Essential Roles of TIM-1 and TIM-4 Homologs in Adaptive Humoral Immunity in a Zebrafish Model Free

The TIM proteins represent one of the most important type I transmembrane glycoprotein families that share a characteristic N-terminal Ig domain of the V subset, a heavily glycosylated mucin region, a single transmembrane domain, and an intracellular cytoplasmic tail (1, 2). The TIM family in mice comprises four members (TIM-1, -2, -3, and -4), three of which are conserved in humans (TIM-1, -3, and -4) (3). These proteins were initially described as a hepatitis A virus receptor and a kidney injury molecule 1, which is an epithelial cell adhesion molecule upregulated during kidney ischemia (4–6). TIM proteins serve important functions in various biological activities, such as cellular proliferation, apoptosis, autophagy, migration, tissue regeneration, and immune regulation (7–13). Dysfunction of TIM proteins results in various diseases, such as rheumatic arthritis, asthma, nephritis, systemic lupus erythematosus, and intestinal allergy, most of which are autoimmune-related disorders. This finding suggests that these proteins play diverse roles in immunologic surveillance, homeostasis, and tolerance (14–18).

Among the TIM family members, TIM-1 and TIM-4 were particularly important in immunological activities. These proteins have long been implicated in the maintenance of immune homeostasis. Both TIM-1 and TIM-4 serve as phosphatidylserine receptors for the clearance of apoptotic cells and consequent prevention of excessive inflammation and autoimmunity (8, 19). TIM-4 expressed by medullary macrophages regulates respiratory tolerance by mediating the phagocytosis of Ag-specific T cells and contributes to the degradation of dying tumor cells by activating autophagy; hence, TIM-4 reduces Ag presentation and increases immune tolerance (9, 10). TIM-1 also acts as a P-selectin ligand with a specialized function in Th1 and Th17 cell trafficking during inflammatory responses and autoimmune disease induction (11). TIM-1 and TIM-4 have recently attracted attention because of their regulatory functions in innate and adaptive immunity. TIM-1 is predominantly expressed in activated CD4⁺ T cells in human and mouse models (20). Administration of an anti–TIM-1 mAb or a TIM-1–Ig fusion protein enhances the proliferation of CD4⁺ T cells and the secretion of IL-4 and IL-10, which are typical Th2 cytokines in adaptive immunity (7). The ectopic expression of TIM-1 during T cell differentiation promotes the production of IL-4 but not IFN-γ, a typical Th1 cytokine, in T cells (21). These findings suggest that TIM-1 preferentially participates in Th2-typic immune responses, whereas TIM-4 is exclusively expressed on APCs (7). Similar to anti–TIM-1 mAb, TIM-4–Ig fusion protein triggers the hyperproliferation of T cells (7). Treatment of T cells with TIM-4-Ig–conjugated beads or TIM-4–expressed Chinese hamster ovary cells together with anti-CD3 and anti-CD28, which mimics activated APCs, can induce T cell expansion and TIM-1 phosphorylation in T cells (22). Furthermore, TIM-4 is a natural ligand for TIM-1, and their interactions can be significantly stimulated by phosphatidylserine on exosomes (23). The above findings partially suggest that TIM-1 and TIM-4 are reciprocal molecules closely associated with the activation of adaptive humoral immunity.

However, controversial observations were reported in different research models. For example, administration of various TIM-1 mAbs to different epitopes of TIM-1 protein results in opposite effects on Th2 cell responses (2, 24). The effect of TIM-4 on T cell activation is dose-dependent; in particular, low concentrations of this protein inhibit T cell activation, whereas high concentrations stimulate this process (7). TIM-4 also inhibits naive CD4⁺ T cells but exerts no effect on activated T cells (25). Additionally, TIM-4 modulates T cell activity by limiting the proliferation of Ag-specific T cells, and either TIM-4 or TIM-1 promotes regulatory T cell development, which can functionally suppress Th2 cell differentiation and response (9). These observations suggest that the exact functions of TIM-1 and TIM-4 in adaptive immunity remain elusive. Further investigation is warranted to clarify whether these two molecules are required for the full activation of Th2-typic immunity, which includes the establishment of interactions between APCs and CD4⁺ T cells, the activation of CD4⁺ T and B cells, and the subsequent production of Ab. Clarification of this issue is indispensible to understand the function of TIM-1 and TIM-4 as costimulatory regulators in adaptive immune responses, particularly in T cell responses related to autoimmunity, peripheral tolerance, and relevant disorders.

Different research models, including lower vertebrates such as fish, should be used to characterize the functions and probe the evolutionary history of the TIM family. Accordingly, the present study determined the molecular and functional characteristics of TIM-1 and TIM-4 homologs from zebrafish (Danio rerio TIM [DrTIM]-1 and DrTIM-4), an ecologically important nonmammalian vertebrate model organism for the study of comparative immunology. As expected, DrTIM-1 and DrTIM-4 served as costimulatory molecules that were required for the full activation of adaptive humoral immunity. To the best of our knowledge, this study is the first to prove that functional TIM-1 and TIM-4 homologs exist in a lower vertebrate and to highlight a novel costimulatory mechanism underlying adaptive immunity. The findings expand the current knowledge on TIM-mediated immunity and provide a cross-species understanding of the evolutionary history of costimulatory systems from fish to mammals.

Wild-type AB zebrafish were bred in our laboratory and maintained in circulating water at 28°C under standard conditions, as previously described (26). One-year-old male and female fish with body lengths of 2–3 cm and weights of 1.0–1.5 g that exhibited healthy appearance and activity were used for the study. All animal experiments were performed in accordance with legal regulations and were approved by a local ethics committee.

The Genome and Expressed Sequence Tags databases maintained by the National Center for Biotechnology Information, Ensembl, and DFCI were used to predict TIM-1 and TIM-4 homologs in fish by using mammalian TIM-1 and TIM-4 as queries, aided by GENSCAN, SMART, and BLAST software programs as previously described (27). Total RNA was isolated from kidney, spleen, and brain tissues by using an RNAiso Plus kit (Takara Bio), and DrTIM-1, DrTIM-4, and soluble (s)DrTIM-4 cDNAs were generated through RT-PCR by using primers (Supplemental Table I) and 3′ full RACE core sets (Takara Bio) in accordance with the manufacturer’s instructions (28). The PCR products were purified from agarose gel (1.2%) by using a gel extraction kit (Qiagen), inserted into the pGEM-T EASY vector (Promega), and then transformed into competent Escherichia coli TOP 10 (Invitrogen). The positive plasmid DNA was purified following the Miniprep protocol (Omega Bio-tek) and sequenced on an ABI 3730 sequencer (Invitrogen) as previously described (29).

