A Novel Inhibitory Receptor (ILT3) Expressed on Monocytes, Macrophages, and Dendritic Cells Involved in Antigen Processing

THP1 is a human myeloid cell line. C1R and 721.221 are MHC class I–deficient, EBV-transformed human B cell lines. Jurkat is a human T cell line. All of these cells were grown in RPMI, 10% FCS. Monocytes were prepared from PBMCs by adhesion to plastic. In brief, PBMCs were isolated from whole blood by Ficoll-Hypaque density gradient, washed with RPMI, resuspended at 5 × 10⁶/ml in RPMI, 10% FCS and cultured in six-well plates for 1 h at 37°C. Nonadherent cells were removed by washing twice with RPMI. Adherent cells were cultured overnight in RPMI, 10% FCS supplemented with 50 ng/ml GM-CSF and 1,000 U/ml IL-4. Dendritic cells (DCs) were obtained from monocytes as previously described (32). Macrophages were obtained by culturing monocytes for 10 d in six-well plates at a concentration of 10⁶ cells/ml in RPMI, 20% human serum supplemented with 1 ng/ml macrophage CSF (M-CSF) (33, 34).

The 3′ end of ILT3 was cloned by rapid amplification of cDNA 3′ end (35). In brief, RNA was prepared from 721.221 and C1R cells as described (36) and single-strand cDNA was synthesized by reverse transcription of 1 μg of total RNA using Moloney murine leukemia virus reverse transcriptase and a (dT)₁₇ adaptor in a reaction volume of 20 μl. cDNA was amplified by two nested PCRs. The 5′ oligonucleotides were CCCCCAGCGACCCCCTGGA (nucleotide 926–944 of ILT2) and ACGGTGGCCTCAGGAGAGA (nucleotide 1003–1021 of ILT2) (25). In both PCRs, the 3′ primer corresponded to the adaptor primer. PCRs (50 μl volume) contained 1 μl of total cDNA sample, PCR primers (0.5 μM each), 2′-deoxynucleoside 5′-triphosphates (125 μM each), 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2.5 mM MgCl₂, 0.01% gelatin, and 0.25 units of Taq polymerase. PCR was carried out for 35 cycles, each consisting of 1-min steps at 94, 55, and 72°C. Amplified products were gel purified in 1.2% agarose gel, cloned into pCRII (Invitrogen, San Diego, CA), and sequenced. The 5′ end of ILT3 was cloned by two consecutive cycles of 5′ rapid amplification of cDNA ends (35).

To produce ILT3 as a soluble fusion protein, we constructed a chimeric gene consisting of ILT3 extracellular domains and human IgG1 constant regions. The cDNA fragment encoding the ILT3 extracellular region was amplified by PCR from cloned plasmid DNA. The 5′ primer (GATCGAATTCATGATCCCCACCTTCACGGCT) contained an EcoRI restriction site (italic) and the ILT3 start codon. The 3′ primer (CTGATCAAGCTTATACTTACCCTCCCAGTGCCTTCTCAGACC) provided a HindIII restriction site (italic), a splice donor sequence, and seven ILT3 codons preceding the transmembrane domain. The ∼800-bp PCR product was cut with EcoRI and HindIII, and ligated into an expression vector containing the exons for hinge, CH2 and CH3 region of human IgG₁, the guanosine phosphotransferase gene conferring resistance to mycophenolic acid, and the κ promoter for the expression in mouse myeloma (37). Transfections, selection of secreting transfectants, and purification of fusion protein from culture supernatants were performed as described (38).

10-wk-old, female BALB/c mice (Iffa-Credo, L'Arbresle, France) received an initial injection of 100 μg of ILT3 human IgG₁ fusion protein (ILT3-HuIgG₁), mixed 1:1 (vol/vol) with Alu-Gel-S (Serva Biochemicals, Paramus, NJ), behind the neck. 4 wk later, they were given a booster immunization with the same immunogen, followed after 2 wk by a final injection of 100 μg of purified ILT3-HuIgG₁. 3 d later, mice were killed and draining lymph node cells were isolated and fused with the myeloma fusion partner, Ag8.653, using polyethylene glycol 4000. Hybridoma supernatants were screened in two steps. First, an ELISA was performed using ILT3-HuIgG₁ in the coating step and human-adsorbed alkaline phosphatase–labeled goat anti–mouse IgG as secondary antibody. Supernatants from clones that were positive in ELISA were then tested by FACS^® analysis for staining C1R cells by flow cytometry.

