, 5ϫ SSC, 2.5ϫ Denhardt's, 250 g/ml Escherichia coli tRNA) to 1 ϫ 10 5 cpm/l. Ten million cpm was then hybridized to membrane cross-linked plasmid DNA of pGEM-4Z as empty vector control, murine CD40, murine IRF-1, as a positive control, and murine GAPDH for reference in 0.5 ml of hybridization buffer at 42°C for 48 h. The membranes were then washed twice with 2ϫ SSC at 56°C for 15 min, once with 2ϫ SSC at 42°C, digested with 0.05 g/ml RNase T1/A mixture in 2ϫ SSC at 37°C for 30 min, washed once with 2ϫ SSC at 37°C for 1 h, and then exposed to x-ray film. Quantification of bound labeled RNA was performed by scanning with the PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Values for CD40 expression were normalized to GAPDH levels for each experimental condition.

IFN-␥-induced CD40 Gene Expression Is Mediated at the Transcriptional Level and Requires STAT-1␣-
We have shown previously that IFN-␥ is the most potent inducer of CD40 expression in microglia, the resident macrophage of the brain (17). Because microglia and macrophages are derived from the same lineage (32), we wished to compare the extent of IFN-␥ induction of CD40 expression in macrophages with that of microglia. The microglial cell line EOC13 (27), macrophage cell line RAW264.7, and primary murine microglia were incubated without or with IFN-␥ (75 units/ml) for 48 h, and then surface CD40 expression was assessed by flow cytometry. Both the microglial cell line EOC13 and primary murine microglia constitutively express extremely low levels of surface CD40, whereas the macrophage cell line RAW264.7 expresses moderate constitutive levels of surface CD40. All three cell types up-regulate CD40 expression in response to treatment with IFN-␥ (Fig. 1A). To determine if the observed increase in surface expression of CD40 protein correlates with CD40 mRNA expression, EOC13 cells were incubated without or with IFN-␥ for 20 h, and RAW264.7 and primary murine microglia were incubated with IFN-␥ for 8 h. Total RNA was harvested and analyzed by ribonuclease protection assay. Fig. 1B shows that microglia/macrophages up-regulate CD40 mRNA levels (ϳ40fold) in response to IFN-␥ treatment. Next, we determined if the observed increase in steady state levels of CD40 mRNA reflects an increase in the transcriptional rate of the CD40 gene. EOC13 and RAW264.7 cells were treated with IFN-␥ for 6 or 2 h, respectively, and then nuclei were isolated and analyzed by nuclear run-on assay. The results demonstrate that IFN-␥ enhances CD40 transcription in microglia and macrophage cell lines by ϳ15-fold (Fig. 1C). The IRF-1 gene was used as a positive control, because IFN-␥ enhances transcription of this gene (33), and GAPDH serves as a reference gene. Collectively, these data indicate that IFN-␥ induces CD40 expression in macrophages and microglia by enhancing transcription of the CD40 gene.
FIG. 3. dGAS and mGAS elements are important for IFN-␥ induction of human CD40 promoter activity. A, GAS elements are functional in the human CD40 promoter. Wild type, indicated mutants, or empty vector pGL3-Basic constructs were cotransfected with the reference plasmid ␤-galactosidase into RAW264.7 cells. The transfected cells were treated with medium or murine IFN-␥ (75 units/ml) for 12 h and then analyzed for luciferase and ␤-galactosidase activity. Values of each construct are plotted as percent of the wild type promoter, which is set at 100%. Above each mutation construct is the sequence of the wild type and mutation, with the mutated sequence depicted in lowercase. Data shown are the mean Ϯ S.E. of five experiments. B, IFN-␥ induces STAT-1␣ binding to the mGAS element. EMSA experiments were performed with mGAS probe and nuclear extracts from RAW264.7 and EOC13 cell lines unstimulated (Ϫ) or stimulated (ϩ) with IFN-␥ for 30 min. For competition and supershift, a 100-fold molar excess of indicated oligonucleotides (Table I)  IFN-␥-induced expression of class II transactivator, ICAM-1, and IRF-1 is mediated mainly through STAT-1␣ binding to GAS elements in the promoters of these genes (31, 34 -36). To determine if STAT-1␣ also mediates IFN-␥ induction of CD40 gene expression, microglia from wild type or STAT-1␣-deficient CD1 mice (29) were treated with medium or IFN-␥ for 48 h, and then surface expression of CD40 was analyzed by FACS. IFN-␥ induced the expression of surface CD40 on wild type microglia, whereas microglia from STAT-1␣-deficient mice were refractive to IFN-␥ induction of CD40 expression (Fig. 1D). These results demonstrate that IFN-␥ induction of CD40 requires the transcription factor STAT-1␣.
