Identification and characterization of a novel human myeloid inhibitory C-type lectin-like receptor (MICL) that is predominantly expressed on granulocytes and monocytes.

Inhibitory and activatory C-type lectin-like receptors play an important role in immunity through the regulation of leukocytes. Here, we report the identification and characterization of a novel myeloid inhibitory C-type lectin-like receptor (MICL) whose expression is primarily restricted to granulocytes and monocytes. This receptor, which contains a single C-type lectin-like domain and a cytoplasmic immunoreceptor tyrosine-based inhibitory motif, is related to LOX-1 (lectin-like receptor for oxidized low density lipoprotein-1) and the beta-glucan receptor (Dectin-1) and is variably spliced and highly N-glycosylated. We demonstrate that it preferentially associates with the signaling phosphatases SHP-1 and SHP-2, but not with SHIP. Novel chimeric analyses with a construct combining MICL and the beta-glucan receptor show that MICL can inhibit cellular activation through its cytoplasmic immunoreceptor tyrosine-based inhibitory motif. These data suggest that MICL is a negative regulator of granulocyte and monocyte function.

The functional balance of the immune system is regulated, in part, by inhibitory and activatory receptors found on all leukocytes as well as on many non-immune cells (1). Within these activatory and inhibitory receptor families, cytoplasmic consensus motifs have been identified and shown to associate with particular signaling mechanisms. These motifs include the immunoreceptor tyrosine-based activation motifs (ITAMs), 1 which become phosphorylated upon stimulation, sending their activatory signals through a variety of intracellular enzyme cascades and resulting in a wide range of cellular outcomes, and the immunoreceptor tyrosine-based inhibitory motifs (IT-IMs), which suppress such activatory signals by recruiting phosphatase enzymes that dephosphorylate ITAMs as well as other components of the activatory pathways.
The array of inhibitory and activatory receptors expressed by natural killer cells has recently been the focus of much investigation (2). In humans, these receptors can be divided into two families: the Ig superfamily, including the killer cell Ig-like receptors (3) that bind classical major histocompatibility complex class I or HLA G (non-classical), and the C-type lectin-like family (4), including the NKG2 receptors, which bind the nonclassical HLA E or other distantly related molecules that are up-regulated on stressed, transformed, or infected cells (5,6).
The C-type lectin and lectin-like proteins are classified into 14 groups based on the arrangement of their C-type lectin-like domains (CTLDs). 2 In particular, group V receptors include the lectin-like NKG2 family mentioned above, but also a distinct subgroup of these receptors that are expressed predominantly on myeloid and endothelial cells (7). The members of this subgroup appear to bind diverse ligands, as exemplified by LOX-1 (lectin-like receptor for oxidized low density lipoprotein-1) (8) and the ␤-glucan receptor (BGR), which is an activatory phagocytic receptor for ␤-glucans (9). 3 Although they share structuralhomologywithclassicalcarbohydrate-bindingC-type(Ca 2ϩdependent) lectins (10), the group V molecules are termed C-type lectin-like because they lack the residues involved in Ca 2ϩ coordination (11).
Here, we describe the identification and characterization of a myeloid inhibitory C-type lectin-like receptor (MICL), a novel group V C-type lectin-like receptor that contains an ITIM in its cytoplasmic tail and is most homologous to the LOX-1/BGR subgroup. We demonstrate that it is expressed at the cell surface and is highly N-glycosylated and that its expression is restricted to polymorphonuclear leukocytes (PMNs) and monocytes. We also show that it associates with the signaling phosphatases Src homology-2 domain-containing tyrosine phosphatase (SHP)-1 and SHP-2 and that it can inhibit cellular activation.

EXPERIMENTAL PROCEDURES
Cell Lines and Growth Conditions-NIH3T3 fibroblasts, RAW264.7 monocytes, and the PT67 and HEK293T-based Phoenix ecotropic retroviral packaging cell lines were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% heat-inactivated fetal calf serum, 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 2 mM L-glutamine. CHO-K1 cells were grown in Ham's F-12 nutrient mixture, and CHO-Lec1 cells were grown in ␣-minimal essential medium with ribonucleosides and deoxyribonucleosides, both with the supplements described above. All cell lines were obtained from the cell bank of the Sir William Dunn School of Pathology, for except CHO-Lec1 (American Type Culture Collection CRL-1735) (12,13), Phoenix ecotropic (a gift from Dr. Gary Nolan, Stanford University), and PT67 (Clontech).
