An Inhibitory Ig Superfamily Protein Expressed by Lymphocytes and APCs Is Also an Early Marker of Thymocyte Positive Selection¹

Activation of lymphocytes through Ag receptors takes place within the context of numerous other cell surface proteins that can modulate the response. This has been best characterized in terms of costimulatory signals needed for mature lymphocyte activation, although development of lymphocytes also involves Ag receptor recognition to promote or inhibit continued maturation.

T cell development in the thymus involves a TCR-initiated differentiation program, characterized by changes in a number of cell surface and intracellular proteins, which leads to functional and phenotypic maturation of the cell (reviewed in Ref. 1). Although significant progress has been made in identifying the signaling molecules and transcription factors involved in thymocyte differentiation, not all the proteins involved in this developmental process have been identified. To determine what other cell surface proteins might be regulated during thymocyte positive selection, we have studied the changes in gene expression associated with the earliest phases of this process (2). We report in this work the identification of an Ig superfamily cell surface protein that is induced during the early stages of positive selection and remains expressed on mature T cells. The gene encoding this protein was independently isolated recently from mature T cells, and was designated B and T lymphocyte attenuator (Btla)³ (3). In addition to its expression in the thymus, we show that the BTLA protein is progressively up-regulated during the pre- to mature B cell transition in the bone marrow (BM) and is expressed to varying degrees on a wide variety of hemopoietic cell types, including APC. The expression of BTLA is further up-regulated on activated T cells and down-regulated on activated B cells. We further demonstrate that a splice variant of BTLA (BTLAs), which is predicted to lack ligand binding, but maintain signaling capability, has the potential to be expressed at the cell surface. Engagement of BTLA on T cells with specific Ab inhibits TCR-mediated activation. Mice that lack the full-length form of BTLA, but maintain the splice variant, have no gross defects in hemopoietic cell development. However, T cells from these mice are hyperresponsive to TCR-mediated activation. The widespread distribution of the BTLA protein coupled with regulated changes in expression and negative signaling capability suggest a role in regulating a wide variety of lymphocyte, and possibly APC, functions.

C57BL/6J (B6), BALB/c, BALB.K, and AND TCR-transgenic (Tg) mice (4) on an H-2^b or H-2^b/k background were used in these studies. Rats were used in production of mAbs. All animals were bred at Scripps Research Institute and maintained under specific pathogen-free conditions. Experiments were conducted in accordance with National Institutes of Health guidelines for the care and use of animals and with an approved animal protocol from Scripps Research Institute Animal Care and Use Committee.

Gene microarray analysis was used to identify changes in gene expression in isolated thymocytes activated with PMA and ionomycin, as previously described (2). Two spleen-derived expressed sequence tags (ESTs), AA184189 and AA177302, were highly up-regulated upon activation of thymocytes. These ESTs could be linked through a series of other ESTs identified by sequential basic local alignment search tool searches (5). The 5′ EST AA177302 was used to synthesize a downstream gene-specific primer. A clone containing the complete coding sequence of the gene was obtained by RACE (SMART RACE cDNA amplification kit; Clontech Laboratories, Palo Alto, CA) using cDNA prepared from stimulated thymocytes as template. The full-length human homologue of BTLA was obtained from IMAGE clone 1554187.

The portion of the gene encoding the external domain of the BALB/c allele was cloned from spleen cells by RT-PCR. Total RNA was isolated using RNeasy RNA kit (Qiagen, Valencia, CA) and oligo(dT)-primed first-strand cDNA prepared with the Superscript First-Strand Synthesis System (Invitrogen, San Diego, CA). Primer pairs (Invitrogen) used were 5′-GGTACCATGAAGACAGTGCCTGCCATGC and 5′-GGATCCGCAGTCCTGCCTGGCCTCTCTTC, which add KpnI and BamHI sites to the 5′ and 3′ ends of the PCR product, respectively. The PCR product was TA cloned (Invitrogen) and sequenced (Scripps Research Institute Protein and Nucleic Acid Core Facility). The sequence encoding a splice variant of BTLA was obtained from B6 spleen cells by high fidelity RT-PCR and cloning as above, using primers 5′-GGAGATCTATGAAGACAGTGCCTGCCAT and 5′-AGATCTGCACTTCTCACACAAATGGA, which add BglII sites to the 5′ and 3′ ends. Constructs encoding fusions of yellow fluorescence protein (YFP) and BTLA were produced by high fidelity PCR using the constructs described above as template. Sequence-verified PCR products were cloned in frame into the BamHI site of pEYFP-N1 expression vector (Clontech Laboratories).

