|Home | About | Journals | Submit | Contact Us | Français|
Granulocyte-macrophage colony-stimulating factor (GM-CSF) and macrophage colony-stimulating factor (M-CSF) independently stimulate the proliferation and differentiation of macrophages from bone marrow progenitor cells. Although the GM-CSF and M-CSF receptors are unrelated, both couple to Ras-dependent signal transduction pathways, suggesting that these pathways might account for common actions of GM-CSF and M-CSF on the expression of macrophage-specific genes. To test this hypothesis, we have investigated the mechanisms by which GM-CSF and M-CSF regulate the expression of the macrophage scavenger receptor A (SR-A) gene. We demonstrate that induction of the SR-A gene by M-CSF is dependent on AP-1 and cooperating Ets domain transcription factors that bind to sites in an M-CSF-dependent enhancer located 4.1 to 4.5 kb upstream of the transcriptional start site. In contrast, regulation by GM-CSF requires a separate enhancer located 4.5 to 4.8 kb upstream of the transcriptional start site that confers both immediate-early and sustained transcriptional responses. Results of a combination of DNA binding experiments and functional assays suggest that immediate transcriptional responses are mediated by DNA binding proteins that are constitutively bound to the GM-CSF enhancer and are activated by Ras. At 12 to 24 h after GM-CSF treatment, the GM-CSF enhancer becomes further occupied by additional DNA binding proteins that may contribute to sustained transcriptional responses. In concert, these studies indicate that GM-CSF and M-CSF differentially utilize Ras-dependent signal transduction pathways to regulate scavenger receptor gene expression, consistent with the distinct functional properties of M-CSF- and GM-CSF-derived macrophages.
Macrophages originate from multipotent progenitor cells in bone marrow and serve critical physiological roles in host immunity, wound healing, and inflammation (10). The colony-stimulating factors (CSFs), through a network of cell surface receptors, cytoplasmic signaling molecules, and transcription factors, regulate macrophage differentiation and function (23). Macrophage CSF (M-CSF) and granulocyte-macrophage CSF (GM-CSF) were originally identified based on their ability to promote macrophage growth and differentiation in colony-forming assays using primary bone marrow progenitor cells (6, 40). M-CSF exerts effects on gene expression through a specific, high-affinity M-CSF receptor encoded by the c-fms proto-oncogene (36, 38, 39). The binding of M-CSF to this receptor leads to dimerization and autophosphorylation of tyrosines within the cytoplasmic tail, resulting in the activation of multiple intracellular signaling pathways, including Ras, phosphatidylinositol 3-kinase, and phospholipase C, via the docking of specific SH2 domain-containing proteins (11, 21, 31, 34, 44). Activation of the Ras pathway has been proposed to play an important role in regulating macrophage differentiation by controlling the activities of downstream nuclear targets, including members of the AP-1 and Ets families of transcription factors (12, 17, 18, 49).
GM-CSF, in addition to promoting macrophage differentiation, also promotes the formation of granulocyte and mixed granulocyte-macrophage colonies in soft agar (22). The GM-CSF receptor is related to the cytokine family of receptors, consisting of a specific α subunit and a common β subunit that is shared with the interleukin 3 (IL-3) and IL-5 receptors (15, 24). These receptors lack intrinsic tyrosine kinase activity but interact with and activate Janus kinase 2 (JAK2) in response to the binding of GM-CSF (4, 32). As a consequence, the receptor becomes tyrosine phosphorylated at several locations within the cytoplasmic domain of the β chain and interacts with several specific SH2-containing proteins. The distal C terminus of the β subunit has been demonstrated to couple to Ras-dependent signal transduction pathways (16, 37), leading to activation of AP-1 (1, 50). Membrane-proximal regions of the GM-CSF receptor are involved in the docking and activation of members of the STAT family of transcription factors, including STAT1, STAT3, and STAT5 isoforms (3, 27, 33, 35). While the membrane-proximal region of the β common subunit has been found to be sufficient to mediate proliferative responses to GM-CSF (37, 45), the mechanisms by which the GM-CSF receptor promotes macrophage differentiation remain poorly understood. The observation that both the M-CSF and GM-CSF receptors activate Ras-dependent signal transduction pathways suggests that these pathways may account for at least some of the common effects of M-CSF and GM-CSF on macrophage differentiation.
