TIGIT regulates apoptosis of risky memory T cell subsets implicated in belatacept-resistant rejection

Abstract

Belatacept confers increased patient and graft survival in renal transplant recipients relative to calcineurin inhibitors, but is associated with an increased rate of acute rejection. Recent immunophenotypic studies comparing pre-transplant T cell phenotypes of patients who reject vs. those that remain stable on belatacept identified three potential “risky” memory T cell subsets that potentially underlie belatacept-resistant rejection: CD4⁺ CD28⁺ T_EM, CD8⁺ CD28^null and CD4⁺ CD57⁺ PD1⁻ subsets. Here, we compared key phenotypic and functional aspects of these human memory T cell subsets, with the goal of identifying additional potential targets to modulate them. Results demonstrate that TIGIT, an increasingly well-appreciated immune checkpoint receptor, was expressed on all three risky memory T cell subsets in vitro and in vivo in the presence of belatacept. Co-culture of human memory CD4⁺ and CD8⁺ T cells with an agonistic anti-TIGIT mAb significantly increased apoptotic cell death of all three risky memory T cell subsets. Mechanistically, TIGIT-mediated apoptosis of risky memory T cells was dependent on FOXP3⁺ Treg, suggesting that agonism of the TIGIT pathway increases FOXP3⁺ Treg suppression of human memory T cell populations. Overall, these data suggest that TIGIT agonism could represent a new therapeutic target to inhibit belatacept-resistant rejection during transplantation.

Introduction

Transplantation is a curative, life-saving therapy for end-stage organ failure. However, the incidence of significant morbidity and graft loss due to immunosuppression-induced toxicities remains unacceptably high (1, 2). CD28 costimulation blockade in the form of belatacept is a CTLA-4Ig fusion protein and the first new primary immunosuppressive agent in transplantation in over 20 years (3, 4). Belatacept offers a significant benefit to renal transplant recipients in that it carries a 43% reduced risk of death or graft loss after 7 years as compared to calcineurin inhibitor-based regimens (5). Nonetheless, belatacept confers a significantly increased risk of acute rejection as compared to calcineurin inhibitors (6–8), a fact that has limited its uptake in the clinical transplant community. Our work and that of many others has revealed that belatacept-resistant rejection is likely the result of the activation of memory T cells that have a diminished requirement for CD28 signaling (9–12). The resistance of memory T cells to the effects of costimulatory blockade may therefore markedly limited patient access to the considerable benefits of belatacept.

To optimize the potential of belatacept in clinical transplantation, several groups have endeavored to identify “risky” memory T cell phenotypes that may predict the development of belatacept-resistant rejection. First, it has been hypothesized that CD28^null CD4⁺ and CD8⁺ T cell populations contribute to belatacept-resistant rejection, owing to their lack of expression of the target of belatacept (3, 4) and the fact that patients with kidney failure awaiting transplantation have been described to accumulate CD28^null cells in peripheral blood (13, 14). However, whether these cells are functionally capable of mediating belatacept-resistant rejection remains a topic of ongoing debate in the field (15–17). In contrast, our previous findings suggest that patients possessing higher frequency of CD28⁺ CD4⁺ T_EM prior to transplant are more likely to experience acute rejection following treatment with a belatacept-based immunosuppressive regimen, and pre-transplant frequencies of CD28⁺ CD4⁺ T_EM have the potential to be used as a biomarker to predict the risk of rejection following treatment with belatacept (18). Further, a population of CD57⁺ PD1⁻ CD4⁺ T cells present prior to transplantation has also been reported to correlate with belatacept-resistant rejection, and likewise this immunophenotype has been proposed to identify patients at higher risk for acute rejection on belatacept-based therapy (19). The similarities and differences, as well as potential overlap, between these memory T cell subsets has not been investigated. Moreover, identification of molecules and pathways that regulate their activation and functionality in the setting of CD28 blockade remains an important goal for immunotherapy in clinical transplantation.

