The Unique Cytoplasmic Domain of Human FcγRIIIA Regulates Receptor-Mediated Function

Ligand specificity characterizes receptors for Abs and many other immune receptors, but the common use of the FcR γ-chain as their signaling subunit challenges the concept that these receptors are functionally distinct. We hypothesized that elements for specificity might be determined by the unique cytoplasmic domain (CY) sequences of the ligand-binding α-chains of γ-chain–associated receptors. Among Fcγ receptors, a protein kinase C (PKC) phosphorylation consensus motif [RSSTR], identified within the FcγRIIIa (CD16A) CY by in silico analysis, is specifically phosphorylated by PKCs, unlike other FcRs. Phosphorylated CD16A mediates a more robust calcium flux, tyrosine phosphorylation of Syk, and proinflammatory cytokine production, whereas nonphosphorylatable CD16A is more effective at activation of the Gab2/PI3K pathway, leading to enhanced degranulation. S100A4, a specific protein-binding partner for CD16A-CY newly identified by yeast two-hybrid analysis, inhibits phosphorylation of CD16A-CY by PKC in vitro, and reduction of S100A4 levels in vivo enhances receptor phosphorylation upon cross-linking. Taken together, PKC-mediated phosphorylation of CD16A modulates distinct signaling pathways engaged by the receptor. Calcium-activated binding of S100A4 to CD16A, promoted by the initial calcium flux, attenuates the phosphorylation of CY, and, acting as a molecular switch, may both serve as a negative feedback on cytokine production pathways during sustained receptor engagement and favor a shift to degranulation, consistent with the importance of granule release following conjugate formation between CD16A+ effector cells and target cells. This switch mechanism points to new therapeutic targets and provides a framework for understanding novel receptor polymorphisms.

Ligand specificity characterizes receptors for Abs and many other immune receptors, but the common use of the FcR g-chain as their signaling subunit challenges the concept that these receptors are functionally distinct. We hypothesized that elements for specificity might be determined by the unique cytoplasmic domain (CY) sequences of the ligand-binding a-chains of g-chainassociated receptors. Among Fcg receptors, a protein kinase C (PKC) phosphorylation consensus motif [RSSTR], identified within the FcgRIIIa (CD16A) CY by in silico analysis, is specifically phosphorylated by PKCs, unlike other FcRs. Phosphorylated CD16A mediates a more robust calcium flux, tyrosine phosphorylation of Syk, and proinflammatory cytokine production, whereas nonphosphorylatable CD16A is more effective at activation of the Gab2/PI3K pathway, leading to enhanced degranulation. S100A4, a specific protein-binding partner for CD16A-CY newly identified by yeast two-hybrid analysis, inhibits phosphorylation of CD16A-CY by PKC in vitro, and reduction of S100A4 levels in vivo enhances receptor phosphorylation upon cross-linking. Taken together, PKC-mediated phosphorylation of CD16A modulates distinct signaling pathways engaged by the receptor. Calciumactivated binding of S100A4 to CD16A, promoted by the initial calcium flux, attenuates the phosphorylation of CY, and, acting as a molecular switch, may both serve as a negative feedback on cytokine production pathways during sustained receptor engagement and favor a shift to degranulation, consistent with the importance of granule release following conjugate formation between CD16A + effector cells and target cells. This switch mechanism points to new therapeutic targets and provides a framework for understanding novel receptor polymorphisms. The Journal of Immunology, 2012, 189: 4284-4294. R eceptors for Abs play a crucial role in immune complex clearance and initiation of multiple cell programs including cytokine synthesis (1) and Ab-dependent cellmediated cytotoxicity (ADCC) (2). Playing a prominent role in immune complex-mediated inflammatory and anaphylactic responses (3), these receptors engage different subclasses of Ab preferentially (4), and structural polymorphisms affecting receptor function in humans (5) are implicated in susceptibility to some autoimmune diseases, including systemic lupus erythematosus (6).
Multiple FcRs associate with a homo-or heterodimer of FcεRI g-chain or z-chain both for expression and for signal transduction (7,8), which engages tyrosine phosphorylation events mediated through an ITAM (9,10). However, studies using mouse models suggest that these FcRs are not simply functionally redundant (3,11,12), and we therefore hypothesized that the unique FcgR cy-toplasmic sequences of the ligand-binding a-chains play a role in determining receptor specificity. Indeed, within a myeloid environment, total truncation mutants of the cytoplasmic domain of FcgRI suggest that the CY affects g-chain tyrosine-based signaling, receptor-induced specific gene expression, and phagocytosis (13,14). Similarly, deletion of the CD16A cytoplasmic domain results in altered receptor/z-chain complex-mediated signaling in T cell lines (15).
In silico analysis indicates that, unlike the highly homologous extracellular domains, each a-chain cytoplasmic domain sequence is without homologs in the human genome, and phosphorylation of a protein kinase C (PKC) motif, unique to CD16A among FcgR within the cytoplasmic domain, plays a critical role in modulating cellular responses following receptor-specific activation. PKC specifically mediates phosphorylation of CD16A-CY in vitro. CD16A phosphorylation is induced in cells upon receptor crosslinking, concurrent with enhanced calcium flux, tyrosine phosphorylation of Syk, and proinflammatory cytokine production. However, nonphosphorylatable CD16A mutants are more effective at activation of the Gab2/PI3K pathway, resulting in enhanced degranulation.
Recognizing that CD16A-CY domain binding partners might modulate receptor-specific functions, we identified S100A4 as specifically interacting with CD16A-CY. The association is direct, specific, and Ca 2+ -dependent. S100A4, a member of the S100 family of calcium-binding proteins (16), widely distributed in normal cells and tumor cells (reviewed in Ref. 17), interacts with a number of intracellular protein partners (18)(19)(20). S100A4 exerts its biological effects by inhibition of the phosphorylation of its target proteins by PKC and/or casein kinase II (20)(21)(22)(23), and we show that the presence of S100A4 inhibits phosphorylation of CD16A-CY by PKC in vitro. Reduction of S100A4 levels in vivo enhances CD16A phosphorylation upon cross-linking. Taken to-gether, we propose a molecular switch that regulates the dual pathways of effector functions for CD16A. Although initially activating cytokine production, CD16A undergoes a shift to more efficient granule release, favored by S100A4-mediated inhibition of receptor phosphorylation, as might occur during conjugate formation between CD16A + effector cells and target cells. Distinct from other FcRs, this switch mechanism highlights new therapeutic targets and provides a framework for understanding novel receptor polymorphisms.

