Donor γδT Cells Promote GVL Effect and Mitigate aGVHD in Allogeneic Hematopoietic Stem Cell Transplantation

Disease relapse and graft-versus-host disease (GVHD) are the major complications affecting the outcomes of allogeneic hematopoietic stem cell transplantation (allo-HSCT). While the functions of αβT cells are extensively studied, the role of donor γδT cells in allo-HSCT is less well defined. Using TCRδ-/- donors lacking γδT cells, we demonstrated that donor γδT cells were critical in mediating graft-versus-leukemia (GVL) effect during allo-HSCT. In the absence of donor γδT cells, IFN-γ production by CD8+ T cells was severely impaired. Vγ4 subset was the major γδT cell subset mediating the GVL effect in vivo, which was partially dependent on IL-17A. Meanwhile, donor γδT cells could mitigate acute GVHD in a murine allo-HSCT model by suppressing CD4+ T cell activation and the major γδT cell subset that exerted this protective function was also Vγ4 γδT cells. Therefore, our findings provide evidence that donor γδT cells, especially Vγ4 subset, can enhance GVL effect and mitigate aGVHD during allo-HSCT.


INTRODUCTION
Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is one of the most curative options for treating leukemia and other hematopoietic malignant diseases (1,2). But its efficacy is limited by graft-versus-host disease (GVHD) and disease relapse (3). GVHD is induced by an immune response of donor T cells against recipient healthy tissues (4). T cells are comprised of two major subpopulations, identified by their expression of either ab or gd TCR heterodimers. Donor abT cells are thought to be the primary T cell subpopulation responsible for mediating GVHD and graft versus leukemia (GVL) responses during allo-HSCT (5,6). Nevertheless, a number of recent studies suggest that gdT cells might also play a critical role in mediating the outcomes of allo-HSCT (1,7,8).
gdT cells are present in relatively smaller numbers and percentages in most tissues of mouse and human compared to abT cells (9). Generally, only a small portion of gdT cells express CD4 or CD8 co-receptors. They can be activated by stressinduced ligands without the antigen presentation via major histocompatibility complex (MHC). The ligands of gdTCR include MHC-related and MHC-unrelated molecules. It is not clear which endogenous ligands activate gdT cells in most disease conditions. gdT cells also exhibit similar recognition mechanisms as NK cells. They can express NKG2D and KIRs, and recognize target cells expressing stress-induced ligands (10). Binding of ligands to activating receptors on gdT cells triggers cytotoxicity by releasing cytotoxic granules and induces immune regulatory functions by producing cytokines (11).
Previous studies demonstrated that gdT cells might facilitate allogeneic engraftment and contribute to anti-viral immunity (12). A recent study showed that human gdT cells were quickly reconstituted with radically altered but stable TCR repertoires after HSCT (13). In this study, they also observed a few individual gdT cell clones (mainly but not exclusively within the Vg9 and Vd2 fraction) underwent additional massive proliferation in response to cytomegalovirus (CMV). In another study, the T cell receptor gamma (TRG) repertoire of gdT cells within peripheral blood stem cells was analyzed by using next-generation sequencing technology. The results showed that the grafts from CMV + donors presented a reshaped TRG repertoire, and the TRG composition was not associated with aGVHD development (14). It has been reported that Vd2 -gdT cells were significantly expanded in CMVseropositive transplant recipients and these cells can directly lyse CMV-infected cells (15). Adoptive transfer of human Vg9Vd2 T cells expanded with phosphorylated antigens could effectively prevent the progress of Epstein-Barr virus-induced lymphoproliferative disease in humanized mice (16). These studies explored the gdT cell responses in anti-viral immunity and the potential of using adoptive gdT cell immunotherapy in allogeneic transplantation recipients.