Identification of IFN-␥-responsive Elements in the CD40 Promoter-IFN-␥-induced expression of class II transactivator, ICAM-1, and IRF-1 is regulated at the transcriptional level (31, 34 -36). Thus far, we have shown that STAT-1␣ is required for IFN-␥ induction of CD40 gene expression (Fig. 1D) and that such induction is transcriptional (Fig. 1C). To study the transcriptional regulation of CD40 gene expression by IFN-␥, we characterized putative IFN-␥-responsive elements in the CD40 promoter. The human CD40 5Ј-flanking region was isolated by Rudert et al. (22); however, no attempt has been made to analyze the promoter of this important molecule in any cell type. To analyze the CD40 promoter, we initially isolated the 5Ј-flanking region of human CD40 by PCR as described under "Experimental Procedures." The sequence of the human CD40 promoter is shown in Fig. 7. Numerous potential cis-regulatory elements are illustrated in Fig. 2 (23). To determine the regions responsible for IFN-␥ inducibility in this 758-bp fragment of DNA, eight serial deletion constructs were generated. The deletion constructs (1 g) were cotransfected with the reference plasmid pCMV-␤-galactosidase (0.2 g) into RAW264.7 cells and then treated with medium or IFN-␥ for 12 h. Transfection studies were not performed in the EOC13 cell line or primary murine microglia as these cells could not be transfected by numerous protocols (electroporation, LipofectAMINE Plus, calcium phosphate, DEAE-dextran). Moreover, we used the human CD40 promoter in a mouse cell line, because the sequence of the mouse CD40 promoter was not available at the time this study was initiated (to be discussed further). IFN-␥-induced activation of the hCD40p0.7 promoter was ϳ10-fold above that of untreated cells (Fig. 2). IFN-␥ also increased the activity of the empty vector pGL3-Basic by ϳ2-fold. The first two deletions, ⌬1 and ⌬2, did not have any significant effect on IFN-␥ inducibility of the hCD40 promoter. The removal of an additional 66 bp (⌬3) reduced IFN-␥ inducibility of this promoter by ϳ75%, suggesting that the fragment between Ϫ570 and Ϫ504 is important for IFN-␥ induction of CD40 promoter activity. This fragment is named as the D3 region. Contained within the D3 region is a potential GAS element designated as dGAS at Ϫ521 (Fig. 2). The elimination of an additional 48 bp (⌬4) decreased IFN-␥ inducibility of this construct to background levels (similar to that of pGL3-Basic). These results indicate that the fragment between Ϫ504 and Ϫ456 also contains potential IFN-␥-responsive elements, particularly the mGAS element at Ϫ483. Additional deletions (⌬5-⌬8) did not further affect IFN-␥-induced activity of the promoter. These data suggest that the region between Ϫ570 and Ϫ456 of the human  (Table I)  CD40 promoter, encompassing two of the three potential GAS elements, is critical for IFN-␥ induction of CD40 expression.
