PILRα, a Novel Immunoreceptor Tyrosine-based Inhibitory Motif-bearing Protein, Recruits SHP-1 upon Tyrosine Phosphorylation and Is Paired with the Truncated Counterpart PILRβ*

SHP-1-mediated dephosphorylation of protein tyrosine residues is central to the regulation of several cell signaling pathways, the specificity of which is dictated by the intrinsic affinity of SH2 domains for the flanking sequences of phosphotyrosine residues. By using a modified yeast two-hybrid system and SHP-1 as bait, we have cloned a human cDNA, PILRα, encoding a 303-amino acid immunoglobulin-like transmembrane receptor bearing two cytoplasmic tyrosines positioned within an immunoreceptor tyrosine-based inhibitory motif. Substrate trapping in combination with pervanadate treatment of 293T cells confirms that PILRα associates with SHP-1 in vivo upon tyrosine phosphorylation. Mutation of the tyrosine residues in PILRα indicates the pivotal role of the Tyr-269 residue in recruiting SHP-1. Surface plasmon resonance analysis further suggests that the association between PILRα-Tyr-269 and SHP-1 is mediated primarily via the amino-terminal SH2 domain of the latter. Polymerase chain reaction amplification of cDNA in combination with genomic sequence analysis revealed a second gene, PILRβ, coding for a putative activating receptor as suggested by a truncated cytoplasmic tail and a charged lysine residue in its transmembrane region. The PILRα and PILRβ genes are localized to chromosome 7 which is in contrast with the mapping of known members of the inhibitory receptor superfamily.

The initiation of cell signaling pathways relies on a dynamic interaction between activating and inhibiting processes that can include, among other things, changes in the phosphorylation status of certain tyrosine residues within the target proteins. Dephosphorylation of these residues is mediated by such phosphatases as the cytosolic SHP-1 (also known as SHP, PTP1C, SH-PTP1, or PTPN6 (1)) which is expressed in hematopoietic cells, and to a lesser extent in non-hematopoietic cells, and which contains tandem amino-terminal Src homology 2 (SH2) 1 domains. The importance of SHP-1 in cellular signal delivery is underscored by the motheaten mouse that carries a natural mutation in the SHP-1 locus and is characterized by widespread autoimmune phenomena resulting from an inability to modulate immune responses (2,3).
Affinity for the SH2 domains is pivotal for interaction of substrates with SHP-1. The flanking sequence of the phosphotyrosine residue promotes the recruitment of specific SH2containing phosphatases and, thus, determines the specificity of the signaling pathway. The consensus sequence (S/L/I/ V)XYXX(L/V), based on sequences originally deduced from several receptors known to bind to the carboxyl-terminal SH2 domain of the protein tyrosine phosphatase SHP-1 (4 -6), defines all immunoreceptor tyrosine-based inhibitory motif (ITIM)-bearing receptors (7) including natural killer cell, B cell, and monocyte and dendritic cell inhibitory receptors (reviewed in Refs. 8 and 9). Members of the inhibitory receptor superfamily (10,11) can be divided into two groups. The immunoglobulin (Ig) superfamily includes luminal amino-terminal (e.g. type I) transmembrane glycoproteins with two or more Ig-like domains such as p58/KIR2DL3, Fc␥RIIB, Ig-like transcripts as well as the mouse PIR-B. The other group is comprised of CD94/NKG-2, CD72, and the mouse Ly49 and NKR-P1 which are cytoplasmic amino-terminal (e.g. type II) transmembrane proteins expressing C-type lectin extracellular architecture.
The existence of complementary proteins expressing similar extracellular domains as the inhibitory receptors, while having distinctive transmembrane and frequently truncated cytoplasmic tails, suggests similar ligand-binding specificities but contrasting signaling capabilities for each individual of the protein pair. The truncated protein would have cellular activating properties, in contrast with its ITIM-bearing counterpart which would inhibit cell signaling through recruitment of SHP-1, SHP-2, or SHIP phosphatases (9) seemingly via their respective carboxyl-terminal SH2 domains (12,13). Human immunoreceptors map to the "complex" or "leukocyte receptor cluster" located on chromosome region 19q13.4 or to the natural killer complex on chromosome 12 (11). Mouse immunoreceptors are located on chromosomes 7 and 6 in regions syntenic to human chromosome 19 and 12, respectively (11), or to band B4 of mouse chromosome 10 (14) which does not appear to have any conserved linkage homology to either human chromosome 19 or 12.
