Invariant NKT Cells Promote the Development of Highly Cytotoxic Multipotent CXCR3⁺CCR4⁺CD8⁺ T Cells That Mediate Rapid Hepatocyte Allograft Rejection

This article is featured in Top Reads, p.

Liver transplant is the optimal therapy for end-stage liver disease, although access to this life-saving treatment is limited by the shortage of donor organs. Strategies to increase the long-term survival of the available livers as well as alternatives to whole liver transplant would mitigate the donor organ shortage (1). For patients with end-stage liver disease, hepatocyte transplant can serve as a bridge to liver transplant or alleviate the loss of liver function (2, 3). Hepatocytes can be isolated from heart-beating or non–heart-beating donors with excellent ex vivo viability (4). Clinical hepatocyte transplant has been successful in partially treating both acute liver failure (5–8) and liver-based metabolic disorders (2, 3). However, in contrast to the tolerogenicity of liver allografts (9), human hepatocyte transplants are vulnerable to both Ab-mediated rejection (10, 11) and cellular-mediated rejection (12) despite the use of conventional immunosuppression that successfully prevents acute rejection after solid organ transplantation. Thus, the clinical experience to date indicates that despite demonstrable initial hepatocyte transplant function, all hepatocellular allograft recipients eventually required whole liver transplantation because of immune-mediated rejection. These, and other challenges, hinder widespread application of this cellular therapy.

Cellular-mediated rejection of hepatocellular transplants has been investigated in our laboratory using a well-characterized mouse model of hepatocyte transplant (13, 14). We have reported that hepatocyte rejection is T cell dependent (13–15) and that CD8-mediated rejection occurs with or without CD4⁺ T cell help (16–18). CD4-independent, CD8⁺ T cell–mediated allograft rejection is resistant to costimulatory blockade and implicated in the rejection of hepatocellular allografts (13, 19), intestinal allografts (20), skin allografts (21, 22), and cardiac allografts (23, 24). In addition, we have reported that the liver microenvironment is sufficient to prime CD8⁺ T cells in the absence of both CD4⁺ T cells and secondary lymphoid tissue (16). Alloprimed CD4-independent CD8⁺ T cells in hepatocyte allograft recipients are distinguished by TNF-α−dependent cytotoxic effector function that peaks on day 5 posttransplant compared with CD4-dependent, CD8⁺ T cell–mediated Fas ligand (FasL)– and perforin-dependent cytotoxicity that peaks on posttransplant day 7 (17, 18). Given these data and the known abundance of invariant NKT (iNKT) cells in the liver, we hypothesized that liver iNKT cells may contribute to the activation of potent CD8⁺ T cells that mediate rejection of allogeneic hepatocytes transplanted to the liver.

iNKT cells represent a large portion, 20–30%, of murine liver mononuclear cells (LMNCs) (25, 26) and have been identified as a key regulator of adaptive immune responses (27). iNKT cells are characterized by the expression of invariant TCR with activation following recognition of glycolipid Ag presented by CD1d on APCs (primarily dendritic cells [DCs] or B cells) (27). In addition, iNKT cells can be indirectly activated by inflammatory cytokine milieus, such as with IL-12–rich microenvironments (28). iNKT cells are primarily localized in the liver with smaller populations in the blood, spleen, bone marrow, thymus, and lymph nodes in mice and humans (29, 30).

Current literature regarding iNKT cell–mediated effects on cellular immunity are mixed on whether these cells enhance or suppress CD8⁺ T cells. iNKT cells are reported to enhance (31–34) or inhibit (35, 36) CD8⁺ T cell effector function toward tumor, hapten, or OVA Ag. One caveat is that these studies exclusively used exogenous sea sponge glycolipid α-galactosylceramide (α-GalCer) to stimulate iNKT cells and did not investigate endogenously activated iNKT cell activity. An additional confounding factor is that a single dose of exogenous α-GalCer stimulates Th1 responses acutely, and subsequent doses skew the immune responses toward Th2 responses (37).

In this study, we investigated the extent to which iNKT cells influence the development and effector function of CD8⁺ cytolytic T cells (in the presence or absence of CD4⁺ T cells) in response to allogeneic hepatocyte transplant without the use of exogenous α-GalCer stimulation. Studies by Toyofuko et al. (38) have reported that iNKT cell–mediated acceleration of islet allograft rejection occurs when islet allografts are transplanted via intraportal injection but not when transplanted by kidney subcapsular injection, suggesting their differential impact based on the local intrahepatic immunity. We hypothesized that iNKT cells in the liver may contribute to alloreactive cell–mediated immunity by enhancing the development and/or function of CD8⁺ T cytolytic cells (with or without CD4⁺ T cell help) after allogeneic hepatocyte transplant. The aim of this study was to investigate the contribution of iNKT cells to the development of alloprimed, CD8-mediated allocytotoxicity and hepatocyte rejection under CD4⁺ T cell–replete and CD4⁺ T cell–deficient conditions and to determine their influence on strength of cytotoxicity, hepatocyte allograft rejection, and compartmentalization of CD8⁺ T cell responses in the liver versus in lymphoid tissue.

C57BL/6 (wild-type [WT]), CD8 knockout (KO), and RAG1 KO (all H-2^b; The Jackson Laboratory) as well as FVB/N (H-2^q; Taconic) mouse strains (all 6–12 wk of age) were used in this study. Jα18 KO mice (H-2^b, backcrossed more than eight times onto a C57BL/6 background) were provided to Dr. R. Brutkiewicz by Dr. L. van Kaer (Vanderbilt University, Nashville, TN) with permission from Dr. M. Taniguchi (Chiba University, Chiba, Japan). Transgenic FVB/N mice expressing human α-1 antitrypsin (hA1At) were the source of donor hepatocytes, as previously described (39). All experiments were performed in compliance with the guidelines of the Institutional Animal Care and Use Committee of The Ohio State University (Protocol 2019A00000124).

Hepatocyte isolation and purification was performed, as previously described (39). Hepatocyte viability and purity was consistently >95%. hA1At transgenic FVB/N donor mice (H-2^q) were the source of hepatocytes for transplantation. Donor hepatocytes (2 × 10⁶) were transplanted by intrasplenic injection with circulation of donor hepatocytes to the host liver, where they engraft, as previously described (39). FVB/N hepatocyte allograft rejection was determined by detection of secreted hA1At in serial recipient serum samples by ELISA (13, 39). The reporter protein hA1At does not elicit an immune response and syngeneic, hA1At-expressing hepatocytes survive long term (39).

Cohorts of WT and Jα18 KO recipients were depleted of circulating CD4⁺ T cells using mAb (GK1.5; Bioexpress Cell Culture Services, West Lebanon, NH) by i.p. injection (250 µg, days −4, −2, 7, 14, 21, 28, 35, and 42 relative to hepatocyte transplant). In other cohorts, CD8⁺ T cells were depleted by i.p. injections of 100 µg anti-CD8 depleting Ab on days −2 and −1 prior to in vivo cytotoxicity assay analysis (clone 53.6.72; Bioexpress Cell Culture Services) (see experimental design, Fig. 1). Depletion was confirmed through flow cytometric analysis of recipient splenocytes, PBMCs, and LMNCs (Supplemental Fig. 1A, 1B). Abs to CD4 and CD8 also depleted CD4⁺ iNKT and CD8⁺ iNKT cell subsets, respectively (Supplemental Fig. 1C).

