Five to 8-week-old male C57BL/6J, BALB/c, and nude mice were purchased from Beijing Hua Fukang Bioscience Co., Ltd. (Beijing, China). TCRδ^−/− mice were a gift from Dr. Zhinan Yin (Nankai University). CD45.1⁺ mice were a gift from Dr. Zhigang Tian (University of Science and Technology of China). All mice were housed in a specific pathogen-free facility under ethical conditions. Experiments were performed with age- and sex-matched animals according to the guidelines for experimental animals from Shandong University and were approved by the Committee on the Ethics of Animal Experiments of Shandong University (Jinan, China) (2017-D023).

The pAAV plasmid and the pAAV-HBV1.2 plasmid containing the full-length HBV DNA were kindly provided by Dr. Peijer Chen (National Taiwan University College of Medicine, Taipei, China). An acute HBV infection model was established by hydrodynamic injection of 20 μg pAAV-HBV1.2 plasmid in 1× phosphate buffer solution (PBS) equivalent to 10% of the mouse body weight by way of the tail veil into wild-type (WT) C57BL/6J and TCRδ^−/− mice (24).

Parabiosis was established as described previously (25). In brief, 5- to 6-week-old male CD45.1⁺ and CD45.2⁺ C57BL/6J mice were matched for body weight and anaesthetized prior to surgery. An incision along the lateral aspect was performed from 0.5 cm above the elbow to 0.5 cm below the knee joint, and the skin was gently detached from the subcutaneous fascia. The knee joints and skins of the paired mice were attached from the elbow to the knee with nonabsorbable silk. Neomycin was added to the drinking water for 7 days. Two weeks after parabiosis, two mice shared the blood circulation.

Mice were sacrificed, and the bone marrow (BM), inguinal lymph node (iLN), spleen, liver, thymus, intestine, and skin were collected. MNCs from each organ were separated and mononuclear cells (MNCs) were collected. In brief, iLN and thymus were passed through a 200-gauge steel mesh and centrifuged at 400×g for 5 min in 1 × PBS. Spleens were first passed through a 200-gauge steel mesh, lysed with red blood cell (RBC) lysis buffer and washed with 1 × PBS. Bone marrows were washed by syringe, and MNCs were harvested after RBC lysis and washing. Livers were passed through a 200-gauge steel mesh, and the cell pellets were collected. MNCs were isolated by gradient centrifugation with 40% Percoll at 800×g for 25 min and harvested after RBC lysis and washing. Skins were excised and minced into small pieces, then digested with 0.04% collagenase I (Gibco, Carlsbad, CA, USA) for 2 h at 37°C. The large pieces were removed by filtration, and the MNCs were obtained by gradient centrifugation with 40% and 70% Percoll. IELs were isolated from the small intestine as previously described (26, 27). Peyer’s patches (PP) were surgically removed from the intestines and excised into small pieces. The specimens were then digested with prewarmed IEL digestive juice and incubated with stirring for 40 min at 37°C and passed over two nylon wool columns to remove undigested tissue debris. IELs were isolated by gradient centrifugation with 40% and 70% Percoll (GE Healthcare, Uppsala, Sweden).

The mice were anesthetized prior to a laparotomy to obtain the LSECs. The liver was perfused with EGTA/HBSS solution through the portal vein. The inferior vena cava was rapidly cut after the liver turned completely pale. Next, the liver was perfused with 37°C prewarmed Collagenase IV (Gibco) solution for 5 min, excised, and digested in collagenase solution for 15 min in a 37°C incubator. LSECs were obtained by gradient centrifugation with 25% and 50% Percoll. The cellular precipitate was resuspended in 1 ml DMEM+10% FBS medium and seeded into 24-well plates at a density of 1 × 10⁵ cells/well. After 4 h, the supernatant was gently removed and replaced with 500 μl fresh DMEM medium.