We now present evidence of a novel pair of receptors express-ing a single extracellular Ig-like domain that we have designated as "paired immunoglobulin-like receptor" (PILR)␣ and PILR␤ to distinguish their putative inhibitory and activating gene products, respectively. The ITIM-bearing receptor PILR␣ recruits SHP-1 via its amino-terminal SH2 domain and is likely to have cellular inhibitory potential. The lack of a cytoplasmic tail and the presence of the transmembrane lysine residue in the second receptor PILR␤ suggest its potential activating function. Both genes map cytogenetically to human chromosome 7.

EXPERIMENTAL PROCEDURES
Cells and Culture Conditions-All cell lines were obtained from American Type Culture Collection except where indicated. The human embryonic kidney epithelial cell line 293T (Edge Biosystems) was maintained in Dulbecco's modified Eagle's medium, 10% fetal bovine serum (FBS). The T cell-derived KG-1, Jurkat and K562 cell lines were maintained in RPMI, 10% FBS. The B cell-derived cell lines WIL2-NS, Namalwa, Daudi, and Raji were maintained in RPMI, 10% inactivated FBS. The 6F11 B cell line was cultured in Iscove's modified Dulbecco's medium containing 2 mM L-glutamine and 15% FBS. The macrophage 28SC cell line was maintained in Iscove's medium supplemented with 0.03 mM thymidine, 0.1 mM hypoxanthine, 0.05 mM ␤-mercaptoethanol, and 10% FBS. NK-92ci cells (ImmuneMedicine, Vancouver, British Columbia, Canada) transformed with a pCep-IL-2 construct were maintained by selection in interleukin-2-free Myelocult medium (Stem Cell Technologies, Vancouver, British Columbia, Canada).
Antibodies-SHP-1 was precipitated and detected using a monoclonal antibody (mAb; P17320; Transduction Laboratories) or an inhouse polyclonal antibody (pAb; 237, see Ref. 15). PILR␣ cDNA was subcloned in-frame with a carboxyl-terminal triple HA tag. The HA epitope was precipitated with a high affinity rat anti-HA mAb (3F10; Roche Molecular Biochemicals) and detected with a mouse mAb (12CA5; Roche Molecular Biochemicals). Separate membranes were also probed with anti-phosphotyrosine 4G10 mAb (Upstate Biotechnology, Inc.) and anti-ubiquitin pAb (U-5379; Sigma). Secondary antibodies were horseradish peroxidase-conjugated to goat anti-mouse or goat anti-rabbit antibody (Bio-Rad) and detection relied on enhanced chemiluminescence (NEN Life Science Products).
Yeast Two-hybrid Screen-Screening of a human mammary gland cDNA library (CLONTECH) with the SHP-1-C455S catalytic mutant (16) was performed as follows. The full-length SHP-1-C455S was cloned into SalI-restricted pBTM116-Src by standard polymerase chain reaction (PCR) procedures and thus was fused in-frame with the carboxyl terminus of the DNA binding domain of the bacterial activator LexA. pBTM116-Src contains a mutant c-Src kinase expression cassette designed for the phosphorylation of protein tyrosine residues in yeast (17). The pBTM116-Src-SHP-1-C455S construct was then used to transform the yeast reporter strain, L40␣ (MATa trp1 leu2 his3 LYS2::lexA-HIS3 URA3::lexA-lacz). Following transformation using Li 2ϩ acetate (18), L40␣ expressing pBTM116-Src-SHP-1-C455S was further transformed with the library. Positive interactors were isolated according to the manufacturer's specifications (CLONTECH Laboratories, Inc.). The positive control consisted of a pBTM-116-cnx1ϩ and pGADGH-hus5ϩ combination kindly provided by M. Pelletier and D.Y. Thomas 2 (NRC Biotechnology Research Institute, Montreal, Quebec, Canada). cDNA sequences were determined and subjected to the BLAST search of the NCBI data bases. The target sequence of interest was used as a probe for screening of cDNA libraries.
Screening of Human cDNA Libraries-32 P-Labeled probes were prepared by random labeling of the PILR␣ PCR product using Ready-to-Go DNA Labeling Beads (Amersham Pharmacia Biotech) and [␣-32 P]dCTP. Placenta, acute myelocytoma leukemia (KG-1), bone marrow, and leukocyte cDNA libraries were screened using the labeled probes. Positive clones were subjected to PCR using GT11-specific primers.