To examine this question, we have used the scavenger receptor A (SR-A) gene as a model. The SR-A gene encodes a trimeric integral membrane protein that has been proposed to play roles in Ca2+-independent cell adhesion events and in the uptake of a variety of polyanionic macromolecules, including oxidized low-density lipoprotein (19, 20). Disruption of the SR-A gene in mice results in impaired host responses to some bacterial and viral pathogens and in partial resistance to the development of atherosclerosis (41). The SR-A gene is selectively expressed in macrophages and related cell types and is positively regulated by both M-CSF and GM-CSF (7, 9, 28, 42). Previous studies in monocyte/macrophage cell lines and transgenic mice have demonstrated that expression of the SR-A gene is dependent on the B-cell- and macrophage-specific transcription factor PU.1 (13, 25). In addition, binding sites for members of the Ets family of transcription factors that form ternary complexes with AP-1 proteins are required for activation of the SR-A promoter during macrophage differentiation of THP-1 monocytic leukemia cells in response to phorbol esters (50). Promoter and upstream enhancer elements containing PU.1 and AP-1/Ets binding sites are sufficient to support macrophage-specific gene expression and transcriptional responses to M-CSF in transgenic mice (13, 14).
In the present study, we have investigated the sequences required for transcriptional activation of the SR gene by M-CSF and GM-CSF in transgenic mice and GM-CSF-responsive cell lines. We find that while GM-CSF is capable of transiently activating AP-1 and Ets transcription factors in cell lines, this activity is not sufficient to activate the SR-A gene in primary bone marrow progenitor cells cultured in the presence of GM-CSF. GM-CSF-dependent activation of SR regulatory elements in transgenic mice requires an additional 300-bp enhancer element that is located adjacent to the M-CSF-dependent enhancer 4.5 kb upstream of the transcriptional start site. Twelve hours following GM-CSF treatment, this enhancer region becomes maximally occupied by GM-CSF-induced proteins that are immunologically distinct from STAT1 and STAT5. Intriguingly, the transcriptional activity of the GM-CSF-dependent enhancer is strongly inhibited by dominant negative forms of STAT5 and JAK2, suggesting that at least some of the late effects of GM-CSF on SR-A gene expression are mediated by a set of intermediate transcription factors that bind directly to the GM-CSF-responsive enhancer. Thus, while M-CSF and GM-CSF are both capable of activating AP-1 and cooperating Ets domain transcription factors, they ultimately regulate the expression of the SR-A gene, and presumably other macrophage-specific genes, by different mechanisms. These observations are consistent with the distinct functional properties of M-CSF- and GM-CSF-derived macrophages and with the phenotypic characteristics of macrophage populations observed in M-CSF-deficient mice.
The SR-A human growth hormone (hGH) transgenes used in this study are illustrated in Fig. Fig.1A.1A. The silencer-enhancer-promoter (SEP)-hGH and EP (enhancer-promoter)-hGH constructs have been described previously (13, 14). The EXP-hGH transgene was generated by removing the minimal promoter (−245 to +45 bp) from EP-hGH and replacing it with the −696 to +46 promoter. The GMEXP-hGH transgene was generated by replacing the −4.4 to −4.1 kb enhancer in EXP-hGH with the region from −4.8 to −4.1 kb. DNA microinjection, screening, breeding of founder animals, and measurement of hGH levels were carried out as previously described (13). For transient transfection experiments, the hGH reporter gene was replaced by a firefly luciferase cDNA. Mutations in the AP-1 and Ets binding sites were generated by oligonucleotide-directed mutagenesis and confirmed by dideoxy sequence analysis as previously described (49). Experiments to identify GM-CSF-responsive enhancer sequences were generated from a SEP template as restriction fragments or PCR-amplified products and subcloned in front of the −696 to +46 promoter-luciferase expression vectors.
Bone marrow progenitor cells were purified and cultured as previously described (13). Recombinant cytokines were obtained from R&D Systems (M- and GM-CSF) and used at the following concentrations unless otherwise noted: human M-CSF, 20 ng/ml; mouse GM-CSF, 5 ng/ml. U-937 and THP-1 cells (both from the American Type Culture Collection) and Ba/F3 parental cells and derivatives (generous gift of A. D. D’Andrea) were cultured in RPMI 1640 supplemented with 10% heat-inactivated fetal calf serum (Gemini), IL-3, penicillin (100 U/ml), and streptomycin (100 mg/ml). Total RNA was isolated by the guanidium thiocyanate method (2). RNase protection assays were performed as described previously (26). The antisense RNA probe for the murine SR corresponds to codons 84 to 158 that are common to both type I and II SR. The antisense probe for hGH corresponds to the BglII/SmaI fragment of exon 5 of the hGH gene.