Construction of expression plasmids
The S100A4/pcDNA3 expression construct was created by cloning S100A4 cDNA from pGEX2T/S100A4 into pcDNA3 expression vector. The triple alanine mutant GST-fusion protein CD16A-CY [RAAAR] was generated using plasmid purification kits from Qiagen (Valencia, CA). To generate GST-fusion proteins, the DNA fragment encoding human CD16A-CY was subcloned into pGEX-2T (Amersham Biosciences). DNA mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene). GST-CD89-CY, GST-CD32a-CY, and GST-CD64-CY were constructed in our laboratory. The GST-S100A4 fusion was generated by cloning human S100A4 cDNA into the EcoRI site of pGEX2T.

Cell lines and mononuclear cell preparation
The mouse macrophage cell line P388D1 and rat mast cell line RBL-2H3 (American Type Culture Collection) were stably transfected by electroporation and cultured in selective medium. Comparable quantitative receptor expression for individual constructs was established by FACS. The HEK cell line 293 was transiently transfected using Lipofectamine 2000 (Invitrogen) and harvested for analysis 48 h after transfection. Mononuclear cells from peripheral blood of healthy donors were obtained by Ficoll-Hypaque density gradients after informed consent.

Cell activation and immunoblotting
Cells were resuspended in ice-cold HBSS ++ (HBSS buffer [pH 7.3] with 20 mM HEPES, 0.1% BSA, 1 mM CaCl 2 , and 1mM MgCl 2 ) at 40 3 10 6 cells/ml. The cells were incubated with 10 mg/ml F(ab9) 2 3G8 or F(ab9) 2 mouse IgG on ice for 15 min. The cells were washed and incubated with F(ab9) 2 goat or rabbit anti-mouse at 20 mg/ml in HBSS ++ for various time points at room temperature. Cells were then lysed with ice-cold Triton X-100 lysis buffer (24) on ice for 10 min. Cell lysates or immunoprecipitates were analyzed by immunoblotting.