MNCs in a single-cell suspension were blocked with anti-CD16/32 (eBioscience, San Diego, CA, USA) and stained with a cocktail of antibodies specific for different cell types for 30 min at 4°C. The foxp3 staining kit (eBioscience) was used to stain the nucleus transcription factor. To detect expression of IL-17A and IFN-γ, the cells were stimulated with phorbol 12-myristate 13-acetate (PMA, 30 ng/ml, Sigma-Aldrich Co., St. Louis, MO, USA) and ionomycin (1 μg/ml, Sigma-Aldrich) for 4 h and treated with monensin (1 μg/ml, Sigma-Aldrich) and brefeldin A (BFA, 1 μg/ml, Biolegend, San Diego, CA, USA) after 1 h of stimulation. After surface staining, cells were fixed, permeabilized, and stained with the indicated intracellular antibodies. Flow cytometry was performed using a fluorescence-activated cell sorter (FACS) Aria III (BD Biosciences, San Jose, CA, USA), and data were analyzed with FlowJo software (TreeStar Inc., Ashland, OR, USA). At least 3 × 10⁵ total events were needed for the FACS analysis. Data are presented in Supplementary Table S1 .

Hepatic MNCs were stained with anti-CD3e, anti-TCRγ/δ, anti-CXCR3, and anti-CXCR6 antibodies to sort CXCR3⁺CXCR6⁺ γδT cells using a FACS Aria III (BD Biosciences). The purity of the sorted cell populations was >95%. Approximately 50,000 purified CXCR3⁺CXCR6⁺ γδT cells in 200 μl 1 × PBS were intravenously injected into sublethally irradiated TCRδ^−/− mice (6.5 Gy given 1 day before adoptive transfer). An acute HBV infection model was established in WT, TCRδ^−/−, and TCRδ^−/− mice after adoptive transfer of CXCR3⁺CXCR6⁺ γδT cells.

Total γδT-cell RNA from the indicated tissues was extracted using TRIZOL reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. The RNA concentration was quantified using Nanodrop 2000 (Thermo Fisher, Waltham, MA, USA). cDNA was generated using a FastQuant RT Kit (Tiangen Biotech Co. Ltd., Beijing, China) and real-time polymerase chain reaction was performed using a SYBR Green SuperMix (Roche, Basel, Switzerland) on a LightCycler 480 platform (Roche). Relative mRNA levels were normalized to β-actin mRNA levels. The primers are described in Supplementary Table S2 .

Hepatic MNCs were seeded at 5 × 10⁵ in 100 μl serum-free RPMI-1640 media in the above inserts with a pore size of 0.4 μm. The lower chamber contained 500 μl DMEM medium with untreated LSECs or LSECs incubated with neutralizing chemokine-targeting antibodies at 37°C for 2 h. After a 2-h incubation, absolute numbers of the migrated cells were detected by FACSAria III.

The LSECs were inoculated into 24-well plates (1 × 10⁵ cells/well). After culturing for 4 h, the supernatant was absorbed, washed three times with 1 × PBS, and replaced with fresh DMEM medium (500 μl). Cell culture supernatants were collected at 24 h. CXCL9, CXCL10, CXCL11, and CXCL16 chemokine levels were detected in the LSEC culture supernatants by ELISA (PeproTech, East Windsor, New Jersey, USA) according to the manufacturers’ instructions.

Serum hepatitis B surface Ag (HBsAg) and hepatitis BeAg (HBeAg) from mice with acute HBV infection were assayed using the HBsAg and HBeAg detection kits (Autobio, Zhengzhou, China) according to the manufacturer’s instructions.

Serum HBV DNA copies were extracted from 50 μl serum and detected by quantitative PCR using a diagnostic kit for HBV DNA (Da An Gene, Guangzhou, China) according to the manufacturer’s instructions.

