Extracellular mutations of non-obese diabetic mouse FcgammaRI modify surface expression and ligand binding.

The non-obese diabetic mouse (NOD) expresses a unique form of the high affinity receptor for IgG (FcγRI), containing multiple mutations that result in substitutions and insertions of amino acids and a truncated cytoplasmic tail. As a result of these major changes, receptor affinity for IgG increases 10-fold over that of wild-type FcγRI from BALB/c mice, while the specificity for ligand is retained. Kinetic analysis revealed that while the association rate of IgG with FcγRI from NOD mice (FcγRI-NOD) and FcγRI from BALB/c mice (FcγRI-BALB) is similar, IgG bound much more tightly to FcγRI-NOD as revealed by a profoundly diminished dissociation rate. Transfection studies demonstrated that FcγRI-NOD was expressed at one-tenth of the level of FcγRI-BALB. Although mouse FcγRI was previously not known to associate with the FcεRI γ-subunit, transfection of COS-7 cells demonstrates that like human FcγRI, mouse FcγRI is also able to associate with this signaling subunit. Furthermore, expression levels of FcγRI-NOD were not restored by the presence of the FcεRI γ-subunit. The difference in the levels of expression was mapped to mutations in the extracellular region of FcγRI-NOD as replacement of the extracellular domains with those of human FcγRI or FcγRI-BALB restored expression to that of human FcγRI or FcγRI-BALB.

The non-obese diabetic mouse (NOD) expresses a unique form of the high affinity receptor for IgG (Fc␥RI), containing multiple mutations that result in substitutions and insertions of amino acids and a truncated cytoplasmic tail. As a result of these major changes, receptor affinity for IgG increases 10-fold over that of wild-type Fc␥RI from BALB/c mice, while the specificity for ligand is retained. Kinetic analysis revealed that while the association rate of IgG with Fc␥RI from NOD mice (Fc␥RI-NOD) and Fc␥RI from BALB/c mice (Fc␥RI-BALB) is similar, IgG bound much more tightly to Fc␥RI-NOD as revealed by a profoundly diminished dissociation rate.
Transfection studies demonstrated that Fc␥RI-NOD was expressed at one-tenth of the level of Fc␥RI-BALB. Although mouse Fc␥RI was previously not known to associate with the Fc⑀RI ␥-subunit, transfection of COS-7 cells demonstrates that like human Fc␥RI, mouse Fc␥RI is also able to associate with this signaling subunit. Furthermore, expression levels of Fc␥RI-NOD were not restored by the presence of the Fc⑀RI ␥-subunit. The difference in the levels of expression was mapped to mutations in the extracellular region of Fc␥RI-NOD as replacement of the extracellular domains with those of human Fc␥RI or Fc␥RI-BALB restored expression to that of human Fc␥RI or Fc␥RI-BALB.
The mouse high affinity receptor for IgG, Fc␥RI (CD64), consists of three Ig-like extracellular domains and is the only Fc␥ receptor that binds monomeric IgG (1,2). Expressed on monocytes and macrophages and induced by interferon-␥ (IFN-␥) 1 on neutrophils (3)(4)(5), Fc␥RI functions by linking the humoral and cellular responses. Although functions mediated by mouse Fc␥RI are less well characterized, cross-linking of human Fc␥RI on myeloid cells leads to events such as tyrosine phosphorylation (6), Ca 2ϩ flux (7), superoxide generation (8), inflammatory mediator release (9), antibody-dependent cellular cytotoxicity (10), and internalization of small immune complexes (11). Many of these sequelae may also be true in the mouse. While human Fc␥RI is associated with a homodimer of the ␥-subunit from Fc⑀RI (12)(13)(14), receptor co-association has not been shown in the mouse, and indeed, there are some differences between the function of human and mouse Fc␥RI (e.g. mouse Fc␥RI is constitutively phosphorylated, while human Fc␥RI is not) (15,16).
Mouse Fc␥RI is structurally homologous to human Fc␥RI and exhibits high affinity binding of monomeric IgG (2). Genetic mapping studies revealed that the single gene encoding mouse Fc␥RI (Fcg1) is situated on chromosome 3, and studies in the non-obese diabetic mouse (NOD) revealed a linkage with the Idd-3 diabetes susceptibility genetic marker on chromosome 3 (17)(18)(19). The Fcg1 allele found in NOD mice has 24 single base differences (compared with the BALB/c allele), 17 of which encode changes in the predicted amino acid sequence, including a four-amino acid insertion between domains 2 and 3 and a four-nucleotide deletion leading to a frameshift and premature translation termination codon that essentially eliminates the cytoplasmic domain (truncated by 73 amino acids) (19). Studies on NOD peripheral blood MAC-1 ϩ cells demonstrated a lower surface expression of Fc␥RI compared with cells from C57BL/10SnJ mice. Binding and turnover studies revealed that Fc␥RI from NOD cells demonstrated a 73% reduction in the turnover of bound mIgG2a compared with C57BL/ 10SnJ mice (19).
