The expanding roles of ITAM adapters FcRγ and DAP12 in myeloid cells

Introduction

The immunoreceptor tyrosine-based activation motif (ITAM) intracellular signaling pathway represents perhaps the dominant mechanism by which immune cells respond to their environment. ITAM signaling was first recognized to be critical in lymphocyte stimulation by antigens through both B-cell and T-cell receptors (BCR and TCR) (reviewed in (1, 2)). Indeed, signaling through these receptors provides the overall conceptual framework for how ITAM pathways function in immune cells. In this framework, the receptor subunits, which mediate ligand binding, physically associate with different ITAM-containing adapter proteins (the CD79 subunits for the BCR and the CD3 subunits for the TCR) to communicate intracellular signaling. The ITAM itself consists of a highly conserved sequence D/ExxYxxL/I(x_6-8)YxxL/I (3). The ITAM-containing signaling adapter is coupled to the ligand binding receptor through charged amino acid interactions within the transmembrane regions of each protein (commonly aspartic acid in the ITAM-containing subunit and lysine or arginine in the receptor). Following engagement of the receptor by ligand (for example antigens in the case of the BCR or TCR), Src-family kinases become activated (by a yet unclear mechanism), which in turn phosphorylate the two tyrosine residues of the ITAM. These phospho-tyrosine residues form a docking site for ζ-associated protein of 70 kD (ZAP70) kinase or spleen tyrosine kinase (Syk) through interactions with Src-homology 2 (SH2) domains on the kinases. This induces activation of the kinase, which in turn phosphorylates a number of downstream substrates (for example, LAT in T-cells, SLP-76 in B cells) allowing for formation of a signaling complex that communicates to a number of pathways in the cell (MAPK, PLCγ/Ca²⁺, Vav/Rac or NF-κB as examples), leading to cellular activation (Fig. 1). Because this pathway was first defined for receptors containing immunoglobulin-like domains (such as the BCR and TCR), it is commonly referred to as the immunoreceptor (or ITAM) pathway. In this review, we define the term immunoreceptor as Ig superfamily or C-type lectin receptors that associate directly with ITAM-adapters through a positively charged amino acid in their transmembrane region.

In myeloid cells, the primary ITAM-containing signaling adapters are DAP12 and the FcεRIγ chain (referred to as FcRγ) (4, 5). These adapters were first defined by their role in signaling downstream of immunoreceptors present in NK cells or myeloid cells, respectively. All ITAM signaling adapters, including DAP12 and FcRγ, are structurally characterized as small (15 – 20 kD) proteins with very short extracellular regions that contain a cysteine residue to allow the molecules to form homodimers though disulfide bonds. Within the cytoplasmic region, both DAP12 and FcRγ have a single ITAM which, when dually phosphorylated, provides a high affinity binding site for the Syk family kinases ZAP-70 and Syk. The ITAM sequences of DAP12 and FcRγ are remarkably conserved through evolution, and related orthologs of these proteins are found in all vertebrates, demonstrating their central role immune cell signaling. Mutation of either of the ITAM tyrosines (or the respective SH2 domains in ZAP70/Syk kinases) completely blocks the signaling function of all associated immunoreceptors in the cell. Complete deletion of either DAP12 or FcRγ results in loss of cell surface expression of the majority of immunoreceptors to which the adapter pairs, demonstrating that formation transmembrane charge bridge stabilizes the receptor at the cell surface.

DAP12 and FcRγ are found in most cells of the innate immune system, including neutrophils, macrophages, various dendritic cell (DC) types, basophils, eosinophils, mast cells and NK cells. Within these cell types, these signaling adapters pair and mediate signaling from a remarkably large repertoire of immunoreceptors (Fig. 1) (6). Originally, both DAP12 and FcRγ were defined as pairing with immunoreceptors that induce cellular activation; for example DAP12 mediates NK cell activation through Ly49D and Ly49H in the mouse (7) or the CD94/NKG2C complex in the human (8), whereas FcRγ mediates macrophage, neutrophil and mast cell activation following engagement of their Fcγ or Fcε receptors (9, 10). The repertoire of immunoreceptors to which these adapters pair to mediate innate immune cell activation has since grown considerably. The ligands for some, but not all, of these immunoreceptors are known, and consist of both protein or carbohydrate structures present on many pathogenic organisms as well as endogenous cellular macromolecules.

A notable development in this field is the relatively recent description of the involvement of DAP12 and FcRγ in signaling pathways mediated by non-immunoreceptors, which we define as those receptors that do not directly associate with these signaling adapters using positively charged amino acids in their transmembrane regions (reviewed in (11-13)). Most of these studies have evolved out of phenotypic characterization of non-immunoreceptor signaling pathways in innate immune cells derived from DAP12 and/or FcRγ deficient mice. Examples of non-immunoreceptors in myeloid cells that rely on ITAM adapters for signaling include integrins, growth factor receptors and antigen presenting (MHC) proteins. Perhaps even more remarkable, it is now well established that DAP12 and FcRγ serve an inhibitory function in innate immune signaling as well (reviewed in (12, 14)). This discovery also emanates from phenotypic characterization of cells from DAP12 and/or FcRγ deficient mice or from cross-linking of associated receptors with antibodies.

This review will cover the new aspects of both the positive and negative (or inhibitory) regulatory function of these ITAM adapters in a variety of recently described signaling responses.

Positive signaling by ITAM adapters

As described above, both DAP12 and FcRγ were first defined by their ability to transmit signals leading to cellular activation by coupling to Ly49 receptors in NK cells or Ig (Fcγ, Fcε and Fcα) receptors in myeloid cells, respectively. However, the spectrum of receptors to which these molecules couples has now expanded dramatically and includes many non-immunoreceptor type proteins. FcRγ has now been demonstrated to mediate signaling through immunoreceptors such as Mincle (15-18), Dectin-2 (19), and MAIR-IV (20), but even more surprisingly from non-immunoreceptors such as the IL-3 receptor (21) or receptors involved in mycobacterial sensing (22). The spectrum of DAP12-associated receptors is even larger, extending way beyond just activating receptors in NK cells (reviewed in (4)). Included in this list are many immunoreceptors such as TREM-2 (23), MDL-1 (24) and PDC-TREM (25), as well as a surprising number of non-immunoreceptor molecules such as the c-Fms receptor (26, 27). In many cases, signaling from non-immunoreceptors requires both DAP12 and FcRγ, as loss of function is only seen in cells derived from mice lacking both adapters. This is certainly the case in the ITAM-mediated signaling of adhesion receptors such as integrins (reviewed in (28)) and P-selectin glycoprotein ligand (PSGL-1) as well as the involvement of DAP12 and FcRγ in IFNγ priming/activation (29, 30) or cross presentation of particulate antigens by MHC I (31). Whereas most immunoreceptors pair with either DAP12 or FcRγ exclusively, the myeloid and B cell expressed MAIR-II receptor has been shown to pair with both ITAM-containing signaling adapters (32).