In Silico Analysis-Sequence analyses were performed using several on-line tools. 4 Sequences were aligned with ClustalX (14); dendrogram analysis and percentage identities were calculated by DNAMAN Version 4.0 (Lynnon BioSoft).
Cloning and Generation of Stable Cell Lines-All routine nucleic acid manipulation techniques were performed essentially as described by Sambrook et al. (15). The complete MICL open reading frame was originally cloned into the pBluescript SK II ϩ vector (Stratagene) from human peripheral blood mononucleocyte cDNA using the Advantage HF2-PCR kit (Clontech) with primers 5Ј-AAAGGATCCTCTTTA-CATATTCATCAATG-3Ј and 5Ј-AAACTCGAGACACTCCTTAAATG-TATTTG-3Ј. Hemagglutinin (HA) and V5 epitope tags (used as described below) were generated using adaptor duplex oligonucleotides as described previously (16). 5 All tags were inserted at the 3Ј-end of each sequence, corresponding to the C-terminal extracellular domain of each receptor. The MICL/BGR chimera was generated using overlap extension PCR (17) with primers 5Ј-CTACAACTGATGAGTAACTTCCTAT-CAAGAAATAAAG-3Ј and 5Ј-TTTATTTCTTGATAGGAAGTTACTCAT-CAGTTGTAG-3Ј such that the MICL 5Ј-sequence ran into the murine BGR sequence after 95 amino acids, translating as MICL88 NISLQLMS MICL95 / BGR83 NFLSRNKEN BGR91 . PCR was used to generate cytoplasmic mutant constructs from the V5-tagged full-length MICL/BGR chimera, including the truncated mutant, in which the 5Ј-end of the gene encoding the cytoplasmic tail (amino acids 1-39) was replaced with a start codon (AUG) and a Kozak sequence, and the ITIM mutant, in which the ITIM (VTYADL, amino acids 5-10) was mutated to VTFADL to inactivate the signaling domain. A schematic of all constructs (without epitope tags) that were used in these experiments is presented in Fig. 1. The fidelity of all clones was confirmed by sequencing.
To obtain stable cell lines, constructs were subcloned into the pFBneo (Stratagene) or pMXs-IP (a gift from Professor Toshio Kitamura, University of Tokyo) (18) retroviral vector; packaged into virions using HEK293T-based Phoenix ecotropic or PT67 cells as described previously (19); and transduced into NIH3T3, CHO-K1, CHO-Lec1, or RAW264.7 cells. RAW264.7 and CHO-K1 cells were pretreated with 0.2 g/ml tunicamycin to increase transduction efficiency. All cell lines were used as non-clonal populations to reduce any founder effects and were generated and tested at least twice to confirm their phenotype. Stable cell lines were selected and maintained in 0.6 mg/ml Geneticin or 3 g/ml puromycin.
RNA Blot Analysis and Reverse Transcription (RT)-PCR-Northern blotting was performed using a full-length MICL cDNA probe as described previously (19,20). RNA isolation and RT-PCR were also performed as described previously (19), except that random hexamer primers were used for first-strand cDNA synthesis. Leukocytes were isolated from buffy coats (Bristol Blood Donor Services) and separated into adherent and non-adherent mononuclear cells using Ficoll-Paque TM Plus (Amersham Biosciences) and from fresh peripheral blood and separated into PMNs and total mononuclear leukocytes using Polymorphprep TM (Axis-Shield, Oslo, Sweden) according to the manufacturers' protocols.
Fluorescent Zymosan Binding and Internalization and Tumor Necrosis Factor-␣ (TNF-␣) Assays-Fluorescence-based binding assays using fluorescein isothiocyanate (FITC)-labeled zymosan (Molecular Probes, Inc.) were performed as follows. NIH3T3 transfectants were plated at 2 ϫ 10 5 cells/well in 24-well plates the day prior to each experiment. The cells were washed three times; 100 g/ml laminarin (Sigma) was added when appropriate; and the cells were incubated for 20 min at 37°C to allow inhibition of BGR (21). Following the addition of FITClabeled zymosan (25 particles/cell), the cells were incubated at 37°C in 5% CO 2 for 60 min. After washes to remove unbound particles, the cells were lysed in 3% Triton X-100, and the amount of FITC-labeled zymosan bound by the cells was quantified using a Titertek Fluoroskan II (Labsystems Group Ltd.) as described (21).