Splenocytes and LMNCs were isolated from transplant-recipient mice at the time of peak in vivo allocytotoxicity. Previous kinetic studies identified peak cytotoxicity on day 5 for CD4-depleted recipients and on day 7 for CD4-replete recipients (18). CD4-replete hosts yielded ∼50 × 10⁶ splenocytes and ∼3 × 10⁶ LMNCs. CD4-depleted hosts yielded ∼30 × 10⁶ splenocytes and ∼2 × 10⁶ LMNCs. Isolation of CD8⁺ T cells was performed by negative selection magnetic beads as per the manufacturer’s recommendations (purity routinely >95%; STEMCELL Technologies, Vancouver, Canada). CD8⁺ T cells make up ∼10–15% of splenocytes and 30–40% of LMNCs from transplant recipients.

iNKT cells were isolated from LMNCs obtained from syngeneic WT mice. LMNC isolation was performed as previously described (40). LMNCs were purified by a 33.75% Percoll gradient. iNKT cell staining and sorting was performed, as previously described (41). Isolated LMNCs were Fc receptor blocked (2.4G2 hybridoma supernatant), washed, and stained with PBS-57–loaded, APC-conjugated CD1d tetramers (1:2000; National Institutes of Health Tetramer Facility, Emory University Vaccine Center, Atlanta, GA). Additional samples of LMNCs were unstained or stained with unloaded APC-conjugated CD1d tetramers (1:2000) for flow cytometric gating purposes. Cells were sorted at The Ohio State University Comprehensive Cancer Center’s Flow Cytometry Core Laboratory using FACSAria III (Becton Dickinson, Franklin Lake, NJ). PBS-57–loaded, CD1d-positive cells represented the iNKT cell population. In general, this method yielded ∼3 million LMNCs per liver. LMNCs ranged from 10–30% tetramer-positive iNKT cells (Supplemental Fig. 1C). iNKT cells (>98% pure) were pooled from multiple mice and adoptively transferred into hepatocyte recipient mice (AT) on the day of transplant as bulk iNKTs or flow sorted for iNKT subsets based on expression of CD4 (clone GK1.5) or CD8 (clone 53-6.7), yielding CD4⁺CD8⁻ (CD4⁺), CD4⁻CD8⁺ (CD8⁺), and CD4⁻CD8⁻ (double-negative [DN]) iNKT cell subsets. The quantity of bulk iNKT cells transferred was 1 × 10⁶, whereas the quantity of iNKT cell subsets transferred was proportional to their composition within the bulk population (60% DN CD4⁻CD8⁻iNKT cells [6 × 10⁵], 30% CD4⁺CD8⁻iNKT cells [3 × 10⁵], and 10% CD4⁻CD8⁺ iNKT cells [1 × 10⁵]). The presence and persistence of iNKT cell subsets after AT was confirmed following transplant rejection by retrieving LMNCs on day 21 following transplant and flow cytometric analysis (data now shown).

Detection of cytolytic T cell function in vivo through clearance of CFSE (Molecular Probes, Eugene, OR)–stained allogeneic and syngeneic target cells was modified from published methods (42, 43) and has been previously described (18). Syngeneic target splenocytes were isolated from WT mice and were stained with 0.2 μM CFSE (CFSE^low). Allogeneic target splenocytes were isolated from FVB/N mice and were stained with 2.0 μM CFSE (CFSE^hi). Allograft-recipient mice and control naive mice received 20 × 10⁶ CFSE^low syngeneic target splenocytes and 20 × 10⁶ CFSE^high allogeneic target splenocytes by tail vein injection on day 5 (CD4-depleted hosts) or day 7 (CD4-replete hosts). Splenocytes from recipients of CFSE-labeled allogeneic and syngeneic splenocyte targets were retrieved 18 h after CFSE-labeled target cell injection and were analyzed by flow cytometry, gating on CFSE-positive splenocytes. Percentage of allospecific cytotoxicity was calculated as the percentage of yield of allogeneic target cells relative to syngeneic target cells in experimental and control naive mice, as previously described (18).

Alloprimed CD8⁺ T cells (effector cells) were purified from transplant-recipient mice at the time of peak in vivo allocytotoxicity: posttransplant day 5 or day 7 for CD4-depleted hosts and CD4-replete hosts, respectively (18). Splenocytes from naive FVB/N mice were used as allogeneic target cells. WT target splenocytes were used as syngeneic controls. Target cells were stained with CFSE. CD8⁺ T cells and target splenocytes were cocultured at a 10:1 E:T ratio for 4 h along with propidium iodide. Percentage of cytotoxicity (propidium iodide⁺ targets) was analyzed by flow cytometry (LSRII; Becton Dickinson).

CD8⁺ T cell intracellular and extracellular staining were performed based on vendor recommendations (BD Biosciences). Splenocytes were isolated from transplant recipients (day 5 for CD4-depleted hosts and day 7 for CD4-replete hosts) and incubated with Leukocyte Activation Cocktail (Becton Dickinson). Splenocytes were subsequently stained with Abs specific to CD44 (clone IM7), CD3 (clone 145-2C11), CD8 (clone 53-6.7), CXCR3 (clone 1C6), CCR4 (clone 1G1), IFN-γ (clone B27), granzyme B (clone NGZB), perforin (clone eBioOMAKD), FasL (clone MFL3), and TNF-α (clone MP6-XT22). Flow cytometry studies were performed by gating for lymphocytes, single cells, and CD8⁺ T cells. In some cohorts, CD8⁺ T cells were further gated on CD44, IFN-γ, CXCR3, and/or CCR4.

Continuous outcomes measured at one time point were compared between relevant groups using general linear models. Outcomes measured at multiple time points on the same mouse were compared between relevant groups using linear mixed-effects models with random intercepts accounting for the correlation of the repeated measurements within each mouse. Paired t tests were used to assess differences in continuous outcomes across liver- or spleen-derived cells within experimental groups of mice. Hepatocyte allograft rejection between experimental groups was compared using Kaplan–Meier survival curves and log-rank statistics. All analyses were conducted using SAS statistical Software Version 9.4 (SAS Institute, Cary, NC). To demonstrate the distribution of the data, results are listed as the mean ± SE. Hypothesis testing was conducted at a 5% type I error rate (α = 0.05), and p < 0.05 was considered statistically significant. Summary statistics are listed as the mean ± SE.