gdT cells can mediate innate anti-tumor activity by direct cytotoxicity and IFN-g production (17). However, gdT cells have also been reported to promote tumor growth by producing IL-17 (18,19). Many studies in clinical trials have demonstrated the anti-leukemia effect of human gdT cells in haematological malignancies after allo-HSCT. An eight years' follow-up study indicated a survival advantage in patients with increased gdT cells after allo-HSCT (20). AML and ALL patients recovered with high gdT cell numbers displayed a better leukemia-free survival (LFS) and overall survival (OS) compared with those with low gdT cell numbers. Interestingly, there was no increase in the incidence of acute GVHD (aGVHD) associated with high gdT cell numbers. Moreover, human gdT cells from blood of patients showed significant cytotoxicity against multiple myeloma or lymphoma cells (21)(22)(23). Treatment of paediatric ALL patients with zoledronate was associated with an increase of Vd2 gdT cells and an increase of the cytotoxicity against primary leukemia blasts (24). Although the anti-tumor function of gdT cells has been suggested by many studies, it is still not clear which gdT subset possesses a strong anti-tumor effect and whether this effect is also mediated through regulation of abT cells besides direct cytotoxicity after allo-HSCT.
There is evidence suggesting that gdT cells are not the primary initiators of GVHD (25). Although an increased number of gdT cells were found in patients who developed aGVHD up to three months after allo-HSCT (26), a subsequent study found no significant correlation between gdT cell recovery and the incidence of GVHD in the first 12 months post HSCT (27). In fact, a recent study showed improved OS, LFS, and less GVHD in patients with high immune reconstitution of gdT cells two months after allo-HSCT (8). In murine studies, donor gdT cells have been shown to exacerbate aGVHD and the elimination of gdT cells from donor mice significantly reduced the lethality of GVHD (28). Similarly, another study showed that co-infusion of in vitro expanded donor-derived gdT cells and naïve abT cells on the same day post allo-HSCT significantly exacerbated GVHD (29). However, donor-derived gdT cell infusion resulted in reduced GVHD and improved survival when the administration of naïve abT cells was delayed for 2 weeks. This protective effect of gdT cells is mediated indirectly via donor BM-derived abT cells. Therefore, donor-derived gdT cells could exert anti-leukemia effect while protecting the host from GVHD. However, this notion has not been fully examined in animal models and the detailed mechanism is not known.
In this study, by performing allo-HSCT using TCRd -/donors and gdT cell infusions, we investigated the role of donor gdT cells in both GVL and aGVHD murine models. Our results suggest that donor Vg4 gdT cells could promote GVL and suppress aGVHD in allo-HSCT through the regulation of abT cell immune responses.
Cell Lines A20 (H2K d ) lymphoma cell line was purchased from American Type Culture Collection (Manassas, VA). Luciferase-expressing A20 cells were generated by a lentiviral system. Briefly, the

Murine GVL and aGVHD Models
GVL model: BALB/c recipient mice received lethal TBI (750cGy: 2 doses of 375cGy with 4 h interval) from a 137 Cs source. Three hours later, 5×10 6 BMCs from WT or TCR-d -/-C57BL/6 mice plus 1×10 6 A20 lymphoma cells or 5×10 6 A20-luc+/yfp cells were intravenously injected into lethally irradiated BALB/c recipients. The survival was monitored and the body weights of recipients were assessed every other day. aGVHD model: BALB/c recipient mice received lethal TBI (750cGy: 2 doses of 375cGy with 4 h interval) from a 137 Cs source. Three hours later, 1×10 7 BMCs plus 5×10 6 splenocytes from WT or TCR-d -/-C57BL/6 mice were intravenously injected into lethally irradiated BALB/c recipient. The survivals were observed, and the body weights and clinical scores were assessed every two or three days. The severity of aGVHD was assessed with a clinical GVHD scoring system as described by Cooke et al. in a blinded fashion. The degree of systemic GVHD was assessed by the sum of changes in five clinical parameters: weight loss, posture (hunching), activity, fur texture, and skin integrity (30).

Cytotoxicity Assay
Cytotoxicity assay was carried out using a cytotoxicity detection kit (LDH) (Roche, Basel, Switzerland). Expanded gdT/Vg1/Vg4 cells or isolated CD8 + T cells from the spleen of recipient mice were obtained and cocultured with A20 cells at different E:T ratios for 6 hours. The killing capability was assessed according to the manufacturer's protocol. The percentage of cytotoxicity at each E:T ratio was calculated using the following formula: percentage of cytotoxicity = (experimental − effector spontaneous − target spontaneous)/(target maximum − target spontaneous) × 100%.