A 700-bp DNA fragment derived from the 3′ end of a Btla cDNA including ∼400 bp of coding sequence was radiolabeled with [α-³²P]dCTP using a random primer labeling kit (Roche, Basel, Switzerland). The probe was hybridized to a normalized poly(A)⁺ RNA Northern blot of mouse tissues (Origene Technologies, Rockville, MD) in ULTRAhyb hybridization buffer (Ambion, Austin, TX) overnight at 42°C and washed, according to manufacturer’s instructions. Amounts of mRNA on this blot have been previously normalized using a β-actin probe (Origene Technologies), but the blot was also stripped and reprobed with a GAPDH housekeeping probe. Hybridization was visualized using a Storm 860 imaging system (Molecular Dynamics, Sunnyvale, CA).

RT-PCR was performed as above. Other primer pairs used were as follows: CD4, 5′-CTGATGTGGAAGGCAGACAAGGATTC and 5′-CAGCACGCAAGCCAGAACACTGTCT; β-actin, 5′-GGCAACGAGCGGTTCCGATGCCCTGA and 5′-GCCACCATGGAGCCACCGATCCACA.

The DPK thymocyte cell line (6) was infected with recombinant retrovirus encoding a BTLA-YFP fusion protein and cells selected in 2 μg/ml puromycin. A hybridoma generated from the D10 T cell clone (7) was transfected by electroporation with a construct encoding a BTLAs-YFP fusion protein by electroporation. Stable transfectants were selected in 500 μg/ml G418 compound (Invitrogen). Fluorescence photomicrographs were obtained using a Zeiss Axiovert (Oberkochen, Germany) inverted fluorescence microscope and a SPOT color digital camera (Diagnostic Instruments, Sterling Heights, MI).

A soluble form of the BTLA protein was used to produce mAbs in rats and mice. Only rat mAbs PK3 (IgMκ) and PK18 (IgG1κ) are used in these studies. A construct encoding a fusion of the external domain of BTLA and the H chain constant region of human IgG (8) was transfected into 293 cells. Secreted BTLA protein was purified by protein A affinity chromatography (Pierce, Rockford, IL). Abs were produced by a modification of a previously published method (9). Rats were immunized at the base of the tail with an emulsion (0.2 ml) of CFA (Sigma-Aldrich, St. Louis, MO) containing recombinant protein (50 μg). Two weeks after immunization, medial iliac lymph node (LN) cells were harvested and fused to YB2/0 myeloma cells (10) by standard methodology. mAbs were screened by differential binding to BTLA-transfected, but not parental, cells.

The following mAbs were used in these studies (all from eBioscience, unless stated otherwise): PE-conjugated anti-CD4 (RM4-5), allophycocyanin-conjugated anti-CD4 (GK1.5), FITC- or allophycocyanin-conjugated anti-CD8α (53-6.7), FITC- or allophycocyanin-conjugated anti-B220 (RA3-6B2), FITC- or allophycocyanin-conjugated anti-TCRβ (H57-597), biotinylated anti-CD5 (53-7.3), biotinylated anti-CD69 (H1.2F3), biotinylated anti-CD86 (GL1), biotinylated anti-CD11b (M1/70), biotinylated rat IgG2a,κ isotype control, biotinylated hamster IgG isotype control, biotinylated anti-I-A(b) MHC class II (M5/114.15.2), and PE anti-CD11c (HL3; BD PharMingen, San Diego, CA). Secondary staining reagents included PE anti-rat IgM (G53-238; BD PharMingen) and fluorochrome-labeled streptavidin.

Cell surface staining was performed, as previously described (2). In some instances, cells were incubated with rat anti-mouse CD16/CD32 (eBioscience, San Diego, CA) in PBS for 10 min at 4°C to block FcR before addition of primary Abs. Stained cells were analyzed on a FACScan or FACSort, using CellQuest software (BD Biosciences, San Diego, CA). A total of 5,000–20,000 viable cells was analyzed, and the log fluorescence is shown.