This study characterizes Fc␥RI from NOD mice (Fc␥RI-NOD) and investigates the influence the mutations have on the cell-surface expression of Fc␥RI-NOD, association with the Fc⑀RI ␥-subunit, and interaction with ligand.

Fc␥RI cDNA Constructs and Generation of Chimeric Receptors-The
Fc␥RI-NOD cDNA was generated by reverse transcription-PCR from RNA isolated from NOD/Lt spleen cells. Briefly, total RNA was isolated from 10 7 spleen cells using guanidinium thiocyanate (20), and firststrand cDNA was synthesized using reverse transcription (Pharmacia Biotech Inc.). PCR was used to generate Fc␥RI-NOD and Fc␥RI-BALB cDNA clones as described (21) using 500 ng of oligonucleotide primers MDH3 and TISM6 and 2 units of Taq polymerase (Amplitaq, Perkin-Elmer) for 30 amplification cycles. The PCR products were subcloned into the expression vector pKC4 (22). The clone FcRI.1 contained the nucleotide sequence of Fc␥RI-NOD (19), and the clone FcRI.3 contained the nucleotide sequence of Fc␥RI-BALB (2). The predicted amino acid sequence of Fc␥RI-NOD (19) is shown in Fig. 1 and is compared with the amino acid sequence of Fc␥RI-BALB (2) and human Fc␥RI (23).
Chimeric receptors were generated by exchanging extracellular domain sequences between Fc␥RI-NOD, Fc␥RI-BALB, and human Fc␥RI using splice overlap extension-PCR (24). The exchange points in the extracellular domains were Leu 268 (NOD), Leu 263 (BALB/c), and Leu 263 (human) and are illustrated in Fig. 1. Briefly, two PCRs were used to amplify the cDNA fragments encoding the extracellular domains of either human Fc␥RI (cDNA gift from B. Seed) (5Ј-primer T-9 and 3Ј-primer T-10) or Fc␥RI-BALB (5Ј-primer MDH3 and 3Ј-primer T-10) and the transmembrane and cytoplasmic domains of either Fc␥RI-NOD or Fc␥RI-BALB (both using 5Ј-primer T-11 and 3Ј-primer T-5). A third PCR was performed to splice the two fragments together and to amplify the spliced product. The cDNA sequences were checked by dideoxynucleotide sequencing (25). The receptor construct containing Fc␥RI-BALB extracellular domains and transmembrane and cytoplas-mic domains of Fc␥RI-NOD origin was designated BALB-NOD. The chimeric receptors containing human Fc␥RI extracellular domains and transmembrane and cytoplasmic domains of either Fc␥RI-NOD or Fc␥RI-BALB origin were designated Hu-NOD and Hu-BALB, respectively.
The mouse Fc⑀RI ␥-subunit cDNA was a kind gift from Dr. U. Blank and was subcloned into the expression vector pRc/CMV (Invitrogen). The ␥-actin cDNA was used as a control (26).
Antibody Reagents-The monoclonal antibody 49-11.1 (anti-Ly2.1) (28) was used as a source of mouse IgG2a and FabЈ fragments. Human IgG1 were purified from a myeloma patient, and antibodies were assessed as monomeric after size fractionation chromatography (Superose 12, Pharmacia). Polyclonal rabbit anti-sheep erythrocyte serum was used to sensitize sheep erythrocytes for EA rosetting studies of complexed IgG binding, and mouse IgG1 anti-trinitrobenzene sulfonate (29) was used to sensitize sheep erythrocytes coated with trinitrobenzene sulfonate (Fluka Chemika, Buchs, Switzerland). Polyclonal rabbit anti-Fc⑀RI ␥-subunit serum used in immunoblotting was a kind gift from Dr. J.-P. Kinet. The monoclonal antibodies 2.4G2 (rat anti-mouse Ly17 mAb) (30) and F4/80 (rat anti-mouse macrophage marker) and human IgG1 myeloma IgG were used in FACS analysis and were detected with FITC-conjugated sheep anti-mouse Ig or FITC-conjugated sheep antihuman Ig (Silenus, Melbourne, Australia).
Monomeric IgG was radiolabeled using Na 125 I (Amersham International, Buckinghamshire, United Kingdom) with chloramine T as described (31); unincorporated iodine was removed using pre-packed G-25 (PD-10) columns (Pharmacia); and the radiolabeled antibody was stored at 4°C for up to 1 week. Radiolabeled IgG was shown to compete equally with unlabeled IgG in binding to Fc␥RI-expressing COS-7 cells.
Northern and Southern Analyses-Total RNA was harvested from COS-7 cells 48 h post-transfection as described (20), and the samples were treated with DNase I (Pharmacia) for 1 h before electrophoresis in agarose containing formaldehyde and Northern transfer to a nylon membrane as described (32). Northern blots were hybridized with either the complete cDNA of Fc␥RI-BALB (2) or ␥-actin cDNA (26) and autoradiographed.