ILT1, ILT2, and ILT3 cDNAs were subcloned into pCDNA3 (Invitrogen) and transfected into COS7 cells by lipofectin (GIBCO BRL, Gaithersburg, MD). Cell surface expression of ILTs was assessed 48 h after transfection by FACS^® analysis. A stable transfectant of ILT3 in Jurkat T cells was obtained as previously described (28).

Before staining, all the cells were preincubated with PBS, 20% human serum for 30 min on ice, to block Fc receptors. In two- and three-color staining experiments, cells were stained with the following primary mAbs: OKT3 (anti-CD3, IgG2a; American Type Culture Collection, Rockville, MD), CD16 (IgG2a; Milan Analytica AG, La Roche, Switzerland), 3C10 (anti-CD14, IgG2b; American Type Culture Collection), 1F5.4 (anti-CD20, IgG2a; American Type Culture Collection), M-T102 (anti-CD1a, IgG2b; PharMingen, San Diego, CA), L243 (anti-HLA-DR, IgG2a; American Type Culture Collection), and HB15a (anti-CD83, IgG2b; provided by Dr. T. Tedder, Duke University, Durham, NC). As secondary antibodies, we used human-adsorbed FITC- or PE-conjugated goat anti–mouse IgG1, IgG2a, and IgG2b (Southern Biotechnology Assoc., Inc., Birmingham, AL). In three-color stainings, biotin-labeled goat anti–mouse IgG2a was followed by streptavidin-allophycocyanin (SBA). Stained cells were analyzed by flow cytometry on a FACStar^® Plus (Becton Dickinson, Mountain View, CA) using the LYSYS II software.

Cells were surface labeled with 1 mCi of Na¹²⁵I using the sulfosuccinimidyl-3-(4-hydroxyphenyl)propionate method (39). For metabolic labeling with [³²P]orthophosphate, cells were incubated in phosphate-free DMEM (Sigma Chemical Co., St. Louis, MO) containing 10% Tris-buffered saline–dialyzed FCS for 1 h. [³²P]orthophosphate (Amersham Corp., Arlington Heights, IL) was then added at 0.5 mCi/ml and incubation was continued for 5 h. After surface or metabolic labeling, cells were lysed in 1% Triton X-100, 100 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM PMSF, 10 μg/ml aprotinin, and 10 μg/ml leupeptin. After overnight preclearing with protein G–sepharose, lysates were incubated with culture supernatants of either ZM3.8 (anti-ILT3, IgG₁) or control IgG1 mAbs (5.133, anti-NKAT4 [40] and 1B7.11, anti-2,4,6 TNP, American Type Culture Collection) at +4°C for 4 h, and immune complexes were precipitated by addition of protein G–Sepharose for 1 h at +4°C. Precipitates were washed 3 times with lysis buffer, followed by a final wash with 10 mM Tris-HCl, pH 7.4, and resuspended in nonreducing or reducing sample buffer. SDS-PAGE analysis was performed according to a standard procedure. After the run, gels were dried and exposed to autoradiography film (Amersham Corp.) for 2–5 d.

Monocytes and macrophages were loaded with Indo-1 AM (Sigma Chemical Co.) as described (41). In stimulation experiments, anti-CD11b mAb (44aabc; American Type Culture Collection) and anti-CD16 mAb (B73.1; American Type Culture Collection) were added to monocytes and macrophages, respectively, followed by a F(ab′)₂ goat anti–mouse IgG (Milan Analytica) as cross-linker. Cells were then analyzed on a flow cytofluorimeter (FACS^® Vantage or Coulter Elite Enhanced Sort Performance [Coulter Corp., Hialeah, FL]) to detect Ca²⁺ fluxes. Stimulation of monocytes with the anti-HLA-DR mAb 3.8B1 (Cella, M., and A. Lanzavecchia unpublished data) was performed without a cross-linker. In inhibition experiments, cells were preincubated for 10 min at room temperature either with ZM3.8 mAb, or with the anti–MHC class I W6/32 mAb. After washing with PBS, either anti-CD11b, anti-HLA-DR, or anti-CD16 mAbs were added followed by the cross-linking antibody, and cells were analyzed by flow cytometry to detect Ca²⁺ fluxes. Only live (based on forward scatter criteria) and Indo-1–loaded cells (based on 405 nM versus 525 nM emission spectra) were included in the analysis.