Activation by FcRγ-associated receptors – new immunoreceptor associations

While most intensely studied in Fc receptor signaling pathways, it is now clear that the FcRγ adapter is important in other innate immune signaling responses. Indeed, the requirement for FcRγ in signaling through receptors other than FcRs was first realized in 1997 by Watson’s group in their studies of the platelet collagen receptor GPVI (33). FcRγ was subsequently shown to associate with a variety of NK and myeloid-expressed receptors, including OSCAR, PIR-A and activating ILT receptors (6). Recently, Sato et al showed that FcRγ associates with the fungal receptor Dectin-2 (19). This receptor is present at high levels on activated DCs and macrophages where it is responsible for binding the hyphal forms of various fungi such as Candida albicans, Microsporum and Trichophyton. Binding of hyphae from these organisms to macrophages activates a tyrosine kinase pathway that leads to pro-inflammatory cytokine release and phagocytic uptake of the hyphae. Dectin-2 is physically associated with the FcRγ chain and crosslinking of Dectin-2 leads to Src-family kinase dependent phosphorylation of FcRγ followed by Syk kinase recruitment in a classical immunoreceptor pathway. Interestingly, the other major fungal receptor on innate immune cells, Dectin-1 does not use an associated adapter like DAP12 or FcRγ but instead has its own intrinsic ITAM in its cytoplasmic tail with which it initiates intracellular signaling in a Src-kinase/Syk pathway (34).

Macrophage-inducible C-type lectin (Mincle) was first identified as a LPS-induced target of the NF-IL6 (C/EBPβ) transcription factor in macrophages. Mincle has a short 21 amino acid cytoplasmic domain with no apparent signaling capacity (35). Recently, Yamasaki et al. showed that Mincle associates with FcRγ in macrophages and that signaling through Mincle is dependent upon this ITAM adapter (18). Cross-linking of Mincle with a mAb induces TNF and MIP-2 (CXCL2) secretion as well as Syk and ERK1/2 phosphorylation in wild-type (WT) macrophages, but not in FcRγ-deficient macrophages. Interestingly, the authors found that Mincle binds a component of snRNPs, spliceasome-associated protein 130 (SAP130). SAP130 is typically a nuclear protein, but is released from late apoptotic or necrotic cells. Consistent with this finding, late apoptotic cells or supernatant from these cells can activate reporter cells expressing Mincle/FcRγ. In contrast, treatment of FcRγ-deficient macrophages with recombinant SAP130 results in less MIP-2 secretion compared to WT cells (18).

The in vivo function of Mincle in recognition of dead cells was addressed using a blocking antibody to this receptor and whole body irradiation of mice, which induces rapid thymocyte apoptosis. In response to irradiation, macrophages in the thymus secrete MIP-2, which causes neutrophils to accumulation within the organ. In the presence of blocking anti-Mincle mAb, both MIP-2 secretion by thymic macrophages and neutrophil accumulation is reduced. Blocking Mincle also reduces MIP-2 and TNF release and the accumulation of neutrophils in the peritoneal cavity after the injection of dead cells. These data show that Mincle is an important receptor for recognition of dead cells both in vitro and in vivo.

Mincle has also been shown to be a receptor for pathogenic and non-pathogenic fungi (15, 16, 18). Wells et al. found that recombinant Mincle binds to soluble extracts from Candida albicans and Sacchromyces cerevisiae; treatment of macrophages with a blocking anti-Mincle mAb reduces TNF secretion induced by C. albicans (15, 16). This group also generated Mincle-deficient mice and found that macrophages lacking Mincle secrete less TNF in response to C. albicans compared to WT cells, but show normal TNF production to zymosan, a S. cerevisiae cell wall extract. After systemic challenge with C. albicans, Mincle-deficient mice manifest higher titers of fungi in their kidneys than WT mice, illustrating the importance of Mincle to the immune response to C. albicans in vivo (16). Yamasaki et al. also found that Mincle is a receptor for pathogenic fungi using a functional assay to screen a panel of fungi with Mincle/FcRγ-expressing reporter cells (17). In this assay Mincle specifically recognizes Malassezia species, but not species of Candida, Aspergillus and other fungi. This binding requires calcium and the EPN motif of the carbohydrate recognition domain of Mincle, suggesting that Mincle has true lectin activity. Notably, the binding of Mincle to SAP130, described by the same group, does not depend upon calcium or the EPN motif (18). As expected, Mincle-deficient macrophages produce less MIP-2, TNF, KC and IL-10 than WT macrophages in response to M. pachydermatis. Similarly, following intraperitoneal challenge with M. pachydermatis, Mincle-deficient mice produce less IL-6 and TNF compared to WT mice. Therefore, in a theme common to other FcRγ and DAP12-associated immunoreceptors, Mincle recognizes both self (SAP130) and foreign (fungal) ligands.

Interestingly, a second C-type lectin receptor, CLEC9A (also known as DNGR-1), has recently been found to mediate activation of CD8α⁺ DCs following uptake of dead cells (36). Signaling through this receptor is required for efficient cross priming of CD8⁺ T-cells that recognize neo-antigens derived from necrotic cells. Like Mincle (and TREM-2 as well as TREMl-4, which also recognize dead cell material, see below), CLEC9A signaling is mediated by the ITAM/Syk pathway. However, in contrast to these other receptors, CLEC9A mediates this signaling through a partial ITAM sequence embedded directly in its C-terminal tail – mutation of the critical tyrosine in the “hem-ITAM” cytoplasmic domain of CLEC9A blocks it ability to signal cross presentation by DCs. The molecular identity of the CLEC9A ligand is unclear and whether this receptor may also recognize pathogen molecules is unknown. Hence, in analogy to Dectin-1 and Dectin-2, innate immune cells use different receptors that either couple to ITAM adapters or contain ITAM sequences within the receptor itself, to mediate inflammatory responses to necrotic cells.

Activation by FcRγ-associated receptors – non-immunoreceptor associations

In addition to the immunoreceptors described above, FcRγ has recently been found to be an important component of the IL-3 receptor signaling pathway (21). This discovery was made through studies of basophil function in FcRγ-deficient mice. Though basophils develop normally in FcRγ-deficient mice, they fail to produce IL-4 when stimulated with IL-3, resulting in poor differentiation of Th2 cells in co-cultures of basophils and T-lymphocytes. The involvement of FcRγ in IL-4 production seems to be selective, as other functions of IL-3, such as the promotion of basophil proliferation, is not affected in FcRγ-deficient mice. Remarkably, these investigators found that FcRγ directly associates with the βc signaling subunit of the IL-3R through transmembrane portions which are distinct from that used for association of FcRγ with FcεRI or FcαRI. Mutation of the ITAM of FcRγ ablates its ability to signal IL-4 release following IL-3 stimulation of basophils. These studies demonstrate that ITAM/Syk kinase type signaling responses may be just as important for aspects of cytokine responses as the traditional JAK/STAT pathways that we think of as being the dominant mechanism of cytokine signaling.