The T cell immunoreceptor with Ig and ITIM domains (TIGIT) has recently been identified as an important immune checkpoint receptor. TIGIT is a CD28 family member (20, 21) that pairs with DNAM (CD226) to form an increasingly well-appreciated cosignaling pathway: DNAM and TIGIT compete for binding to the same set of ligands (CD155 and CD112), but TIGIT is coinhibitory for T cells while DNAM is costimulatory (22, 23). These relationships are analogous to the CD28/ CTLA-4/ CD80/ CD86 family of costimulatory/ coinhibitory molecules.

TIGIT can inhibit T cell responses by binding the ligand CD155 on DCs and thereby inhibiting IL-12 while inducing IL-10 production, resulting in a cell-extrinsic mechanism of inhibition of T cell responses (24, 25). Furthermore, recent studies have revealed that TIGIT expression defines a population of highly suppressive regulatory T cells (T_reg) in both mouse and humans (20, 26, 27), and contributes to the selective T_reg cell-mediated suppression of pro-inflammatory Th1 and Th17 cells but not Th2 responses (27). Importantly, TIGIT is currently being targeted therapeutically in several ongoing clinical trials in cancer (28–31) (#NCT02794571, #NCT03119428, #NCT03628677 ClinicalTrials.gov). However, the role of TIGIT signaling in regulating alloreactive immune responses, especially these risky memory T cell subsets implicated in belatacept-resistant rejection in the context of solid organ transplantation has not been fully elucidated. Furthermore, the ability of TIGIT to inhibit memory/effector T cells in a cell-intrinsic fashion remains controversial. For instance, in a naïve mouse tumor model, TIGIT primarily suppressed CD8⁺ antitumor immunity via T_reg cells and not a cell-autonomous effect on CD8⁺ T cells (32). Conversely, TIGIT has also been reported to directly regulate the expansion and function of tumor antigen-specific CD8⁺ T cells in vitro in the absence of T_reg (33).

Here, we endeavored to define the expression of TIGIT on risky human memory T cell subsets associated with belatacept-resistant rejection, and show that TIGIT agonism attenuates risky memory T cell responses by increasing apoptotic cell death in a FOXP3⁺ Treg-dependent fashion.

Materials and Methods

Human study approval, sample collection, and patient immunosuppression

PBMCs were collected from healthy volunteers and renal transplant patients treated with belatacept following protocols approved by the Emory University Institutional Review Board (IRB #00006248) after informed consent was obtained. Patients treated with belatacept received intravenous infusion of belatacept (10mg/kg) during surgery and on post-transplant days 28, 56, and 84 with subsequent doses (5mg/kg) given every 4 weeks thereafter. Belatacept-treated patients also received anti-IL-2R induction (basiliximab 20mg iv on days 0 and 3 or 4), mycophenolate mofetil (MMF, 1g twice daily), a short steroid taper (methylprednisolone 500mg iv intra-operatively, 250 mg iv d1, 125mg iv d2, and prednisone 5mg d3 and daily thereafter), and a tacrolimus taper over the first 3-9 months (target trough levels 5-12ng/ml) as previously described (8). Patients who were Epstein Barr virus (EBV) seronegative, HIV⁺, had a history of post-transplant lymphoproliferative disease (PTLD), lymphoma or other hematologic malignancy, were undergoing treatment for latent tuberculosis or were the recipient of a simultaneous extra-renal organ were excluded from belatacept treatment. PBMC were obtained from patients who were stable on belatacept 6-12 months post-transplantation. Pre-transplant samples (n=5) consisted of 4 males and 1 female, average age 54.0 (range 44-63), 60% Black and 40% White, and underlying disease of diabetes mellitus (DM) (n=3), hypertension (n=1), and FSGS (n=1). Post-transplant samples (n=7) consisted of 6 males and 1 female, average age 54.4 (range 41-72), 57.1% Black, 28.6% White, and 14.4% American Indian or Alaskan Native; and underlying disease of DM (n=2), hypertension (n=4), and familial nephropathy (n=1). None of the pre- vs. post-transplant samples were from the same donor.

