Characterization of Lnk. An adaptor protein expressed in lymphocytes.

Stimulation of the T cell antigen receptor (TCR) activates a set of non-receptor protein tyrosine kinases that assist in delivering signals to the cell interior. Among the presumed substrates for these kinases, adaptor proteins, which juxtapose effector enzyme systems with the antigen receptor complex, figure prominently. Previous studies suggested that Lnk, a 38-kDa protein consisting of a single SH2 domain and a region containing potential tyrosine phosphorylation sites, might serve to join Grb2, phospholipase C-gamma1, and phosphatidylinositol 3-kinase to the TCR. To elucidate the physiological roles of Lnk in T cell signal transduction, we isolated the mouse Lnk cDNA, characterized the structure of the mouse Lnk gene, and generated transgenic mice that overproduce Lnk in thymocytes. Here we report that although Lnk becomes phosphorylated during T cell activation, it plays no limiting role in the TCR signaling process. Moreover, we have distinguished p38(Lnk) from the more prominent 36-kDa tyrosine phosphoproteins that appear in activated T cells. Together these studies suggest that Lnk participates in signaling from receptors other than antigen receptors in lymphocytes.

The earliest detectable biochemical event following stimulation of the T cell receptor (TCR) 1 is the activation of several protein-tyrosine kinases, resulting in transient tyrosine phosphorylation of numerous intracellular substrates. These substrates include the ␥, ␦, and ⑀ components of the CD3 complex (1), the subunit of the TCR (2), phospholipase C-␥1 (PLC-␥1) (3)(4)(5), the Vav proto-oncogene product (6,7), the Ras-GTPaseactivating protein (Ras-GAP) (8), ezrin (9), several cytoskeletal proteins, and the non-receptor protein-tyrosine kinases themselves, Lck, Fyn, and ZAP-70 (10 -13). Phosphorylation of Lck (14,15), ZAP-70 (16), and PLC-␥1 (17) alters the catalytic activity of each protein, but a detailed explication of the machinery that links ligand occupancy of the TCR to changes in cell behavior remains unapproachable. In part, this process remains mysterious because of an incomplete appreciation of the total set of enzymes that associate with, and respond to Lck, Fyn, and ZAP-70 as they interact with the TCR complex.
Among the substrates that become rapidly phosphorylated on tyrosine following T cell stimulation is a set of so-called adaptor proteins, molecules which lack catalytic function (insofar as is known) but which possess interaction domains, e.g. SH2 domains that bind tightly to phosphotyrosine residues or SH3 domains that bind proline-rich sequences (18). An appreciation for the importance of these proteins derives from studies of growth factor receptor kinases, where initial binding and phosphorylation of the Shc adaptor creates a binding site for a second adaptor, Grb2, that then directs Sos, a guanine nucleotide exchange factor, to the membrane where it can act upon Ras (19,20). In a similar way, the adaptor proteins of T cells, by binding to tyrosine-phosphorylated components of the TCR, should serve to juxtapose other SH2 domain-containing proteins with the TCR, thereby regulating the interaction of downstream signaling pathways with the ligand-engaged receptor. A thorough enumeration of the normal complement of adaptor proteins, and an analysis of how these proteins behave, should therefore help to illuminate mechanisms of signal transduction in lymphoid cells.
Lymphocytes themselves contain Shc, which becomes phosphorylated and binds both TCR subunits and the Grb2-Sos complex following TCR stimulation (21). Although several observations support the view that Shc may help to link Grb2 and perhaps other proteins to the TCR complex, the importance of this interaction remains controversial, in part because only a small proportion of total cellular Grb2 binds to phosphorylated Shc after TCR stimulation (22,23). SLP-76 (SH2 domain-containing leukocyte protein of 76 kDa) contains many tyrosinephosphorylation sites, an SH2 domain, and a proline-rich region that may interact with SH3-containing proteins (24). TCR ligation stimulates SLP-76 phosphorylation, and augmented expression of SLP-76 in the Jurkat cell line improves the nuclear translocation of the NFAT transcription factor (a key regulatory event in IL-2 expression) following TCR cross-linking (25). Moreover, SLP-76 has been shown to bind to the Vav oncoprotein, a putative exchange factor for Rho/Rac GTP-binding proteins (26), which is also phosphorylated by protein tyrosine kinases upon TCR stimulation. Coexpression of SLP-76 with Vav in Jurkat cells produces synergistic increases in NFAT activation following TCR stimulation (27). Yet another adaptor protein, the 120-kDa product of the cbl gene, becomes tyrosine phosphorylated following TCR activation and thereaf-ter associates with a variety of signaling molecules including Crk, Grb2, Fyn, Lck, ZAP-70, Ras-GAP, PLC-␥1, and the phosphatidylinositol 3-kinase (28 -34). While the significance of these interactions remains obscure, the documented transforming ability of v-cbl, in which normal cbl-encoded sequences are partially truncated, supports the view that this molecule plays an active role in TCR-mediated signal transduction (35). A fourth presumed adaptor protein, HS1, which was recently shown to behave as an Lck-binding protein (36), contains an SH3 domain and also becomes tyrosine phosphorylated following antigen receptor cross-linking (37,38). T lymphocytes from mice lacking HS1 exhibit impaired proliferative responses and a disturbance in normal repertoire selection (38). At a minimum, then, the adaptor proteins Shc, Grb2, SLP-76, Cbl, and HS1 all behave as molecules that may link activation of kinase components of the T cell antigen receptor complex to subsequent biochemical responses.
