Classical STAT1 activation in response to TLR agonists occurs by phosphorylation of the Y701 and S727 residues through autocrine type I IFN signaling and p38 MAPK signaling, respectively. In this study, we report that the TLR9 agonist CpG DNA induced Ifn-β mRNA, as well as downstream type I IFN-dependent genes, in a MyD88-dependent manner in mouse myeloid dendritic cells. This pathway was required for maximal TNF and IL-6 secretion, as well as expression of cell surface costimulatory molecules. By contrast, neither A- nor B-type CpG-containing oligonucleotides induced Ifn-β in mouse bone marrow-derived macrophages (BMM) and a CpG-B oligonucleotide did not induce IFn-β in the macrophage-like cell line, J774. In BMM, STAT1 was alternatively activated (phosphorylated on S727, but not Y701), and was retained in the cytoplasm in response to CpG DNA. CpG DNA responses were altered in BMM from STAT1S727A mice; Il-12p40 and Cox-2 mRNAs were more highly induced, whereas Tlr4 and Tlr9 mRNAs were more repressed. The data suggest a novel inhibitory function for cytoplasmic STAT1 in response to TLR agonists that activate p38 MAPK but do not elicit type I IFN production. Indeed, the TLR7 agonist, R837, failed to induce Ifn-β mRNA and consequently triggered STAT1 phosphorylation on S727, but not Y701, in human monocyte-derived macrophages. The differential activation of Ifn-β and STAT1 by CpG DNA in mouse macrophages vs dendritic cells provides a likely mechanism for their divergent roles in priming the adaptive immune response.
Activation of innate immune cells such as macrophages and dendritic cells (DC)3 commonly involves concerted signaling by cytokines such as IFNs, and microbial products such as LPS and CpG DNA (1). Many cytokines act through members of the hemopoietin receptor family, and their actions are coupled to the JAK-STAT signaling cascade. Microbial products act through the TLR family, which are coupled via adaptor proteins such as MyD88 to downstream effectors such as NF-κB. IFNs prime for maximal activation in response to the TLR agonists LPS and CpG DNA, and the IFN-responsive STAT family member STAT1 can integrate the two pathways; IFNs and LPS both induce STAT1 phosphorylation and STAT1-dependent gene expression (1, 2).
TLRs differ in the use of specific adaptors and consequent induction of target genes. The response to LPS requires TLR4 (3), which signals through both MyD88-independent (mediated by the TIR domain-containing adaptor proteins 1 and 2 and MyD88-dependent pathways (reviewed in Ref. 4). By contrast, CpG DNA-dependent signaling via TLR9 absolutely requires MyD88 (5). Consequently, LPS, but not CpG DNA, induces a set of MyD88-independent genes in macrophages (6, 7). One of the MyD88-independent genes induced by LPS in macrophages encodes the type I IFN, IFN-β (8, 9). LPS-induced IFN-β acts in an autocrine fashion to activate tyrosine phosphorylation of STAT1, which in turn induces a set of target genes, including inducible NO synthetase in macrophages (9, 10).
TLR signaling also modifies STAT1 more directly through phosphorylation at S727 (11, 12), which occurs via a p38 MAPK-dependent pathway (13, 14, 15). The contributions of S727 phosphorylation to the actions of STAT1 were analyzed using STAT1S727A mutant mice, in which an alanine is substituted for the serine at residue 727 of STAT1 (12). These mice were resistant to LPS shock, and macrophages from these mice were less sensitive to the combined effects of IFN-γ plus LPS in vitro (12). Phosphorylation of STAT1 on Y701 is required for homo- and heterodimerization and for nuclear translocation and transcriptional activation. Nevertheless, STAT1 phosphorylated at S727, but not Y701, can still regulate gene expression (16, 17, 18, 19, 20). STAT1 serine phosphorylation was required, whereas tyrosine phosphorylation and homodimerization were dispensable, for maximal constitutive expression of caspase-2 and -3 in fibroblasts (16). Furthermore, Y701 mutants of STAT1 still regulated the constitutive expression of several genes, including CycA, Hsp70, and Lmp2, in fibroblasts (17).