The FcRγ to Syk kinase pathway has also been found to be critical in DC and macrophage responses to Mycobacterium tuberculosis glycolipids (22). The mycobacterial cell wall component Trehalose-6,6-dimycolate (TDM), or its less toxic, synthetic analogue Trehalose-6,6-dibehenate (TDB) are both potent activators of cytokine release and nitric oxide (NO) production by macrophages. However, both of these analogs fail to induce NO release from FcRγ-deficient and Syk-deficient macrophages, whereas DAP12-deficient cells respond normally. The FcRγ/Syk mediated response to these glycolipids, as well as to a number of fungal agents, seems to be most important in activating the CARD9/MALT1 inflammasome pathway (leading to NF-κB activation), rather than coupling to the more traditional Ras/Erk pathway as described for most immunoreceptors (37). This novel function of FcRγ-initiated signaling highlights the growing diversity of receptors and pathways in which this ITAM adapter participates. Clearly, definition of the receptors involved in the innate immune cell response to mycobacterial lipids is of paramount importance for vaccine research.

Activation by DAP12-associated receptors – new immunoreceptor associations

There have been many advances in understanding the function of DAP12 and its associated receptors in the activation of myeloid cells. Recently, the myeloid DAP12-associating lectin (MDL-1), also known as CLEC5, has been found to function as a receptor for dengue virus; an MDL-1-Fc fusion protein can directly bind dengue virions as determined by direct ELISA and immunoprecipitation (38). Though MDL-1 is a C-type lectin receptor, binding to dengue is not dependent upon Ca⁺², but is partially inhibited by excess fucose and mannose. This is in contrast to DC-SIGN, another myeloid-expressed C-type lectin receptor, which also binds the dengue envelope but in a Ca⁺²-dependent manner that is blocked by mannans and mannose (39, 40). Treatment of human monocyte-derived macrophages with dengue virions induces phosphorylation of DAP12. ShRNA-mediated knockdown of MDL-1 expression in these cells blocks DAP12 phosphorylation and reduces inflammatory cytokine production following dengue virus challenge (38). In contrast, MDL-1 appears to have no role in the infection of macrophages by dengue, which was dependent upon DC-SIGN. Therefore, these two C-type lectin receptors for dengue virus participate in different processes—DC-SIGN in viral uptake and MDL-1 in viral-induced inflammatory mediator production. Blocking MDL-1 in a mouse model of dengue virus infection partially protectes mice from lethal dengue virus disease by reducing serum TNF and IP-10 levels. These data show that MDL-1 participates in the inflammatory response to dengue virus and may be an attractive drug target for treating dengue shock syndrome and dengue hemorrhagic fever, two life-threatening diseases associated with secondary dengue virus infection.

MDL-1 has also recently been shown to participate in osteoclast function in mice, through association with both DAP12 and DAP10 (24). DAP10 is a small signaling adapter expressed in myeloid and NK cells. It also uses a negatively charged aspartic acid in its transmembrane to associate with immunoreceptors, but the DAP10 cytoplasmic domain contains a PI-3 kinase binding site rather than an ITAM (4). DAP12-deficient mice have modest osteopetrosis (or bone overgrowth), which is likely due to reduced fusion capacity of bone marrow precursors during formation of multinucleate osteoclasts (41-43). DAP10-deficient mice also have mild osteopetrosis and reduced osteoclast fusion in culture at limiting doses of osteoclast growth factors M-CSF and RANKL (24). Co-immunoprecipitation experiments demonstrate that DAP10 and DAP12 both associate with MDL-1 in osteoclast precursors, suggesting that in these cells, the adapters may heterodimerize. Consistent with pairing of both DAP10 and DAP12 to MDL-1, surface expression of the receptor is decreased in DAP10-deficient or DAP12-deficient osteoclasts, and is absent on DAP10/DAP12 double-deficient cells. Confirming the role of MDL-1 in osteoclast development, knockdown MDL-1 in pre-osteoclasts reduces their ability to generate mature osteoclasts (24).

The idea of dual signaling through DAP10 and DAP12 co-associated receptors has also been suggested in NK cells, where both of these adapters bind to the NKG2D activating receptor. However, biochemical experiments suggest that NKG2D pairs only with one homodimer of each signaling adapter, not heterodimeric DAP10/DAP12 complexes (4). More recently, several groups have shown that DAP10 can associate with and contribute to the function of other NK cell-expressed C-type lectin receptors previously thought only to use DAP12, including Ly49H and Ly49D (44-46). The potential dual utilization of DAP10 and DAP12 by other myeloid innate immune receptors remains to be explored.

Several new studies have added to our understanding of the function of the DAP12-associated TREM-2 receptor. TREM-2 is expressed on tissue macrophages, (including osteoclasts and microglia), bone marrow-derived macrophages and DCs, where it mediates both positive and negative signals in myeloid cells (see below). TREM-2 and DAP12 have also been shown to have both positive and negative effects on osteoclastogenesis in vitro (47). Interestingly, TREM-2 associates with Plexin-A1 on DCs and Plexin A1-deficient mice show osteoclast defects similar to DAP12-deficient mice (48). A new study from Helming et al. found that TREM-2/DAP12 also regulates a different form of macrophage fusion into multinucleate giant cells (49). Mature macrophages fuse into multinucleate giant cells in vivo during a foreign body response or granulomatous inflammation, such as is seen during mycobacterial infection. This fusion occurs in vitro if mouse thioglycollate-elicited macrophages or bone marrow-derived macrophages are cultured in the presence of IL-4 or IL-4 and GM-CSF, respectively. Helming et al. found that both murine and human monocyte-derived macrophages lacking DAP12 or DAP12 signaling, either by knockout or knockdown, are defective in in vitro fusion compared to WT macrophages (49). Generation of multinucleate giant cells in vivo during a foreign body response is also reduced in mice lacking DAP12. Knockdown of TREM-2 in mouse bone marrow-derived macrophages severely reduces fusion, showing that TREM-2 is the principal DAP12-associated receptor required for in vitro fusion. The authors conclude that TREM-2 and DAP12 allow macrophages to reach a fusion-competent state as they regulate the expression of several proteins known to be involved in macrophage fusion, including DC-STAMP and E-cadherin. Though this observation is consistent with the majority of the work showing a positive role for TREM-2 and DAP12 in osteoclast fusion in mice and humans (41, 42, 50-52), it is not consistent with the fact that TREM-2-deficient bone marrow shows accelerated in vitro osteoclastogenesis or necessarily with the bone cysts seen in patients lacking these proteins (see below) (47). Analysis of macrophage and osteoclast fusion in TREM-2-deficient mice will be of future interest as will identification of endogenous TREM-2 ligands that mediate these effects.