LN T cells were enriched by negative selection using anti-B220 and anti-A^b class II MHC (Y3P) Abs and anti-rat IgG magnetic beads (Qiagen). The purity of the resultant cell population was routinely >90%. A total of 2.5 × 10⁵ LN T cells was cultured a final volume of 200 μl in 96 flat-bottom plates previously coated with a 1:1 ratio of anti-CD3ε (145-2C11; eBioscience) and rat IgG or anti-CD3ε and PK18 mAbs. Cell proliferation was measured after 48 h by pulsing with [³H]thymidine (1 μCi) for the final 16 h of culture.

For Ag-specific responses, splenocytes were isolated from H-2^b/k AND TCR-Tg mice, depleted of RBC by hypotonic lysis, and cultured at a density of 1.5 × 10⁶ cells/ml in 200 μl final volume with various concentrations of pigeon cytochrome c peptide (aa 88–104). Cells were harvested at 71 h and analyzed by three-color FACS staining.

For RT-PCR analysis of Ag-activated T cells, CD4⁺ LN T cells were purified from H-2^b AND TCR-Tg mice by magnetic bead negative selection using anti-B220, anti-CD8α, and anti-A^b class II MHC (Y3P) Abs. T cells were cultured at 2.5 × 10⁵ cells/ml with 2.5 × 10⁶ cells/ml irradiated (3300 rad) B10.BR splenocytes in the presence of 100 nM pigeon cytochrome c peptide. After 3 days, viable cells were isolated over Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) and total RNA was prepared.

In other experiments, B6 splenocytes were stimulated with Con A (2.5 μg/ml) or LPS (50 μg/ml) for 48 h and then analyzed.

BMDC were produced, as previously described (11). Briefly, 2 × 10⁶ BM cells were cultured in 10 ml of complete RPMI 1640 medium (Invitrogen) containing 100 U/ml GM-CSF (RDI Research Diagnostics, Flanders, NJ). Fresh medium containing 100 U/ml GM-CSF was added on day 3 of culture. On days 6 and 8, half of the culture medium was removed and replaced with fresh medium containing 60 U/ml GM-CSF. On day 10, nonadherent cells were collected, spun down, and resuspended in fresh medium containing 100 U/ml GM-CSF with or without 1 μg/ml LPS (Sigma-Aldrich). Cells were analyzed 24 h later.

Cells (1 × 10⁷) were treated with 2 mM pervanadate for 5 min at room temperature and lysed in 1 ml of lysis buffer (PBS, 2 mM pervanadate, 1% Nonidet P-40 (v/v), 1 mM sodium fluoride, 1 mM PMSF, 2 μg/ml leupeptin, 2 μg/ml aprotinin) for 30 min on ice. Nuclei were removed by centrifugation at 2000 rpm for 10 min at 4°C, and the supernatants were further clarified by centrifugation for 30 min at 14,000 rpm. Lysates were precleared three to four times using rat IgG-coupled agarose and immunoprecipitated with rat IgG- or mAb-coupled agarose overnight at 4°C. Agarose beads were washed six times in lysis buffer and eluted with SDS sample buffer under reducing conditions. Immunoprecipitates were analyzed by Western blot, as previously described (2). Blots were probed with anti-phosphotyrosine (1/1000) (clone 4G10; Upstate Biotechnology, Lake Placid, NY), followed by HRP-conjugated goat anti-mouse κ-chain secondary Ab (1/2500; Southern Biotechnology Associates, Birmingham, AL). Blots were stripped and reprobed sequentially with anti-green fluorescence protein (1/500) (Clontech Laboratories) and anti-SH2-containing tyrosine phosphatase 2 (SHP-2) (1/2000) (BD Biosciences) Abs and appropriate secondary Abs.

PCR primers corresponding to sequences in exon 2 of the Btla gene were used to screen a 129SvJ bacterial artificial chromosome (BAC) library (ES Library Release II; Incyte Genomics, Palo Alto, CA). A cloned BAC was subsequently isolated that contained exons 1 and 2 of the Btla gene. A 13-kb KpnI fragment derived from this BAC that contained exon 1 and a 5′ portion of exon 2 was cloned into the KpnI site of pKOScrambler NTKV-1906 (Stratagene, La Jolla, CA). This targeting construct contains both positive and negative selectable markers. A BAC-derived 2.5-kb KpnI-HindIII fragment that contained the 3′ portion of exon 2 and additional downstream sequence was then cloned into the SmaI and NotI sites of the pKO-modified construct. The result is disruption of exon 2 by insertion of a neo^r cassette into the KpnI site in reverse transcriptional orientation. The targeting construct was electroporated into CMTI-1 (129 S6/SvEv) embryonic stem cells (Specialty Media, Lavellette, NJ), and gancyclovir and G418-resistant clones were screened by PCR (primers: 5′-TGGCGCTACCGGTGGATGTGGAATG and 5′-TGGAGAACAAAAACCGGAACTGATTGA). Southern blots were used to confirm gene targeting of positive clones. Chimeric mice were produced by standard means and bred to B6 mice one generation before breeding to obtain homozygous mutant animals.

