Engineering donor lymphocytes with Fas ligand protein effectively prevents acute graft-versus-host disease

Allogeneic hematopoietic stem cell transplantation (HSCT) is an effective treatment for hematological malignancies, inherited hematological disorders, and immunodeficiencies.^1-3 Acute graft vs host disease (aGVHD) orchestrated by mature donor T cells is one of the major barriers to the broad and successful application of HSCT.^2,4,5 Exclusion of mature T cells by various ex vivo manipulations from the donor graft reduces the efficacy of engraftment and slows the pace of posttransplant immune recovery in recipients with myeloablative conditioning.⁶ More restricted approaches to deplete various donor T-cell subsets ex vivo have proven inefficient.^6-8 A prevalent procedure is the posttransplant administration of cyclophosphamide at the peak of reciprocal donor-host sensitization to reduce both host-versus-graft and graft-versus-host reactions.⁹ However, cyclophosphamide impairs immune reconstitution, thereby necessitating delayed donor lymphocyte infusion.¹⁰ Regardless of recent advances, aGVHD remains the leading cause of morbidity and mortality associated with HSCT.^2,5 

We herein report a novel strategy to eliminate alloreactive mature donor Teffs in myeloablated graft recipients by taking advantage of Fas ligand (FasL)–mediated apoptosis as a critical homeostatic immune control mechanism.^11-13 Fas pathway serves as a feedback mechanism to terminate an ongoing immune response and is critical to the maintenance of peripheral tolerance.^12,14 Activated Teff cells upregulate Fas expression and become sensitive to FasL-mediated apoptosis after repeated interaction of T-cell receptor with the target antigen, defined as restimulation-induced cell death (RICD). The current study assessed RICD as a potential mechanism to eliminate alloreactive Teffs in the graft recipients for the prevention of aGVHD. We previously reported a novel form of FasL protein chimeric with streptavidin (SA-FasL). There are 2 unique features of this construct: (1) SA-FasL exists as tetramers and oligomers with robust apoptotic activity on Fas-expressing cells; and (2) SA-FasL can be positionally and transiently displayed on biotinylated biologic or nonbiologic surfaces with an in vivo half-life of ∼3.5 days with demonstrated immunomodulatory efficacy in various transplantation models.^15-21 The underlying hypothesis of this study was that SA-FasL on donor alloreactive T cells will engage Fas upregulated on the same cell in response to host alloantigens, leading to autocrine apoptosis and prevention of aGVHD.

Spleen cells were engineered with SA-FasL or streptavidin (SA) proteins per our previously reported protocols.¹⁶ Briefly, splenocytes were incubated in 5 μM EZ-Link Sulfo-NHS-LC biotin (hereafter referred as biotin) solution (Pierce) in sterile phosphate-buffered saline (PBS) at room temperature for 30 minutes. After washing, cells were incubated in PBS containing SA or SA-FasL proteins for 30 minutes, washed to remove unbound proteins, and then used for the indicated studies.

Human peripheral blood mononuclear cells (PBMCs) were isolated as previously reported from healthy donors undersigned informed consent approved by the Institutional Review Board of University of Louisville. Peripheral blood was collected in heparin-containing vacutainer (BD Bioscience, catalog# 364606). PBMCs were isolated from buffy coats by Ficoll-paque (GE, catalog# 17-1440-03) density centrifugation and washed by sterile PBS before further use. Human PBMCs were engineered with SA-FasL, as described for the mouse cells.¹⁶ 

4C.SJL or C57BL/6^hCD2 splenocytes were engineered with various amounts of SA-FasL or SA proteins. Engineered cells were labeled with 2.5 μM CTV, and 4C.SJL (5 × 10⁶ cells/mouse) or C57BL/6^hCD2 (10 × 10⁶ cells/mouse) cells were intravenously injected into F1 (C57BL/6xBALB/c; H-2^b/d). Mice were euthanized 48 or 72 hours after cell infusion to monitor donor cell proliferation using flow cytometry. For xenogeneic settings, human neutrophil-deplete PBMCs were engineered with SA-FasL (25 ng/10⁶ cells) or equimolar of SA protein as control. Engineered cells (5 × 10⁶ cells) were injected IV into NSG mice 4 hours after total body irradiation (200 cGy). Animals were euthanized 5 days after cell infusion to assess the frequency and absolute numbers of various lymphoid cell populations using flow cytometry by gating on human CD45⁺ cells.