Mutant Expression, Immunoprecipitation, and Immunoblotting-Single and double tyrosine to phenylalanine mutants of PILR␣-HA, i.e. Y269F, Y298F, Y269F/Y298F, were generated by PCR-based site-directed mutagenesis. 293T cells were transfected either alone or in combination with SHP-1 by calcium phosphate precipitation. Fortyeight hours after transfection the cells were treated with pervanadate (100 M, 30 min at 37°C), a phosphatase inhibitor that induces maximal tyrosine phosphorylation (19), and then washed with phosphate-buffered saline and lysed on ice. Solubilized extracts were immunoprecipitated with the appropriate antibodies followed by protein A-or protein G-Sepharose. The immunoprecipitates were resolved by SDS-PAGE and revealed by standard immunoblotting techniques.
Glycosylation and Ubiquitination Status of PILR␣-HA-293T cells were transiently transfected with PILR␣-HA. Following immunoprecipitation with an HA-directed antibody, immune complexes were washed and denatured, and N-glycosylation status was determined by overnight incubation at 37°C with 5 milliunits of N-glycosidase F (PNGase:F, Roche Molecular Biochemicals) while a parallel series of experiments determined O-glycosylation status by incubation with 1 milliunit of neuraminidase (to remove sialic acid residues, Roche Molecular Biochemicals) and/or 1 milliunit of endo-␣-N-acetylgalactosaminidase (O-glycosidase, Roche Molecular Biochemicals). The various reactions were resolved on 10% SDS-PAGE gels. To test for the presence of covalently bound ubiquitin in the 55-kDa expressed PILR␣-HA species, transiently transfected 293T cells were treated with pervanadate, and HA-bound immunoprecipitates were collected, resolved by SDS-PAGE, and probed with an anti-ubiquitin pAb.
Preparation and Expression of GST Fusion Proteins Containing the SH2 Domains of SHP-1-A 300-base pair segment encoding either the amino-terminal SH2 domain (SH2(N)) or the carboxyl-terminal SH2 domain (SH2(C)) of SHP-1 were generated by PCR amplification of human SHP-1 cDNA (20) and subcloned into pGEX-2T (Amersham Pharmacia Biotech). The SHP-1.SH2(N) and SHP-1.SH2(C) GST fusion proteins were produced in Escherichia coli DH5␣ cultures transformed with the corresponding plasmid and induced with 25 M isopropyl-␤-Dthiogalactoside for 22 h at 28°C. The GST fusion proteins were isolated using glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech). Protein expression and purity were determined by SDS-PAGE and Coomassie Blue staining.
Surface Plasmon Resonance Measurements-Surface plasmon resonance (SPR) was performed on a BIAcore apparatus using CM5-sensor chips (Biosensor AB, Uppsala, Sweden). The tyrosyl-phosphorylated (pY) synthetic peptides, KDDGIV(pY)ASLALSSSTS and PQNETL(pY)-SVLKA, corresponding, respectively, to the Tyr-269-and Tyr-298-based motifs contained in PILR␣ were immobilized at 0.5 mg/ml on the Biosensor chip. Immobilization efficiency was verified using the anti-phosphotyrosine 4G10 mAb. GST-SH2 domain fusion proteins (see above) were initially dialyzed in 10 mM Hepes buffer (pH 7.4) containing 150 mM NaCl and 3.4 mM EDTA (HBS). The GST fusion proteins (0.5-2000 nM), diluted in HBS, 0.05% P20, were injected over the test surfaces at a flow rate of 5 l/min. Regeneration of the Biosensor chip using 46 mM HCl, 1 M NaCl did not result in loss of subsequent signal. Dissociation constants (K D ) were calculated by saturation (nonlinear regression) analysis as well as by kinetic analysis (e.g. the ratio of k off and k on determined using BIAevaluation software; Biosensor AB, Uppsala, Sweden).
