Butyrophilin-like 2 (BTNL2) is a butyrophilin family member with homology to the B7 costimulatory molecules, polymorphisms of which have been recently associated through genetic analyses to sporadic inclusion body myositis and sarcoidosis. We have characterized the full structure, expression, and function of BTNL2. Structural analysis of BTNL2 shows a molecule with an extracellular region containing two sets of two Ig domains, a transmembrane region, and a previously unreported cytoplasmic tail. Unlike most other butyrophilin members, BTNL2 lacks the prototypical B30.2 ring domain. TaqMan and Northern blot analysis indicate BTNL2 is predominantly expressed in digestive tract tissues, in particular small intestine and Peyer’s patches. Immunohistochemistry with BTNL2-specific Abs further localizes BTNL2 to epithelial and dendritic cells within these tissues. Despite its homology to the B7 family, BTNL2 does not bind any of the known B7 family receptors such as CD28, CTLA-4, PD-1, ICOS, or B and T lymphocyte attenuator. Because of its localization in the gut and potential role in the immune system, BTNL2 expression was analyzed in a mouse model of inflammatory bowel disease. BTNL2 is overexpressed during both the asymptomatic and symptomatic phase of the Mdr1a knockout model of spontaneous colitis. In functional assays, soluble BTNL2-Fc protein inhibits the proliferation of murine CD4+ T cells from the spleen, mesenteric lymph node, and Peyer’s patch. In addition, BTNL2-Fc reduces proliferation and cytokine production from T cells activated by anti-CD3 and B7-related protein 1. These data suggest a role for BTNL2 as a negative costimulatory molecule with implications for inflammatory disease.
The regulation of T cell responses is achieved through Ag-specific TCRs acting together with coreceptors that can modulate activation. One of the principal costimulatory pathways for naive T cells is initiated when CTLA-4 and CD28 engage B7 class molecules during TCR engagement by a primary signal (1). These secondary costimulatory signals can either inhibit or promote T cell responses, depending on the receptor they engage. For example, B7-1 and B7-2 can bind either CD28, resulting in T cell stimulation, or CTLA-4, resulting in T cell inhibition (2). Negative signals can contribute to the attenuation of T cell responses and so are important in regulating T cell tolerance. The balance between stimulatory and inhibitory signals is critical to an effective immune response.
In the past several years, new members of the B7 ligand and the corresponding CD28/CTLA-4 receptor families have been identified, increasing our understanding of immune cell function. A second activating receptor in T cells is ICOS (3); its ligand B7-related protein 1 (B7RP1)3 (also known as LICOS, GL50, B7-H2, or B7h) (4, 5, 6, 7, 8) is expressed in both lymphoid and nonlymphoid cells (9). A second inhibitory receptor PD-1 (10, 11) binds to either of two B7-related ligands PD-L1 (also known as B7-H1) (12, 13) and PD-L2 (or B7-DC) (14, 15). Another inhibitory receptor, B and T lymphocyte attenuator (BTLA), binds to HVEM, previously identified as a binding partner for LIGHT and lymphotoxin α. B7-H3, another B7 homolog (16, 17), binds an unidentified receptor on activated T cells and serves as a negative regulator of T cell activation (18). A more recently described B7 family member, B7-H4 (also called B7S1 or B7x) (19, 20, 21), acts as a novel costimulator and regulates the threshold of T cell activation.
Butyrophilin-like 2 (BTNL2) has been identified as a B7-like molecule within the butyrophilin gene family, located in the MHC locus in mice and humans (22). Recently, several studies have linked the gene for BTNL2 to two inflammatory diseases, sporadic inclusion body myositis and sarcoidosis (23, 24, 25); however, no function has been demonstrated for BTNL2. Although the butyrophilin family of proteins share homology with the B7 family, no immune responses have been associated with BTNL1, the first and most broadly expressed butyrophilin. Because of its close structural homology to B7-1, BTNL2 has been hypothesized to have T cell costimulatory properties (25).
