BTLA−HVEM Checkpoint Axis Regulates Hepatic Homeostasis and Inflammation in a ConA-Induced Hepatitis Model in Zebrafish

This information is current as Wei Shi, Tong Shao, Jiang-yuan Li, Dong-dong Fan, Ai-fu of September 27, 2021. Lin, Li-xin Xiang and Jian-zhong Shao J Immunol published online 27 September 2019 http://www.jimmunol.org/content/early/2019/09/26/jimmun ol.1900458 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2019/09/26/jimmunol.190045 Material 8.DCSupplemental http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

by guest on September 27, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2019 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published September 27, 2019, doi:10.4049/jimmunol.1900458 The Journal of Immunology

BTLA–HVEM Checkpoint Axis Regulates Hepatic Homeostasis and Inflammation in a ConA-Induced Hepatitis Model in Zebrafish

Wei Shi,* Tong Shao,* Jiang-yuan Li,* Dong-dong Fan,* Ai-fu Lin,* Li-xin Xiang,* and Jian-zhong Shao*,†

The BTLA2HVEM checkpoint axis plays extensive roles in immunomodulation and diseases, including cancer and autoimmune disorders. However, the functions of this checkpoint axis in hepatitis remain limited. In this study, we explored the regulatory role of the Btla–Hvem axis in a ConA-induced hepatitis model in zebrafish. Results showed that Btla and Hvem were differentially expressed on intrahepatic Cd8+ T cells and hepatocytes. Knockdown of Btla or Hvem significantly promoted hepatic inflamma- + tion. Btla was highly expressed in Cd8 T cells in healthy but was downregulated in inflamed liver, as evidenced by a Downloaded from disparate proportion of Cd8+Btla+ and Cd8+Btla– T cells in individuals without or with ConA stimulation. Cd8+Btla+ T cells showed minimal cytotoxicity to hepatocytes, whereas Cd8+Btla– T cells were strongly reactive. The depletion of Cd8+Btla– T cells reduced hepatitis, whereas their transfer enhanced hepatic inflammation. These observations indicate that Btla endowed Cd8+Btla+ T cells with self-tolerance, thereby preventing them from attacking hepatocytes. Btla downregulation deprived this tolerization. Mechanistically, Btla–Hvem interaction contributed to Cd8+Btla+ tolerization, which was impaired by Hvem + – knockdown but rescued by soluble Hvem administration. Notably, Light was markedly upregulated on Cd8 Btla T cells, http://www.jimmunol.org/ accompanied by the transition of Cd8+Btla+Light– to Cd8+Btla–Light+ T cells during hepatitis, which could be modulated by Cd4+ T cells. Light blockade attenuated hepatitis, thereby suggesting the positive role of Light in hepatic inflammation. These findings provide insights into a previously unrecognized Btla–Hvem–Light regulatory network in hepatic homeostasis and inflammation, thus adding a new potential therapeutic intervention for hepatitis. The Journal of Immunology, 2019, 203: 000–000.

cell activation is strictly regulated by numerous costim- CTL-associated protein 4 (CTLA-4) in humans. This is ulatory and coinhibitory molecules that comprise diverse a type I membrane that contains a type V Ig domain T axes of adaptive immunity (1). These in its extracellular region and three functional residues by guest on September 27, 2021 immune checkpoint axes are crucial for self-tolerance, which (Tyrs) embedded in one growth factor –bound protein 2 prevents the immune system from indiscriminately attacking cells (Grb2) binding site and two immunoreceptor tyrosine-based mo- to maintain homeostasis (2, 3). The dysregulation of immune tifs in its cytoplasmic tail, which are essential for the recruitment checkpoint axes leads to various diseases, including cancer, in- of Grb2 and Src homology phosphatase-1/-2 (6). HVEM is a flammation, and autoimmune disorders (4, 5). Among the growing member of the TNFR superfamily and is a type I membrane family of checkpoint inhibitors, the B and T lymphocyte attenu- protein with an N terminus extracellular region and a cytoplas- ator (BTLA), which is associated with the herpesvirus entry me- mic segment closely associated with TNFR-associated factors diator (HVEM), is a promising target for immunotherapy because (TRAFs) and STAT3 signaling pathways (8–10). The ectodomain of its great potential use in multiple types of cancer and other of HVEM is composed of four cysteine-rich domains (CRDs). diseases (5–7). CRD1 is combined with BTLA and CD160, whereas CRD2 and BTLA is an Ig superfamily member that mainly serves as a CRD3 are mainly responsible for recognizing lymphotoxin-a and coinhibitor similar to programmed cell death 1 (PD-1) and LIGHT (lymphotoxin, exhibits inducible expression and competes

*Key Laboratory for Cell and Engineering of Zhejiang Province, College of Address correspondence and reprint requests to Prof. Jian-zhong Shao and Associate Prof. Life Sciences, Zhejiang University, Hangzhou 310058, People’s Republic of China; Li-xin Xiang, College of Life Sciences, Zhejiang University, 866 YuHangTang Road, and †Laboratory for Marine Biology and Biotechnology, Qingdao National Labora- Hangzhou 310058, People’s Republic of China. E-mail addresses: [email protected] tory for Marine Science and Technology, Qingdao 266071, People’s Republic of (J.-z.S.) and [email protected] (L.-x.X.) China The online version of this article contains supplemental material. ORCIDs: 0000-0002-2236-4708 (W.S.); 0000-0003-3020-6264 (T.S.). Abbreviations used in this article: ALT, alanine aminotransferase; AST, aspartate Received for publication April 24, 2019. Accepted for publication August 29, 2019. aminotransferase; Bcl-2, lymphoma 2; BHMT, Bhmt, betaine homocysteine S-methyltransferase; BTLA, Btla, B and T lymphocyte attenuator; Co-IP, coimmu- This work was supported by grants from the National Natural Science Foundation of noprecipitation; CRD, cysteine-rich domain; CsA, cyclosporin A; EGFP, enhanced China (31630083, 31572641), the National Key Research and Development Program GFP; Fasl, ; FCM, flow cytometry; Grb2, growth factor receptor–bound of China (2018YFD0900503, 2018YFD0900505, 2016YFA0101001), the Open Fund protein 2; HVEM, Hvem, herpesvirus entry mediator; Lamp-1, lysosomal-associated of the Laboratory for Marine Biology and Biotechnology, the Qingdao National 1; Lck, lymphocyte protein tyrosine kinase; LIGHT, Light, lym- Laboratory for Marine Science and Technology, Qingdao, China (OF2017NO02), photoxin, exhibits inducible expression and competes with HSV glycoprotein D for the Open Funding Project of the State Key Laboratory of Bioreactor Engineering, herpesvirus entry mediator; LV, lentivirus; mLIGHT, membrane-bound LIGHT; PFA, and the Zhejiang Major Special Program of Breeding (2016C02055-4). paraformaldehyde; qRT-PCR, quantitative real-time PCR; sBtla, soluble Btla; The sequences presented in this article have been submitted to GenBank (http://www. shRNA, short hairpin RNA; siRNA, small interfering RNA; sLIGHT, soluble LIGHT; ncbi.nlm.nih.gov/genbank/) under accession numbers MK112054, MK112055, and TRAF, TNFR-associated factor; TU, transducing unit. MK112056. Copyright Ó 2019 by The American Association of Immunologists, Inc. 0022-1767/19/$37.50