We have previously reported that following allogeneic hepatocyte transplant, CD8⁺ T cells develop allocytotoxicity both in the presence or absence of CD4⁺ T cells (13, 16–18). To determine if iNKT cells impact the development of CD8⁺ T cell–mediated allocytotoxicity in CD4-replete transplant recipients, WT and iNKT-deficient Jα18 KO mice (C57BL/6 background; both H-2^b) underwent allogeneic FVB/N (H-2^q) hepatocyte transplant. A cohort of Jα18 KO recipients received AT of WT iNKT cells on the day of transplant. Hepatocyte transplant–recipient mice were tested for in vivo allocytotoxicity on day 7 posttransplant (see experimental design, Fig. 1). We found that iNKT cell–deficient, Jα18 KO recipients exhibited a 3-fold decrease of in vivo allocytotoxicity compared with WT recipients (p < 0.0001; (Fig. 2A). When iNKT cells were transferred into Jα18 KO recipients, we observed significantly increased in vivo allocytotoxicity compared with Jα18 KO recipients without cell transfer (p = 0.004), partially restoring allocytotoxicity to WT levels. To determine if the observed in vivo allocytotoxicity is CD8 dependent, a cohort of WT recipients were treated with anti-CD8 mAb 1–2 d prior to the day 7 in vivo allocytotoxicity assay; depletion of recipient CD8⁺ T cells significantly reduced allogeneic allocytotoxicity (p < 0.0001). This confirmed that in vivo allocytotoxicity is predominantly CD8 mediated and that low residual allocytotoxicity in WT hosts is likely Ab mediated, as previously reported (16). In contrast, Jα18 KO recipients do not produce significant alloantibody after transplant (44). These findings support the hypothesis that CD8⁺ T cell–dependent allocytotoxicity in CD4-replete recipients is enhanced by iNKT cells.

Next, we investigated if iNKT cells also enhance in vitro CD8-mediated allocytotoxicity. Alloprimed CD8⁺ T cells were isolated and purified from transplant-recipient splenocytes on day 7 and cocultured with allogeneic FVB/N splenocytes in an in vitro allocytotoxicity assay. CD8⁺ T cells from iNKT cell–deficient Jα18 KO recipients exhibited a 2-fold decrease of in vitro allocytotoxicity compared with CD8⁺ T cells from WT recipients (p < 0.0001; (Fig. 2B). CD8⁺ T cells from Jα18 KO recipients that received AT of iNKT cells mediated a nearly 2-fold increase of in vitro allocytotoxicity compared with CD8⁺ T cells from Jα18 KO recipients (p = 0.004), partially restoring allocytotoxicity to the level mediated by CD8⁺ T cells from WT recipients. Thus, we observed concordance between iNKT cell enhancement of both in vitro and in vivo allocytotoxicity in CD4-replete transplant recipients. Next, we evaluated the influence of iNKT cells on CD8-mediated allocytotoxicity in the absence of CD4⁺ T cells.

We have previously reported that CD8⁺ cytolytic T cells develop in the absence of CD4⁺ T cells in multiple models of transplant-recipient CD4⁺ T cell deficiency, including recipient mice that are CD4 KO, MHC class II KO, and WT mice treated with CD4-depleting mAb (14, 16, 18, 45). To determine the influence of iNKT cells on the development of CD4-independent CD8⁺ T cell–mediated in vivo allocytotoxicity responses, WT and Jα18 KO mice were CD4 depleted prior to allogeneic hepatocyte transplant (days −4 and −2). A cohort of Jα18 KO recipients received AT of WT iNKT cells on the day of transplant. Recipient mice were assessed for in vivo allocytotoxicity [determined in previous studies to peak on day 5 posttransplant (18)]. We found that depletion of CD4⁺ T cells led to a significant decrease in in vivo allocytotoxicity across all conditions (p < 0.0001). In addition, we found that in the absence of iNKT cells, CD4-depleted Jα18 KO recipients exhibited complete abrogation of in vivo allocytotoxicity compared with CD4-depleted WT recipients (p < 0.0001; (Fig. 3A). The allocytotoxicity of CD4-depleted Jα18 KO recipients was comparable with CD8-depleted, CD4-depleted WT recipients (negative control). When iNKT cells were transferred into CD4-depleted Jα18 KO recipients, we observed significant enhancement of in vivo allocytotoxicity (p = 0.02), partially restoring allocytotoxicity to WT levels.

We also tested the in vitro cytotoxicity of alloprimed CD4-independent CD8⁺ T cells in the presence or absence of iNKT cells. Splenic CD8⁺ T cells were retrieved from recipient mice on day 5 posttransplant and cocultured in an in vitro allocytotoxicity assay. CD8⁺ T cells from CD4-depleted, iNKT cell–deficient (Jα18 KO) recipients exhibited an abrogation of in vitro allocytotoxicity compared with CD8⁺ T cells from CD4-depleted WT hosts (p = 0.005; (Fig. 3B). CD8⁺ T cells from CD4-depleted Jα18 KO recipients that received AT of WT iNKT cells exhibited a significant increase in in vitro allocytotoxicity compared with CD8⁺ T cells from CD4-depleted Jα18 KO recipients (p = 0.03). The CD8-mediated allocytotoxicity pattern observed between in vivo and in vitro assays were similar, indicating concurrence of in vitro and in vivo alloreactive CD8⁺ T cell effector function under these experimental conditions.

Notably, both the in vitro and in vivo allocytotoxicity were significantly reduced in CD4-depleted hosts compared with their CD4-replete counterparts (p < 0.0001 for all comparisons). Altogether, these findings suggest that the most robust alloreactive CD8⁺ cytolytic T cells develop when both iNKT and CD4 help are provided. We next assessed the impact of iNKT cells upon hepatocyte allograft rejection.

We have previously reported that CD8⁺ T cells efficiently reject hepatocyte allografts in both CD4-sufficient and CD4-deficient hosts (13, 18). To determine hepatocyte rejection kinetics in the presence or absence of recipient iNKT cells, we transplanted WT and Jα18 KO mice with FVB/N hepatocytes and monitored hepatocyte allograft survival. Cohorts of Jα18 KO recipients received AT of WT iNKT cells. We observed similar rejection kinetics in CD4-replete WT hosts, iNKT cell–deficient Jα18 KO hosts, and Jα18 KO hosts that received AT of WT iNKT cells (p = NS; (Fig. 4).