The data were analyzed using GraphPad Prism 7 software (GraphPad Software Inc., San Diego, CA, USA). Data are presented as the mean ± standard error of the mean (SEM). Differences between more than two groups were statistically analyzed by one-way analysis of variance (one-way ANOVA). A two-way ANOVA test was used to determine differences in HBsAg, HBeAg, and HBV DNA tests. A p-value <0.05 was considered statistically significant (^* p < 0.05; ^** p < 0.01; ^*** p < 0.001).

IELs and skin are rich in γδT cells, where they regulate inflammation, pathogen invasion, and tumor surveillance (22, 28, 29). In this study, the liver also had a higher proportion of γδT cells than other immune organs and tissues, including the BM, inguinal LN, spleen, and thymus ( Supplementary Figure S1 and Figures 1A, B ). This finding correlates with results from a recent study (23).

This study also assessed the phenotype of γδT cells in liver, first by analyzing molecules associated with γδT-cell-induced cytokine production. Hepatic γδT cells expressed low levels of CD27, which is a fate determinant of γδT cells to express IFN-γ but not IL-17A. The expression was similar in γδT cells from the IEL but significantly lower than γδT cells from the spleen and thymus (30) ( Figure 1C ). Hepatic γδT cells also expressed higher levels of IL-2 receptor beta CD122, which is expressed on IFN-γ⁺ γδT cells, compared with γδT cells from other immune organs and tissues (31) ( Figure 1C ). Hepatic γδT cells exhibited higher NKG2D expression, indicating that these cells may be involved in immune defense and tissue protection in the liver ( Figure 1D ). γδT cells induce effector−target cell apoptosis through the Fas–FasL pathway (32, 33). Hepatic γδT cells have a higher expression of Fas (CD95) than intestinal epithelial γδT cells but lower expression than thymic γδT cells ( Figure 1D ). Importantly, hepatic γδT cells expressed a high level of the adhesion and retention molecule, CD44, and a low level of the lymph node homing molecule, CD62L ( Figure 1E ). Overall, these data indicate that hepatic γδT cells exhibited a unique phenotype that was distinct from γδT cells found in other organs.

Chemokines and chemokine receptors regulate the trafficking and accumulation of lymphocytes in homeostasis and disease. The mRNA level of chemokine receptors in γδT cells was measured in multiple organs. As shown in Figure 2A , CXCR3, CXCR6, and CCR2 mRNA levels were higher in hepatic γδT cells than in other tissues and organs. Hepatic γδT cells from nude mice still retained high CXCR3 and CXCR6 expression ( Figure 2B ), indicating that thymus deficiency did not impact expression. CXCR3 and CXCR6 expression was further examined by flow cytometry. Results showed that hepatic γδT cells expressed higher levels of CXCR3 and CXCR6 than those from spleen, IEL, and thymus in C57BL/6J mice ( Supplementary Figure S2 and Figure 2C ). Furthermore, the CXCR3⁺CXCR6⁺ γδT-cell subset was found in liver but not in the spleen, IEL, or thymus ( Figure 2C ). These data indicate that hepatic γδT cells exhibited a unique phenotype and the CXCR3⁺CXCR6⁺ γδT subset specifically exists in liver.

Tissue-resident lymphocytes often express CD103, CD69, CD44, and CD49a, which mediate adhesion and retention, and lack the lymphoid homing markers, CD62L and CCR7 (6, 34). In this study, hepatic CXCR3⁺CXCR6⁺ γδT cells expressed high levels of CD69, CD103, CD44, and CD49a ( Figure 3A ) and low levels of CD62L and CCR7 ( Figure 3B and Supplementary Figure S3 ). These cells also expressed a high level of the transcription factor, PLZF, which is important for innate tissue-resident lymphocytes like iNKT cells and MAIT cells (35), and a low level of BlIMP1, which correlates with diverse tissue-resident lymphocyte populations such as T_RM and NKT cells ( Figure 3C ). These data indicate that hepatic CXCR3⁺CXCR6⁺ γδT cells have tissue-resident characteristics.