Flow Cytometry
Single cell suspensions from spleens, livers, lungs, and IELs were obtained according to the methods previously described (31)

Statistical Analysis
One-way ANOVA was used to determine statistically significant differences among more than two experimental groups. The unpaired Student t-test was used to determine statistically significant differences between the two experimental groups. Data were analyzed using GraphPad Prism 5 software for Windows (GraphPad Software, San Diego, CA). P-value <0.05 was considered statistically significant (*), the significance levels are marked as *p <0.05, ** p<0.01, *** p<0.001 and **** p<0.0001.

Donor gdT Cells Exert GVL Effect During allo-HSCT
To investigate the role of donor gdT cells during allo-HSCT, we established a murine GVL model. BALB/c mice were lethally irradiated and received bone marrow cells (BMCs, 5×10 6 cells/ mouse, iv.) from WT or gdT deficient (TCRd -/-) C57BL/6 donor mice. A20 cells (1 × 10 6 cells/mouse, iv.) were injected into the recipients intravenously on the day of transplantation ( Figure  1A). The survival of the recipient mice was monitored and weighed every other day. The results showed that in the absence of donor gdT cells, the survival and body weights of the recipient mice were significantly reduced ( Figures 1B), suggesting that donor-derived gdT cells could enhance GVL effect during allo-HSCT.
Donor gdT Cells Are Essential for the Production of IFN-g in CD8 + T Cells Other than direct cytotoxicity against tumor cells, donor gdT cells may have anti-leukemia effect by regulating abT cell functions (32). To explore the regulatory role of donor gdT cells in vivo, we examined the immune phenotypes of T lymphocytes on day 10 post allo-HSCT in mice receiving either WT or TCR d -/-BMs (5×10 6 cells/mouse, iv.) together with A20 cells (1 × 10 6 cells/mouse, iv.) (Figures 2A-C). By CD62L and CD44 expression, naïve, effector, and memory subsets of CD4 + and CD8 + T subsets from the spleen, liver, and lung were examined. The results showed that there was no obvious difference in the activation of CD4 + or CD8 + T cells between the mice receiving WT BMCs and those receiving TCR d -/-BMCs.
IFN-g is critical in mediating anti-leukemia activity of T lymphocytes (33,34). We then examined the IFN-g production by both CD4 + and CD8 + T cells in the spleen and liver of the recipient mice ( Figures 2D, E). CD8 + T cells were the main IFNg producers, and the results showed that the capability of CD8 + T cells producing IFN-g was severely impaired in the absence of donor gdT cells. The percent of IFN-g-producing CD8 + T cells decreased from 39.7 to 4.0% in the spleen and from 49.0 to 0.4% in the liver. To exclude the possibility of intrinsic low IFN-g production by CD8 + T cells in TCR-d -/mice, we examined IFN-g production by both CD4 + and CD8 + T cells in the spleen and liver of naïve WT and TCR-d -/mice ( Figures S1A, B). The results showed that there was no difference in the percent of CD4 + and CD8 + T cells, or the IFN-g-producing cells between WT and TCR-d -/mice. To investigate whether the different IFNg production by CD8 + T cells in the recipients was affected by the changes in regulatory T cells, we examined the proportion of regulatory T cells in the spleen and liver of the host mice that received BMCs from WT or TCR-d -/mice. We found that there was no difference in the percentages of regulatory T cells in the spleen and liver of the recipient mice receiving WT or TCR-d -/-BMCs ( Figure S1C). Thus, donor gdT cells could exert GVL effect via regulating IFN-g production by CD8 + T cells in allo-HSCT without affecting regulatory T cells. There are two major gdT cell subsets in the mouse periphery tissues, Vg1 and Vg4 cells. These two subsets can have different functions in various diseases (10). To investigate the roles of donor Vg1 and Vg4 cell subsets in mediating GVL effect post allo-HSCT, we first compared the cytotoxicity of the two cell subsets against A20 cells in vitro ( Figure 3A). We found that Vg1 cells exhibited a significantly higher level of cytotoxicity against A20 cells compared with Vg4 cells or total gdT cells. To investigate whether Vg1 cells are also the main cell subset mediating GVL effect in vivo, we adoptively transferred in vitro expanded Vg1, Vg4, or gdT cells (1 × 10 7 cells/mouse, iv.) and BMCs from TCR-d -/mouse (5 × 10 6 cells/mouse, iv.), as well as A20 cells (1 × 10 6 cells/mouse, iv.) on day 0 of allo-HSCT ( Figure 3B). The results showed Vg4 cell adoptive transfer significantly prolonged the survival of the recipients that received no gdT cell infusion, while Vg1 or total gdT cell infusion had no significant effect on their survival, suggesting Vg4 cells were the main cell subset mediating GVL effect during allo-HSCT.