For antiphosphotyrosine blots, monocytes (2 × 10⁶) were incubated for 2 min at 37°C with medium, ZM3.8 mAb, 3.8B1 mAb, or with both mAbs in the absence or in the presence of a secondary cross-linker. Cells were washed in icecold PBS and lysed as previously described (20). Cell lysates were boiled, separated on SDS-PAGE, transferred to nitrocellulose, and probed with horseradish peroxidase (HRP)-coupled 4G10 (500 ng/ml; Upstate Biotechnology Inc., Lake Placid, NY). Immunoblotted proteins were visualized by chemiluminescence using the enhanced chemiluminescence detection reagents (Amersham Corp.). For anti-SHP blots, 10⁷ monocytes were plated on six-well tissue culture plates previously coated with purified ZM3.8 or 5.133 mAbs at 50 μg/ml, centrifuged for 2 min, shifted to 37°C for 2 min, collected in cold PBS, 0.5 mM EDTA, and washed twice. Identical number of cells stimulated with ZM3.8 or 5.133 mAbs were lysed and immunoprecipitated with ZM3.8 and protein G as described above. Precipitated samples were boiled for 5 min, separated by SDS-PAGE, and transferred to nitrocellulose (Hybond-C extra; Amersham Corp.). Membranes were blocked with PBS, 3% nonfat, dry milk and probed anti-SHP-1 rabbit antiserum (Santa Cruz Biotechnology, Santa Cruz, CA) followed by HRP-labeled goat anti–rabbit Ig (SBA). Immunoblotted proteins were visualized by chemiluminescence. Membranes were stripped of antibody by incubating for 30 min at 50°C in 100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM TrisHCL, pH 6.7, according to the protocol supplied by Amersham Corp. Stripped blots were reprobed after blocking with antiSHP-2 rabbit antiserum (Santa Cruz Biotechnology).

Monocytes were cultured for 24–48 h in culture medium containing GM-CSF and IL-4, as described above. One aliquot of monocytes was fixed for 1 min at room temperature in RPMI, 0.05% glutaraldehyde. Another aliquot was used without fixation. Both samples were stained with ZM3.8 mAb for 40 min on ice, washed twice with ice-cold PBS, 1% FCS, and subsequently stained with biotin-labeled F(ab′)₂ goat anti–mouse IgG for 1 h on ice. After washing with ice-cold PBS, 1% FCS, cells were incubated at 37 °C in prewarmed RPMI, 10% FCS for various times to allow internalization. Cells were then cooled on ice, washed once with PBS, 0.1% NaN₃, stained with PE-conjugated streptavidin for 30 min on ice, washed, and analyzed by FACS^®. As a measure of internalization in nonfixed cells, we used the percental decrease of the median fluorescence intensity (MFI) as compared to control samples kept at 4°C. The percentage decrease of MFI observed in fixed cells was taken as a measure of the off-rate of the antibody at 37°C.

4 × 10⁴ irradiated monocytes (3,000 rad) were co-cultured with 8 × 10⁴ cells of the B13 T cell clone (42) in 96-well flat-bottom microplates in the presence of serial dilutions of IgG1 mAbs. mAbs used in the assay were the following: ZM3.8 (anti-ILT3, IgG₁) or 5.133 (anti-NKAT4 KIR, IgG₁), 1B7.11 (anti-2,4,6 TNP, IgG₁; American Type Culture Collection), and 32.2 (anti-FcγRI, IgG₁; American Type Culture Collection). After 48 h, the cultures were pulsed with [³H]thymidine (1 μCi/well, specific activity 5 Ci/mmol), and the radioactivity incorporated was measured after additional 16 h. The data were plotted against the concentration of mAbs determined by ELISA using a purified mouse IgG₁ (anti-CD68, Dako Corp., Carpinteria, CA) as a standard.