Genome assemblies and locations were retrieved from the University of California at Santa Cruz genome bioinformatics Web site and mapviewer in the National Center for Biotechnology Information. Gene organizations (intron/exon boundaries) were elucidated by comparing DrTIM-1 and DrTIM-4 cDNAs with genome sequences, and the figures were drawn using GeneMapper 2.5 (28). Full-length cDNAs of DrTIM-1 and DrTIM-4 were assembled using the CAP3 sequence assembly program (30). Multiple alignments and phylogenetic trees were generated using the ClustalX program (version 3.0) and the BlastP algorithm (31, 32). The signal peptides, transmembrane domain, and potential functional motifs were predicted using the SignalP 4.1 server (33), TMHMM server 2.0 (34), and PROSITE (35), respectively. N-linked glycosylation sites, O-linked glycosylation sites, and phosphorylation sites were predicted using the NetNGlyc 1.0 server, NetOGlyc 4.0 server, NetPhos 2.0 server, and GPS 2.1.1 (36, 37). Secondary and three-dimensional structures were analyzed using SMART (38, 39), SWISS-MODEL (40), and ESyPred3D Web server 1.0 (41).

The coding sequences for the extracellular region of DrTIM-1 (DrTIM-1ex) or DrTIM-4 (DrTIM-4ex) and the open reading frame (ORF) of sDrTIM-4 were subcloned into pET28a and pET41b (Invitrogen) to construct prokaryotic expression vectors (pET28a–DrTIM-1ex, pET28a–DrTIM-4ex, pET41b–DrTIM-4ex, and pET41b–sDrTIM-4) with His-tags and GST-tags (29). The ORFs of DrTIM-1 and DrTIM-4 were inserted into pEGFP-N1 (Clontech), pCMV-Tag2B (Stratagene), and pcDNA6/myc-His B (Invitrogen) to construct eukaryotic expression vectors (pEGFP–DrTIM-1, pEGFP–DrTIM-4, pCMV–DrTIM-1, and pcDNA6–DrTIM-4) with enhanced GFP, flag-tag, and myc-tag, respectively (42). The primers used in plasmid constructions are shown in Supplemental Table I. The Golgi apparatus subcellular localization plasmid (pDsRed2-Golgi) was provided by Prof. Thomas Kietzmann (University of Kaiserslautern, Kaiserslautern, Germany) (43). All of the plasmids used for transfection were prepared free of endotoxin by using the Endo-free Plasmid mini kit II (Omega Bio-tek).

pET28a–DrTIM-1ex, pET28a–DrTIM-4ex, pET41b–DrTIM-4ex, and pET41b–sDrTIM-4 were transformed into E. coli Rosetta (DE3) pLysS. Positive colonies were inoculated into Luria–Bertani medium for protein expression in accordance with a previously described method (29). The recombinant DrTIM-4, DrTIM-1, and sDrTIM-4 proteins were detected via 12% SDS-PAGE and purified through Ni-NTA agarose affinity chromatography (Qiagen) or GSTrap FF (GE Healthcare) in accordance with the manufacturers’ manuals. Six-week-old male New Zealand White rabbits were immunized with 200 μg purified recombinant DrTIM-4 and DrTIM-1 proteins, respectively, in CFA initially and then in incomplete Freund's adjuvant four times thereafter at biweekly intervals. Male BALB/c mice were immunized with 40 μg purified recombinant DrTIM-4 and DrTIM-1 proteins. One week after the final immunization, blood was collected from the rabbits and mice when the Ab titers were >1:10,000 as determined by ELISA (29). The anti–DrTIM-1 and anti–DrTIM-4 Abs were affinity purified into IgG by using a protein A–agarose column (Qiagen) and a membrane-based Ag-absorbent protocol as previously described (44, 45). Western blot was used to characterize the specificity of the Abs (26). The Abs against D. rerio MHC class II (DrMHC-II), DrIgM, DrCD4, DrCD40, and DrCD154, including mouse anti–DrMHC-II, mouse anti–DrIgM, rabbit anti–DrIgM, mouse anti–DrCD4, rabbit anti–DrCD4, rabbit anti–DrCD40, and rabbit anti–DrCD154, were produced from our previous studies (26, 29, 46).

Total RNA was extracted from tissues and leukocytes from the spleen, kidney, and peripheral blood. The transcripts of DrTIM-1 and DrTIM-4 were analyzed via quantitative real-time PCR on a Mastercycler ep realplex instrument (Eppendorf) (47). In brief, all PCR experiments were performed in a total volume of 10 μl by using a SYBR premix ex taq kit (Takara Bio). The PCR program was as follows: 1) 94°C for 2 min; 2) 40 cycles of denaturation at 94°C for 20 s, annealing at 55–65°C (according to the target genes) for 20 s, and extension at 72°C for 20 s; 3) melting curve analysis at 95°C for 15 s, 60°C for 15 s, 60°C up to 95°C for 20 min, and 95°C for 15 s; and 4) cooling at 40°C for 30 s. The relative expression levels were calculated using the 2^−∆∆CT method with β-actin for normalization. In all cases, the sample was run in triplicate parallel reactions. The experiments were repeated independently at least three times.

The distributions of DrTIM-1 and DrTIM-4 on CD4⁺ T cells and APCs were determined on the basis of the colocalization of these molecules with DrCD4 and DrMHC-II on zebrafish leukocytes via immunofluorescence staining. Through Ficoll-Hypaque (1.080 g/ml) centrifugation, leukocytes were separated from fish stimulated by 10 μg keyhole limpet hemocyanin (KLH; Sigma-Aldrich) plus 10 ng LPS (Sigma-Aldrich) from E. coli serotypes O55:B5 or unstimulated fish (26). The cells were fixed with 2% paraformaldehyde at room temperature for 10 min, blocked with 5% normal goat serum, and then incubated with primary Abs (mouse anti–DrMHC-II along with rabbit anti-DrCD4, mouse anti–DrTIM-1, or rabbit anti–DrTIM-4) at 4°C for 1 h. After washing with PBS, the cells were incubated with secondary PE- and FITC-conjugated anti-mouse or anti-rabbit Abs (Santa Cruz Biotechnology) in accordance with the manufacturer’s instructions. The cells were also incubated with 100 ng/ml DAPI (Invitrogen) at room temperature for 5 min for nucleus staining. Imaging was performed under an LSM 710 two-photon laser-scanning microscope (Zeiss, Jena, Germany) at ×630 magnification.