Early studies on CpG DNA signaling in macrophages implied that this stimulus did not elicit type I IFN production (21, 22), consistent with the inability of CpG DNA to activate MyD88-independent signaling pathways (5). However, recent reports demonstrated that CpG DNA triggered IFN-β production in myeloid DC in a MyD88-dependent fashion via IFN regulatory factor (IRF)1 (23, 24). In addition, in plasmacytoid DC (pDC), endosomal retention of CpG DNA enabled MyD88-dependent activation of the IRF family member, IRF7, leading to type I IFN production (25, 26). CpG-containing oligonucleotides that effectively induce type I IFN in pDC are termed A type and have a palindromic central CpG motif, which can form a double-stranded region, and strings of G residues at both the 3′ and 5′ ends. A-type and conventional single-stranded B-type CpG-containing oligonucleotides traffic differentially in pDC; A-type oligonucleotides are retained in early endosome-like structures and initiate signaling leading to type I IFN production, whereas B-type oligonucleotides traffic to a late endosome/lysosome compartment and more strongly induce inflammatory cytokines, such as TNF (26). Differences between macrophages and DC in their response to MyD88-dependent signals could be important in constraining pathological consequences of macrophage activation (e.g., septic shock), while permitting activation of acquired immunity.
In this study, we confirm that macrophages differ from DC in their response to MyD88-dependent signaling by CpG DNA. The MyD88-dependent production of IFN-β was essential for CpG DNA-mediated DC maturation, and provides an important insight into the powerful adjuvant activity of CpG DNA. Conversely, the failure of macrophages to respond to CpG DNA with inducible IFN-β expression reveals a novel cytoplasmic function for STAT1, which limits the production of a subset of inflammatory mediators. We propose that the divergent responses of macrophages and DC to CpG DNA could derive from differential expression of members of the IRF family.
Materials and Methods
Recombinant mouse IFN-γ (R&D Systems) and IFN-β (PBL) were used at a final concentration of 500 pg/ml and 100 U/ml, respectively. LPS (from Salmonella minnesota; Sigma-Aldrich) was used at a final concentration of 10 ng/ml, unless otherwise stated. Lipoteichoic acid (LTA) from Staphylococcus aureus was a gift from S. Morath (Konstanz, Germany); PAM3CSK4 was obtained from EMC Microcollections. R837 (imiquimod), a specific TLR7 agonist (27, 28), was purchased from InvivoGen. Human rCSF-1 (a gift from Chiron) was used at a final concentration of 1 × 104 U/ml (100 ng/ml). Unless otherwise stated, the well-characterized B-type phosphorothioate-modified oligonucleotide 1668S (5′-tccatgacgttcctgatgct-3′) (29) was used in all experiments at a final concentration of 0.1 μM. The GC inversion of this oligonucleotide, 1668S-GC, 5′-tccatgagcttcctgatgct-3′, was used to control for CpG DNA-independent effects on cells (29). In experiments in which A-type oligos were examined, the D19 oligonucleotide (5′-ggTGCATCGATGCAgggggG-3′) (26) and the 1585 oligonucleotide (5′-ggGGTCAACGTTGAgggggG-3′) (30) were used. Upper and lower case letters indicate bases with phosphodiester- and phosphorothioate-modified linkages, respectively. All oligonucleotides were synthesized by Geneworks. Abs used in Western blotting were purchased from Cell Signaling Technology (anti-STAT1, anti-phospho-Y701-STAT1, anti-p38 MAPK, anti-phospho-p38 MAPK, anti-ERK-1/2, anti-phospho-ERK-1/2, anti-CSF-1R, goat anti-rabbit IgG), Calbiochem (anti-phospho-S727-STAT1), or Santa Cruz Biotechnology (anti-PU.1). All primary Abs were used at 1:1000, except anti-PU.1, which was used at 1:5000, whereas the secondary Ab was used at 1:3000.
Preparation of macrophages and DC
Bone marrow-derived macrophages (BMM) were obtained by ex vivo differentiation from mouse bone marrow progenitors in the presence of rCSF-1 over 7 days, as described previously (31, 32). C57BL/6 male mice (6–8 wk old) kept under specific pathogen-free conditions were used in BMM generation. For experiments with wild-type (WT) vs STAT1S727A mice, bone marrow progenitors were pooled from equal numbers of age-matched male and female mice in each experiment. J774 macrophages (American Type Culture Collection) were maintained in RPMI 1640 supplemented with 5% FCS and antibiotics (100 IU/ml penicillin G and 100 IU/ml streptomycin sulfate). GM-CSF-derived (CD11c+, B220−) bone marrow-derived DC (BMDC) were prepared, as described previously (33), from C57BL/6 female 6- to 8-wk-old mice. Where the effect of IFNAR1 or MyD88 genotype was examined, SV129 IFNAR1+/+, SV129 IFNAR1−/−, BALB/c MyD88+/+, or BALB/c MyD88−/− mice were used as a source of DC and BMM. All experiments with mouse primary cells were conducted in accordance with local animal ethics guidelines.