dGAS and mGAS Elements Are Involved in IFN-␥ Induction of the CD40 Promoter-Because we have determined that STAT-1␣ is required for IFN-␥ induction of CD40 (Fig. 1D), and there are three potential GAS elements in the hCD40 promoter (Fig. 2), we wished to ascertain the role of each GAS element in IFN-␥ induction of CD40 promoter activity in the context of the full-length construct. We utilized site-directed mutagenesis techniques to mutate the three potential GAS elements individually (Fig. 3A). Mutation of the proximal GAS (pGAS) element at Ϫ129 (mpGAS construct) did not affect IFN-␥-induced activation of the CD40 promoter, suggesting that pGAS does not participate in IFN-␥ induction of CD40 expression. Mutation of the mGAS element (mmGAS construct) abrogated IFN-␥-induced promoter activity, reducing activity levels to that of pGL3-Basic. Mutation of the dGAS element (mdGAS construct) had an intermediate effect, with a reduction of ϳ50% of CD40 promoter activity compared with the full-length promoter construct. These results suggest that the mGAS element has a pivotal role in IFN-␥ induction of CD40 promoter activity, whereas the dGAS element is also important in this response.
To confirm that STAT-1␣ actually binds to the GAS sites identified functionally as important for CD40 promoter activity, we treated RAW264.7 and EOC13 cells with IFN-␥ for 30 min, prepared nuclear extracts, and performed EMSA using pGAS, mGAS, and dGAS oligonucleotides as probes and/or competitors (Table I). Fig. 3B shows that there is no binding to the mGAS oligonucleotide using nuclear extracts from untreated cells (lanes 1 and 10). Upon treatment with IFN-␥, there is complex formation over the mGAS probe in both cell types (lanes 2 and 11). The complex can be competed by a 100-fold excess of the mGAS oligonucleotide, although a residual complex is still very weakly detected (lane 3). Complex formation is partially competed by the dGAS oligonucleotide (lane 4), suggesting that the complex binds to mGAS with higher affinity than to dGAS. In fact, a weak band appears in EMSA performed using labeled dGAS as a probe, which can be supershifted with anti-STAT-1␣ antibody but not anti-STAT-3 or STAT-6 antibodies (data not shown). A 100-fold excess of pGAS oligonucleotide did not affect binding of the complex to the mGAS probe (lane 5), indicating the specificity of binding. In addition, no IFN-␥-induced complex formation was observed when pGAS was used as a probe (data not shown), correlating with the observation in Fig. 3A that mutation of the pGAS element does not affect the IFN-␥-induced activity of the promoter. The identity of the IFN-␥-induced complex was confirmed by supershifting with anti-STAT-1␣ (lane 7) but not with normal rabbit serum, anti-STAT-3, or anti-STAT-6 antibodies (lanes 6, 8, and 9). Furthermore, we have previously shown that nuclear extracts from IFN-␥-treated EOC13 cells bind the authentic GAS element of the human ICAM-1 promoter, and complex formation is totally abolished upon inclusion of anti-STAT-1␣ antibody (17). These results indicate that STAT-1␣ mediates IFN-␥ induction of CD40 by binding to the mGAS element, and to a lesser extent, to the dGAS element.
Maximal IFN-␥ Induction of CD40 Requires Additional cis-Acting Elements-STAT-1␣ is known to cooperate with other transcription factors for induction of gene expression. For example, IFN-␥-induced expression of class II transactivator and ICAM-1 requires cooperation between STAT-1␣ and USF-1, and STAT-1␣ and Sp-1, respectively (20,37). We suspected this might also be the case for IFN-␥-induced CD40 expression. This notion arose from the incongruity between the promoter activity of the ⌬3 construct (Fig. 2) and that of the site-directed mutation of the dGAS element (Fig. 3, mdGAS construct). The almost complete abolition of the ⌬3 promoter activity suggested that there may be other cis-acting elements besides the dGAS element in the D3 region that also participate in IFN-␥ induction of CD40 promoter activity. To resolve this issue, we searched for additional cis-acting element(s) in the D3 region by systematically mutating the D3 fragment starting from Ϫ568 bp. Four different mutant constructs, designated as D3M1 through D3M4, were transfected into RAW264.7 cells and then assayed for luciferase activity (Fig. 4A). The IFN-␥induced activity of each mutated construct was compared with the wild type promoter, which was set at 100%. The M2 and M3 constructs both reduced IFN-␥-induced activity of the promoter by ϳ50%, whereas M1 and M4 had no effect (Fig. 4A). These results suggest that the sequence between Ϫ541 and Ϫ559 contains cis-regulatory element(s) that are important for IFN-␥ inducibility of CD40 promoter activity. Upon closer examination of this 18-bp region, we identified the sequence 5Ј-GGAAG-3Ј, which is a potential binding site for Ets family members (for review see Ref. 38) spanning across the junction of M2 and M3. This site is designated as etsA. Selective mutation of this etsA site resulted in a decline in IFN-␥-induced luciferase activity by ϳ50%, similar to that of the M2 and M3 constructs ( Fig. 4A; metsA construct). These data indicate that the etsA element located upstream of dGAS is also involved in IFN-␥-induced CD40 promoter activity.