TREM-2 paired with DAP12 has also been found to mediate phagocytosis (53-55). Takahashi et al. originally implicated TREM-2 and DAP12 in the phagocytosis of apoptotic neurons by microglia (55). shRNA-mediated knockdown of TREM-2 or DAP12 in mouse microglia blocks their ability to take up apoptotic neurons, whereas overexpression of TREM-2 induces increases phagocytosis. Hsieh et al. have expanded on this finding by showing that neurons express a ligand for TREM-2 that is increased during the induction of apoptosis, shown by staining cells with a fusion protein of TREM-2 and human IgG1 (53). In fact, several cell lines have increased staining with the TREM-2 Fc reagent after the induction of cell death (53, 56). When non-phagocytic CHO cells are transfected with a TREM-2/DAP12 chimera they can endocytose apoptotic neurons, showing TREM-2 is a bona fide phagocytic receptor for these dying cells (53). One endogenous ligand for TREM-2 has recently been identified as Hsp60, a mitochondrial chaperone which can be released from both healthy and dying cells (57). Hsp60 present on the surface of neuroblastoma N2A cells serves as a ligand for TREM-2/Fc fusion. Moreover treatment of N9 microglial cells with purified Hsp60 increases their phagocytosis of dying cells in a TREM-2-dependent manner. Interestingly, a new DAP12-associated TREM family member, TREM-like 4 (TREMl4), has also been shown to bind dying cells. TREMl4 is specifically expressed on DCs and certain macrophage populations in the spleen and is most highly expressed on CD8-expressing DCs that efficiently phagocytose apoptotic cells (56). Its ligand is unknown.

TREM-2 has also binds several species of bacteria and fungi (58). Recently, N’Diaye et al. reported that TREM-2 serves as a phagocytic receptor for these microbes (54). Similar to what was shown for apoptotic neurons, transfection of CHO cells with a TREM-2/DAP12 chimera allows these cells to phagocytose E. coli, S. aureus and F. tularensis, but not zymosan, a S. cerevisiae cell wall particle. This phagocytosis in CHO cells requires Src and Syk kinase activity as well as the activity of the Rho-family GTPases Rac1 and Cdc42. DAP12-deficient macrophages have reduced phagocytosis of all 3 bacteria tested compared to WT macrophages; expression of the TREM-2/DAP12 chimeric receptor in the DAP12-deficient cells restores their phagocytic activity to WT levels. TREM-2-deficient macrophages or WT TREM-2 knockdown cells also have reduced E. coli phagocytosis. Interestingly, Charles et al. found that TREM-2 and DAP12 are involved in regulating the macrophage response to Salmonella typhimurium though TREM-2 does not directly bind this organism (59). Collectively, these observations demonstrate that the TREM-2/DAP12 receptor complex binds both endogenous and exogenous ligands to induce phagocytosis. Hence, this dual function of TREM-2 resembles the ability of Mincle to recognize both pathogen and host-derived ligands.

PDC-TREM is another newly described TREM family member that associates with DAP12. Plasmacytoid DCs (PDCs) are specialized for type 1 IFN production in response to viral infection and among the TLRs, they express exclusively TLR7 and TLR9, which sense ssRNA and CpG-containing DNA, respectively. PDC-TREM is specifically expressed on PDCs after stimulation with TLR7 or TLR9 agonists (25). Similar to TREM-2, PDC-TREM associates with Plexin-A1 and participates in signal transduction by the Plexin-A1 ligand, Sema6D. In PDCs stimulated with low levels (0.1μM) of CpG, co-stimulation with Sema6D-Ig fusion leads to slow enhancement of IFNα production. But in PDCs that are stimulated with optimum levels of CpG (1 μM) PDC-TREM expression increases 10-fold to the point where the co-associated PlexinA1 receptor can potentially interact with endogenous Sema6D to mediate production of IFNα. The production of IFNα following optimal CpG stimulation is reduced in DAP12-deficient PDCs, in cells following knockdown of PDC-TREM by shRNA, and in cells treated with a mAb that blocks PDC-TREM/Plexin A1 association. These results suggest that the complex of PDC-TREM/DAP12/Plexin A1 is required for production of IFNα by TLR agonists in PDCs.

Activation by DAP12-associated receptors – non-immunoreceptor associations

Outside of the immunoreceptor field, there is now compelling evidence that DAP12 plays a central role in signaling through the c-Fms or M-CSF receptor (CSF-1R) in mononuclear cells (26, 27). These studies are particularly revealing because they provide some molecular understanding of the consequences of DAP12 deficiency in both mice and humans. As mentioned above, mice lacking DAP12 develop a modest form of osteopetrosis (41-43). This phenotype is much worse in mice lacking both FcRγ and DAP12 (60, 61), potentially as a result of both poor cell fusion events during osteoclastogensis and reduced adhesion of osteoclasts to bone matrix through the αvβ3 integrins leading to poor bone resorption (61). In contrast, humans lacking DAP12 develop polycystic osteodysplasia (bone cysts) leading to spontaneous fractures (62). These patients (which have a syndrome referred to as Nasu-Hakola disease or Polycystic Lipomembranous Osteodysplasia with Sclerosing Leukoencephalopathy (PLOSL)) also develop progressive demyelination and CNS gliosis associated with potential defects in microglia function. The involvement of DAP12 in signaling downstream of the CSF-1R may explain these mononuclear cell defects in DAP12-deficient individuals. Mononuclear precursors lacking DAP12 show reduced proliferative responses to M-CSF as well as poor survival at limiting doses of the cytokine (26). This results in a relatively low “fitness” of DAP12 mononuclear cells in competitive repopulation assays in irradiated recipient mice. Otero et al. found that M-CSF-induced signaling through DAP12 in macrophages leads to activation of Pyk2 and the translocation of β-catenin protein to the nucleus where it promotes cell survival and proliferation (26). In osteoclasts, DAP12 is required for coupling M-CSF stimulation to Syk activation, which is also important in osteoclast differentiation and function (26, 27). More importantly, a careful analysis of the DAP12 KO mice reveals that they too have microglial degeneration and apoptosis in the CNS, which would correlate with the neurologic findings in Nasu-Hakola patients (26). The mechanism by which DAP12 is coupled to the CSF-1R is unclear, but mutations within the transmembrane portion of DAP12 block its ability to function in the M-CSF pathway (27). However, the CSF-1R does not have a corresponding charged amino acid with which DAP12 can interact. Thus DAP12 either associates directly with the CSF-1R, through some unusual transmembrane mechanism or via an intermediary protein, such as an immunoreceptor.

Besides signaling in the M-CSF and IL-3 receptor pathways, it is also likely that DAP12 and/or FcRγ have major roles in other cytokine/growth factor pathways. Indeed, we have found that GM-CSF-induced DC differentiation from bone marrow precursors is dramatically reduced when cells from DAP12/FcRγ double mutant mice are used in comparison with WT cells (Fig. 2A). This defect is likely due to poor activation of Syk kinase in the ITAM-deficient cells, because the same growth impairment is seen in syk^-/- cells. Cells lacking FcRγ or DAP12 alone have an intermediate growth impairment (data not shown). Similar to the M-CSF pathway, when DAP12/FcRγ-deficient bone marrow cells are cultured in GM-CSF, they show increased rates of apoptosis (Fig. 2B). Whether DAP12 and/or FcRγ are directly associated with the GM-CSF receptor or become phosphorylated following GM-CSF stimulation of DCs remains to be seen, but this is likely as the IL-3 and GM-CSF receptors share the βc signaling subunit. Nevertheless, it seems quite clear that these ITAM adapters play a fundamental role in a number of cytokine signaling pathways.