Next, cohorts of both WT and Jα18 KO recipients were depleted of CD4⁺ T cells. In contrast to what was observed in CD4-replete recipients, rejection of hepatocyte allografts was significantly impaired in CD4-depleted, iNKT cell–deficient Jα18 KO recipients (p < 0.0001) compared with CD4-depleted WT recipients and CD4-depleted Jα18 KO hosts that received AT of WT iNKT cells. It is not unexpected that hepatocyte rejection in CD4-replete mice did not differ in the presence or absence of iNKT cells given the reduced, but still vigorous, in vivo cytotoxic responses observed in iNKT-deficient, CD4-replete hepatocyte transplant recipients (Fig. 2A). However, in the absence of both iNKT and CD4⁺ T cells, hepatocyte survival was significantly prolonged for the extent of CD4-depletion, correlating with the abrogated CD8-mediated in vivo and in vitro cytotoxicity. After cessation of anti-CD4 mAb treatment and reconstitution of CD4⁺ T cells, all Jα18 KO recipients eventually do reject hepatocyte allografts. In prior studies, we have reported that deficiency of the CD4⁺ T cell compartment alone (CD4 KO, MHC class II KO, and CD4-depleted WT recipients) does not significantly affect the kinetics of CD8-mediated hepatocyte allograft rejection (18). Altogether, our data are consistent with the interpretation that depletion of CD4⁺ T cells (and the subpopulation of CD4⁺ iNKT cells) does not delay hepatocyte allograft rejection. However, when both the CD4⁺ T cell and iNKT cell compartments are absent, CD8-dependent hepatocyte allograft rejection is severely impaired.

Because we hypothesized that iNKT cells impact the local intrahepatic microenvironment where transplanted hepatocytes engraft, we analyzed the composition, activation, phenotype, and cytotoxicity of CD8⁺ T cells from LMNCs and splenocytes retrieved from allogeneic hepatocyte transplant recipients. WT and Jα18 KO mice underwent allogeneic FVB/N hepatocyte transplant. A cohort of Jα18 KO recipients received AT of WT iNKT cells. LMNCs and splenocytes retrieved on day 7 posttransplant were analyzed for CD8⁺ T cells and phenotyped for expression of activation markers IFN-γ and CD44. Liver-derived CD8⁺ T cells from CD4-replete hosts had a significantly higher proportion of activated CD44⁺IFN-γ⁺CD8⁺ T cells compared with splenic-derived CD8⁺ T cells from CD4-replete hosts (p < 0.0001 for all; (Fig. 5A, 5B) as a percentage of total CD8⁺ T cells.

In the absence of iNKT cells (Jα18 KO recipients), both liver- and splenic-derived activated CD44⁺IFN-γ⁺CD8⁺ T cells were significantly reduced compared with WT recipients (p < 0.0002 for both). However, AT of iNKT cells into Jα18 KO recipients significantly increased both liver- and splenic-derived activated CD8⁺ T cells (p < 0.02 for both), restoring activated CD8⁺ T cells to similar composition observed in the WT recipients’ livers, but not the spleens (Fig. 5B). Thus, the presence of host iNKT cells significantly impacted the proportion of activated CD8⁺ T cells in both the liver and the spleen in CD4-replete recipients.

The same analysis was performed in CD4-depleted recipients. Both liver- and splenic-derived CD8⁺ T cells retrieved on posttransplant day 5 from CD4-depleted WT hosts were comprised of a significantly higher percentage of activated CD44⁺IFN-γ⁺CD8⁺ T cells compared with respective populations in CD4-depleted iNKT cell–deficient Jα18 KO recipients (p < 0.0001 for both; (Fig. 5A, 5C). AT of iNKT cells into CD4-depleted Jα18 KO recipients increased both liver- and splenic-derived activated CD44⁺IFN-γ⁺CD8⁺ T cells (p < 0.0001 for both; (Fig. 5C). These data demonstrate that in the absence of CD4⁺ T cells, iNKT cells critically influence the activation of CD8⁺ T cells. Taken together, these data indicate that iNKT cells significantly enhance the activation of both liver- and splenic-derived CD8⁺ T cells in CD4-replete and CD4-depleted recipients. These results also reveal that iNKT cells’ contribution to activation of CD8⁺ T cells in the liver and the spleen could not be replaced by host CD4⁺ T cells.

Given the significantly greater proportion of activated CD44⁺IFN-γ⁺CD8⁺ T cells noted in the liver-derived lymphocytes compared with splenocytes from transplant recipients, we next compared the in vitro cytotoxic effector function of liver- versus splenic-derived bulk CD8⁺ T cells. WT mice were transplanted with FVB/N (H-2^q) hepatocytes. On day 7 posttransplant, bulk CD8⁺ T cells were isolated from the liver and spleen and cocultured in an in vitro cytotoxicity assay against allogeneic or third-party (A/J; H-2^k) splenocyte targets. Cocultures of alloprimed CD8⁺ T cells isolated from LMNCs of FVB/N hepatocyte recipients mediated significantly higher cytotoxicity (3-fold) against allogeneic targets compared with alloprimed CD8⁺ T cells isolated from splenocytes (p < 0.0001). Neither alloprimed liver- nor splenic-derived CD8⁺ T cells mediated cytotoxicity against third-party splenocyte targets (Fig. 5D). Collectively, these data are consistent with the interpretation that iNKT cells increase the development of alloreactive CD44⁺IFN-γ⁺CD8⁺ T cells in both the liver and the spleen. However, liver-derived alloreactive CD8⁺ T cells display more potent cytotoxic effector function compared with alloreactive CD8⁺ T cells from the spleen.

Given the more pronounced effect of iNKT cells on liver-derived compared with splenic alloreactive CD8⁺ T cell cytotoxic effector function, we considered the possibility that iNKT cells activate a distinct and unique subset of CD8⁺ T cells in the liver. Therefore, we analyzed activated (CD44⁺IFN-γ⁺) CD8⁺ T cells from the liver or spleen of CD4-replete and CD4-depleted hepatocyte transplant recipients for the expression of chemokine receptors known to be important for T cell trafficking to the liver, including CXCR3 and CCR4 (46, 47).

Liver- and splenic-derived mononuclear cells were retrieved on day 5 (CD4-depleted recipients) or 7 (CD4-replete recipients) after hepatocyte transplant and analyzed by flow cytometry for CD8⁺ T cell expression of activation markers (CD44 and IFN-γ) and chemokine receptors (CXCR3 and CCR4; (Fig. 6A). We found that in CD4-replete recipients, there was a significantly higher proportion of liver-derived activated CXCR3⁺CCR4⁺ CD8⁺ T cells in iNKT-sufficient compared with iNKT-deficient recipients (p < 0.0011; (Fig. 6B). A similar pattern was observed for splenic-derived CD8⁺ T cells because the proportion of activated CXCR3⁺CCR4⁺CD8⁺ T cells in the spleen was significantly higher in iNKT-sufficient compared with iNKT-deficient recipients (p < 0.03).

In CD4-depleted recipients, we also observed a higher proportion of liver-derived activated CXCR3⁺CCR4⁺CD8⁺ T cells in iNKT-sufficient recipients compared with iNKT-deficient recipients (p < 0.0001; (Fig. 6C). A similar pattern was observed for splenic-derived CD8⁺ T cells from CD4-depleted recipients because the proportion of activated CXCR3⁺CCR4⁺CD8⁺ T cells in the spleen was significantly higher in CD4-depleted, iNKT-sufficient recipients compared with CD4-depleted, iNKT-deficient recipients (p < 0.035).