Using gene microarray technology, we have characterized changes in gene expression in immature CD4⁺8⁺ double-positive thymocytes activated by PMA and ionomycin under conditions that elicit the differentiation of these cells (2). Normalized hybridization signals of cDNA to two spleen-derived ESTs, AA184189 and AA177302, were increased by 9.1- and 6.6-fold, respectively, following thymocyte activation. The mitogen-activated protein kinase (MAPK) signaling pathway has been shown to be obligatory for positive selection (reviewed in Ref. 12). Thus, we also compared gene expression in thymocytes activated in the absence or presence of the MAPK kinase inhibitor U0126. The normalized hybridization signal to both ESTs was 2.7-fold greater when cells were activated in the absence of U0126 than in its presence, indicating involvement of the MAPK signaling pathway in regulating expression of the gene(s) involved. These findings were confirmed by RT-PCR analysis of differentiating cultured thymocytes (data not shown).

EST “walking” indicated that both spleen ESTs were derived from a single gene and the full-length gene was subsequently cloned. Tissue distribution of the gene was examined by Northern analysis and, as expected, highest levels were detected in the thymus, and interestingly the spleen, with lower levels found in some other tissues (Fig. 1,A). Sequencing revealed that the gene encoded a 306-aa protein with a single Ig-like domain, a transmembrane region, and a cytoplasmic domain (Fig. 1 B). This gene was recently independently isolated by Murphy and colleagues (3) and has been designated Btla.

The predicted BTLA protein from B6 mice (accession NP_808252) has been reported to contain 305 aa, while that from 129 strain of mice (accession AAP44002) has 306 aa, as does the B6-derived protein reported in this work (Fig. 1). The difference lies in the presence or absence of a CAG codon that begins the fourth exon (Fig. 1,B). In the genomic sequence of the Btla gene, this codon is preceded by an additional CAG triplet. Thus, splicing to the acceptor AG consensus site in the first CAG codon yields an mRNA that encodes the sequence -LITVS-, while splicing to the second CAG codon would yield the shorter translated sequence -LIIS-. We have observed both sequences by RT-PCR, and thus it is possible that a single cell expresses both isoforms of the protein. Six potential N-linked glycosylation sites are present in the predicted sequence (Fig. 1 B), and glycosidase treatment of immunoprecipitated protein verified that BTLA is indeed a glycoprotein (data not shown), consistent with a previous report (3).

Database searches identified a human kidney-derived clone (accession AI792952 and AA931122 for 5′ and 3′ sequencing, respectively) that upon sequencing was found to encode the full-length human homologue of Btla (data not shown). The predicted sequence of the human protein external domain sequence is ∼50% identical with its murine counterpart. In addition, both proteins contain three tyrosines in the cytoplasmic domain that are embedded in conserved motifs of 8, 9, and 13 aa, respectively (Fig. 1 B). The residues surrounding the tyrosine at position 274 fit the consensus sequence (V/I)xYxxL of an immunoreceptor tyrosine-based inhibitory motif (ITIM) (13). The sequence surrounding tyrosine 299, TxYxxI, might also function in this regard, although it is also similar to motifs in the cytoplasmic domains of CD150 family members that bind SLAM-associated protein and EAT-2 adaptor proteins (14, 15).

An alternatively spliced variant of BTLA (BTLAs) has also been detected by us (Fig. 1, B and C) and others (3). The spliced transcript eliminates the second exon that encodes the Ig-like domain (Fig. 1,B). The Btla-s transcript was detected in isolated B, CD4 T, and CD8 T cells (Fig. 1,C). To determine whether this transcript could produce a cell surface protein, a construct encoding a fusion of BTLAs and YFP was produced. BTLAs-YFP was expressed at the cell surface in transfected cells (Fig. 1 D). An additional larger mRNA species was also detected by RT-PCR in some experiments (data not shown). This transcript is caused by a cryptic splice site that results in insertion of intronic sequence 5′ to exon 3. A translational stop codon in this sequence precludes this transcript from coding for a cell surface protein.