F1 (C57BL/6xBALB/c; H-2^b/d) mice (10-12 weeks old) were subjected to 1000 cGy total body irradiation (Gammacell 40 Extractor, ¹³⁷Cs source) followed by IV infusion of 10 × 10⁶ C57BL/6 nonengineered bone marrow cells admixed with 20 × 10⁶ nonengineered or SA-FasL–engineered splenocytes. Animals were monitored twice weekly for general health and body weight, and acute GVHD clinical score was calculated per published literature.²² A clinical score of >6 and total body weight loss >25% were considered as experimental end point and animals were euthanized for various analysis. Long-term mice without aGVHD underwent transplantation with the skin from BALB/c and C3H third-party grafts (H-2^k) and monitored for rejection twice weekly, as previously reported.²³ 

NSG females (8-10 weeks old) were subjected to 200 cGy total body irradiation and injected via tail vein 4 hours later with 10 × 10⁶ SA-FasL–engineered or -nonengineered human PBMCs. Fresh human PBMCs were used in all experiments. Animals were monitored for body weight twice weekly, and the development of xenogeneic GVHD was assessed as published.²⁴ Animals with >25% body weight loss and a clinical aGVHD score >6 were euthanized as an experimental end point.

The comparison of the survival curves was done using the log-rank (Mantel-Cox) test. Data are shown as individual data points or as mean ± standard error mean, as depicted in the figure legends. Unpaired, two-tailed or Mann-Whitney t test was performed as indicated. For multiple comparison, one-way analysis of variance with Tukey post hoc test was performed. Statistical significance was defined as P < .05 (GraphPad prism v.8).

We first tested the impact of SA-FasL on T-cell proliferation in responses to alloantigens ex vivo. Splenocytes from 4C mice (H-2^b), transgenic for a T-cell receptor that recognizes BALB/c MHC I-A^d antigen,²⁵ were engineered with various amounts of SA-FasL or equimolar SA as a control protein and used as responders against irradiated BALB/c splenocytes in a standard [³H]-thymidine–based proliferation assay. There was a dose-dependent display of SA-FasL on the cell surface, as assessed using flow cytometry (supplemental Figure 1A). Nonengineered 4C and SA-engineered 4C T cells showed robust proliferation (Figure 1A). In marked contrast, there was an absence/minimal proliferation of SA-FasL–engineered 4C cells across all engineering levels. These observations were further confirmed using a carboxyfluorescein succinimidyl ester (CFSE)–based proliferation assay. Analysis of live cells at various times after culture using flow cytometry showed a significant reduction in the frequency of alloreactive CD4⁺ and CD8⁺ cells expressing the transgenic Vβ13 TCR as compared with control nonengineered 4C or SA-engineered 4C cells (Figure 1B; supplemental Figure 1B).

To investigate whether SA-FasL eliminates activated T cells via an autocrine mechanism by engaging with an upregulated Fas receptor on the same cell, CFSE-labeled SA-FasL–engineered 4C cells were comixed with equal numbers of CTV-labeled nonengineered cells and used as responders at various ratios to a fixed number of irradiated BALB/c stimulators. Flow cytometry analysis at 72 hours after culture showed ∼20% live CFSE⁺ vs ∼80% live CVT⁺ 4C cells across all stimulator/responder ratios used (Figure 1C). Although nonengineered 4C cells showed robust proliferation with accumulated daughter cells at each proliferation cycle, SA-FasL–engineered 4C cells showed minimal accumulation of proliferating cells. Similar results were found 48 hours after culture (supplemental Figure 1C). These observations implicate autocrine apoptosis as the primary mechanism of cell death, as paracrine apoptosis would have resulted in an increased level of death in naïve 4C T cells at the highest responder cell density (1:1) because of the greatest probability of cell-to-cell interaction. In fact, nonengineered 4C T cells in coculture with SA-FasL–engineered cells showed almost the same level of viability (∼80%) at all responder cell densities.