For the internalization assays, NIH3T3 transfectants were plated at 2 ϫ 10 5 cells/well in 12-well plates the day prior to each experiment. At the start of the assay, the cells were washed three times with culture medium, and when necessary, cytochalasin D (1 M; Sigma) was added to the cells 40 min prior to the start of the assay and then maintained throughout the experiment. Following the addition of FITC-labeled zymosan (5 particles/cell), the cells were incubated at 37°C in 5% CO 2 for 30 min, washed to remove unbound particles, and then incubated for a further 90 min at 37°C. External zymosan was stained with rabbit anti-zymosan antibody (Molecular Probes, Inc.) after blocking with phosphate-buffered saline (PBS) containing 1% bovine serum albumin and 5% heat-inactivated goat serum and then detected with allophycocyanin-conjugated goat anti-rabbit antibody (Molecular Probes, Inc.). The cells were lifted and fixed in 1% formaldehyde, and flow cytometry analysis (carried out according to conventional protocols) was performed by gating on the FITC-positive cell populations, which had bound zymosan. The percentage of phagocytosis was determined by comparing the allophycocyanin-negative (internalized particles) and allophycocyanin-positive (non-internalized particles) cell populations.
TNF-␣ production in RAW264.7 transfectants stimulated with zymosan was determined as described previously (16). After taking samples for anti-TNF-␣ enzyme-linked immunosorbent assay, the amount of FITC-labeled zymosan bound by the cells was quantified as described above. All graphical data are presented as the means Ϯ S.D. from a representative experiment.
Immunoprecipitation-To identify MICL-associated phosphatases in RAW264.7 transfectants, cells were washed twice with PBS and resuspended in PBS containing 0.5% bovine serum albumin and 2 mM NaN 3 . Aliquots containing 5 ϫ 10 7 cells were incubated for 2 h at 4°C with 5 g of anti-HA antibody. Excess antibody was then removed by washing, and the cells were resuspended in 2.25 ml of PBS and prewarmed to 37°C for 3 min before stimulation with 250 l of pervanadate (or PBS-only control), followed by incubation at 37°C for 10 min. Pervanadate (22) was prepared from 1 mM activated Na 3 VO 4 6 and 10 mM H 2 O 2 (final concentrations). Cells were lysed in Nonidet P-40 buffer (1% Nonidet P-40, 150 mM NaCl, 10 mM EDTA, 10 mM NaN 3 , 10 mM Tris-HCl (pH 8), 2 mM Na 3 VO 4 , 10 mM NaF, and complete EDTA-free protease inhibitors (Roche Applied Science)) and incubated on ice for 45 min. Nuclei were pelleted at 12,000 ϫ g for 20 min at 4°C. Supernatants were added to 5 l of anti-mouse Dynabeads® (precoated with 2% bovine serum albumin, Dynal Biotech) and incubated overnight at 4°C. Beads were then washed four times with Nonidet P-40 buffer, and precipitated proteins were denatured by boiling for 5 min in reducing sample buffer. Aliquots were subjected to SDS-10% PAGE, and proteins were transferred to Hybond-C Extra membranes (Amersham Biosciences) for 90 min at 100 V. Proteins were probed with primary antibody in PBS containing 2% skimmed milk powder for 2 h at room temperature, followed by peroxidase-conjugated secondary antibody for 60 min at room temperature and detection using ECL substrate (Amersham Biosciences). The primary antibodies used were mouse anti-HA (HA.11, Covance), rabbit anti-SHP-1 (Upstate Biotechnology, Inc.), mouse anti-SHP-2 (clone 79, BD Biosciences) and rabbit anti-Src homology-2 domain-containing inositol phosphatase (SHIP) (a gift from Dr. Gerry Krystal).

MICL Is a C-type Lectin-like Transmembrane
Receptor with a Cytoplasmic ITIM-We identified a novel human immunoreceptor that we named MICL ( Fig. 2A) through data base searches for molecules homologous to BGR. The MICL gene consists of six exons and is located on the (ϩ)-strand of chromosome 12 (locus 12p13.31) within the natural killer gene complex (23), spanning ϳ14 kb on GenBank TM /EBI genomic contig NT_009714 (Fig. 2B). It is positioned at the telomeric end of a cluster of homologous genes that include LOX-1, BGR, CLEC-1, and CLEC-2 ( Fig. 2C) (7). The predicted open reading frame encodes a 265-amino acid type II transmembrane polypeptide with an expected mass of ϳ31 kDa and comprises one CTLD, a stalk/neck region, a single transmembrane domain, and a cytoplasmic tail containing an archetypal ITIM (Figs. 1A and 2A). The CTLD contains the six canonical cysteines, present in almost all C-type lectins; and in the stalk, there are two extra cysteines that may allow homo-or heterodimerization (24). The ectodomain also contains several putative N-glycosylation sites and one predicted O-glycosylation site. 7 Two single nucleotide polymorphisms in the coding region have been identified (at 255 and 790 bp in Fig. 2A), although only the latter one results in an amino acid substitution (Q244K). The Lys 244 allele was previously deposited in the GenBank TM /EBI Data Bank under accession number NM_138337, but its functional significance is not yet known.