Human macrophages were derived from monocytes isolated from blood donations obtained from the Australian Red Cross Blood Bank. PBMC were separated from buffy coat by Ficoll-Paque (Amersham) density centrifugation. CD14+ monocytes were isolated from the PBMC by positive selection for CD14 using MACS magnetic beads (Miltenyi Biotec), according to the manufacturer’s instructions. Monocytes were seeded on tissue culture plastic at 15 × 106 per plate in IMDM (Invitrogen Life Technologies) with 10% FCS, CSF-1, and antibiotics, as above, and allowed to differentiate into macrophages for 6 days. Macrophages were supplemented with CSF-1-replete medium on day 5, and both the adherent and nonadherent cells were reseeded on day 6 and harvested for analysis on day 7.
Determination of mRNA expression by quantitative PCR (qPCR)
Cellular RNA was extracted using RNeasy kits (Qiagen) or HighPure RNA kits (Roche). Contaminating genomic DNA was removed using DNA-Free (Ambion), and cDNA was synthesized using Superscript III (Invitrogen Life Technologies) or reverse transcriptase (MBI Fermentas) and oligo(dT) primer. Negative control reactions were performed as above, with the omission of the enzyme. cDNA was quantitated using the Platinum SYBR Green quantitative PCR system (Invitrogen Life Technologies) or TaqMan reagents (AbGene) and fluorogenic probes (FAM/TAMRA; Eurogentec). Negative control reactions were cycled alongside test samples to ensure the absence of contaminating genomic DNA. Data were analyzed using ABI Prism software. Expression was determined relative to the abundance of housekeeping genes (SYBR Green, hypoxanthine guanine phosphoribosyltransferase Hprt; TaqMan, β-actin). Transcript abundance (Gene/Hprt or Gene/β-actin) and SD were determined, as recommended by Applied Biosystems. Where inducible expression in cells of different genotypes was compared, expression was determined as a fold induction relative to basal expression within the genotype. Profiles were quantitated in at least three separate experiments using completely independent preparations of cells and RNA extracts, unless otherwise stated. Real-time PCR primer pairs were designed and tested as described previously (34). Primer sequences are available on request.
The levels of secreted cytokines (IFN-γ-inducible protein-10 (IP-10), TNF, and IL-6) in culture supernatants were determined using ELISA kits (OptEiA; BD Biosciences).
Cell surface proteins were stained using anti-CD80 and anti-CD86 Abs (BD Biosciences) and analyzed by flow cytometry using a FACS Canto (BD Biosciences).
Western analysis of phospho-specific protein forms included phosphatase inhibitors (1 mM sodium vanadate, 1 mM sodium pyrophosphate, 1 mM sodium molybdate, 10 mM sodium fluoride) in all steps of protein extract preparation and immunodetection. The fractionation of nuclear and extranuclear proteins was performed at 4°C, and all solutions for extract preparation were prechilled and included 5 μM PMSF and 1× protease inhibitors (Roche). Cells were harvested and washed twice with PBS. The cell pellet was resuspended in 225 μl of plasma membrane lysis buffer (10 mM HEPES (pH 7.4), 1.5 mM MgCl2, 10 mM KCl, 0.1% Nonidet P-40) per 10 × 106 cells, and incubated for 5 min. Nuclei were separated from the solubilized extranuclear extract by centrifugation, and the nuclei were washed three times with plasma membrane lysis buffer. Nuclei were resuspended in 75 μl of nuclei lysis buffer (20 mM HEPES (pH 7.8), 420 mM NaCl, 20% glycerol, 0.2 mM EDTA, 1.5 mM MgCl2) per 10 × 106 cells and incubated for 10 min. The nuclear extract was clarified by centrifugation. Total protein in whole cell, nuclear, and extranuclear extracts was quantitated using the BCA Protein Assay Kit (Pierce) to normalize total protein loading. Whole cell extracts were prepared by boiling lysis (35), and whole cell and fractionated extracts were analyzed by Western blot, as described previously (35, 36).