In this study we characterize the expression of BTNL2 in normal and diseased tissue, and show evidence that BTNL2 down-regulates T cell activation and may play a role in maintaining tolerance and resetting the balance of the immune response in the gut.
Materials and Methods
Murine and human BTNL2 (accession nos. NM_023145 and NM_019602, respectively) were originally identified from the National Center for Biotechnology Information (NCBI) database based on homology to other B7 family molecules. RACE PCR was performed to confirm 5′ and 3′ ends of the human and mouse BTNL2 genes based on the original deposits using RACE-Ready Marathon cDNA from human colon and mouse spleen (BD Clontech). In addition, cDNAs from human T84 (colon carcinoma) cells, Caco2 (colon carcinoma) cells, and mouse colon were also used to confirm the final cDNA sequences. The final full-length coding region cDNAs of human and mouse BTNL2 were then amplified as single cDNA contigs, with the new additional 3′ sequence, and then subcloned into pCR4-Topo (Invitrogen Life Technologies) and appropriate mammalian expression vectors. Independent PCR analyses were performed to confirm each final murine and human BTNL2 sequence. The sequence results were compared with both the predicted sequence published by Stammers et al. (22) and the NCBI database. Transmembrane and signal peptide predictions were predicted with TMHMM and SignalP (www.cbs.dtu.dk/services/TMHMM-2.0 and www.cbs.dtu.dk/services/SignalP).
Quantitative real-time PCR
Expression of murine BTNL2 was measured by quantitative real-time RT-PCR using the ABI PRISM 7900HT sequence detection system (Applied Biosystems) and normalized to the expression of a housekeeping gene (HPRT). Three PCR primer sets spanning the 1/2 exon junction (forward) 5′-GCTGACTATAAAGCACCCAGATGAC-3′ (nucleotides 60–84), (reverse) 5′-GAAGATGCCCTGCTAACGTGTC-3′ (130–151), (probe) 5′-AGTGGTCGGTCCTAACCTCCCAATCTTG-3′ (90–117); spanning the 5/6 exon junction (forward) 5′-ACCCTGCAGTCCACGTGTATG-3′ (863–883), (reverse) 5′-GGCATCAGTCACCAATGAAGTC-3′ (933–954), (probe) 5′-CACGTGGCTGGAGAGCAGATGGTAGAATAC-3′ (895–924); and spanning the 7/8 exon junction (forward) 5′-GGACCTGATCAAGGTGAAACG-3′ (1422–1442), (reverse) 5′-ACACACAGCAGCAATCAGGAAA-3′ (1465–1486), (probe) 5′-ACGGACCAATGAACA-3′ (1449–1463) were used to determine relative expression of BTNL2 cDNA and confirm expression patterns. All RNA was DNase-treated (DNA-free; Ambion) before generation of cDNA using a TaqMan Reverse Transcription kit (Applied Biosystems). Using 20 ng of total RNA equivalent cDNA, samples were subject to quantitative PCR using 2× TaqMan Universal buffer (Applied Biosystems) with 200 nM forward and reverse primers and a 900 nM probe with the following PCR conditions 50°C for 2 min; 95°C for 10 min; 95°C for 15 s; 60°C for 1 min (40 times). Each PCR analysis was run in triplicate for each biological sample included in the study.
Preparation of murine BTNL2-Fc
A cDNA was constructed encoding the extracellular domain of murine BTNL2 in-frame with the human IgG1-Fc mutein (BTNL2-Fc). The human IgG1-Fc mutein encodes 98–330 aa of SWISSProt P01857 with the following mutations: L117A, L118E, and G120A. These amino acids have been shown to be critical for high affinity binding to Fc receptors (26, 27). Mutation of these amino acids reduces affinity to Fc receptors >100-fold, described in U.S. Patent Nos. 5,457,035 and WO 93/10151 (28). COS protein kinase B cells (E5) were transfected with the BTNL2-Fc mammalian expression construct cDNA with Lipofectamine 2000 (Invitrogen Life Technologies) and cultured in complete DMEM with 0.5% low Ig serum, as described by Ettehadieh et al. (29). Seven days posttransfection, supernatants were harvested and purified by protein A-Sepharose Fast Flow (Amersham Biosciences) column chromatography.