Ex vivo analysis of T cell phenotype and apoptosis

Standard extracellular staining (20 mins at 24°C) was performed using the following fluorophore-labeled antibodies: CD3-BUV737 (BD Biosciences), CD4-BUV395 (BD Biosciences), CD8-BV711 (BD Biosciences), CD14-BV510 (BD Biosciences), CD19-BV510 (BioLegend), Live/Dead Fixable Aqua (ThermoFisher), CCR7-PE-CF594 (BD Biosciences), CD45RA-APC-H7 (BD Biosciences), CD28-APC-R700 (BD Biosciences), CD57-PE (BioLegend), PD-1-BV605, BV421 (BioLegend), CD69-BV650 (BioLegend), CD38-BV421 (BioLegend), HLA-DR-PerCP-Cy5.5 (BD Biosciences), TIGIT-PE-Cy7 (BioLegend), DNAM-BB515 (BD Biosciences), FcγRIIB-BV786 (BD Biosciences). Fluorescence minus one (FMO) controls were used to determine negative expression of FcγRIIB. Apoptosis was measured with Caspase-3/7 Green Flow Cytometry Assay Kit (ThermoFisher) or FITC Annexin V with 7-AAD Viability Staining Solution (BioLegend), according to the manufacturer’s instructions.

Ex vivo polyclonal stimulation assay

Fresh PBMCs were isolated from healthy donors under IRB approval using CPT tubes and incubated in 24-well flat-bottomed plates in culture medium (R10) consisting of 1640 RPMI medium supplemented with 10% heat-inactivated FBS (Mediatech, Herdon, VA), 1% L-glutamine (200mM), 1% penicillin/streptomycin (100×), 1% Hepes (1M), 1% 2-ME (5mM). Cells were processed unstimulated, stimulated with anti-CD3/CD28 Dynabeads (ThermoFisher) per manufacturer instructions, or stimulated with 3 μg/mL plate-bound functional grade anti-CD3 (OKT3; eBiosciences) with clinically therapeutic concentrations of belatacept (10 μg/mL; Bristol-Myers Squibb, NY) (34, 35) for 3 days at 37°C and 5% CO₂, humidified atmosphere. Where indicated, cultures were treated with 10 μg/mL LPS- and azide-free agonistic αTIGIT (A15153G; BioLegend) or mouse IgG2a isotype control (MOPC-173; BioLegend). Where indicated, cultures were washed twice with media and restimulated with 50 ng/mL PMA (Sigma) and 1 μg/mL Ionomycin (Sigma) in the presence of GolgiStop (BD Biosciences) for 4 h at 37°C, 5% CO₂. Where indicated, conventional CD4⁺ or CD8⁺ T cells or CD25⁺ Treg were purified by MACS according to the manufacturer’s instructions (Miltenyi Biotech).

Alloreactive proliferation assay

One-way mixed lymphocyte reactions (MLRs) were performed using human PBMCs from 8 stimulator-responder pairs. Irradiated stimulator PBMCs (8 × 10⁵) were labeled with CellTrace Violet dye (CTV, ThermoFisher) with 1 μL of 5mM CTV per 10⁷ cells at RT for 20 minutes, and then cultured with responder PBMCs (4 × 10⁵) in culture medium R10 as described above with 10% plasma from the responders for 6 days at 37°C in a 5% CO₂, humidified atmosphere. Where indicated, cultures were treated with 10 μg/mL LPS- and azide-free agonistic αTIGIT or isotype control in the presence or absence of belatacept.

Ex vivo intracellular cytokine staining

Cells were cultured in round-bottom 96-well plates in R10 media. PMA/Ionomycin and Golgi Stop were also added for 4 hours as described above at 37°C, 5% CO₂. Intracellular staining was performed after fixation and permeabilization using FOXP3 Staining Buffer Set (eBioscience) according to manufacturer’s instructions, utilizing fluorophore-labeled antibodies to Ki-67-FITC (BD Biosciences), Foxp3-APC (eBioscience), IL-2-BV605 (BioLegend), IFN-γ-BV785 (BioLgend), TNF-BV650 (BioLegend), IL-17-APC-R700 (BD Biosciences) following extracellular staining as described above. Samples were acquired on a Fortessa flow cytometer (BD Biosciences), and data were analyzed using FlowJo software (9.9.6 FlowJo, LLC) and GraphPad Prism Version 7.