During the process of cloning the 3′ end of ILT2 cDNA from EBV-B cell lines (25), we obtained an additional cDNA sequence that differed from ILT2 by several amino acid residues. Amplification of the 5′ end of this fragment yielded a new cDNA, called ILT3, which contains a single open reading frame encoding a transmembrane protein of 447 amino acids with a predicted molecular mass of ∼47 kD (Fig. 1). The amino acid sequence begins with a hydrophobic signal peptide of 23 amino acids followed by an extracellular region composed of two C2 type Ig-SF domains (43). Each domain shows two characteristic cysteines that are 49 and 50 residues apart, flanked by conserved residues (Val-x-Leu/Ile-x-Cys and His/Tyr-x-Gly-x-Tyr-x-Cys-Tyr/Phe, respectively, where x is any amino acid). The putative transmembrane domain of ILT3 consists of 21 amino acids, followed by a long cytoplasmic region of 167 amino acids, which is characterized by the presence of one Tyr-x-x-Val motif followed by two Tyr-x-x-Leu motifs spaced by 26 amino acid residues. These Tyr-x-x-Leu pairs and their spacing are reminiscent of the Tyr-x-x-Leu motifs identified in KIRs as binding sites for protein tyrosine phosphatase SHP-1 (19–23).

Comparison of the predicted amino acid sequence of ILT3 with Ig-SF protein sequences, revealed that ILT3 is closely related to ILTs (64% similarity with ILT2), bovine Fcγ2R (45%), murine cell surface antigen gp49 isoforms (39–43%), and, to a lesser extent, to KIRs (23–32%) and human FcαR (28%). Genomic DNA analysis of human– hamster hybrid cell lines, each with a different partial complement of human chromosomes, showed hybridizing bands only in samples containing human chromosome 19 (data not shown). Thus, ILT3 belongs to the ILT family which maps to human chromosome 19, as do the FcαR and the KIR genes. Expression of ILT3 by RNA blot analysis revealed a major transcript of ∼1.6 kb (data not shown). Interestingly, this transcript was weakly expressed in the B cell lines C1R or 721.221, from which ILT3 was initially cloned, whereas it was predominantly expressed in monocytes purified from PBMCs and in the myeloid cell line THP1. No transcripts were detected in the T cell line Jurkat.

To characterize the ILT3-encoded cell surface molecule, we produced anti-ILT3 mAbs using a soluble ILT3HuIgG₁ fusion protein as an immunogen. The ZM3.8 mAb specifically recognized ILT3 on transiently transfected COS cells (data not shown). In PBMCs, ZM3.8 mAb stained only CD14⁺ monocytes, whereas no staining was observed in CD3⁺ T cells, CD16⁺ NK cells, or CD20⁺ B cells (Fig. 2,A). Interestingly, in monocyte-enriched populations derived from peripheral blood, ZM3.8 also stained CD83⁺ cells (Fig. 2,A), as well as CD14-/HLA-DR^high cells (Fig. 2,B). Both of these cell populations correspond to circulating primary DCs (44–46). In addition, ILT3 was expressed by both DCs and macrophages derived from monocytes cultured with GM-CSF plus IL-4 or M-CSF, respectively (Fig. 2, C and D; 32–34).

We also tested ILT3 expression in several cell lines. Among four myelo-monocytic cell lines (U937, Monomac6, THP1, and HL-60), only THP1 reacted with ZM3.8. All tested EBV-B cell lines (n = 10) were negative, with the exception of C1R. No cell surface expression was detected either on purified fresh B cells or on B cells activated by CD40L in the presence of IL-4, IL-2, and IL-10 (data not shown). Finally, ZM3.8 did not stain neutrophils (data not shown). Thus, ILT3 is selectively expressed on myeloid APCs.

To define the biochemical characteristics of ILT3, we carried out immunoprecipitations from ILT3-transfected Jurkat T cells and monocytes (Fig. 3). A prominent band of ∼55 kD in ILT3-transfected T cells was detected under nonreducing and reducing conditions, while a ∼58–60 kD band was detected in monocytes (Fig. 3,A) and in THP1 (data not shown). N- and O-deglycosylation of immunoprecipitates did not produce any change in the size of the bands. After labeling cells with [³²P]orthophosphate, ILT3 was immunoprecipitated as a constitutively phosphorylated molecule (Fig. 3 B). The discrepancy in apparent molecular mass of ILT3 between ILT3-transfected T cells and monocytes may be due to a different degree of phosphorylation in distinct cell types.