To further confirm the GVL function of Vg4 cells, we depleted the Vg1 or Vg4 cells using specific anti-Vg1 or anti-Vg4 antibodies in WT BMC recipient mice ( Figure 3C). Anti-Vg1 or anti-Vg4 antibodies (100 µg/200 µl/mouse, ip.) were administered once a week for 4 weeks. The depletion efficiency of either cell subset was confirmed by flow cytometry ( Figure  S2). The results showed that the depletion of Vg4 cells A B D E C FIGURE 2 | Activation phenotypes and IFN-g production of CD4 + and CD8 + T cells in the recipients of WT or TCRd -/grafts after allo-HSCT. The allo-HSCT was performed as described in Figure 1A. The lymphocytes were isolated from host spleen or liver on day 10 post allo-HSCT and analyzed by flow cytometry. Percentages of naïve, effector and memory CD4 + and CD8 + T cells in spleen (A), liver (B), and lung (C) were shown. The production of IFN-g by CD4 + and CD8 + T cells in the spleen (D) and liver (E) post allo-HSCT were examined (NC-Negative control: grey shaded, WT: solid line, TCRd -/-: dotted line). All data are representative of at least 3 independent experiments with n = 7 mice per group. All graphs display mean ± SEM. Significance was determined by unpaired 2-tailed Student's t tests. **p < 0.01, ****p < 0.0001. significantly reduced the survival of the recipients compared to WT group, while the depletion of Vg1 cells did not affect the survival of allo-HSCT recipients. To visualize and quantify leukemia growth in the hosts, we performed bioluminescent imaging by establishing a murine leukemia model using A20-luc cells ( Figure 3D). The results showed that the depletion of Vg1 cells displayed a comparable tumor burden to that of WT group. However, the depletion of Vg4 cells resulted in a higher tumor burden than WT or anti-Vg1 group, which was comparable to that of TCR-d -/-BMC recipients ( Figures 3D, E). These results demonstrated that Vg4 cells were the major gdT cell subset mediating GVL effect during allo-HSCT.

The GVL Function of Vg4 cells Is Partially Dependent on IL-17A Production
To investigate the mechanism of Vg4 cells mediating GVL function in vivo, we first compared the phenotypes of in vitro expanded Vg1, Vg4, and gdT cells (Figures S3A-G). Flow cytometry analysis revealed that the percentage of IL-17Aproducing cells was higher in Vg4 cells than in Vg1 and total gdT cells. We then established murine GVL model and adoptively transferred donor Vg1, Vg4, or gdT cells (1 × 10 7 cells/mouse, iv.) derived from CD45.1-TCR-b -/mice together with BMCs (5 × 10 6 cells/mouse, iv.) from TCRd -/mice and A20 cells (1 × 10 6 cells/mouse, iv.). CD8 + T cells from spleens of the recipient mice were isolated on day 7 post-transplantation and the cytotoxicity of CD8 + T cells against A20 cells was measured ( Figure 4A). The results showed that CD8 + T cells from the mice receiving Vg4 cells displayed increased cytotoxicity against A20 cells compared to the mice receiving no adoptive transfer. The total number of CD8 + T cells in the spleen and liver of the recipient mice showed no difference among different groups ( Figure S3H). Flow cytometry results demonstrated that the percentage of IL-17A-producing cells was significantly higher in adoptively transferred Vg4 cells than in Vg1 or gdT cells in the liver of the recipient mice ( Figure 4B), which is consistent with their phenotypes in vitro. IL-17A has been shown to promote anti-tumor T cell response in murine models and human tumors (35,36). It might be the effector molecule mediating the regulatory function of Vg4 cells.