Notably, the proportion of activated CXCR3⁺CCR4⁺CD8⁺ T cells in both CD4-replete and CD4-depleted recipients is significantly higher in the liver (65.0 ± 14.7%) than in the spleen (45.1 ± 20.1%) in all groups of mice (p = 0.0002 for all recipient groups combined and compared between liver versus spleen). Furthermore, iNKT cells significantly enhanced the proportion of this cell subset in the liver in both CD4-replete and CD4-depleted recipients. In contrast, the presence or absence of CD4⁺ T cells alone did not significantly influence the proportion of activated CXCR3⁺CCR4⁺CD8⁺ T cells in the liver or spleen in WT recipients (p = NS for both; (Fig. 6B, 6C).

Given the high proportion of dual chemokine receptor–positive, activated CXCR3⁺CCR4⁺CD8⁺ T cells in the liver and the high magnitude of allocytotoxicity mediated by liver-derived CD8⁺ T cells, we next investigated their effector molecule expression profile. WT mice were transplanted with allogeneic FVB/N hepatocytes, and on day 7 posttransplant, liver- and splenic-derived CD8⁺ T cells were isolated. CD8⁺ T cell subsets were analyzed by flow cytometry for expression of granzyme B, perforin, Lamp-1, FasL, IFN-γ, and TNF-α. CXCR3⁺CCR4⁺CD8⁺ T cells were compared with CXCR3⁺CCR4⁻CD8⁺ T cells. Granzyme B, perforin, Lamp-1, and FasL were highly expressed by both CD8⁺ T cell subsets without significant differences between them (Fig. 7A–C). In contrast, both liver- and splenic-derived CXCR3⁺CCR4⁺CD8⁺ T cells had significantly higher proportion of cells expressing TNF-α compared with CXCR3⁺CCR4⁻CD8⁺ T cells (p < 0.0001 for all comparisons; (Fig. 7B, 7C). Similarly, both liver- and splenic-derived CXCR3⁺CCR4⁺CD8⁺ T cells had significantly higher proportion of cells expressing IFN-γ compared with CXCR3⁺CCR4⁻CD8⁺ T cells (p < 0.0001 for all comparisons, both liver and spleen subsets; (Fig. 7B, 7C). The expression of TNF-α by mean fluorescence intensity was higher for CXCR3⁺CCR4⁺CD8⁺ T cells compared with CXCR3⁺CCR4⁻CD8⁺ T cells (p = 0.026; (Fig. 7D). However, no difference in IFN-γ expression was detected between the two cell subsets (p = NS). Altogether, CXCR3⁺CCR4⁺CD8⁺ T cells are distinguished by a multipotent cytotoxic effector phenotype.

Given the significant influence of iNKT cells upon in vivo development of activated CXCR3⁺CCR4⁺CD8⁺ cytolytic T cells after hepatocyte transplant, we next analyzed the cytotoxic potency of this cell subset. We used splenic-derived CXCR3⁺CCR4⁺CD8⁺ T cells for these in vitro cytotoxicity assays given the common effector molecule profile for both liver- and splenic-derived subsets. The in vitro cytotoxicity of alloprimed splenic-derived CXCR3⁺CCR4⁺CD8⁺ T cells was compared with CXCR3⁻CD8⁺ T cells, CXCR3⁺CCR4⁻CD8⁺ T cells, and naive CD8⁺ T cells (negative control). CXCR3⁺CCR4⁺CD8⁺ T cells mediate 2-fold higher allocytotoxicity compared with CXCR3⁺CCR4⁻CD8⁺ T cells (p < 0.0001). CXCR3⁻CD8⁺ T cells display negligible allocytotoxicity comparable to the negative control group with naive CD8⁺ T cells (p = NS; (Fig. 8A).

Next, we investigated the in vivo cytotoxic effector function of CXCR3⁺CCR4⁺CD8⁺ T cells. Day 7 alloprimed CXCR3⁺CCR4⁺CD8⁺ T cells or CXCR3⁺CCR4⁻CD8⁺ T cells were transferred into naive WT mice. In vivo allocytotoxicity in mice that received AT with CXCR3⁺CCR4⁺CD8⁺ T cells was significantly higher (2.5-fold) compared with mice that received CXCR3⁺CCR4⁻CD8⁺ T cells (p < 0.0001; (Fig. 8B).

Next, the capacity of CXCR3⁺CCR4⁺CD8⁺ cytolytic T cells to reject hepatocyte allografts was investigated. To do this, RAG1 KO mice (H-2^b; deficient in T cells, B cells, and iNKT cells) were transplanted with FVB/N hepatocytes. After hepatocyte allograft function was established for 2 wk, RAG1 KO recipients received AT of CXCR3⁺CCR4⁺CD8⁺ or CXCR3⁺CCR4⁻CD8⁺ T cell subsets. RAG1 KO mice with AT of CXCR3⁺CCR4⁺CD8⁺ T cells had rapid hepatocyte allograft rejection compared with mice that received AT of CXCR3⁺CCR4⁻CD8⁺ T cells (median survival time [MST] of 5 d versus 16 d, p < 0.0001; (Fig. 8C). These data demonstrate that CXCR3⁺CCR4⁺CD8⁺ T cells have high cytotoxic potency that correlates with more rapid hepatocyte allograft rejection.