MICL Shares Homology with LOX-1, BGR, CLEC-1, and CLEC-2 as Well as Other Inhibitory Receptors-MICL is most homologous to other human C-type lectin-like molecules on chromosome 12 that are found in the same cluster (Fig. 2C), including two novel transcripts with GenBank TM /EBI acces- sion numbers AY358265 and AY358810 (Fig. 3A). The predicted MICL CTLD sequence is most similar to those of AY358265, LOX-1, and BGR (Fig. 3B). Alignments of the CTLDs of MICL and related proteins show that they all contain the six conserved cysteines, but lack the Ca 2ϩ -binding residues that are present in classical C-type lectins such as rat MBP-A (Fig. 3A). Furthermore, none of the known sugar-binding motifs present in classical C-type lectins, e.g. EPN/QPD (11), are found in MICL. Analysis of the MICL cytoplasmic tail revealed homology to other ITIM-containing group V C-type lectin-like molecules (Fig. 3C) such as AY358810, CD72, and murine Ly49C. The ITIM is also similar to those of certain better characterized type I transmembrane receptors, viz. signal-regulatory protein-␣ (SIRP-␣), ILT5, KIR3DL1, and KIR2DL4 (data not shown).
Recently, the murine ortholog of MICL was identified (Gen-Bank TM /EBI accession number XM_149798). This gene is located within the equivalent gene cluster on the syntenic region of mouse chromosome 6. The encoded protein has 49% overall identity (ϳ70% similarity) to human MICL and shares many of the features mentioned above, e.g. exon structure, N-glycosylation sites, and extra cysteines in the stalk. Similar MICLrelated sequences are also present in other vertebrate genomes, including Rattus norvegicus, Bos tauros, and Danio rerio (data not shown).
MICL Has at Least Three Alternatively Spliced Transcripts, and Expression Is Restricted to PMNs and Monocytes-We examined the expression of human MICL by northern blotting and RT-PCR analysis. By northern blotting using a full-length cDNA probe, we were able to detect MICL transcripts in several tissues, especially those rich in leukocytes. One predominant transcript of 1.5 kb was observed in positive tissues, as were several transcripts of greater length. High levels of MICL transcripts were detected in peripheral blood leukocytes and bone marrow, with lower levels in spleen, fetal liver, heart, colon, placenta, and lung (Fig. 4A) as well as testis (data not shown). Blotting with an alternative MICL cDNA probe and increased hybridization stringency gave identical results, thus confirming the specificity of the hybridizing bands (data not shown).
We then determined the peripheral blood cell populations expressing MICL by RT-PCR and detected high levels of MICL transcripts in both PMNs and total mononuclear leukocytes (Fig. 4B). We separated the mononuclear leukocyte populations into non-adherent cells (predominantly lymphocytes) and adherent cells (predominantly monocytes), and the latter were also allowed to mature (from 1 to 7 days) into macrophages. MICL was detected in immature monocytes, but expression was gradually lost upon maturation toward macrophages; lymphocytes were MICL-negative. This experiment was repeated using blood from three separate donors, and comparable results were obtained, although the degree of MICL down-regulation concomitant with monocyte maturation was variable.
Several human cell lines from a range of sources were also analyzed for MICL expression. Consistent with previous data, only the monocytic/promyelocytic cells tested (U937, HL-60, . RNA loading controls for this blot were described previously (19). and MonoMac6) were clearly MICL-positive (Fig. 4C). Moreover, maturation of the monocytic/promyelocytic cell lines THP-1 and HL-60 by phorbol 12-myristate 13-acetate, which induces a macrophage-like phenotype, down-regulated MICL expression as shown by northern blotting (Fig. 4D). In contrast, when HL-60 cells were treated with all-trans-retinoic acid and Me 2 SO, which induce a PMN-like phenotype (25), MICL expression was maintained. This PMN/monocyte-restricted mRNA expression was further corroborated by SAGE, 8 which identified MICL transcripts in only myeloid cells, viz. granulocytes, monocytes, and immature dendritic cells (26,27). 9 MICL was not detected in any SAGE library of lymphoid origin or in mature dendritic cells. All non-immune tissue libraries, except lung and liver, were also MICL-negative (28). 10 Thus, MICL appears to be expressed preferentially by granulocytes and monocytes.