We next assessed the efficacy of SA-FasL in eliminating alloreactive cells in vivo by adoptive transfer of CTV-labeled 4C.SJL cells engineered with various doses of SA-FasL or SA control protein into F1 mice. SA-FasL significantly (P < .01) reduced the frequency of live donor (7AAD^- CD45.1⁺) cells as well as CD45.1⁺CD4⁺ T cells as compared with SA-engineered controls (Figure 2A). Decrease in donor cells engineered with SA-FasL was because of apoptosis (CD45.1⁺AnnexinV⁺) that translated into significantly (P < .01) reduced frequency and absolute number of proliferating cells as compared with the SA control group (Figure 2A; supplemental Figure 2A-B). These differences were not unique to the 4C TCR transgenic model as in vivo cell tracking assay in a regular parent-to-F1 model [C57BL/6-to-F1(BALB/cxC57BL/6)] resulted in similar observations (Figure 2B; supplemental Figure 2C). To assess if alloreactive T cells undergo apoptosis in an autocrine fashion in vivo, donor splenocytes engineered with SA-FasL were labeled with CFSE and comixed at 1:1 ratio SA-engineered and CTV-labeled cells and transplanted into lethally irradiated F1 recipients. A second group of mice that underwent transplantation with nonengineered and CFSE- or CTV-labeled cells at 1:1 ratio served as controls. Analysis of donor cells 72 hours after transplantation demonstrated specific elimination of SA-FasL–engineered cells (Figure 2C; supplemental Figure 2D-E), further confirming autocrine apoptosis as the underlying mechanism of alloreactive T-cell elimination. Autocrine apoptosis was not because of engineering with SA-FasL altering the Fas expression as the level of this receptor was similar to that on nonengineered cells (supplemental Figure 3). The in vivo turnover kinetic studies established ∼108 minutes as the half-life of SA-FasL on donor T cells transplanted into lethally irradiated F1 recipients (Figure 2D). Collectively, these results demonstrate that engineering a donor graft with SA-FasL protein is an effective approach to physically eliminate alloreactive Teffs primarily through autocrine apoptosis both ex vivo and in vivo.

To investigate whether engineering donor T cells prevents aGVHD, we used a haploidentical, myeloablative transplant model. C57BL/6 splenocytes (2 × 10⁷) engineered with various doses of SA-FasL protein comixed with whole bone marrow cells (1 × 10⁷) were transplanted into lethally irradiated F1 recipients. Recipients of control splenocytes (nonengineered or engineered with SA protein) showed high-grade clinical GVHD scores and significant weight loss, expiring within 46 days (Figure 3A). In marked contrast, >70% of mice that underwent transplantation with splenocytes engineered with SA-FasL doses exceeding 10 ng/10⁶ cells survived (P < .0001) with improved clinical aGVHD scores and body weight (Figure 3A). The efficacy of SA-FasL was dose-dependent, with the highest dose (25 ng/10⁶ cells) showing ∼73% survival, whereas the lowest dose (5 ng/10⁶ cells) resulted in ∼20% survival (Figure 3A) for a 100-day experimental end point.

Long-term survivors (>100 days) displayed full donor chimerism in the peripheral blood, bone marrow, and spleen (Figure 3B). These mice also showed efficient reconstitution of Treg (CD4⁺CD25⁺FoxP3⁺), Teff (CD4⁺CD44^hiCD62L^–), and NK (NK1.1⁺CD3^–) cells, with the Treg/Teff cell ratio being comparable to that of control GVHD-free mice receiving only BM cells (Figure 3C; supplemental Figure 4A-B). Long-term survivors were immunocompetent with established tolerance to parental antigens as they accepted BALB/c (H-2^d) skin graft while promptly rejecting C3H (H-2^k) third-party grafts within 24 days, mice that underwent transplantation with only bone marrow cells showed a similar rejection tempo (Figure 3D). These results demonstrate that transplantation with SA-FasL–engineered allogeneic cells is effective in preventing aGVHD without a detectable negative impact on immune tolerance and reconstitution.