One prominent substrate of the TCR-activated protein tyrosine kinase(s) is a 36 -38-kDa protein (pp36) that associates with the SH2 domains of numerous signaling molecules, including Grb2, PLC-␥1, phosphatidylinositol 3-kinase, GAP, and the Src family kinases (22, 39 -42). Although other phosphoproteins, notably Shc, have been shown to associate with the Grb2 SH2 domain, pp36 is the predominant species that binds the SH2 domain of Grb2 in anti-TCR-stimulated T lymphocytes. Moreover, tyrosine phosphorylation of pp36 correlates well with inositol phosphate production, a measure of PLC-␥1 activity (43). Transfection of a chimeric CD45 phosphotyrosine phosphatase containing the Grb2 SH2 domain into Jurkat T lymphoblasts yielded cells in which TCR stimulation failed to induce tyrosine phosphorylation of pp36, while phosphorylation of other proteins remained intact (43). These results were interpreted to suggest that phosphorylated pp36, once bound to the chimeric Grb2/CD45 protein, became rapidly dephosphorylated. Interestingly, inositol phosphate production and calcium mobilization were compromised in the transfected cells, further suggesting a role for pp36 in coupling the TCR to phosphoinositide metabolism. Efforts have therefore focused on biochemical characterization of pp36.
Recently, cDNA clones encoding a 38-kDa molecule called Lnk, which shares many properties with pp36, were isolated (44). Lnk consists of a single SH2 domain juxtaposed with a sequence of unknown function in which are embedded potential tyrosine-phosphorylation sites. In an initial report, Lnk expression was documented in lymphoid tissues and the Lnk protein was shown to become tyrosine phosphorylated after TCR crosslinking. Intriguingly, phosphorylated Lnk appeared to interact (as judged by coimmunoprecipitation) with Grb2, PLC-␥1, and phosphatidylinositol 3-kinase. These observations suggested that Lnk was in fact the pp36 protein, the phosphorylation of which correlates so well with phosphoinositide turnover (43,44).
To investigate the role of Lnk in TCR signal transduction and in T cell development, we have characterized the structure of mouse Lnk, the gene that encodes it, and its expression in various tissues and lymphocyte subsets. In addition, we generated transgenic mice expressing Lnk at very high levels in thymocytes. These studies demonstrate that the availability of Lnk protein does not restrict normal T cell development or responsiveness. Thymocytes from transgenic mice proliferated normally and showed an equivalent pattern of tyrosine-phosphorylated proteins when compared with normal thymocytes stimulated via TCR cross-linking. By studying the behavior of proteins bound by GST-Grb2 fusion proteins, and by anti-Lnk heteroantisera, we learned that pp36 and Lnk are distinct entities, although both become phosphorylated during T cell activation. Thus Lnk represents a novel adaptor protein expressed in lymphocytes, which almost certainly interacts with phosphoproteins other than those that comprise the T cell antigen receptor.

EXPERIMENTAL PROCEDURES
Isolation of cDNA and Genomic Clones-A mouse Lnk cDNA fragment was amplified from thymocyte cDNA by PCR using the following primers, which were synthesized based on the published rat Lnk cDNA sequence: forward (FWD) primer: 5Ј-TCACTTCCTGTCCTGCTAC-CCCTG, reverse (REV) primer: 5Ј-GCACAGCTGTGAGAGAGGGGT-GTA. The cDNA fragment was labeled using the random-primer method with [␣-32 P]dATP (DuPont NEN), and used as a probe to screen a library previously constructed with mouse thymocyte cDNA in the ZAP II vector (Stratagene, La Jolla, CA) by plaque hybridization. Five positive plaques were isolated from 1 ϫ 10 6 plaques screened, and phage DNAs were excised in vivo to rescue phagemids containing the cDNA fragment. The DNA sequences of these clones were determined using the dideoxy chain termination method employing Sequenase 2.0 (U. S. Biochemical Corp.). Genomic clones were isolated from a FIX II library of 129SV liver DNA partially digested with MboI (Stratagene) using [␣-32 P]dATP-labeled cDNA fragments as probes. A total of 1 ϫ 10 6 plaques were screened and 28 positive plaques were isolated. Five independent clones of these were expanded and each NotI restriction fragment was subcloned into pBluescript KS(ϩ) (Stratagene). Physical maps of subcloned plasmids were determined by restriction enzyme digestion, Southern blotting, PCR, and DNA sequencing.