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1900458 2 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION with HSV glycoprotein D for herpesvirus entry mediator, a receptor interaction by upregulated Light on Cd8+ T cells activates the expressed on T cells) (11–13). BTLA is predominantly expressed cytotoxicity of Cd8+ T cells, thereby leading to the disruption of in B cells and various T cells, including Th1, Tfh, Th17, gd T, and liver homeostasis and occurrence of hepatitis. These findings NKT cells (14–16). HVEM is extensively expressed in dendritic suggest the importance of the Btla–Hvem axis and Btla–Hvem– cells, epithelial cells, hematopoietic cells, B lymphocytes, and Light network in the liver and reveal a previously unrecognized hepatoma cells (12, 17, 18). The engagement of BTLA by HVEM mechanism underlying hepatic homeostasis and inflammation. triggers the inhibitory activity of the former on T and B cell ac- tivation by decreasing the accumulation of phosphorylated TCR and BCR signals. This occurrence leads to the tolerance or anergy Materials and Methods of CD4+ T and CD8+ T cells and inhibition of cytokine release of Experimental fish NKT cells to avoid excessive inflammatory reactions and tissue One-year-old wild-type AB zebrafish (D. rerio) with body weight of injury (19–21). The BTLA–HVEM interaction can be regulated by 0.5–1.0 g were raised and maintained at 28˚C on a 12 h/12 h /dark LIGHT, the latter of which is also known as TNF superfamily cycle in a standard circulating system, as previously described (39). All fish used in experiments were siblings generated after at least two gen- member 14 (TNFSF14) (22). LIGHT, a type II membrane protein erations of inbreeding. Only healthy fish, as determined by general ap- with a C terminus extracellular TNF homology domain (THD), as- pearance and activity level, were used. All the animal experiments were sembles in homotrimers and modulates immune responses by conducted in accordance with the guiding principles for the care and use interacting with three TNFR superfamily members, namely, HVEM, of laboratory animals and were approved by a local ethics committee. lymphotoxin b receptor, and decoy receptor 3 (22–24). Two LIGHT Molecular cloning isoforms, namely, membrane-bound LIGHT (mLIGHT) and soluble Downloaded from The genome databases maintained by the National Center for Biotech- LIGHT (sLIGHT), have been identified in humans and mouse nology Information, the Genome Browser of the University of California models. mLIGHT can break the interaction between BTLA and Santa Cruz, and Ensembl were used to predict zebrafish , hvem, and HVEM because its affinity to HVEM is higher than that of BTLA light homologs. Total RNA was extracted from zebrafish by using to HVEM. By contrast, sLIGHT tends to combine with the an RNAiso Plus (Takara Bio). The btla, hvem,andlight cDNAs were BTLA–HVEM complex and firms the link between the two amplified by RT-PCR with primers shown in Supplemental Table I. The cDNA products were ligated into pGEM-T easy vectors (Promega) and components because of the opposite binding region of BTLA and sequenced on a 3730 XL sequencer (Applied Biosystems), as previously http://www.jimmunol.org/ LIGHT to HVEM (25–27). Thus, mLIGHT disrupts the inhibi- described (39). tory effect of BTLA on T cells by competitively binding to Bioinformatics analysis HVEM, whereas sLIGHT enhances the functional role of BTLA (25, 27). Genome locations of btla, hvem, and light were retrieved from the BTLA, HVEM, and LIGHT are closely associated with various Genome Data Viewer in the National Center for Biotechnology Informa- tion database. Primers for gene cloning were predicted by the Primer- cancers, such as melanoma; hepatocellular carcinoma; diffuse large BLAST program. Gene organization was elucidated by comparing B-cell lymphoma; invasive breast and colon cancers (5, 28–30); cDNAs of btla, hvem, and light with genome sequences by using BLAT autoimmune disorders, including systemic lupus erythematosus, and drawn by GeneMapper 2.5. Multiple alignments were analyzed using experimental autoimmune encephalomyelitis, type 1 diabetes, Clustal X (version 1.8) and GeneDoc. Putative extracellular, transmem- by guest on September 27, 2021 spontaneous urticaria, lung fibroblasts, and dermatitis (6, 31, 32); brane, and cytoplasmic regions were predicted by TMHMM Server (ver- sion 2.0). Phylogenetic trees were generated by MEGA 7.0 with the and inflammatory diseases, such as intestinal inflammation and neighbor-joining or maximum likelihood method. Potential tertiary struc- diet-induced obesity (4, 33). However, investigations on the tures and functional domains were predicted using PROSITE, SWISS- functional roles of BTLA, HVEM, and LIGHT in hepatitis re- MODEL, and PyMOL software (40). main limited. Although the involvement of BTLA and HVEM in Plasmid constructions hepatitis has been preliminarily explored, the results have been elusive because of some controversial conclusions from differ- The encoding sequences for Btla, Hvem, and Light and 2 2 their ectodomains were amplified by RT-PCR with primers shown in ent research groups. For example, BTLA-deficient (BTLA / ) + Supplemental Table I. After gel extraction and digestion, the sticky mice display autoimmune hepatitis-like features with CD4 T fragments were inserted into pET41a (Invitrogen), pcDNA6/myc-His and NKT cell infiltration, spotty necrosis in the liver, and ele- (Invitrogen), pEGFPN1 (BD Biosciences), pEGFPC1 (BD Biosci- vated transaminase levels in serum (34, 35). In accordance with ences), and pAcGHLTc (AB Vector) vectors to construct prokary- this observation, HVEM-deficient (HVEM2/2) mice exhibit otic or eukaryotic expression vectors with GST, Myc, enhanced GFP (EGFP), and His tags, respectively. The resulting constructs desig- increased morbidity and mortality, with high levels of multiple nated as pET41a-btla, pAcGHLTc-hvem, pEGFPN1-btla, pEGFPC1-btla, proinflammatory cytokines in a ConA-induced hepatitis model, pEGFPC1-hvem, pEGFPC1-light,pcDNA6/myc-His-hvem,andpcDNA6/ depending on the presence of CD4+ T cells (36). However, myc-His-light were used for recombinant protein expression and subcellular HVEM2/2 mice also show low serum transaminase and IFN-g localization examination. levels, high protective IL-22 serum levels, and attenuated liver Preparation of recombinant proteins histopathology in a liver invariant NKT cell–dependent manner For prokaryotic expression of soluble Btla (sBtla) protein with ectodomain, (37). These observations suggest that intensive studies are pET41a-btla was transformed into BL21 (DE3; TransGen Biotech) com- warranted to further explore the complex regulatory mecha- petent cells, cultured in Luria–Bertani medium containing kanamycin nisms underlying BTLA, HVEM, and LIGHT interactions dur- (50 mg/l; Sangon Biotech) at 37˚C with 200 rpm shaking, and induced by ing hepatic inflammation. isopropyl-b-D-thiogalactoside (IPTG, 0.5 mM; Sangon Biotech) at 20˚C In this study, the functional roles of Btla, Hvem, and Light in for 12 h. After ultrasonication, the supernatants were collected for pu- rification. For eukaryotic expression, pAcGHLTc-hvem and baculovirus liver homeostasis and inflammation were explored in a ConA- vector DNA (AB Vector) were cotransfected into Sf9 (Spodoptera inducedhepatitismodelinzebrafish(Danio rerio), which is at- frugiperda) cells under the assistance of polyethylenimine (branched tractive for the study of comparative immunology and diseases PEI; Sigma-Aldrich) in a T25 flask with SIM HF medium (Sino Biolog- (38). The Btla–Hvem interaction between intrahepatic Cd8+ ical) containing penicillin–streptomycin (Thermo Fisher Scientific). The cells were cultured at 28˚C for 5 d, harvested by centrifugation (2000 rpm), T cells and hepatocytes contributes to liver homeostasis. The and dissolved in lysing buffer (200 mM Tris-HCl, pH 8, 150 mM NaCl, engagement of Btla by Hvem triggers inhibitory signals for the 1% Nonidet P-40, 1 mM PMSF). The recombinant Btla–GST and tolerization of Cd8+ T cells. The breakdown of the Btla–Hvem Hvem–His proteins were purified by nickel–nitrilotriacetic acid agarose The Journal of Immunology 3 affinity chromatography (Qiagen), following the manufacturer’s manual, were separated from the liver digested with type IV collagenase (1.0 U/ml, and then detected by SDS-PAGE (41). at room temperature for 30 min; Sigma-Aldrich) by filtration through a 40-mm strainer (Falcon; BD Biosciences) and centrifugation at 100 3 g at Preparation of polyclonal Abs 4˚C for 5 min. The cells were fixed with 4% PFA at room temperature for The recombinant sBtla and sHvem proteins and an epitope peptide with 10 min, blocked with 2% BSA (BSA; Sigma-Aldrich), and incubated 16 aa (SLHPKIPRPSIENSFL) predicted from the ectodomain of the Light with primary Abs in combinations of rabbit anti-Cd8a and mouse anti- protein were used to prepare Abs against Btla, Hvem, and Light proteins. Btla, rabbit anti-Cd8a and mouse anti-Light, and rabbit anti-BHMT and The epitope peptide was chemically synthesized and conjugated with OVA mouse anti-Hvem Abs at 4˚C for 2 h. After washing with PBS, the cells (BankPeptide). Four-week-old male Institute of Cancer Research mice were combined with FITC-conjugated goat anti-rabbit IgG and PE- (∼15 g) or 6-wk-old male New Zealand rabbits (∼1.5 kg) were immunized conjugated goat anti-mouse IgG secondary Abs (Thermo Fisher Scientific), with the proteins or peptide (20 mg or 0.5 mg) each time in CFA (Sigma- following the manufacturer’s instructions. The cells were washed with PBS Aldrich) initially and then in IFA (Sigma-Aldrich) for four times thereafter and stained with DAPI (100 ng/ml) at room temperature for 5 min. Fluo- rescence images were captured using a two-photon laser confocal scanning at biweekly intervals, as previously described (39). Seven days after the 3 final immunization, serum samples were collected. Abs were affinity pu- microscope (LSM-710; Zeiss) with 630 magnification. rified by using Protein A Agarose Columns (Thermo Fisher Scientific), and Coimmunoprecipitation and Western blot analysis their titers were examined by ELISA. The validity and specificity of the Abs were determined by Western blot analysis. Rabbit anti-Cd8a Ab and Coimmunoprecipitation (Co-IP) was performed to detect the interaction mouse anti-Cd4 Ab were produced in our previous studies (39, 42), and between Btla and Hvem. HEK293T cells were cotransfected with pcDNA6/ rabbit anti–betaine homocysteine S-methyltransferase (anti-BHMT) Ab myc-His-hvem (3 mg) and pEGFPN1-btla (3 mg) in a 10-cm dish under the was purchased from Thermo Fisher Scientific. assistance of PEI. At 48 h posttransfection, the cells were lysed with precooling cell lysis buffer (Beyotime). The lysates were centrifuged Quantitative real-time PCR (10,000 rpm at 4˚C for 10 min), and the supernatants were incubated with Downloaded from The transcript abundance of the target genes was analyzed via quantitative mouse anti-EGFP mAb (Abmart) at 4˚C overnight. The mixture was in- real-time PCR (qRT-PCR) on a CFX Connect Real-Time PCR Detection cubated with 50 ml of protein A agarose beads (Thermo Fisher Scientific) System (Bio-Rad). Total RNA was extracted from the samples by using an for 4 h. The beads were washed three times with lysis buffer, mixed with RNAiso Plus kit (Takara Bio) and reverse transcribed into cDNAs. The PCR loading buffer, and heated to 100˚C for 10 min to denature the proteins. experiments were performed in a total volume of 10 ml by using an iTaq After centrifugation, the proteins were separated by SDS-PAGE and Universal SYBR Green Supermix (Bio-Rad). The reaction mixtures were transferred onto a 0.22-mm polyvinylidene difluoride membrane (PVDF; incubated for 2 min at 95˚C, followed by 40 cycles of 15 s at 95˚C, 15 s at EMD Millipore) for Western blot analysis. After blocking with 2% BSA at 60˚C, and 20 s at 72˚C. Relative expression levels were calculated using 4˚C for 1 h, the membrane was incubated with primary Abs at 4˚C for 2 h, http://www.jimmunol.org/ 22Δ cycle threshold and 22ΔΔ cycle threshold methods with rps18 (40S ribosomal washed with TBST, and incubated with HRP-conjugated goat anti-mouse/ protein S18) for normalization (43). Each PCR trial was run in triplicate rabbit IgG mAb (Abmart) at 4˚C for 1 h. Detection was performed on a gel parallel reactions and repeated three times. The primers used were listed in imaging system (Tanon 4500). Supplemental Table I and were checked to have well efficiency. ConA-induced hepatitis model in zebrafish Subcellular localization ConA-induced hepatitis model was established by i.p. injection of zebrafish with various doses of ConA (50 mg to 300 mg/g body weight; Sigma- HEK293T cells were seeded into 12-well plates (Corning) with cover glass Aldrich) for different time periods (12–72 h) for optimization. Hepatitis and cultured in high-glucose DMEM (Life Technologies) in which 10% (v/v) was evaluated by the release levels of alanine aminotransferase (ALT) and FBS (Life Technologies) was added at 37˚C in 5% CO to allow growth until 2 aspartate aminotransferase (AST) in sera; histopathologic symptoms; ex- 50–60% confluence. The cells in each well were transfected with 0.8 mgof by guest on September 27, 2021 pression of proinflammatory cytokines (Il-1b, Il-6, Tnf-a, and Ifn-g); pEGFPC1-btla, pEGFPC1-hvem, or pEGFPC1-light plasmid DNA com- perforin; lysosomal-associated membrane protein 1 (Lamp-1); apoptotic bined with PEI reagent (3.2 mg per well) in accordance with the manu- regulators, including Fas ligand (Fasl) and B cell lymphoma 2 (Bcl-2) in facturer’s protocol. At 48 h after the transfection, the cells were fixed by livers; and of hepatocytes through in situ detection and FCM 4% (mass/volume) paraformaldehyde (PFA; Sigma-Aldrich) and analysis. Serum ALT and AST levels were detected using ALT and AST stainedwithCM-DiI(1mM; Thermo Fisher Scientific) and DAPI (100 Assay Kits (Jiancheng Bioengineering Institute, Nanjing). The liver ng/ml; Sigma-Aldrich). Fluorescence images were captured using a two- tissues were fixed in 4% PFA overnight. The paraffin sections were photon laser confocal scanning microscope (LSM-710; Zeiss) with 3630 collected for H&E, immunohistochemical staining, and TUNEL assay magnification. for histopathology and in situ hepatocyte apoptosis analyses using a Generation of short hairpin RNAs encoding lentivirus TUNEL Detection Kit (Beyotime). For immunohistochemical staining, the sections were blocked in 2% BSA at 25˚C for 1 h and incubated with Short hairpin RNAs (shRNAs) carrying the small interfering RNAs primary Abs (mouse anti-Btla/Hvem/Light and rabbit anti-Cd8a)or (siRNAs) targeting btla and hvem mRNAs and shRNAs encoding len- isotype control mouse/rabbit IgG (Sangon Biotech) at 4˚C overnight tiviruses (LVs) were designed and produced, as previously described after retrieving Ags and blocking endogenous peroxidase. Secondary (39). shRNAs were constructed into a pSUPER plasmid (pSUPER.retro.puro; Abs (HRP-conjugated goat anti-mouse/rabbit IgG Abs, Abmart) were OligoEngine) downstream of the H1 promoter. The constructs were transfected incubated at 4˚C for 1 h, and color was developed using a DAB mixture into HEK293T cells with pEGFPC1-btla or pEGFPC1-hvem for effi- (Beyotime). Hematoxylin was stained to show the nuclei. Pictures were ciency evaluation. The U6 promoter cassette in a lentiviral plasmid caught under a Zeiss microscope (Zeiss Axiostar Plus). (pLB) was replaced by an H1–shRNA cassette from the constructs har- boring effective shRNAs to produce pLB-sibtla and pLB-sihvem FCM analysis and sorting vectors. The shRNAs encoding LVs were generated by cotransfecting Cells under detection or sorting were blocked with 1% goat serum for 1 h at HEK293T cells with pLB-sibtla or pLB-sihvem and packaging vectors 4˚C and incubated with the defined primary Abs for 1 h at 4˚C. Nonspecific (pCMV-VSVG and pCMV-dR8.2). The viral supernatant was concen- rabbit or mouse IgG was served as the isotype control. After washing three trated by ultracentrifugation (25,000 rpm, 90 min, 4˚C). Viral titers were times with D-Hank’s buffer, the cells were incubated with secondary Abs detected through EGFP signature in HEK293T cells under the fluores- (PE-conjugated goat anti-mouse IgG mAb and FITC-conjugated goat anti- cent microscope or by flow cytometry (FCM) analysis. The silencing rabbit IgG mAb) for 1 h at 4˚C. The cells were detected or sorted by the activity of the resulting LVs was determined in zebrafish intrahepatic flow cytometer (FACSCalibur or FACSJazz; BD Biosciences). FCM leukocytes and Cd8+ T cells and hepatocytes by qRT-PCR or Western analysis for lymphocytes was performed following previously described blot analysis after fish were i.p. injected with the LVs (2 3 105 trans- protocols (39, 42). At least 10,000 cells were acquired from the gate for ducing units [TU] per fish) once every 24 h for three to five times with or analysis. FlowJo 7.6 software (BD Biosciences) was used for data pro- without ConA stimulation. cessing. For hepatic apoptosis analysis, the hepatocytes were separated Immunofluorescence staining from the livers digested with type IV collagenase as described above, washed three times with PBS by centrifugation, resuspended in 1 3 Colocalizations of Cd8a and Btla, Cd8a and Light, and Bhmt and Hvem annexin-binding buffer, and labeled by Annexin V and propidium iodide were determined by immunofluorescence staining. Leukocytes were iso- following the protocol recommended by the manufacturer (Thermo Fisher lated from the liver of zebrafish by Ficoll–Hypaque (1.080 g/ml; Sangon Scientific). At least 10,000 cells were acquired from the gate for FCM Biotech) centrifugation at 2500 rpm at 25˚C for 25 min. The hepatocytes analysis. 4 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION

Deletion and adoptive transfer assays Results Deletion and adoptive transfer assays were performed for functional Identification of zebrafish btla, hvem, and light genes + evaluation of the cytotoxic activity of Cd8 T cells and the regulatory role With human (Homo sapiens) BTLA (HsBTLA), HVEM (HsHVEM), of the Btla–Hvem axis to the tolerization of Cd8+ T cells. For deletion assay, zebrafish was i.p. injected with anti-Cd8a Ab (20 mg/g body and LIGHT (HsLIGHT) gene sequences as queries, the corre- weight) three times in a 24-h interval, as described in our previous sponding zebrafish btla, hvem, and light homologous genes were studies (39, 42). At the third administration, the fish were coinjected with predicted from the zebrafish genome database. The btla/hvem/ Ab and ConA. One day after the last injection, the deletion efficiency of light gene was located within a 2.98/4.20/12.68 kb genomic Cd8+ T cells was examined by FCM, and the level of hepatitis was assessed, as described above. For adoptive transfer assay, intrahepatic fragment on 1/8/3 and contained 7/8/4 exons and Cd8+,Cd8+Btla+,Cd8+Btla–,Cd8+Light+,andCd8+Light– T cells were 6/7/3 introns, respectively. The genes adjacent to btla/hvem/light sorted from the donor zebrafish livers with or without ConA stimulation loci shared an overall conserved chromosome synteny to human and i.p. transferred into the ConA stimulation–recipient fish at different 4 6 genes, but some of the genes were in reverse order at the cell dosages (10 –10 cells per fish). In a specified blockade assay, upstream or downstream of the btla/hvem/light gene relative to Cd8+Btla+ and Cd8+Btla– or Cd8+Light+ and Cd8+Light– Tcellswere incubated with anti-Btla Ab or anti-Light Ab at 4˚C for 1 h before that of humans (Fig. 1A left; Fig. 1A right; Fig. 1D). Exon or- transfer. In a knockdown assay, Cd8+Btla+ T cells were transferred into ganization and encoding sequences for functional domains in the recipient fish receiving ConA (200 mg ConA/g body weight), sihvem-LV btla/hvem/light gene were also similar to those in the HsBTLA/ 3 5 and ConA (2 10 TU per fish three times at a 24-h interval in com- HsHVEM/HsLIGHT gene or mouse (Mus musculus) BTLA/HVEM/ bination with ConA at the last time of viral injection), and ConA, sih- vem-LV, and sHvem (ConA and sihvem-LV in combination with sHvem LIGHT (MmBTLA/MmHVEM/MmLIGHT) gene (Fig. 1B, 1C, 1E). + atadoseof2mg/g body weight at the last time of viral injection). Cd8 , The cloned btla/hvem/light cDNA consisted of 1408/1604/3536 bp Downloaded from Cd8+Btla+, and Cd8+Btla– T cells were sorted from the liver by flow with an 80/433/327 bp 59UTR, 927/813/708 bp open reading cytometer (FACSJazz; BD Biosciences), as mentioned above. Mock PBS frame encoding 308/270/235 aa, and 401/358/2501 bp 39UTR was administered in the control groups. The effect of adoptive transfer was determined by the changes in serum ALT and AST levels and ap- (Supplemental Fig. 1A, 1B), respectively. One alternatively optosis of hepatocytes. spliced variant of Btla, which lacked 5 aa in the extracellular re- gion, was cloned from zebrafish instead of two other variants Cytotoxicity assay lacking the extracellular Ig domain or transmembrane region in http://www.jimmunol.org/ The cytotoxicity of Cd8+Btla+ or Cd8+Btla– T cells to hepatocytes was mammalian BTLAs (Fig. 1F). examined by the release of lactate dehydrogenase by using a CytoTox 96 Non-Radioactive Cytotoxicity Assay Kit (Promega). The sorted intra- Structural characterization of btla, hvem, and light proteins hepatic Cd8+Btla+ or Cd8+Btla– T effector cells were cocultured with the target hepatocytes in 96-well plate in different proportions (1:40, 1:20, Btla was predicted as a type I transmembrane protein with the 1:10, or 1:5) at 28˚C for 4 h. In a specified blockade assay, Cd8+Btla+ typical structural features of the Ig superfamily. The protein T cells were pretreated with anti-Btla Ab (5 mg/ml) or nonspecific mouse contained three major functional domains, including a single ex- IgG (isotype control, 5 mg/ml) at 4˚C for 1 h. After incubation, the plate tracellular V-type Ig-like domain (40–136 aa), a transmembrane was centrifuged at 250 3 g for 4 min, and the supernatant (50 ml) was collected from each well and mixed with the CytoTox 96 reagent (50 ml) region (167–189 aa), and a cytoplasmic tail (190–308 aa). One for enzymatic assay. After the 30-min reaction in an opaque box, the stop ITIM and one immunoreceptor tyrosine-based switch motif were by guest on September 27, 2021 solution (50 ml) was added to each sample. Absorbance (OD490) was de- present in the distal cytoplasmic tail region, which also included tected by a Synergy H1 Hybrid Multi-Mode Microplate Reader (BioTek). two conserved Tyrs (Tyr277 and Tyr302) (Supplemental Fig. 2A). The cytotoxicity (percentage) of the effector cells was calculated following a previously described protocol (44). The functional domains of Btla exhibited overall conserved ter- tiary structures compared with those of HsBTLA and MmBTLA, + Functional evaluation of Cd4 T cells but they shared moderate identities (30–32%) and minor diver- Coculture assay with Cd4+,Cd8+,andCd8+Btla+ T cells was performed gence with their mammalian counterparts. For example, the in a Transwell device (Corning) to evaluate the functional role of Cd4+ N-terminal Ig-like domain of Btla was folded with two a helices T cells in the transition of Cd8+Btla+ to Cd8+Btla– T cells and + + – + – + + + + + and two b sheets with eight strands. One sheet was composed of Cd8 Btla Light to Cd8 Btla Light T cells. Cd4 , Cd8 , or Cd8 Btla B, E, and D strands, whereas the other consisted of A9,G,F,C, T cells were isolated from the liver with or without ConA stimulation through FACS (FACSJazz; BD Biosciences). Cd4+ T cells (1 3 104) were and C9 strands. The comparison indicated that the analogy model added into each Transwell chamber with a 5.0-mm pore polycarbonate of HsBTLA displayed one a helix and two b sheets with a total of + + + 4 membrane. Cd8 or Cd8 Btla T cells (2 3 10 ) were loaded per well in a nine strands, including B, E, and D in one sheet and A9,C,C9,F, 12-well plate. The cells were cultured in L-15 medium (Thermo Fisher G0, and G in another sheet. Six cysteine residues were conserved Scientific) containing 10% FBS and penicillin–streptomycin. After 1 or 3 d of culture, the expression of Ifn-g, Fasl, perforin, and Lamp-1 in Cd8+Btla+ between Btla and HsBTLA. These cysteine residues formed three T cells was examined by qRT-PCR. T-bet, Il-12 (p35), Il-12 (p40), Ifn-g, disulfides between C33 and C64, C57 and C120, and C73 and Il-4, Cd154, and lymphocyte protein tyrosine kinase (Lck) in Cd4+ T cells C80. The C73–C80 disulfide connected strands C and C9, whereas with or without ConA stimulation were also examined by qRT-PCR. the C57–C120 disulfide connected strands B and F, similar to the The expression level of Btla on Cd8+Btla+ T cells and the proportion changes in Cd8+Btla+Light– and Cd8+Btla–Light+ T cells were deter- case in HsBTLA. Residues 34–42 (DVKLKVPRQ) in the Ig-like mined by FCM with primary (rabbit anti-Btla and mouse anti-Light) and domain of Btla were similar to residues 35–43 (DVQLYIKRQ) secondary (FITC-conjugated goat anti-rabbit IgG and PE-conjugated in HsBTLA, the binding site to HVEM (Supplemental Fig. + + goat anti-mouse IgG) Abs. Cd8 T cells cultured alone without Cd4 1C). Some imperceptible differences still existed between T cells were examined to show the initial expression levels of Btla and Light in Cd8+ T cells. Unlabeled Cd8+ T cells were adopted as negative Btla and HsBTLA/MmBTLA, despite their numerous similar- control in the FCM analysis. ities. For example, the Grb2 binding site (YDND-containing motif), which was present in the cytoplasmic tail of HsBTLA Statistical analysis and MmBTLA, was absent in Btla and some other fish Btlas, Statistical differences among means of experimental groups were evaluated such as carp Btla. This observation suggests a slight functional by ANOVA and multiple Student tests. All data were presented as the difference in intracellular signaling between Btla and mam- mean 6 SD of each group. Statistical significance was considered when p , 0.05 or p , 0.01. The sample number for each group was at least malian BTLAs. 20 fish of equal mean body weight. All data represented the means of at Hvem was also predicted as a type I transmembrane protein least three independent experiments. containing an extracellular region (1–195 aa) rich in cysteines The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 1. Molecular characterization of zebrafish (D. rerio) btla, hvem, and light genes. (A) Chromosomal localization analysis between btla and human (H. sapiens) BTLA genes and between hvem and human HVEM genes. The arrows indicate the transcriptional direction of the genes. (B and C) Gene organization of zebrafish btla, human, and mouse (M. musculus) BTLA genes (B) and zebrafish hvem, human, (Figure legend continues) 6 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION

(22 residues), a single transmembrane region (196–218 aa), and Preparation of recombinant proteins, Abs, and siRNA- a cytoplasmic region (219–270 aa) (Supplemental Fig. 2B). The encoding LVs extracellular region of Hvem possessed four CRDs in different Recombinant Btla, Hvem, and Light proteins with either the lengths, including CRD1 (24–59 aa), CRD2 (61–103 aa), CRD3 ectodomain or intact molecules were prepared from Escherichia (111–146 aa), and CRD4 (149–187 aa). These structural features coli, Sf9, and HEK293T cell lines with a GST or Myc tag are typically observed in mammalian TNF superfamily members (Supplemental Fig. 3A–C). Abs against Btla (anti-Btla), Hvem occupying one to six CRDs of ∼40 aa each. Hvem shared 41% (anti-Hvem), and Light (anti-Light) were prepared from immu- sequence identity with that of HsHVEM. The domain nized mouse or rabbit sera into IgG isotypes by employing protein architecture of CRDs was highly conserved between HsHVEM A affinity purification. The results showed that Abs exhibited high and Hvem. In HsHVEM, CRD1 was critical for the interaction specificities to Btla, Hvem, and Light proteins from liver leuko- of HsHVEM with HsBTLA through residues 73–77 (TVCEP), cytes or hepatocytes, with average titers above 1:10,000 based on which was mainly located on CRD1. A loop structure formed by ELISA and Western blot analyses (Supplemental Fig. 3A–C). the disulfide bonds was also present within C57 and C75 as well Rabbit anti-BHMT Ab was also confirmed to be effective and as C54 and C67, a position named DARC (glycoprotein D and specific to zebrafish Bhmt molecules, as determined by Western BTLA binding site on the TNFR HVEM in CRD1). A similar blot and FCM analyses (Supplemental Fig. 3D, 3E). To prepare structure of DARC was also predicted in Hvem, in which a siRNA-encoding LVs against btla and hvem, we predicted a total potential Btla binding site (residues 57–61, TTCVP) and a loop of six candidate siRNAs (sibtla-1 to sibtla-3 and sihvem-1 to sihvem-3) targeting different regions of btla and hvem mRNAs. composed by the disulfide bonds within C41 and C59 as well as Downloaded from sibtla-3 and sihvem-2 were siRNAs with the highest effectiveness C38 and C51 were present (Supplemental Fig. 1D). The CRD1 in inducing the degradation of btla and hvem mRNAs. These and CRD2 of Hvem individually contained one sheet with two siRNAs were selected for the development of siRNA-encoding strands, resembling those of HsHVEM. A threonine phos- LVs (sibtla-LV and sihvem-LV, Supplemental Fig. 3F–H). The ti- phorylation site (Thr251) was predicted in the cytoplasmic re- ters of the two generated LVs were .1 3 106 TU/ml when gion of Hvem, whose counterpart (Thr277) in HsHVEM is assessed in HEK293T cells by EGFP-based FCM analysis crucial for the recruitment of TRAFs (TRAFs 1, 2, 3, and 5) and (Supplemental Fig. 3I, 3J). The interference activities of the two http://www.jimmunol.org/ activation of PKCu downstream HsHVEM signaling pathway. LVs were determined by in vivo knockdown evaluation. Zebrafish Hence, Hvem may activate a signaling pathway (such as the under ConA stimulation were i.p. administered with sibtla-LV and TRAFs–PKC axis) similar to that of HsHVEM, thereby leading sihvem-LV (2 3 105 TU per fish) or scrambled control siRNA-LV to NF-kB activation. for three times within a 24-h interval. One day after the last in- Light was predicted as a type II transmembrane protein con- jection, the expression levels of btla and hvem mRNAs and pro- sisting of a cytoplasmic region with N terminus (1–38 aa), a single teins in the liver were determined by qRT-PCR and Western blot transmembrane region (39–61 aa), and an extracellular region with analyses. The results showed that the btla and hvem mRNAs were C terminus (63–235 aa) (Supplemental Fig. 2C). Light shared 43% significantly (p , 0.01) downregulated by 80–90% in the intra- by guest on September 27, 2021 amino acid sequence identity to HsLIGHT. Light and HsLIGHT hepatic leukocytes and Cd8+ T cells as well as hepatocytes that were composed of three protomers, with each protomer containing received sibtla-LV and sihvem-LV compared with those adminis- one a helix and two b sheets. The two b sheets contained a total tered with the scrambled control siRNA-LV (Supplemental Fig. of 10 b strands. Light showed the typical “jelly-roll” fold model 3K–M). The Btla and Hvem proteins were also dramatically similar to HsLIGHT, in which two b sheets were formed by downregulated in leukocytes and hepatocytes of the livers that strands A9,A,H,C9, and F (inner b sheet) and strands B9,B,G,D, received the sibtla-LV and sihvem-LV (Supplemental Fig. 3N). and E (outer b sheet). A GH loop structure that linked the inner Subcellular localization and hepatic distribution of btla, hvem, and outer b sheets and contributed to the interaction between and light proteins HsLIGHT and HsHVEM in humans also existed in Light, Subcellular localization analysis showed that the green fluores- thereby suggesting the potential interaction of Light with Hvem cence from the Btla and Light fusion proteins in HEK293T cells in zebrafish (Supplemental Fig. 1E). In the distal cytoplasmic was predominantly colocalized with the red fluorescence from region of Light, a potential serine phosphorylation site (Ser10) the membrane indicator CM-DiI, thereby suggesting that Btla and was predicted, whose counterpart in HsLIGHT (Ser10) is respon- Light are typical membrane proteins. By contrast, the green signals sible for the activation of PKC-mediated ERK, PI3K, and NF-kB from Hvem fusion protein was distributed on the cell membrane signaling pathways. This finding implies that Light plays similar and in the cytoplasm (Fig. 2A–C). Interestingly, when the cells functional roles by conserved signaling pathways. The phylogenetic were stimulated with sBtla, the majority of the Hvem proteins analysis showed that Btla, Hvem, and Light were clustered with displayed a distinct punctate cytoplasmic distribution on the lat- other BTLAs, HVEMs, and LIGHTs in different species with high eral area of the nuclear membrane and a loss of their original bootstrap probability (Supplemental Fig. 1F, 1G). membrane localization and random cytoplasmic distribution