SA-FasL may prevent aGVHD by 2 interlinked mechanisms; apoptosis of alloreactive Teffs and the release of apoptotic bodies from dying cells initiating a feed-forward immunoregulatory pathway resulting in the generation/expansion of Treg cells.²⁶ Activation and extensive proliferation of naïve alloreactive T cells in secondary lymph organs early after HSCT followed by homing to target tissues, skin, liver, intestine, are hallmarks of aGVHD.^27,28 When analyzed on day 7 after transplantation, recipients of SA-FasL–engineered splenocytes had reduced numbers of CD4⁺ and CD8⁺ Teffs in the liver and mesenteric lymph nodes as compared with mice grafted with nonengineered splenocytes (Figure 4A). Importantly, the SA-FasL group also had increased absolute numbers of CD4⁺CD25⁺FoxP3⁺ Tregs in the lymph nodes (P < .001), which translated into an increased Treg/Teff ratio (P < .0001, Figure 4B). Analysis on day 21 after transplantation revealed no significant differences in T cells subsets in the mesenteric lymph nodes of the SA-FasL and control groups, but the absolute intrahepatic numbers of CD4⁺ and CD8⁺ Teff cells were significantly reduced (P < .05) in the recipients of SA-FasL–engineered splenocytes. The absolute number of CD4⁺CD25⁺FoxP3⁺ Tregs was also increased (P < .05) in the liver of control group as compared with the liver of the SA-FasL group, which did not translate into a significantly increased ratio of Tregs/Teffs because of the higher numbers of Teffs (Figure 4B). The T-cell subset frequencies showed a similar pattern to absolute cell numbers in the lymph nodes and liver, with some variations (supplemental Figure 5). Analysis of the large intestine on day 21 after transplantation using immunohistochemistry showed extensive tissue damage and significantly higher numbers of CD4⁺ T cells in the control group than in the SA-FasL–engineered group, which showed normal tissue structure with minimal CD4⁺ T-cell infiltration (Figure 4C-D; supplemental Figure 5B-C). Taken together, these data demonstrate that SA-FasL eliminates Teffs in lymphoid tissues with a null effect on Tregs during the induction phase of the disease, shifting the balance of immune responses toward aGVHD prevention.

Proinflammatory cytokines and chemokines, released from activated donor immune cells as well as damaged host tissues owing to conditioning, play a critical role in the development of aGVHD. The transcript profile of cytokines and chemokines in aGVHD target tissues was assessed using quantitative reverse transcription polymerase chain reaction (qRT-PCR). We observed decreased levels of transcripts for IL-1β, IL-6, TNF-α, IFN-γ, and IL-23 proinflammatory cytokines implicated in GVHD^29,30 in the liver and small and large intestine of animals that underwent transplantation with SA-FasL–engineered splenocytes as compared with the recipients of nonengineered splenocytes on day 7 after transplantation, except for IL-6, which showed increased expression in large intestine (Figure 5A). The increases in the transcripts for proinflammatory cytokines were much more pronounced and significant (P < .05) on day 21 effector phase as compared with day 7 initiation phase of GVHD (Figure 5B). Transcripts for CCL2 were significantly reduced in the liver (P < .01) and small intestine (P < .05) of the SA-FasL-engineered splenocyte group on day 21 after transplantation, consistent with the role of this chemokine in the recruitment of CD8⁺ Teff cells into target tissues for GVHD.³¹ The transcripts for granulocyte macrophage colony stimulating factor, involved in the pathology of aGVHD,³² were also reduced in the liver (P < .001; day 7) of the SA-FasL-engineered splenocyte group as compared with the nonengineered splenocyte group, with the exception of the small intestine on day 21 that showed a significant increase (P < .001).