To determine the expression pattern of BTLA in hemopoietic cells, mAbs were generated by immunizing rats and mice with a recombinant secreted form of the protein. Both sets of mAbs gave comparable results, although rat monoclonal PK3 was found to be the best reagent for cell surface staining and was used for FACS analysis, while PK18 was used for biochemical and functional assays. The specificity of PK3 and PK18 mAbs was confirmed by specific staining of a transfected thymocyte cell line that expressed a BTLA-YFP fusion protein (Fig. 2,A). In addition, there was a correlation between expression of YFP and mAb staining of transfected cells, further confirming the specificity of the Abs (Fig. 2 A).

PK3 (Fig. 2,B) and PK18 (data not shown) mAb stained cells derived from B6, but not BALB/c mice. This was found to be the result of an allelic variation in the external domain of BTLA between these mouse strains (Fig. 1 B). The B6 allele has been designated Btla^b (encoding BTLA.2), and that found in BALB/c mice Btla^a (encoding BTLA.1).

PK3 was used to analyze expression of BTLA during T and B cell development. In a wild-type thymus, a subset of thymocytes was PK3⁺ (Fig. 3,A). In contrast, PK3⁺ thymocytes were absent in TCRα chain-deficient mice (Fig. 3 A). These animals lack single-positive (SP) thymocytes due to a failure of TCR-mediated positive selection. Conversely, the majority of thymocytes in AND TCR-Tg mice on a positively selecting background were PK3⁺, consistent with the up-regulation of BTLA as a consequence of positive selection.

BTLA is expressed by a subset of double-positive thymocytes, and is up-regulated coincident with the positive selection marker CD69 (16) (Fig. 3,B). Similar results were found when TCR up-regulation was used as a marker of positive selection (data not shown). However, unlike CD69, expression of BTLA was maintained on all TCR⁺ CD4 and CD8 SP thymocytes, with somewhat higher levels on CD4 SP cells (Fig. 3,B, and data not shown). BTLA was not detected on precursor CD4⁻CD8⁻ thymocytes (Fig. 2 and data not shown).

Staining of BM cells for BTLA revealed that the protein was first detected at low levels in B220⁺IgM⁻ cells, a subset that contains pro- and pre-B cells. Somewhat higher expression was found in B220⁺IgM^high cells (immature B cells) with the highest level of expression in mature B cell subsets (Fig. 3 C).

The staining of CD4 and CD8 SP thymocytes with PK3 suggested that BTLA would also be expressed by T lymphocytes in peripheral lymphoid tissue. Indeed, all T and B lymphocytes in the spleen (Fig. 4,A) and LN (data not shown) were PK3⁺. B cells express ∼10-fold higher amounts of BTLA on the cell surface than T cells (Fig. 4,A). As was observed in the thymus, CD4 T cells express somewhat higher amounts of BTLA than do CD8 T cells (Fig. 4,A). BTLA is also expressed by minor subsets of T cells in the spleen, including γδ T cells (Fig. 4,B) and CD25⁺CD4⁺ T cells (Fig. 4,C), at comparable levels to the majority αβ T cell population. In contrast, NK cells have low levels of BTLA expression (Fig. 4 D).

BTLA expression in B cells decreased 3- to 10-fold following activation with LPS, while expression on Con A-activated CD4⁺ or CD8⁺ T cells increased ∼10-fold (Fig. 4,E). AND TCR-Tg CD4 T cells activated by specific Ag also up-regulated BTLA, and the extent of up-regulation was dependent on the concentration of Ag (Fig. 4,F). Interestingly, a variable frequency (ranging from ∼1 to 8%) of ex vivo CD4 T cells expressed high levels of BTLA (Fig. 4,C). The majority of these cells were negative for activation markers CD25 and CD69 (75 and 90% CD25⁻ and CD69⁻, respectively, in Fig. 4,C). However, approximately half of the BTLA^high CD4 T cells were larger cells, as assessed by forward light scatter (45% in Fig. 4 C).

To determine whether BTLAs was also expressed by activated T cells, RT-PCR was performed on AND TCR-Tg T cells activated by peptide Ag. Both full-length and spliced mRNAs were detected in activated CD4⁺ T cells (Fig. 4,G), and no significant difference in the ratio of the two isoforms was apparent when resting and activated cells were compared (compare with Fig. 1 C).