Next, we pursued studies to identify specific iNKT cell subsets with capacity to promote development of CXCR3⁺CCR4⁺CD8⁺ T cells. In prior studies, we noted that CD4-depleted CD8 KO recipients do not acutely reject hepatocyte allografts (and lack alloantibody responses) (13–15). These mice are devoid of CD4⁺ T cells, CD8⁺ T cells, CD4⁺ iNKT cells, and CD8⁺ iNKT cells but retain DN iNKT cells (Supplemental Fig. 2A, 2B). Because the AT of naive CD8⁺ T cells into CD4-depleted CD8 KO recipients initiates acute rejection (MST = 13 d, p = 0.0002; Supplemental Fig. 2C), this suggests that DN iNKT cells are sufficient to provide help in the absence of CD4⁺ T cells to stimulate the development of alloprimed CD8⁺ cytolytic T cells and subsequent allogeneic hepatocyte rejection. To directly investigate the capacity of iNKT subsets to promote the development of alloprimed CD8⁺ cytolytic T cells, we performed AT studies in CD4-depleted Jα18 KO hepatocyte transplant recipients. Recipients received AT of sorted iNKT cell subsets in proportion to their composition in bulk iNKT cell populations (6 × 10⁵ DN iNKT cells or 1 × 10⁵ CD8⁺ iNKT cells). These recipients were then analyzed for CD8-mediated in vitro allocytotoxicity, CD8⁺ T cell immunophenotype, and hepatocyte allograft rejection. CD8⁺ T cells retrieved from CD4-depleted Jα18 KO recipients that received AT with either CD8⁺ iNKT cells or DN iNKT cells displayed robust CD8-mediated in vitro allocytotoxicity that was significantly greater in comparison with CD8⁺ T cells from CD4-depleted Jα18 KO mice without iNKT cell transfer (p = 0.0005 and p = 0.0001, respectively; (Fig. 9A) and similar to allocytotoxicity observed in recipients that received transfer of unsorted bulk iNKT cells (p = NS for both groups). AT of CD8⁺ iNKT or DN iNKT subsets stimulated the development of a greater proportion of activated CD44⁺IFN-γ⁺CD8⁺ T cells compared with CD4-depleted Jα18 KO recipients without iNKT cell transfer (p < 0.0001 for both comparisons; (Fig. 9B). AT of CD8⁺ iNKT cells promoted an increase in the proportion of alloprimed CXCR3⁺CCR4⁺CD8⁺ T cells compared with CD4-depleted Jα18 KO recipients without iNKT cell transfer (p = 0.03; (Fig. 9C). AT of DN iNKT cells did not increase the proportion of alloprimed CXCR3⁺CCR4⁺CD8⁺ T cells compared with CD4-depleted Jα18 KO recipients without iNKT cell transfer (p = NS). However, there was not a significant difference in the proportion of alloprimed CXCR3⁺CCR4⁺CD8⁺ T cells between groups that received AT of DN iNKT cells versus CD8⁺ iNKT cells (p = NS). AT of either CD8⁺ iNKT cell or DN iNKT cell subsets enhanced the tempo of CD8⁺ T cell–mediated hepatocyte allograft rejection in CD4-depleted Jα18 KO recipients (p < 0.0001 for both; (Fig. 9D). Prior studies in CD4-depleted CD8 KO recipients suggest that DN iNKT cells do not directly mediate hepatocyte rejection (13–15). To investigate whether the accelerated allograft rejection observed in CD4-depleted Jα18 KO recipients that received AT with CD8⁺iNKT cells could be attributed to CD8⁺ iNKT cells directly mediating hepatocyte allograft damage, we transplanted Rag1 KO mice with allogeneic FVB/N hepatocytes. Following 14 d of established hepatocyte allograft survival, recipient mice received AT of 1 × 10⁵ CD8⁺ iNKT cells. Rag1 KO mice that received AT of CD8⁺ iNKT cells did not reject hepatocellular allografts (p = NS; (Fig. 9E).

To directly investigate the capacity of CD4⁺ iNKT cells to stimulate the development of alloprimed CD8⁺ cytolytic T cells, Jα18 KO hepatocyte transplant–recipient mice received AT of CD4⁺ iNKT cells and were analyzed for in vitro allocytotoxicity and composition of CD8⁺ T cell subsets. CD8⁺ T cell–mediated in vitro allocytotoxicity was not significantly different in the group of Jα18 KO recipients that received AT of CD4⁺ iNKT cells (3 × 10⁵) compared with CD8⁺ T cells from Jα18 KO recipients without iNKT cell transfer (p = NS; Supplemental Fig. 3A). Similarly, CD4⁺ iNKT cells did not enhance the development of activated CD44⁺IFNγ⁺ CD8⁺ T cells or alloprimed CXCR3⁺CCR4⁺CD8⁺ T cells when compared with Jα18 KO recipients without iNKT cell transfer (p = NS; Supplemental Fig. 3B, 3C). Collectively, these data demonstrate that CD8⁺ iNKT cells and DN iNKT cells, but not CD4⁺ iNKT cells, enhance the development and cytotoxic function of alloprimed CD44⁺IFNγ⁺ CD8⁺ T cells. Furthermore, CD8⁺ iNKT cells significantly increase the proportion of multipotent CXCR3⁺CCR4⁺CD8⁺ cytotoxic T cells.

We have previously reported that hepatocyte allografts engrafted in the liver are susceptible to rejection by alloreactive CD8⁺ cytolytic T cells that develop in both CD4-sufficient and CD4-deficient conditions (13, 16–18). The current study aimed to investigate the role of iNKT cells, known to be abundant in the liver, upon the development of alloreactive CD8⁺ cytolytic T cells. It is important to note that no exogenous foreign glycolipids (e.g., α-GalCer) were used in these studies to activate iNKT cells. Thus, our results uniquely analyze endogenous activation of iNKT cells after allogeneic hepatocyte transplant in contrast to published studies that use exogenous α-GalCer to activate iNKT cells. We found that iNKT cells significantly enhance the magnitude of both in vivo and in vitro CD8-mediated allocytotoxicity. However, iNKT cells are critically required for allocytotoxicity by CD4-independent CD8⁺ T cells because when neither iNKT nor CD4⁺ T cells are present, the development of alloreactive CD8⁺ cytolytic T cells is severely impaired. This impairment of CD8-mediated cytotoxic effector function is accompanied by impaired rejector function as manifested by significant prolongation of hepatocyte allograft survival in CD4-depleted, iNKT-deficient Jα18 KO recipients. iNKT-mediated enhancement of CD8⁺ cytolytic T cells was not limited to alloimmune responses because similar results were observed after syngeneic membrane-bound OVA transgenic hepatocyte transplantation. We found that WT CD8⁺ T cell–mediated OVA-peptide specific cytotoxicity was similarly enhanced by iNKT cells in CD4-replete recipients and critically dependent on iNKT cells when CD4⁺ T cells were absent (data not shown). Collectively, our data suggest that there are two independent mechanisms (CD4⁺ T cell mediated or iNKT cell mediated) that can provide help to CD8⁺ T cells, and each are sufficient to promote the development of alloprimed CD8⁺ cytolytic T cells capable of mediating hepatocyte allograft rejection. Whether iNKT cells provide help to CD8⁺ T cells directly through their cognate interaction with APCs and/or noncognate interaction with CD4⁺ T cells is not clear and requires further study. Nevertheless, interference with either the iNKT or CD4⁺ T cell–dependent pathway reduces the magnitude of CD8-mediated cytotoxicity but does not significantly impact the kinetics of hepatocyte allograft rejection. However, when both mechanisms are impaired, the development of alloprimed CD8⁺ T cell cytotoxic effector function and hepatocyte allograft rejection are abrogated.

Interestingly, when we analyzed activated CD8⁺ T cell subsets by expression of chemokine receptors CXCR3 and CCR4, we discovered that iNKT cells significantly enhance the proportion of a subset of activated CD8⁺ T cells expressing CXCR3 and CCR4 that are detected in both the liver and spleen. We found that this dual chemokine receptor–positive CXCR3⁺CCR4⁺CD8⁺ cytolytic T cell subset is multipotent and highly expresses granzyme B, perforin, Lamp-1, and FasL and is differentiated from CXCR3⁺CCR4⁻CD8⁺ T cells by a higher proportion expressing TNF-α and IFN-γ effector molecules. CXCR3⁺CCR4⁺CD8⁺ T cells also demonstrate higher expression of TNF-α (but not IFN-γ) when compared with CXCR3⁺CCR4⁻CD8⁺ T cells. This novel subset stimulated by endogenous activation of iNKT cells is distinguished by its potent in vitro and in vivo cytotoxic effector function that, to our knowledge, has not been previously reported. This high level of allospecific cytotoxicity observed for the CXCR3⁺CCR4⁺CD8⁺ cytolytic T cell subset is correlated with vigorous and more rapid hepatocyte rejection. Furthermore, this subset comprises a larger proportion of activated CD8⁺ T cells in the liver compared with in lymphoid tissue, such as the spleen, and likely poses a significant barrier to long-term survival of allogeneic hepatocytes transplanted to the liver.