Cells under examination were blocked with 5% normal goat serum for 1 h at 4°C and then incubated with the defined primary Abs for 1 h at 4°C. Nonspecific rabbit or mouse IgG served as the negative control. After washing, the cells were incubated with secondary Abs (PE-goat anti-mouse and FITC-goat anti-rabbit) for 1 h at 4°C, and the fluorescence signals were determined using a FACScan flow cytometer (BD Biosciences) at 488 nm. At least 10,000 events were collected from the myeloid or lymphocyte gate (48). CellQuest software (BD Biosciences) was used for flow cytometry (FCM) analyses, and ModFit LT software was used for T cell proliferation assays (26).

Fish were i.p. injected with 10 μg KLH plus 10 ng LPS. Leukocytes were enriched from the spleen, kidney, and peripheral blood through Ficoll-Hypaque (1.080 g/ml) centrifugation as previously described (26, 47). After blocking with 5% normal goat serum, leukocytes (1 × 10⁸ /ml) were incubated with mouse anti–DrMHC-II or rabbit anti-DrCD4 Ab for 15 min at 10°C. The cells were gently washed thrice with MACS buffer (PBS containing 2 mM EDTA and 0.5% BSA), incubated with anti-mouse or anti-rabbit IgG magnetic beads (Miltenyi Biotec) for 15 min at 10°C, and then applied to an LS separation column in accordance with the manufacturer’s instructions. MHC-II⁺ and CD4⁺ cells (APC_KLH and KLH-primed CD4⁺ T cells [CD4⁺T_KLH]) were collected and cultured in L-15 medium (Life Technologies) containing 10% FBS (Life Technologies), 1% l-glutamine, 1.5% HEPES, 100 U/ml penicillin, and 100 μg/ml streptomycin at 28°C overnight to detach the magnetic beads. The purity of the sorted cells was detected through FCM as described above.

HEK293T or HeLa cells ectopically expressing DrTIM-1 and DrTIM-4 were used for subcellular localization analysis. Cells (3–6 × 10⁵ /ml) were seeded into six-well plates (Corning) and then cultured in DMEM (Biochrom, Berlin, Germany) supplemented with 10% FBS at 37°C in 5% CO₂ to allow growth until 70–80% confluence was reached. The cells were transiently transfected with pEGFP–DrTIM-1, pEGFP–DrTIM-4, and pDsRed2-Golgi plasmid DNAs (1–2 μg) combined with FuGENE HD transfection reagent (Roche, 3 μl/well) in accordance with the manufacturer’s instructions. At 48 h posttransfection, the cells were fixed in 2% paraformaldehyde for 10 min and then stained with 10 μM DiI (Beyotime) and 100 ng/ml DAPI (Sigma-Aldrich) at 37°C for 5 min. Fluorescence images of DrTIM-1, DrTIM-4, and Golgi apparatus proteins in cells were obtained using a laser scanning confocal microscope (LSM 710; Zeiss) (43, 47). For trafficking analysis, the magnetically sorted CD4⁺T_KLH cells were incubated with Ag-stimulated APC_KLH for 48 h before examination. Trafficking of DrTIM-1 in CD4⁺T_KLH cells was detected via immunofluorescence staining as described above.

Small interfering RNAs (siRNAs) against DrTIM-1 or DrTIM-4 as well as short hairpin RNAs (shRNAs) containing the selected siRNAs or the nonrelated scrambled siRNA were designed as previously reported (49). The synthesized DNA oligonucleotides that encoded for the shRNAs were constructed into the pSUPER vector downstream of the H1 promoter (47, 50). The generated constructs or control pSUPER, along with pCMV–DrTIM-1 or pcDNA6–DrTIM-4, was cotransfected into HEK293T cells. The U6 promoter cassette in lentiviral vector pLB was replaced by the H1-siRNA cassette excised from highly effective siRNA constructs to produce pLB–DrTIM-1 and pLB–DrTIM-4 lentiviral vectors. The constructed vectors were cotransfected with pCMV-dR8.2 dvpr and pCMV-VSVG packaging vectors into HEK293T cells using Lipofectamine 2000 (Invitrogen) (47, 50). The lentiviral supernatant was concentrated by ultracentrifugation, and the viral titers were determined through FCM analysis of GFP expression in HEK293T cells. The HEK293T cells were infected with lentiviruses (LVs) (10 μl) for 72 h and the zebrafish were injected with LVs (1 × 10⁶ transducing units [TU]/fish) once every 24 h for three times (26, 47). Afterward, the silencing efficacy of the constructed LVs was determined in HEK293T cells transfected with pCMV–DrTIM-1 or pcDNA6–DrTIM-4, as well as in peripheral blood, and in kidney leukocytes by real-time PCR.

Ag-induced expressions of DrTIM-4 and DrTIM-1 on APCs and CD4⁺ T cells were confirmed by an increase in DrMHC-II⁺DrTIM-4⁺ and DrCD4⁺DrTIM-1⁺ double-positive cells. In brief, fish were injected with 10 μg KLH plus 10 ng LPS for 24, 48, and 72 h. The double-positive cells in leukocytes from the spleen, kidney, and peripheral blood were labeled with Abs (rabbit anti-DrCD4 and mouse anti–DrTIM-1 or mouse anti–DrMHC-II and rabbit anti–DrTIM-4) and then analyzed via FCM as described above.

The effects of DrTIM-1 and DrTIM-4 on CD4⁺ T cell activation were examined using in vivo knockdown and blockade assays. For the knockdown assay, fish were injected thrice with siRNA-encoding LVs (1 × 10⁶ TU/fish) at a 24-h interval. At the last viral administration, the samples were coinjected with 10 μg KLH plus 10 ng LPS. Scrambled siRNA-encoding LV was administered as a negative control. For the blockade assay, fish were i.p. injected with 10 μg KLH plus 10 ng LPS and then administered thrice with anti–DrTIM-1 or anti–DrTIM-4 Ab (10 μg/fish). Nonrelated rabbit IgG was injected (10 μg/fish) as a negative control. Fish were challenged with 10 μg KLH plus 10 ng LPS and then administered thrice with sDrTIM-4 at different concentrations (1, 5, and 10 μg) to determine the effect of sDrTIM-4 on CD4⁺ T cells. GST-tag protein (10 μg) was injected as a negative control. The leukocytes were isolated from the spleen, kidney, and peripheral blood tissues 3 d after KLH stimulation. The proliferation and activation of CD4⁺ T cells were assessed via FCM, and the upregulation of CD154 and Lck was determined via real-time PCR (26). The CD4⁺ T cells were magnetically sorted from the stimulated leukocytes. Then, the expression of Th1-typic (IFN-γ, T-bet) and Th2-typic (IL-4, GATA3) cytokines and the transcription factors in the CD4⁺ T cells were examined by real-time PCR to evaluate the balance of Th1/Th2 differentiation (45).