Northern Blot Analysis-Human multiple tissue Northern blots (CLONTECH Laboratories, Inc.) containing 2 g of poly(A) ϩ RNA per lane were hybridized consecutively with 32 P-labeled full-length PILR␣ cDNA and with 32 P-labeled ␤-actin cDNA probes. The level of expression in a series of human cells was also examined. Total RNAs from the individual cell lines were extracted by the TriZOL method (Life Technologies, Inc.) and enriched in poly(A) ϩ RNA by passage on oligo(dT) spun columns (Amersham Pharmacia Biotech). Ten g of poly(A) ϩ RNA were loaded per lane, transferred to membrane, probed as above, and then washed (maximum stringency being 2 ϫ 15 min, 0.1ϫ SSC, 0.1% SDS, 55°C).
Long Range PCR-The sequence of the 5Ј-untranslated region was amplified by 5Ј-rapid amplification of cDNA ends from Marathon-Ready cDNA (CLONTECH), whereas the genomic organization of PILR␣ and PILR␤ genes was partially determined using the proofreading Taq polymerase GenomeWalker TM kit (CLONTECH).

The SHP-1 Catalytic Mutant SHP-1-C455S Interacts with
Several ITIM-bearing Protein Fragments-A SHP-1-C455S catalytic mutant was used as bait in a yeast two-hybrid screening of a human mammary gland cDNA library. 4.5 ϫ 10 6 clones were screened to reveal 24 interactors able to grow on Trp Ϫ / Leu Ϫ /His Ϫ medium and able to induce ␤-galactosidase activity (Fig. 1). These positive clones were sequenced and identified as a novel ITIM-bearing protein, subsequently named PILR␣, as well as known interactors of SHP-1 namely SHPS-1 (21), EGFRBP-Grb2 (22), and the leukocyte-associated Ig-like recep-tor LAIR-1 (23). The sequence of the novel interactor was extended by PCR amplification of the target cDNA library and was ultimately confirmed following its use as a probe in the cDNA library screening which yielded the full-length sequence.
The nucleotide sequence and deduced amino acid sequence, indicating a theoretical mass of 31.8 kDa, are depicted in Fig.  2. Structurally the gene product does not contain an IgG domain per se, although the single extracellular cysteine residue (Cys-125) is flanked by a motif, e.g. YXCXVXL, reminiscent of the carboxyl-terminal cysteine-based motif, e.g. (F/Y)XCX(V/ A)XH, involved in the intradomain disulfide bond of immunoglobulin and major histocompatibility complex (MHC) proteins (PROSITE data base; reference number PS00290). This same segment bears a slight homology to extracellular regions of sialoadhesin (24). In addition, there is a potential N-glycosylation site (NX(T/S)), numerous serine and threonine residues capable of being O-glycosylated, a transmembrane domain, and three potential tyrosine phosphorylation sites, two of which, e.g. Tyr-269 and Tyr-298, reside within an immunoreceptor tyrosine-based inhibitory motif.
PILR␣-HA migrated with an apparent molecular mass of 55 kDa rather than the expected theoretical molecular mass of 36 kDa. The 55-kDa molecular mass includes 4.6 kDa contributed FIG. 1. SHP-1 interacts with a number of known proteins and a novel inhibitory receptor PILR␣. Interactions were assayed in the yeast strain, L40␣, which requires Trp, Leu, and His to grow. pBTM-116-Src-SHP-1-C455S (bait) and pACT2 (target, library) constructs carry Trp and Leu, respectively, as their selective markers. Library plasmids from positive clones were cured and used in combination with irrelevant (A, negative control), pBTM-116-Src (B, vector control) or pBTM-116-Src-SHP-1-C455S (C, the bait) plasmid, to retransform yeast. These yeast transformants were randomly plated in triplicate onto Trp Ϫ /Leu Ϫ /His Ϫ master plates, grown to a visible biomass, and then submitted to colony-lift ␤-galactosidase assay. The positive control (ϩ) was a pBTM-116-cnx1ϩ (bait)/pGADGH-hus5ϩ (target) combination. FL-cDNA is an unknown interactor described in the NCBI dbEST data base as "fetal lung cDNA."  (Fig. 4).
Mutation of Tyrosine Residues of PILR␣-HA Alters Its Interaction with SHP-1-The role of the ITIM-based tyrosine residues of PILR␣-HA in recruiting SHP-1 was investigated by mutational analysis. Immunodetection of SHP-1 following coimmunoprecipitation with an HA-directed antibody indicated that the Tyr-269 motif, e.g. IVYASL, was the target for binding with SHP-1 (Fig. 5A). Indeed, whereas the Y298F substitution seems to have had only a slight effect on the association with SHP-1, the Y269F mutant and the Y269F/Y298F double mutant significantly decreased the SHP-1 signal. PILR␣-HA (Fig.   5C) and SHP-1 (Fig. 5B) were equally expressed in the corresponding lysates.