Generation of BTNL2 polyclonal Abs
Three synthetic peptides representing the extracellular murine BTNL2 protein sequence were independently KLH-conjugated, pooled, and then injected into New Zealand White rabbits for antisera production. Before immunization, preimmune serum was collected as a negative control. Antisera were collected from the immunized rabbits and reactivity verified by ELISA against the original peptides and the murine BTNL2-Fc protein.
Generation of BTNL2 mAbs
Two Lewis rats were immunized with 20 μg of murine BTNL2-Fc in equal volume with Titermax (Cytogen), in a total volume of 100 μl. Three weeks later, animals were boosted with 20 μg of the same material emulsified in IFA. Sixteen days later, one rat was i.v. boosted with 8 μg of murine BTNL2-Fc. Three days later, the rat was sacrificed and spleen removed. Spleen cells were fused to NS1 myelomas using PEG1500. Supernatants were screened by ELISA against the murine BTNL2-Fc protein and an irrelevant Fc protein (p7.5-Fc). Abs were purified over protein A-agarose (Amersham Biosciences).
Western blot analysis
Cell lysates (50 × 106 cells/ml) from splenocytes, mesenteric lymph node (MLN) cells, and Peyer’s patch (PP) cells were prepared by lysis in PBS 1% Triton X-100, with a protease inhibitor mixture (Boehringer Mannheim). Proteins were separated on 4–20% reducing Tris-glycine gel, transferred to nitrocellulose, and blocked overnight at 4°C in TBST plus 3% nonfat milk. After washing in TBST, the membranes were incubated with the murine BTNL2–77p polyclonal antisera or control sera for 1 h at 4°C, followed by goat anti-rabbit HRP (1/5000; Amersham Biosciences) and developed by chemiluminescence (Amersham Biosciences).
Frozen sections were fixed in acetone at 4°C for 10 min, washed in TBS (pH 8.0; Sigma-Aldrich) three times, then blocked for endogenous peroxidase in glucose oxidase solution (b-D+ glucose (G-5250; Sigma-Aldrich), glucose oxidase (Sigma-Aldrich), sodium azide (Sigma-Aldrich), and TBS (pH 8.0)) for 30 min at 37°C. Sections were then treated with avidin and biotin blocking solutions (Vector Laboratories) for 15 min each, rinsed in TBST and blocked in Tris/NaCl/blocking solution (TNB; PerkinElmer) for 30 min. Primary Ab was applied using anti-BTNL2 at 19.5 μg/ ml (xmuRIBD006, rat IgG2a, clone M517; Amgen) or 20 μg/ml (xmuRIBD006, rat IgG2a, clone M518; Amgen), anti-B220 at 0.025 μg/ml (01121D, rat IgG2a; BD Pharmingen), or using goat anti-human Fc (Jackson ImmunoResearch Laboratories), and incubated overnight at 4°C. Slides were washed in TBST followed by a 30-min incubation in the appropriate secondary Ab (biotinylated rabbit anti-rat IgG or anti-goat IgG; Vector Laboratories) at 2.5 μg/ml. Slides were washed in TBST then incubated for 30 min in streptavidin HRP (NEL750; PerkinElmer), followed by chromogen development for 5 min (liquid diaminobenzidine; BioGenex). Slides were counterstained with hematoxylin solution (DakoCytomation), dehydrated, cleared, and mounted in synthetic resin (PolyMount; Poly Scientific).
Naked DNA studies
Naked DNA injections were used to induce expression of murine BTNL2-Fc in vivo, as previously described (30, 31). The 10 μg of DNA (huFc.pEF100G or IgK-BTNL2-huFc.pEF100G) in sterile saline was delivered in a volume equal to 10% of the mouse body weight by tail vein injection to female BALB/c mice (The Jackson Laboratory). The construct containing BTNL2-Fc in this experiment used an IgK leader in place of the native leader to improve expression. The vector, pEF100G is Gateway adapted and uses the human elongation factor 1α promoter. Endotoxin levels were kept below 2 UEq/injection. Blood samples were collected via the tail vein from mice on days 1, 3, 7, 10, 14, and 20 (mice were injected on day 0), and serum was analyzed by ELISA for Fc expression. Tissue was collected from a subset of mice on day 7 for analysis by immunohistochemistry for the presence of Fc protein. All tissues were frozen in OCT and stored at −70°C. All studies using mice were performed in accordance with the guidelines and approval of the Amgen Institutional Animal Care and Use Committee.