We next determined if protein complex(es) can bind to the etsA element. Labeled etsA probe (Table I) was incubated with nuclear extracts from untreated or IFN-␥-treated RAW264.7 cells (Fig. 4B). Two complexes bound to this probe (lane 1), and IFN-␥ treatment did not alter expression of either complex (lanes 11). A 100-fold excess of unlabeled estA oligonucleotide completely competed away both complexes (lane 2), whereas none of the oligonucleotides with mutations in the etsA element (Table I) had any effect on complex formation (lanes [3][4][5], indicating that the binding is specific. The identity of the complex was determined by the use of antisera against various ets proteins. Normal rabbit serum or anti-ets-1/2 antiserum did not affect complex formation (lanes 6 and 7). Anti-Spi-B antiserum blocked formation of the upper complex (lane 8), whereas anti-PU.1 affected the lower complex (lane 9), and the inclusion of both Spi-B and PU.1 antisera blocked formation of both complexes (lane 10). A similar binding pattern was also obtained using nuclear extracts from EOC13 cells (data not shown). These data suggest that binding of PU.1 and/or Spi-B to the etsA element is involved in IFN-␥ induction of CD40 expression. We have excluded other proteins (activator protein-2, C/EBP␦, C/EBP␤, c-Myb, c-Myc, Fra-1, FosB, c-Fos, Jun, Ets-1 and 2, Elk-1, IRF-1, IK-1, nuclear factor-B (p65), Sp1, Sp3, USF-1, and USF-2) as possible components of the complexes because antibodies to these proteins failed to affect complex formation ( Fig. 4B and data not shown).
Because we determined that binding of PU.1 and/or Spi-B to the etsA element is important for IFN-␥-induced CD40 promoter activity, we searched for other potential Ets protein binding sites in the human CD40 promoter. The DNA fragment between the mGAS and pGAS elements (Ϫ456 bp to Ϫ155 bp) contains at least five potential Ets protein binding sites. To ascertain the role of these potential Ets protein binding sites, we created two internal deletion constructs from the hCD40p0.7 construct designated as 3⌬1 and 3⌬2. The 3⌬1 and 3⌬2 constructs were generated by differential deletion of the intervening sequence between the mGAS and pGAS elements (Fig. 5A). All potential Ets protein binding sites located between the dGAS and the mGAS element were removed in the 3⌬1 construct, whereas one potential Ets protein binding site, named etsB (5Ј-CTTCC-3Ј) at Ϫ447 was left intact in the 3⌬2 construct. When transfected into RAW264.7 cells, IFN-␥-induced luciferase activity of the 3⌬1 construct was reduced by ϳ50%, whereas the IFN-␥-induced luciferase activity of the 3⌬2 construct was similar to that of the wild type construct (Fig. 5A). These results suggest that there are cis-regulatory elements residing in the 45-bp region between Ϫ456 and Ϫ411,

FIG. 5. The etsB cis-regulatory element enhances IFN-␥-induced human CD40 promoter activity.