CpG DNA induces type I IFN and downstream target genes via MyD88 in mouse DC, but not macrophages
CpG DNA did not induce IFN-β expression in mouse bone marrow-derived macrophages (BMM) or RAW264.7 cells (26), but recent studies have identified a MyD88-dependent pathway leading to Ifn-β production in BMDC (8, 23, 24). Fig. 1 shows a direct comparison of the effects of the B-type CpG DNA, 1668S, on mouse macrophages and DC. CpG DNA strongly induced Ifn-β mRNA expression (Fig. 1,A), as well as expression of the type I IFN target gene, Ip-10 (Fig. 1,B) in BMDC. In contrast, the mouse macrophage cell line J774 responded to LPS, but not CpG DNA, with inducible Ifn-β mRNA expression (Fig. 1,C), as well as downstream responses such as induction of Ip-10 mRNA (Fig. 1,D) and protein (data not shown). The responsiveness of J774 cells to both stimuli was confirmed by assessing Tnf mRNA expression (data not shown). Similarly, stimulation of BMM with LPS, but not CpG DNA, for 3 h elicited robust Ifn-β mRNA expression (Fig. 1,E). In these cells, the TLR2 ligands, Pam3CSK4 and LTA, were also ineffective at inducing Ifn-β gene expression (Fig. 1,E). Even after prolonged incubation with CpG DNA, there was no induction of Ifn-β (Fig. 1,F), whereas Il-12p40 (Fig. 1 G) and Tnf (data not shown) were highly induced.
Distinct classes of CpG-containing oligonucleotides vary in their capacity to elicit type I IFN production. In pDC, A-type oligonucleotides such as D19 and 1585 are a more potent stimulus for Ifn-β expression than B-type oligonucleotides (26, 37). In BMM, however, neither A-type nor B-type oligonucleotides induced robust Ifn-β expression (Fig. 1,H) or expression of the IFN-β target gene Ip-10 (Fig. 1,I). The activity of A-type oligonucleotides was confirmed by assessing Tnf expression (Fig. 1,J). Because the nature of the CpG DNA did not affect Ifn-β induction and the B-type oligonucleotide was a more potent activator of macrophage function (Fig. 1 J), the B-type CpG DNA 1668S was used for all further studies.
The response to CpG DNA and other TLR agonists is dependent upon the adaptor protein MyD88. Although LPS elicited IP-10 production in both BMM (Fig. 2,A) and BMDC (Fig. 2,C) from MyD88-deficient mice, production of IP-10 in response to CpG DNA was absolutely dependent on MyD88 in BMDC. As expected, TNF production from both BMM (Fig. 2,B) and BMDC (Fig. 2 D) was either partially dependent (LPS) or totally dependent (CpG DNA) on MyD88. Hence, CpG DNA-triggered IFN-β production in the myeloid DC lineage is MyD88 dependent. By extension, because macrophages can respond to CpG DNA (and TLR2 agonists) in a MyD88-dependent manner, the signaling blockade in these cells lies specifically within the pathways linking MyD88 to IFN-β production.
The type I IFN pathway is required for both cytokine production and cell surface costimulatory molecule expression in response to CpG DNA in BMDC
Autocrine actions of CpG DNA-induced IFN-β have not been widely implicated in DC maturation. We assessed the phenotype of BMDC derived from mice in which the IFN-αβ receptor is nonfunctional due to targeted deletion of the receptor subunit IFNAR1. As expected, induction of Ip-10 mRNA by LPS and CpG DNA was impaired in BMDC from IFNAR1-deficient animals (Fig. 3,A). CpG DNA- and LPS-mediated up-regulation of surface expression of the costimulatory molecules CD80 (Fig. 3,B) and CD86 (Fig. 3,C), as well as production of both TNF (Fig. 3,D) and IL-6 (Fig. 3,E), was also dependent on type I IFN signaling. However, this pathway did not play a major role in the response to Pam3Cys (Fig. 3, D and E).