Total DNA was harvested 48 h post-transfection from COS-7 cells by incubating cells in lysis buffer (50 mM Tris-Cl, pH 7.5, 100 mM EDTA, 100 mM NaCl, 1% SDS, and 500 g/ml proteinase K) for 4 h at 55°C prior to phenol and chloroform extractions and precipitation of DNA with an equal volume of isopropyl alcohol. The DNA was recovered by centrifugation and sheared through 23-gauge needles prior to overnight digestion with EcoRI. 10 g of DNA was electrophoresed and transferred to a nylon membrane as described (32). Southern blots were hybridized with the complete cDNA of Fc␥RI-BALB (2) and autoradiographed.
Cell lysates were incubated with 30 l of packed Sepharose beads coupled with either whole mouse IgG2a or FabЈ fragments of mouse IgG2a for 1 h at 4°C with rotation before seven washes with lysis buffer. Samples were boiled in the presence of ␤-mercaptoethanol and analyzed by SDS-PAGE.
Western Blotting and Fc⑀RI ␥-Subunit Association-Transfected COS-7 cell lysates were incubated with whole mouse IgG2a-Sepharose and mouse IgG2a FabЈ fragment-Sepharose as described above, washed, and denatured under nonreducing conditions prior to SDS-PAGE and Western blotting onto an Immobilon-P membrane (Millipore Corp., Bedford, MA). After transfer, the membrane was blocked with 2% casein in Tris-buffered saline (10 mM Tris-Cl, pH 7.5, 150 mM NaCl) for 2 h before overnight incubation with polyclonal rabbit anti-Fc⑀RI ␥-subunit antiserum in Tris-buffered saline containing 1% casein. After washing with Tris-buffered saline containing 0.1% Tween 20, the blots were incubated with peroxidase-conjugated swine anti-rabbit IgG (DAKO-PATTS) and then washed again prior to chemiluminescent detection (ECL, Amersham International).
Ligand Binding Analysis-Various concentrations of radiolabeled human IgG1 ( 125 I-hIgG1) and mouse IgG2a ( 125 I-mIgG2a) (radiolabeled as described above) were incubated with transfected COS-7 cells at 1 ϫ 10 6 cells/ml for 150 min on ice prior to centrifugation through phthalate oils (3:2 (v/v) dibutyl phthalate/dioctyl phthalate) (Fluka Chemika), and bound 125 I-IgG (cellular pellets) and free 125 I-IgG (supernatant) were determined. Nonspecific 125 I-IgG binding was determined either by the amount of 125 I-hIgG1 bound in the presence of 100-fold excess unlabeled IgG or by nonspecific binding to mock-transfected COS-7 cells, and the counts were subtracted from total binding to give specific IgG binding levels. Ratios of bound and free fractions were calculated, and Scatchard analysis was performed.
Association and dissociation kinetics of IgG binding were performed at 4°C with monomeric 125 I-hIgG1 on COS-7 cells transfected with the Fc␥RI constructs. The association of IgG was determined using transfected COS-7 cells (1 ϫ 10 6 cells/ml) and incubating with 125 I-hIgG1 (5 g/ml) for the indicated times before assaying cell-bound 125 I-hIgG1 by pelleting cells through phthalate oils. To compensate for the various levels of surface expression of the two receptors, the data are presented as percent maximum bound IgG, where 100% is the level of binding demonstrated after 3 h of incubation on ice.
The dissociation of bound human 125 I-hIgG1 was measured by incubating either transfected COS-7 cells or mouse bone marrow-derived macrophages (ϮIFN-␥) with 125 I-hIgG1 (5 g/ml) for 2 h before adding 200-fold excess unlabeled hIgG (1 mg/ml) to the cells at time 0. Cellbound 125 I-hIgG1 was assayed at the indicated times by pelleting cells through phthalate oils. To compensate for the various levels of surface expression, the data are presented as percent maximum bound IgG, where 100% is the level of binding demonstrated at time 0. The dissociation of bound 125 I-hIgG1 from Fc␥RI-transfected COS-7 cells was performed at both 4 and 22°C, while the dissociation of 125 I-hIgG1 from Fc␥RI on bone marrow-derived macrophages was measured at 4°C.
Bone Marrow-derived Macrophages-NOD/Lt and BALB/c bone marrow-derived macrophages were generated as adherent cells from their nonadherent progenitors essentially as described before (33), but with minor modifications so that they could be removed from the culture surface. The macrophages were eventually grown on Petrie dishes for 4 days in RPMI 1640 medium supplemented with 50 M 2-mercaptoethanol, 20 mM HEPES, 15% fetal calf serum, and 20% L-cell conditioned medium (as a crude source of colony-stimulating factor-1 or macrophage colony-stimulating factor). The bone marrow-derived macrophages were a relatively pure and homogeneous population, with Ն95% of adherent cells binding colony-stimulating factor-1. Cells were washed twice with phosphate-buffered saline and where stated were treated with 1000 units/ml recombinant murine IFN-␥ (Nippon-Roche, Tokyo, Japan) 18 h before harvesting by vigorous washing.