A, internal deletion constructs (3⌬1, 3⌬2) of the human CD40 promoter or the wild type construct were co-transfected with the reference plasmid pCMV-␤-galactosidase into RAW264.7 cells. The transfected cells were then stimulated with medium or murine IFN-␥ for 12 h prior to assaying for luciferase and ␤-galactosidase activity. Values are plotted as percent of the wild type construct, which is set at 100%. Data shown are the mean Ϯ S.E. of five experiments. B, the etsB element is functional in the hCD40 promoter. The wild type or etsB mutant construct (metsB) was co-transfected with the reference plasmid ␤-galactosidase into RAW264.7 cells and treated similar to that of A. Data shown are mean Ϯ S.E. of three experiments. C, PU.1 constitutively binds to the etsB probe. EMSA experiments were performed with the etsB probe and nuclear extracts from RAW264.7 and EOC13 cell lines unstimulated (Ϫ) or stimulated (ϩ) with IFN-␥. For competition analysis, a 100-fold molar excess of indicated oligonucleotides (Table I)  particularly etsB, that may also participate in IFN-␥ activation of the CD40 promoter.
To confirm that the etsB element is involved in IFN-␥-induced CD40 promoter activity, this site was mutated within the hCD40p0.7 construct and transfected into RAW264.7 cells. Mutation of the etsB element reduced IFN-␥ inducibility of the human CD40 promoter activity by ϳ50% (Fig. 5B), comparable to that observed with deletion construct 3⌬1, suggesting that the etsB element is functional and participates in IFN-␥ induction of CD40 expression. To identify the protein(s) that bind to this site, EMSA was performed using labeled etsB as probe (Table I), and nuclear extracts from RAW264.7 and EOC13 cells (Fig. 5C). A single complex bound to the probe constitutively (lanes 1 and 8) and IFN-␥ treatment did not alter the binding pattern (lanes 7 and 9). The binding of this complex was specific because it could be competed away with a 100-fold excess of etsB oligonucleotide (lane 2) but not with the mutated oligonucleotide metsB (lane 3). We determined that the complex contains PU.1, as anti-PU.1 antibody completely supershifted the complex (lane 6), whereas normal rabbit serum or antibody that recognizes both ets-1 and ets-2 did not affect complex formation (lanes 4 and 5). These results suggest that PU.1 binding to the etsB element is required for maximal induction of CD40 expression by IFN-␥.
Both PU.1 and STAT-1␣ Are Required for IFN-␥-induced CD40 Expression-We have determined that there are at least four different cis-elements on the human CD40 promoter that contribute to IFN-␥ induction of CD40 expression. However, certain discrepancies remain unresolved. The removal of 207 bp at the 5Ј-end in the deletion construct hCD40p⌬3, which contains both the dGAS and etsA elements, reduced promoter activity to background levels (Fig. 2), whereas individual mutations of the dGAS and etsA elements decreased activity only by ϳ50% (Figs. 3A and 4A). Would a double mutation of both dGAS and etsA reflect the reduction observed with hCD40p⌬3? Indeed this is the case; as shown in Fig. 6A, the combined mutation of both dGAS and etsA (construct mdGAS/metsA) decreased IFN-␥-induced promoter activity to background levels, suggesting that the loss of both dGAS and etsA can account for the loss of promoter activity observed in the hCD40p⌬3 deletion construct. Interestingly, double mutation of dGAS and etsB (construct mdGAS/metsB), or etsA and etsB (construct metsA/metsB) also diminished IFN-␥-induced CD40 promoter activity down to the level of the promoterless vector pGL3-Basic (Fig. 6A). These results suggest that the dGAS and the ets elements (etsA and etsB) cooperate for optimal IFN-␥ activation of the CD40 promoter.
To examine the nature of cooperation between PU.1 and/or Spi-B at the etsA element and STAT-1␣ at the dGAS element, we altered the orientation of the etsA site by inserting 5 bp between the dGAS and etsA elements in the context of the hCD40p0.7 construct. Fig. 6B shows that a 5-bp insertion (Ins5N construct) reduces IFN-␥-induced CD40 promoter activity by ϳ50%. This decrease is comparable to that observed when the etsA element was mutated (Fig. 4A, constructs M2, M3, and metsA). Upon re-establishment of the orientation of the etsA element by inserting an additional 5 bp into the Ins5N construct, IFN-␥-induced activity of the promoter was restored (Fig. 6B, construct Ins10N). These data indicate that the function of the etsA element is dependent on its spatial orientation rather than the spacing between the etsA and dGAS elements.