RT-PCR analysis of peripheral blood leukocytes (Fig. 4B) identified three alternatively spliced isoforms of ϳ800, 700, and 1100 bp (MICL-␣, -␤, and -␥, respectively). These were cloned and sequenced, and their predicted structures are depicted in Fig. 2B. MICL-␣ is the isoform characterized in this work; MICL-␤ lacks exon 2 (encoding the transmembrane region); and in the MICL-␥ transcript, the second intron is unspliced, resulting in the introduction of a stop codon after 18 codons. Transcripts similar to the MICL-␥ isoform were also found in several human expressed sequence tags. MICL Is Highly Glycosylated and Is Expressed at the Cell Surface-To investigate MICL protein expression and function, we generated HA epitope-tagged MICL transfectants in a variety of cell lines. Because several similar C-type lectin and lectin-like proteins are not expressed at the cell surface when transfected into certain cell lines (29,30), 11 presumably due to the lack of accessory molecules, we examined the surface expression of MICL transfected into NIH3T3 fibroblasts (Fig. 5A). 8 E. J. Evans, personal communication. 9 Available at bloodsage.gi.k.u-tokyo.ac.jp/. 10  Analysis by flow cytometry of live transfected cells demonstrated high levels of MICL expression at the cell surface, indicating that accessory molecules were not required.
We also examined HA-tagged MICL expression by western blotting and observed bands between 40 and 75 kDa, which are much higher molecular masses than predicted from the polypeptide sequence alone (Fig. 5B). These bands were detected similarly under both reducing and nonreducing conditions in NIH3T3 and RAW264.7 lysates (data not all shown), suggesting they were due to glycosylation of MICL rather than dimerization. To confirm that MICL was glycosylated, we expressed MICL-HA in CHO-Lec1 cells, which are N-glycosylation-deficient (12,13), as well as in wild-type CHO-K1 cells. CHO-K1 cells expressed glycosylated MICL in a similar fashion to NIH3T3 cells, but in CHO-Lec1 cells, the molecular mass of MICL was reduced to near its predicted unglycosylated mass (ϳ31 kDa). Therefore, MICL is expressed at the cell surface and is highly N-glycosylated.
MICL Is Associated with SHP-1 and SHP-2, but Not with SHIP-ITIM-containing receptors are known to exact their inhibitory functions through the recruitment of phosphatase enzymes, including SHP-1, SHP-2, and SHIP (1). We therefore wanted to determine which, if any, of these phosphatases are recruited by the ITIM of MICL. To this end, we expressed HA-tagged MICL in RAW264.7 cells because this cell line is known to express all three phosphatases (31,32) and isolated MICL-positive signaling complexes from these transfectants by anti-HA immunoprecipitation following treatment with and without pervanadate (Fig. 6), which stimulates recruitment of signaling molecules (22). Specific phosphatases that might associate with the MICL ITIM were identified by probing western blots of the immunoprecipitated complexes with selected antiphosphatase antibodies. MICL-HA was observed to be expressed at the RAW 264.7 cell surface by flow cytometry (data not shown), and equal loading of MICL immunoprecipitates both with and without pervanadate was confirmed by anti-HA western blotting. Both SHP-1 and SHP-2 phosphatases were detected in pervanadate-stimulated MICL-positive immunoprecipitates, but not in unstimulated or MICL-negative controls. SHIP, which was present in multiple isoforms (33) in the untreated cell lysates, could not be detected in the MICL immunoprecipitates. These results indicate that MICL preferentially associates with SHP-1 and SHP-2.