Our studies indicate that iNKT cells have a more significant impact than conventional helper CD4⁺ T cells on the development of this novel activated CXCR3⁺CCR4⁺CD8⁺ T cell subset. The proportion of activated CXCR3⁺CCR4⁺CD8⁺ T cells that develops after transplant is severely reduced by the absence of host iNKT cells, whereas depletion of CD4⁺ T cells has minimal impact. It is notable that both iNKT cells and activated CXCR3⁺CCR4⁺CD8⁺ T cells are more prominent in the liver than other immune locales such as the spleen. This may explain our prior findings that alloreactive CD8⁺ cytolytic T cells develop in response to hepatocyte transplant even in recipients that are deficient of both lymphoid tissue and CD4⁺ T cells (16). Collectively, these findings suggest that iNKT cell help for development of activated CXCR3⁺CCR4⁺CD8⁺ T cells constitute an important mechanism of CD8-mediated immunity in the liver microenvironment. Further investigation revealed that CD8⁺ iNKT and DN iNKT cells, but not CD4⁺ iNKTs, promote the development and function of alloprimed CXCR3⁺CCR4⁺CD8⁺ T cells.

The dual positive CXCR3⁺CCR4⁺CD8⁺ T cells in these studies are unlike dermal CCR4⁺CD8⁺ T cells reported by Kondo et al. (48) that are noncytotoxic cells that produce high amounts of IFN-γ, IL-2, IL-4, and TNF-α cytokines but do not express perforin or granzyme B effector molecules (48, 49). Other studies investigating cutaneous inflammation also report enhanced quantities of skin-infiltrating CCR4⁺CD8⁺ T cells (50–53) or peripheral blood CCR4⁺CD8⁺ T cells (50, 51, 53, 54). CCR4⁺CD8⁺ T cells largely display a Tc2 phenotype associated with IL-4 expression (55). Furthermore, Teraki et al. (56) reported a high percentage of infiltrating CCR4⁺CD8⁺ T cells but few CXCR3⁺CD8⁺ T cells, suggesting that CXCR3⁻CCR4⁺CD8⁺ T cells comprise the majority of skin-infiltrating CD8⁺ T cells in psoriatic lesions. This is in stark contrast to the liver-derived CXCR3⁺CCR4⁺CD8⁺ T cells in this study that are highly cytotoxic and multipotent (perforin⁺granzyme B⁺FasL⁺TNF-α⁺IFN-γ⁺). Whether these differences arise from tissue-specific features of the immune microenvironment of the skin versus the liver, the nature of the inflammatory stimulus or systemic factors is unclear. For example, CCR4⁺CD8⁺ T cells associated with autoimmune disease (cutaneous lupus erythematosus) (51) and tumor immunity (nasopharyngeal carcinoma) (57) are reported to have a cytotoxic phenotype (granzyme B⁺). The association between CCR4 expression and high CD8 cytotoxic effector function is also noted by the work of Semmling et al. (58) that reported WT, OVA-specific CD8⁺ T cells (OT-I) were significantly more cytotoxic than CCR4-deficient, OVA-specific CD8⁺ T cells. However, none of the aforementioned published studies investigated CD8⁺ T cells with dual expression of CXCR3 and CCR4 or their enhancement by iNKT cells.

We previously reported that CD8⁺ T cells, developing in CD4-sufficient recipients, use multiple cytotoxic mechanisms (including perforin, FasL, and TNF-α), whereas CD4-independent CD8⁺ T cells do not develop perforin- and FasL-dependent cytotoxicity and, instead, are critically dependent on TNF-α–mediated effector function (17, 18). In the current study, the dual chemokine receptor–positive CXCR3⁺CCR4⁺CD8⁺ T cells are a distinct subset that express multiple cytotoxic effector molecules and mediate very high in vitro and in vivo allocytotoxicity and vigorous hepatocyte allograft rejection. They develop in both CD4-replete and CD4-depleted hepatocyte transplant recipients and are significantly reduced in the absence of iNKT cells.

These data, to our knowledge, provide first evidence that the development of this novel and highly cytotoxic CXCR3⁺CCR4⁺CD8⁺ T cell subset is induced by iNKT cells. Furthermore, this iNKT cell–enhanced CD8⁺ T cell subset comprises a greater proportion of liver-derived compared with splenic-derived CD8⁺ T cells. Although we did not investigate how iNKT cells impact the development of this novel subset, others have reported that iNKT cells recruit and activate CCR4⁺CD8⁺ T cells (46). This recruitment is thought to occur through a cytokine-dependent mechanism in which IL-4⁺ iNKT cells stimulate DCs to release CCL17 (also known as the IL-4/CCL17 axis), a known chemokine ligand for CCR4. iNKT cells may also engage in CD1d-dependent cognate interaction with DCs and license them for cross-priming of extracellular Ags by upregulating DC secretion of CCL17 and enhancing selective recruitment of CCR4-expressing CD8⁺ T cells (58). This iNKT cell and CCR4-CCL17 mechanism of CD8⁺ T cell cross-priming is distinct from CD4⁺ T cell–facilitated DC licensing and cross-priming that recruit CCR5-expressing CD8⁺ T cells (59). Similarly, cytotoxic CXCR3⁺CCR4⁺CD8⁺ T cells may use these chemokine receptors for recruitment and positioning as reported for regulatory CD4⁺ T cells in the inflamed human liver. Oo et al. (60) demonstrated that CD4⁺ regulatory T cells (Tregs) use CXCR3 to traffic to the liver and CCR4 to migrate toward APCs. Unlike what we observed for the multipotent CXCR3⁺CCR4⁺ CD8⁺ T cell subset, the proportion of CXCR3⁺CD8⁺ T cells expressing other chemokine receptors, such as CCR5 or CXCR6 that may be important for T cell recruitment to the liver, was not influenced by the presence or absence of host iNKT cells (Supplemental Fig. 4). Interestingly, in experimental models using exogenous α-GalCer to activate iNKT cells, iNKTs did not enhance the recruitment, proliferation, or quantity of Ag-specific CD8⁺ T cells in the liver but rather enhanced effector function (IFN-γ production) of liver CD8⁺ T cells (61). However, in tumor models, activation of iNKT cells with exogenous α-GalCer often resulted in enhanced proliferation and quantity of tumor-infiltrating, Ag-specific CD8⁺ T cells (34, 62, 63).