Cell lines were transiently transfected by electroporation (5, 50) in 200 μl of OptiMEM (Gibco) with 10 μg of luciferase reporter plasmid and 0.1 μg of β-actin-lacZ reporter plasmid, as an internal transfection control, at 250 V and 960 μF, using a Bio-Rad electroporator with capacitance extender. Cells were resuspended in RPMI 1640 and incubated with additional GM-CSF (5 ng/ml) for 12 to 72 h. Salmon sperm DNA was used to balance all transfection groups to equal amounts of transfected DNA. Luciferase and β-galactosidase enzymatic activities were determined as previously described (25), and luciferase activity was normalized to the β-galactosidase standard.
Nuclear extracts were prepared from exponentially growing U-937 and Ba/F3 cell types maintained in the absence or presence of GM-CSF. The cells were then washed with ice-cold phosphate-buffered saline and incubated in buffer A (10 mM HEPES [pH 7.9], 1.5 mM MgCl2, 10 mM KCl, 1 mM dithiothreitol) on ice for 10 min. After centrifugation, the cells were resuspended in buffer A plus 0.2% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 4 mM benzamidine, 10 mg of leupeptin per ml, and 10 mg of aprotinin per ml and placed on ice for 5 to 10 min to allow lysis of the plasma membrane. The nuclei were isolated by centrifugation and resuspended in buffer C (20 mM HEPES [pH 7.9], 20% glycerol, 400 mM KCl, 0.2 mM EDTA, 1 mM dithiothreitol, proteinase inhibitors) and incubated for 30 min at 4°C. After centrifugation at 65,000 rpm for 30 min in a Beckman TLA-100 rotor at 4°C, aliquots of nuclear extract protein were frozen at −80°C.
DNase footprinting studies were performed with 10 to 20 fmol of DNA probe end labeled with 32P to a specific activity of 10,000 cpm/fmol containing 100,000 cpm of radioactivity. The probe was combined with 0 to 30 μg of nuclear extract protein diluted in 20 μl of binding buffer (40 mM KCl, 30 mM HEPES [pH 7.9], 12.5 mM MgCl2, 1 mM dithiothreitol, 20% glycerol, 0.1% Nonidet P-40) as described above and incubated at room temperature for 30 min. Limited DNase digestions were performed at room temperature after addition of 50 μl of a solution containing 10 mM MgCl2 and 5 mM CaCl2 and then 2 to 20 ng of DNase I. After 1 min, the digestion was stopped by addition of 90 μl of a solution containing 20 mM EDTA (pH 8.0), 1% sodium dodecyl sulfate, 0.2 M NaCl, and 125 μg of tRNA per ml. Samples were extracted twice by phenol-chloroform, precipitated, and resolved on a denaturing 10% polyacrylamide gel. The patterns of DNase footprints were visualized by autoradiography.
Electrophoretic mobility shift assays were performed with approximately 1 ng of synthetic double-stranded DNA probes labeled to a specific activity of 100,000 cpm/ng. Probes were combined with 0 to 8.5 μg of nuclear extract protein in 30 μl of total binding buffer [final composition, 5 mM HEPES (pH 7.9), 40 mM KCl, 2.5 mM EDTA, bovine serum albumin (0.15 μg/μl), poly(dI-dC) (0.15 μg/μl)] and incubated on ice for 30 min. Antibodies to specific STAT proteins (Santa Cruz Biotechnology) were added to DNA binding reactions and incubated on ice 20 min prior to electrophoresis. Protein-DNA complexes were resolved by nondenaturing polyacrylamide gel electrophoresis and detected by autoradiography. The sequences of the sense strands of synthetic probes were as follows: footprint B, 5′-GAT CCT GGC TTA AAT TAA TTT TGG ACA CTT TTA CAA CTC ATT TA-3′; footprint C, 5′-GAT CCC TCA TTT AAA CTA CAT GGG TCT TTC CTA CAA TTC TGA AGC CAT AGT TCC CT-3′; and gamma interferon activation sequence (GAS) consensus, 5′-GAT CCA GAT TTC TA GGA ATT CAA TCC A-3′.