TCR⁻B220⁻ class II MHC-positive cells in the spleen expressed BTLA, suggesting that APC may also express this protein (Fig. 5,A). Indeed, CD11b⁺ macrophages were found to be uniformly PK3⁺, with levels of expression comparable to resting T cells (Fig. 5 A).

To determine whether DC also express BTLA, BMDC were prepared. The majority of cells in these preparations are CD11c⁺, class II MHC^−/low, and CD86⁻, the phenotype of immature DC (Fig. 5,B). Various inflammatory and other stimuli can activate DC, leading to a maturation process that up-regulates expression of class II MHC and costimulatory molecules of the B7 family, and results in increased T cell stimulatory potential. Thus, LPS treatment of BMDC resulted in a majority cell population that was class II MHC^high and CD86⁺ (Fig. 5,B). PK3 stained only a minor subset of unactivated BMDC, possibly correlating with the few mature DC that arise spontaneously in these cultures (Fig. 5,B). In contrast, LPS-mediated maturation of DC was associated with dramatic up-regulation of PK3 staining (Fig. 5 B).

The conserved tyrosine-containing motifs in the cytoplasmic domain of BTLA were considered as likely sites of phosphorylation. To test this, cells were transfected with a construct encoding a BTLA-YFP fusion protein. Although BTLA was not constitutively phosphorylated, treatment of transfected cells with pervanadate induced tyrosine phosphorylation of the protein (Fig. 6,A). ITIM motifs mediate negative signaling as a consequence of activation-induced phosphorylation and recruitment of tyrosine phosphatases, including SHP-2 (17). Indeed, SHP-2 coimmunoprecipitated with BTLA in pervanadate-treated cells (Fig. 6 A), consistent with the identification of an ITIM motif(s) in the BTLA cytoplasmic tail. A weaker association with SHP-1 was also detected (data not shown).

These results strongly suggest a negative signaling function for BTLA. To test this, the effect of anti-BTLA Ab on TCR-mediated T cell activation was assessed. Coimmobilized PK18 inhibited anti-CD3-mediated proliferation of T cells (Fig. 6,B). PK18 reacts with the BTLA.2, but not BTLA.1 protein. PK18 had no effect on activation of BALB/c (BTLA.1) T cells, thereby demonstrating the specificity of the Ab-mediated inhibition (Fig. 6 B). In this experiment, equal concentrations of anti-CD3 and PK18 Abs were used to coat plates. Interestingly, the inhibition of T cell activation was particularly pronounced at higher concentrations of Ab, perhaps reflecting the threshold of negative signaling that is necessary to overcome a productive response. Alternatively, this pattern of inhibition could be due to differences in the affinity of anti-CD3 and PK18 Abs for their respective ligands.

Because engagement of BTLA inhibited T cell activation, one would predict that loss of BTLA would result in an enhanced response. To address this, mice deficient in BTLA were produced. Exon 2, which encodes the Ig-like domain, was targeted for disruption (Fig. 6,C). Spleen cells from homozygous mutant animals (BTLA^−/−) failed to express full-length Btla mRNA (Fig. 6,D). The anti-BTLA Abs described in this work also fail to bind homozygous mutant cells (data not shown). However, because these Abs are not able to recognize the 129 strain allele of BTLA, this result only confirms that mutant mice are homozygous for the 129 Btla locus. In contrast to loss of full-length Btla mRNA, the Btla-s transcript, which splices out exon 2, was still present in homozygous mutant cells (Fig. 6 D).

Both B and T cells develop normally in BTLA^−/− mice, and no abnormalities in peripheral lymphoid organ cell subpopulations were detected in these mutant animals (data not shown). However, BTLA^−/− T cells were hyperresponsive to TCR-mediated activation (Fig. 6 E), as predicted from the ability of anti-BTLA Abs to inhibit T cell activation.

Our analysis of changes in gene expression in differentiating cultured thymocytes has led to the identification of an inhibitory member of the Ig superfamily of cell surface proteins that is up-regulated in the early stages of thymocyte positive selection. The gene for this protein was independently isolated recently and has been designated Btla (3). We describe in this study mAbs that recognize BTLA and allow a more complete picture of the expression pattern of this cell surface protein.