Generation of Anti-Lnk Antibodies-Rabbit polyclonal antisera were raised against a synthetic peptide spanning amino acids 281-309 of mouse Lnk, or against the GST-Lnk fusion proteins. The antibodies were purified by chromatography on Affi-Gel-10 conjugated with either the synthetic peptide, GST-Lnk-N, or GST-Lnk-C fusion protein, followed by elution using 0.1 M glycine, pH 2.5. Antibodies against GST were absorbed from sera against GST-Lnk fusion proteins using GSTcoupled glutathione-Sepharose-4B. Affinity-purified antibodies to GST-Lnk fusion proteins were coupled to Affi-Gel-10, and used for immunoprecipitation studies. Antibodies against the Lnk peptide were used for immunoblotting.
Northern Blotting and Semi-quantitative Reverse Transcriptase-PCR-Poly(A) ϩ RNA was isolated from mouse tissues using the Micro-Fast Track kit (Invitrogen, Carlsbad, CA), fractionated on a 1% agarose/ formaldehyde gel, and transferred onto nitrocellulose membranes (Amersham). These membranes were hybridized with 32 P-labeled cDNA fragments encoding mouse Lnk (amino acids 36 -268) or mouse ␤-actin (amino acids 160 -375). For quantitative reverse transcriptase-PCR analysis, poly(A) ϩ RNA was isolated from purified cells using the Micro-Fast Track kit, and first strand cDNA template was synthesized using avian myeloblastosis virus reverse transcriptase (Stratagene) with random primers. Serial dilutions of these cDNA templates were subjected to PCR amplification using sets of primers spanning several introns of Lnk (FWD primer, 5Ј-ATGCCTGACAACCTCTACAC; REV primer, 5Ј-ATTCACACGTCTGCCTCTCT) or ␤-actin (FWD primer, 5Ј-ACACTGTGCCCATCTACGAG; REV primer, 5Ј-CTAGAAGCACTT-GCGGTGCA). Cycling parameters were 1 min at 94°C, 2 min at 64°C, and 3 min at 72°C for 34 cycles to detect Lnk mRNA, and 27 cycles for ␤-actin. PCR products were separated by electrophoresis on a 1.0% agarose gel and visualized by staining with ethidium bromide.
Chromosomal Localization-Southern blot filters of C57BL/6J ϫ Mus spretus backcross DNA panels digested with HindIII were purchased from the Jackson Laboratory (Bar Harbor, ME). The filters were hybridized with a 32 P-labeled 2-kb BglII-BamHI fragment of the Lnk gene containing exon 1. The probe detected a restriction fragment length polymorphism, previously defined using test digestions, that distinguishes C57BL/6J mice from the M. spretus strain (a 10.5-kb fragment in C57BL/6J DNA, and 8-kb and 2.5-kb fragments in M. spretus DNA, respectively). The results of these hybridizations were scored blindly, and forwarded to the Jackson Laboratory Backcross DNA Panel Map Service, which used these data to position the Lnk gene.
In Vitro Phosphorylation Assay-Baculovirally expressed murine p56 lck was prepared as described (46). High Five cells containing baculovirally expressed affinity-tagged ZAP-70 (47) were generously provided by Dr. L. E. Samelson (National Institutes of Health, Bethesda, MD). ZAP-70 was isolated via immunoprecipitation with the 9E10 monoclonal antibody, which recognizes the Myc affinity tag, as described elsewhere (48).