Donor gdT Cells Mitigate aGVHD During allo-HSCT aGVHD is one of the major complications post allo-HSCT causing patients' mortality. Generally, aGVHD is induced by abT cells that are also critical for GVL function (4). To determine whether donor gdT cells could induce aGVHD while promoting GVL effects, we established a murine aGVHD model by using BMCs (1 × 10 7 cells/mouse, iv.) and splenocytes (5 × 10 6 cells/mouse, iv.) from WT or TCR-d -/mice (C57BL/6) ( Figure 5A). The results showed that the deficiency of gdT cells in donor grafts resulted in accelerated aGVHD-related death in recipient mice. The recipients that received TCR-d -/grafts A B D C FIGURE 4 | The enhanced GVL effect of Vg4 cells was partially dependent on the production of IL-17A. Recipient mice were lethally irradiated and received A20 lymphoma cells (1 × 10 6 cells/mouse, iv.) plus BMCs (5 × 10 6 cells/mouse, iv.) from TCRd -/mice. Vg1, Vg4, or total gdT cells (1 × 10 7 cells/mouse, iv.) from CD45.1-TCRb -/mouse were adoptively transferred into the recipients on day 0. (A) The CD8 + T cells were isolated from the recipient mice (purity >99%) that received the A20 cells (1 × 10 6 cells/mouse, iv.) plus BMCs (5 × 10 6 cells/mouse, iv.) from TCRd -/mice with or without the adoptive transfer of Vg1, Vg4, or gdT cells (1 × 10 7 cells/ mouse, iv.) on day 7 post allo-HSCT. Cytotoxicity against A20 cells (at E:T ratio of 20:1 and 10:1) was measured by LDH assay. (B) The activation marker expression and cytokine production of adoptively transferred Vg1, Vg4, or gdT cells in the liver were examined by flow cytometry on day 7 post allo-HSCT. (C) Experimental design: BALB/c recipients were lethally irradiated and received A20 lymphoma cells cells (1 × 10 6 cells/mouse, iv.) and BMCs (5 × 10 6 cells/mouse, iv.) from WT or TCRd -/mice. Vg4 cells (1 × 10 7 cells/mouse, iv.) expanded from WT or IL-17A -/mouse were adoptively transferred into recipients on day 0. (D) The survival of recipients that received WT-Vg4 or IL-17A -/-Vg4 cells. All data are representative of at least 3 independent experiments with n ≥5 mice per group. All graphs display mean ± SEM. Significance was determined by one-way ANOVA test (A-B) and log-rank (Mantel-Cox) survival test (D). *p < 0.05, **p < 0.01, ***p < 0.001. displayed lower body weight and higher GVHD clinical scores compared with the WT group ( Figures 5B, C). In addition, we performed histopathology with livers, lungs, and small intestines from the recipients on day 7 post-transplantation. Compared with the WT recipients, the TCR-d -/recipient mice displayed much more severe tissue damage in the livers, lungs, and small intestines known as the typical characteristics of aGVHD, including the damage of parenchymal hepatic cells and pulmonary alveoli, the infiltration of lymphocytes, incomplete intestinal villus epithelial structure, and epithelial cell shedding ( Figure 5D). These results indicated that donor gdT cells could mitigate aGVHD during allo-HSCT.
To investigate whether abT cell activation could be affected by donor gdT cells in the murine aGVHD model, we examined the immune phenotypes of abT cells from aGVHD target organs. In the absence of donor gdT cells, percent of activated CD4 + T cells was significantly increased in the liver and lung, while the changes were not significant in the spleen or intestinal intraepithelial lymphocytes (IELs) ( Figure 5E). Moreover, the percentage of CD44 + CD62Leffector CD4 + T cells displayed a significant increase in the IELs and a trend of increase in the spleen in the TCR-d -/graft recipients compared with WT recipients ( Figure 5F). These results suggest that donor gdT cells may suppress aGVHD by inhibiting CD4 + T cell activation.