The Cytoplasmic Tail of MICL Can Function in an Inhibitory Fashion-The ligands for MICL are not known; and therefore, to investigate the function of MICL, we generated a MICL/BGR chimera that could bind the BGR ligand zymosan. This chimera comprises the cytoplasmic tail, the transmembrane region, and part of the stalk of MICL fused with most of the murine BGR ectodomain (Fig. 1B). Cells stably expressing this chimeric construct have enabled us to dissociate the signaling functions of the MICL cytoplasmic tail (and ITIM) from those of its CTLD. We analyzed MICL/BGR chimeric transfectants by both flow cytometry, which demonstrated surface expression (Fig.  5A), and by western blotting, which revealed that the chimera has a molecular mass slightly lower than that of full-length MICL (Fig. 5B). Because this chimeric molecule contains the two most membrane-proximal N-glycosylation sites from the MICL stalk and given that the murine BGR component of this chimera does not contribute significantly to this N-glycosylation (34) (data not shown), we concluded that these two sites are responsible for the majority of the MICL glycosylation.
We first determined whether this chimeric construct (expressed in NIH3T3 cells) is functional for binding of ␤-glucan ligands. Cells transfected with MICL alone were unable to bind zymosan, whereas MICL/BGR chimeric transfectants were able to bind zymosan to a level comparable with that of cells expressing intact BGR (Figs. 1A and 7A). The BGR dependence of this binding was confirmed by specific blocking with the soluble ␤-glucan laminarin, demonstrating that the ectodomain of the MICL/BGR chimera is functionally intact.
We also utilized the MICL/BGR chimeric transfectants to examine the endocytic/phagocytic potential of MICL. Transfected NIH3T3 cells were allowed to bind and internalize FITClabeled zymosan, and the amount of internalization was quantified as described under "Experimental Procedures." Cytochalasin D, which blocks actin polymerization, was included as a control in these experiments. Whereas Ͼ50% of the BGRtransfected cells were able to internalize the bound FITClabeled zymosan, the MICL/BGR chimeric transfectants were unable to internalize their zymosan significantly above control levels (Fig. 7B). Furthermore, by fluorescence microscopy, FITC-labeled zymosan could be seen in rosettes around the MICL/BGR chimeric transfectants with little internalized, whereas almost all the BGR only-expressing cells had FITClabeled zymosan within them (data not shown). Thus, the cytoplasmic tail of MICL does not mediate ligand uptake.
To demonstrate that the ITIM of MICL is a functional inhibitory domain, we also assessed the ability of the MICL/BGR chimera to block cellular activation. For these experiments, we coexpressed full-length BGR together with the MICL/BGR chi-FIG. 6. MICL associates with phosphatases SHP-1 and SHP-2 in RAW264.7 transfectants. Shown are western blots of MICL-HA-or vector only-transfected RAW264.7 lysates and immunoprecipitates (IP) probed with antibodies as indicated. The anti-phosphatase (SHP-1/ SHP-2/SHIP) Western blots indicate association of SHP-1 and SHP-2 (but not SHIP) with MICL upon pervanadate (PV) stimulation. Lysates, sufficient or deficient in the appropriate phosphatase enzyme, were also included as controls, except for a SHP-2-negative control. Untreated RAW264.7 transfectant cell lysates were analyzed to demonstrate the presence of each phosphatase. The anti-SHIP autoradiograph was overexposed to show that SHIP could not be detected in the immunoprecipitates. These data are representative of three independent experiments. mera in RAW264.7 cells to examine the effect of the MICL ITIM on BGR ITAM-mediated cellular activation. Flow cytometric analysis confirmed that the doubly transfected cell line was expressing the desired constructs at the cell surface (Fig.  8A). The expression of BGR alone resulted in increased zymosan binding (Fig. 8B) compared with negative control transfectants and a concomitant increase in TNF-␣ production (Fig. 8C) as expected (16). However, although zymosan binding increased when the MICL/BGR chimera was coexpressed with BGR, the result of an increased number of zymosan-binding sites at the cell surface, TNF-␣ production was largely blocked, suggesting that the ITIM of the chimera was suppressing the activatory BGR signal. Coexpression of the MICL/BGR chimera with full-length BGR also significantly inhibited the level of zymosan internalization (data not shown).