In vitro analysis of CD4+ T cell proliferation
Single-cell suspensions were prepared from murine splenocytes, MLN, and PP cells. Tissues were harvested from at least five female C57BL/6 mice per experiment, and the CD4+ cells were purified using the Spin Sep CD4+ negative selection kit (Stem Cell Sciences). Purity of CD4+ cells was >90% as assessed by FACS analysis. A total of 1–2 × 105 CD4+ cells/well were added to precoated flat-bottom 96-well plates. Plates were coated with variable concentrations of anti-CD3 mAb (clone 2C11; BD Pharmingen) and 10 μg/ml goat anti-human Fc (Jackson ImmunoResearch Laboratories) in PBS at 4°C overnight. Wells were then washed with PBS and coated with 10 μg/ml of the specified amount of the indicated fusion protein for 4 h at room temperature. In the human B7RP1-Fc costimulation assay, two fusion proteins were plated simultaneously to the same well, to a final concentration of 10 μg/ml (unless indicated otherwise) in PBS. After a final PBS wash, purified CD4+ splenocytes, or MLN or PP cells were added to wells in a final volume of 200 μl of complete RPMI 1640 medium containing 10% FBS and 55 pM 2-ME. Proliferation of CD4+ cells was determined by incorporation of 1 μCi/well of [3H]thymidine during the last 6 h (splenocytes) or 16 h (MLN and PP cells) of the 72-h culture.
Human T cells were purified from human PBMC using a CD4+ T cell isolation kit from Miltenyi Biotec, resulting in a population of cells containing >90% CD4+ cells. Plates were precoated by adding 100 μl of PBS containing varying concentrations of anti-CD3 (clone HIT3a; BD Pharmingen) with or without BTNL2-Fc (10 μg/ml) or negative control protein Chinese hamster ovary (CHO)-Fc (10 μg/ml). Plates were incubated at 4°C overnight and then washed twice with PBS. Purified T cells (1 × 105) were added to precoated 96-well flat-bottom plates in a volume of 200 μl of complete RPMI 1640 medium containing 10% FBS. Proliferation of CD4+ cells was determined by incorporation of 1 μCi/well [3H]thymidine (Amersham Biosciences) during the last 6–8 h of a 72-h assay.
Proteinase K treatment of BTNL2-Fc
In a final volume of 50 μl, BTNL2-Fc protein (15 μg) was incubated with a 25% slurry of Glycine-Affi-10 (control) or Proteinase K-Affi-10 coupled resin overnight at 4°C. The resin was removed by spinning and supernatant collected. The “degraded” BTNL2-Fc protein was diluted to a final concentration of 10 μg/ml for plate coating based on pretreatment equivalent and used in an in vitro analysis of mouse CD4+ T cell proliferation as described.
In a standard anti-CD3 proliferation assay as earlier described, murine CD4+ cells were isolated from spleens and costimulated with anti-CD3, human B7RP1-Fc (5 μg/ml) and/or inhibited with murine BTNL2-Fc (10 μg/ml). After 64 h of stimulation, 100 μl of supernatant was harvested from each condition. The supernatants (25 of 100 μl) were then assayed for cytokine levels using the Linco Research mouse 22-Plex (no. MCYTO-70K-PMX22) kit containing premixed beads capable of detecting a variety of T cell produced cytokines. The plate was read with a Luminex 100 system. Each experiment was performed with triplicate biological replicates. The data shown are representative of at least two independent experiments. All results were expressed as the mean ± SEM of the biological replicates in a single experiment. Values for probability were calculated by unpaired two-tailed t test using Prism software version 4.01 (GraphPad Software). Those cytokines that demonstrated a statistically significant value of p < 0.05 (two-tailed t test) in at least two independent experiments and those with cytokine values above the reported least-detectable dose for the assay are included.