SA-FasL–engineered splenocyte group also showed reduced expression of T-bet transcription factor, a master regulator of Th1 differentiation,³³ in all GVHD target organs, reaching significance (P < .05) on day 21 after transplantation (supplemental Figure 6). The level of transcripts for day 7 were not significant between the 2 groups, except small intestine where SA-FasL–engineered splenocyte group showed a significant (P < .05) increase. RORγt, transcriptional regulator of Th17, showed a mixed pattern of expression with reduced level of transcripts in the small intestine (P < .01, day 7) and large intestine (P < .01, day 21) of SA-FasL–engineered splenocyte group with increased levels (P < .01) on day 21 in the liver and small intestine as compared with those of the nonengineered splenocyte group (supplemental Figure 6). The SA-FasL–engineered splenocyte group also showed decreased levels of transcripts for FoxP3 in all 3 tissues on day 7 and only the large intestine on day 21 after transplantation, which is consistent with the paucity of CD4⁺ T cells detected using immunohistochemistry (Figure 4C-D; supplemental Figure 5C). Taken together, SA-FasL blunts the inflammatory responses implicated in the pathogenesis of aGVHD.

Given the demonstrated role of donor CD4⁺FoxP3⁺ Treg cells in the prevention of aGVHD^34,35 and increased ratio of Treg-to-Teffs in lymph nodes and liver of SA-FasL group on day 7 after transplantation, we investigated the role of donor Treg cells in protection against aGVHD in our model. C57BL/6 splenocytes were engineered with SA-FasL and subjected to negative selection of CD4⁺CD25⁺FoxP3⁺. Donor splenocytes depleted for CD25⁺ cells had marked reduction of FoxP3⁺ Treg cells as compared with unmanipulated splenocytes (0.33 vs 7.12%; supplemental Figure 7). Transplantation of SA-FasL–engineered whole splenocytes into a lethally irradiated haploidentical GVHD model resulted in the long-term survival of ∼70% mice (>60-day observation period; Figure 6A), characterized by a marked reduction in clinical scores (Figure 6B) and preserved body weight (Figure 6C). In contrast, all recipients of Treg depleted SA-FasL–engineered splenocytes developed severe GVHD, presenting mortality and disease indices similar to recipients of nonengineered allogeneic splenocytes. These studies implicate donor Treg cells as an important regulatory mechanism that confers SA-FasL–mediated prevention of aGVHD in our model.

We next assessed the efficacy of SA-FasL in eliminating human T cells infused into immunocompromised NOD-scid-IL2γR^null (NSG) mice as an important step to clinical translation. When analyzed 5 days after infusion, the spleen of SA-engineered PBMC recipients had significantly higher numbers (Figure 7A) and frequencies (supplemental Figure 8) of human total CD45⁺ cells as well as CD4⁺ and CD8⁺ T cells as compared with the recipients of SA-FasL–engineered cells. A similar pattern was also observed in the liver (Figure 7A; supplemental Figure 8), a major target for human T-cell–mediated xenogeneic aGVHD. Infusion of 10⁷ SA-engineered human PBMC into sublethally irradiated NSG mice resulted in aGVHD with severe clinical scores and a median survival time of 15 days (Figure 7B). In marked contrast, recipients of SA-FasL–engineered PBMCs showed improved survival along with a significant decline in clinical GVHD scores and preserved body weight. Notably, the improved survival was dose-dependent, as a higher concentration of SA-FasL protein (50 ng/10⁶ cells) was more effective in preventing aGVHD than a lower concentration (25 ng/10⁶ cells), 60% vs 25%, over the 60-day experimental end point. Thus, it remains to be investigated if higher concentrations of SA-FasL result in further improved efficacy in abrogating aGVHD in this preclinical model of the human disease.

Ex vivo pan T-cell depletion from the donor inoculum to mitigate aGVHD is associated with compromised engraftment and delayed posttransplant immune reconstitution.⁶ Alloreactive T cells upregulate the Fas receptor after activation and become susceptible to Fas-mediated apoptosis.^12,36,37 Thus, the Fas pathway provides a unique opportunity to preferentially eliminate alloreactive Teffs as a prophylactic approach for aGVHD. Here, we tested the efficacy of SA-FasL transiently displayed on donor lymphocytes to induce apoptosis in alloreactive Teffs and prevent aGVHD. SA-FasL on T cells was effective in inhibiting their proliferation in response to alloantigens in vitro and in vivo primarily through autocrine apoptosis, where SA-FasL engages with Fas on the same cell. Mice receiving SA-FasL–engineered allogeneic splenocytes showed significantly lower frequency and absolute number of donor cells as compared with the control group adoptively transferred with SA-engineered cells. The reduction in donor cells correlated with increased apoptosis and occurred at all SA-FasL protein doses used for engineering. These observations are in line with studies reporting elimination of alloreactive T cells in ex vivo cultures treated with soluble FasL protein^1,38,39 and single-cell studies demonstrating autocrine apoptosis as the major driver of FasL-mediated cell death.^40,41 However, we cannot rule out paracrine apoptosis that may occur in vivo because of the close physical proximity of alloreactive Teffs in the target tissues.