Activation by both FcRγ- and DAP12-associated receptors

Perhaps one of the earliest examples of the involvement of FcRγ and DAP12 in non-immunoreceptor signaling was the description of their requirement for leukocyte β1, β2, and β3 integrin signaling (63, 64). These studies evolved out of the realization that many of the signaling molecules described as having a central role in leukocyte immunoreceptor signaling were also required for the integrin pathways. This included Src-family and Syk kinases (65, 66) as well as SLP-76 (67) and Vav guanine exchange factors (68). The direct examination of FcRγ and DAP12 deficient cells was definitive; neutrophils and macrophages from these mice show complete blocks in integrin-mediated activation (64). In neutrophils, the lack of integrin signaling in DAP12/FcRγ double mutant cells translates to poor activation of superoxide production, absent release of granule contents and poor cytoskeletal responses following engagement of surface integrins when cells are plated on adhesive surfaces. In macrophages, deficiency of DAP12 alone is sufficient to block integrin signaling responses, resulting in poor adhesive signaling, reduced cytokine production and poor migration, whereas FcRγ-deficient macrophages behave normally (Abram and Lowell, unpublished). In neutrophils, loss of integrin signaling also results in poor phagocytosis of complement opsonized bacteria through the CD18/integrin/complement receptors, but surprisingly does not seem to affect cell migration either in vitro or in vivo, mainly in response to strong infectious or inflammatory stimuli (Van Ziffle and Lowell, unpublished and (69)). As in the CSF-1R pathway, it remains unclear how the ITAM adapters are coupled to the integrins, whether via direct binding or through an associated immunoreceptor(s), or even through lipid raft co-localization. However, it is clear that in both the CSF-1R pathway and the integrin pathway, the transmembrane arginine residue of DAP12 is required for its signaling capacity, suggesting that an associated immunoreceptor mediates the interaction with integrins.

DAP12 and FcRγ have also been found to signal in the PSGL-1 adhesion receptor pathway (70, 71). PSGL-1 is primary adhesion receptor on neutrophils that mediates attachment of cells to vascular P- and E-selectins. Engagement of PSGL-1 on neutrophils induces a tyrosine kinase signaling response, which in cooperation with leukocyte integrins (mainly αLβ2 or LFA-1) initiates slow rolling of cells over inflamed endothelium (72). This signaling response depends on Syk kinase, and in particular the SH2/ITAM binding function of Syk. These observations lead us to ask whether DAP12 and/or FcRγ were involved. Not surprisingly, neutrophils from the double deficient mice show an impaired ability to mediate slow rolling over surfaces coated with E-selectin and ICAM-1 (70). Though this signaling response may reflect the role of these ITAM adapters in the integrin pathway because ICAM-1 is required for slow rolling, the observation that DAP12/FcRγ double deficient neutrophils manifest reduced activation of Syk and p38 MAPK when plated on E-selectin alone suggests that PSGL-1 signaling is the major contributor. In this pathway, it appears that the Src-family kinase Fgr is the primary enzyme that phosphorylates DAP12 and/or FcRγ to initiate the PSGL-1 response.

In the theme of regulation of cytokine responses in innate immune cells by DAP12 and FcRγ, there is clear evidence that these adapters, along with Syk, play a signaling role in IFNγ priming of macrophages (29). Pre-incubation of macrophages with IFNγ dramatically increases their responses to subsequent IFNα stimulation as measured changes in inflammatory cytokine production (TNFγ or IL-1RA) and phosphorylation of STAT1. In part, this is due to increased expression of STAT-1 in IFNγ stimulated cells. However, IFNγ priming of DAP12/FcRγ double mutant macrophages fails to induce dramatic phosphorylation responses to subsequent IFNα stimulation, even though STAT1 expression increases as expected. The downstream pathway from the ITAMs that leads to increased IFNγ priming involves the calcium dependent kinases CaMK and Pyk2 since inhibitors of these enzymes block the IFNγ effect (30). The fact that the enhanced priming effect of IFNγ is dependent on cell adhesion (macrophages kept loosely in suspension have reduced IFNγ priming) suggests that that integrin-mediated signaling through DAP12/FcRγ is the primary pathway that leads to enhanced priming. In this regard, tonic signaling by cell adhesion, mediated through the integrin/ITAM/Syk pathway may serve as a general amplifier of a number of signaling responses besides just the IFNγ response (reviewed in (13)).

ITAM signaling through DAP12/FcRγ is also involved in maximizing antigen presentation responses in DCs (31). DCs derived from DAP12/FcRγ double mutant mice manifest reduced cross-presentation of particulate, but not soluble, antigens to cognate T-cells. This defect does not seem to be mediated by poor uptake of the particulate antigens, but instead reflects some other signaling pathway involving the Vav GEF proteins since DCs lacking these enzymes also have defects in particulate antigen presentation. Indeed, tyrosine phosphorylation of Vav proteins is reduced in DAP12/FcRγ deficient DCs following stimulation with particulate antigens. The relevant receptors that mediate this antigen presentation defect are unclear.

Inhibitory signaling by ITAM adapters

In addition to contributing to activation of myeloid cells, the DAP12 and FcRγ ITAMs also mediate signals that inhibit cell function. There were some descriptions of this type of inhibitory activity by FcRγ-paired Fc receptors in the 1990’s (73-75), however in the last 5 years there have been many more receptors implicated in inhibition, including those that associate with DAP12 and/or FcRγ. Generally, this inhibitory activity is seen when the cells are simultaneously activated through a DAP12 or FcRγ-associated receptor and a distinct receptor system, such as Toll-like receptors (TLRs) or Fc receptors, leading to a reduced response when compared to cells activated through the heterologous receptor alone. The inhibition requires association of DAP12 and/or FcRγ with a particular immunoreceptor(s) and is not thought to be mediated through direct association of the ITAM adapters with the heterologous receptor(s). Inhibition of activation by ITAM-containing adapters has also been shown in B and T cells, through Igα and the CD3 molecules respectively (reviewed in (12)), showing that ITAM-based inhibitory signals are widespread in the immune system. Several different mechanisms for ITAM-based inhibition in myeloid and lymphoid cells have been proposed, and it is as yet unclear if there will be a single mechanism for inhibition, or whether it will vary by cell type, ITAM adapter, and/or inhibited signaling receptor.

Inhibition by FcRγ-associated receptors

One of the earliest descriptions of the inhibition of myeloid cell responses by a receptor associated with an ITAM-containing adaptor was that of the IgE receptor, FcεRI (75). IgE binds to the α chain of the receptor, which pairs with 2 ITAM-containing chains, FcεRIβ and FcRγ. Torigoe et al. showed that cross-linking of FcεRI on the RBL mast cell line simultaneously with a low affinity and a high affinity ligand for the IgE pre-bound to FcεRI resulted in the reduction in mast cell degranulation when compared with cells treated with either the low or the high affinity ligand alone. Cells treated with both the low and high affinity Ags manifest a reduction in the phosphorylation of downstream signaling molecules Syk, Pyk2 and Erk as well as a reduction in the phosphorylation of the receptor ITAMs. In this system, there is no heterologous receptor, the model is that the FcεRI ligated by the low affinity hapten inhibits the signal from the FcεRI ligated by the high affinity hapten.