and mouse HVEM genes (C). The black squares, fold lines, and numbers represent exons, introns, and size of corresponding exons or introns, respectively. (D) Chromosomal localization analysis between zebrafish light and human LIGHT genes. The arrows indicate the transcriptional direction of the genes. (E) Gene organization of zebrafish light, human, and mouse LIGHT genes. The black squares, fold lines, and numbers represent exons, introns, and size of corresponding exons or introns, respectively. (F) Structural characterization of alternative spliced variants of Btla. The signs above the schematic show the amino acid residues or amino acid sequence encoded by the variants. The GenBank accession numbers of the sequences are as follows: H. sapiens BTLA, NM_181780.3; M. musculus BTLA, NM_001037719.2; D. rerio btla variant 1, MK112054; D. rerio btla variant 2, MK112055; H. sapiens HVEM, NM_001297605.1; M. musculus HVEM, NM_178931.2; D. rerio hvem, MK112056; H. sapiens LIGHT, NM_003807.4; M. musculus LIGHT, NM_019418.3; and D. rerio light, NM_001281995.1. CP, cytoplasmic region; SP, signal peptide; THD, TNF homology domain; TM, transmembrane region; YCR, tyrosine conserved region. The Journal of Immunology 7 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 2. Subcellular localization analysis of Btla, Light, and Hvem proteins in HEK293T cells. (A–C) Detection of subcellular localization of Btla, Light, and Hvem proteins in HEK293T cells transfected with pEGFPC1-btla/-light/-hvem for 48 h. The green, blue, and red fluorescence shows Btla–/ Light–/Hvem–EGFP fusion proteins, DAPI-labeled nuclei, and CM-DiI–labeled cell membranes. (D and E) Perinuclear distribution of Hvem proteins in HEK293T cells enhanced by sBtla protein stimulation (E) compared with that in the control group without sBtla stimulation (D). Scale bar(s), 10 mm.

(Fig. 2D, 2E). Hence, Hvem acts as a trafficking protein under cells colocalized with Bhmt, which is a surface hallmark of hepa- Btla stimulation. Immunohistochemical staining assay showed tocytes in zebrafish. This finding suggests the distribution of Hvem that considerable nonparenchymal cells in the liver sections in zebrafish hepatocytes (Fig. 3F). were marked with anti-Btla and anti-Light (Fig. 3A, 3B). Most of the hepatocytes in the sample were positive for anti-Hvem ConA-induced hepatic inflammation in zebrafish (Fig. 3C). For further clarification, the hepatocytes and other ConA-induced hepatic inflammation, one of the most widely used nonparenchymal cells were separated from the liver. The dis- hepatitis models, was developed in zebrafish to investigate the tribution of Btla, Light, and Hvem were examined by an indirect functional roles of Btla, Hvem, and Light in liver inflammatory immunofluorescence assay. Btla and Light clearly displayed dot- reactions. ConA induced hepatic inflammation in a dose-dependent like signatures on the surface of cells coexpressed with Cd8a, manner, with an optimal concentration of ∼200 mg/g of body suggesting the membrane localization of Btla and Light on Cd8+ weight. The induction of hepatic inflammation was determined by T cells (Fig. 3D, 3E). Hvem was mainly detected on the surface of the increased levels of ALT and AST in sera; upregulation of 8 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 3. Cellular distribution of Btla, Light, and Hvem proteins in liver. (A–C) Immunohistochemical staining with anti-Btla (A), anti-Light (B), and anti-Hvem (C) Abs in zebrafish liver paraffin sections. Tan spots marked by red arrowheads show the positive cells. Scale bar(s), 50 mm. (D–F) Cellular distribution of Btla, Light, and Hvem proteins in cells isolated from liver. Double immunofluorescence staining shows the colocalization of Cd8a and Btla, Cd8a and Light, and Bhmt and Hvem on the surfaces of intrahepatic lymphocytes and hepatocytes. The nuclei of cells were labeled by DAPI. Scale bars are present in the first diagram of each line.

Il-1b, Il-6, Tnf-a, and Ifn-g in the liver; and enhanced apoptosis ConA-induced livers, were significantly diminished by CsA of hepatocytes accompanied by the upregulation of Fasl and treatment in a dose-dependent manner. In accordance with the downregulation of Bcl-2. Hepatic inflammation peaked at 24 h decreased hepatocyte apoptosis, Bcl-2 mRNA was upregulated under ConA stimulation at a dose of 200 mg/g of body weight. The (Fig. 4H–K). most strikingly upregulated proinflammatory cytokines, such as Il-1b and Tnf-a, were increased by more than 50-fold (Fig. 4A–F, Btla and hvem contribute to liver homeostasis and 4J). Histopathological observation through H&E staining showed hepatitis suppression evident hepatic plate structure disorder, punctate/lytic necrosis, Given the immune tolerance nature of the liver, this organ satisfies and edema of hepatocytes in the liver after ConA treatment. the requirements for studying the negative regulatory role of Lymphocyte infiltration markedly occurred in the liver (Fig. 4G). coinhibitors (such as BTLA) in the maintenance of immune ho- Given that ConA-induced hepatitis in mouse models is a T cell– meostasis, the disruption of which boosts inflammatory reactions dependent disorder, this characteristic was also evaluated in a and leads to various diseases, including hepatitis. Thus, the po- zebrafish model by using the T cell inhibitor cyclosporin A (CsA). tential roles of Btla in the maintenance of liver homeostasis were Expectedly, hepatic inflammatory indicators, including serum examined by siRNA-based knockdown assays. Transaminases ALT and AST levels; Il-1b,Il-6,Tnf-a, Ifn-g, perforin, Fasl, (ALT and AST), inflammatory cytokines (Il-1b,Il-6,Tnf-a,and and mRNA levels; and percentage of apoptotic hepatocytes in Ifn-g), perforin, Lamp-1, and Fasl were significantly induced The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 4. ConA-induced hepatic inflammation in zebrafish. (A–F) Changes in ALT and AST release levels in sera and apoptosis of hepatocytes after ConA treatment (200 mg/g body weight) at different doses and time periods. Mock PBS–injected groups served as controls. (G) H&E staining of zebrafish liver paraffin sections of PBS-treated control group (left) and ConA- (200 mg/g body weight) stimulated group (middle). The area in the red, dashed rectangle frame is enlarged for a detailed view (right). Red arrow indicates the area of punctate necrosis of hepatocytes accompanying lymphocyte infiltration. Scale bar(s), 50 mm. (H–K) Changes in ALT and AST release levels in sera; transcription levels of proinflammatory factors, perforin, Lamp-1, and apoptosis-related factors in the livers; and apoptosis of hepatocytes induced by ConA (200 mg ConA/g body weight) or ConA plus CsA (2 or 10 mg CsA/g body weight) administration. Mock PBS treatment served as the control. Broken y-axis with disparate scales was adopted to display the disproportionate data. Error bars indicate SD. *p , 0.05, **p , 0.01. 10 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION

FIGURE 5. Evaluation of the effect of Btla or Hvem on liver homeostasis and ConA-induced hepatitis. (A–D) Changes in

ALT and AST release levels in sera, tran- Downloaded from scription levels of proinflammatory factors, perforin, Lamp-1, and apoptosis-related factors in the livers and apoptosis of he- patocytes after knockdown of Btla by ad- ministration of sibtla-LV for 3 and 5 d. (E) H&E staining shows the histopathological changes in zebrafish livers that received http://www.jimmunol.org/ sibtla-LV injection for 3 d. Red arrow in- dicates the area of punctate necrosis of hepatocytes accompanying lymphocyte in- filtration. Scale bar(s), 50 mm. (F–H) Changes in ALT and AST release levels in sera and transcription levels of proin- flammatory factors, perforin, Lamp-1, and apoptosis-related factors in the livers of

ConA- (200 mg/g body weight) stimulated by guest on September 27, 2021 zebrafish after the knockdown of Btla and Hvem by sibtla-LV and sihvem-LV admin- istration, respectively, for 3 d. (I and J) TUNEL assay and FCM analysis show the increased apoptotic hepatocytes in sibtla-LV– and sihvem-LV–treated groups. The tan spots indicate the apoptotic hepatocytes. Mock PBS-injected group served as the control. Broken y-axis with disparate scales was adopted to display the disproportionate data. Error bars represent SD. Scale bar(s), 50 mm. *p , 0.05, **p , 0.01.

(p , 0.05 or p , 0.01) in the sera and livers of zebrafish in Hepatocyte apoptosis was significantly increased, accompanied which sibtla-LV had been administered compared with those by Lamp-1 and Fasl upregulation (Fig. 5D). The livers also injected with the scrambled control siRNA-LV (Fig. 5A–C). showed evident hepatic plate structure disorder, punctate/lytic necrosis, The Journal of Immunology 11 lymphocyte infiltration, and edema of hepatocytes after Btla knock- optimal effector/target ratio of 1:20. After treatment with anti-Btla down for 3 d (Fig. 5E). These observations indicate that Btla greatly Ab, Cd8+Btla+ T cells exhibited considerable cytotoxic activity to contributes to the immune homeostasis of the liver. Furthermore, fish target cells (Fig. 7B). To indicate that the Cd8+Btla+ T cells are with Btla or Hvem knockdown markedly exacerbated ConA-induced anergic rather than just naive, we examined the expression levels hepatic inflammation, as shown by the significantly higher levels of of Il-2, Lck, Cxcr3, and Ccr7, a set of markers for Cd8+ T cell ALT and AST in sera and Il-1b,Il-6,Tnf-a,andIfn-g in the livers of activation in Cd8+Btla+ T cells. Intrahepatic Cd8+Btla+ T cells fishinwhichsibtla-LV and/or sihvem-LV had been administered than exhibited high expression levels of Il-2, Lck, and Cxcr3 but low those in scrambled siRNA-LV and ConA-treated fish (Fig. 5F–H). The Ccr7 expression. These values were comparable with those of the TUNEL assay showed a remarkably increased number of apoptotic activated Cd8+Btla2 T cells. This expression pattern was opposite hepatocytes in the livers of fish in which sibtla-LV or sihvem-LV had of that of the inactivated naive Cd8+ T cells from the head been administered under ConA stimulation (Fig. 5I). The FCM anal- (a hematopoietic tissue) of fish without any stimulation (Fig. 7C). ysis results indicated that the apoptotic cell rate was 30.16 6 0.15% or In vivo adoptive transfer assay was performed for further evalu- 31.13 6 0.49% in ConA-induced livers or ConA- and scramble ation. Cd8+Btla+ and Cd8+Btla– T donor cells sorted from the siRNA–treated livers, which significantly increased to 43.63 6 0.67% inflamed livers were transferred (1 3 105 cells per fish) into the and 44.60 6 0.50% with the administration of sibtla-LVand sihvem- recipient fish that received the optimized ConA treatment (200 mg/ LV, respectively (Fig. 5J). These results suggest the inhibitory role of g body weight for 1 d). Hepatitis was significantly enhanced by Btla and Hvem in hepatic inflammation, whose disruption would the transfer of Cd8+Btla– T cells, as determined by the increased augment hepatic inflammatory responses. serum levels of ALT and AST and the occurrence of hepatic ap-

optosis. By contrast, hepatic inflammation stopped developing in Downloaded from Involvement of Cd8+ T cells in ConA-induced the ongoing ConA-induced hepatitis in the Cd8+Btla+ T cell– hepatic inflammation transferred liver, as evident by the maintained basal levels of se- ConA-induced hepatitis in zebrafish is T cell dependent, and im- rum levels of ALT and AST and the apoptosis of hepatocytes in munohistochemical results showed that a considerable percentage comparison with those in the control groups. However, after the of Cd8+ T cells infiltrated into the ConA-induced liver (Fig. 6A). transfer of Cd8+Btla+ T cells that were pretreated with anti-Btla + This finding suggests that Cd8 T cells are largely involved in the Ab for the blockade of Btla, hepatic inflammation resumed http://www.jimmunol.org/ ConA-induced hepatitis model. To provide support for this hy- (Fig. 7D–F). The treatment of Cd8+Btla– T cells with anti-Btla Ab pothesis, we performed a Cd8+ T cell deletion assay by using anti- did not influence the cytotoxic activity of Cd8+Btla– T cells, Cd8a Ab. The FCM analysis showed that the number of Cd8+ suggesting that Btla is the direct target of anti-Btla Ab. In this T cells was significantly decreased by ∼91% in the ConA-induced case, the ratio of hepatic apoptosis in fish that received a transfer liver after anti-Cd8a Ab was administered three times within a of Cd8+Btla– T cells that were pretreated with anti-Btla Ab was 24-h interval (Supplemental Fig. 3O). The Cd8+ T cell–deleted 39.03 6 0.46%, similar to the 39.23 6 0.31% ratio in the control fish displayed lower ALT and AST release, downregulation of group without anti-Btla Ab treatment. These results indicate inflammatory cytokines (Il-1b,Tnf-a, and Ifn-g) and perforin, and that Cd8+Btla– T cells contributed to cytotoxicity during he- decreased apoptotic hepatocytes from 30.17 6 0.25% to 16.17 6 0. patic inflammation and Cd8+Btla+ T cells were anergic because by guest on September 27, 2021 50% compared with the fish that received ConA treatment alone of Btla expression. Moreover, the cytotoxic effector molecules (Fig. 6B–E). The TUNEL assay showed a remarkably decreased perforin, Lamp-1, and Fasl were significantly upregulated in number of apoptotic hepatocytes in the livers of fish that received Cd8+Btla– T cells in the liver that suffered from inflammation anti-Cd8a Ab under ConA stimulation (Fig. 6F). To support this (Fig. 7G). finding, we performed an adoptive transfer assay by sorting Cd8+ Btla–hvem association contributes to the tolerization of T cells from the ConA-induced livers and transferring them into Cd8+Btla+ T cells Cd8+ T cell–depleted fish at different doses (104,105, and 106 cells per fish) under ConA stimulation. As expected, the adopted The association of Btla with Hvem was initially examined by Co-IP fish showed significant restoration of inflammation, as determined assay to evaluate whether they act as reciprocal molecules in a by the upregulated serum ALT and AST levels, along with the manner similar to that in mammals. The result showed that Btla– increased dosage of Cd8+ T cells (Fig. 6G). EGFP interacted with Hvem, whereas EGFP–tag protein did not