To confirm that the inhibition by the chimera was dependent on the MICL ITIM and was not an artifact due to factors such as steric hindrance, we generated a MICL/BGR chimeric construct containing a truncated cytoplasmic domain (Fig. 1B,  trunc) and a chimera in which the ITIM of MICL had been mutated (ITIM⌬). As described above, these constructs were coexpressed alongside full-length BGR, and cell-surface expression was confirmed by flow cytometry (Fig. 8A). For both of these transfectants, zymosan binding was increased as before (Fig. 8B), but TNF-␣ production was not inhibited (Fig. 8C). This demonstrates that MICL is a functional inhibitory receptor whose activity is dependent on its ITIM. DISCUSSION We present here the identification and initial characterization of a novel human inhibitory C-type lectin-like immunoreceptor. We have given it the acronym MICL for myeloid inhibitory C-type lectin-like receptor because we have shown that it bears a group V CTLD, has a predominantly PMN/monocyterestricted expression pattern, and is functional as an inhibitory receptor. MICL shares many features with other C-type lectinlike receptors such as the NKG2 family (4), including a CTLD that conforms to the basic structure of classical C-type lectins, but in which the Ca 2ϩ -binding residues are not conserved (Fig.   3A). MICL is, however, most homologous to a subgroup of C-type lectin-like molecules that are distinct from the NKG2 family and that are less well characterized, including BGR, LOX-1, and several novel receptors (7,29). The ways in which the functions of the receptors in this subgroup relate to each other and to those of the NKG2 family, if at all, are not understood. The basic characterization of each member is a necessary first step toward elucidating the various in vivo roles of these molecules.
In the initial characterization of MICL expression, we have identified three isoforms (MICL-␣, -␤, and -␥) that result from alternative splicing (Figs. 2B and 4B). Several different transcripts of up to 6 kb were detected by northern blotting (Fig.  4A), suggesting that there are splice variants other than MICL-␣/␤/␥ (cf. GenBank TM /EBI cDNA clone BC027967.1). Many molecules homologous to MICL are also variably spliced (5,19,(35)(36)(37), although the functional significance of most of their alternative isoforms has not yet been investigated. The MICL-␤ splice variant, which lacks the transmembrane region, is equivalent to isoforms of other C-type lectins that have been reported to be either secreted (38) or withheld in the cytosol (39). It has been suggested that alternative transcripts such as MICL-␤ and MICL-␥ may encode dominant-negative inhibitors of full-length polypeptides (40). The relative levels of these alternative transcripts may be variably regulated in different cells under different conditions. For example, it appears that human BGR is expressed in monocytes/macrophages in only one of its two main isoforms, whereas both mRNA transcripts are expressed by PMNs (19). Similarly, CLECSF6, a group II C-type lectin expressed in several leukocyte populations including PMN and monocytes (41), has stalked and stalkless isoforms that appear to be oppositely regulated by anti-and pro-inflammatory stimuli (35).
The particular PMN/monocyte-restricted expression pattern of MICL (Fig. 4) is shared by few other, if any, ITIM-containing receptors; and therefore, MICL is likely to have a role distinct from those of other myeloid inhibitory molecules. Other human ITIM-containing receptors expressed on PMNs/monocytes as  (53), and cancer (54,55). The variety of inhibitory effects reported to be mediated by these receptors includes inhibition of cell maturation (56), calcium mobilization (57), secretion/exocytosis (58), phagocytosis (59), cytokine production (56), and cellular proliferation (57). It is also notable that a few of the above receptors have seemingly "activatory" roles (some are suggested to promote cell migration (45), cell fusion (60), or apoptosis (55)) presumably in correlation with the function of the pathways they are inhibiting. As an inhibitory receptor preferentially expressed on PMNs/monocytes (Fig. 4), we can speculate that MICL might play a role similar to those above, such as preventing inappropriate migration or activation. It is striking that MICL was down-regulated concomitant with monocyte maturation (Fig. 4, B and D), a pattern that is inverse to that observed for the related activatory receptor BGR (19,61,62). This may provide a mechanism for removing the inhibitory effect of MICL on unstimulated monocytes so that they can become activated. The possibility of a similar down-regulation upon PMN activation has not yet been examined.
Analysis of the predicted MICL polypeptide sequence revealed that the cytoplasmic tail contains a consensus ITIM (Fig. 3C). The ITIMs of receptors such as MICL orchestrate their inhibitory effects by associating with one or more phosphatases, including SHP-1, SHP-2, and SHIP (1). Other inhibitory C-type lectin or lectin-like receptors have been found to preferentially associate with SHP-1 and/or SHP-2 (63), but not usually with SHIP (MAFA-L is one exception (37)). Consistent with this bias, we observed that MICL bound only SHP-1 and SHP-2 (Fig. 6). It is almost certainly via these phosphatases that the MICL ITIM exerts its inhibitory potential, as described below.