Transfections and FACS
293-MSR cells were transfected with the full-length murine cDNAs for B7RP1, CD80, PD-L2, BTLA, BTNL2, or vector only with Lipofectamine 2000 (Invitrogen Life Technologies) per the manufacturer’s instructions. Two days posttransfection, cells were harvested with nonenzymatic cell dissociation buffer (Invitrogen Life Technologies). One million cells were then stained on ice for 30 min with each murine BTNL2-Fc, HVEM-Fc, ICOS-Fc, PD-1-Fc, CTLA-4-Fc, and CD28-Fc at 10 μg/ml. ICOS-Fc, PD-1-Fc, and CD28-Fc were purchased from R&D Systems; all others were developed in-house. Following a wash, bound Fc protein was detected with PE-conjugated F(ab′)2 goat anti-human Fc (Jackson ImmunoResearch Laboratories). Samples were fixed and analyzed using a FACSCalibur (BD Immunocytometry Systems).
Cloning of BTNL2
The gene encoding BTNL2 is contained in the MHC locus in both mouse and human. BTNL2 was originally described in a genomic analysis as a putative soluble butyrophilin protein with an unknown function (22). The butyrophilin and B7 family members are ancestrally related, and past genomic analysis has identified other butyrophilin-like molecules later determined to be new B7 family members with immunoregulatory activity. Overall, amino acid sequence similarity of B7 and butyrophilin is low, but domain structure is conserved (Fig. 1 A). All members contain a signal peptide, two Ig-like domains (IgV and IgC), a transmembrane domain, and a cytoplasmic domain. Two distinguishing features of most butyrophilins from B7 family members is the inclusion of a 170 aa B30.2 domain in the cytoplasmic domain (32). Unlike most butyrophilins, this domain is not found in BTNL2, but like butyrophilins a 7 aa heptad sequence linker separating the two sets of IgC/IgV domains is present.
Full-length cloning and RACE PCR analysis was performed to confirm the 5′ and 3′ end of the putative human and mouse BTNL2 genes. Human and mouse BTNL2 are encoded by eight exons, and have a 64% identify score (Fig. 1,B). The original published GeneScan analysis theorized only a soluble protein containing the signal peptide, Ig-like domains and heptad linker. We have identified a full-length transmembrane cDNA of murine BTNL2 encoding a signal peptide, two Ig-like domains, heptad linker, two additional Ig-like domains, a transmembrane domain, and a cytoplasmic tail (Fig. 1, B and C). A splice variant missing the second Ig-like domain in the extracellular region (exon 3) was also identified. The significance of this finding is unknown but a similar human variant was also identified. Cloning of the human counterpart yielded two full-length transmembrane alternative splice variants with either exon 2 or exons 2 and 3 deleted. A single cDNA representing a full-length transmembrane version of human BTNL2 with all four Ig-like domains was not identified. PCR of various exon combinations (i.e., exons 1–3, 2–6, 4–6) yielded products that suggest a full-length version does exist in human, although it may not be the dominant form, as it appears to be in the mouse.
Expression analysis of BTNL2 in normal tissue
Quantitative RT-PCR was used to define the expression pattern of BTNL2 in normal C57BL/6J mouse tissue (Fig. 2,A). As previously reported (22, 25), expression of BTNL2 mRNA is detected in a number of tissue types including spleen, lymph node, stomach, MLN, bone marrow, small intestine, cecum, lung, large intestine, PP, and thymus. The highest levels of expression are detected in small intestine, PP, and cecum tissue. Northern blot analysis of murine tissue detects BTNL2 predominantly in digestive tract tissue especially in the colon and small intestine (data not shown). Analysis by immunoblot with an anti-BTNL2 polyclonal Ab confirms expression of BTNL2 protein in PP (Fig. 2,B). Murine BTNL2 runs at a higher m.w. than predicted by amino acid sequence, likely due to further glycosylation or other posttranslational modifications. Western blot analysis of cell lysates of 293-EBNA cells transfected with full-length transmembrane BTNL2 detected BTNL2 protein with a m.w. consistent with the BTNL2 band detected in PP cells using the same BTNL2 polyclonal antisera (Fig. 2 C). Together, these data suggest that the additional mass is likely due to posttranslational modifications.