Fish were i.p. injected with 10 μg KLH plus 10 ng LPS 5 d before sacrifice. CD4⁺ T cells were magnetically sorted from the spleen, kidney, and peripheral blood leukocytes, stained with 5 μM CFSE (Beyotime), and then terminated by supplementing with 10% FBS as previously described (26). MHC-II⁺ APCs were also magnetically sorted and further stimulated with KLH (100 μg/ml) for 8 h (51). The APCs were cocultured with CD4⁺ T cells for 72 h, during which anti–DrTIM-1 (5 μg/ml), anti–DrTIM-4 (5 μg/ml), and control rabbit IgG (5 μg/ml) were added to the cultures every 24 h. The proliferation and activation of Ag-specific CD4⁺ T cells (CD4⁺T_KLH) were examined via FCM, and the expression of CD154 and Lck was evaluated using real-time PCR.

For the B cell activation assay, fish were injected with KLH and then administered thrice with anti–DrTIM-1 or anti–DrTIM-4 as described above. Leukocytes from the spleen, kidney, and peripheral blood were collected 4 d after KLH stimulation, and the proliferation and activation of B cells were assessed via FCM with mouse anti-DrIgM and rabbit anti–DrCD40. For the IgM production assay, fish were i.p. immunized with 10 μg KLH plus 10 ng LPS, administered thrice with anti–DrTIM-1 or anti–DrTIM-4 at a 24-h interval, and then further immunized on days 5 and 28. Serum samples were collected at 35 d postimmunization, and the level of IgM against KLH was measured using ELISA as previously reported (29, 46). Briefly, 96-well plates were coated with KLH (5 μg/ml) overnight at 4°C. After blocking with 2% BSA for 1 h at 37°C and washing with PBS (containing 0.05% Tween 20), the wells were loaded with serially diluted serum samples at 37°C. After 2 h of incubation, the plates were washed and rabbit anti-DrIgM Ab was added and incubated for 1 h at 37°C. After washing, the HRP-conjugated goat anti-rabbit IgG Ab was added. Color was developed using tetramethyl benzidine and measured at 450 nm on a Synergy H1 hybrid reader (BioTek Instruments). Ab titer is defined as the highest serum dilution at which the A₄₅₀ ratio (A₄₅₀ of postimmunization sera/A₄₅₀ of preimmunization sera) is >2.1.

Fish were divided into two groups. One group was i.p. immunized with a bacterial vaccine (1 × 10⁶ CFU) derived from 0.5% formaldehyde-inactivated Aeromonas hydrophila, a pathogen of infectious sepsis in fish (52). The other group was also immunized with the same bacterial vaccine at the same dosage, with the exception of the administration of anti–DrTIM-1 and anti–DrTIM-4 (10 μg/fish) as described above. After 35 d, both immunized groups and unimmunized control groups were challenged with live A. hydrophila (2 × 10⁷ CFU/fish). The mortality of each group was recorded during a week, and the statistics of survival were analyzed.

Association between DrTIM-1 and DrTIM-4 was examined via quantitative ELISA and FCM (23, 29). For ELISA, various concentrations (0–780 nM) of recombinant DrTIM-4 protein were coated on a 96-well plate, followed by the administration of recombinant DrTIM-1 at 440 nM or PBS as negative control. The experiment was performed according to the protocol described above. For FCM, HEK293T cells were transfected with pcDNA6–DrTIM-4. At 48 h posttransfection, the cells were blocked with 5% normal goat serum and then incubated with FITC-conjugated DrTIM-1 (5 μg/ml) or FITC-conjugated DrTIM-1 (5 μg/ml) plus lipidosome (1 mM, prepared in the laboratory) (53). FCM analyses were performed as described above.

All data are presented as the mean ± SD of each group. Statistical evaluation of differences between means of experimental groups was performed using a Student t test. Statistical significance was considered at p < 0.05 or p < 0.01. The sample number for each group of fish was >10. All experiments were replicated at least three times.

With human (Homo sapiens) TIM-1 (HsTIM-1) and TIM-4 (HsTIM-4) sequences as queries, homologous genes DrTIM-1 and DrTIM-4 were predicted from the zebrafish genome database, respectively. Through a linkage search, a TIM-3 (DrTIM-3) homologous gene was also predicted downstream of the DrTIM-4 gene. Genes adjacent to the DrTIM cluster were retrieved from zebrafish chromosome 21 by using Genscan and BLAST. Similar to the genes around the TIM cluster on human chromosome 5 or mouse chromosome 11, the CXXC5, PURA, TTC1, FABP6, CCNJL, and GABRA1 genes were all found to be clustered on zebrafish chromosome 21, although the synteny of CXXC5 and PURA gene loci was in a reverse order (Fig. 1A). Similar results were observed in other teleost fish, such as Tetraodon nigroviridis (data not shown). Thus, an overall conservation of genome synteny exists between the TIMs of zebrafish and other species. The organization of the DrTIM-1 and DrTIM-4 genes was elucidated by comparing the DrTIM-1 and DrTIM-4 cDNAs with the corresponding genomic sequences. The DrTIM-1 and DrTIM-4 genes comprised seven and eight exons located within an 8.8-kb and a 7.5-kb genomic fragment, respectively. Exons 1 and 2 in the DrTIM-1 and DrTIM-4 genes were predicted to encode the signal peptides and the IgV domains. Meanwhile, exons 3 and 4 and exons 5–7 in the DrTIM-1 gene and exons 3–6 and exons 7 and 8 in the DrTIM-4 gene were predicted to encode the mucin domains and intracellular regions, respectively. The DrTIM-1 and DrTIM-4 genes slightly differed from their human counterparts. In particular, the DrTIM-1 and DrTIM-4 genes lack a mucin domain–encoding exon (exon 3 in HsTIM-1 and HsTIM-4 genes), making the mucin domains in the DrTIM-1 and DrTIM-4 proteins shorter than those in the HsTIM-1 and HsTIM-4 proteins (Fig. 1B). The cloned DrTIM-1 cDNA consists of 1479 bp comprising a 135-bp 5′-untranslated region (5′-UTR), a 564-bp 3′-UTR, and a 780-bp ORF that encodes 259 aa (Fig. 2A, GenBank accession no. KM373786). The DrTIM-4 cDNA consists of 1139 bp comprising a 66-bp 5′-UTR, a 185-bp 3′-UTR, and an 888-bp ORF that encodes 295 aa (Fig. 2B, GenBank accession no. KM373787).