+ combine with Hvem (Fig. 7H). The contribution of Btla–Hvem Downregulation of btla on Cd8 cells enhances association to the tolerization of Cd8+Btla+ T cells to hepatocytes hepatic inflammation was functionally evaluated by adoptive transfer assays. The Immunohistochemical staining and FCM analyses showed two Cd8+Btla+ T cells from ConA-induced donor fish were transferred subsets of Cd8+ T cells (i.e., Cd8+Btla+ and Cd8+Btla–)inthelivers into Hvem-knockdown recipient fish that received sihvem-LV and with or without ConA stimulation. The proportion of Cd8+Btla+ ConA stimulation in accordance with the above-mentioned T cells to the total Cd8+ T cells decreased from 15.13 6 0.96% to method. The tolerization of Cd8+Btla+ T cells in hepatitis was 2.46 6 0.26% in the ConA-induced liver compared with that of the significantly impaired, along with the downregulation of Hvem untreated liver. The ratio of Cd8+Btla– T cells increased from in the liver. This finding suggests that the cytotoxic anergy of 83.13 6 0.75% to 97.03 6 0.15%, accompanied with the decline Cd8+Btla+ T cells on hepatocytes was abolished when the degree of Cd8+Btla+ T cells (Fig. 7A). These observations suggest that of association between Btla and Hvem decreased. The attenuation Cd8+Btla– T cells are cytotoxic in hepatic inflammation, whereas of Cd8+Btla+ T cell tolerization was determined by the increased Cd8+Btla+ T cells are anergic because of the expression of the ALT and AST release and the occurrence of hepatic apoptosis inhibitory Btla protein. To provide evidence for this hypothesis, from 31.90 6 1.08% in the untreated liver to 50.80 6 0.40% in we isolated Cd8+Btla+ and Cd8+Btla– T cells and hepatocytes Hvem-knockdown liver. When the Hvem-knockdown recipi- from the ConA-induced livers for in vitro cytotoxicity assay. As ent fish were coadministered Cd8+Btla+ T cells and a soluble predicted, Cd8+Btla+ T cells showed no significant cytotoxicity to recombinant Hvem protein (sHvem, 2 mg/g body weight), the hepatocytes at a wide range of effector/target ratios (1:5 to 1:40), cytotoxic anergy of Cd8+Btla+ T cells was significantly restored, whereas Cd8+Btla– T cells displayed high cytotoxicity with an as shown by the decline in ALT and AST release and apoptotic 12 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 6. Determination of the involvement of Cd8+ T cells in ConA-induced hepatic inflammation. (A) Immunohistochemical staining with anti-Cd8a Ab shows the increased infiltration of Cd8+ T cells (red arrowheads) in ConA- (200 mg/g body weight) treated group (right) or mock PBS–injected control group (left). Scale bars, 50 mm. (B–E) Changes in ALT and AST release levels in sera; transcription levels of proinflammatory factors, perforin, Lamp-1, apoptosis-related factors in the livers; and apoptotic hepatocytes after deletion of Cd8+ T cells by anti-Cd8a Ab administration. (F) TUNEL analysis was performed for apoptosis determination. The tan spots indicate the apoptotic hepatocytes. Scale bar(s), 50 mm. (G) Adoptive transfer of Cd8+ T cells from the inflamed livers shows restoration of hepatic inflammation, as determined by the upregulation of serum ALT and AST levels. Broken y-axis with disparate scales was adopted to display the disproportionate data. Error bars represent SD. **p , 0.01. hepatocytes from 50.80 6 0.40% to 34.20 6 2.26% (Fig. 7I, 7J). did not receive sHvem administration. This finding suggests that By contrast, when the Hvem-knockdown recipient fish were the suppressive axis of Btla2Hvem in the Hvem-knockdown coadministered Cd8+Btla2 T cells and sHvem (2 mg/g body recipient fish with Cd8+Btla2 T cell transfer could not be re- weight), the cytotoxic activity of Cd8+Btla2 T cells remained stored, even when sHvem was replenished. In addition, the unchanged. In this case, the ratio of hepatic apoptosis in the re- maintenance of the cytotoxic anergy of Cd8+Btla+ T cells de- cipients was 49.00 6 0.66%, similar to the 50.70 6 2.01% ratio in pends on the association of Btla with Hvem largely provided by the control group in which Cd8+Btla– T cells were transferred but hepatocytes. The soluble Hvem protein plays a regulatory role The Journal of Immunology 13 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 7. Evaluation on the contribution of Btla and Btla–Hvem association to the tolerization of Cd8+ T cells in the liver. (A) The ratio of Cd8+Btla+ and Cd8+Btla2 T cells in the control liver without stimulation and inflamed liver under ConA (200 mg/g body weight) stimulation suggests that down- regulation of Btla on Cd8+ T cells enhanced hepatic inflammation. (B) In vitro cytotoxicity assay showing that Cd8+Btla+ T cells are anergic, whereas Cd8+Btla2 T cells are cytotoxic. After treatment with anti-Btla Ab (5 mg/ml), the Cd8+Btla+ T cells displayed cytotoxicity to the target cells. (C) Different expression levels of Il-2, Lck, Cxcr3, and Ccr7 in Cd8+Btla+ and Cd8+Btla2 T cells sorted from ConA-stimulated liver. The inactivated Cd8+ T cells sorted from the head kidney of fish without stimulation were selected as a control. (D–F) In vivo adoptive transfer assay showing Cd8+Btla– T cells, but not Cd8+Btla+ T cells, promoted hepatic inflammation, as determined by the release levels of ALT and AST and changes (Figure legend continues) 14 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION in hepatic inflammation, which is of potential importance for support for this hypothesis, we abolished the functional role of clinical therapeutic purposes. Light on Cd8+Btla–Light+ T cells by a blockade assay via the administration of anti-Light Ab (20 mg/g body weight) into fish. Functional role of Cd4+ T cells for the downregulation of btla As predicted, the blockade of Light markedly impaired ConA- on Cd8+ T cells induced hepatic inflammation, as shown by the significantly + CD4 T cells play an important role in hepatitis in humans and decreased levels of ALT and AST in the sera and Il-1b,Il-6, ConA-induced mouse models. Thus, we explored whether the Tnf-a, Ifn-g, perforin, Lamp-1, and Fasl in the livers, accom- + downregulation of Btla on Cd8 T cells during hepatic inflam- panied by the upregulation of Bcl-2 (Fig. 8H–J). The adoptive + + + + mation was modulated by Cd4 T cells. Cd4 ,Cd8Btla ,and transfer of Cd8+Light+ T cells (1 3 105 cellsperfish)fromthe + – Cd8 Btla T cells were sorted from ConA-induced livers by inflamed livers into the ConA-stimulated recipient fish signifi- + + + – FACS. Cd8 Btla and Cd8 Btla T cells were cocultured with cantly aggravated hepatitis and increased the levels of ALT and + Cd4 T cells through a Transwell chamber. The transcription AST released. However, this aggravation was impaired when the + + level of btla mRNA in Cd8 Btla T cells significantly declined. transferred Cd8+Light+ T cells were pretreated with anti-Light Ab In contrast, Ifn-g, Fasl, and Lamp-1 were upregulated after 24 or (Fig. 8K). No significant enhancement of hepatitis was observed 72 h of coculture (Fig. 8A, 8B). The transcripts of T-bet, Il-12, after transferring Cd8+Light– T cells with or without the treatment and Ifn-g, which are three representative Th1-typic factors, were of anti-Light Ab. These observations support the notion that the + remarkably increased in Cd4 T cells from the ConA-induced upregulation of Light on Cd8+ T cells exacerbates hepatitis. When + livers compared with those of Cd4 T cells from the unstimu- Cd8+ T cells were cocultured with Cd4+ T cells through a Transwell + lated livers (Fig. 8C). These observations suggest that Cd4 T cells chamber for 24 h, the percentage of CD8+Btla+Light– Tsubsets Downloaded from underwent Th1 cell differentiation under hepatic inflammation and decreased from 41.33 6 0.95% to 0.51 6 0.05%, which was ac- + + secreted Th1-typic cytokines to suppress Btla expression in Cd8 Btla companied with an increase in the number of Cd8+Btla–Light+ + – T cells, which then differentiated into cytotoxic Cd8 Btla T cells. T cells from 1.41 6 0.41% to 67.27 6 0.42% (Fig. 8L). This Supporting this hypothesis, the mean fluorescence intensity of result suggests that Cd4+ T cells also play a modulatory role in the + + Btla in Cd8 Btla T cells was significantly (p , 0.01) declined upregulation of Light on Cd8+ T cells, in addition to their function + + after 24 or 72 h of coculture with Cd4 T cells, as determined by in the downregulation of Btla in Cd8 T cells. http://www.jimmunol.org/ the FCM analysis (Fig. 8D). Discussion Involvement of light in btla–hvem–mediated The BTLA, HVEM, and LIGHT immune checkpoint regulators hepatic inflammation have become increasingly explored because of their important LIGHT, another ligand of HVEM, serves as a costimulatory modulatory roles in immunity and extensive involvement in various molecule during T cell activation in humans and other mammalian diseases of humans and other mammals (28, 32). However, the models. Finally, we explored whether Light plays a regulatory role occurrence and functions of these checkpoint molecules in non- in Btla–Hvem–mediated hepatic inflammation in the zebrafish mammalian species remain poorly understood (45). In the current

model. The qRT-PCR results showed that light mRNA was study, we identified Btla, Hvem, and Light homologs from by guest on September 27, 2021 expressed in low quantities in Cd8+Btla+ T cells but was signifi- zebrafish. These homologs were characterized to share similar cantly (p , 0.01) upregulated in Cd8+Btla– T cells (Fig. 8E). The structural features with their mammalian counterparts, thereby FCM analysis showed that the ratio of Cd8+Light+ to total Cd8+ suggesting their functional conservation in vertebrate evolutionary T cells was significantly (p , 0.01) enhanced by ∼2.4-fold in the history. For functional study, a ConA-induced hepatitis model was ConA-stimulated livers compared with that in the control group established in zebrafish. The regulatory functions of Btla, Hvem, treated with mock PBS (20.03 6 1.17% versus 8.23 6 1.00%, and Light homologs were then extensively investigated in this Fig. 8F). Accordingly, a Cd8+Btla–Light+ and a Cd8+Btla+Light– inflammatory disease model. T subset population were detected in the sorted intrahepatic Cd8+ A number of experimental lines substantially supported the T cells. The number of Cd8+Btla–Light+ T cells significantly in- conclusion that the Btla–Hvem–Light checkpoint network reg- creased (p , 0.01) from 2.15 6 0.45% in Cd8+ T cells sorted ulates liver homeostasis and hepatic inflammation in the ConA- + from healthy livers without ConA stimulation (Cd8 TCtrl cells) to induced hepatitis model. For example, the knockdown of Btla and 38.35 6 2.90% in those from the inflamed livers with ConA Hvem in healthy or ConA-stimulated livers significantly induced + + + – stimulation (Cd8 TConA cells). The number of Cd8 Btla Light hepatic inflammatory reactions. These results provide initial in- T cells significantly decreased (p , 0.01), accompanied by the sights into the negative regulation of the Btla–Hvem axis in liver increase in the percentage of Cd8+Btla–Light+ T cells, from homeostasis and inflammation, given that the disruption of this 12.37 6 2.82% in healthy livers to 1.22 6 0.67% in the ConA- axis augmented the occurrence of hepatitis. Immunohistochemical stimulated livers (Fig. 8G). The disparate ratio of Cd8+Btla–Light+ staining showed the presence of a considerable number of Cd8+ and Cd8+Btla+Light– T cells in the healthy and hepatitis livers T cells in ConA-induced liver. The depletion of these Cd8+ T cells suggests the different roles between Light and Btla. The upregu- decreased hepatic inflammation and tissue injury, which could be lation of Light on Cd8+Btla–Light+ T cells may contribute to the restored by the adoptive transfer of Cd8+ T cells from the livers hepatic inflammation in response to ConA stimulation. To provide that suffered from hepatitis. This result suggests the involvement