Phosphorylation of GST-Lnk-N with recombinant p56 lck and 9E10 precipitates of ZAP-70 was performed for 1 h at 30°C in 50 mM Tris, pH 7.5, 10 mM MnCl 2 , 0.1% Nonidet P-40, 250 M ATP with [␥-32 P]ATP added to 4400 dpm/pmol. Phosphoproteins were separated, and unreacted [␥-32 P]ATP removed, by SDS-PAGE on 12% gels, and transferred to nitrocellulose. Phosphoproteins of interest were identified by autoradiography. Tryptic digestion was performed on the membrane. The resulting peptides were separated on two-dimensional phosphopeptide maps and then subjected to IMAC (immobilized metal affinity chromatography)-HPLC (high performance liquid chromatography)-ESI (electrospray ionization)-MS (mass spectrometry) analysis as described elsewhere (48) with the following modification: the second dimension ascending chromatography buffer for the two-dimensional phosphopeptide mapping was 55:45:40:9, butan-1-ol, pyridine, water, acetic acid Transgenic Mice-The DNA sequences surrounding the initiation codon of the mouse Lnk cDNA were modified by PCR to provide a better consensus sequence (49) using the following two primers: FWD primer (containing a BamHI site), 5Ј-AAGGATCCACCATGCCTGACAACCTC-TACACC, and the Lnk-N REV primer. A BamHI-SacI fragment of the modified cDNA was used to replace a SacI fragment containing the initiation codon of the original Lnk cDNA. The Lnk transgene was produced by ligation of the combined Lnk cDNA (containing 8 base pairs of modified 5Ј-untranslated region, the entire coding region and 50 base pairs of 3Ј-untranslated region) into the BamHI cloning site of the p1017 vector (50), which consists of the murine lck proximal promoter and a human growth hormone gene cassette. The Lnk transgene, purified as a NotI fragment, was injected into (C57BL/6J ϫ DBA/2J) F2 mouse zygote pronuclei as described previously (51). Transgenic founders were detected by hybridization of genomic DNA with an human growth hormone probe, and stable lines of mice were generated by backcrossing founders with C57BL/6J mice.

RESULTS
Isolation of the Mouse Lnk cDNA-To pursue the importance of Lnk in immune physiology, we determined the structure of mouse Lnk, thereby providing entree to the set of immunological reagents available in this species. A full-length cDNA encoding mouse Lnk was isolated from a thymocyte cDNA library and its nucleotide sequence determined using standard techniques. Not surprisingly, Lnk is well conserved between mouse and rat. The mouse Lnk cDNA sequence is 95% identical with that of rat (data not shown; the mouse Lnk cDNA sequence is available through GenBank, accession number U89992). The deduced amino acid sequences of mouse and rat Lnk share 96% identity and 98% similarity overall, with com- plete identity within the SH2 domain (Fig. 1A). Possible tyrosine-phosphorylation sites suggested in this paper (see below) and by Huang et al. (44) are also perfectly conserved. By comparing the Lnk amino acid sequence with the GenBank data base (translated in all possible reading frames), we identified several proteins with significant similarities to either the N-terminal domain of Lnk or to its SH2 domain (Fig. 1B). Among these, the recently characterized SH2-B protein, which was isolated in a yeast "tribrid" screen by virtue of its ability to bind the immunoreceptor tyrosine-based activation motif (ITAM) of the Fc⑀RI␥ subunit (52), was especially similar. Although no functional data have yet been reported regarding SH2-B, simple structural comparison suggests that Lnk might also interact with phosphorylated ITAM-like motifs to link signal-transducing molecules to cell surface receptors. Besides SH2-B, the Lnk SH2 domain also appears related to those of the protein-tyrosine phosphatase corkscrew, the adaptor proteins Shc and Nck, and the protein-tyrosine kinases Srm and Abl.
Expression of Lnk mRNA-RNA blotting was performed to characterize the tissue distribution of the Lnk mRNA ( Fig. 2A). Two transcripts of 5 and 3 kb were strongly expressed in spleen, lymph node, and bone marrow, and found at low levels in thymus. A fair amount of Lnk mRNA was also detected in lung, liver, and kidney. Lnk has been reported to be expressed preferentially in lymphocytes (44); however, the level of transcript accumulation in lung and liver is consistent with the existence of other populations of Lnk-expressing cells in those organs. To assess the pattern of Lnk expression during lymphocyte development, we purified thymocyte subpopulations based on their discrete patterns of CD4 and CD8 expression (Fig. 2B). CD4 Ϫ 8 Ϫ , CD4 ϩ 8 ϩ , CD4 ϩ 8 Ϫ , and CD4 Ϫ 8 ϩ thymocytes were purified by sorting, and expression of Lnk in each subpopulation was analyzed using a semi-quantitative reverse transcriptase-PCR approach. The PCR primers employed in amplification spanned five introns of the mouse Lnk gene (see below), and hence the products amplified from cDNA (1.0-kb) could be easily distinguished from those arising through amplification of contaminating genomic DNA (1.6-kb). In each case, the amount of template cDNA present was normalized between samples using PCR detection of ␤-actin cDNA. Interestingly, Lnk transcripts were present at essentially equivalent levels throughout thymocyte development, including in the CD4 Ϫ 8 Ϫ population where few cells express the T cell antigen receptor. Analysis of splenic subpopulations revealed that Lnk mRNA accumulates maximally in B lymphocytes (B220 ϩ cells), reaching levels nearly 5-fold greater than those seen in mature CD3⑀ ϩ T cells (Fig. 2B). As was noted in studying rat Lnk (44), several alternatively spliced transcripts of unknown significance were found at low levels in lymphoid tissues. To gain an appreciation for these transcripts we also obtained clones of the mouse Lnk gene.