We have recently cloned and sequenced 1100 bp of the mouse CD40 promoter. All four identified functional sites (dGAS, mGAS, etsA, and etsB) are conserved between the human and mouse promoter, as is the spacing between these sites (Fig. 7). The conservation of the organization of these four identified elements suggests that these sites are likely to function in a similar manner in both mouse and human for CD40 expression.
Binding to etsA and etsB Is Only Observed in CD40 Positive Cells-We have shown that IFN-␥-induced CD40 promoter activity requires the activation of STAT-1␣ and the constitutive expression of PU.1 and/or Spi-B. The failure of IFN-␥ to induce CD40 expression in some cells (Table II) could be explained by two possible mechanisms; either IFN-␥ fails to activate STAT-1␣, or PU.1 and/or Spi-B are not expressed in these cells. To test this hypothesis, we first examined the ability of these cells to activate STAT-1␣ when treated with IFN-␥. RAW264.7, EOC13, HeLa, U251-MG astroglioma cells, NIH 3T3 fibroblasts, and A20 B-cells were treated with IFN-␥ for 30 min. Cell extracts were prepared and analyzed by EMSA using mGAS as probe (Fig. 8A). Complex formation was observed in all cell types upon treatment with IFN-␥ (lanes 2, 4, 6, 8, and 10), except for the mature B cell line A20 (lane 12). The complex has been identified as STAT-1␣ by supershift analysis (data not shown). IFN-␥ also enhances surface expression of class I MHC and ICAM-1 in these cells (data not shown), indicating that the IFN-␥ signaling pathway is functional. There was no change in the pattern of complex formation in A20 cells upon IFN-␥ stimulation (lanes 11 and 12), and IFN-␥ did not enhance class I MHC or ICAM-1 expression in these cells (data not shown).
The same nuclear extracts were analyzed by EMSA using etsA or etsB as probes. PU.1 and Spi-B binding to the etsA probe and PU.1 binding to the etsB probe was observed only in cells that can be induced to express CD40 in response to IFN-␥ treatment (Fig. 8, B and C, lanes 1-4, and Table II). Interestingly, nuclear extracts from A20 cells show weak binding to the etsA probe and strong binding to the etsB probe (Fig. 8, B and  C, lanes 11 and 12), despite the fact that these cells are refractive to IFN-␥-enhanced CD40 expression. However, A20 cells constitutively express CD40, suggesting that PU.1 and/or Spi-B are involved in constitutive expression of CD40 in Bcells. These results suggest that PU.1 and/or Spi-B may confer cell-specific expression of CD40. DISCUSSION Direct involvement of CD40 has been demonstrated in many Th1-mediated autoimmune diseases such as experimental allergic encephalomyelitis, an animal model for multiple sclerosis; collagen-induced arthritis, an animal model for rheumatoid arthritis; and lupus nephritis (for review see Ref. 2). There are numerous reports focusing on the functional effects of CD40; however, little attention has been paid toward cytokine regulation of CD40 gene expression. We and others have reported previously that IFN-␥ is the most potent inducer of CD40 expression in microglia/macrophages (16,17). Herein, we report the transcriptional mechanism by which IFN-␥ induces CD40 expression in these cells. Data obtained from FACS analysis demonstrate that IFN-␥ induces CD40 protein expression in both macrophages and microglia (Fig. 1A). This increase in sur-face expression of CD40 is accompanied by elevated levels of steady state CD40 mRNA (Fig. 1B). Nuclear run-on assays confirm that the IFN-␥-induced elevation of CD40 mRNA is because of an increase in the transcription of the CD40 gene (Fig. 1C).