Around the same time Sutterwala et al. (73) showed that ligating IgG receptors on mouse macrophages with IgG-coated sheep red blood cells (SRBC) inhibits LPS-induced IL-12 p40 production. This inhibition depends upon FcRγ, but not FcγRIII (CD16), suggesting that either FcγRI (CD64) or FcγRIV is mediating inhibition (74). Treatment with IgG-coated SRBC also increases LPS-induced IL-10 production, but IL-10 does not mediate the inhibition of IL-12 p40 production (76). This was the first description of the inhibition of TLR responses by ITAM-containing proteins.

Another Fc receptor that pairs with FcRγ to inhibit myeloid cell function is human FcαRI (CD89), which has no mouse equivalent (77). FcαRI is one of many FcRs for IgA, but is the only one found on leukocytes -- it binds specifically to the IgA subtype found in serum. Over the years, there had been many descriptions in the literature showing that monomeric IgA can inhibit macrophage functions including phagocytosis, chemotaxis, and respiratory burst (reviewed in (78)). In addition, many patients with IgA deficiency have autoimmunity or atopic diseases, supporting the idea that IgA serves to limit the immune response (79). Similar to what had been shown previously for IgA itself, Pasquier et al. found that a Fab fragment of an anti-FcαRI mAb can inhibit IgG-mediated phagocytosis, implicating FcαRI as the receptor mediating IgA-induced inhibition (77). In FcαRI-transfected RBL mast cells, both monomeric IgA and the anti-FcαRI Fab fragment also inhibit IgE-induced degranulation responses. However, this only works when the IgA or the Fab are monomeric; if the IgA or anti-FcαRI Fab fragments are cross-linked with secondary antibody, the inhibitory signal switches to enhanced FcεRI-mediated responses. In this system, the FcαRI-mediated inhibition requires association with FcRγ as well both the tyrosines in the FcRγ ITAM.

By examining the signaling responses downstream of monomeric versus oligomeric engaged FcαRI, Pasquier et al. provided a model for how this ITAM could limit signaling from the FcεRI receptor. Cells treated with monomeric IgA (or anti-FcαR Fab) have less phosphorylation of the FcRγ ITAM, Syk, LAT and ERK1/2 when compared to cells exposed to oligomeric ligands (anti-FcαRI F(ab)’₂ plus rat-anti mouse secondary) in conjunction with FcεRI cross-linking (77). Less Syk is recruited to FcRγ under inhibitory (monomeric) conditions in comparison with activating (oligomeric) conditions, while recruitment of the tyrosine phosphatase SHP-1 is seen only when inhibitory signals are transduced with the anti-FcαRI Fab. These data suggest a model where low avidity cross-linking of FcαRI with monomeric ligand results in weak phosphorylation of the FcRγ ITAM, leading to preferential recruitment of SHP-1, which in turn causes reduced phosphorylation of signaling components downstream of the heterologous activating signal (in this case FcεRI). However, following engagement by high avidity oligomeric ligands, the FcRγ ITAM is fully phosphorylated leading to Syk kinase recruitment, maximal signaling and full cellular activation through the co-engaged heterologous FcεRI (Fig 1) (77, 80).

More recently, the Monteiro laboratory has extended the number of heterologous receptors that can be inhibited by monovalent ligands for FcαRI to include signaling induced by the chemokine MCP-1 (CCL2), the cytokine TNF, and the TLR4 agonist LPS, all in human monocytes (81). Treatment with anti-FcαRI Fab inhibits monocyte chemotaxis to MCP-1 and TNF, as well as downstream signal transduction, in an SHP-1-dependent fashion, consistent with previous findings. To extend these findings to an in vivo model, these investigators generated transgenic mice expressing human FcαRI; treatment of these animals with the anti-FcαRI Fab ameliorates kidney disease in a passive transfer model of anti-glomerular basement membrane IgG-mediated glomerulonephritis. Taken together these studies clearly show that signaling through FcαRI/FcRγ with monovalent ligands can inhibit a broad array of macrophage functions (77, 81).

The Monteiro group has also investigated inhibitory signals through FcRγ-associated FcγRIII (CD16) in mouse macrophages (82). After caecal ligation and puncture (CLP)-induced sepsis, FcRγ-deficient mice and FcγRIII-deficient mice have markedly prolonged survival compared to WT mice -- all WT mice succumb to shock within 70 hours, whereas only 50% of WT mice succumb by 180 hours. This protection is associated with less recovered E. coli from the peritoneal cavity, as well as less serum and peritoneal TNF in FcRγ-deficient mice compared to WT mice. In vitro, phagocytosis of E. coli (but not zymosan or S. aureus) is increased in peritoneal neutrophils and macrophages lacking FcRγ compared to WT cells. FcRγ-deficient macrophages are at least as efficient, if not more efficient, as WT macrophages in killing phagocytosed E. coli. The authors conclude that the reason for increased survival of FcRγ-deficient mice during CLP is that the peritoneal neutrophils and macrophages clear the bacteria more efficiently and therefore there is reduced systemic TNF and other inflammatory cytokines that mediate shock. The authors suggest a model whereby in WT macrophages recognition of E. coli by FcγRIII leads to weak FcRγ phosphorylation, recruitment of SHP-1 to the receptor, decreased PI-3 kinase signaling and subsequent impaired function of the scavenger receptor MARCO, all of which cause prolonged exposure of macrophages to E. Coli LPS and increased TLR4 responses, hence more TNF leading to shock (82). In FcRγ-deficient cells, this inhibitory function is reversed and E. coli phagocytosis through MARCO proceeds robustly.

There have also been several studies showing that FcRγ-associated receptors inhibit TLR responses in human plasmacytoid dentritic cells (PDCs) (83-86). Blood dendritic cell antigen (BDCA)-2 was originally described as a C-type lectin-like receptor specifically expressed on human PDCs that when cross-linked with antibody could inhibit IFNα induced by Influenza virus or DNA-containing immune complexes (86). Though BDCA-2 has a small cytoplasmic region of 21 amino acids with no known signaling motifs, cross-linking of BDCA-2 induces protein tyrosine phosphorylation and intracellular calcium mobilization. Cao et al. found that BDCA-2 associates with FcRγ to transduce signals (85). In this study, the authors show that PDCs express many signaling molecules in common with B cells to transduce ITAM signals, including Syk, Lyn, BLK, PLCγ2 and BCAP. Cross-linking BDCA-2 in PDCs, which potently suppresses TLR7- and TLR9-induced IFNα, TNF and IL-6 secretion, also induces the phosphorylation of Src family kinases, Syk, Vav1, PLCγ2 and ERK. Remarkably, mAb cross-linking of BDCA-2 up to 4 hours after CpG DNA treatment is able to inhibit PDC cytokine secretion (85), though the BDCA-2 ligand that mediates this effect is unknown.