To further localize BTNL2 expression within tissue, an immunohistochemical survey across an array of mouse tissues was performed with a mAb to BTNL2. BTNL2+ cells are found in the tissues indicated by the RNA analysis, including small intestine, large intestine, PP, inguinal lymph node, and spleen (Fig. 2 D). In the small intestine, BTNL2 is predominantly expressed in the epithelial cells of the villi but not the crypts. Interestingly, colonic epithelial cells have been previously implicated in the suppression of T cells in the gut, and it is possible that the same may be true for epithelial cells from the small intestine (33). In other areas of the intestine and other tissues analyzed, BTNL2 is localized to cells with the morphological appearance of dendritic cells.
Expression analysis of BTNL2 in diseased tissue
Numerous mouse models of inflammatory bowel disease have been developed to mimic acute and chronic inflammatory bowel diseases in human. In the Mdr1a knockout model of inflammatory bowel disease, mice spontaneously develop colitis in a specific pathogen-free environment (34, 35). It is thought that an aberrant immune response to commensal bacterial Ags leads to the intestinal inflammation, characterized by dysregulated epithelial cell growth and massive leukocyte infiltration. Before the onset of clinical signs, changes in mRNA levels of inflammatory mediators can be detected (35). Using quantitative RT-PCR to compare the total RNA from the colons of Mdr1a knockout mice to the parental strain, FVB wild-type mice, BTNL2 is found to be up-regulated in both asymptomatic and colitic Mdr1a knockout mice (Fig. 3 A).
An increase in BTNL2 expression in the inflamed colon was further investigated at the protein level by immunohistochemistry. Numbers of BTNL2+ cells with the morphological appearance of dendritic cells were increased in the lamina propria of the inflamed colon of an Mdr1a knockout mouse compared with the expression in FVB wild-type mice (Fig. 3 B). Affymetrix array analysis of human BTNL2 expression levels in RNA isolated from patients with Crohn’s disease and ulcerative colitis is inconclusive due to low expression levels of BTNL2 in these samples (data not shown).
BTNL2 does not bind to known B7 family receptors
By analogy with other family members that have immune function, we predicted that BTNL2 might bind to cell surface receptors on T cells and/or other cells at sites of inflammation and Ag presentation. To confirm that known B7 family ligands and receptors are not a binding partner for BTNL2, cells were transfected with the full-length transmembrane version of murine BTNL2 and tested for binding to many known recombinant soluble B7 family member receptors by FACS, including CTLA-4, CD28, ICOS, PD-1, and HVEM or membrane-bound BTLA. No binding to BTNL2 was detected for any of these molecules, although each receptor bound to its known ligands (Fig. 4).
BTNL2 binds to a counterstructure present in PP and the liver
To identify tissues expressing the binding partners for BTNL2, we induced in vivo expression of Fc-tagged BTNL2, and then used immunohistochemistry to identify which tissues contained bound BTNL2-Fc. Mice were injected with either naked DNA encoding mouse BTNL2-Fc, with an IgK leader sequence to force expression of BTNL2-Fc, or a control vector for expression of Fc only. At various time points following injection of BTNL2-Fc or control Fc DNA, serum levels in individual mice were measured by an ELISA for Fc (Fig. 5 A). Equivalent serum concentrations of Fc protein were achieved in mice injected with either the BTNL2-Fc construct or the Fc control construct.