Expression of the SR-A gene serves as a marker of macrophage differentiation. This gene has previously been demonstrated to contain information in both proximal promoter and distal enhancer elements that target reporter gene expression to macrophages of transgenic mice (13). Within these regions, sites for AP-1, AP-1/Ets ternary complexes, and PU.1, as diagrammed in Fig. Fig.1A,1A, are required for macrophage-specific expression and transcriptional activation of reporter genes in response to phorbol esters (25, 49). To investigate mechanisms responsible for transcriptional activation by M-CSF and GM-CSF, we examined the expression of both the endogenous scavenger receptor gene and SR-A–hGH fusion gene transcripts during differentiation of macrophages from bone marrow progenitor cells. In an RNase protection assay, bone marrow cells were found to exhibit little or no expression of either endogenous scavenger receptor mRNA or reporter gene expression, whether directed by information extending to −6.5 kb from the transcriptional start site (SEP-hGH [Fig. 1B]) or a minimal EP element (EP-hGH [Fig. 1B, lanes 4 and 9]).
Culture of progenitor cells in the presence of M-CSF to promote macrophage proliferation and differentiation resulted in marked increases in both scavenger receptor mRNA and transgenic reporter (hGH) mRNA directed by either the SEP or EP regulatory element (Fig. (Fig.1B,1B, lanes 5, 6, 10, and 11). Thus, the minimal EP regulatory elements were sufficient to confer M-CSF responsiveness, consistent with previous studies (14). Culture of progenitor cells in GM-CSF to promote differentiation of granulocytes and macrophages also resulted in a marked increase in expression of the endogenous scavenger gene in the adherent macrophage population (Fig. (Fig.1B,1B, lanes 7, 8, 12, and 13). However, while the longer SEP-hGH transgene was also responsive to GM-CSF (Fig. (Fig.1B,1B, lanes 7 and 8), the minimal EP-hGH transgene was not (Fig. (Fig.1B,1B, lanes 12 and 13). The differential responses of the SEP-hGH and EP-hGH transgenes were confirmed by quantitation of reporter (hGH) protein levels in progenitor cells treated for 3 days with either M-CSF or GM-CSF (Fig. (Fig.1C).1C). Thus, regulatory elements in the minimal EP of the SR-A gene were sufficient for responsiveness to M-CSF but were not sufficient to mediate a transcriptional response to GM-CSF.
We next evaluated the effects of combinations of M-CSF and GM-CSF on the expression of SEP-hGH and EP-hGH transgenes. Bone marrow progenitor cells were cultured in the presence of 20 ng of M-CSF per ml and increasing concentrations of GM-CSF. As illustrated in Fig. Fig.1D,1D, treatment of cells with the combination of M-CSF and GM-CSF led to a further increase in expression of the SEP-hGH transgene. These factors may act synergistically, because the half-maximal response to the addition of GM-CSF occurred at a concentration that alone has no effect on SEP expression (~0.05 ng/ml) and is insufficient to promote macrophage differentiation (data not shown). Surprisingly, GM-CSF strongly inhibited the response of the minimal EP-hGH transgene to M-CSF. Thus, GM-CSF exerted opposite effects on M-CSF-dependent transcription, dependent on the presence or absence of GM-CSF regulatory elements. These observations were confirmed in assays using several independently derived transgenic lines (data not shown) and therefore do not represent artifacts of the site of integration of the transgene. To determine whether the inability of the minimal EP-hGH transgene to respond to GM-CSF was due to removal of proximal sequences (−697 to −235) or to removal of sequences flanking the distal enhancer, we generated transgenic mice in which the minimal enhancer (−4.1 to −4.5 kb) was linked to promoter information extending from −697 to +45 bp (EXP-hGH [Fig. 1A]). This transgene was responsive to M-CSF but did not respond to GM-CSF (Fig. (Fig.1C).1C). These results indicated that sequences required for GM-CSF responsiveness resided in the distal enhancer.
To identify cis-active elements responsible for transcriptional activation of the SR gene in response to GM-CSF, a series of deletion mutants of the upstream regulatory elements were evaluated in transfected bone marrow progenitor cells cultured in the presence of GM-CSF. These experiments suggested that regulatory elements necessary for transcriptional activation by GM-CSF resided in a 300-bp region immediately upstream of the M-CSF-dependent enhancer element present in the EXP construct (Fig. (Fig.2A).2A). To determine whether these sequences were sufficient to confer GM-CSF-dependent activation to the minimal EP regulatory elements in vivo, two transgenic lines were established. Both of these lines demonstrated GM-CSF-dependent induction of hGH reporter gene expression in bone marrow progenitor cells (Fig. (Fig.2B),2B), thus defining the 300-bp region from −4.5 to −4.8 kb as a GM-CSF-responsive enhancer element.