The Tnf and Il-6 genes are classically regarded as primary response genes in TLR signaling. Because CpG DNA induced Ifn-β mRNA expression in BMDC, but not macrophages (Fig. 1), we predicted that exogenous IFN-β would enhance TNF and IL-6 production in response to CpG DNA in macrophages, but not DC. This was indeed the case (Fig. 3, F–I). As expected, IFN-β did not amplify BMM (Fig. 3, F and G) or BMDC (Fig. 3, H and I) responses to LPS, because this agonist induced Ifn-β mRNA expression in both cell types (Fig. 1). Thus, autocrine IFN-β allows DC, but not macrophages, to elicit maximal cytokine production and costimulatory molecule expression in response to CpG DNA.
CpG DNA activates serine, but not tyrosine, phosphorylation of STAT1 in BMM
TLR agonists activate STAT1 phosphorylation on S727 via a p38-dependent pathway (13). Since IFN-β was not produced in response to CpG DNA in macrophages (Fig. 1), CpG DNA induced an alternative STAT1 activation state in which S727, but not Y701, was phosphorylated in BMM (Fig. 4,A). In contrast, both LPS (Fig. 4,B) and IFN-γ (Fig. 4 C) triggered dual Y701/S727 phosphorylation of STAT1 in BMM, as previously reported (38, 39, 40). The time course of Y701 phosphorylation in response to LPS was delayed, reflecting the autocrine nature of IFN-β signaling.
STAT1 homo- and heterodimerization are mediated by phosphorylation of Y701 at the C-terminal end of the STAT1 Src homology 2 domain (41). Fig. 5 shows that STAT1 translocated to the nucleus during classical activation (Y701, S727 diphosphorylation) by LPS or IFN-γ, but was retained in the cytoplasm following alternative activation (monophosphorylation at S727) by CpG DNA. The integrity of the nuclear and extranuclear extracts used in this analysis was confirmed by immunoblotting for PU.1 (a nuclear transcription factor) and the exclusively extranuclear protein, CSF-1R (a plasma membrane receptor). Total STAT1 levels served as an internal control within the experiment. We and others showed that LPS and CpG DNA caused a loss of cell surface CSF-1R due to cleavage of the extracellular domain, whereas IFN-γ only modestly down-regulated cell surface CSF-1R expression (34, 36, 42, 43). Consistent with these previous findings, CSF-1R levels in extranuclear extracts were markedly reduced in response to LPS and CpG DNA, but were only modestly affected by IFN-γ (Fig. 5), thus confirming the activity of all stimuli used. Hence, BMM responding to CpG DNA provide a model system in which the function of S727 phosphorylation can be examined in isolation.
Alternatively activated STAT1 regulates gene expression in response to CpG DNA in BMM
The function of STAT1 S727 phosphorylation in the CpG DNA response was examined using mice in which the S727 residue was replaced with an alanine (STAT1S727A mice) (12). BMM from these mice showed a selective defect in serine phosphorylation in response to CpG DNA or IFN-γ, whereas IFN-γ-dependent Y701 phosphorylation was intact (Fig. 6,A). STAT1-independent signaling by CpG DNA was unaffected by the S727 mutation, because ERK-1/2 activation in response to CpG DNA was comparable between WT and STAT1S727A BMM (Fig. 6 B).
Given that CpG DNA activated the type I IFN pathway in DC, but not macrophages (Fig. 1), we predicted that STAT1, phosphorylated on S727, but not Y701, would have a unique function in macrophages. This was examined by comparing the expression of known CpG DNA-responsive genes in BMM from WT and STAT1S727A mutant animals. Induction of both Il-12p40 and Cox-2 mRNAs was enhanced 3- to 4-fold in BMM from the mutant mice (Fig. 7, A and B), whereas the induction of Il-6 and Tnf mRNA by CpG DNA was unaffected (Fig. 7, C and D). CpG DNA also selectively represses expression of some genes, including certain receptors and signal transducers. Both Tlr4 and Tlr9 mRNAs were transiently repressed by CpG DNA in WT BMM, and were hyperrepressed in STAT1S727A BMM (Fig. 7, E and F). Hence, phosphorylation of STAT1 at S727 constrains both the induction (Il-12p40, Cox-2) and the repression (Tlr4, Tlr9) of certain genes in response to CpG DNA. The data therefore support a regulatory role for serine-phosphorylated STAT1 in limiting macrophage responses to CpG DNA.