in hepatocyte apoptosis. (G) Different transcription levels of perforin, Ifn-g, Lamp-1, and Fasl in Cd8+Btla+ and Cd8+Btla2 T cells. (H) Co-IP assay showing the association of Btla with Hvem. Btla–EGFP and Hvem–Myc fusion proteins were coexpressed on HEK293T cells. Mouse anti-EGFP and anti- Myc Abs and nonrelevant control IgG isotype were used for the examination. (I and J) Functional evaluation on the negative regulatory role of Btla–Hvem association in ConA-induced (200 mg/g body weight) hepatic inflammation by the adoptive transfer of Cd8+Btla+ T cells into fish that had or had not been administered sihvem-LV and/or soluble Hvem protein. The changes in ALT and AST release levels and apoptotic hepatocyte ratios were determined in the transfer assays to evaluate whether the tolerization of Cd8+Btla+ T cells is dependent on Hvem receptor proteins. Broken y-axis with disparate scales was adopted to display the disproportionate data. Error bars indicate SD. *p , 0.05, **p , 0.01. The Journal of Immunology 15 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 8. Determination of the functional role of Cd4+ T cells for the downregulation of Btla on Cd8+ T cells and involvement of Light in Btla–Hvem– mediated hepatic inflammation. (A and B) Changes in the transcriptional levels of btla and some other proinflammatory and apoptosis-related factors, including Ifn-g, Fasl, perforin, and Lamp-1, which are expressed in Cd8+Btla+ T cells after coculture with Cd4+ T cells for 24 or 72 h. (C) Changes in the transcriptional levels of Th1- or Th2-associated factors in Cd4+ T cells from the livers with or without ConA stimulation (200 mg/g body weight). (D) Mean fluorescence intensity (MFI) of Btla in Cd8+Btla+ T cells after coculture with Cd4+ T cells for 24 or 72 h. (E) The light mRNA is differentially expressed in Cd8+Btla+ and Cd8+Btla2 T cells. (F) The ratio of Cd8+Light+ and Cd8+Light2 T cells in the livers without (mock PBS–injected control) or with ConA stimulation (200 mg/g body weight). (G) The ratio of Cd8+Btla+Light2 and Cd8+Btla2Light+ T cells in Cd8+ T subsets sorted from the livers without (mock + + PBS–injected control, Cd8 TCtrl cells) or with ConA stimulation (200 mg/g body weight, Cd8 TConA cells). (H–J) Evaluation on the functional role of Light by blockade of Light on Cd8+Btla–Light+ T cells through anti-Light Ab. Changes in ALT and AST levels in sera and transcription levels of proinflammatory factors, perforin, Lamp-1, and apoptosis-related factors in the livers were examined in the blockade assays. Mock PBS and nonrelated IgG were used in the control groups. Broken y-axis with disparate scales was adopted to display the disproportionate data. (K) Adoptive transfer assay of Cd8+Light+ and Cd8+Light– T cells or Cd8+Light+ T cells that were pretreated with anti-Light Ab. Changes in serum ALT and AST levels were examined. (L) Ratios of intrahepatic Cd8+Btla+Light– and Cd8+Btla–Light+ T cells after coculture with Cd4+ T cells for 24 h. Unlabeled Cd8+ T cells served as negative control group. Error bars represent SD. *p , 0.05, **p , 0.01. 16 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION of Cd8+ T cells in ConA-induced hepatitis. The FCM analysis explored. As predicted, Btla expression was significantly declined indicated that Btla was differentially expressed on Cd8+ T cells in at the mRNA and protein levels when Cd8+Btla+Light– T cells the livers with or without ConA stimulation. Btla was highly were cocultured with Cd4+ T cells prestimulated to express Th1-typic expressed in Cd8+ T cells in healthy liver without stimulation. By cytokines, such as T-bet, Il-12, and Ifn-g, in the hepatitis-burdened contrast, Btla expression was dramatically decreased in Cd8+ liver. This process was accompanied by the upregulation of Light in T cells in liver with hepatitis. This observation was determined by Cd8+Btla–Light+ T cells, as determined by the increased per- the high proportion of Cd8+Btla+ T cells in healthy liver, which centage of Cd8+Btla–Light+ T subset and the decreased propor- decreased in the ConA-induced liver, accompanied with the tion of Cd8+Btla+Light– T subset. These results suggest that increase in the number of Cd8+Btla– T cells. In the in vitro Cd4+ Th1 cells modulate the polarization of Cd8+ Tcellsfrom cytotoxicity assay, Cd8+Btla+ T cells showed no significant Cd8+Btla+Light– to Cd8+Btla–Light+ T subsets, which is a key cytotoxicity to hepatocytes, whereas Cd8+Btla– T cells exhibited event in hepatitis. To our knowledge, this work is the first to report strong cytotoxicity. Similarly, the in vivo transfer of Cd8+Btla– the existence of Cd8+Btla+Light– and Cd8+Btla–Light+ T subsets T cells into ConA-induced liver enhanced hepatic inflammation in the liver. The Cd8+Btla+Light– T subset acts as an anti- and liver injury, whereas the transfer of Cd8+Btla+ T cells did inflammatory population and contributes to liver homeostasis not. These results indicate that Btla endowed Cd8+ T cells with and inhibition of hepatitis. The Cd8+Btla–Light+ T subset serves self-tolerance, which prevented Cd8+ T cells from attacking he- as a proinflammatory population and promotes hepatitis. However, patocytes, thereby maintaining liver homeostasis. However, Btla the precise mechanisms underlying Cd4+ T cell regulation for downregulation in Cd8+ T cells deprived the tolerization of Cd8+ Cd8+ T subset cell polarization remain to be further clarified.