Previous experiments in the THP-1 monocytic leukemia cell line suggested that AP-1 and Ets protein binding sites were necessary for activation of the SR-A gene during macrophage differentiation induced by phorbol ester (50). To evaluate the roles of AP-1 and Ets proteins in regulating SR-A expression in vivo, mutations were introduced into either the three AP-1 sites or the two Ets sites in the context of the EXP-hGH transgene (Fig. (Fig.3A).3A). The resulting transgenic lines were first assayed for expression in the peritoneal macrophage population. Mutation of each of the AP-1 sites and each of the Ets sites resulted in measurable hGH expression in peritoneal macrophages in one of six and two of six independently derived transgenic lines, respectively, while the control EXP transgene was expressed in six of nine independently derived lines. Although the differences in the percentage of expression observed for the mutant SR-A promoters and the wild-type promoter did not reach statistical significance (EPmAP1-hGH versus EXP-hGH, P = 0.1477; EPmets-hGH versus EXP-hGH, P = 0.6889, using two tailed P values adjusted for multiple comparisons), we noted that only high-copy-number (10 to 100) mutant promoter constructs expressed hGH, whereas even single-copy integrations of the wild-type EP-hGH construct expressed hGH. To determine whether transgenes containing mutations in AP-1 or Ets sites were regulated by M-CSF, experiments were performed in bone marrow progenitor cells. Transgenic lines that exhibited no reporter gene expression in peritoneal macrophages also exhibited no expression in M-CSF-treated bone marrow progenitor cells (data not shown). Each of the transgenes containing selective mutations in AP-1 or Ets binding sites that were expressed in peritoneal macrophages was also found to be activated in progenitor cells in response to M-CSF (Fig. (Fig.3B).3B). We therefore next evaluated the expression of transgenes in which the AP-1 and Ets sites were mutated in combination (EXPmAE-hGH). Only one of nine independently derived founder animals was found to be expressed in peritoneal macrophages, which represented a statistically significant difference in penetrance compared to the EXP transgene (P = 0.045). Furthermore, the one transgenic line that was found to be expressed in peritoneal macrophages exhibited no response to M-CSF in bone marrow progenitor cells. In concert, these results indicate that AP-1 and cooperating Ets domain transcription factors are required for the transcriptional responses of the SR-A gene to M-CSF.
Because previous studies have demonstrated that GM-CSF can activate AP-1 and Ets-dependent promoters, we wished to further explore the basis for the differential effects of GM-CSF and M-CSF on SR-A transgenes that were observed in bone marrow progenitor cells. Experiments were performed in Ba/F3-GMR cells, which were derived from an IL-3-dependent B-cell leukemia by stable transfection of an expression vector encoding the α subunit of the GM-CSF receptor (50). In contrast to the parental Ba/F3 cells, Ba/F3-GMR cells confer transcriptional responses to GM-CSF. As illustrated in Fig. Fig.4A,4A, GM-CSF not only stimulated transcription from promoters containing the GM-CSF-responsive enhancer (GM-XP-luc and GMEXP-luc) but also activated the M-CSF-responsive promoter (EXP-luc). However, the durations of the transcriptional responses of the two classes of promoters were significantly different. The response of the M-CSF-responsive promoter, EXP-luc, was very transient, being maximal 12 h after GM-CSF treatment and returning to basal levels by 24 h. Induction of EXP-luc by GM-CSF was dependent on the AP-1 and Ets binding motifs, because point mutations of either of these classes of binding sites abolished the GM-CSF response (data not shown). In contrast, reporter genes containing the GM-CSF-responsive enhancer element defined in vivo (GM-XP-luc and GMEXP-luc) continued to exhibit high levels of expression for as long as 72 h following treatment with GM-CSF.