Our data demonstrate that BTLA can be added to a very short list of useful markers of positive selection. The coincident expression of CD69 and BTLA may reflect the involvement of MAPK signaling in up-regulation of both these proteins (18). One intriguing possibility is that the varying degrees of expression of BTLA in CD4 and CD8 thymocytes reflect differences in the strength of the MAPK signal that is thought to be important in the development of these lineages (reviewed in Ref. 12). Despite the early expression of BTLA during positive selection in the thymus, there is no gross abnormality in T cell (or B cell) development in BTLA-deficient animals. Whether this reflects compensatory mechanisms, more subtle defects in development, or a lack of function for this cell surface protein in the thymus remains to be determined. In contrast, mice lacking programmed death 1, another T cell inhibitory receptor with proposed functional and structural similarity to BTLA (3), exhibit alterations in T cell development, including enhanced negative selection and generation of TCR⁺ CD4⁻CD8⁻ T cells in a TCR-Tg model system (19).

Expression of BTLA also marks B cell developmental progression, first appearing in pre-B cells. The progressive increase in cell surface expression during B cell development may reflect continued signaling during this process, a change in expression of the transcriptional regulators that control BTLA gene expression, and/or changes in protein sorting or internalization.

BTLA is also widely expressed on peripheral mature lymphocytes, including B, αβ T, γδ T, and CD25⁺CD4⁺ T cells. The latter cell population contains regulatory T cells (reviewed in Ref. 20), and thus BTLA has the potential to impact their function as well. In contrast, NK cells express little BTLA. NK cells have ITIM-containing inhibitory receptors that bind class I MHC and prevent cytolysis of normal healthy tissue (reviewed in Ref. 21). It is possible that expression of BTLA on NK cells would otherwise interfere with the careful balance of negative and positive signaling that is critical for preventing indiscriminant killing of normal cells by these cytolytic cells.

BTLA is more highly expressed on the cell surface of B cells than T cells. There is up-regulation of the protein on the cell surface upon activation of T cells and down-regulation upon activation of B cells. The extent of BTLA up-regulation on TCR-Tg CD4⁺ T cells was more sensitive to the concentration of Ag than was up-regulation of CD69 or CD25. Thus, ∼90% of activated T cells were BTLA^high and CD69⁺ or CD25⁺ at high Ag concentration, while ∼50% of activated T cells had this phenotype at lower concentrations of Ag (Fig. 4 F). The extent of up-regulation of expression of this negative regulator on T cells may therefore be proportional to the initial activating signal and may suggest a role for the protein in dampening responses.

Interestingly, we also detected a low and variable frequency of BTLA^high CD4⁺ T cells in LN and spleen. Based on an increased cell size compared with the bulk population of T cells, BTLA^high cells may include in vivo recently activated cells that have down-regulated other activation markers. Alternatively, these cells could represent memory cells or a distinct functional subset of T cells.

BTLA is also expressed by APC, including splenic macrophages and BMDC. Interestingly, mature, but not immature BMDC express BTLA. These data indicate that DC maturation is not only associated with changes that increase the potency of the DC as a T cell activator, but also leads to changes in cell surface proteins with the potential for negative signaling. DC express other ITIM-containing proteins, including lectin-like and Ig superfamily proteins (22, 23, 24). The Ig-superfamily inhibitory receptor, FDF03 (also known as paired Ig-like receptor α), was similarly shown to be highly expressed by mature DC following activation. However, unlike BTLA, this protein was also expressed by populations of immature DC (24). In contrast, the ITIM containing lectin-like protein CLECSF6 is down-regulated upon DC maturation (23). This suggests that the particular complement of inhibitory receptors expressed by DC is altered upon maturation. The change in cell surface signaling receptors may coordinate with changes in the microenvironment, including changes in expression of appropriate ligands, as DC migrate from tissue to lymphoid organ.

It was reported previously that the Btla gene was expressed by B cells and activated T cells, but not naive T cells (3). In addition, Btla mRNA was not detected in macrophages or thymocytes in that study. Our data clearly demonstrate that BTLA is expressed by maturing thymocytes, naive T cells, and APC. Our finding that resting T cells express BTLA is also consistent with the ability of Abs to BTLA to inhibit T cell activation in a short-term assay. The difference between our results and those reported previously is likely to be the result of low sensitivity of detection of gene expression and the lack of reagents to assess protein expression in the latter study.