Both DrTIM-1 and DrTIM-4 were predicted to be type I membrane proteins with molecular masses of 28.9 and 32.8 kDa, respectively. Both proteins possess an extracellular region (197 aa for DrTIM-1 and 216 aa for DrTIM-4), a transmembrane domain (23 aa for both DrTIM-1 and DrTIM-4), and an intracellular tail (39 aa for DrTIM-1 and 56 aa for DrTIM-4). The extracellular regions contain a putative signal peptide (27 aa for DrTIM-1 and 23 aa for DrTIM-4), an IgV domain (104 aa for DrTIM-1 and 108 aa for DrTIM-4), and a mucin domain (66 aa for DrTIM-1 and 85 aa for DrTIM-4) (Supplemental Fig. 1A). The IgV domains of DrTIM-1 and DrTIM-4 share high amino acid identities with those of their mammalian counterparts (49% to HsTIM-1 and 58% to HsTIM-4) and with those of other TIM family members (47–54%) (TIM-2 and TIM-3). However, the amino acid sequences of DrTIM-1 and DrTIM-4 share only 28–35% and 31–33% identities with their mammalian counterparts, respectively (data not shown). The above findings suggest that the IgV domain in TIM molecules is more conserved than others. An Arg-Gly-Asp (RGD) motif unique in the mammalian TIM-4 IgV domain was also found in DrTIM-4 but not in DrTIM-1. The mucin domains of DrTIM-1 and DrTIM-4 are rich in threonine, serine, and proline. Moreover, these domains predictably contain 18 (DrTIM-1) and 8 (DrTIM-4) O-linked glycosylation sites, as well as 1 (DrTIM-1) and 3 (DrTIM-4) N-linked glycosylation site/sites. The mucin domains of DrTIM-1, DrTIM-3, and DrTIM-4 vary considerably in size. The mucin domain of DrTIM-4 is longer than those of DrTIM-1 and DrTIM-3 by 19 and 16 aa, respectively. Similar observations were made in human and mouse TIM proteins, with the longest and shortest mucin in TIM-4 and TIM-3, respectively. At least one and four tyrosine phosphorylation sites were predicted in the intracellular tail of DrTIM-1 and DrTIM-4, respectively. The latter slightly differs from mammalian TIM-4, wherein no putative signaling motifs were detected (Fig. 3). These findings suggest the involvement of DrTIM-4 in intracellular signaling and imply a slight functional diversity between zebrafish and mammalian TIM-4 molecules.

Given its various unique structural features that distinguish TIM proteins from other Ig superfamily members, the IgV domain in TIM proteins was selected for further comparative analysis. The tertiary structures of the DrTIM-1 and DrTIM-4 IgV domains were modeled using the human and mouse TIM-1 and TIM-4 IgV domains as templates. As expected, DrTIM-1 and DrTIM-4 share typical hallmarks in IgV structures with their mammalian counterparts (Fig. 4A, 4B). For example, DrTIM-1 and DrTIM-4 are composed of two stacked pleated β-sheets (i.e., GFC β-sheet in one face and BED β-sheet in the opposite face), each formed by six (A, G, F, C, C′, and C″) and three (B, E, and D) anti-parallel β-strands, the latter of which (B, E, and D strands) are short. They also contain two α helices and several hydrophobic loop structures (BC, CC′, and FG). The CC′ and FG loops were predicted to form a deep pocket for ligand binding. DrTIM-1 IgV contains six Cys residues (Cys⁴⁰, Cys⁵³, Cys⁵⁸, Cys⁶³, Cys¹¹⁰, and Cys¹¹¹), whereas DrTIM-4 IgV contains four Cys residues (Cys³⁹, Cys⁵², Cys⁶³, and Cys¹¹¹). These residues were completely conserved in the TIM-1 and TIM-4 proteins in different species. The Cys residues are located on the BC loop (Cys⁴⁰), CC′ loop (Cys⁵⁸, Cys⁶³), C strand (Cys⁵³), and F strand (Cys¹¹⁰, Cys¹¹¹). The Cys residues form three disulfide bridges (Cys⁴⁰–Cys¹¹¹, Cys⁵³–Cys⁶³, and Cys⁵⁵–Cys¹¹⁰) between the BC loop and F strand, CC′ loop and C/F strands in DrTIM-1 IgV, and two disulfide bridges (Cys³⁹–Cys¹¹¹ and Cys⁵²–Cys⁶³) between the BC loop and F strand and CC′ loop and C strand in DrTIM-4 IgV. A slight difference between DrTIM-4 and DrTIM-1 (even other TIM proteins) is the replacement of Cys⁵⁸ and Cys¹¹⁰ by Ile⁵⁸ and Gly¹¹⁰, which leads to the absence of one disulfide bond between the CC′ loop and F strand (Fig. 4C, 4D). Root mean square deviation analysis showed that the deviation between DrTIM-1 and HsTIM-1 (0.285928 nm) or between DrTIM-4 and HsTIM-4 (0.382893 nm) was significantly lower than that between DrTIM-1 and HsTIM-4 (0.796561 nm) or between DrTIM-4 and HsTIM-1 (0.780437 nm). This result suggests that the DrTIM-1 and DrTIM-4 IgV domains are highly identical to their human counterparts. Two other functional residues (Arg¹¹² and Lys¹²³) in F and G strands, which are important for the binding of IgV to its ligands, are also completely conserved in the DrTIM-1 and DrTIM-4 IgV domains. A conserved metal ion–dependent ligand binding site inside the CC′-FG loop–formed pocket, which is crucial for the trafficking of TIM-1 to the cell surface, was also observed in both DrTIM-1 (FFND) and DrTIM-4 (WFND) (Fig. 2).

Phylogenetic analysis showed that DrTIM-4 was clustered with a cavefish (Astyanax mexicanus) TIM-4 homolog and then merged with mammalian counterparts to form a solitary branch. This branch was further clustered with a sister branch that comprised other teleost TIM-4 homologs to merge into a larger group. DrTIM-3 was grouped with its teleost counterparts and clustered with mammalian TIM-3 homologs. DrTIM-1 was classified in a solitary cluster with other teleost counterparts to form a supported clade adjacent to the mammalian TIM-1 branch. This branch separated from the mucosal vascular addressin cell adhesion molecule 1 molecules (MAdCAM-1), which had the closest genetic relationship to the TIM family (Supplemental Fig. 1B). The MAdCAM-1 molecule was included in the analysis because of its extracellular domain structure that was similar to TIM family members, including the Ig domain and the mucin domain.