To further confirm whether these cells specifically reside in liver, CD45.1⁺ and CD45.2⁺ mice were surgically joined by parabiosis ( Figure 3D ). After 2 weeks, the percentages of CD45.1⁺ or CD45.2⁺ cells in peripheral blood lymphocytes (PBLs) were equivalent, indicating that the parabiosis models were successfully established ( Figure 3E ). The redistribution of γδT cells in each mouse in the CD45.1/CD45.2 parabiotic mouse pairs was assessed ( Figure 3F ). While conventional T cells were mutually exchanged, hepatic CXCR3⁺CXCR6⁺ γδT cells from the CD45.1⁺ parabiont mice were almost entirely CD45.1⁺, with very few CD45.2⁺, while CXCR3⁺CXCR6⁺ γδT cells from the CD45.2⁺ parabiont mice were primarily CD45.2⁺ ( Figures 3G, H ). These data suggest that hepatic CXCR3⁺CXCR6⁺ γδT cells were seldom exchanged between CD45.1⁺ and CD45.2⁺ parabiont mice. In addition, 70%–80% CXCR3⁺CXCR6⁻ γδT cells and CXCR3⁻CXCR6⁺ γδT cells were CD45.1⁺ in CD45.1⁺ parabiont mice, whereas the reverse was observed in CD45.2⁺ parabiont mice ( Supplementary Figure S4 ). In summary, the parabiosis assay indicated that CXCR3⁺CXCR6⁺ γδT cells were predominantly retained in the liver, while CXCR3⁺CXCR6⁻ γδT cells and CXCR3⁻CXCR6⁺ γδT cells have less liver-resident features.

In general, γδT cells are exported from the thymus at defined periods of fetal and neonatal development, and then migrate to and populate different peripheral tissues in adult animals (36). However, intestinal γδT cells can also develop in intestinal crypts and reside in the gut (37). To explore whether the existence of hepatic γδT cells was dependent on the thymus, the proportion of γδT cells in athymic nude and WT BALB/c mice were assessed. The proportion of intestinal γδT cells in nude mice was nearly identical to the proportion of intestinal γδT cells in WT BALB/c mice. Importantly, the liver still contained γδT cells in athymic nude mice despite the lack of γδT cells in the spleen ( Supplementary Figures S5A, B ). These data suggest that some hepatic γδT cells are not dependent on the thymus. Assessment of γδT cells during different stages of growth showed that they are found in liver and thymus from embryo to adulthood. As the mice grew, the percentage of γδT cells decreased in the thymus and increased in the liver ( Figure 4A and Supplementary Figure S6 ). More importantly, CXCR3⁺CXCR6⁺ γδT cells were present in the liver from embryo to adulthood and the proportion increased as the mice aged; however, they were absent in the thymus during particular stages of growth ( Figure 4B ).

CXCR3⁺CXCR6⁺ γδT cells expressed high levels of the IFN-γ-associated transcriptional factor, T-bet, from embryo to adulthood, while rarely expressing the IL-17A-associated transcriptional factor, RORγt ( Figure 4C ). In addition, hepatic CXCR3⁺CXCR6⁺ γδT cells expressed a high level of tissue-resident-associated molecules like CD103 and CD69 and the transcription factor, PLZF, but rarely expressed CD62L during development ( Figure 4D and Supplementary Figure S7 ).

In summary, these data collectively indicate that CXCR3⁺CXCR6⁺ γδT cells are retained in liver from embryo to adulthood and have tissue-resident characteristics during mouse development.