The CCL17/CCL22-CCR4 chemokine/receptor axis is implicated in the pathogenesis of various diseases including skin diseases (64, 65), asthma (66), and CNS autoimmune disease (multiple sclerosis) (67–71). CCR4 is known to be expressed on activated T cells, especially CD4⁺ Th2 cell subsets along with CCR8 (72) and on Th17 cells along with CCR6 (73–75). CCR4 is also expressed on Tregs and has been associated with CCL22-driven recruitment of Tregs and suppression of tumor immunity (76). In contrast, enhanced CCL22-directed Treg recruitment ameliorated autoimmune diabetes (77). CCL17 is an inflammatory chemokine with a highly organ- and DC-restricted expression profile, as CCL17 expression is enriched in mature DCs (CD8⁻CD11b⁺DEC205⁺) in secondary lymphoid tissue (not the spleen) (75, 78). Thus, iNKT-enhanced CD8⁺ T cell responses may be promoted through the CCL17/CCL22-CCR4 chemokine/receptor axis via cytokine-mediated and/or cognate interaction with DCs to recruit CCR4-expressing CD8⁺ T cells and CD4⁺ T cells. However, the recruitment of CCR4⁺CD4⁺ T cells is not required for CD8⁺ T cell activation. Activated CD8⁺ cytolytic T cells also upregulate the expression of CD1d and in turn can activate iNKT/CD8⁺ T cell cross-talk and enhance in vitro CD8⁺ cytolytic T cell IFN-γ production, proliferation, and cytotoxicity (79). Because iNKT cell subsets have specialized localization in tissues that impacts their cytokine profile and iNKT cells enriched in the liver are IFN-γ producing (29, 80), it will be of interest in future studies to determine the mechanism by which CD8⁺ iNKT and DN iNKT cells impact the development and function of CXCR3⁺CCR4⁺CD8⁺ cytolytic T cells.

In these studies, treatment of transplant recipients with anti-CD4 mAb to depleted CD4⁺ T cells also depleted populations of CD4⁺ iNKT cells. Indeed, we found that recipient mice treated with anti-CD4 mAb were depleted of both CD4⁺ T cells and CD4⁺ iNKT cells. In CD4-depleted recipients, we noted a significant population of CD8⁺ iNKT cells and DN iNKT cells. Because activated CXCR3⁺CCR4⁺CD8⁺ T cells developed in mice depleted of both CD4⁺ T cells and CD4⁺ iNKT cells, this would imply that CD4⁺ iNKT cells are not essential for iNKT-mediated enhancement of activated CXCR3⁺CCR4⁺CD8⁺ cytolytic T cells investigated in this study. Indeed, when we investigated the efficacy of sorted iNKT cells subsets to promote in vivo development of CXCR3⁺CCR4⁺CD8⁺ cytolytic T cells, we found that CD8⁺ iNKT cells and DN iNKT cells, but not CD4⁺ iNKT cells, increase the proportion of CXCR3⁺CCR4⁺CD8⁺ T cells enhance CD8-mediated cytotoxicity and rapidity of hepatocyte allograft rejection. CD8⁺ iNKT and DN iNKT cells do not mediate direct damage of hepatocyte allografts.

The importance of iNKT-enhanced CD8⁺ T cell responses may extend beyond intrahepatic cell transplantation. For example, despite the underlying immunosuppressed state and associated reduced quantity of CD4⁺ T cells, HIV-positive transplant recipients experience a higher rate of liver and kidney transplant rejection compared with HIV-negative transplant recipients (81–83). One explanation for these unexpected results could be rejection mediated by CD4-independent CD8⁺ T cells that have been reported by our group and others to be resistant to immunosuppressive strategies (including costimulatory blockade) that readily regulate alloreactive CD4⁺ T cell–dependent CD8⁺ T cells (19, 21, 23, 24, 84–86). Our current data suggest that another explanation may be that CD8⁺ iNKTs and DN iNKT-mediated help may promote the expansion of highly cytotoxic CD8⁺ T cells in HIV-positive recipients despite their impaired CD4⁺ T cell function. As such, it will be of interest to test the susceptibility of this novel CXCR3⁺CCR4⁺CD8⁺ T cell subset to conventional and experimental immunosuppressive strategies.

The CCR4-CCL17/22 axis is a target for immunotherapy. Mogamulizimab, an anti-CCR4 mAb, potentially depletes pathogenic CCR4⁺ T cells through an Ab-dependent cellular cytotoxicity mechanism and also stimulates proinflammatory immunity through reduction of CCR4⁺ Tregs. Mogamulizimab has been tested for the treatment of relapsed adult T cell leukemia/lymphoma (87, 88), cutaneous T cell lymphoma (89), and HTLV-1–associated myelopathy (90). Mogamulizimab treatment was associated with improved clinical outcomes, but skin-related proinflammatory adverse effects associated with depletion of CCR4⁺CD4⁺ Tregs have been implicated. Small molecule competitive antagonists of the CCL17/CCL22/CCR4 axis have been developed to block CCL17- and CCL22-mediated recruitment of Th2 cells and Tregs with ameliorating effect in disease models for asthma, atopic disease, and tumor growth (91–95). Although mAb therapy targeting the CCR4 ligands CCL17 and CCL22 has not been tested clinically, murine studies have demonstrated therapeutic benefit in inflammatory osteoarthritis (96), pulmonary invasive aspergillosis (97), and experimental autoimmune encephalomyelitis (98). Given the spectrum of CCR4 expression on both proinflammatory and anti-inflammatory T cells, approaches to achieve direct targeting of specific cell subsets are likely to optimize therapeutic effect and minimize adverse effects.

Our studies provide rationale to further investigate CD8-dependent immunity promoted by CD8⁺ iNKT and DN iNKT cells in the setting of intrahepatic cell transplantation and other conditions such as infection, malignancy, and autoimmunity. However, caution regarding generalization of results derived solely from murine models is warranted in view of differences in the abundance of innate like T cell subsets such as iNKT cells in the liver of mice and humans (99). Although iNKT cells represent a large population of intrahepatic lymphocytes in mice (10–30%), they comprise a smaller proportion in humans (1.0%) (25, 99, 100). Whereas mice have low proportions of mucosal associated invariant T cells in the liver (<1%), humans have a much higher proportion (40%) (101). Nevertheless, the conceptual innovation from our study is that innate immune cell subsets abundant in the liver augment Ag-specific, CD8⁺ T cell–mediated cytotoxic responses.

Altogether, our published and current data demonstrate that despite the tolerogenicity observed after whole liver transplant, the unique cellular composition of the liver gives rise to an immunologically proinflammatory microenvironment in the setting of allogeneic hepatocyte transplantation. The current studies expand upon this theme by identifying that CD8⁺ iNKT and DN iNKT cells enhance the development of a novel and highly cytotoxic subset of CXCR3⁺CCR4⁺CD8⁺ T cells that mediate rapid allogeneic hepatocyte rejection. Future research to optimize hepatocyte and other cell transplant to the liver warrants development of novel immunosuppressive strategies that prevent rejection by iNKT-enhanced alloreactive CD8⁺ cytolytic T cell–mediated immune responses.

The authors have no financial conflicts of interest.