Donor Vg4 gdT Cells Are the Main Cell Subset Mitigating aGVHD
To investigate which gdT cell subset plays the protective role in aGVHD, we adoptively transferred the in vitro expanded Vg1, Vg4, or total gdT cells (1 × 10 7 cells/mouse, iv. from TCR-b -/-C57BL/6 mice) in the murine aGVHD model described before (Figures 6A, B). Vg4 cell infusion significantly prolonged the survival of aGVHD recipients of TCR-d -/grafts and reduced the severity of aGVHD symptoms, while Vg1 or total gdT cell FIGURE 5 | Donor-derived gdT cells could mitigate aGVHD during allo-HSCT. Recipient mice (BALB/c) were lethally irradiated and given splenocytes (5 × 10 6 cells/ mouse, iv.) plus BMCs (1 × 10 7 cells/mouse, iv.) from WT or TCR-d -/-C57BL/6 donor mouse to establish murine aGVHD model after allo-HSCT. Survival of recipients (A), body weight changes (B), and clinical scores (C) were monitored over time. (D) Histopathology of livers, lungs and small intestines of the recipients of WT or TCR d -/grafts on day 7 post transplantation. The activation phenotype of lymphocytes was examined on day 7 post allo-HSCT. (E) The percentages of activated CD4 + T cells in the spleen, liver, lung and IEL of the recipients. (F) The percentages of effector CD4 + T cells in the spleen, liver, lung and IEL of the recipients. All data are representative of at least 3 independent experiments with n = 4-6 mice per group. All graphs display mean ± SEM. Significance was determined by log-rank (Mantel-Cox) survival test (A) and unpaired 2-tailed Student's t tests (B-E). *p < 0.05, **p<0.01, ***p < 0.001.
To further confirm this finding, Vg1 or Vg4 gdT cells were depleted with specific anti-Vg1 or anti-Vg4 antibodies (100 µg/ 200 µl/mouse, ip.) in the WT recipients ( Figures 6C, D). Depletion of Vg4 cells aggravated the progress of aGVHD and the survival was similar to that of the TCR-d -/recipients, while Vg1 depletion exhibited no effect on aGVHD progression in the WT recipients. These results indicated that donor Vg4 cells were the main gdT cell subset mitigating aGVHD during allo-HSCT.

DISCUSSION
gdT cells have been reported to reconstitute faster than abT cells after allo-HSCT (13), thus might play an important role in modulating GVL and aGVHD at the early stage of allo-HSCT. gdT cells and NK cells share a series of features, including the expression of surface receptors and non-MHC-restricted recognition (22). Donor NK cell infusion can promote engraftment, enhance GVL effect and suppress aGVHD after allo-HSCT (37). In the current study, we found that donor gdT cells could also promote GVL effect and mitigate aGVHD during allo-HSCT. Further analysis revealed that Vg4 gdT cells were the main cell subset mediating both functions by regulating CD4 + and CD8 + abT cell responses.
gdT cells have been shown to have direct cytotoxicity against tumor cells. Human gdT cells can directly kill CML blasts and other tumor cells (15,38,39). In vitro expanded human Vg9Vd2 T cells can efficiently kill EBV-transformed autologous lymphoblastic B cell lines (16). The adoptive transfer of these expanded human Vg9Vd2 T cells significantly prevents disease progression in humanized mice. gdT cells also display direct cytotoxicity against solid tumors, such as melanoma, prostate cancer, breast cancer, and lung carcinomas (23,(40)(41)(42). gdT cells exert the direct anti-tumor effect by the engagement of surface receptors, including gdTCR and NKG2D. The expanded donor gdT cells in our experimental system also exhibited direct cytotoxicity against A20 cells. However, this in vitro killing capacity was not associated with the in vivo antileukemia activity of the gdT cell subsets we examined. Vg4 gdT cells showed a lower level of cytotoxicity in vitro but superior GVL effect in vivo compared to Vg1 gdT cells, suggesting the immune regulatory role of gdT cells may be more critical than their direct killing capacity in regulating GVL effect in vivo after allo-HSCT.