T cells, leading to the disruption of liver homeostasis and the Liver is a metabolic and an immune organ with numerous Downloaded from occurrence of hepatitis. Cd8+Btla+ T cells showed high expression metabolites and exogenous toxins (49, 50). Hence, the liver is in an levels of Il-2, Lck, and Cxcr3 but a low level of Ccr7. These values immunotolerant state to prevent the improper activation of im- were similar to those of the activated Cd8+Btla2 T cells. Given the mune reactions in healthy individuals. Liver self-tolerance is upregulation of Il-2, Lck and Cxcr3 combined with the down- sustained through numerous approaches, including cellular and regulation of Ccr7 are typical markers for Cd8+ T cell activation molecular mechanisms (51). For example, T regulatory/Th17 cells + + (39, 46–48), indicating that Cd8 Btla T cells are anergic rather maintain the balance of the liver, in which T regulatory cells ex- http://www.jimmunol.org/ than just naive, and their activity is largely suppressed by Btla panded by Kupffer cells or primed by liver sinusoidal endothelial expression. The Co-IP assay showed that Btla clearly associated cells maintain the tolerant state by inhibiting Th17 and Th1 cell with Hvem in a manner similar to that in mammals. The toler- activation (52). In addition, PD-1 and CTLA-4 coinhibitory ization of Cd8+Btla+ T cells in hepatitis was dramatically impaired molecules expressed on exhausted T cells help sustain homeo- with Hvem knockdown in the liver but could be rescued by the stasis (53). In the current study, we provided new (to our knowl- administration of a recombinant soluble Hvem protein. These edge) insights into the contribution of the Btla–Hvem checkpoint observations suggest that the maintenance of Cd8+ T cell toleriza- axis to liver homeostasis by maintaining the tolerization of Cd8+ tion depends on the association of Btla with Hvem. The transition of T cells, whose disruption leads to the occurrence of hepatitis. Cd8+Btla+ T cells to Cd8+Btla– T cells in the ConA-induced liver Hepatitis is the inflammation of liver tissue and a severe clinical by guest on September 27, 2021 may largely contribute to hepatic inflammation. disease worldwide. The common causes of hepatitis include The costimulatory signal for the cytotoxic activity of Cd8+Btla– viruses, heavy alcohol use, nonalcoholic steatohepatitis, toxins, T cells was further explored. In mammals, LIGHT is another li- certain medications, and autoimmune diseases (50, 54–56). To gand of HVEM, which plays a costimulatory role in the activation date, the main research animals for hepatitis are mammals, such as of several T cell models. Thus, the involvement of zebrafish Light mouse and tree shrew (Tupaia belangeri), which are used as viral, in the Btla–Hvem axis and Cd8+Btla– T cell–elicited hepatitis was alcoholic, and drug-induced hepatitis models (55–57). ConA- investigated. light mRNA was weakly detected in Cd8+Btla+ induced hepatitis is one of the most widely used drug-induced T cells but was highly expressed in Cd8+Btla– Tcells.This hepatitis models that mimic T cell–dependent acute or chronic observation was supported by the detection of Cd8+Light+ hepatitis in humans (58). In the current study, we established a T cells, which were markedly increased in liver with hepatitis. ConA-induced hepatitis model in zebrafish. The model was sub- For clarification, Cd8+Btla–Light+ and Cd8+Btla+Light– Tsub- stantially verified by various pathological characteristics similar to sets were further distinguished from the sorted total liver Cd8+ those in mouse models. Such characteristics include lymphocyte T cells. The number of Cd8+Btla–Light+ T cells significantly infiltration; elevation of serum ALT and AST; upregulation of increased in the ConA-stimulated liver, accompanied with the proinflammatory cytokines, such as IL-1b, TNF-a, and IFN-g; decrease in the percentage of Cd8+Btla+Light– T cells and the and increased apoptosis of hepatocytes and tissue injury in the occurrence of hepatitis. The disparate expression of Btla and liver. Together with other liver disease models, such as alcoholic Light on Cd8+ T cells during hepatitis suggests that the latter hepatitis, which was effectively established in zebrafish, we con- plays a costimulatory role in the initiation of the cytotoxic activity of clude that zebrafish is a suitable nonmammalian model organism Cd8+ T cells, whose performance was opposite to the inhibitory for studying the mechanisms of liver diseases. effect of Btla on Cd8+ T cells. Thus, the transition of Cd8+Btla+- Various cells, such as liver sinusoidal endothelial cells, Kupffer Light– to Cd8+Btla–Light+ T cells under ConA stimulation may cells, NK, NKT, and CD4+ T cells, are involved in ConA-induced largely contribute to hepatic inflammation. Cd8+Btla–Light+ T cells hepatitis. CD4+ T cells are considered the predominant population were abolished by blockade of Light to provide functional support. for the occurrence of hepatitis because they initiate humoral au- As predicted, the blockade of Light attenuated the ConA-induced toimmune reactions through association with liver sinusoidal en- hepatitis. By contrast, the adoptive transfer of Cd8+Light+ T cells dothelial cells and Kupffer cells (59–61). However, the functional exacerbated the hepatitis. performance of CD8+ T cells in ConA-induced hepatitis remains Given that CD4+ T cells play crucial roles in ConA-induced elusive because controversial observations have been obtained by hepatitis in mouse models, the modulatory function of Cd4+ different research groups. For example, a mouse administered with T cells for the downregulation of Btla in Cd8+Btla+Light– T cells anti-CD4 mAb can be effectively protected from ConA-induced and the upregulation of Light in Cd8+Btla–Light+ T cells were liver injury, whereas anti-CD8 mAb treatment fails in the The Journal of Immunology 17 protection (58). The activated CD8+ T cells under ConA stimu- 17. Huang, Y., Y. Zhao, X. Ran, and C. Wang. 2014. Increased expression of her- + pesvirus entry mediator in 1,25-dihydroxyvitamin D3-treated mouse bone lation induce the activation of ST2 type 2 innate lymphoid cells, marrow-derived dendritic cells promotes the generation of CD4+CD25+Foxp3+ which can promote hepatic inflammation (62). Consistent with our regulatory T cells. Mol. Med. Rep. 9: 813–818. results, CD8+ T cells contribute to hepatic inflammation provoked 18. Chen, W., X. Lv, C. Liu, R. Chen, J. Liu, H. Dai, and G. M. Zou. 2018. He- matopoietic stem/progenitor cell differentiation towards myeloid lineage by ConA. Nevertheless, knowledge on the mechanisms underlying is modulated by LIGHT/LIGHT receptor signaling. J. Cell. Physiol. 233: + CD8 T cell regulation is still incomplete and requires extensive in- 1095–1103. vestigations. In the current study, we described the differential 19. Wu, T. H., Y. Zhen, C. Zeng, H. F. Yi, and Y. Zhao. 2007. B and T lymphocyte + + 2 + 2 + attenuator interacts with CD3zeta and inhibits tyrosine phosphorylation of functional roles of Cd8 Btla Light and Cd8 Btla Light Tsubsets TCRzeta complex during T-cell activation. Immunol. Cell Biol. 85: 590–595. in the pathogenesis of hepatitis. We added two new (to our 20. Miller, M. L., Y. Sun, and Y.-X. Fu. 2009. Cutting edge: B and T lymphocyte knowledge) Cd8+ T cell populations into the intrahepatic Cd8+ attenuator signaling on NKT cells inhibits cytokine release and tissue injury in early immune responses. J. Immunol. 183: 32–36. T cell family. This finding uncovered a previously unknown (to 21. Vendel, A. C., J. Calemine-Fenaux, A. Izrael-Tomasevic, V. Chauhan, D. Arnott, our knowledge) cellular mechanism underlying hepatic inflam- and D. L. Eaton. 2009. B and T lymphocyte attenuator regulates B cell receptor mation, which will benefit the development of therapeutic in- signaling by targeting Syk and BLNK. J. Immunol. 182: 1509–1517. 22. Granger, S. W., and S. Rickert. 2003. LIGHT-HVEM signaling and the regula- terventions for hepatic inflammatory disorders. tion of T cell-mediated immunity. Cytokine Growth Factor Rev. 14: 289–296. 23. Zhai, Y., R. Guo, T.-L. Hsu, G.-L. Yu, J. Ni, B. S. Kwon, G. W. Jiang, J. Lu, J. Tan, M. Ugustus, et al. 1998. LIGHT, a novel ligand for lymphotoxin beta Acknowledgments receptor and TR2/HVEM induces apoptosis and suppresses in vivo tumor for- mation via gene transfer. J. Clin. Invest. 102: 1142–1151. We thank Ying-ying Huang, Xing-hui Song, and Jing-yao Chen for techni- 24. Liu, W., C. Zhan, H. Cheng, P. R. Kumar, J. B. Bonanno, S. G. Nathenson, and cal support for FACS and immunohistochemical staining and She-long S. C. Almo. 2014. Mechanistic basis for functional promiscuity in the TNF and Downloaded from Zhang and Xin-hang Jiang for two-photon laser confocal scanning micro- TNF receptor superfamilies: structure of the LIGHT:DcR3 assembly. Structure scope capture and ultracentrifugation. 22: 1252–1262. 25. Cheung, T. C., I. R. Humphreys, K. G. Potter, P. S. Norris, H. M. Shumway, B. R. Tran, G. Patterson, R. Jean-Jacques, M. Yoon, P. G. Spear, et al. 2005. Disclosures Evolutionarily divergent herpesviruses modulate T cell activation by targeting the herpesvirus entry mediator cosignaling pathway. Proc. Natl. Acad. Sci. USA The authors have no financial conflicts of interest. 102: 13218–13223. 26. Cheung, T. C., M. W. Steinberg, L. M. Oborne, M. G. Macauley, S. Fukuyama, H. Sanjo, C. D’Souza, P. S. Norris, K. Pfeffer, K. M. Murphy, et al. 2009. http://www.jimmunol.org/ References Unconventional ligand activation of herpesvirus entry mediator signals cell 1. Chen, L., and D. B. Flies. 2013. Molecular mechanisms of T cell co-stimulation survival. [Published erratum appears in 2009 Proc. Natl. Acad. Sci. USA 106: and co-inhibition. [Published erratum appears in 2013 Nat. Rev. Immunol. 13: 16535–16536.] Proc. Natl. Acad. Sci. USA 106: 6244–6249. 542.] Nat. Rev. Immunol. 13: 227–242. 27. Murphy, T. L., and K. M. Murphy. 2010. Slow down and survive: enigmatic 2. Lucca, L. E., and D. A. Hafler. 2017. Co-inhibitory blockade while preserving immunoregulation by BTLA and HVEM. Annu. Rev. Immunol. 28: 389–411. tolerance: checkpoint inhibitors for glioblastoma. Immunol. Rev. 276: 9–25. 28. Ritthipichai, K., C. L. Haymaker, M. Martinez, A. Aschenbrenner, X. Yi, 3. Mitsuiki, N., C. Schwab, and B. Grimbacher. 2019. What did we learn from M. Zhang, C. Kale, L. M. Vence, J. Roszik, Y. Hailemichael, et al. 2017. + CTLA-4 insufficiency on the human immune system? Immunol. Rev. 287: 33–49. Multifaceted role of BTLA in the control of CD8 T-cell fate after antigen en- 4. Steinberg, M. W., O. Turovskaya, R. B. Shaikh, G. Kim, D. F. McCole, counter. Clin. Cancer Res. 23: 6151–6164. K. Pfeffer, K. M. Murphy, C. F. Ware, and M. Kronenberg. 2008. A crucial role 29. Tsang, J. Y. S., K. W. Chan, Y. B. Ni, T. Hlaing, J. Hu, S. K. Chan, S. Y. Cheung, for HVEM and BTLA in preventing intestinal inflammation. J. Exp. Med. 205: and G. M. Tse. 2017. Expression and clinical significance of herpes virus entry by guest on September 27, 2021 1463–1476. mediator (HVEM) in breast cancer. Ann. Surg. Oncol. 24: 4042–4050. 5. Zhao, Q., Z.-L. Huang, M. He, Z. Gao, and D.-M. Kuang. 2016. BTLA identifies 30. Quan, L., X. Lan, Y. Meng, X. Guo, Y. Guo, L. Zhao, X. Chen, and A. Liu. 2018. dysfunctional PD-1-expressing CD4+ T cells in human hepatocellular carci- BTLA marks a less cytotoxic T-cell subset in diffuse large B-cell lymphoma with noma. OncoImmunology 5: e1254855. high expression of checkpoints. Exp. Hematol. 60: 47–56.e1. 6. Watanabe, N., M. Gavrieli, J. R. Sedy, J. Yang, F. Fallarino, S. K. Loftin, 31. Inuo, M., K. Ihara, T. Matsuo, H. Kohno, and T. Hara. 2009. Association study M. A. Hurchla, N. Zimmerman, J. Sim, X. Zang, et al. 2003. BTLA is between B- and T-lymphocyte attenuator gene and type 1 diabetes mellitus or a lymphocyte inhibitory receptor with similarities to CTLA-4 and PD-1. Nat. systemic lupus erythematosus in the Japanese population. Int. J. Immunogenet. Immunol. 4: 670–679. 36: 65–68. 7. Sedy, J. R., M. Gavrieli, K. G. Potter, M. A. Hurchla, R. C. Lindsley, K. Hildner, 32. da Silva Antunes, R., A. K. Mehta, L. Madge, J. Tocker, and M. Croft. 2018. S. Scheu, K. Pfeffer, C. F. Ware, T. L. Murphy, and K. M. Murphy. 2005. B and TNFSF14 (LIGHT) exhibits inflammatory activities in lung fibroblasts com- T lymphocyte attenuator regulates T cell activation through interaction with plementary to IL-13 and TGF-b. Front. Immunol. 9: 576. herpesvirus entry mediator. Nat. Immunol. 6: 90–98. 33. Kim, H.-J., H.-M. Kim, C.-S. Kim, C.-S. Jeong, H.-S. Choi, T. Kawada, B.-S. Kim, 8. Montgomery, R. I., M. S. Warner, B. J. Lum, and P. G. Spear. 1996. Herpes and R. Yu. 2011. HVEM-deficient mice fed a high-fat diet are protected from simplex virus-1 entry into cells mediated by a novel member of the TNF/NGF adipose tissue inflammation and glucose intolerance. FEBS Lett. 585: 2285–2290. receptor family. Cell 87: 427–436. 34. Oya, Y., N. Watanabe, T. Owada, M. Oki, K. Hirose, A. Suto, S. Kagami, 9. Marsters, S. A., T. M. Ayres, M. Skubatch, C. L. Gray, M. Rothe, and H. Nakajima, T. Kishimoto, I. Iwamoto, et al. 2008. Development of autoim- A. Ashkenazi. 1997. Herpesvirus entry mediator, a member of the tumor necrosis mune hepatitis-like disease and production of autoantibodies to nuclear antigens factor receptor (TNFR) family, interacts with members of the TNFR-associated in mice lacking B and T lymphocyte attenuator. Arthritis Rheum. 58: 2498–2510. factor family and activates the transcription factors NF-kappaB and AP-1. 35. Oya, Y., N. Watanabe, Y. Kobayashi, T. Owada, M. Oki, K. Ikeda, A. Suto, J. Biol. Chem. 272: 14029–14032. S. Kagami, K. Hirose, T. Kishimoto, and H. Nakajima. 2011. Lack of B and 10. Shui, J.-W., A. Larange, G. Kim, J. L. Vela, S. Zahner, H. Cheroutre, and T lymphocyte attenuator exacerbates autoimmune disorders and induces Fas- M. Kronenberg. 2012. HVEM signalling at mucosal barriers provides host de- independent liver injury in MRL-lpr/lpr mice. Int. Immunol. 23: 335–344. fence against pathogenic bacteria. Nature 488: 222–225. 36. Wang, Y., S. K. Subudhi, R. A. Anders, J. Lo, Y. Sun, S. Blink, Y. Wang, 11. Carfı´, A., S. H. Willis, J. C. Whitbeck, C. Krummenacher, G. H. Cohen, J. Wang, X. Liu, K. Mink, et al. 2005. The role of herpesvirus entry mediator as a R. J. Eisenberg, and D. C. Wiley. 2001. Herpes simplex virus glycoprotein D negative regulator of T cell-mediated responses. J. Clin. Invest. 115: 711–717. bound to the human receptor HveA. Mol. Cell 8: 169–179. 37. Wahl, C., U. M. Wegenka, F. Leitha¨user, R. Schirmbeck, and J. Reimann. 2009. 12. Murphy, K. M., C. A. Nelson, and J. R. Sedy´. 2006. Balancing co-stimulation IL-22-dependent attenuation of T cell-dependent (ConA) hepatitis in herpes and inhibition with BTLA and HVEM. Nat. Rev. Immunol. 6: 671–681. virus entry mediator deficiency. J. Immunol. 182: 4521–4528. 13. Ware, C. F., and J. R. Sedy´. 2011. TNF Superfamily Networks: bidirectional and 38. Huang, C., and P. Niethammer. 2018. Tissue damage signaling is a prerequisite interference pathways of the herpesvirus entry mediator (TNFSF14). Curr. Opin. for protective neutrophil recruitment to microbial infection in zebrafish. Immu- Immunol. 23: 627–631. nity 48: 1006–1013.e6. 14. Otsuki, N., Y. Kamimura, M. Hashiguchi, and M. Azuma. 2006. Expression and 39. Shao, T., W. Shi, J. Y. Zheng, X.-X. Xu, A. F. Lin, L. X. Xiang, and J.-Z. Shao. function of the B and T lymphocyte attenuator (BTLA/CD272) on human 2018. Costimulatory function of Cd58/Cd2 interaction in adaptive humoral T cells. Biochem. Biophys. Res. Commun. 344: 1121–1127. immunity in a zebrafish model. Front. Immunol. 9: 1204. 15. Kashiwakuma, D., A. Suto, Y. Hiramatsu, K. Ikeda, H. Takatori, K. Suzuki, 40. Li, J. Y., K. Gao, T. Shao, D. D. Fan, C. B. Hu, C. C. Sun, W. R. Dong, A. F. Lin, S. Kagami, K. Hirose, N. Watanabe, I. Iwamoto, and H. Nakajima. 2010. B and L. X. Xiang, and J. Z. Shao. 2018. Characterization of an NLRP1 inflammasome T lymphocyte attenuator suppresses IL-21 production from follicular Th cells from zebrafish reveals a unique sequential activation mechanism underlying and subsequent humoral immune responses. J. Immunol. 185: 2730–2736. inflammatory caspases in ancient vertebrates. J. Immunol. 201: 1946–1966. 16. Gertner-Dardenne, J., C. Fauriat, F. Orlanducci, M.-L. Thibult, S. Pastor, 41. Feng, W., Y.-F. Gu, L. Nie, D.-Y. Guo, L. X. Xiang, and J. Z. Shao. 2016. J. Fitzgibbon, R. Bouabdallah, L. Xerri, and D. Olive. 2013. The co-receptor Characterization of SIGIRR/IL-1R8 homolog from zebrafish provides new BTLA negatively regulates human Vg9Vd2 T-cell proliferation: a potential way insights into its inhibitory role in hepatic inflammation. J. Immunol. 197: of immune escape for lymphoma cells. Blood 122: 922–931. 151–167. 18 REGULATION OF BTLA–HVEM AXIS IN HEPATIC INFLAMMATION