Cytokine analysis
Cytokine production was measured as described previously (13) with modification. Cells were stimulated in 24-well tissue culture plates (Costar) with 1 mg/ml LPS, surface-absorbed mAb 3G8 F(ab9) 2 , mouse IgE, or F(ab9) 2 mouse IgG . Wells were coated with absorbed F(ab9) 2 Ab, mouse IgG (40 mg/ml), or mouse IgE (40 mg/ml) overnight at room temperature. After washing, cells (5 3 10 5 cells/ml) were added to the wells and cultured for varying periods of time. Levels of murine or rat cytokines in culture supernatants were quantitated by ELISAs following manufacturers' protocols.

Degranulation assay
Degranulation of RBL-2H3 cells was determined by measuring the release of the granule enzyme b-hexosaminidase (25). Briefly, cells were plated on 24-well plates at 1 3 10 5 cells/well overnight. Monolayer cells were washed with Tyrode's buffer (130 mM NaCl, 5 mM KCl, 1.4 mM CaCl 2 , 1 mM MgCl 2 , 5.6 mM glucose, 10 mM HEPES, and 0.1% BSA [pH 7.4]). The cells were incubated with 100 ml 5 mg/ml F(ab9) 2 3G8, F(ab9) 2 mouse IgG, or mouse IgE on ice for 45 min. After washing with Tyrode's buffer, cells were incubated with 100 ml F(ab9) 2 goat anti-mouse at 20 mg/ml at 37˚C for various time points. Cell supernatants (10 ml) were incubated with 30 ml substrate (3 mM p-nitrophenyl-N-acetyl-b-D-glucosaminide in 0.1 M citrate [pH 4.5]) at 37˚C for 1 h. The reaction was stopped with 150 ml 100 mM NaCO 3 -NaHCO 3 (pH 10.0) and absorbance at 405 nm was measured. Results are expressed as percentages of the total b-hexosaminidase contents of the cells after subtracting the spontaneous release in the absence of receptor cross-linking.

Measurement of intracellular calcium
Intracellular Ca 2+ concentration ([Ca 2+ ] i ) was determined as described previously (26). Briefly, suspensions of cells at 10 7 /ml in Ca 2+ -and Mg 2+ -free Hanks' PBS [pH 7.4] were incubated with 5 mM Indo-1/AM at 37˚C for 40 min. Following incubation, cell suspensions were incubated another 25 min at room temperature, washed, and resuspended to 10 7 cells/ml in Hank's PBS with 1 mM Ca 2+ and Mg 2+ , 1 mg/ml BSA, and 10 mM HEPES (Hanks +/+ ). Cells were opsonized with 7.5 mg/ml mAb 3G8 F(ab9) 2 for 40 min at 4˚C, washed once, resuspended in Hanks +/+ , and 2 3 10 6 cells were immediately transferred to the SLM 8000. With excitation at 355 nm, the simultaneous fluorescence emission at 405 and 490 nm was measured, integrated, and recorded each second. After establishing a baseline for 60 s, goat anti-mouse F(ab9) 2 was added (20 mg/ml final concentration) and data acquisition was continued for an additional 3.5 min. Each sample was individually calibrated by lysing cells in 1% Triton X-100 to determine the maximal emission ratio and by adding EDTA (20 mM final concentration) to determine the minimal ratio. Alternatively, 2.5 3 10 6 RBL-2H3 cells expressing [RSSTR] or [RAAAR] CD16A were loaded with 2 mM Fluo-4 and opsonized with 10 mg/ml murine IgE for 20 min on ice. After establishing a baseline, cells were stimulated with goat anti-mouse Ig (20 mg/ml) and further analyzed by FACS analysis for intracellular calcium release.   (27). Labeled cells were incubated with 3G8 F(ab9) 2 at 10 mg/ml in phosphate-free DMEM at 4˚C for 10 min. After washing and resuspending in ice-cold phosphate-free DMEM, cells were incubated with 20 mg/ml F(ab9) 2 rabbit anti-mouse at room temperature for 5 min. The control samples were treated the same but incubated on ice all the times, or without addition of F(ab9) 2 rabbit anti-mouse . Cells were lysed in Triton X-100 lysis buffer and CD16A was precipitated with 3G8 overnight at 4˚C. Alternatively, CD16A was precipitated with F(ab9) 2 goat anti-mouse and F(ab9) 2 goat anti-rabbit at 4˚C overnight. Precipitates were analyzed by SDS-PAGE and autoradiography. Parallel samples without [ 32 P] labeling were analyzed by Western blotting to control protein loading.