We next evaluated the effects of combinations of GM-CSF and independent activation of the Ras signaling pathway on M-CSF- and GM-CSF-responsive promoters in Ba/F3-GMR cells. Reporter gene expression was evaluated at 24 and 72 h, by which time GM-CSF activation of AP-1 and Ets proteins is no longer measurable. Cotransfection of the Val12-Ras vector resulted in a sustained increase in expression of each of the luciferase reporter genes (Fig. (Fig.4B),4B), indicating that activation of the Ras pathway is alone sufficient to activate the M-CSF-responsive enhancer in these cells. Val12-Ras also activated GM-XP, suggesting that the GM-CSF-responsive enhancer is also activated by Ras, as the −696 to +45 promoter is relatively unresponsive by itself in this assay (reference 49 and data not shown). When cells were cotransfected with Val12-Ras and treated with GM-CSF, additive or slightly more than additive effects were observed for the GM-XP-luc and GMEXP-luc reporter genes. In contrast, the combination of GM-CSF and cotransfected Val12-Ras resulted in less activity of the EXP-luc reporter gene at 72 h than was observed in the presence of cotransfected Val12-Ras alone. Thus, at extended time points, GM-CSF inhibited, rather than potentiated, Ras-dependent activation of the M-CSF-dependent enhancer, similar to the effects of GM-CSF on M-CSF-dependent expression of EP-hGH and EXP-hGH observed in transgenic bone marrow progenitor cells.
Although an immediate-early response of transgenes containing the GM-CSF-dependent enhancer to GM-CSF was not observed in undifferentiated bone marrow progenitor cells (Fig. (Fig.1B),1B), macrophage precursor cells represent a very small percentage (<3%) of this cell population. To determine whether the GM-CSF enhancer can also confer an immediate-early response to GM-CSF in macrophages, bone marrow progenitor cells were harvested from GMEXP-hGH transgenic mice and cultured in M-CSF for 48 h to establish a homogeneous population of adherent macrophages. The cells were then treated with GM-CSF for 12, 24, 48, and 72 h and harvested for analysis of hGH content. These experiments demonstrated a significant increase in hGH expression as early as 12 h following GM-CSF treatment, consistent with the immediate-early response observed in Ba/F3-GMR cells (Fig. (Fig.44C).
The functional characteristics of the GM-CSF-responsive enhancer suggested that it should be recognized by proteins accounting for both immediate and sustained transcriptional responses to GM-CSF. DNase I footprinting experiments were therefore performed with nuclear proteins extracted from Ba/F3-GMR cells at various times following treatment with GM-CSF. As illustrated in Fig. Fig.5A,5A, footprinting analysis of the sense and antisense strands revealed several hypersensitive sites and the presence of five footprints, labeled A through E, that encompassed nearly the entire GM-CSF-dependent enhancer. There was no significant change in the footprint pattern in nuclear extracts prepared from cells treated with GM-CSF for 1 h. However, by 12 h of GM-CSF treatment, occupancy of footprint C was significantly increased, the extents of footprints A, B, D, and E had expanded, and an altered profile of hypersensitive sites was apparent. These footprints remained mostly unchanged after 24 h of GM-CSF treatment, although a slight decrease in footprints D and C was noted. Thus, these experiments were consistent with a class of proteins that were constitutively bound and could potentially mediate immediate responses to GM-CSF, as well as a second class of proteins induced by GM-CSF that either replace or complement the constitutively bound proteins and contribute to the delayed transcriptional responses.
To investigate the possible identities of the factors binding to the GM-CSF-responsive enhancer, we performed a series of electrophoretic mobility shift assays using synthetic DNA probes corresponding to specific footprints. Representative experiments obtained for probes corresponding to footprints B and C are illustrated in Fig. Fig.6.6. Nuclear extracts prepared from untreated Ba/F3-GMR cells contained DNA binding activities for both of these probes that were not altered by treatment with GM-CSF for 30 min, consistent with the results of DNase I footprinting experiments. Similar results were obtained with DNA probes corresponding to footprints A and D (data not shown). In contrast, treatment of Ba/F3-GMR cells with GM-CSF for 30 min strongly induced a DNA binding activity for an oligonucleotide containing a consensus GAS element that could be partially supershifted by an antibody directed against STAT5a and STAT5b (Fig. (Fig.6C).6C). The GAS element failed to compete the binding of proteins to the footprint B and C oligonucleotides, and the antibody that recognizes STAT5a and STAT5b had no effect on the binding of these factors (Fig. (Fig.6A6A and B). Anti-STAT1 antibodies also had no influence on these DNA binding activities (data not shown). Thus, the factors that bind to footprints B and C of the GM-CSF-dependent enhancers are unlikely to be STAT proteins that are capable of recognizing the consensus GAS element, and they are immunologically distinct from STAT5a/b and STAT1. The protein DNA complexes forming on the footprint B and C oligonucleotides exhibited significantly different electrophoretic mobilities, and the footprint B oligonucleotide had no effect on the binding of proteins to footprint C. At high concentrations, the footprint C oligonucleotide reduced binding of proteins to footprint B, suggesting that it contains a weak binding site for the footprint B DNA binding activity. In concert, these experiments suggest that the GM-CSF-dependent enhancer is recognized by several distinct factors that together mediate both immediate and delayed responses to GM-CSF. The specific identities of these DNA binding proteins remain to be determined.