FACS Analysis of Bone Marrow-derived Macrophages-Bone marrow-derived macrophages that had been stimulated with IFN-␥ were incubated with mAb 2.4G2 (anti-Fc␥RII/Fc␥RIII), mAb F4/80 (antimacrophage marker), and/or human IgG1 for 30 min on ice prior to washing and incubation with specific FITC-conjugated secondary antibodies. Since there are no mAbs that detect mouse Fc␥RI, the "specific" staining of mouse Fc␥RI was performed by blocking Fc␥RII/Fc␥RIII binding with mAb 2.4G2 (100 g/ml) prior to the addition of monomeric human IgG (20 g/ml). Human IgG bound by Fc␥RI was detected using FITC-conjugated sheep anti-human Ig, which does not cross-react with rat IgG. Binding of antibodies was measured as surface fluorescence using the FACScan (Becton Dickinson).

RESULTS
Biochemical Characterization of Fc␥RI-NOD-To establish the molecular mass of Fc␥RI-NOD, COS-7 cells were transfected with either Fc␥RI-NOD or Fc␥RI-BALB cDNAs, and cells were radioiodinated and lysed 48 h post-transfection (Fig.  2). Immunoprecipitation studies were performed on cell lysates from COS-7 cells transfected with Fc␥RI-NOD (Fig. 2, lanes a  and b) and Fc␥RI-BALB (lanes c and d) using whole mouse IgG2a-Sepharose (lanes a and c) and mouse IgG2a FabЈ fragment- Sepharose (lanes b and d). The 70-kDa moiety precipitated by whole mIgG2a from Fc␥RI-BALB-transfected COS-7 cells (Fig. 2, lane c) was the expected size for Fc␥RI-BALB (16). The molecular mass of Fc␥RI-NOD, however, was expected to be smaller than 70 kDa due to the truncation of the cytoplasmic domain and was predicted to be Ϸ45 kDa. No such moiety was observed from immunoprecipitations with whole mIgG2a (Fig.  2, lane a), even after prolonged exposure of the autoradiograph. FabЈ fragments did not precipitate any material from either the Fc␥RI-NOD-transfected (Fig. 2, lane b) or Fc␥RI-BALB-transfected (lane d) lysates. The failure to immunoprecipitate Fc␥RI-NOD was possibly due to low expression levels or failure to bind whole mIgG2a. These possibilities are addressed below.
Expression of Fc␥RI and Binding of IgG-To assess if ligand binding differences were the reason for the observed immunoprecipitation difference between Fc␥RI-NOD and Fc␥RI-BALB, binding studies using monomeric 125 I-mIgG2a and 125 I-hIgG1 were performed (Fig. 3). COS-7 cells transfected with Fc␥RI-NOD (Fig. 3A) or Fc␥RI-BALB (Fig. 3B) were incubated with various concentrations of 125 I-hIgG1 or 125 I-mIgG2a. It is clear that both Fc␥RI-NOD and Fc␥RI-BALB are able to bind IgG; however, the total saturable binding of IgG2a to Fc␥RI-NOD was one-tenth that of IgG2a binding to Fc␥RI-BALB. This difference was apparent in the binding of both hIgG1 and mIgG2a by Fc␥RI-NOD and was ϳ0.5 ng bound per 5 ϫ 10 4 cells, 10-fold less than cells transfected with Fc␥RI-BALB (5-6 ng bound per 5 ϫ 10 4 cells). This 10-fold difference in expression levels was reproducible in all 10 experiments performed.
Transcription and Transfection Efficiency-To establish that the difference in surface expression was due to Fc␥RI-NOD receptor difference and not variability in the assay systems, the efficiency of transfection and mRNA levels were investigated. Total RNA (treated with DNase I) was harvested from COS-7 cells transfected with Fc␥RI-NOD (Fig. 4A, lane a), Fc␥RI-BALB (lane b), or the Fc⑀RI ␥-subunit (negative control) (lane c) and hybridized with Fc␥RI-BALB cDNA (upper panel) and ␥-actin cDNA (lower panel). Hybridization of the mouse Fc␥RI cDNA revealed equivalent amounts of RNA in cells transfected with either Fc␥RI-NOD (Fig. 4A, lane a) or Fc␥RI-BALB (lane b). Moreover, hybridization was specific as no Fc␥RI mRNA was detected in cells transfected with Fc⑀RI ␥-subunit cDNA only (Fig. 4A, lane c). RNA probed with ␥-actin cDNA revealed that similar amounts of total RNA were present in each sample.
Southern blot analysis of total DNA was performed to determine the levels of plasmid DNA in the transfected cells (Fig.  4B). Total DNA, digested with EcoRI (which linearizes transfected plasmid to 4.76 kilobases) and hybridized with Fc␥RI-BALB cDNA, indicated that approximately equivalent amounts of Fc␥RI plasmid DNA were present in samples from cells transfected with Fc␥RI-NOD (Fig. 4B, lane a) and Fc␥RI-  a and b) or Fc␥RI-BALB (lanes c and d), transfected cells were cell surface-iodinated, lysed, and then incubated with Sepharose beads coupled with either whole mouse IgG2a (lanes a and c) or FabЈ fragments derived from mouse IgG2a (lanes b and d). BALB (lane b). Specificity was demonstrated by the failure to detect any hybridizing material in cells transfected with the Fc⑀RI ␥-subunit (Fig. 4B, lane c). No difference in levels of mRNA or transfected DNA was observed.