The presence of putative ITIMs suggested an inhibitory role of ILT3 on APC activation. To test this hypothesis, we triggered monocytes and macrophages with mAbs that have been shown to induce Ca²⁺ mobilization, such as anti-CD11b (47, 48), antiHLA-DR (49–51), and anti-FcγRIII (CD16; 52) mAbs, and tested whether recruitment of ILT3 to the triggering receptors inhibits Ca²⁺ mobilization. As shown in Fig. 4,A, ligation of surface CD11b with a specific mAb followed by a secondary cross-linking antibody elicited a rapid and transient rise in intracellular calcium concentration ([Ca²⁺]_i). On the contrary, preincubation of monocytes with ZM3.8 mAb, followed by addition of anti-CD11b and co-crosslinking of CD11b and ILT3 by a secondary antibody, resulted in complete inhibition of the [Ca²⁺]_i increase (Fig. 4 D). In control experiments, the [Ca²⁺]_i increase induced by CD11b was not affected by co-ligation of CD11b with MHC class I. Also, cross-linking of ILT3 alone did not evoke Ca²⁺ mobilization, indicating that ILT3 is not a stimulatory receptor (data not shown).

It has been reported that some anti–MHC class II mAbs can trigger functional responses in B lymphocytes (49–51). We have recently developed an anti-HLA-DR mAb, called 3.8B1, which triggers a rapid rise in [Ca²⁺]_i, even in the absence of a secondary cross-linking antibody (Fig. 4,B). Co-ligation of ILT3 and HLA-DR resulted in significant inhibition of [Ca²⁺]_i increase (Fig. 4,E). Interestingly, HLA-DR–mediated stimulation of monocytes was not affected by preincubation of monocytes with anti-ILT3 antibody in the absence of the secondary cross-linking antibody, indicating that recruitment of ILT3 to the stimulatory receptor is necessary to block the [Ca²⁺]_i increase (Fig. 4 G).

The low affinity receptor for IgG, FcγRIII, is expressed on NK cells and macrophages and has been shown to trigger Ca²⁺ mobilization in NK cells (52, 53). Similarly, we showed that cross-linking of FcγRIII in macrophages evokes a strong increase of [Ca²⁺]_i (Fig. 4,C). Despite the low expression of ILT3 on macrophages (see Fig. 2,D), co-ligation of ILT3 and FcγRIII inhibited the [Ca²⁺]_i increase, especially at early time points (Fig. 4 F). Together, these results demonstrate that co-ligation of ILT3 with a triggering receptor results in signal extinction.

To test whether inhibition of Ca²⁺ mobilization responses was paralleled by inhibition of tyrosine phosphorylation, we performed antiphosphotyrosine immunoblotting on whole cell lysates of monocytes stimulated via HLA-DR in the absence or in the presence of co-ligation with ILT3. As shown in Fig. 5 A, monocytes triggered via HLA-DR displayed a substantial increase of tyrosine phosphorylation (lane 3), as compared to monocytes incubated with medium alone (lane 1) or with anti-ILT3 mAb (lane 2). Co-ligation of ILT3 with HLA-DR resulted in a dramatic reduction of the intensity and number of tyrosine phosphorylated species (lane 5). In the absence of secondary cross-linking antibody, ILT3 did not block HLA-DR–triggered tyrosine phosphorylation increase (lane 4).

Since in NK cells, KIRs downregulate calcium mobilization and tyrosine phosphorylation by recruiting SHP-1 and SHP-2 (19–23), we tested whether these phosphatases could be also involved in the negative signaling pathway mediated by ILT3. ILT3 was cross-linked on monocytes by ZM3.8 mAb attached to the plastic surface of tissue culture plates and subsequently immunoprecipitated from cell lysates. In control experiments, ILT3 was immunoprecipitated from unstimulated cells. Immunoblotting of immunoprecipitates with anti-SHP-1 antibodies demonstrated association of SHP-1 with ILT3 that significantly increases upon ILT3 cross-linking (Fig. 5 B). On the contrary, no association with SHP-2 was detected (data not shown). These results implicate SHP-1 as a cytosolic component of the signal extinction mediated by ILT3.