LSECs secrete a variety of chemokines that promote the recruitment of immune cells (38, 39). Results from the present study showed that the liver-resident γδT-cell subset specifically expressed the chemokine receptors, CXCR3 and CXCR6. Thus, we hypothesized that chemotaxis may play an important role in the recruitment and retention of CXCR3⁺CXCR6⁺ γδT cells in the liver. LSECs were isolated and chemokines, including the ligands of CXCR3 (CXCL9, CXCL10, CXCL11) and the ligand of CXCR6 (CXCL16), were measured in the culture supernatants. LSECs produced high levels of CXCL9, CXCL11, and CXCL16 ( Figure 5A ). To further demonstrate the recruitment function of LSECs, a transwell assay was performed to test the chemotactic ability of chemokines secreted by LSECs on CXCR3⁺CXCR6⁺ γδT cells ( Figure 5B ). When the lower chamber contained untreated LSECs, CXCR3⁺CXCR6⁺ γδT cells were recruited from the above inserts. Chemotaxis was significantly reduced when CXCL9 and CXCL11 were neutralized, indicating that the chemokines, CXCL9 and CXCL11, play important roles in the retention of CXCR3⁺CXCR6⁺ γδT cells ( Figure 5C ). When CXCL16 was neutralized, the level of chemotaxis between the LSECs and CXCR3⁺CXCR6⁺ γδT cells was also weakened ( Figure 5C ). These findings indicate that the chemokines, CXCL9, CXCL11, and CXCL16, are all required for the retention of CXCR3⁺CXCR6⁺ γδT cells. Taken together, these results indicate that LSECs play an important role in retaining liver-resident CXCR3⁺CXCR6⁺ γδT cells through the interaction between the chemokine receptors, CXCR3 and CXCR6, and their ligands.

Liver-resident CXCR3⁺CXCR6⁺ γδT cells produced high levels of IFN-γ and expressed a high level of the IFN-γ transcription factor, T-bet ( Supplementary Figure S8 and Figures 6A, B ). However, these cells produced very little IL-17A and expressed low levels of the IL-17A transcription factor, RORγt ( Figures 6A, B ). The production of IFN-γ indicated that liver-resident CXCR3⁺CXCR6⁺ γδT cells may be involved in protective immunity against viruses and intracellular pathogens. In an acute HBV infection model, the proportion and absolute numbers of CXCR3⁺CXCR6⁺ γδT cells increased on day 7 and returned to normal levels as the infection progresses ( Figure 6C ). The expression of Ki67 was measured at different time points. On the third day after infection, CXCR3⁺CXCR6⁺ γδT cells showed significantly higher expression of Ki67, indicating that the proliferative capacity of CXCR3⁺CXCR6⁺ γδT cells was enhanced postinfection ( Figure 6D ). Notably, IFN-γ levels increased after 7 days postinfection ( Figure 6E ). These results indicate that liver-resident CXCR3⁺CXCR6⁺ γδT cells proliferate during acute HBV infection and IFN-γ from these cells may be involved in eliminating HBV. IL-17A secretion increased at the early stage of infection, and decreased after 10 days, and was then maintained at a low level ( Figure 6E ).

To further investigate the role of CXCR3⁺CXCR6⁺ γδT cells during HBV infection, the acute HBV infection model was established in WT, TCRδ^−/− mice, and TCRδ^-/- mice that were adoptively transferred with CXCR3⁺CXCR6⁺ γδT cells. HBsAg, HBeAg, and HBV DNA levels were measured in serum at different time points. TCRδ^−/− mice that had received CXCR3⁺CXCR6⁺ γδT cells exhibited markedly lower serum HBsAg, HBeAg, and HBV DNA levels than TCRδ^−/− mice that did not receive CXCR3⁺CXCR6⁺ γδT cells ( Figure 6F ). These data suggest that liver-resident CXCR3⁺CXCR6⁺ γδT cells may be involved in the clearance of HBV. Interestingly, TCRδ^−/− mice had lower serum HBsAg, HBeAg, and HBV DNA than WT mice, suggesting that other subsets of γδT cells may have different or even opposing effects during acute HBV infection. The roles of other subsets of γδT cells in acute HBV infection deserved further research.

These findings demonstrate that liver-resident CXCR3⁺CXCR6⁺ γδT cells exert a protective role during acute HBV infection through production of IFN-γ.