Immunohistochemistry was performed for the Fc tag in mice injected with BTNL2-Fc or control Fc expression vectors 7 days after DNA injection. Staining is evident along the sinusoidal endothelium of the liver and the vascular endothelium of PP (Fig. 5,B); this staining was not present in control mice expressing Fc alone, although similar expression levels of Fc were detected in the serum (Fig. 5 A). Immunohistochemistry was also performed using a mAb against BTNL2. Similar patterns of staining are observed, although it is important to note that this Ab detects endogenous BTNL2, as well as the tissue-bound BTNL2-Fc. Although we have not identified the binding partner for BTNL2, it appears to be expressed on sinusoidal endothelium in the liver and in the vascular endothelium within the PP.
BTNL2 is a negative regulator of T cell proliferation and cytokine production
B7 family proteins have been shown to have activity in T cell costimulatory assays that involve activating T cells through their T cell receptors with varying doses of Ag. Using costimulation with immobilized anti-CD3 mAb, the effect of murine BTNL2-Fc protein was assessed on murine CD4+ T cell responses. Proliferation of freshly purified CD4+ T cells from spleen, MLN, or PP by anti-CD3 mAb is inhibited by the addition of BTNL2-Fc, but not an irrelevant Fc protein, CHO-Fc (Fig. 6,A). BTNL2-Fc can inhibit T cell proliferation across a range of anti-CD3 doses (Fig. 6,B), and the inhibition by BTNL2-Fc is dose-dependent (Fig. 6,C). To ensure the effects of T cell proliferation were not due to copurification of contaminants or toxic products, proteinase K digestion was performed on the BTNL2-Fc protein. Digestion of the protein with proteinase K abolishes the inhibitory activity of BTNL2-Fc, demonstrating that the observed inhibitory activity is protein specific (Fig. 6,D). Viability studies demonstrate that BTNL2-Fc does not promote death of the T cells (data not shown). BTNL2-Fc can similarly inhibit human T cell proliferation (Fig. 6 E). Human and mouse B lymphocyte proliferation is not inhibited by BTNL2-Fc (data not shown).
B7RP1, typically expressed by APCs, behaves as a potent costimulatory factor in the presence of anti-CD3, through binding to ICOS, expressed by activated T cells. Addition of B7RP1-Fc and anti-CD3 to murine CD4+ T cell cultures results in enhanced proliferation over anti-CD3 alone. This costimulation can be effectively blocked by the addition of BTNL2-Fc (Fig. 7 A) but not a control protein, CHO-Fc.
We further explored the functional effect of BTNL2-Fc in this assay by analyzing cytokine production in the different culture conditions. Cytokine profiling of murine splenocyte CD4+ T cell supernatants by multiplexed Luminex-based analysis demonstrates statistically significant increases in the expression of TNF-α, GM-CSF, IL-10, IL-4, IL-6, IL-17, and IFN-γ by anti-CD3/B7RP1 costimulation of CD4+ T cells. The addition of BTNL2-Fc, when compared with the inclusion of a control protein, CHO-Fc, effectively blocks these increases (p < 0.05) (Fig. 7 B). The presence of a human Fc tag alone (CHO-Fc) does appear to have some inconsistent inhibitory activities in these assays, although no statistically significant differences are observed. Together, the inhibitory effects on both proliferation and cytokine expression demonstrate a functional role for BTNL2-Fc in decreasing the activation state of CD4+ T cells in the presence of costimulatory molecules.
Recently, BTNL2 has been linked in separate genetic studies to sarcoidosis and sporadic inclusion body myositis (23, 25). The BTNL2 SNP that results in a premature exon-splice site is associated with an elevated risk of sarcoidosis, a multisystem disorder characterized by granulomas. Originally established in a white German population (25), this association has also been independently demonstrated in an African American population (24). Sporadic inclusion body myositis is an acquired autoimmune inflammatory muscle disease diagnosed by chronic muscular weakness. Both diseases are characterized by inappropriate T cell activation (36, 37, 38). Although the functional effects of the polymorphisms in BTNL2 have yet to be established, these genetic linkages combined with the homology to the B7 family would predict a role for BTNL2 in the activation of T cells.