Organization of the Mouse Lnk Gene-We screened a 129SV genomic library in bacteriophage using the Lnk cDNA fragments as probes. The restriction map of a 30-kb region containing the mouse Lnk gene was constructed by analysis of five overlapping bacteriophage clones (Fig. 3A). Restriction maps for BamHI, EcoRI, HindIII, and BglII were determined and analyzed at greater resolution by Southern blotting with cDNA probes and specific oligonucleotides, or by PCR. In all cases, the sizes of subcloned restriction fragments corresponded with the sizes of cross-hybridizing bands seen on genomic blots analyzed under high stringency conditions (data not shown). In addition, we found no clones harboring intron-containing or processed pseudogenes. These results suggest that Lnk exists as a single copy in the haploid mouse genome. All exons and introns span-ning the translated region of the Lnk gene were sequenced (Fig.  3B). For convenience, the most 5Ј exon, which contains untranslated region sequences mapped by PCR and DNA blotting, has been tentatively designated as exon 1, however, the in vivo transcription start site has not yet been defined. Comparison of genomic and cDNA sequences revealed that the Lnk gene is divided into at least 7 exons and 6 introns. The coding region of Lnk resides on 6 exons (exons 2-7), encompassing just 1.6 kb of the mouse genome. The sequences surrounding all exon-intron boundaries fulfilled typical AG/GT rules (53). In previous studies of rat Lnk, several rarely expressed cDNAs termed Lnk3 and Lnk4 2 (GenBank accession numbers U24654 and U24655, respectively) were identified. Alignment of these sequences demonstrated that each contains one or more of the small introns which punctuate the mouse sequence, confirming that these rare rat sequences represent partially processed transcripts (data not shown). We were unable to adduce evidence that these rare transcripts direct the synthesis of a protein product.
Chromosomal Localization of the Mouse Lnk Gene-We determined the chromosomal localization of the mouse Lnk gene 2 J. Hayashi, unpublished data.

FIG. 2. Expression of Lnk mRNA in various tissues and lymphocytes.
A, tissue distribution of Lnk mRNA. Poly(A) ϩ RNA from the indicated tissues was analyzed by Northern blotting using a mouse Lnk cDNA fragment as a probe (upper panel). The same filter was stripped and rehybridized with a ␤-actin probe to estimate the relative amount of RNA in each lane (lower panel). The relative mobilities of 28 S and 18 S rRNA are shown on the left. B, semi-quantitative reverse transcriptase-PCR analysis of Lnk expression in lymphocyte subsets. Thymocytes or splenocytes were separated by flow cytometry on the basis of the expression of the indicated surface markers. cDNA was prepared from purified cells, and serial dilutions (3-fold) of the cDNA template were subjected to PCR analysis using sets of primers to amplify Lnk (upper panel), or ␤-actin (lower panel) cDNA for comparison of cDNA quantities in each sample. Purity of separated thymocyte subpopulations was: CD4 Ϫ 8 Ϫ , 99%; CD4 ϩ 8 ϩ , 99%; CD4 ϩ 8 Ϫ , 99%; and CD4 Ϫ 8 ϩ , 96%, and that of splenocyte subpopulations were B220 ϩ : 99%, and CD3 ϩ : 99%.
using an interspecific backcross DNA panel (54). Restriction fragment length polymorphisms within Lnk that distinguish C57BL/6J DNA from M. spretus DNA were scored by blotting and hybridization using a probe containing exon 1, and used to follow the segregation of Lnk alleles within an interspecific backcross that has been characterized at the Jackson Laboratory. This analysis permitted assignment of Lnk to mouse chromosome 5. Complete co-segregation was observed with Tbx3 (T/omb homologous domain containing gene 3 (55)) and NOS1 (neuronal nitric oxide (NO) synthase (56)); all three genes lie approximately 17 cm centromeric to Epo, the gene encoding erythropoietin (Fig. 4). There are no known mouse mutants, other than those bearing targeted disruptions of NOS1, that map to this precise region (57). Interestingly, the human NOS1 gene has been reported to localize within the 12q24.2 region (58, 59), which adjoins a site of frequent chromosomal abnormalities in some malignant diseases (see "Discussion").