Other than direct cytotoxicity against tumor cells, gdT cells can also function as antigen presenting cells to stimulate adaptive immune responses. Human Vg9Vd2 T cells expanded in vitro can present exogenous soluble protein epitopes via MHC class I complexes to antigen-specific CD8 + abT cells (32). Due to their early reconstitution after allo-HSCT, gdT cells may serve as antigen presenting cells at the early stage of immune reconstitution to activate leukemia-specific CD8 + T cell response. We found that IFN-g production in CD8 + T cells was severely impaired in the absence of donor gdT cells post allo-HSCT. By using IL-17Adeficient donor Vg4 gdT cells, we also demonstrated that the GVL effect mediated by Vg4 gdT cells was partially dependent on IL-17A.
The role of IL-17A-producing gdT cells in tumor development is controversial and could be tumor model-dependent. They are found to promote tumor growth and metastasis in both mice and humans (43). Vg4 gdT cells in the liver can enhance the development of murine hepatocellular carcinoma by producing chemokines that recruit MDSCs in the tumor microenvironment (19). Consistently, IL-17-producing gdT cell infiltration is positively correlated with the severity of human colorectal carcinoma (44). However, there are also studies showing that the anti-tumor CD8 + T cell response can be facilitated by IL-17-producing gdT cells (35,45). Therefore, the function of IL-17-producing gdT cells could be tumor type-and environment-dependent. In the current study, we discovered that IL-17A produced by donor Vg4 gdT cells might be involved in promoting the GVL effect after allo-HSCT. Further studies are needed to investigate the mechanism of such an effect mediated by IL-17A.
Interestingly, our previous study showed that IL-17A was protective in murine aGVHD models by modulating CD4 + T cell responses (46). In fact, donor Vg4 gdT cells, which produce higher levels of IL-17A than other gdT cell subsets, mitigated aGVHD in the murine model of allo-HSCT. Donor gdT cells significantly inhibited CD4 + T cell activation, which is the main cellular event for aGVHD responses. However, one study showed that the depletion of donor gdT cells prevented aGVHD during allo-HSCT (47). A recent study reported donor gdT cells alleviated aGVHD when the administration of abT cells was delayed for two weeks and the mitigation of aGVHD by donor gdT cells occurred only at high doses (25). This low efficacy of donor gdT cell infusion in mitigating aGVHD could be due to the heterogeneity of the in vitro expanded gdT cells. We demonstrated by using TCRd -/donors that donor gdT cells are critical in mitigating aGVHD during allo-HSCT. However, only infusion of Vg4 gdT cells exhibited prolonged survival in recipient mice, while adoptively transfer of total gdT cells had no effect on the progression of aGVHD, which is consistent with the previous study. Nevertheless, the detailed mechanism of Vg4 gdT cells mitigating aGVHD warrants further studies. Different subsets of gdT cells have been reported to have different, even opposite roles in various diseases. In B16 melanoma model, activated CD44 high Vg4 cells but not Vg1 cells exert dominant anti-tumor function by producing IFN-g and perforin (17). These two subsets were also reported to play distinct and opposing functions in the EAE model. Vg4 cells exacerbate disease symptoms by producing IL-17A, while Vg1 subset plays a protective role by secreting CCR5 ligands to regulate the Treg-Th17 balance (48). In human studies, high total gdT cell numbers after HSCT are associated with a favorable clinical outcome but not with aGVHD incidence (49). A recent study in 105 allo-HSCT recipients showed that the higher proportions of CD8 + gdT cells in the graft were associated with an increased incidence of aGVHD, while high proportions of CD27 + gdT cells had a trend of an inverse association with the relapse (50). Although there are studies indicating the links between the murine Vg gdT subsets and human Vd gdT subpopulations, how the functions of the murine gdT cell subsets can be correlated with human gdT cell populations needs further investigations.
Disease relapse and aGVHD are the main complications leading to the failure of allo-HSCT. Novel strategies are urgently needed to prevent aGVHD while preserving or promoting GVL effect. Our findings provide evidence supporting the notion that donor gdT cell infusion could be a potentially effective therapeutic strategy to enhance GVL and mitigate aGVHD during allo-HSCT.

DATA AVAILABILITY STATEMENT
All datasets presented in this study are included in the article/ Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Laboratory Animal Care and Use Committee of Soochow University and National University of Singapore.