Cao, Liu and colleagues also have studied Immunoglobulin-like transcript (ILT) 7, which similar to BDCA-2, is specifically expressed on human PDCs and associates with FcRγ (83). Like BDCA-2, mAb mediated cross-linking of ILT7 potently inhibits TLR7 and TLR9-mediated IFNα and TNF production by PDCs. More recently, this group has identified a ligand for ILT7 that can mediate this inhibitory effect (84). They found that several breast cancer and melanoma cell lines express a cell surface ligand for ILT7 that can activate a reporter cell line expressing ILT7/FcRγ. By generating blocking mAbs against these cancer cell lines, they identified the ILT7 ligand as bone marrow stromal cell antigen (BST) 2. Recombinant BST2 binds to ILT7 with an affinity of approximately 1 μM, causing inhibition of PDC IFNα secretion induced by CpG DNA or influenza virus. Interestingly, BST2 expression is induced by IFNα treatment in many cell lines and is even expressed on PDCs themselves, suggesting the possibility for a cis interaction between ligand and receptor (84, 87). BST2 is also expressed on mouse PDCs, though there is no direct ortholog of ILT7 in mouse. Though not directly shown by these authors, it is possible that ILT7 or BDCA-2 is inhibiting the function of PDC-TREM following treatment of cells with TLR7 or TL9 agonists.

FcRγ has also been implicated in the inhibition of conventional DC function induced by cross-linking of MHC class II molecules (88). Though not heavily studied, it is now well recognized that engagement of Ag loaded MHC II by T-cells can lead to signaling responses in the antigen presenting B cells, monocytes and DCs. MHC II does not have any recognized signaling motifs in its cytoplasmic tail, however in B cells MHC II associates with the BCR ITAM adapters Igα and Igβ (CD79a and CD79b) to transmit activating signals to the B-cell (89). Engagement of MHC II on DCs by the lymphocyte activation gene (LAG)-3, a CD4-related protein expressed on regulatory T cells, has recently been found to inhibit up-regulation of CD86 expression (88). mAb cross-linking of MHC II mimics the effect of LAG-3 binding, causing reduction of LPS-induced CD86 expression and IL-12 production, as well as diminishing the ability of peptide-pulsed DCs to activate antigen-specific T cells. MHC II cross-linking of FcRγ-deficient DCs fails to down modulate CD86 expression, suggesting that this inhibitory signal is mediated by ITAM signaling. Similar to inhibition by FcαRI monovalent cross-linking, down-regulation of CD86 by MHC II requires ERK activity and SHP-1 (88). Low levels of SHP-1 are constitutively associated with FcRγ and MHC II in DCs, which may mediate the inhibitory effect of mAb engaged MHC II. Overall, these observations suggest a mechanism by which regulatory T-cells, through expression of LAG-3, down modulate DC function. While the observation that recruitment of SHP-1 by MHC II/FcRγ complexes is consistent with the model proposed by Pasquier et al. in their studies of FcαRI, there are several surprising points to this modulatory effect of MHC II (mainly the constitutive rather than induced association of SHP-1 with FcRγ) that require further investigation. Nevertheless, these studies provide an important example of a non-immunoreceptor (MHC II) that mediates inhibitory signaling through FcRγ.

Inhibition by DAP12 associated receptors

As with FcRγ, DAP12 and its associated receptors are capable of inhibiting myeloid cell function. We originally described DAP12’s inhibitory properties by examination of DAP12-deficient macrophages (90). Both bone marrow and peritoneal macrophages lacking DAP12 display increased inflammatory cytokine production (TNF, IL-6, IL-12 p40) compared to WT cells following treatment with a panel of TLR agonists or with zymosan, which signals through TLR2 and Dectin-1. As a result of loss of this inhibitory pathway, DAP12-deficient mice develop higher serum levels of TNF than WT mice after LPS injection, and as a result are more susceptible to endotoxic shock. The increased response to TLR agonists is, however, protective following bacterial infection, since DAP12-deficient mice have improved clearance of Listeria monocytogenes compared to WT mice. DAP12-deficient mice also have increased responses to infection with Mycobacterium tuberculosis (L. Lanier, personal communication, (91)).

Loss of DAP12 not only affects macrophage TLR responses, but also results in increased signaling through other macrophage receptor systems. Cross-linking of FcγRIII with plate-bound mAb results in the secretion of more TNF and IL-6 from DAP12-deficient macrophages than from WT macrophages (90). Similarly, DAP12-deficient macrophages show enhanced phagocytosis of IgG-opsonized SRBC compared to WT macrophages over a range of particle doses (Fig. 3A). DAP12-deficient macrophages also phagocytose zymosan through the Dectin-1 receptor more efficiently than WT macrophages (Fig. 3B). Besides inhibiting heterologous immunoreceptor (FcγRIII) or C-type lectin receptor (Dectin-1) function, we have found that, we have found that DAP12 can also inhibit signaling through the IFNγR in macrophages. DAP12-deficient macrophages show significantly higher upregulation of MHC II following low dose IFNγ treatment than do WT macrophages (Fig. 3C). Therefore, similar to inhibition via FcαRI/FcRγ, DAP12 can inhibit a variety of heterologous signaling pathways in macrophages.

The exact mechanism by which DAP12 inhibits TLR and other signaling pathways in macrophages is still being elucidated. The ITAM residues of DAP12 are required for this inhibitory function, since reintroduction of an ITAM tyrosine mutant of DAP12 into DAP12-deficient macrophages fails to restore normal responses, while re-expression of WT DAP12 functions normally (90). Macrophages deficient in Syk show similarly increased TLR-induced inflammatory cytokine production as DAP12-deficient macrophages, again supporting the role of the DAP12 ITAM in the inhibition of TLR responses. To begin to understand how DAP12 inhibits TLR signaling, we compared signal transduction pathways that are activated downstream of TLRs in WT and DAP12-deficient macrophages. The phosphorylation of ERK1/2 is consistently higher in DAP12-deficient macrophages than in WT cells. In contrast, the phosphorylation of p38 MAPK and JNK and the degradation of IκBα, a measure of activation of the NF-κB pathway, are similar between WT and DAP12-deficient macrophages (90). More recently, we have found that LPS-induced AKT phosphorylation is also increased in macrophages that lack DAP12, suggesting an increase in PI3-kinase activity (Fig. 4). These data show that DAP12 regulates multiple signaling pathways important in TLR responses in macrophages. It will be interesting to examine if ERK and AKT phosphorylation are also increased downstream of FcγRIII cross-linking in DAP12-deficient macrophages and to fully delineate the inhibitory signaling pathway.

Of the many DAP12-associated receptors expressed in macrophages, TREM-2 is the single receptor that has been shown to be required for inhibition of TLR and FcR-signaling in macrophages (92, 93). Expression of a chimeric receptor composed of the extracellular domain of TREM-2 fused to DAP12, in which the transmembrane aspartic acid was mutated to alanine, in DAP12-deficient macrophages restores normal signaling responses to CpG DNA, zymosan, or mAb cross-linking of FcγRIII (92). In contrast, chimeric receptors containing the extracellular region of TREM-1, MAIR-II and PILRβ have no effect. Additionally, shRNA-mediated knockdown of TREM-2 increases the CpG DNA- and zymosan-induced TNF production of WT macrophages. Most importantly, macrophages from TREM-2-deficient mice show the same exaggerated TLR-induced inflammatory cytokine production as those from DAP12-deficient mice (93). The nature of the ligand for TREM-2 that mediates inhibitory signaling is not known, though the finding that a TREM-2/Fc chimeric protein binds directly to macrophages would suggest than these cells posses an endogenous ligand (92). Both microbial and endogenous ligands for TREM-2 have been identified (53, 54, 56-58, 94), but whether these mediate the inhibitory signal has not been tested. Together, these studies clearly implicate TREM-2 as the major inhibitory DAP12-associated receptor on mouse macrophages. Further questions of interest include whether the TREM-2/DAP12 receptor complex is also inhibitory in human macrophages and the nature of the TREM-2 ligand that mediates inhibition. Interestingly, TREM-2 also inhibits in vitro osteoclast formation in mice -- TREM-2-deficient bone marrow precursors display accelerated osteoclastogenesis in comparison with WT bone marrow cells (47), though this is in direct contrast to what is seen in human in vitro osteoclastogenesis assays with TREM-2-deficient monocytes precursors (50, 52) or with knockdown of TREM-2 in RAW cells (51).