IFN-␥ induction of numerous genes has been shown to require the transcription factor STAT-1␣ (for review see Ref. 18). This is also the case for IFN-␥-induced CD40 expression, because STAT-1␣-deficient microglia are refractive to IFN-␥-induced CD40 expression (Fig. 1D). This dependence on STAT-1␣ can be accounted for by the presence of three potential GAS elements in the human CD40 promoter: dGAS (Ϫ521), mGAS (Ϫ483), and pGAS (Ϫ129) (Fig. 2). Of the three GAS elements, mGAS has the central role in IFN-␥ activation of CD40 expression because mutation of the mGAS element abolishes IFN-␥induced CD40 promoter activity (Fig. 3A). Furthermore, IFN-␥-activated STAT-1␣ directly binds to the mGAS element (Fig.  3B). The dGAS element, despite its low affinity for STAT-1␣, enhances the induction of CD40 by IFN-␥ ( Fig. 3A and data not shown). Variations in the sequences of the mGAS and the dGAS elements may explain the differential binding of STAT-1␣ to these GAS elements (for review see Ref. 39). STAT-1␣ cannot bind to hybrid GAS elements that we created by exchanging the core sequence of the dGAS (5Ј-TTCCTTT-GAA-3Ј) and mGAS (5Ј-TTCCTTGAA-3Ј) probes (data not shown), suggesting that the number of intervening nucleotides in the palindromes and the sequence of the flanking nucleotides determines the binding affinity of STAT-1␣ to the GAS elements in the CD40 promoter. pGAS does not appear to participate in IFN-␥ induction of CD40 expression because STAT-1␣ does not bind to the oligonucleotide that contains the pGAS sequence (data not shown), and mutation of pGAS has no effect on the ability of IFN-␥ to activate CD40 promoter activity (Fig. 3A). Thus, we have shown that STAT-1␣ directly mediates IFN-␥-induced CD40 expression and does so through two of the GAS elements in the CD40 promoter.
STAT-1␣-mediated gene expression usually is accomplished through the cooperation between STAT-1␣ and other transcription factors (20,21,37). We have shown that optimal IFN-␥induced CD40 promoter activity requires the preservation of GAS (dGAS and mGAS) elements as well as two ets elements (etsA and etsB) (Figs. 3-5). The ets elements, when mutated individually, decrease IFN-␥-induced CD40 promoter activity by ϳ50% (Figs. 4A and 5B). Gel shift analysis shows that constitutively expressed PU.1 binds to both etsA and etsB, whereas Spi-B, which is also expressed in a constitutive manner, binds only to the etsA element (Figs. 4B and 5C). Both PU.1 and Spi-B are members of the ets transcription factor family that mediate cell-specific expression of many genes by binding to cis-regulatory elements with the common core 5Ј-GGAA/T-3Ј (for review see Refs. 38 and 40). Spi-B is expressed only in B and T cells but not in promonocytic U937 cells (24). In our hands, both Spi-B and PU.1 are expressed constitutively in the macrophage cell line RAW264.7, the microglial cell line EOC13, and the mature B cell line A20, but neither protein is expressed in U937 cells (ATCC no. CRL-1593.2) as determining by EMSA and Western blot (Figs. 4B and 8 and data not shown). The expression of Spi-B and PU.1 in RAW264.7 cells, but not U937 cells, may reflect differences in their stages of differentiation.
The preservation of the GAS and PU.1 cis-acting elements has been shown to be necessary for cell-specific expression of FcR and IL-18 genes (21,41). However, optimal expression of IFN-␥-induced CD40 promoter activity is dependent not only on the conservation of all four sites identified in this study but also on the relative spatial orientation between the etsA and the dGAS element (Fig. 6B). The importance of the spatial orientation between the etsB and the mGAS element could not be studied as twisting the DNA between these two sites also affects the orientation of the upstream sites (e.g. dGAS and etsA), making data interpretation difficult. However, it appears that the spatial orientation between two GAS and two ets Upon binding of IFN-␥ to its receptor, STAT-1␣ is phosphorylated, dimerizes, translocates into the nucleus, and binds to GAS elements (dGAS and mGAS). The cooperation between STAT-1␣ and PU.1/ Spi-B at etsA and PU.1 at etsB to activate CD40 transcription may be mediated by direct interaction between the transcription factors or by interacting with an integrator protein, possibly CBP, to facilitate CD40 transcription. elements is important as illustrated by the conservation of this organization between the human and mouse CD40 promoters (Fig. 7).