In a haploidentical mouse model, transient display of SA-FasL on donor lymphocytes effectively prevented GVHD in over 70% of graft recipients. This immunomodulatory scheme was also effective in mitigating aGVHD in a humanized mouse model of xenogeneic aGVHD. The efficacy of SA-FasL was dose-dependent, as higher doses of the protein resulted in better protection. These observations are consistent with several studies using soluble FasL for ex vivo elimination of alloreactive T cells as an approach to mitigate aGVHD.^1,38,39 Our strategy targeting the elimination of alloreactive Teffs in the graft recipient has 3 distinct advantages over studies using soluble FasL for ex vivo treatment of the donor graft. First, in vitro cultures represent a contrived system that lacks the intricate cellular and molecular networks occurring in a 3-dimensional space that dictate Teff activation requirements, effector function, expansion, and establishment of long-term memory. Considerable differences in T-cell activation between in vitro and in vivo systems have been reported.^42,43 Particularly, there is activation-independent significant T-cell death in vitro,⁴³ which may restrict the T-cell repertoire and immune reconstitution after transplantation into allogeneic host subjected to myeloablative conditioning. The direct display of SA-FasL on donor graft overcomes these limitations and has the potential to specifically eliminate Teffs responding to host alloantigens after transplantation, leaving alloantigen-unrelated repertoire unaltered. In support of this notion, long-term mice showed full donor chimerism, efficient immune engraftment and reconstitution, and accepted parental skin grafts while rejecting third-party grafts, demonstrating immune tolerance and competency. Thus, long-term animals are expected to generate an effective immune response to tumors in line with studies demonstrating ex vivo treatment of donor cells with soluble FasL as a prophylactic approach preserves the graft-vs-tumor effect.^7,8 In marked contrast, the efficacy of treating donor HSC ex vivo with soluble FasL is limited to the elimination of donor T memory cells in the graft, irrespective of their antigen specification, without an impact on naïve alloreactive T cells that have been shown to be the major driver of aGVHD.^8,44,45 Second, apoptosis of alloreactive T cells in vivo may initiate a feed-forward regulatory pathway involving immunoregulatory cytokines and Tregs^20,46-48 that contributes to the prevention of aGVHD. Furthermore, apoptotic bodies can alter the maturation of antigen-presenting cells (APCs), which is critical to the priming and initiation of lethal aGVHD, and polarize them toward tolerance induction.^49-51 In line with these reports, we observed a high ratio of Tregs to Teffs and significantly reduced levels of inflammatory cytokines, particularly granulocyte macrophage colony stimulating factor involved in the activation and maturation of APCs,³² in the SA-FasL group. Third, pluripotent hematopoietic stem cells are shown to be resistant to Fas-mediated apoptosis and respond to the FasL trophic effect for improved engraftment.^52,53 Thus, the presence of SA-FasL on the donor graft may also facilitate hematopoietic stem cell engraftment, an additional advantage over the ex vivo treatment of the graft with soluble FasL.