Determination of the K D Values of Interactions between PILR␣ Phosphotyrosyl Peptides and SH2 Domains of SHP-1 by SPR-The importance of the Tyr-269 motif in recruiting SHP-1 is supported using BIAcore detection. The immobilized phosphorylated peptide corresponding to this motif, e.g. KDD-GI(pY)-ASLALSSSTS, demonstrated high affinity for the SHP-1-SH2(N) domain (Fig. 6, top) and intermediate affinity for the SHP-1-SH2(C) domain (Fig. 6, bottom). Interestingly, the SHP-1-SH2(C) domain selectively recognized the Tyr-298 phosphopeptide, e.g. PQNETL(pY)SVLKA, but with much lower affinity (Fig. 6, bottom). The dissociation constants (K D ) for the various interactions were determined by saturation analysis (nonlinear regression analysis) and supported by those obtained by kinetic analysis (e.g. K D ϭ k off /k on : Table I) following binding of the SHP-1-SH2(N) and -SH2(C) GST fusion proteins, respectively, to the Tyr-269 phosphopeptide. A K D value of 3777 Ϯ 447 nM was obtained upon binding of the SHP-1-SH2(C) GST fusion protein to the Tyr-298 phosphopeptide. GST itself did not bind to any of the test surfaces (data not shown) and neither of the GST-SH2 domain fusion proteins bound significantly (e.g. less than 5 resonance units) to the control Biosensor chip surface.
The Genomic Organization of PILR Reveals Two Distinct, yet Structurally Related, Receptors-Rapid amplification of cDNA ends designed to clone the 5Ј end of the PILR␣ cDNA revealed the presence of a second cDNA displaying long regions of near sequence identity to PILR␣ but differing in its 5Ј non-coding sequence. The existence of a related gene was also suggested from long range PCR performed on human genomic DNA (results not shown) and was confirmed by the sequence of a 200-kilobase segment of human chromosome 7 deposited subsequently in GenBank TM under the accession number RG161A02. The PILR␣ gene consists of seven exons and six introns spanning approximately 26.7 kb. The second gene, designated PILR␤, is located 5.6 kb upstream of the PILR␣ gene and consists of four exons and three introns spanning approximately 9.8 kb (Fig. 7B). The nucleotide sequence of the first three exons of the two genes is extremely similar, displaying more than 90% sequence identity (Fig. 7A), and suggests that the two genes share a common origin. Tables II and III summarize the nucleotide sequence of the splice sites of the PILR␣ and PILR␤ genes, respectively, and indicate that in both cases the intron-exon boundaries conform to the GT-AG rule. PILR␤ codes for a protein with similar extracellular features as those of PILR␣ but with a short cytoplasmic tail and a charged lysine residue within its transmembrane domain (Fig. 7A).
Expression of PILR in Tissues and Selected Cell Lines-Northern blot analysis of selected human tissues revealed a strong signal in peripheral blood leukocytes and lower intensity signals in lung, spleen, and placenta (Fig. 8A). Analysis of various human cell lines revealed a signal in a two B cell lines, e.g. WIL2-NS and 6F11 (Fig. 8B), whereas all other cells tested were negative for PILR. Due to the high nucleotide sequence identity of the two genes and the similar length of the two transcripts, the Northern blots do not allow us to discriminate which of the two forms, e.g. PILR␣ or PILR␤, is expressed in these tissues and cell lines. To resolve this question, PCR with a forward oligonucleotide primer whose sequence was common to both PILR␣ and PILR␤ (HUKF1, 5Ј-GGGAAGCTTAT-GGGTCGGCCCCTGCTGCTGCCC) and reverse primers specific for each form, HUKR3 for PILR␣ (5Ј-CCATCTCGAGTTT-TGAGAGGGCTG) and HUKR5 for PILR␤ (5Ј-TCCTCCGGGG-CTAATACACATCC), were performed on reverse transcribed single-stranded cDNA from various tissues including colon, leukocyte, ovary, prostate, small intestine, spleen, testis, thymus, placenta, and mammary gland. The results indicate that both PILR␣ and PILR␤ were expressed, e.g. were paired, at the mRNA level in the individual samples (results not shown).