In Vitro Tyrosine Phosphorylation of Lnk by Lck and ZAP-70 -Lnk is reported to become rapidly tyrosine-phosphorylated upon TCR stimulation. Among those kinases that might catalyze this event, Lck and ZAP-70 are especially prominent. Both of these kinases associate physically and functionally with the TCR, and both become rapidly activated after TCR cross-linking (10 -13). We therefore examined whether purified GST-Lnk fusion proteins containing the N-terminal (GST-Lnk-N) or the C-terminal domain of Lnk (GST-Lnk-C) can serve as substrates for Lck and Zap-70 in vitro. As shown in Fig. 5A, GST-Lnk-N was phosphorylated by both purified Lck and ZAP-70. Two-dimensional phosphopeptide mapping showed that both kinases phosphorylated the same peptide from GST-Lnk-N (Fig. 5B). Further analysis using the IMAC-HPLC-ESI-MS system (48) identified tyrosine 6 as the residue phosphorylated by Lck and ZAP-70 ( Fig. 5C and data not  shown). GST-Lnk-C also appeared to be phosphorylated by Lck (data not shown), however, we were unable to identify the phosphorylated tyrosine residue unambiguously in this case.
Tyrosine 297 would appear to be an attractive target for phosphorylation within the C-terminal domain, since this residue lies embedded within a sequence that resembles the consensus sequence of ITAMs (Y-X-X-L/I), a known in vitro target for Lck. Moreover, a synthetic phosphotyrosine 297-containing peptide bound efficiently to Grb2, PLC-␥1, and phosphatidylinositol 3-kinase, molecules that also bind rat Lnk (44).
Consequences of Lnk Overexpression in Thymocytes-To investigate the function of Lnk in vivo, we generated transgenic mice expressing high levels of Lnk in thymocytes using the thymocyte-specific lck proximal promoter (50). In all, we examined 7 independent transgenic founder animals and 13 lines of mice expressing transgene-encoded Lnk protein. To assess the importance of Lnk abundance on lymphocyte behavior, we used antisera directed against a peptide to quantitate the level of Lnk expression by immunoblotting, and compared the values obtained in this way with a standard immunoblot prepared using serial dilutions of heterologously expressed GST-Lnk-C fusion protein. Whereas normal mouse thymocytes contain 2.5 ϫ 10 1 Lnk molecules per cell, the levels in transgenic animals ranged from 1.2 ϫ 10 2 to 1.5 ϫ 10 4 molecules per cell. As expected, expression of the p1017lnk transgene was confined almost entirely to thymocytes, with some modest expression also observed in peripheral T cells. Thymuses from p1017lnk transgenic mice exhibited normal cellularity, and the proportion of thymocyte subpopulations was in most cases also grossly normal as assessed by CD4 and CD8 expression. However, two lines of transgenic mice, expressing especially high levels of transgene-derived Lnk, showed a decreased number of single-positive thymocytes as compared with normal littermate control mice. Data obtained from a representative individual of one of those transgenic lines (the highest-expressing line) are shown in Fig. 6, and our experience with such mice is summarized in Table I. This reduction in single-positive thymocytes was only observed in mice expressing Ͼ1.2 ϫ 10 4 Lnk molecules per cell.
Thymocytes from all transgenic mice proliferated as well in response to anti-CD3 stimulation as did those from normal mice, and the levels of CD3 and CD69 were comparable with those seen in normal thymocytes (data not shown). In the highest expressing line, Lnk levels were 600-fold elevated over normal on a per cell basis. It should be noted, however, that the molar abundance of Grb2 in such thymocytes (2.3 ϫ 10 4 molecules per cell) exceeded that of transgenic Lnk (1.5 ϫ 10 4 molecules per cell) by an appreciable margin (data not shown). Hence we would not expect the Lnk SH2 domain to compete effectively for binding sites ordinarily occupied by Grb2. As was seen in thymocytes, splenocytes expressing transgene-derived Lnk were present in normal numbers, and responded normally to anti-CD3 stimulation (data not shown).