Following expression analyses and the identification of the signaling phosphatases associating with MICL, we demonstrated the inhibitory function of MICL using a MICL/BGR chimeric construct in which most of the ectodomain of MICL was replaced with that of BGR. Similar approaches have been employed previously, for example, to demonstrate the functioning of inhibitory receptors that can suppress activatory Fc receptors or B-cell activation (64,65). We have made use of an established activatory pathway in RAW264.7 cells in which BGR, following binding of zymosan, is able to induce cellular activation, including TNF-␣ production and phagocytosis (16). In other immunoreceptor systems, co-clustering of an inhibitory receptor with an activatory receptor results in suppression FIG. 8. MICL can inhibit cellular activation through its ITIM. A, live flow cytometric analysis to indicate the surface coexpression of BGR and various MICL/BGR chimeric constructs stably expressed in RAW264.7 cells. The BGR constructs were HA-tagged, and the chimeric constructs were V5-tagged. wt, the MICL/ BGR chimera containing a wild-type MICL cytoplasmic tail and the murine BGR ectodomain; trunc, the MICL/BGR chimera with a truncated cytoplasmic tail; ITIM⌬, the MICL/BGR chimera with a mutated ITIM sequence; Ab, antibody. B, quantification of FITC-labeled zymosan bound by each double transfectant, showing that the levels for the BGR and chimeric cotransfectants are above those of the BGR and vector-only control (Ϫ). C, concentration of TNF-␣ released by the RAW264.7 double transfectants following incubation with zymosan (zym) as determined by anti-TNF-␣ enzyme-linked immunosorbent assay. Only the chimera with a wild-type MICL ITIM (wt) was able to inhibit BGR-induced TNF-␣ production. These data are representative of three independent experiments in at least two sets of separately generated stable cell lines (74). GluP, glucan phosphate. of activatory signals (66). Because both BGR and our MICL/ BGR chimera could bind zymosan (Fig. 7A), coexpression of the two constructs allowed us to assess the ability of the chimera, following zymosan-induced co-clustering with BGR, to suppress BGR-dependent TNF-␣ induction and phagocytosis. By this approach, we were successful in demonstrating MICLmediated inhibitory function. We also tested two modified MICL/BGR chimeras, one with the cytoplasmic tail truncated and another with a mutated ITIM, to show that inhibition of BGR signaling by the MICL component of the chimera was not due to steric hindrance, for example, but required active inhibition mediated by the MICL ITIM. Although this in vitro model demonstrates that MICL can function as an inhibitory receptor, the true function of MICL in vivo can be established only following the identification of its ligand.
Several other group V C-type lectin-like molecules (the NKG2 receptors) appear to bind non-classical major histocompatibility complex class I or related ligands, but because the MICL CTLD has low sequence identity to members of this subfamily (data not shown), it is probable that it does not bind a similar ligand. For example, CLEC-2, which is homologous to MICL, could not be demonstrated to bind a range of soluble class I tetramers (29). From what little is known about the ligands of the LOX-1, BGR, and CLEC cluster of receptors, it appears that, unlike those of the NKG2 subgroup, they are structurally unrelated. LOX-1 is reported to bind aged apoptotic cells, activated platelets, bacteria, and oxidized low density lipoprotein (8,67), whereas BGR binds ␤-glucans as well as an unidentified T-cell antigen (19,34). It can also not be assumed that MICL will bind protein rather than carbohydrate (9, 68 -70) or act as a scavenger receptor (8). It is even possible that MICL binds another C-type lectin, as was recently found for Nkrp1d and Clrb (71).
Consistent with our ideas above concerning the function of MICL, we could suggest that an endogenous ligand for MICL might be expressed in a context in which PMN/monocytes should not normally be activated, such as at sites of immune privilege or within the blood. We have unsuccessfully screened various cell lines and tissues for a ligand using the MICL ectodomain multimerized on fluorescent beads, as previously employed in other ligand searches (72). We are currently employing alternative approaches to identify a ligand. The fact that MICL is substantially N-glycosylated (Figs. 5B and 6) may well have implications for ligand binding, as previously demonstrated for LOX-1 (70) and Ly49 (69), especially because the level of MICL N-glycosylation is much higher than the Nglycosylation level of homologous receptors, e.g. LOX-1 (70) and BGR (34) (data not shown). Another determinant of the ligandbinding potential of MICL is the possibility of homo-or heterophilic dimerization. Many other group V receptors dimerize through cysteines in the stalk region, 2 and although MICL does contain the necessary residues (Fig. 2), we have not identified any multimerized forms of this receptor. In conclusion, our preliminary characterization of MICL, its expression, and the demonstration of its functional potential provide the basis for future investigations aimed at identifying the ligand(s) and physiological function of MICL.