The finding that STAT5 proteins did not bind directly to the GM-CSF enhancer raised the question of whether JAK2 and/or STAT5 was required for transactivation. To address this question, we initially evaluated the function of the GM-CSF enhancer in Ba/F3 cells that stably express either the wild-type human common β subunit of the GM-CSF/IL-3/IL-5 receptor or C-terminal deletion mutants. Transactivation of the β subunit at residue 544 abolishes activation of the Ras pathway, while further transactivation to residue 455 prevents association of JAK2. As illustrated in Fig. Fig.7A,7A, truncation of the β subunit at residue 544 almost completely abolished the transcriptional response of the GM-CSF-responsive promoter to GM-CSF treatment, consistent with the ability of constitutively active Ras to activate this promoter.
We next evaluated effects of wild-type and dominant negative forms of STAT5 and JAK2 (16a) on the transcriptional activity of the GM-CSF-dependent promoter GM-XP-luc in Ba/F3 cells expressing the murine GM-CSFα subunit. As illustrated in Fig. Fig.7B,7B, overexpression of STAT5a and JAK2, but not STAT1α, potentiated the activity of the GM-CSF-dependent enhancer at 24 h, while dominant negative forms of STAT5 and JAK2 strongly inhibited transcription. Similar effects were observed at 72 h (data not shown). In contrast, dominant negative STAT5 did not inhibit Ras-dependent activation of a promoter containing the scavenger receptor AP-1/Ets sites (Fig. (Fig.77C).
Cells of the macrophage lineage serve diverse homeostatic and immunologic functions by differentiating into functionally distinct end cells that proliferate and differentiate under the influence of M-CSF, GM-CSF and other locally produced regulatory molecules (10). Studies of the osteopetrotic op/op mouse indicate that the macrophage lineage can be considered to consist of M-CSF-dependent and M-CSF-independent populations (47, 48). Mice homozygous for the op mutation lack M-CSF (46) and exhibit striking defects in some but not all macrophage populations. Osteoclasts, peritoneal macrophages, and lymph node subcapsular macrophages, for example, are severely reduced or absent. Other populations, however, including phagocytes in the splenic red pulp, thymic cortex, Kupffer cells, alveolar macrophages, and microglia, are largely unaffected in the absence of M-CSF (47, 48). These observations indicate the presence of multiple pathways for macrophage differentiation that give rise to cells that have overlapping but nonidentical functions.
The CSFs responsible for M-CSF-independent macrophage populations appear to be highly redundant. Disruptions of the genes encoding GM-CSF or the common β subunit of the GM-CSF, IL-3, and IL-5 receptors have had relatively little effect on any of the macrophage lineages, with the exception of a functional defect in the ability of alveolar macrophages to clear surfactant components (8, 29, 30). Thus, GM-CSF is not essential for the development of M-CSF-independent populations of macrophages. Nevertheless, the potent activities of GM-CSF in promoting the development of macrophage colonies in vitro and its expression by many cell types in vivo support the idea that it plays important physiologic roles in establishing M-CSF-independent macrophage populations.
Although M-CSF and GM-CSF promote different sets of specialized functions, many of the phenotypic characteristics of M-CSF- and GM-CSF-derived macrophages are similar. At the gene level, M-CSF and GM-CSF induce the expression of many macrophage- or myeloid-specific genes in parallel, including the SR-A gene. Because the receptors for M-CSF and GM-CSF are not related, these observations have raised the question of whether coordinate regulation of these genes reflects (i) activation of distinct sets of transcription factors that function independently or (ii) convergence of signal transduction pathways on a common set of transcription factors.