A unique DrTIM-4 cDNA with a 498-bp ORF that encodes 165 aa was also cloned and was observed to be shorter by 390 bp in ORF compared with the complete DrTIM-4 cDNA. It was an alternatively spliced variant, in which exon 1–3 sequences in the DrTIM-4 gene were preserved whereas exon 4–8 sequences were spliced from the precursor mRNA during splicing (Fig. 5A). Interestingly, partial 5′ nucleotides (GTACTTAA) in intron 3 with a stop codon (TAA) was introduced into the exon 3 sequence in this variant. This phenomenon can be attributed to the existence of an extra exon/intron boundary (AG/GT) at the 3′ end of GTACTTAA. The observed splicing mechanism (partial intron retention) was distinct from the previously known intron retention mechanism, in which the intron sequence was completely introduced into the variant cDNA. Accordingly, the variant encodes a DrTIM-4 protein with a signal peptide, an IgV, and a partial mucin domain, but it lacks most of the mucin domain and the whole transmembrane and cytoplasmic regions (Fig. 5B). This protein was hypothesized to exist in soluble form and hence was designated as sDrTIM-4 (GenBank accession no. KM373789).

The results of real-time PCR showed that the transcripts of both DrTIM-1 and DrTIM-4 were expressed in most of the examined tissues, particularly in immune-relevant tissues such as the spleen, intestine, kidney, liver, and leukocytes (from the spleen, kidney, and peripheral blood) (data not shown). Double-immunofluorescence staining showed that a considerable number of leukocytes were DrMHC-II and DrTIM-4 double-positive (DrMHC-II⁺DrTIM-4⁺) or DrCD4 and DrTIM-1 double-positive (DrCD4⁺DrTIM-1⁺) (no positive cells in the negative control groups using nonrelated rabbit or mouse IgG as primary Abs) (Fig. 6A–D). In contrast, few cross-reactive cells (DrCD4⁺DrTIM-4⁺ or DrMHC-II⁺DrTIM-1⁺) were detected (Fig. 6E, 6F). Kinetic alterations of DrCD4⁺DrTIM-1⁺ and DrMHC-II⁺DrTIM-4⁺ cells in the leukocytes were examined via FCM every 24 h after Ag (KLH plus LPS) stimulation. The percentage of DrCD4⁺DrTIM-1⁺ cells was significantly upregulated (p < 0.05; from 2.68 ± 0.96% to 10.48 ± 2.46%) at 72 h. Meanwhile, the percentage of DrMHC-II⁺DrTIM-4⁺ cells was significantly upregulated (p < 0.05; from 6.2 ± 1.33% to 14.27 ± 2.15%) at 24 h and reached its peak (16.35 ± 2.07%; p < 0.01) at 48 h (Fig. 6G, 6H). These results indicate that DrTIM-1 and DrTIM-4 are predominantly expressed on CD4⁺ T cells and MHC-II⁺ APCs under the induction of Ag stimulation.

The generated pEGFP–DrTIM-1 and pEGFP–DrTIM-4 constructs were transfected into HEK293T or HeLa cells. The green fluorescence from the DrTIM-4 fusion protein was clearly colocalized with the red fluorescence from the membrane indicator DiI (Fig. 7A). This result suggests that DrTIM-4 is a typical membrane protein. In contrast, most of the DrTIM-1 signals were distributed in the cytoplasm (Fig. 7B), which could be well merged with the signal from the DsRed-tagged proteins located on the Golgi apparatus (Fig. 7C) but not with the signal from the lysosome probe (LysoTracker Red from Beyotime; Fig. 7D). These results imply that the DrTIM-1 protein is distributed on the Golgi apparatus (at least partially) but not on lysosomes. During cell culture, considerable DrTIM-1 proteins translocated onto the membrane surface at the intercellular junctions when the cells grew into contact confluence (Fig. 8A–C). Similar results were observed in contact CD4⁺ T cells (Fig. 8D). These observations suggest that DrTIM-1 is a trafficking protein that is originally located on the Golgi apparatus of resting cells but translocates onto the cell surface after different forms of stimulation, including cell–cell interactions. To further elucidate the trafficking nature of DrTIM-1 and determine its role in adaptive immunity, an induced trafficking experiment was performed in zebrafish CD4⁺ T cells in response to stimulatory signals provided by cell–cell interactions. The TIM-1⁺CD4⁺ T cells and the TIM-4⁺MHC-II⁺ APCs were sorted from Ag (KLH plus LPS)–treated fish to enhance the expression of DrTIM-1 and DrTIM-4 on these cells. As expected, DrTIM-1 was mainly distributed in the cytoplasm of CD4⁺ T cells without the stimulation of Ag-activated APCs (Fig. 9A). However, the addition of activated APCs to CD4⁺ T cells caused significant trafficking of DrTIM-1 onto the cell surface (Fig. 9B, 9C, 9F, 9G). In contrast, DrTIM-1 showed no significant trafficking in CD4⁺ T cells treated with resting APCs (low TIM-4 expression) from unstimulated fish (Fig. 9D, 9E). These observations suggest that the traffic-inducing signal between CD4⁺ T cells and APCs may be provided by DrTIM-1–DrTIM-4 interaction.

LV is a powerful siRNA delivery system for the knockdown of various immune-relevant molecules, such as CD80/86, CD83, and DIGIRR, in fish (26, 47). Recombinant LVs against DrTIM-1 and DrTIM-4 were constructed for functional investigation. After the in vitro assay in HEK293T cells, two siRNAs (DrTIM-1siRNA-1 and DrTIM-4siRNA-2) with the highest inhibitory efficiency (86–89%) against DrTIM-1 and DrTIM-4 mRNAs were selected from the candidates (Supplemental Fig. 2A, 2B). These siRNAs were used to generate two recombinant LVs (DrTIM-1siRNA-LV and DrTIM-4siRNA-LV), which showed high infectiousness (≥10⁵ TU/μl) and strong interference (56–68%), as determined by the fluorescence assay (with GFP as an indicator) (Supplemental Fig. 2C, 2D) and real-time PCR in vitro (Supplemental Fig. 2E, 2F). For the in vivo knockdown, the two LVs (1 × 10⁶ TU/fish) were i.p. administered thrice with a 24-h time interval. Real-time PCR showed that the expression levels of DrTIM-1 and DrTIM-4 in the DrTIM-1siRNA-LV– and DrTIM-4siRNA-LV–administered fish dramatically decreased (by 47–68%) in the peripheral blood and kidney leukocytes compared with those in the control groups. In contrast, few cross-interference reactions occurred between DrTIM-1siRNA-LV/DrTIM-4siRNA-LV and DrTIM-4/DrTIM-1 (Supplemental Fig. 2G–H). Abs against DrTIM-1 (anti–DrTIM-1) and DrTIM-4 (anti–DrTIM-4) were prepared for the Ab blockade assay. The extracellular regions of the DrTIM-1 and DrTIM-4 proteins were expressed in E. coli, purified using Ni-NTA agarose affinity chromatography, and then examined via SDS-PAGE (Supplemental Fig. 3A). Abs were affinity purified from the immunized rabbit sera into IgG isotypes by using protein A–agarose columns and nitrocellulose membrane immunosorbent protocols. ELISA and Western blot analyses showed that the purified Abs had an average titer >1:10,000 and showed specificity to DrTIM-1 and DrTIM-4 because few cross-reactions occurred between anti–DrTIM-1/anti–DrTIM-4 and other lymphocyte proteins (Supplemental Fig. 3B–G).