Yeast two-hybrid studies
Yeast two-hybrid studies (DupLEX-A yeast two-hybrid system; OriGene Technologies) were performed as detailed in the manufacturer's protocols. Briefly, the bait plasmid, pEG202-CD16A-CY, was used to screen the human peripheral blood leukocyte library. Yeast cells were transformed first with the bait plasmid and selected on leucine-free plates to produce a stable LexA-CD16A-CY strain. The cells were then transformed with the library and selected on leucine-, tryptophan-, and histidine-free plates and media. Expression of LacZ was determined by the appearance of blue colonies within 24 h. Interaction of CD16A-CY mutants with S100A4 was measured by cell growth in liquid medium lacking leucine, tryptophan, and histidine. Positive yeast clones from the two-hybrid system were screened for identity. The identity of target plasmid was determined by amplification using vector-specific upper and lower primers provided in the DupLEX-A system. An S100A4-specific upper primer was designed to be compatible with the vector-specific lower primer to screen all of the positive yeast clones. S100A4 primer was 59-GGG CAA AGA GGG TGA CAA-39. The interaction of CD16A-CY with S100A4 was confirmed by cotransforming S100A4 and CD16A-CY expression plasmids into a less sensitive yeast strain, EGY 188.

In vitro binding assays
GST-CD16A-CY (12 mg) was incubated with 12 mg GST-S100A4, GST-CD64-CY, or GST in 100 ml KTT buffer (28) with 1 mM CaCl 2 or 10 mM EDTA at 4˚C overnight. CD16A was immunoprecipitated with rabbit anti-CD16A-CY overnight at 4˚C. The precipitates were washed five times with KTT buffer with 1 mM CaCl 2 or 10mM EDTA and analyzed by SDS-PAGE and Western blotting. To pull down S100A4 in cell lysate, U937 cells were lysed in Nonidet P-40 lysis buffer (29) containing no EDTA (Roche) with Ca 2+ (1.09 mM CaCl 2 ) or without Ca 2+ but containing 10 mM EDTA. The supernatant was incubated with various GST fusion proteins on glutathione-Sepharose 4B beads overnight at 4˚C. The beads were washed with the lysis buffer with or without Ca 2+ for three times and analyzed by SDS-PAGE and Western blotting with anti-S100A4 mAb 5C7. Blots were then stripped and reprobed with an anti-GST mAb to control protein loading.

Coimmunoprecipitation assays
NK3.3 cells, 293 cells transiently expressing S100A4 and CD16A or CD64, or primary mononuclear cells were lysed in Nonidet P-400 lysis buffer with or without Ca 2+ but containing 10 mM EDTA as above. CD16A or CD64 was immunoprecipitated from the supernatants with 3G8 or 197 overnight at 4˚C. Immunoprecipitates were analyzed by Western blotting with anti-S100A4. Blots were then stripped and reprobed with polyclonal anti-CD16A-CY Abs and/or a polyclonal anti-CD64-CY Ab.