DISCUSSION
The cytosolic phosphatase SHP-1 contains tandem SH2 domains that bind tyrosine-phosphorylated proteins and, thus, by virtue of its catalytic subunit, gets recruited as an effector enzyme in a signaling pathway initiated by activated tyrosine kinase receptors (26). Specificity in tyrosine kinase signaling pathways is critical and is often dictated by the intrinsic affinity of SH2 domains for the flanking sequences of phosphotyrosine residues. The ITIM consensus sequence (S/L/I/V)-XYXX(L/V), via its recruitment of phosphatases, is central to inhibiting the signaling cascade initiated by activating receptors and, thus, plays a pivotal role in regulating cells of the immune system (10).
PILR␣-HA is expressed as a membrane-bound N-and Olinked glycoprotein that contains within its cytoplasmic domain three potential tyrosine phosphorylation sites. Two of these tyrosines, e.g. Tyr-269 and Tyr-298, are positioned in YXXL consensus sequences in a manner very reminiscent of the ITIM (D/E)XXYXXL(X) 26 YXXL originally described for numerous inhibitory receptors (7-9). In the case of PILR␣-HA the Curves such as these were used to obtain the kinetic data presented in Table I. The control surface did not bind either of the SH2 domains, and thus the resultant data appear as straight lines on both graphs. spacer between the two portions of the ITIM is 25 amino acids long. By using an SHP-1-D421A substrate-trapping mutant attenuated in its catalytic function but sufficiently stabilized to permit isolation of itself and the "trapped" substrate (25) indicates that tyrosine phosphorylation is required for in vivo association with SHP-1. Mutational analysis reveals that Tyr-269 in the motif IVYASL is the essential residue for interaction with SHP-1 although Tyr-298 in the ETLYSVL motif may also contribute marginally to the interaction. Parenthetically, both co-immunoprecipitation and surface plasmon resonance anal-yses indicate that the membrane-proximal tyrosine residue, e.g. Tyr-246, which lies outside of the ITIM consensus sequence and which does not conform to the motif thought to bind to SHP-1 SH2 domains (27) does not, in fact, contribute to the recruitment of SHP-1 by PILR␣ (data not shown). A parallel series of experiments indicate that PILR␣-HA also interacts with the protein tyrosine phosphatase SHP-2 (data not shown) perhaps through its Tyr-298 motif which resembles the tyrosine-based motif, e.g. TXYXX(V/I), found in the human NK cell receptor 2B4 as well as in the activation molecule SLAM FIG. 7. Genomic organization of the PILR␣ and PILR␤ genes and comparison of their respective deduced amino acid sequences. A, the deduced amino acid sequences of PILR␣ and PILR␤ proteins are presented in single letter code. Only those amino acid residues that differ between the two sequences are shown in the PILR␤ sequence. The minus symbol has been added to facilitate alignment (spacing determined by the respective nucleotide sequences). The amino-terminal signal peptide sequence is underlined. The transmembrane domain is displayed as a shaded box in both cases, and the charged lysine residue located within this region in PILR␤ is highlighted in bold. B, schematic representation of the PILR␣ and PILR␤ genes as well as the respective receptors. The numbers above the genes indicate the individual exons. The hydrophobic signal peptide (solid), the transmembrane region (hatched), and the three cytoplasmic tyrosine residues (e.g. Y) are indicated. The bar represents the scale in base pairs.  4, 195185-195218; exon 5, 196597-196646; exon 6, 196887-196918; exon 7, 197091-197398 (polyadenylation signal at 197381). The "gt-ag" delimiting each intron is underlined. and thought to be essential for recruitment of SHP-2 and adaptor signaling proteins such as SAP (see Ref. 28 and references therein). Our in vivo demonstration of an interaction between SHP-1 and PILR␣ is supported by our surface plasmon resonance analyses that indicate the high affinity of the aminoterminal SH2 domain of SHP-1 for the Tyr-269 phosphopeptide. A number of receptors, via their (pY)XX(F/P/L/Y) motif, bind the amino-terminal SH2 domain of SHP-1 (27) and, at least in the case of the erythropoietin receptor (29), the interaction with SHP-1 mediates termination of signaling initiated by the receptor itself. In contrast, immunoreceptors such as Fc␥RIIB (4), CD22 (30), and p58/KIR2DL3 (31,32), which inhibit signaling initiated by other receptors, apparently do so by recruiting the carboxyl-terminal SH2 domain of SHP-1 (5,33), an exception being pp130, the product of p91/PIR-B in activated macrophages (34), which does so by interaction with the amino-terminal SH2 domain of SHP-1. PILR␣-associated SHP-1 probably exists in an active conformation since activation of SHP-1 resides in amino-terminal SH2 domain occupancy (35,36). The length of the spacer between Tyr-269 and Tyr-298 of PILR␣-HA (25 amino acids), and recognition of the amino-terminal and carboxyl-terminal SH2 domains of SHP-1 by Tyr-269 and Tyr-298, respectively, suggest the potential for tandem occupancy and, thus, maximal phosphatase activity (35,37). Although no definitive ligand for PILR␣ has been identified at this time, the lack of homology, extracellular or otherwise, between PILR␣ and known immunoreceptors, does not necessarily preclude MHC molecules, glycoproteins on selfcomponents, and complex carbohydrates on microbial pathogens as potential ligand candidates.