The findings from this study allow us to envision a model as shown in Fig. 9. In this scenario, PU.1 and/or Spi-B constitutively bind to the etsA element, and PU.1 binds to the etsB element. This constitutive binding only occurs in cells that express CD40, either constitutively or upon induction by IFN-␥ (Table II and Fig. 8). Binding of IFN-␥-activated STAT-1␣ to the mGAS element of the CD40 promoter may facilitate STAT-1␣ to bind to dGAS with higher affinity. The cooperation between STAT-1␣ and PU.1 and/or Spi-B depends on the spatial orientation, but not on the position of the cis-acting elements, allowing us to speculate that a yet unidentified protein may bridge PU.1 and/or Spi-B and STAT-1␣. The occupancy of both the GAS and the ets sites by their respective transcription factors may facilitate the formation of a complex containing STAT-1␣, PU.1, and/or Spi-B, and possibly an unknown integrator. The cAMP-response element-binding protein-binding protein (CBP) is a likely candidate because it has been shown that CBP can interact with both PU.1 and STAT-1␣ (42,43). In this hypothetical complex, CBP requires at least four contact points with the DNA-binding proteins to be able to activate gene transcription. The four contact points are provided by two STAT-1␣ dimers at the dGAS and mGAS elements, one PU.1 or Spi-B at the etsA element, and one PU.1 at the etsB element. With the loss of any one contact point, such as mutation in the dGAS, etsA, or etsB site, IFN-␥-induced CD40 promoter activity is reduced by ϳ50%, and with the loss of any two sites, promoter activity is abolished (Fig. 6). Mutation of the mGAS site completely abrogates IFN-␥-induced CD40 promoter activity, suggesting that the lack of STAT-1␣ binding to the mGAS element may result in a loss of STAT-1␣ binding at the dGAS element.
Because the expression of PU.1 and/or Spi-B is cell-restricted, IFN-␥ induction of CD40 is also restricted to cell types that can activate STAT-1␣ while constitutively expressing PU.1 and/or Spi-B (Table II). In Fig. 8, we show that CD40negative cells fail to form a complex with the estA and estB probes in EMSA, despite their ability to activate STAT-1␣ when treated with IFN-␥. Moreover, these cells can respond to IFN-␥ by induction of cell surface molecules such as ICAM-1 and class I MHC (data not shown), indicating that the IFN-␥ signaling pathway is intact in these cells. These data suggest that the activation of STAT-1␣ allows the cells to respond to IFN-␥, but PU.1 and/or Spi-B control the cell-specific expression of CD40. However, our preliminary results indicate that overexpression of PU.1 with or without its interacting partner Pip, in CD40-negative cells (NIH 3T3, HeLa, and U251-MG) does not render these cells responsive to IFN-␥-induced CD40 expression (data not shown), suggesting that other regulatory mechanism(s) are involved in the induction of CD40 expression. One possibility is that other cell-restricted transcription factor(s) also regulate the activation of the CD40 promoter, or CD40 expression is actively suppressed in CD40 negative cells by a mechanism similar to that of PU.1 regulation (44).
In this paper, we have identified the importance of GAS and ets elements in the activation of the human CD40 promoter by IFN-␥. As well, cooperation between STAT-1␣ and PU.1/Spi-B may confer cell-specific induction of CD40 by IFN-␥. The nature of cooperation between STAT-1␣ and PU.1/Spi-B and the role of CBP in IFN-␥-induced CD40 expression is under investigation. In conclusion, we report the mechanism of IFN-␥-induced CD40 expression in microglia/macrophages. The novelty of these results is partly in identifying the transcription factors necessary for IFN-␥-induced CD40 expression, but more impor-tantly, uncovering the spatial organization of the regulatory elements/transcription factors that facilitate the regulation of CD40.