Localization of Region Involved in Diminished Surface Expression-Chimeric Fc receptors were generated to localize the region of Fc␥RI-NOD affecting the level of cell-surface expression. These receptors were generated using splice overlap extension-PCR, wherein the extracellular domains of Fc␥RI-NOD were replaced with either Fc␥RI-BALB sequence, to generate chimeric BALB-NOD, or human Fc␥RI sequence, to generate Hu-NOD (Fig. 5A) (see also "Experimental Procedures"). These sequence exchanges were made at the conserved leucine (positions 268, 263, and 263 of NOD/Lt, BALB/c, and human Fc␥RI, respectively) within the conserved sequence LELQVLG on the N-terminal side of the predicted transmembrane region of each receptor (Figs. 1 and 5A).
Surface expression of chimeric receptors was tested by assessing the binding of monomeric human 125 I-IgG1 (Fig. 5B). The binding of 125 I-hIgG1 to BALB-NOD, which has BALB/c extracellular domains but the NOD transmembrane region and cytoplasmic tail, was indistinguishable from that to Fc␥RI-BALB, implying that the reduced expression of Fc␥RI-NOD is likely to be due to differences in the extracellular domains of Fc␥RI-NOD rather than the membrane-spanning region or truncated cytoplasmic tail (see Fig. 5A). Similarly, the binding of IgG to the Hu-NOD chimeric receptor (human Fc␥RI has similar affinity for hIgG as Fc␥RI-BALB) was also indistinguishable from that of Fc␥RI-BALB. Thus, replacement of the Fc␥RI-NOD extracellular domains with either Fc␥RI-BALB or human Fc␥RI sequence restored 125 I-hIgG1 binding levels to those of Fc␥RI-BALB and indicated that the mutant trans-

FIG. 4. Northern analysis of total RNA (A) and Southern analysis of DNA (B) from COS cells transfected with Fc␥RI-NOD (lane a), Fc␥RI-BALB (lane b), or the Fc⑀RI ␥-subunit (lane c).
Total RNA was harvested 48 h post-transfection (A) and was probed with Fc␥RI cDNA (upper panel) or ␥-actin cDNA (lower panel). DNA was harvested 48 h post-transfection (B) and was digested with EcoRI before Southern transfer and hybridization with mouse Fc␥RI cDNA. kb, kilobases. Association of Fc⑀RI ␥-Subunit with Mouse Fc␥RI-Mouse Fc␥RI is not known to associate with protein subunits. However, human Fc␥RI, Fc␥RIII, and Fc⑀RI all share a common subunit (Fc⑀RI ␥-subunit) (12-14, 34 -37). The association with the Fc⑀RI ␥-subunit has been found to be dependent upon membrane-spanning regions of Fc⑀RI and Fc␥RIII (34 -37). While expression of human Fc␥RI is not dependent on the association with the Fc⑀RI ␥-subunit, Fc␥RIII and Fc⑀RI both require the Fc⑀RI ␥-subunit for expression. Thus, the Fc⑀RI ␥-subunit may be required to "optimize" mouse Fc␥RI expression, especially that of Fc␥RI-NOD.
Cotransfection and immunoprecipitation experiments were performed with Fc␥RI-BALB or Fc␥RI-NOD and the Fc⑀RI ␥-subunit to demonstrate any association of the Fc⑀RI ␥-sub-unit with mouse Fc␥RI. The cotransfection of the Fc⑀RI ␥-subunit and Fc␥RI-NOD did not increase cell-surface expression of Fc␥RI-NOD. Mouse IgG2a failed to immunoprecipitate any labeled material from the transfected COS-7 cells (Fig. 6A, lane  a). However, Fc␥RI-BALB was clearly precipitated by IgG2a from cells cotransfected with the Fc⑀RI ␥-subunit (Fig. 6A, lane e). No material was precipitated by IgG2a from cells transfected with the Fc⑀RI ␥-subunit only (Fig. 6A, lanes c and d) or by FabЈ fragment-conjugated beads (lanes b, d, and f). It was clear, however, that the transfected cells were expressing the Fc⑀RI ␥-subunit as Western blots of whole cell lysates indicated the presence of the 22-kDa dimer protein (Fig. 6B)  To establish if the Fc⑀RI ␥-subunit was capable of association  and f). Lysates of these cells were incubated with whole mouse IgG2a- Sepharose (lanes a, c, and e) or FabЈ fragment- Sepharose (lanes b, d, and  f), and Western blots were probed with rabbit anti-Fc⑀RI ␥-chain antibody.