PILR␣-HA is expressed in 293T cells (as well as in Cos-1 and MCF-7 cells; data not shown) as a 55-kDa species that is significantly higher than the molecular mass of 36 kDa expected from the deduced amino acid sequence. We were able to reduce the molecular mass of expressed PILR␣-HA to approximately 42 kDa and thus determine that the major portion of this difference was due to the presence of N-and O-linked carbohydrate residues. An explanation for our inability to reduce completely the mass of PILR␣-HA to 36 kDa may well lie in the substrate specificity of the O-glycosidase (e.g. endo-␣-Nacetylgalactosaminidase) which only digests unsubstituted Gal␤1-3GalNAc moieties while leaving all other types of Oglycans untouched (38). It is therefore very possible that a certain portion of O-linked carbohydrate residues remained bound to PILR␣-HA following treatment.
One of the striking features of the ITIM-bearing family of receptors is the pairing of inhibitory receptors with complementary activating receptors. The importance of a matched receptor pair and their individual cellular regulatory roles is exemplified by the stimulation of natural killer and T cell signaling pathways upon binding of the MHC class I ligand to the killer cell activating receptor which contrasts the disrupting signal initiated by binding of the ligand to the inhibitory counterpart killer cell inhibitory receptor (39). To date, all human immunoreceptors have mapped cytogenetically to the 19q13.4 complex or leukocyte receptor cluster, or to the natural killer complex on chromosome 12. Genomic organization of PILR␣ revealed the presence of two genes, e.g. PILR␣ and PILR␤, with high sequence homology on chromosome 7. The ITIM-bearing gene product encoded by PILR␣ has a non-polar transmembrane domain. In contrast, the truncated protein PILR␤ with the charged residue (e.g. lysine) in its transmembrane domain confirms that it is likely the activating counterpart of PILR␣ (9). Activating receptors such as the mouse PIR-A associate with Fc⑀RI-␥ (40) to deliver activation signals to macrophages. We are presently investigating the possibility that our own activating isoform PILR␤ can associate with proteins bearing immunoreceptor tyrosine-based activation motifs (ITAMs) such as the Fc receptors. Control of cellular signaling pathways probably occurs through a balance between PILR␣-mediated inhibition and PILR␤-mediated activation.
Based on these data and the current literature which pairs inhibitory and activating receptors such as killer cell inhibitory receptor and killer cell activating receptor, Fc␥RIIB and Fc␥RIII, and PIR-B and PIR-A, to name but a few (9, 11), we are confident in stating that PILR␣ and PILR␤ represent a novel ITIM-bearing and non-ITIM-bearing receptor pair. Furthermore, their chromosomal localization and their lack of homology with any of the known inhibitory/activating receptors suggests that they represent a novel family of receptors. Their chromosomal localization may yet reveal itself to include other genes bearing information relevant to our understanding of the regulation of cellular signaling pathways.
Acknowledgments-We are grateful to Dr. J. A. Cooper for supplying us with the pBTM116-Src construct. We also gratefully acknowledge  Drs. E. O. Long and S. Rajagopalan of the NIAID (National Institutes of Health) for insightful discussions. We thank N. Jolicoeur for preparation of the figures and Y. Fortin for technical assistance.