TREM-2 has also been shown to have a negative regulatory effect on microglia, the resident tissue macrophage of the brain. Takahashi et al. found that not only do TREM-2 and DAP12 function positively to regulate the phagocytosis of apoptotic neurons, as discussed above, but that they also negatively regulate transcription of the genes encoding the inflammatory mediators TNF, IL-1β and iNOS (55). Recently, two papers have highlighted the inhibitory role of TREM-2 in vivo in the CNS disease experimental autoimmune encephalomyelitis (EAE), a mouse model for human multiple sclerosis (94, 95). Treatment of mice with a blocking antibody to TREM-2 results in exacerbation of EAE as measured by clinical score (95). In contrast, treatment of mice with GM-CSF derived myeloid cells overexpressing TREM-2 (by lentivirus gene transduction) show an improvement in clinical score as compared with those that received control cells (94). Taken together, these studies demonstrate that TREM-2 is a potent inhibitor of autoimmune inflammatory disease in the EAE model. Interestingly, high levels of a soluble form of the TREM-2 receptor is found in the cerebrospinal fluid of multiple sclerosis patients while patients with non-inflammatory neurological diseases have lower levels of the receptor (96). This finding may suggest that production of soluble TREM-2 interferes with the inhibition of inflammation by membrane associated TREM-2 in multiple sclerosis patients, possibly by blocking access to TREM-2 ligands.

DAP12-associated receptors on both mouse and human PDCs have also been found to inhibit TLR responses, similar to the effects of ILT7/FcRγ and BDCA-2/FcRγ (87, 97). The DAP12-associated receptor, NKp44, which functions as an activating receptor for NK cell mediated cytotoxicity paradoxically functions to inhibit production of cytokines by human tonsillar PDCs (97). Similar to what had been reported for BDCA-2 cross-linking, cross-linking of NKp44 with plate-bound mAb inhibits the CpG DNA-induced secretion of IFNα, MIP-1α (CCL3) and MIP-1β (CCL4). Similar to TREM-2, NKp44 has been described to bind a variety of microbial and endogenously-derived ligands (98-104). The TREM family was originally identified through homology with NKp44 and the genes encoding these proteins in humans are closely linked on chromosome 6 (105).

In mouse PDCs, Siglec-H is a DAP12-associated receptor that can inhibit CpG DNA-induced IFNα secretion (106, 107). Siglecs are sialic acid-binding receptors belonging to the Ig superfamily that predominantly have cytoplasmic ITIMs. In contrast, Siglec-H has a transmembrane lysine and small cytoplasmic region consistent with DAP12 association; indeed, DAP12-deficient PDCs, but not those lacking FcRγ or DAP10, fail to express Siglec-H. Supporting a role for Siglec-H in modulating splenic PDCs in mouse, cells from DAP12-deficient mice produce significantly more IFNα after CpG DNA treatment than WT cells, while in vivo administration of CpG DNA results in higher serum amounts of IFNα in DAP12-deficient mice than WT mice (107). These data suggest that Siglec-H on PDCs, similar to TREM-2 on macrophages, recognizes an endogenously-expressed ligand, possibly on PDCs themselves and disruption of this signal by removal of DAP12 results in increased TLR responses.

Inhibition by both FcRγ- and DAP12-associated receptors

During our studies of inhibition of macrophage responses by TREM-2 and DAP12, we were surprised to find that, unlike DAP12-deficient macrophages, DAP12-deficient bone marrow-derived DCs show very modestly increased TLR responses compared to WT cells, despite the fact that bone marrow-derived DCs express abundant amounts of both DAP12 and TREM-2 (90). We hypothesized that FcRγ may also play a role in inhibition of TLR responses in DCs; indeed, DCs lacking both DAP12 and FcRγ have significantly increased TLR responses in comparison with WT DCs as measured by production of TNF, IL-6, and IL-12 and up-regulation of MHC II, CD80 and CD86 (108). Singly DAP12-deficient or FcRγ-deficient DCs have increased responses that are typically intermediate between the DAP12/FcRγ-deficient DCs. Consistent with the role of ITAM signaling in inhibiting TLR responses in DCs, Syk-deficient DCs also have increased inflammatory cytokine production and maturation compared to WT DCs (108). Furthermore, DAP12/FcRγ-deficient and Syk-deficient DCs both induce more antigen-specific T cell proliferation than WT DCs. These data show that similar to macrophages and PDCs, conventional DC populations have constitutive signaling through ITAM-bearing adapters that inhibits TLR responses. We are currently investigating the DAP12- and FcRγ-associated receptors that participate in these inhibitory signals in DCs.

Summary

The extensive involvement of the ITAM adapters DAP12 and FcRγ in multiple signaling pathways in myeloid cells is now well established. This signaling involves a growing array of immunoreceptors and, surprisingly, many non-immunoreceptors as well, such as integrins and cytokine receptors. In the case of signaling through immunoreceptors, the ITAMs appear to be absolutely critical; deficiency of DAP12 and/or FcRγ ablates signaling. The same is true for integrins, but in the case of cytokine responses, the ITAM signaling seems to modulate and/or be additive to the traditional signaling pathways used by these receptors. In this regard, the molecular details of how either FcRγ or DAP12 couple to non-immunoreceptor proteins remains obscure. Whether these adapters are directly associated with integrin and/or cytokine receptor complexes, or modulate signaling through co-association with other immunoreceptors, needs to be defined. More remarkable is the clear demonstration of the contribution of DAP12 and FcRγ to inhibitory signaling in innate cells. In the case of downmodulation of TLR responses, although the upstream receptor to which DAP12 couples in macrophages is known -- TREM2 -- how the TREM2/DAP12 association mediates an inhibitory signaling is unclear. The current hypothesis in the FcR pathway involves ligand affinity (high affinity ligands mediate activation signaling while low affinity ligands are inhibitory), but that remains to be demonstrated in other inhibitory responses through DAP12 and FcRγ. Overall, it is clear that cross-talk between different receptor signaling pathways can function to allow immune cells to rapidly amplify responses to low levels of antigen, pathogen molecules or cytokines, while also potentially dampening responses to mediate immune tolerance or limit immunopathology. Deciphering the disease related contexts for the dual signal regulation by DAP12 and FcRγ will be critical in design and application of therapeutics that target ITAM signaling.