with mouse Fc␥RI, COS-7 cells were transfected with Fc␥RI-BALB and Fc⑀RI ␥-subunit cDNA (Fig. 6C, lanes a and b), with Fc␥RI-NOD and the Fc⑀RI ␥-subunit (lanes c and d), or with the Fc⑀RI ␥-subunit alone (lanes e and f). Samples were incubated with either intact mouse IgG2a-Sepharose (Fig. 6C, lanes  a, c, and e) or FabЈ fragment- Sepharose (lanes b, d, and f). Following SDS-PAGE under nonreducing conditions, Western blotting revealed the presence of the Fc⑀RI ␥-subunit 22-kDa dimer in association with Fc␥RI-BALB (Fig. 6C, lane a); however, no Fc⑀RI ␥-subunit was detected in precipitates from cells transfected with Fc␥RI-NOD and the Fc⑀RI ␥ subunit (lane c) or with the Fc⑀RI ␥-subunit alone (lane e). As expected, precipitation was specific for Fc␥RI as no material was precipitated from these cells after incubation with FabЈ fragments of mouse IgG2a (Fig. 6C, lanes b, d, and f). Thus, the observation that Fc␥RI-NOD did not appear to associate with the Fc⑀RI ␥-subunit was presumably due to the low surface expression of this receptor (Fig. 6A) and, consequently, the low level of coimmunoprecipitation of the Fc⑀RI ␥-subunit.
Since the membrane-spanning region of the Fc⑀RI ␣-subunit and Fc␥RIIIa have been shown to be crucial for the association with the Fc⑀RI ␥-subunit (34 -36), the question of whether the membrane-spanning region of Fc␥RI-NOD and amino acid changes found therein would influence association with the Fc⑀RI ␥-subunit was assessed. Chimeric receptors (see Fig. 5A) with extracellular domains of human Fc␥RI origin and transmembrane and cytoplasmic domains from either Fc␥RI-BALB (Hu-BALB) (Fig. 6D, lanes a and b) or Fc␥RI-NOD (Hu-NOD) (lanes c and d) were cotransfected into COS-7 cells with the Fc⑀RI ␥-subunit (lanes e and f, Fc⑀RI ␥-subunit alone). Radiolabeled cell lysates were incubated with whole mIgG2a-Sepharose (Fig. 6D, lanes a, c, and e) or mIgG2a FabЈ fragment- Sepharose (lanes b, d, and f) prior to SDS-PAGE under nonreducing conditions. Western blotting revealed that the 22-kDa homodimer of the Fc⑀RI ␥-subunit was coprecipitated with Hu-BALB chimeric Fc␥RI (Fig. 6D, lane a) and also with Hu-NOD Fc␥RI (lane c), which contains only the membranespanning sequence and residual cytoplasmic tail of Fc␥RI-NOD. No material was precipitated from lysates from cells transfected with the Fc⑀RI ␥-subunit alone (Fig. 6D, lane e) or from samples incubated with FabЈ fragment-Sepharose (Fig.  6D, lanes b, d, and f). It is clear from these results that the Fc⑀RI ␥-subunit is able to associate with mouse Fc␥RI and that this association can take place in the presence of BALB/c-or NOD-derived membrane-spanning and cytoplasmic tail sequences. Thus, it is likely that the reduced expression of Fc␥RI-NOD is due to the mutations in the extracellular domains.
Specificity and Affinity of IgG Interactions with Fc␥RI Forms-Since mutations in the extracellular domains affected cell-surface expression but not Fc⑀RI ␥-subunit association, possible additional effects on ligand interactions were tested using both complexed and monomeric IgG (Table I). First, the binding of immune complexes was tested on Fc␥RI-NOD-and Fc␥RI-BALB-transfected COS-7 cells. Both Fc␥RI-NOD and Fc␥RI-BALB exhibited binding of complexed rabbit IgG-sensitized sheep erythrocytes assessed by rosetting, but could not bind complexed mouse IgG1-sensitized sheep erythrocytes, thereby confirming that the extracellular changes did not affect the specificity of mouse Fc␥RI. While the specificity of Fc␥RI-BALB and Fc␥RI-NOD was identical, repeated Scatchard analysis demonstrated that although both receptors bound monomeric IgG, Fc␥RI-NOD bound 125 I-mIgG2a or 125 I-hIgG1 with a 10-fold increase in affinity compared with Fc␥RI-BALB (Table I).
The reasons for the increased affinity were addressed by kinetic analysis of IgG binding (Fig. 7). The association of monomeric 125 I-hIgG1 with Fc␥RI-NOD or Fc␥RI-BALB showed identical kinetics, with 95% of maximum ligand bound by 30 min (4°C) (Fig. 7A). By contrast, the dissociation of monomeric 125 I-hIgG1 from Fc␥RI-NOD and Fc␥RI-BALB was remarkably different (Fig. 7B). Whereas IgG showed a biphasic dissociation from Fc␥RI-BALB with a rapid initial phase and a slower second phase, the dissociation from Fc␥RI-NOD was extremely slow. Only a single phase was apparent, with Ͻ10% of bound IgG dissociated from Fc␥RI-NOD after 45 min compared with ϳ75% from Fc␥RI-BALB. The differences in disso- a Complexed IgG binding by Fc␥RI-transfected COS-7 cells was assessed using sheep erythrocytes sensitized with either rabbit or mouse IgG, and binding was scored (ϩ) when 10 or more erythrocytes were bound by the COS-7 cell. Monomeric IgG binding was assessed by Scatchard analysis. b EA, erythrocyte coated with antibody; ND, not determined. ciation of IgG from Fc␥RI-NOD and Fc␥RI-BALB were observed at both 4 and 22°C.