Since ILT3 is selectively expressed on APCs and displays putative tyrosine-x-x-valine/leucine internalization motifs in the cytoplasmic tail (54–56), we analyzed the ability of ILT3 to endocytose and deliver its ligand to an antigen-processing and loading compartment. As shown in Fig. 6, ZM3.8 mAb bound at 0°C to monocytes was not internalized when the cells were shifted to 37°C. We therefore examined whether internalization occurs in the presence of a secondary cross-linker. ZM3.8 mAb was bound on ice to monocytes and cross-linked by a biotinylated F(ab′)₂ secondary antibody. Cells were subsequently warmed to 37°C to allow internalization. After various time points, cells were returned to ice and the amount of ZM3.8 mAb remaining on the cell surface was determined by flow cytometry using PE-conjugated streptavidin. Freshly isolated monocytes efficiently internalized bound ZM3.8 mAb, as demonstrated by a 60–80% reduction after 30 min at 37°C and by a complete disappearance after 60–90 min (Fig. 6). In control experiments, no significant decrease of surface-bound ZM3.8 was observed using fixed monocytes, ruling out a detachment of the primary or secondary antibodies from the cell surface at 37°C (Fig. 6). In addition, when ¹²⁵I-labeled streptavidin was added as a third step reagent before incubation at 37°C, no significant decrease of cell-bound radioactivity was observed over time, thus excluding shedding of ILT3 from the cell surface (data not shown).

To further investigate a possible role of ILT3 in antigen capture, we evaluated the ability of monocytes to present ZM3.8 mAb to a CD4⁺ class II–restricted T cell clone specific for a mouse IgG1 peptide epitope (42). The presentation of the ZM3.8 mAb was compared to that of IgG1 mAbs that bind to other receptors (anti-FcγRI mAb) or do not bind to surface molecules and are taken up in the fluid phase [anti-TNP and anti-NKAT4 KIR mAbs]. As shown in Fig. 7, monocytes presented ZM3.8 mAb to T cells 50– 100-fold more efficiently than the anti-FcγRI mAb, and 400–500-fold more efficiently than anti-TNP and antiNKAT4 KIR mAbs. Together, these experiments indicate that ILT3 can efficiently deliver its ligand to an intracellular compartment where class II loading can occur.

Previous work has extensively shown that inhibitory receptors regulate cell activation in B, NK, mast cells, and subsets of T cells when co-ligated with ITAM-containing stimulatory receptor complexes. Our data extend this concept to professional APCs, showing that ILT3 negatively regulates APC functional responses triggered via stimulatory receptors, such as CD11b, CD16, and even MHC class II. In addition to its inhibitory function, ILT3 is involved in antigen uptake. It is noteworthy that ILT3 is expressed on DCs. These are a unique type of leukocytes whose primary function is to efficiently capture antigens, process them, and present them to naive T cells (57). Although the uptake and processing of antigen is a major function of DCs, only a few mechanisms and receptors specialized for antigen capture have been identified on DCs (58–60). Our data indicate that ILT3 is a unique receptor expressed in DCs, able to target the ligand into processing and peptide-loading compartments. Thus, ILT3 displays a dual function of inhibitory receptor and antigen-capturing molecule. A similar dual function has been previously shown for the low affinity receptor for IgG, FcγRIIB (7, 8).

The ligand of ILT3 is still unknown. ILT3 is closely related to bovine Fcγ2R and, to a lesser extent, to human FcαR, suggesting that ILT3 may be an Fc receptor. However, we did not detect any significant binding of human monomeric and heat-aggregated IgM, IgG, IgA, and IgE to ILT3-transfected Jurkat T cells (data not shown). ILT3 also shows homology to KIRs, which suggests that it may be a receptor for MHC class I molecules. We have investigated this possibility by analyzing binding of soluble ILT3HuIgG₁ fusion protein to MHC class I transfectants in the class I–deficient mutant 721.221. No significant binding could be detected by FACS^® analysis under the same experimental conditions in which KIR-HuIgG fusion proteins bind to MHC class I molecules (data not shown). We are investigating the possibility that ILT3 is a receptor for cell surface molecules related to MHC class I molecules, such as nonclassical class I molecules (CD1 [61], MR1 [62], MIC [63]), for MHC class II molecules, or for serum factors such as complement components.

ILT3 is a member of the ILT subfamily of Ig-SF molecules, which is located on chromosome 19, most likely in close linkage with KIRs. Another member of this family, ILT2, is characterized by similar cytoplasmic tyrosine-based motifs such as ILT3, and is expressed in B cell lines, suggesting that this may be a putative inhibitory resceptor in B cells. We have also cloned additional cDNAs which encode novel molecules homologous to ILT2 and ILT3 (ILT4 and ILT5, GenBank accession numbers AF000574 and AF000575). Thus, the number of Ig-SF inhibitory receptors with different tissue distribution and specificity is likely to increase.

We thank S. Bahram, K. Campbell, and R. Torres for critical reading of the manuscript.