Lnk Is Not a Major Tyrosine-phosphorylated Protein upon TCR Stimulation-It has been shown that Lnk becomes tyrosine-phosphorylated upon TCR/CD4 stimulation in rat lymph node T cells. We investigated whether the overexpression of Lnk in thymocytes affects TCR-induced phosphorylation of endogenous substrates on tyrosine. Thymocytes were stimulated with a combination of anti-CD3 and anti-CD4 mAbs, and phosphotyrosine-containing proteins were visualized by SDS-PAGE and immunoblotting (Fig. 7A). Augmented expression of Lnk did not measurably affect the induced tyrosine phosphorylation of cellular proteins. Of particular note, although the transgenic animals contain extraordinarily high levels of Lnk, no increase in the abundance of phosphotyrosinecontaining proteins was observed in the 36 -38-kDa region where Lnk migrates. Immunoprecipitation using anti-Lnk antibodies revealed that Lnk is rapidly (albeit weakly) phosphorylated on tyrosine following TCR stimulation of thymocytes (Fig. 7B). However, the abundance of this phosphotyrosinecontaining species was not increased in the p1017lnk transgenic thymocytes. Since the pattern of accumulation of phosphotyrosine-containing species in the 36-kDa range remained unaffected despite massive increases in the intracellular Lnk concentration, we suspected that Lnk and pp36 might be different proteins. To examine this hypothesis, we separated Lnk from pp36 using anti-Lnk antibodies or a GST-Grb2 fusion protein. Antibodies raised against Lnk immunoprecipitated Lnk effectively, but did not absorb or immunoprecipitate pp36 from cell lysates. Reciprocally, a GST-Grb2 fusion protein bound pp36 and absorbed pp36 proteins from lysates, but did not bind Lnk at all (Fig. 7C). Moreover, Lnk was phosphorylated very weakly under the conditions that we employed while pp36 was strongly phosphorylated, and Lnk migrated more slowly in SDS-PAGE than did pp36. These results demonstrate that although Lnk is an SH2 domain-containing protein that becomes phosphorylated on tyrosine during T cell activation, it is not the easily identified, phosphotyrosine-containing species in activated T cells that migrates at 36 kDa. DISCUSSION It has been proposed that Lnk, a cellular adaptor protein expressed preferentially in lymphoid tissues, becomes tyrosinephosphorylated upon TCR stimulation and participates in TCR signaling by associating with Grb2, PLC-␥1, and phosphatidylinositol 3-kinase (44). We therefore sought to elucidate the function of Lnk in lymphocytes by isolating the mouse Lnk cDNA, clarifying the genomic structure and chromosomal localization of the Lnk gene, and defining the basic features of Lnk phosphorylation through in vitro studies and the analysis of transgenic mice. Our studies suggest that although Lnk may participate in TCR signaling, its functions are in no way limiting during T cell development or activation.
Insights into Lnk Function from Structural Studies-The overall structure of Lnk is well maintained between mouse and rat. The SH2 domain and possible phosphorylation sites are especially closely conserved. These characteristic sequences are also conserved in human Lnk. 2 In addition, the putative phosphorylation sites of Lnk served as satisfactory substrates in vitro for two protein-tyrosine kinases known to be involved in TCR signaling, Lck and ZAP-70. These observations support the hypothesis that Lnk participates in a TCR-coupled signaling pathway and that tyrosine phosphorylation is crucial for its functions. However, TCR cross-linking induced only very modest tyrosine phosphorylation of Lnk in both normal thymocytes and in thymocytes expressing high levels of transgene-derived Lnk protein. Phosphorylation of Lnk was also not prominent in splenic T cells from normal or Lnk transgenic mice. 3 In light of these results, we believe that Lnk abundance in no way limits T cell responsiveness, perhaps because Lnk ordinarily participates in relaying signals other than those which we tested. It is also conceivable that the protein-tyrosine kinases responsible for Lnk phosphorylation, or binding sites that provide access to these protein-tyrosine kinases, are limited, such that only a fraction of endogenous Lnk protein can sustain phosphorylation.
Interestingly, the SH2 and N-terminal domains of Lnk are most homologous to another 80-kDa adaptor-like molecule, SH2-B, that was identified by virtue of its ability to bind a tyrosine phosphorylated ITAM of the Fc⑀RI␥ chain (52). Although SH2-B mRNA is not expressed in splenocytes and the function of the SH2-B protein is unknown, the structural similarity between SH2-B and Lnk, particularly in their unique N-terminal domains, suggests that these proteins participate in related signaling pathways. The availability of mouse GST-Lnk fusion proteins should provide a basis for the identification of cellular proteins with which Lnk, and perhaps SH2-B, interact.
Our experiments also help to resolve the relationship between Lnk and the long sought after pp36 protein, the prominent tyrosine-phosphorylated species in TCR-stimulated T cells. These proteins share similar molecular weights, become phosphorylated on tyrosine, and bind (at least in vitro in the case of Lnk) to the SH2 domains of Grb2, PLC-␥1, and phosphatidylinositol 3-kinase. However, our results clearly demonstrate that Lnk and pp36 are distinct. Anti-Lnk antibodies precipitate Lnk, but not pp36. On the other hand, GST-Grb2 fusion proteins bind pp36 efficiently but barely interact with Lnk. Finally, Lnk and pp36 differ with respect to mobility in SDS-PAGE, and the extent to which each is phosphorylated after TCR cross-linking. Enumeration of those proteins phosphorylated during T cell activation has led to the discovery of ZAP-70 and SLP-76, both of which play important roles in TCR signaling. Our results should encourage renewed efforts aimed at the purification of pp36.