Assessment of Expression and Dissociation Rate of Fc␥RI on NOD Macrophages-It was of some concern that the differences between Fc␥RI-BALB and Fc␥RI-NOD had been defined using a somewhat artificial transfection system. Thus, a series of experiments were performed to analyze Fc␥RI from macrophages derived from mice, and indeed, the results of the transfections were confirmed using macrophages. FACS analysis of expression of cell-surface markers and Fc␥RI was performed on BALB/c or NOD/Lt bone marrow-derived macrophages treated with IFN-␥ (to up-regulate mouse Fc␥RI). Using the 2.4G2 antibody, it was found that the expression of the low affinity IgG receptors (Fc␥RII/Fc␥RIII) was approximately the same on both BALB/c-and NOD/Lt-derived macrophages (Fig. 8A). Similarly, both BALB/c-and NOD/Lt-derived macrophages expressed nearly identical levels of a macrophage marker detected by antibody F4/80 (Fig. 8B). In contrast, NOD/Lt-derived macrophages bound less monomeric human IgG via Fc␥RI than did BALB/c-derived macrophages (Fig. 8C). Since no monoclonal anti-mouse Fc␥RI is available, Fc␥RI was detected by using monomeric human IgG in the presence of excess mAb 2.4G2 (which blocks any binding to Fc␥RII/Fc␥RIII) (Fig. 8C). The expression of Fc␥RI on NOD/Lt-derived macrophages was ϳ5-fold lower than that of BALB/c-derived macrophages with (see above) or without IFN-␥ (data not shown), indicating that the expression defect observed in the COS-7 transfection system is also observed in vivo on macrophages, although not to quite the same extent (10-fold versus 5-fold expression difference).
The slower dissociation of ligand from Fc␥RI-NOD expressed on COS-7 cells was also observed using NOD/Lt-derived macrophages (Fig. 9). As previously observed, monomeric human IgG dissociates quickly from BALB/c Fc␥RI and less quickly from NOD/Lt Fc␥RI. Treatment of the macrophages with or without IFN-␥ did not appear to influence the dissociation rate of monomeric IgG from macrophages from either strain of mouse (although Fc␥RI expression levels were altered (data not shown)). As noted previously with the expression phenotype of Fc␥RI from NOD/Lt-derived macrophages, the difference in the dissociation rate of IgG from Fc␥RI on NOD/Lt-and BALB/cderived macrophages was also not as dramatic as that seen in the COS-7 transfection system; however, the interaction of Fc␥RI on NOD/Lt-derived macrophages was still dramatically different than that of Fc␥RI on BALB/c-derived macrophages.

DISCUSSION
The mutations of Fc␥RI from the diabetes-prone NOD mouse have profound effects on receptor function. Whereas many protein mutations may result in loss of function, the Fc␥RI-NOD mutations result in a 10-fold increase in receptor affinity. Since artificially introduced mutations in mouse Fc␥RI are known to alter affinity and specificity (21), the binding of IgG to Fc␥RI-NOD and Fc␥RI-BALB was tested. The specificity of Fc␥RI-NOD and Fc␥RI-BALB was identical, with both receptors binding human IgG1, mouse IgG2a, and rabbit IgG, but not mouse IgG1 (Table I). The 10-fold increase in affinity was observed for both human IgG1 and mouse IgG2a. Through the use of chimeric receptors, the increased affinity of Fc␥RI-NOD for ligand could be attributed to mutations in the extracellular (ligandbinding) domains rather than to a secondary effect of the transmembrane region and residual cytoplasmic tail.
Previous studies of mouse Fc␥RI indicate that extracellular domain 3 is important in modulating the affinity and specificity of the receptor and that the first two domains interact with ligand with low affinity (21). It is interesting to note that there is a four-amino acid insertion between domains 2 and 3 of Fc␥RI-NOD. In addition, the Fc␥RI-BALB His 211 to Fc␥RI-NOD Tyr 216 mutation occurs in the body of domain 3, but is unlikely to affect binding as human Fc␥RI contains tyrosine in this position (see Fig. 1). Many of the amino acid changes found in Fc␥RI-NOD either are identical to normal human Fc␥RI or are conservative changes. However, the insertion of Glu 89 between domains 1 and 2 and the substitution of Asp 135 (Fc␥RI-BALB) for Gly 136 in Fc␥RI-NOD are potentially more disruptive. Indeed, the Asp 135 3 Gly 136 mutation falls within a region homologous to Ig-interactive regions of Fc␥RII (38 -40), Fc␥RIII (41), and Fc⑀RI (27). It is also possible, however, that the change in binding characteristics is due to a combination of the mutations.