PKC phosphorylates CD16A-CY but not other FcR CYs
The a-chain CYs of FcgRIIIa (CD16A), FcgRIa (CD64), and FcaRI (CD89) each have unique sequences without homology in the human genome. ProfileScan (PROSITE database) suggested several putative PKC phosphorylation motifs in their a-chain CYs although in vitro kinase assays showed essentially no phosphorylation of the CYs of CD89 or CD64. However, CD16A-CY, which contains a potential PKC phosphorylation motif [RSSTR] (Fig.  1C), was strongly phosphorylated by PKCs (Fig. 1A, 1B, Supplemental Fig. 1). Mutation of the CD16A [RSSTR] motif to [RAAAR] abolished the phosphorylation. Mutation of Ser 219 and Thr 220 to Ala 219 and Ala 220 also abolished PKC phosphorylation, localizing the phosphorylation site to these two residues (Fig. 1C).
The PKC motif in CD16A-CY differentially regulates induction of proinflammatory cytokines and degranulation Activation of monocytes/macrophages by CD16A results in the synthesis and secretion of TNFa, IL-6, and IL-1b (30,31). The levels of secretion of IL-6, IL-1b, and TNF-a in [RSSTR] CD16A-expressing P388D1 cells following receptor cross-linking were markedly higher than those in [RAAAR] cells expressing equivalent levels of CD16A ( Fig. 2A, 2B). Comparable levels of g-chain were associated with CD16A, consistent with the requirement of g-chain for CD16A surface expression (Fig. 2C). CY-tailless CD16A-expressing cells gave similar results as did [RAAAR] cells (Supplemental Fig. 3). LPS-induced cytokine production was at similar levels in these cell lines (Supplemental Fig. 3 and data not shown). The levels of secretion of IL-4 in [RAAAR] CD16A-expressing RBL-2H3 cells following receptor cross-linking were significantly reduced in comparison with those in [RSSTR] cells expressing equivalent levels of CD16A (Fig. 2D). CD16A cell surface expression was comparable (Fig. 2E) and g-chain association with CD16A was equivalent (Fig. 2F) in the transfectants. In contrast, the [RAAAR]-transfected cells showed enhanced receptor-mediated degranulation compared with [RSSTR] CD16A-transfected RBL-2H3 cells (Fig. 3A). Both [RSSTR] and [RAAAR] CD16A-expressing cells underwent comparable levels of degranulation when activated through endogenous FcεR, indicating equivalent capacities for degranulation (Fig. 3B). As a complementary approach, RBL-2H3 cells stably expressing [RSSTR] or [RAAAR] CD16A were transiently transfected with wild-type PKCu or constitutively active PKCu A148E constructs. Both the wild-type PKCu and the constitutively active PKCu resulted in a significant decrease (40.2 and 39.5% reduction at 30 min; p = 0.0004 and 0.0036, respectively) in receptor-induced degranulation in cells expressing [RSSTR] CD16A when compared with the vector control (Fig. 3C). However, the wild-type and the constitutively active PKC constructs reduced receptor-induced degranulation much less in cells expressing the [RAAAR] receptor (9.4 and 15.2% at 30 min, respectively) (Fig. 3D). Taken together, these results provide strong evidence for the role of PKC-mediated phosphorylation of CD16A a-chain in determining the balance of cell programs initiated by receptor engagement.
The PKC motif in CD16A-CY regulates g-chain-mediated signaling The CD16A-CY regulates receptor-mediated early signaling, including calcium mobilization and tyrosine phosphorylation of Syk (Ref. 15 and data not shown). We investigated the role of the CD16A-CY [RSSTR] motif in receptor-mediated early signaling in cell lines expressing functionally competent endogenous g-chain (13,14,29)

. Receptor cross-linking induced a rise in [Ca 2+ ] i of 241 nM in RBL-2H3 cells expressing [RSSTR] CD16A, but only 127 nM calcium, on average, in RBL-2H3 cells expressing [RAAAR]
CD16A at equivalent levels (Fig. 4A). Endogenous FcεR engagement induced comparable calcium fluxes in both cell lines. Receptor-induced tyrosine phosphorylation of total cellular g-chain and coprecipitated Syk was reduced markedly in P388D1 cells expressing [RAAAR] CD16A in comparison with [RSSTR] CD16A (Fig. 4B). Furthermore, although association of g-chain with the CD16A was equivalent in [RSSTR] and [RAAAR] cells, receptorinduced tyrosine phosphorylation of receptor-associated g-chain and g-chain-associated Syk was significantly reduced in P388D1 cells expressing [RAAAR] CD16A in comparison with [RSSTR] CD16A (data not shown). Thus, key components of g-chainmediated signaling are regulated by the PKC motif in CD16A.
The PKC motif in CD16A-CY regulates degranulation via a Gab2/PI3K-dependent pathway FcR-induced degranulation is likely mediated by both calciumdependent and -independent pathways (32). The adaptor protein Gab2 is critical for activation of the PI3K/Akt pathway (33) and the Fyn/Gab2/RhoA-dependent, calcium-independent pathway in mast cell degranulation (32). Engagement of CD16A on NK cells initiates a cascade of signaling events including PI3K activation (34), which is required for CD16A-mediated granule exocytosis and ADCC in NK cells (35). The enhanced degranulation in the [RAAAR] CD16A-expressing RBL-2H3 cells is unlikely caused by the calcium-dependent pathway because there is impaired calcium signaling in these cells. We therefore investigated the effect of the PKC motif on the calcium-independent Gab2/PI3K/AKT pathway. After CD16A cross-linking, tyrosine phosphorylation of Gab2 was induced in the [RAAAR] but not the [RSSTR] CD16Aexpressing RBL-2H3 and P388D1 cells (Fig. 4C and data not  shown). Similarly, levels of phosphorylation of Ser 473 and Thr 308 of AKT, an indication of maximum activation of the enzyme, markedly increased in the [RAAAR] CD16A cells following receptor cross-linking compared with that in [RSSTR] CD16A cells (Fig. 4D, 4E). Treatment with wortmannin, a highly selective inhibitor for PI3K with an in vitro IC 50 of 2-4 nM for PI3K (36)  that the enhanced degranulation in [RAAAR] CD16A RBL-2H3 cells is likely mediated by a wortmannin-sensitive, PI3K-dependent pathway.