The expression levels of DrTIM-1 and DrTIM-4 on CD4⁺ T cells and MHC-II⁺ APCs suggest that both proteins serve as costimulatory molecules in the activation of Th2-typic adaptive immunity. To verify this hypothesis, the involvement of DrTIM-1 and DrTIM-4 in CD4⁺ T cell activation was examined in vivo through LV-based knockdown and Ab blockade assays. The administration of DrTIM-1siRNA-LV, DrTIM-4siRNA-LV, anti–DrTIM-1, and anti–DrTIM-4 dramatically inhibited (p < 0.05) CD4⁺ T cell activation. The percentage of CD4⁺CD154⁺ T cells in the Ag-stimulated peripheral blood, spleen, and kidney leukocytes decreased from 20.16 ± 2.11% (Ag plus scrambled siRNA-LV group) and 19.37 ± 2.91% (Ag plus IgG group) to 11.18 ± 2.04% (Ag plus DrTIM-1siRNA-LV group), 10.27 ± 2.57% (Ag plus DrTIM-4siRNA-LV group), 6.45 ± 2.85% (Ag plus anti–DrTIM-1 group), and 6.1 ± 1.15% (Ag plus anti–DrTIM-4 group), respectively (Fig. 10A). Similarly, the expression levels of Lck and CD154 in the peripheral blood, spleen, and kidney leukocytes dramatically decreased (p < 0.05) after the knockdown and blockade of DrTIM-1 and DrTIM-4 as determined via real-time PCR (Fig. 10B). Additionally, the blockade of DrTIM-1 and DrTIM-4 significantly inhibited (p < 0.05 or p < 0.01) the expression of the Th2-typic hallmarks (IL-4 and GATA3), but not the Th1-typic ones (IFN-γ and T-bet) in CD4⁺ cells (Fig. 11). These results suggest that DrTIM-1 and DrTIM-4 are essential for the activation of CD4⁺ T cells in Th2-typic adaptive immunity.

Blockade assays were performed in vitro to further understand the functions of DrTIM-1 and DrTIM-4 in APC-initiated, Ag-specific CD4⁺ T cell activation. The magnetically sorted MHC-II⁺ APCs were stimulated with KLH and cocultured with CD4⁺ T cells. During this period, anti–DrTIM-1 and anti–DrTIM-4 Abs were added to the cells. The CD4⁺T_KLH cells were labeled with CFSE, and the proliferation and activation of these cells were examined using FCM and real-time PCR. As shown in Fig. 12A, the proliferation of CD4⁺T_KLH cells primed by KLH-loaded APCs significantly decreased (p < 0.05) in the blockade groups compared with those in the nonrelated IgG-treated control groups. The proportion of activated CD4⁺T_KLH cells (CD4⁺CD154⁺T_KLH) in the DrTIM-1 and DrTIM-4 blockade groups declined from 37.76 ± 3.01% (control groups) to 25.53 ± 1.78% and 21.19 ± 2.34%, respectively (Fig. 12A). Similarly, the expression levels of Lck and CD154 significantly (p < 0.05) decreased in the DrTIM-1 and DrTIM-4 blockade cocultures (Fig. 12B). These results further support that DrTIM-1 and DrTIM-4 play essential roles in APC-initiated CD4⁺T_KLH cell activation.

The observed roles of DrTIM-1 and DrTIM-4 in APC-initiated, Ag-specific CD4⁺ T cell activation indicate that both proteins serve as costimulatory molecules in the initiation of Th2-typic immunity, which is critical for the subsequent B cell activation and Ab production. To address this notion, the inhibition of B cell activation and Ab (IgM) production was further assayed after the in vivo blockade of DrTIM-1 or DrTIM-4. As expected, the percentage of membrane (m)IgM⁺CD40⁺ B cells in the Ag-stimulated peripheral blood, spleen, and kidney leukocytes significantly decreased (p < 0.05) from 18.5 ± 0.82% (Ag plus IgG control group) to 12.32 ± 1.02% (Ag plus anti–DrTIM-1 group) or 13.58 ± 1.33% (Ag plus anti–DrTIM-4 group; Fig. 13A). Correspondingly, the levels of serum IgM against KLH in the DrTIM-1 and DrTIM-4 blockade groups significantly decreased (p < 0.05) compared with those in the nonrelated IgG-administered control group (Fig. 13C). These results support the costimulatory roles of DrTIM-1 and DrTIM-4 in CD4⁺ T-initiated B cell activation and Ab production in adaptive immunity.

To verify the involvement of DrTIM-1 and DrTIM-4 in adaptive immunity, an immunoprotection inhibitory assay was performed by blocking DrTIM-1 and DrTIM-4. An immunoprotective activity was induced after immunizing fish with a bacterial vaccine (inactivated A. hydrophila). As shown in Fig. 13D, 24% and 76.6%, respectively, of the fish in the unimmunized negative control group (without vaccination) and the immunized positive control group (received vaccination) survived after A. hydrophila challenge. This result indicates that adaptive immunity was well established after vaccination. However, the survival rates in the DrTIM-1 or DrTIM-4 blockade groups (with vaccination and anti–DrTIM-1 or anti–DrTIM-4) decreased from 76.6 to 53.3% or 46.7%, respectively. This trend supports the hypothesis that DrTIM-1 and DrTIM-4 play critical roles in adaptive immunity after vaccination.

To elucidate the effect of sDrTIM-4 on CD4⁺ T cell activation, fish were administered with the recombinant sDrTIM-4 protein accompanied by immunization with KLH. The proliferation of CD4⁺ T cells was significantly inhibited (p < 0.05) as the amount of the sDrTIM-4 protein for inoculation was increased from 1 to 10 μg (Fig. 13B). In contrast, the percentages of CD4⁺CD154⁺ T cells in the Ag-stimulated lymphocytes from the spleen, kidney, and peripheral blood decreased from 23.35 ± 1.17% (Ag plus GST control group) to 18.64 ± 1.02% (Ag plus 1 μg sDrTIM-4 group), 16.42 ± 0.91% (Ag plus 5 μg sDrTIM-4 group), and 6.84 ± 1.73% (Ag plus 10 μg sDrTIM-4 group), respectively.