Despite the genetic linkages, no function has been previously demonstrated for BTNL2. Using a soluble Fc-tagged BTNL2 (extracellular domain only), we have shown that BTNL2 is capable of diminishing the proliferative responses of CD4+ lymphocytes from spleen, MLN, and PP in response to anti-CD3. In addition, BTNL2 effectively down-regulates the proliferative and cytokine expression responses to B7RP1 costimulation. As the ligand for ICOS, B7RP1 acts as a costimulatory trigger in combination with anti-CD3 to generate a high state of activation in T cells. Disruption of this pathway has been beneficial in regulating autoimmunity in a number of mouse models of disease, including colitis, multiple sclerosis, and arthritis (1, 39). The observed ability of BTNL2 to prevent B7RP1-ICOS costimulation (both proliferation and cytokine production) suggests that increased BTNL2 presents a novel avenue to modulate T cell activation in inflammatory disease. While under review, a complementary report was published confirming an inhibitory role for BTNL2 in proliferation of mouse T cells exposed to anti-CD3, and demonstrates binding of BTNL2-Fc to activated B and T cells, suggesting that its cognate is present on these cells (40). Nguyen et al. (40) also report a decrease in anti-CD3/CD28-induced IL-2 production by T cells in the presence BTNL2-Fc; this is consistent with our observed blockade of IL-2 production in anti-CD3/B7RP1 stimulated cells.
Inappropriate or inadequate expression of a negative costimulatory molecule could contribute to the abnormal inflammation and/or immune responses observed in a number of diseases. We have shown that in the Mdr1a knockout model of colitis, BTNL2 is up-regulated and widely expressed in cells with the morphology of epithelia and dendritic cells in the colons of mice with signs of colitis. Overexpression of negative costimulatory molecules likely proves valuable in controlling inflammation where down-modulation of the immune response is desirable following the initial inflammatory event. The inhibitory effects of BTNL2-Fc on T cell activation in vitro suggest that the up-regulation of BTNL2 in this mouse model of colitis may contribute to the attempted amelioration of the inflammatory response. Of note, other anti-inflammatory molecules, such as IDO and IL-10, are also up-regulated in concordance with onset of colitis, highlighting the complex balance of proinflammatory and anti-inflammatory signals that accompany disease processes.
Mice lacking either BTLA or PD-1 are predisposed to develop autoimmune phenotypes with signs of progressive arthritis and glomerulonephritis, suggesting their roles in the inhibition of inflammatory responses and in maintaining peripheral tolerance (11, 41), therefore it is possible that mice lacking BTNL2 might be more susceptible to developing spontaneous autoimmune disease. This hypothesis remains to be tested.
In summary, the characterization of a new member of the B7 family provides new insights into regulation of effector T cells, improving our understanding of organ-specific autoimmune diseases and immune evasive strategies used by tumor cells and parasites. The expression patterns of BTNL2 in normal and diseased gut tissue, together with its ability to down-regulate T cell activation in vitro, makes it an attractive protein to be implicated in promoting intestinal tolerance, and point to a role for BTNL2 in regulating T cell-mediated responses in the gut by dampening immune responses.
We thank John Delaney, Melissa Lee, Irene Corpuz, and Jane Carter for the production of BTNL2 protein, Peter Baum for initial contributions to the cloning, Li-ya Huang for immunohistochemistry, Duke Virca for proteinase K digestion of BTNL2-Fc, Dean Thompson for screening polyclonal Abs, Fergus Byrne, Jennitte Stevens, D. J. Caughey, and Todd Juan for HVEM and BTLA reagents, Deanna Hill for assistance in necropsies, and the expertise contributed by Jacque Wilk, Afsi Mozaffarian, Ryan Swanson, and the Amgen Washington Sequencing, FACS, and Array groups.
H. A. Arnett, S. S. Escobar, E. Gonzalez-Suarez, A. L. Budelsky, L. A. Steffen, N. Boiani, M. Zhang, G. Siu, A. W. Brewer, and J. L. Viney all hold stock interest in Amgen.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Abbreviations used in this paper: B7RP1, B7-related protein 1; BTLA, B and T lymphocyte attenuator; BTNL, butyrophilin-like; MLN, mesenteric lymph node; PP, Peyer’s patch; CHO, Chinese hamster ovary.