A number of isoforms of the BTLA protein exist, as a result of both allelic variation and alternative splicing. Duplication of a CAG sequence at the 5′ end of exon 4 causes differential splicing, resulting in two isoforms of the protein that differ in length by a single amino acid and have a minor difference in sequence. This sequence change is in a region of the protein between the Ig-like domain and the transmembrane region. Whether these protein isoforms have any functional significance remains to be determined. We also show in this work that the more radical splice variant BTLAs, which eliminates the Ig-like domain, has the potential to be expressed at the cell surface. The Btla-s transcript is expressed by B and T cells, and we have not observed any gross alteration of the ratio of full-length to Btla-s transcripts in activated T cells. The loss of this external domain of BTLA would be predicted to eliminate ligand binding, but could maintain some signaling function of the protein. The position of the coding sequence in the human gene in relation to consensus splice sites would also allow exon 1 to be spliced in frame with exon 3 (data not shown). It remains to be determined whether the BTLAs variant protein is functional in cells, but it is possible that it could act as an endogenous inhibitor of BTLA function in some circumstances by sequestering intracellular signaling partners.

In contrast to previously reported BTLA-deficient mice (3), the BTLA^−/− mice described in this work have the potential to express the BTLA protein (Fig. 6 C). Nevertheless, loss of full-length BTLA is sufficient to cause enhanced T cell activation. This suggests that the external domain of BTLA is likely to be critical for the ability of the protein to act as a coinhibitory receptor (25) during TCR-mediated activation.

BALB/c and B6 mice express different alleles of Btla (Btla^a and Btla^b encoding BTLA.1 and BTLA.2, respectively) that differ in the external domain coding region. The predicted sequence of the human BTLA external domain described in this work differs from that reported previously (3) by 3 aa in the external domain (data not shown). Thus, it is possible that there are also allelic variations in humans. The ability of the PK3 and PK18 mAbs to distinguish BTLA.1 and BTLA.2 suggests that some of the amino acid differences in these proteins are solvent exposed and could impact ligand as well as Ab binding.

Our data using anti-BTLA Abs and BTLA-deficient mice indicate that BTLA is a negative regulator of TCR-mediated T cell activation. This function is most likely mediated at least in part by recruitment of tyrosine phosphatases to phosphorylated ITIM motif(s) in the cytoplasmic tail. Increased cell surface expression of BTLA protein following T cell activation could enhance avidity for ligand, potentially tipping the balance of positive and negative signaling. A similar effect may account for the particular potency of higher concentrations of anti-BTLA Ab in inhibiting T cell responses in culture.

Interestingly, the Btla gene is located between gene loci for CD200 (Mox2) and its receptor CD200R (Mox2r), on chromosomes 16 and 3 in mice and humans, respectively. Like BTLA, both CD200 and its receptor are Ig superfamily members, and a number of CD200R protein isoforms are produced by alternate splicing (26). There is conservation of tyrosine-based motifs in the human and mouse CD200R cytoplasmic domains that mediate signaling, although they are unrelated to motifs in BTLA. Expression of CD200R is primarily restricted to myeloid cells and T cells in mice and humans (27, 28). In contrast, CD200 is expressed by a broader range of cell types, including neural cells, endothelial cells, follicular DCs, and lymphocytes (29, 30). The interaction of CD200 and CD200R inhibits granulocyte and macrophage activation, and interference in this interaction by gene targeting or Ab inhibition exacerbates autoimmune disease in murine models (27, 31). Similarly, experimental allergic encephalomyelitis is exacerbated in BTLA-deficient mice (3). Whether this effect is mediated by lymphocytes and/or APC remains to be determined. The CD200-CD200R interaction has been proposed to maintain the basal inactivated state of myeloid cells and/or regulate myeloid cell function (31, 32). Like CD200, expression of BTLA is common to B and T lymphocytes, although in this case BTLA is the receptor with inhibitory signaling function. B7x, a member of the B7 family of proteins, has been demonstrated to be a ligand for BTLA (3), but the detailed expression pattern of this protein has yet to be reported, and other ligands for BTLA may yet exist. Whether BTLA, like the linked CD200-CD200R pair, plays a more global role in regulating lymphocyte and myeloid cell interactions with each other or other cell types must await further experimentation.

We thank Parinaz Aliahmad for critical reading of the manuscript. This is manuscript 16201-IMM from the Scripps Research Institute.