S100A4 interacts with CD16A-CY
To identify protein partners that might interact with the CD16A cytoplasmic domain and contribute to the modulation of CY function, we performed yeast two-hybrid studies using a pEG202-CD16A-CY bait vector and the human PBL cDNA library and identified S100A4 as a specific CD16A-CY-interacting protein (Fig. 5A). To map the region of CD16A-CY that is essential for the interaction with S100A4, a series of CD16A mutant variants were constructed ( Fig. 5B and data not shown). These variants were tested for interaction of CD16A-CY with S100A4 using the yeast two-hybrid system. Deletion of the six most C-terminal residues of CD16A abrogated the optimal interaction with S100A4, implicating these to be part of the S100A4 binding site of CD16A (Fig. 5C). To verify interaction between CD16A-CY and S100A4, we used GST fusion proteins and in vitro pull-down assays to demonstrate that that GST-S100A4, but not GST or GST-CD64-CY, was pulled down by GST-CD16A-CY (Fig. 6A). Similarly, GST-CD16A-CY, but not GST-CD64-CY, was pulled down by GST-S100A4 or purified S100A4 (data not shown). Furthermore, GST-CD16A-CY, GST-CD64-CY, GST-CD89-CY, or GST was incubated with U937 cell lysates containing endogenous S100A4, and S100A4 was only pulled down with GST-CD16A-CY in the presence of Ca 2+ (Fig. 6B). These data demonstrated that the interaction between CD16A-CY and S100A4 is direct, Ca 2+ -dependent, and specific. To confirm that these two proteins interact in cells, CD16A was immunoprecipitated from NK3.3 cell lysates and probed for CD16A and S100A4, which coprecipitated with CD16A in these cells in a Ca 2+dependent manner (Fig. 6C). CD16A or CD64, immunoprecipitated from 293 cells transiently expressing CD16A or CD64 and S100A4, showed coimmunoprecipitation of Sl00A4 with CD16A but not with CD64 in a Ca 2+ -dependent manner (Fig. 6D). Furthermore, S100A4 was coimmunoprecipitated with CD16A but not CD64 in lysates from primary mononuclear cells in the presence of calcium (Fig. 6E), indicating specific interaction of the two proteins at physiological expression levels. To study the subcellular localization of both proteins in freshly prepared primary mononuclear cells, cells were stained for S100A4 (green) and CD16 (red) and examined by scanning laser microscopy. About 20-30% of the cells stained positive for CD16A, presumably being NK cells, monocytes, and macrophages. S100A4 was also present in these cells, with a cytoplasmic distribution and a few patches of colocalization with CD16 (Fig. 6F). When CD16 was cross-linked, enhanced colocalization of CD16A and S100A4 was evident with some comigration to the interior of cells. Thus, modest colocalization of S100A4 with CD16A is enhanced following receptor cross-linking. In cells pretreated with BAPTA-AM, however, S100A4 no longer colocalized with CD16A following receptor cross-linking, indicating that the colocalization is Ca 2+ -dependent. CD64 and S100A4 were not colocalized in receptor cross-linked cells.
S100A4 inhibits phosphorylation of CD16A by PKC and affects receptor-mediated degranulation S100A4 inhibits phosphorylation of its target proteins by PKC and/ or CK2 (20,22,23,37). In vitro kinase assays with GST-CD16A-CY and PKC in the presence or absence of S100A4 demonstrated FIGURE 5. CD16A-CY specifically interacts with S100A4 protein in the yeast LexA two-hybrid analysis. Initial screen of 52 3 10 6 yeast cells carrying the CD16A-CY bait vector identified 399 clones that were able to survive on minimal media. A secondary screen using a LacZ reporter containing the Gal4 minimal promoter identified 150 LacZ-positive clones. Sequencing identified 84% of the LacZ-positive clones as S100A4. (A) The colonies for each combination of cDNA were transferred to leucine-, tryptophan-, and histidine-free plates and tested for b-galactosidase activity (indicated by blue staining within 24 h). CD16A-CY bait, but not CD64-CY bait, mediated LacZ expression (n = 3). (B) Schematic representation of mutations in the cytoplasmic domain of CD16A. (C) Deletion of the C-terminal six amino acids greatly reduced the interaction of CD16A with S100A4 (n = 3). that GST-S100A4, but not GST, inhibited phosphorylation of GST-CD16A-CY by PKC (Fig. 7A, Supplemental Fig. 4A). Because GST-S100A4 inhibited phosphorylation of GST-CD16A-CY and p53 but did not affect the phosphorylation of MBP by PKC under the same conditions, we exclude the possibility that GST-S100A4 acts as a competitive substrate or results in substrate inhibition of the kinase (Supplemental Fig. 4B, 4C). Notably, the level of inhibition of CD16A phosphorylation by PKC in the presence of S100A4 was similar to that reported for other target proteins of S100A4m including p53 and liprin-b1 (20)(21)(22)(23). To assess the effect of S100A4 binding on CD16A phosphorylation in vivo, phosphorylation of CD16A was measured in S100A4 small interfering RNA (siRNA) or control RNA-treated RBL-2H3 cells. S100A4 was reduced to an undetectable level in S100A4 siRNAtreated cells (data not shown). The ratio of [ 32 P]CD16A to CD16A protein in anti-CD16 immunoprecipitates increased 40.5% in cells treated with control RNA, whereas the ratio increased 78.6% in S100A4 siRNA-treated cells after receptor cross-linking (Fig. 7B), indicating interference of S100A4 with induced phosphorylation of CD16A. Thus, S100A4 may regulate CD16A-specific functions by modulating its phosphorylation by PKC. Correspondingly, degranulation measured at 10 min after CD16A cross-linking was reduced in cells treated with S100A4 siRNA (Figs. 7C, 8).