A TLR7 agonist triggers alternative STAT1 activation in primary human monocyte-derived macrophages (HMDM)
It is generally accepted that human monocytes do not respond to CpG DNA, which might imply that STAT1 cannot be alternatively activated in human monocytes or macrophages. We hypothesized, however, that any TLR agonist that activates the p38 MAPK pathway without concomitant induction of autocrine IFN-β signaling would initiate alternative STAT1 activation in human macrophages. We therefore examined the activation of this pathway in response to the TLR7 agonist, R837 (imiquimod) (27, 28), in HMDM. Although LPS strongly induced Ifn-β mRNA as expected, R837 was ineffective at inducing Ifn-β mRNA (Fig. 8,A) or the IFN-β target gene Ip-10 (Fig. 8,B) in HMDM. In contrast, this signal triggered robust Il-1β mRNA expression (330-fold, Fig. 8,C) and p38 MAPK phosphorylation (Fig. 8,D). Consistent with the ability of R837 to activate p38 MAPK, but not IFN-β signaling (Fig. 8, A, B, and D), R837 elicited alternative STAT1 activation (monophosphorylation at S727), whereas LPS triggered classical STAT1 activation (S727 and Y701 phosphorylation) in HMDM (Fig. 8 D). Thus, alternative and classical STAT1 activation occur in both mouse and human primary macrophages.
Although Ifn-β was initially identified as a MyD88-independent target gene in TLR3 and TLR4 signaling pathways in macrophages, CpG DNA induces this gene in a MyD88-dependent fashion in pDC (25, 26) and myeloid DC (23, 24). We confirmed that myeloid DC and macrophages are different in this respect; CpG DNA induced Ifn-β mRNA expression as well as IFN-β target genes in a MyD88-dependent fashion only in BMDC (Figs. 1 and 2) (23, 24). Although A-type oligonucleotides such as D19 and 1585 are reportedly a much more powerful stimulus for type I IFN production than B-type oligonucleotides in pDC (26, 37), we found that BMDC responded robustly to the B-type oligonucleotide 1668S in this respect (Fig. 1), consistent with the recent findings of Schmitz et al. (23). This pathway was important for myeloid DC maturation and function, because maximal TNF and IL-6 production and costimulatory molecule expression in response to CpG DNA were dependent upon autocrine IFN signaling (Fig. 3).
In macrophages, even A-type oligonucleotides were ineffective at inducing Ifn-β mRNA expression (Fig. 1,H) and downstream IFN target genes (Fig. 1,I). The endosomal trafficking of A-type oligonucleotides leading to IRF7-dependent induction of type I IFN (26) is likely to be unique to the antiviral function of pDC. The use of the autocrine type I IFN pathway in DC, but not macrophages, may contribute to the enhanced capacity of DC to act as professional APC, and implies that endogenous IFN contributes to functional divergence between macrophages and myeloid DC. The differential responses of DC and macrophages to CpG DNA are summarized in Fig. 9.
The mechanisms responsible for the divergent effects of CpG DNA on IFN-β expression in DC vs macrophages may be linked to differential expression or function of the IRF transcription factor family. The induction of IFN-β by CpG DNA requires IRF7 in pDC (25, 26) and IRF1 in BMDC (23, 24). In BMM, IFN-γ induced IRF1 mRNA expression and enabled the assembly of the MyD88-dependent pathway leading to IFN-β production in response to CpG DNA (23). Thus, a likely explanation is that either IRF1 expression or function is limiting in resting macrophages. However, differential expression of other IRF family members might also explain the absence of the MyD88-IFN-β pathway in this cell type. We recently showed that the mRNA expression of IRF4, as well as IRF8 that has been implicated in CpG DNA-specific signaling in DC (44), was down-regulated by the macrophage growth factor, CSF-1, in BMM (34). A prediction of this finding is that these factors would be more highly expressed in DC than in macrophages. Certainly, IRF4 and IRF8 are functionally required for DC differentiation in response to GM-CSF and FLT-3, respectively (45). We are currently characterizing the expression of the entire IRF family in macrophages vs BMDC to identify differentially expressed family members.
Although CpG DNA did not activate the classical type I IFN-dependent pathway in BMM, the phosphorylation of STAT1 on S727 (Fig. 4) led to an alternative activation state in which STAT1 was retained in the cytoplasm (Fig. 5). Gene expression profiles comparing WT and S727A STAT1 mutant macrophages (Fig. 7) suggest that phosphorylation on this amino acid contributes to CpG DNA-mediated signaling in macrophages. Serine phosphorylation of STAT1 regulates interactions with several binding partners (e.g., BRCA1, CBP, MCM5 (12, 46, 47, 48)). On this basis, we propose that S727-phosphorylated STAT1 functions to sequester transcriptional regulators in the cytoplasm of CpG DNA-stimulated macrophages. Such a mechanism was demonstrated in osteoblasts, where serine-phosphorylated STAT1 sequestered the RUNX2 transcription factor in the cytoplasm in an inducible manner that was independent of Y701 phosphorylation (49).