Immunophenotyping of the mice that underwent transplantation revealed several interesting findings. First, the numbers of both CD4⁺ and CD8⁺ Teff subsets were significantly increased in the mesenteric lymph nodes and the liver of the GVHD control group as compared with the SA-FasL group 1 week after transplantation. After 3 weeks, the increase in the number of these cell types leveled off in the lymph nodes while becoming more amplified in the liver of the control group than in the SA-FasL group. Second, the reduction in the numbers of Teff cells in the SA-FasL group translated into significantly reduced levels of proinflammatory cytokines (IL-1β, IL-6, TNF-α, and IFN-γ) implicated in pathogenicity of aGVHD.^2,29,54 These findings are consistent with the progression of aGVHD, which begins with the activation of donor T cells in response to alloantigens presented by host APCs in local lymph nodes and culminates in a cascade of molecular and cellular interactions followed by immune effector cell trafficking to various target tissues to inflict damage.^27,55 Immunohistochemical analysis of the large intestine on day 21 showed intense infiltration with CD4⁺ T cells and significant tissue damage in the control group, whereas the SA-FasL group had minimal infiltration and tissue damage. Third, the increased Treg numbers in the lymph nodes of the SA-FasL group early after transplantation is consistent with the demonstrated sensitivity of activated alloreactive T cells, but not Treg cells, to FasL-mediated apoptosis, which is regulated by cytokines in an inflammatory milieu.^56,57 Increased proliferation in response to stimulatory cytokines and/or conversion of naïve T cells may also contribute to the increased frequency of Treg cells in our model. This notion is consistent with our recent studies demonstrating that SA-FasL–mediated tolerance to allogeneic islets involves apoptosis of Teff cells that culminates in a cascade of regulatory mechanisms involving phagocyte uptake of apoptotic bodies, secretion of TGF-β, and increased frequency of Treg cells that are critical for induction and maintenance of long-term graft survival.²⁰ Human Treg cells use FasL to induce apoptosis in both APCs and Teff cells as a mechanism of immune suppression.^58-60 Importantly, Treg cells engineered to display SA-FasL on their surface are resistant to apoptosis and have enhanced immunoregulatory function.^61,62 Exclusion of Treg from the parental donor splenocyte inoculum mitigated the protective efficacy of SA-FasL, consistent with published studies reporting a key role for donor Tregs in the prophylaxis of aGVHD.^35,63 Our findings are also consistent with studies using cyclophosphamide to prevent aGVHD by targeting Teffs for physical and functional elimination. Donor Tregs were shown to be required for the efficacy of cyclophosphamide in preventing aGVHD in various mouse models.^5,6 We previously shown that SA-FasL–engineered Tregs show significantly improved efficacy as compared with nonengineered cells in preventing aGVHD in an adoptive transfer preclinical model.⁶⁴ Thus, the direct display of SA-FasL on T cells in donor inoculum provides 2 distinct advantages: preferential elimination of alloreactive Teffs and functional enhancement of Tregs.

We herein demonstrate that transient display of a novel form of FasL protein, SA-FasL, on the donor lymphocytes is effective in eliminating alloreactive Teffs posttransplantation, leading to the prevention of aGVHD in haploidentical and humanized mouse models (visual abstract). The process of donor graft engineering is straightforward and efficient, providing an attractive platform for clinical translation. SA-FasL does not persist in the system because of its short half-life ∼108 minutes on CD3⁺ T cells, thereby minimizing potential off-target effects. Our approach harbors significant potential as monotherapy or in combination with various clinical regimens to improve both the efficacy and safety of mismatched T-cell–replete hematopoietic cell transplants. This approach may also improve donor lymphocyte infusion to manage the relapse of hematological malignancies after HSCT.

Contribution: H.S., E.S.Y., and P.S. conceived and designed studies; P.S., L.B., A.T., A.E.G., Z.S., A.E.G., and H.T. performed the study and collected and analyzed the data; and P.S., A.T., E.S.Y., H.S., and N.A. interpreted the analysis and wrote the manuscript.

Conflict-of-interest disclosure: H.S., E.S.Y., and P.S. have a provisional patent on using SA-FasL–engineered cells as a prophylactic approach for aGVHD. H.S. is CEO of Fascure Therapeutics, LLC, and the scientific cofounder, stockholder, and member of SAB for iTolerance, Inc. E.S.Y. is a consultant for iTolerance. The remaining authors declare no competing financial interests.

Correspondence: Esma S. Yolcu, University of Missouri, Columbia Child Health and Molecular Microbiology and Immunology NextGen Precision Health Bldg, Columbia, MO 65211; e-mail: esma.yolcu@health.missouri.edu; and Haval Shirwan, University of Missouri, Columbia Child Health and Molecular Microbiology and Immunology NextGen Precision Health Bldg, Columbia, MO 65211; e-mail: haval.shirwan@health.missouri.edu.