Discussion
The [RSSTR] sequence in the cytoplasmic domain of CD16A is phosphorylated by PKC in in vitro kinase assays, in human cell lines, and in primary human cells. The phosphorylation is specific to this motif, and the phosphorylated and nonphosphorylatable states differentially regulate receptor-initiated signaling and downstream cell programs. Phosphorylated CD16A favors more robust tyrosine phosphorylation of the FcRg-chain with greater flux in [Ca 2+ ] i and production of IL-1b, IL-6, TNF-a, and IL-4, a provoc-ative observation given the hyperactivation of PKCu in a subset of lupus patients (38). In contrast, the [RAAAR] CD16A leads to greater activation of Gab2 and AKT and greater dependence of degranulation on the PI3K pathway. Furthermore, the calciumsensitive S100A4, known to inhibit the phosphorylation of its target proteins by PKC and CK2, interacts specifically with CD16A (among FcRs), inhibits phosphorylation of CD16A-CY by PKC in vitro, and modulates in vivo CD16A phosphorylation. Taken together, these data suggest that S100A4 may serve as a modulator of CD16A [RSSTR] phosphorylation. Its calcium-dependent binding of CD16A may diminish further phosphorylation, providing both a negative feedback loop for cytokine production and a mechanism favoring degranulation by receptors newly recruited to structures such as the immunological synapse in conjugate formation.
Among the FcRg-chain-associated FcRs, total truncation mutants suggest a contribution of the a-chain cytoplasmic domain to receptor signaling and cell programs (29,39). We hypothesized that Ser/Thr phosphorylation events might provide a window on the mechanism of a-chain contributions, and in silico analysis identified the PKC phosphorylation consensus motif [RSSTR] within the CD16A cytoplasmic domain, whereas FcgRI (CD64) and FcaR (CD89) have CK2 and CK1 sites, respectively. The predicted specificity for kinases has been supported experimentally, suggesting important mechanisms for differential regulation. Further evidence for receptor-specific mechanisms for regulation is provided by the specific interaction of S100A4 with the cytoplasmic domain of CD16A, but not with those of CD64 or CD89. Not only do CD16A and S100A4 coimmunoprecipitate in a Ca 2+ dependent manner in human cell lines and primary mononuclear cells, but they also colocalize in primary mononuclear cells. S100A4 inhibits phosphorylation of its binding partner proteins by PKC and/or CK2 (20,22,23,37). Consistent with these obser- CD16A stable transfectants were transfected with S100A4 siRNA or nonsilencing control RNA (45). After 24 h, phosphorylation of CD16A in [ 32 P] metabolically labeled RBL-2H3 cells was measured as described above. Data are representative of three experiments. (C) RBL-2H3 stable transfectants expressing CD16A were transfected with S100A4 siRNA or control RNA. After 24 h, degranulation was measured at 10 min after CD16A cross-linking at the indicated concentrations. Data represent the means 6 SD for six experiments. **p , 0.01, ***p , 0.001 by Student paired t test.
vations, we show that S100A4 inhibits phosphorylation of CD16A-CY by PKC in vitro. Furthermore, in cells treated with S100A4 siRNA, phospho-CD16A induced by receptor cross-linking was increased, indicating interference of S100A4 with induced phosphorylation of CD16A.