One possibility is that members of the IRF family could be targeted for cytoplasmic sequestration by S727-phosphorylated STAT1 in macrophages. For example, IRF9, which normally resides in both the nucleus and cytoplasm, was retained in the cytoplasm when coexpressed with STAT2 (50). The IRF9 nuclear localization signal is contained within the DNA binding domain and is conserved in IRF8 and IRF4 (50), and it is possible that these IRFs could be inducibly retained by STAT1 in the cytoplasm of macrophages, upon S727 phosphorylation. In support of such a mechanism, IRF8 was reported to bind to nontyrosine-phosphorylated STAT1 (51). Other IRF family members are also candidates. IRF1, which was reported to interact with STAT1 (17), positively regulated the inducible expression of Cox-2 (52) and Il-12-p40 (53), both of which were superinduced by CpG DNA in STAT1S727A BMM (Fig. 7). Similarly, IRF2, which can also form a complex with STAT1 (54), was implicated in driving IL-12p40 expression (53). STAT1S727A BMM may therefore permit the nuclear translocation of one or more IRF family members and the superinduction of Cox-2, Il-12-p40, and coregulated genes in response to CpG DNA; we are currently investigating this hypothesis.
The differing activity profiles of STAT1 in response to CpG DNA and LPS may underpin their very different toxicities. IFN-β signaling and classic STAT1 trans-activator function in macrophages contributes to LPS-mediated toxicity in vivo (55, 56). CpG DNA is less toxic than LPS in vivo, and the apparently anti-inflammatory actions of alternatively activated STAT1 may be protective in this context.
Alternative STAT1 activation was not just restricted to CpG DNA signaling in mouse macrophages. The TLR7 agonist, R837, induced a similar activation profile (monophosphorylation at S727) due to a failure of autocrine IFN-β signaling in HMDM (Fig. 8). It is likely that a range of stimuli that activate the p38 MAPK pathway without inducing autocrine IFN signaling trigger alternative STAT1 activation. Furthermore, this pathway may also operate as a feedback regulator of TLR4 signaling because LPS triggers sustained S727 phosphorylation, but transient Y701 phosphorylation (Fig. 4). That is, a shift from classical to alternative STAT1 function after prolonged activation via TLR4 may act to sequester transcriptional activators in the cytoplasm.
In summary, we propose that the IFN-β/STAT1 axis is a key regulator of CpG DNA signaling, and that divergent activation of this pathway between macrophages and DC dictates cell-specific responses to CpG DNA. CpG DNA induces autocrine IFN-β signaling and classic STAT1 transcription factor activities in myeloid DC, thereby potentiating expression of IFN-target genes, costimulatory molecules (CD80 and CD86), and proinflammatory cytokines (IL-6, TNF). The lack of this autocrine pathway in CpG DNA-stimulated macrophages enables alternative STAT1 activities that may include sequestration of gene regulatory factors to limit CpG DNA responses. This blockade could be overcome by IFN-γ, which can induce STAT1 Y701 phosphorylation and nuclear translocation. In such a scenario, STAT1 would be rapidly transformed from signaling repressor to transcriptional activator to elicit appropriate antimicrobial responses.
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 grants from the National Health and Medical Research Council of Australia (Project Grants 301211 and 301210). T.R. was supported by a research grant for the Scripps NeuroAIDS Preclinical Studies Center from the National Institute of Mental Health, Grant 2P30MH062261-07. A.H.D. was supported by Grants DFG Da592/2 and 592/3 from the German research foundation.
Abbreviations used in this paper: DC, dendritic cell; BMDC, bone marrow-derived DC; BMM, bone marrow-derived macrophage; HMDM, human monocyte-derived macrophage; IP-10, IFN-γ-inducible protein-10; IRF, IFN regulatory factor; LTA, lipoteichoic acid; pDC, plasmacytoid DC; qPCR, quantitative PCR; WT, wild type.