42. Wan, F., C. B. Hu, J. X. Ma, K. Gao, L. X. Xiang, and J. Z. Shao. 2017. 53. Cho, H., H. Kang, H. H. Lee, and C. W. Kim. 2017. Programmed cell death 1 Characterization of gd T cells from zebrafish provides insights into their im- (PD-1) and cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) in viral portant role in adaptive humoral immunity. Front. Immunol. 7: 675. hepatitis. Int. J. Mol. Sci. 18: 1517. 43. Yamaguchi, S., A. Nambu, T. Numata, T. Yoshizaki, S. Narushima, E. Shimura, 54. Liberal, R., and C. R. Grant. 2016. Cirrhosis and autoimmune liver disease: Y. Hiraishi, K. Arae, H. Morita, K. Matsumoto, et al. 2018. The roles of IL-17C current understanding. World J. Hepatol. 8: 1157–1168. in T cell-dependent and -independent inflammatory diseases. Sci. Rep. 8: 15750. 55. Piconese, S., I. Cammarata, and V. Barnaba. 2018. Viral hepatitis, inflammation, 44. Ye, L., Y. Lou, L. Lu, and X. Fan. 2019. Mesothelin-targeted second generation and cancer: a lesson for autoimmunity. J. Autoimmun. 95: 58–68. CAR-T cells inhibit growth of mesothelin-expressing tumors in vivo. Exp. Ther. 56. Bajaj, J. S. 2019. Alcohol, liver disease and the gut microbiota. Nat. Rev. Gastroenterol. Med. 17: 739–747. Hepatol. 16: 235–246. 45. Tian, A. Y., H. J. Yang, S. C. Zhu, Y. S. Zhang, Z. A. Jiang, J. Y. Song, H. Z. Liu, 57. Kouwaki, T., Y. Fukushima, T. Daito, T. Sanada, N. Yamamoto, E. J. Mifsud, and S. Q. Zhang. 2012. Isolation and characterization of LIGHT (TNFSF14) gene C. R. Leong, K. Tsukiyama-Kohara, M. Kohara, M. Matsumoto, et al. 2016. homologue in zebrafish (Danio rerio). Int. Immunopharmacol. 14: 629–634. Extracellular vesicles including exosomes regulate innate immune responses to 46. Tewari, K., J. Walent, J. Svaren, R. Zamoyska, and M. Suresh. 2006. Differential virus infection. Front. Immunol. 7: 335. requirement for Lck during primary and memory CD8+ T cell responses. Proc. 58. Tiegs, G., J. Hentschel, and A. Wendel. 1992. A T cell-dependent experi- Natl. Acad. Sci. USA 103: 16388–16393. mental liver injury in mice inducible by concanavalin A. J. Clin. Invest. 90: 47. Finlay, D., and D. A. Cantrell. 2011. Metabolism, migration and memory in 196–203. cytotoxic T cells. Nat. Rev. Immunol. 11: 109–117. 59. Wang, H.-X., M. Liu, S.-Y. Weng, J.-J. Li, C. Xie, H.-L. He, W. Guan, Y.-S. Yuan, 48. Hensel, M. T., T. Peng, A. Cheng, S. C. De Rosa, A. Wald, K. J. Laing, L. Jing, and J. Gao. 2012. Immune mechanisms of Concanavalin A model of autoimmune L. Dong, A. S. Magaret, and D. M. Koelle. 2017. Selective expression of CCR10 hepatitis. World J. Gastroenterol. 18: 119–125. and CXCR3 by circulating human herpes simplex virus-specific CD8 T cells. 60. Horst, A. K., C. Wegscheid, C. Schaefers, B. Schiller, K. Neumann, J. Virol. DOI: 10.1128/JVI.00810-17. S. Lunemann, A. E. Langeneckert, K. J. Oldhafer, C. Weiler-Normann, 49. Racanelli, V., and B. Rehermann. 2006. The liver as an immunological organ. K. S. Lang, et al. 2018. -related cell adhesion mole- Hepatology 43(2 Suppl. 1): S54–S62. cule 1 controls IL-2-dependent regulatory T-cell induction in immune-mediated 50. Yuan, Z., H. Zhang, M. Hasnat, J. Ding, X. Chen, P. Liang, L. Sun, L. Zhang, and hepatitis in mice. Hepatology 68: 200–214. Z. Jiang. 2019. A new perspective of triptolide-associated hepatotoxicity: liver 61. Zheng, C., S. Yin, Y. Yang, Y. Yu, and X. Xie. 2018. CD24 aggravates acute liver hypersensitivity upon LPS stimulation. Toxicology 414: 45–56. injury in autoimmune hepatitis by promoting IFN-g production by CD4+ T cells. Downloaded from 51. Heymann, F., and F. Tacke. 2016. Immunology in the liver--from homeostasis to Cell. Mol. Immunol. 15: 260–271. disease. Nat. Rev. Gastroenterol. Hepatol. 13: 88–110. 62. Zhang, Y., C. Qi, L. Li, S. Hua, F. Zheng, F. Gong, and M. Fang. 2019. 52. Zhang, H., Z. Jiang, and L. Zhang. 2019. Dual effect of 17 (Th17) CD8+ T cell/IL-33/ILC2 axis exacerbates the liver injury in Con A-induced and (Treg) in liver pathological process: from occurrence to end hepatitis in T cell-transferred Rag2-deficient mice. Inflamm. Res. 68: stage of disease. Int. Immunopharmacol. 69: 50–59. 75–91. http://www.jimmunol.org/ by guest on September 27, 2021 1 2

1

3 SUPPLEMENTAL FIGURE 1. Sequences, structural characterization, and 4 phylogenetic trees (unrooted) of btla (Btla), hvem (Hvem), and light (Light). (A and B) 5 Nucleotide and amino acid sequences of D. rerio btla (or Btla) (A) and hvem (or 6 Hvem) (B). The asterisk (*) represents the position where the translation ends. The 7 circles represent the conserved cysteine residues. The long and short underlines 8 indicate the signal peptides and potential N-glycosylation sites, respectively. The 9 shadow region indicates the leucine-rich region. The rectangular frames represent the 10 transmembrane areas. (C to E) Structural characterization of zebrafish Btla, Hvem, 11 and Light proteins. (C) Tertiary structures of BTLA and Btla ectodomains between 12 human (H. sapiens) and zebrafish (D. rerio). The conserved cysteine residues are 13 represented in light yellow. The HVEM binding site of BTLA and potential Hvem 14 binding site of Btla are indicated in bright orange. (D) Tertiary structures of HVEM 15 and Hvem ectodomains between human and zebrafish. The representative CRD1 and 16 CRD2 structures are presented. The conserved cysteine residues are presented in light 17 yellow. The potential BTLA/Btla binding site (DARC) is indicated in bright orange. 18 (E) Tertiary structures of LIGHT and Light ectodomains between human and 19 zebrafish. The GH loop structures are indicated in bright orange. The tertiary 20 structures were predicted by SWISS-MODEL and constructed by the PyMOL 21 software version 1.3. (F and G) The phylogenetic trees (unrooted) of the amino acid 22 sequences of Btla (F), Hvem (G), and Light (G) and their homologs in other species 23 were constructed using the neighbor-joining method by MEGA 7.0. Numbers 24 represent the percentage bootstrap confidence derived from 1000 replicates. The 25 Genbank accession numbers of the sequences are as follows: Homo sapiens BTLA, 26 NP_861445.3; Pan troglodytes BTLA, XP_001154904.1; Equus caballus BTLA, 27 XP_005602054.1; Ursus maritimus BTLA, XP_008704814.1; Mus musculus BTLA, 28 NP_001032808.2; Alligator sinensis BTLA, XP_006038011.3; Gallus gallus BTLA, 29 XP_429579.5; Columba livia BTLA, XP_005502656.1; Latimeria chalumnae Btla, 30 XP_014350589.1; Xiphophorus maculatus Btla, XP_014329250.2; Stegastes partitus 31 Btla, XP_008296967.1; Danio rerio Btla, MK112054; Cyprinus carpio Btla, 32 XP_018973925.1; Salmo salar Btla, XP_014029481.1; Oncorhynchus mykiss Btla, 33 XP_021453476.1; Salmo salar Ctla4 XP_014030171.1; Oncorhynchus mykiss Ctla4, 34 NP_001118005.1; Danio rerio Ctla4, XP_005167576.1; Xiphophorus maculatus Ctla4, 35 XP_005805334.1; Latimeria chalumnae Ctla4, XP_014345612.1; Gallus gallus 36 CTLA4, NP_001035180.1; Columba livia CTLA4, XP_005504072.1; Alligator 37 sinensis CTLA4, XP_006017712.1; Mus musculus CTLA4, NP_033973.2; Homo 38 sapiens CTLA4, NP_005205.2; Pan troglodytes CTLA4, XP_526000.1; Equus 39 caballus CTLA4, XP_023478240.1; Ursus maritimus CTLA4, XP_008685618.1; 40 Canis lupus familiaris LIGHT (TNFSF14), NP_001271423.1; Callorhinus ursinus 41 LIGHT (TNFSF14), XP_025717251.1; Felis catus LIGHT (TNFSF14), 42 XP_003981774.1; Equus caballus LIGHT (TNFSF14), XP_001496269.1; Tursiops 43 truncatus LIGHT (TNFSF14), XP_004320567.1; Delphinapterus leucas LIGHT 2

44 (TNFSF14), XP_022413059.1; Homo sapiens LIGHT (TNFSF14), NP_003798.2; 45 Pan troglodytes LIGHT (TNFSF14), NP_001315249.1; Macaca mulatta LIGHT 46 (TNFSF14), XP_014978121.1; Macaca fascicularis LIGHT (TNFSF14), 47 NP_001274576.1; Cavia porcellus LIGHT (TNFSF14), XP_003460758.1; Mus 48 musculus LIGHT (TNFSF14), NP_062291.1; Rattus norvegicus LIGHT (TNFSF14), 49 NP_001178732.1; Bos taurus LIGHT (TNFSF14), NP_001095325.1; Capra hircus 50 LIGHT (TNFSF14), NP_001301258.1; Python bivittatus LIGHT (TNFSF14), 51 XP_025019660.1; Danio rerio Light (Tnfsf14), NP_001268924.1; Gallus gallus 52 HVEM (TNFRSF14), XP_418509.4; Homo sapiens HVEM (TNFRSF14), 53 NP_003811.2; Pan troglodytes HVEM (TNFRSF14), XP_513730.2; Equus caballus 54 HVEM (TNFRSF14), NP_001075376.1; Ursus maritimus HVEM (TNFRSF14), 55 XP_008706001.1; Mus musculus HVEM (TNFRSF14), NP_849262.1; Xenopus 56 tropicalis HVEM (TNFRSF14), XP_004916213.1; Latimeria chalumnae Hvem 57 (Tnfrsf14), XP_005986083.1; Alligator sinensis HVEM (TNFRSF14), 58 XP_014378794.1; Columba livia HVEM (TNFRSF14), XP_005514384.2; Salmo 59 salar Hvem (Tnfrsf14), NP_001135338.1; Xiphophorus maculatus Hvem (Tnfrsf14), 60 XP_014329182.2; Oryzias latipes Hvem (Tnfrsf14), XP_023812845.1; Cyprinus 61 carpio Hvem (Tnfrsf14), XP_018925960.1; Danio rerio Hvem (Tnfrsf14), 62 MK112056; Oncorhynchus mykiss Hvem (Tnfrsf14), XP_021472570.1. 63

3

64 65 66 SUPPLEMENTAL FIGURE 2. Multiple amino acid sequence alignment of BTLA (or 67 Btla) (A), HVEM (or Hvem) (B), and LIGHT (or Light) (C). The conserved and 68 partially conserved amino acid residues in each species are marked by black and gray 69 backgrounds, respectively. Ig-like domain, transmembrane region, Grb2 binding sites, 70 immunoreceptor tyrosine-based inhibitory motifs (ITIM), immunoreceptor 71 tyrosine-based switch motifs (ITSM), TNFR family cysteine-rich region, and TNF 72 homology domain (THD) are marked at the top of the sequence. The GenBank 73 accession numbers of the sequences are noted in SUPPLEMENTAL FIGURE 1.

4

74 75 76 SUPPLEMENTAL FIGURE 3. Assays for the effectiveness and specificity of Abs 77 and lentivirus (LV). (A to D) Western blot analyses for mouse or rabbit anti-Btla (A), 78 anti-Hvem (B), anti-Light (C), and anti-BHMT (D) Abs that bind to the corresponding

5

79 recombinant target GST-/Myc-fusion proteins expressed in prokaryotic/eukaryotic 80 systems and natural proteins from liver leukocytes (LL), hepatocytes (Hep), and liver 81 tissues. The molecular weights (MWs) of Btla–GST fusion protein with ectodomain 82 of Btla, natural Btla protein from liver leukocytes, Hvem–Myc fusion protein, natural 83 Hvem protein from hepatocytes, Light–Myc fusion protein, natural Light protein from 84 liver leukocytes, and natural Bhmt protein from hepatocytes and liver tissues were 85 determined to be 41.6, 34.4, 31.6, 29.2, 29.2, 26.8, and 44.0 kDa, respectively, which 86 were consistent with the expected MWs. (E) FCM analysis for the reactive activity of 87 rabbit anti-BHMT to zebrafish hepatocytes showing that the majority (77.90%) of the 88 isolated hepatocytes were marked by the Ab. (F and G) Screening the effective 89 siRNAs targeting btla or hvem mRNAs in HEK293T cells by qRT-PCR. pSUPER 90 plasmid containing the scrambled siRNA was used as the control. (H to J) 91 Determination of the packages and titers of the constructed sibtla-LV and sihvem-LV 92 in HEK293T cells by fluorescence microscopy and FCM analysis. The cells without 93 LVs infection were used as controls, which are presented as outlines in the diagrams. 94 Scale bar: 50 μm. (K to M) In vivo examination of the interference activities of 95 sibtla-LV and sihvem-LV in zebrafish liver leukocytes and Cd8+ T cells (K and L) and 96 hepatocytes (M) by qRT-PCR under ConA stimulation. Scrambled siRNA-LV was 97 used as a control. Error bars represent the standard deviation, *, p < 0.05; **, p < 0.01. 98 (N) Western blot analysis for the interference activities of sibtla-LV and sihvem-LV in 99 ConA stimulated zebrafish liver leukocytes and hepatocytes. Scrambled siRNA-LV 100 was used as the control. (O) Examination of the deletion efficiency of anti-Cd8 Ab 101 on Cd8+ T cells by FCM analysis. (P) Non-related mouse IgG and rabbit IgG used as 102 isotype controls in this study were examined by immunohistochemical staining of 103 liver sections and showed negative reactions. Scale bar: 50 μm. 104

6