Nonphosphorylatable CD16A enhanced degranulation, implying that phosphorylation of the PKC motif downregulates receptormediated degranulation. Indeed, overexpression of a constitutively active PKCu construct decreased receptor-mediated degranulation in cells expressing wild-type, but only minimally in cells with the nonphosphorylatable CY receptor, clearly supporting the notion that phosphorylation of the [RSSTR] motif is responsible for this effect. Although PKCs are important mediators in initiation of many effector functions of Fcg receptors, including NK cell killing and phagocytosis (40,41), their effects are no doubt multiple. For example, PKCd-deficient mast cells exhibited a sig-nificantly higher level of degranulation, suggesting that PKCd is a negative regulator of Ag-induced mast cell degranulation (42). In addition to the well-known calcium-dependent pathway, degranulation in mast cells requires a calcium-independent, Fyn/Gab2/RhoAdependent pathway, which plays a critical role in the microtubuledependent translocation of granules to the plasma membrane (32). Gab2 is critical for PI3K recruitment and activation of the PI3K/ AKT pathway (33) and Rho GTPases, involved in microtubule organization and granule translocation, and it is a downstream effector of PI3K (reviewed in Ref. 43). Engagement of CD16A on NK cells initiates PI3K activation (34), which is required for CD16-mediated granule exocytosis and ADCC by NK cells (35). Hyperresponsive CD16A-initiated degranulation in RBL-2H3 cells with nonphosphorylableCD16A is unlikely due to the effects on the calcium-dependent pathway, which is downregulated in cells with a nonphosphorylatable receptor. In contrast, the Gab2/PI3K/AKT pathway is upregulated in cells with nonphosphorylable CD16A, and degranulation is more sensitive to wortmannin inhibition in these cells. Thus, phosphorylation of CD16A regulates two distinct functions mediated by the receptor via at least two distinct pathways: a calcium-dependent pathway is concurrently upregulated, leading to augmented production of proinflammatory cytokines; and a calcium-independent, PI3K-dependent pathway is downregulated, leading to attenuated degranulation. That activation of PI3K after CD16 cross-linking in human primary monocytes limited the expression of TNFa, IL-1b, and IL-6 (44) supports the notion that PI3K activation plays distinct roles in CD16A-mediated degranulation and production of proinflammatory cytokines.
Our proposed model of PKC-mediated, phosphorylation-dependent balancing of proinflammatory cytokine production and degranulation, with a feedback loop involving the calcium-activated binding of S100A4 (Fig. 8), focuses attention on a critical role of the ligand-binding a-chain of CD16A and other Fc common g-chain-associated receptors. The calcium-activated binding of S100A4 to CD16A may serve as such an inhibitory modulator on the Syk/cytokine pathways during sustained signaling and favor a shift to degranulation upon conjugate formation between CD16A + effector cells and target cells. Such a molecular switch points to new therapeutic targets. Furthermore, the subtle modulation of the contributions of the ligand-binding a-chain through genetic variation in the cytoplasmic region, a theme clearly established within the extracellular domain, may provide additional insights into receptor function and into risk for immune-mediated diseases. FIGURE 8. Roles of CD16A-CY phosphorylation in receptor-mediated functions. Upon engagement, phosphorylation of CD16A regulates two distinct functions mediated by the receptor via at least two distinct pathways: a calcium-dependent pathway is concurrently upregulated, leading to augmented production of proinflammatory cytokines; and a calciumindependent, PI3K-dependent pathway is downregulated, leading to attenuated degranulation. Furthermore, calcium-dependent binding of S100A4 to CD16A diminishes further phosphorylation, providing both a negative feedback loop, potentially restricting cytokine production, and a mechanism favoring degranulation.