Quantitative ELISA revealed a weak interaction between the DrTIM-1 and DrTIM-4 proteins (Supplemental Fig. 3H). To support this interaction, a functional binding assay was performed on DrTIM-4–expressing HEK293T cells by incubation with FITC–DrTIM-1. FCM analyses showed that the DrTIM-4–expressing HEK293T cells bound to FITC–DrTIM-1. Additionally, the administration of phosphatidylserine-derived liposomes (exosome mimics) can significantly stimulate the association of the two molecules. This result supports the previous observation that exosomes promote the association of mammalian TIM-1 and TIM-4 proteins via the phosphatidylserine molecule (Supplemental Fig. 3I). Therefore, further studies are needed to clarify whether the interaction between DrTIM-1 and DrTIM-4 is regulated by exosomes in vivo.

Since TIM-1 was first discovered as a hepatitis A virus receptor and a kidney injury molecule 1 in humans, TIM-1 and other TIM family members have also been characterized in mice (4, 5, 54). However, little is known about their occurrence in other organisms, including lower vertebrates. To uncover the biological functions of TIM proteins in other research models, such as teleosts, and to map the history of the TIM family throughout vertebrate evolution, we described the molecular and functional characteristics of TIM-1 and TIM-4 members in zebrafish and provided partial characterization of these molecules in other teleosts, such as Tetraodon. To the best of our knowledge, this work is the first to describe TIM molecules in a nonmammalian species. Several conserved structural lines among the TIM-1 and TIM-4 proteins of fish and other species support the conclusion that the cloned zebrafish TIM-1 and TIM-4 genes are homologs of the human and mouse TIM families. The evidence includes similar chromosomal synteny, location, and gene organization of the TIM-1 and TIM-4 genes. Additionally, similar protein domains were present, such as the extracellular IgV and mucin structures conserved in the TIM family; the unique stacked pleated β-sheets, α-helices, and hydrophobic loops; the ligand-binding essential amino acids (Arg and Lys) and motifs (metal ion–dependent ligand binding site) conserved in IgV; and the Cys residues critical for the structural integrity of the molecules. However, only 28–37% sequence identity was found among TIM family members in different species, with the most variability in the mucin domains. A few remarkable differences in features were distinguished between TIM-1 and TIM-4. For example, TIM-4 is 36 aa longer than TIM-1. TIM-4 contains an RGD motif in the IgV domain, which is a hallmark that differentiates the protein from other TIM members. TIM-4 is located on the cell surface, whereas TIM-1 is distributed in the cytoplasm before translocating onto the cell surface after stimulation (1, 55). These features are also present between DrTIM-4 and DrTIM-1, which clearly distinguish the proteins from each other. This observation was also supported by root mean square deviation analysis, wherein high identity independently existed between the IgV domains of DrTIM-1 and DrTIM-4 corresponding to their human counterparts.

TIM-1 and TIM-4 regulate immune responses in humans and mice in diverse ways, thereby performing different roles in various immunological activities (2, 55–57). The functions of TIM-1 and TIM-4 in adaptive immunity have received increasing attention, but updated results in mouse models remain elusive. Novel research systems greatly complementary to mammalian models are being developed. Emerging studies have shown that zebrafish is an attractive model organism in investigating immunology and various diseases (58). Therefore, a zebrafish model was used in the present study. Our results showed that TIM-4 and TIM-1 were closely associated with APCs and CD4⁺ T cells in zebrafish. Both proteins are essential for CD4⁺ T cell activation and subsequent B cell activation and Ab production. These observations strongly suggest that TIM-4 and TIM-1 are required for the interaction of APCs and CD4⁺ T cells to establish complete activation of adaptive humoral immunity against foreign Ags. Therefore, the results support the previous implication that TIM-1 and TIM-4 are critical costimulatory molecules in the initiation of adaptive immunity. Additionally, the underlying regulatory mechanisms of TIM-1 and TIM-4 in adaptive immunity have evolved during the early vertebrate evolution. Therefore, zebrafish may prove to be a powerful new model in studying the TIM family, with significant therapeutic implications. The identification of DrTIM-1 and DrTIM-4 in this study and the characterization of other costimulatory molecules, such as CD80/86, CD83, CD40, CD154, and CD209, from zebrafish in other recent studies may improve the current understanding of costimulatory molecules in the network and the evolutionary history of the costimulatory molecule families.

An sTIM-1 protein with an extracellular domain exists in humans. This sTIM-1 protein is released from the cell membrane–bound TIM-1 proteins by the digestion of metalloproteases at the cleavage site in the mucin domain near the transmembrane region (12, 59). However, the existence of other soluble forms of TIM proteins remains unknown, and the function of the sTIM protein has never been reported. In the present study, a novel soluble TIM-4 variant (sDrTIM-4) was identified in zebrafish. This variant is derived from a unique method of alternative splicing at the transcription level; its origin differs from that of the previously reported sTIM-1 at the protein level. The in vivo administration of this sDrTIM-4 protein significantly suppresses the activation of CD4⁺ T cells. This result suggests that this protein negatively regulates adaptive immunity. Such an effect may promote the inhibition of hyperimmune responses elicited by excessive interaction between TIM-1 and TIM-4, with potential applications in controlling TIM-1/TIM-4–based immune diseases. Several previous studies have shown that different anti–TIM-1 mAbs react against different epitopes in the extracellular region of TIM-1 and can differentially regulate T cell activity (2, 60). For example, mAbs against the mucin or stalk region of TIM-1 enhance CD4⁺ T cell activation, whereas the mAb against the IgV domain of TIM-1 decreases CD4⁺ T cell response. Therefore, sDrTIM-4 might function by interacting with the IgV domain via a mechanism similar to that of the mAb against the IgV domain. However, the precise regulatory mechanism of sDrTIM-4 warrants further investigation.

The TIM-1 protein in mouse lymphocytes can be transported from the cytoplasm to the cell surface under stimulation with PMA and ionomycin (61). The proposed underlying mechanism of this process is closely associated with the binding of the FG loop region in the IgV domain to phosphatidylserine in the cell membrane in a calcium-dependent manner (62). However, the biological significance of this trafficking is unclear. The demonstrated trafficking of the DrTIM-1 protein occurs in CD4⁺ T cells only when DrTIM-4–expressing APCs are added into the cells. This result suggests that the trafficking is induced by the interaction between APCs and CD4⁺ T cells, and that the allowing signal may be considerably provided by the DrTIM-4 molecules expressed on the APC surface. This event might provide a positive feedback regulatory mechanism that underlies the initiation of adaptive immunity between APCs and CD4⁺ T cells. This precise regulatory strategy allows CD4⁺ T cells to avoid unnecessary activation before they come into contact with fully activated APCs (with high expression of TIM-4 or other costimulatory molecules). The results also imply that TIM-1–TIM-4 interaction occurs only after other costimulatory molecules have interacted with each other.

The authors have no financial conflicts of interest.