In addition to the observed altered affinity, Fc␥RI-NOD was expressed at approximately one-tenth of the level of Fc␥RI-BALB in the COS-7 cell system. The mechanism of this difference remains unclear and could not be explained by the level of transient transfection efficiency or steady-state RNA amounts. Mouse Fc␥RI is able to associate with the Fc⑀RI ␥-subunit (Fig.  6); however, cotransfection of Fc␥RI-NOD and Fc⑀RI ␥-subunit cDNA does not rescue the expression of Fc␥RI-NOD. The region in the transmembrane domain thought to be involved in the Fc⑀RI ␥-subunit association is conserved between human Fc␥RI and Fc␥RI-NOD and was not expected to be disrupted by the truncation of the cytoplasmic tail of Fc␥RI-NOD. This was formally demonstrated when chimeric Fc␥RI mutants containing this sequence were shown to be able to associate with the Fc⑀RI ␥-subunit.
The mutations within the extracellular domains of Fc␥RI-NOD were shown to be involved in altering cell-surface expression as constructs containing Fc␥RI-NOD transmembrane and residual cytoplasmic tail domains were expressed at levels comparable to those of human Fc␥RI or Fc␥RI-BALB. However, it is not clear how the extracellular domain mutations can affect the surface expression and whether the mutations affect processes such as translation, the assembly of the polypeptide, transport to the cell surface, or even stability of the protein.
By measuring the level of surface expression using total saturable ligand binding, it is possible that the reduced expression of functional Fc␥RI-NOD is due to alterations in the receptor's configuration that allow only a small fraction (10%) of the total receptor pool to bind ligand rather than to a decrease in total receptor protein. Currently, this possibility cannot be directly tested as there are no mouse Fc␥RI-specific antibodies (monoclonal or polyclonal) to evaluate total cell-surface receptor levels. Nonetheless, it is clear that the mutations in the extracellular domains of Fc␥RI-NOD affect the levels of functional receptor on the cell surface.
Transfection of the Fc␥RI cDNAs into COS-7 cells was chosen primarily as a way to study mouse Fc␥RI in the absence of other Fc␥ receptors with which it is normally coexpressed (necessary because of the lack of monoclonal antibody reagents); however, the disadvantage of the system is that COS-7 cells may lack other proteins necessary for mouse Fc␥RI function. Indeed, although bone marrow-derived macrophages from NOD/Lt mice expressed less functional Fc␥RI on their surface (Fig. 8), the expression difference varied between 3-and 5-fold less than that of Fc␥RI from BALB/c-derived macrophages, not the reproducible 10-fold difference observed in the COS-7 system. This implies that other molecules, possibly even other Fc receptors, may play a role in the surface expression of functional Fc␥RI. Indeed, the dissociation of IgG from Fc␥RI on NOD/Lt-derived macrophages was again slower than that of Fc␥RI on BALB/c-derived macrophages; however, the effect was less dramatic than that seen in the COS-7 system, again pointing to other molecules being involved with Fc␥RI in macrophages. Despite this, however, the trend of higher affinity, slower dissociation, and lower expression was still observed in the NOD/Lt-derived macrophages.
The initial study of mutated Fc␥RI-NOD also demonstrated a lower surface expression of functional Fc␥RI on peripheral blood MAC-1 ϩ cells from NOD mice (19). This study reported that immune complexes bound to this receptor were still present on the surface of NOD-derived cells and not on C57BL/ 10SnJ-derived cells after incubation of the cells at 37°C for a few minutes. This observation indicates either that Fc␥RI from NOD mice was unable to internalize the complexes or that the immune complexes bound well and had not dissociated from the receptor on NOD cells. The fact that Fc␥RI-NOD demonstrated markedly reduced dissociation of ligand in the COS-7 system, demonstrated herein, directly supports the latter hypothesis.
The functional capacity of mouse Fc␥RI on cells from NOD mice must be questioned as it appears that the mutations in the extracellular domains lead to reduced expression of functional receptor, and receptors that are expressed exhibit higher affinity for monomeric ligand. Not only is ligand bound, but it does not appear to dissociate normally, implying that mouse Fc␥RI on NOD cells would be permanently saturated or perhaps continually signaling. The fact that Fc␥RI-NOD can associate with the Fc⑀RI ␥-subunit implies that signaling through this pathway may be intact. NOD mice, however, have numerous other defects, including aberrant protein kinase C function (42), and this, in conjunction with the Fc␥RI-NOD phenotype, may affect immune complex and antigen-immune complex handling. As signaling via human Fc␥RI in macrophages can lead to inflammatory mediator release such as interleukin-8 (43), interleukin-6 (44), and tumor necrosis factor (9), the role of dysfunctional mouse Fc␥RI in the exacerbation of the NOD mouse pathology requires further investigation.