In this study, we have examined the mechanisms by which M-CSF and GM-CSF coordinately regulate the expression of the macrophage SR-A gene. SR regulatory elements consisting of a minimal 245-bp promoter and a 400-bp upstream enhancer were sufficient to confer macrophage-specific expression and M-CSF-dependent regulation in transgenic mice. Several lines of evidence indicate that transcriptional activation of the SR-A gene by M-CSF is mediated by a Ras-dependent signal transduction cascade that stimulates the expression and activities of AP-1 and Ets domain transcription factors. We have previously shown that c-Jun, JunB, and Ets2 bind to regulatory elements in the M-CSF-dependent enhancer and act synergistically to stimulate transcription (49). In the present study, we have shown that mutation of these binding sites abolishes SR-A promoter responses to M-CSF in transgenic mice, while expression of a constitutive form of Ras is sufficient to activate the M-CSF-responsive enhancer.
These observations raised the possibility that GM-CSF acts to regulate the SR-A gene in a similar manner, as the GM-CSF receptor also couples to Ras-dependent signal transduction pathways and activates AP-1 and Ets transcription factors (1, 16, 37, 50). Consistent with this possibility, GM-CSF was indeed capable of activating the M-CSF-responsive enhancer in Ba/F3-GMR cells, and this effect was mediated by AP-1 and Ets transcription factors. However, the response of the M-CSF-dependent enhancer to GM-CSF was very transient in Ba/F3 cells, corresponding to the temporal profile of an immediate-early gene. Furthermore, when evaluated in transgenic mice, GM-CSF not only was unable to activate SR-A promoter activity through the M-CSF-responsive enhancer but inhibited activation by M-CSF.
Regulation of the SR-A gene by GM-CSF in transgenic mice was instead found to depend on a distinct GM-CSF-responsive enhancer. This enhancer could be activated by both Ras and GM-CSF in a cell line expressing functional GM-CSF receptors, with maximal activation observed 72 h following GM-CSF treatment. DNase I footprinting experiments of the GM-CSF-responsive promoter using nuclear extracts prepared from these cells indicated the presence of both constitutively expressed DNA binding proteins, as well as GM-CSF-inducible proteins that become maximally expressed 12 h following GM-CSF treatment. The identities of these proteins remain to be established, but they do not appear to correspond to known STAT factors. The constitutively expressed proteins are likely to be subject to posttranslational regulation by Ras-dependent signal transduction pathways, because the GM-CSF-responsive enhancer exhibited relatively little transcriptional activity in Ba/F3-GMR resting cells in which these proteins were found to be present but was very active following forced expression of constitutively active Ras. Similarly, truncation of the distal C terminus of the GM-CSF/IL-3/IL-5 receptor β subunit, which couples to Ras, significantly reduced GM-CSF-dependent transcription. While immediate-early effects of Ras on the GM-CSF-dependent enhancer are likely mediated by intracellular signaling pathways, delayed effects could potentially reflect the induction of other cytokines that indirectly activate the GM-CSF-dependent promoter in an autocrine manner.
In concert, our findings are most consistent with a model in which M-CSF and GM-CSF regulate the expression of the SR-A gene by fundamentally different mechanisms (Fig. (Fig.8).8). While Ras appears to play an important role in regulating both enhancers, cross talk between GM-CSF-dependent signal transduction pathways and targets of the M-CSF receptor that are activated by Ras is insufficient to mediate activation of the M-CSF-responsive enhancer in primary bone marrow progenitor cells. The relatively restricted utilization of Ras signaling pathways by M-CSF and GM-CSF may reflect the differential assembly of receptor-associated signaling complexes that are relatively dedicated to specific downstream targets (43). Differential utilization of Ras signal transduction pathways by M-CSF and GM-CSF is consistent with the distinct phenotypic properties of M-CSF- and GM-CSF-derived macrophages. It will be of considerable interest to identify the transcription factors responsible for activation of the SR-A gene by GM-CSF, as these factors are likely to be important in establishing the phenotypic properties of the M-CSF-independent population of macrophages.
We thank Tanya Schneiderman for assistance with preparation of the manuscript. We also thank Alan D’Andrea for dominant-negative STAT5 and JAK2 expression vectors and helpful discussions.
A.C.L. is supported by a Physician Scientist grant from the NIH. J.S.W. is supported by an NIH Medical Scientist Training Program grant to U.C. San Diego. C.K.G. is an Established Investigator of the American Heart Association. This study was supported by NIH grants CA 52599 and P50HL14197 to C.K.G.