Lnk Overexpression Is Benign-In general, previously studied adaptor proteins, when expressed in cells at high levels, stimulate dramatic changes in cell behavior. For example, augmented expression of Shc triggers substrate-independent growth (i.e. transformation) in NIH3T3 cells (60). Similarly, the magnitude of NFAT translocation following TCR stimula-FIG. 6. Flow cytometric analysis of thymocytes from Lnk transgenic and normal mice. Thymocytes from 5-to 7-week-old p1017lnk transgenic mice (TG) and normal littermate controls were stained with FITC-anti-CD8 and PE-anti-CD4, and analyzed by flow cytometry. Shown are representative dual parameter histograms. Lnk transgenic mice had reduced representation of mature CD4 ϩ 8 Ϫ thymocytes compared with normal mice. tion increases substantially in Jurkat lymphoblasts expressing high levels of SLP-76, especially when the Vav protein is also present at high levels (25,27). We therefore expected that augmented expression of Lnk in the thymocytes of transgenic mice would influence the outcome of TCR stimulation, and thereby perhaps perturb T cell development. In fact, Lnk expression in transgenic thymocytes proved surprisingly benign. Massive accumulation of Lnk, to levels 600-fold that ordinarily present, did not alter the proliferative response of thymocytes to TCR stimulation, the pattern of proteins phosphorylated on tyrosine, or the mobilization of intracellular calcium (Fig. 7 and data not shown). In transgenic mice expressing especially high levels of Lnk, the proportion of CD4 ϩ 8 Ϫ mature thymocytes was slightly decreased. This observation may provide a clue to the normal biological function of Lnk, suggesting that it acts to relay signals received during thymocyte differentiation. This effect cannot be explained by proposing that accumulation of an SH2 domain-containing protein masks important phosphotyrosine sequences indiscriminately. For example, the endogenous level of Grb2 protein (about 2.3 ϫ 10 4 molecules per cell) exceeds that of the transgene-encoded Lnk (at best 1.5 ϫ 10 4 molecules per cell). The isolated SH2 domain of Grb2, but not that of Shc, is capable of inhibiting signal transduction of Jurkat T cell transfectants (61). Regardless of the specific basis for this phenotype, most Lnk transgenic mice manifest no abnormalities, and in view of the very high levels of Lnk protein required to provoke disturbances in thymocyte maturation, we cannot comfortably conclude that T cell development ordinarily requires Lnk. This question can be better addressed through the generation of mice bearing mutations in the Lnk gene.
Interestingly, Lnk is expressed in all thymocyte subsets, and the abundance of Lnk mRNA remains constant throughout T cell development. Hence differential expression cannot serve to regulate thymocyte development. Lnk mRNA proved to be more abundant in B cells than in T cells, which presumably explains why peripheral lymphoid organs and bone marrow contain more Lnk mRNA than does the thymus (Fig. 2). A tyrosine-phosphorylated protein of 38 kDa that binds to the Ig-␣ chain cytoplasmic domain has been observed in a B lymphoma line (62). Our experiments do not permit resolution of whether this protein represents Lnk itself, or yet another protein migrating at a similar apparent molecular weight. If a family of phosphotyrosine-containing proteins of approximately 38 kDa exists in lymphoid cells, the genes encoding these proteins differ very substantially, since exhaustive screening of both cDNA and genomic libraries with the Lnk cDNA yielded only Lnk sequences.
Potential Involvement of the Lnk Gene in Malignant Disease-We mapped Lnk to mouse chromosome 5 at a position coincident with the NOS1 and Tbx3 gene. Given the close proximity of the Lnk gene with the NOS1 gene in the mouse genome, the human Lnk gene is probably located on 12q24 of chromosome 12 where the NOS1 gene has been mapped (58,59). Interestingly, a t(5;12)(q21-q22:q22-q24) has been reported in Richter's syndrome, the most common type of acute transformation seen in patients with chronic lymphocytic leukemia (63), and trisomy 12 is the most common numerical anomaly in chronic lymphocytic leukemia (64). A dup(12)(q13-qter) has been reported in B cell-derived non-Hodgkin's lymphomas that do not carry the IgH/BCL2 recombination often seen in this disease (65). While numerous genes positioned on the long arm of chromosome 12 might contribute to the pathogenesis of these diseases, it is of interest that Lnk transcripts accumulate to high levels in B cells. Hence translocation of another gene into the Lnk locus could result in inappropriate B cell-specific transcription. Alternatively, although augmented expression of Lnk did not affect T cell proliferation, it remains possible that B cell signaling responds more emphatically to the presence of increased levels of Lnk.
Our studies indicate that Lnk, although expressed in T cells, does not subserve an important adaptor function in TCR signaling. Lnk has characteristics of what we take to be a large class of adaptor proteins, at a minimum including Shc, Grb2, SLP-76, Cbl, HS1, and almost certainly SH2-B, that must serve to organize signal transduction complexes from a variety of receptor structures in lymphoid cells. Since massive overexpression of Lnk does not perturb the function of the TCR pathway that recruits other, related adaptor proteins, proper interaction of adaptor proteins with their cognate receptors will require substantial specificity. This in turn suggests that genetic and biochemical strategies will ultimately prove effective in defining the signaling processes mediated by Lnk.