Statistics

Data shown and described depict the mean ± SEM. T cell responses were analyzed using paired, nonparametric, Wilcoxon matched-pairs signed-rank test. All statistical analyses were conducted using GraphPad Prism 7.0 software (San Diego, CA). Significance was determined as *p < 0.05, **p < 0.005. All calculated p values were two-tailed analyses.

Results

CD4⁺ CD28⁺ T_EM subset exhibits high proliferative capacity and cytokine functionality and low frequencies of apoptotic cells following stimulation with alloantigen

To compare phenotypic and functional characteristics of the three “risky” memory T cell subsets associated with belatacept-resistant rejection, human PBMCs were isolated from healthy volunteers, ranging in age from 28 to 63. Responder PBMCs were cultured alone or in the presence of irradiated, CTV-labeled, allogeneic stimulator PBMCs. Subsets were analyzed for proliferation, apoptosis, and cytokine production using flow cytometry. Gating strategies to identify the “risky” memory T cell subsets previously identified as being associated with belatacept-resistant rejection are depicted in Supplemental Figure 1. As depicted in Figure 1A, the CD4⁺ CD28⁺ T_EM subset contained higher frequencies of alloantigen-driven Ki-67⁺ proliferating cells upon allogeneic stimulation as compared to the CD8⁺ CD28^null subset (the CD4⁺ CD57⁺ PD1⁻ subset exhibited a similar trend). We next queried the frequencies of apoptotic cells within risky memory T cell subsets in allogeneic co-cultures. Within the responder alone cultures, the CD4⁺ CD57⁺ PD1⁻ subset contained a significantly higher frequency of Caspase 3/7⁺ 7-AAD⁺ apoptotic cells following ex vivo culture relative to the other two subsets (Figure 1B). Following culture with allogeneic stimulators, both the CD4⁺ CD57⁺ PD1⁻ subset and the CD8⁺ CD28^nul subset contained significantly increased frequencies of Caspase 3/7⁺ 7-AAD⁺ apoptotic cells relative to the CD4⁺ CD28⁺ T_EM subset (Figure 1B). The cytokine-producing functionalities of three risky memory T cell subsets were then evaluated. CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets yielded increased frequencies of TNF and IFN-γ producers compared to CD8⁺ CD28^null subset (Figure 1C). Results for polyclonal stimulations using anti-CD3/CD28 beads as described in the Materials and Methods exhibited similar trends (Supplemental Figure 2). Collectively, these data demonstrate that the CD4⁺ CD28⁺ T_EM subset exhibits the most proliferative capacity combined with the least death relative to the other two subsets, while retaining strong functional capacity.

Figure 1. CD4⁺ CD28⁺ T_EM subset exhibits high proliferative capacity and cytokine functionality and low frequencies of apoptotic cells following stimulation with alloantigen.

Responder PBMCs were stimulated in the presence or absence of irradiated, CTV labeled, allogeneic stimulator PBMCs. Risky memory T cell subsets were defined by the following gating strategy: CD4⁺ CD28⁺ T_EM, CD3⁺ CD4⁺ CCR7⁻ CD45RA⁻ CD28⁺; CD8⁺ CD28^null, CD3⁺ CD8⁺ CD28⁻; CD4⁺ CD57⁺ PD1⁻, CD3⁺ CD4⁺ CD57⁺ PD1⁻. (A) Representative flow plots and summary data of frequencies of Ki-67⁺ proliferating cells within risky memory T cell subsets in the presence or absence of allogeneic stimulation. (*, p < 0.05; data depicted are from six independent stimulator/responder pairs). (B) Representative flow plots and summary data of frequencies of Caspase3/7⁺ 7-AAD⁺ apoptotic cells within risky memory T cell subsets in the presence or absence of allogeneic stimulation (**, p < 0.01; *, p < 0.05; data depicted are from eight independent stimulator/responder pairs). (C) Representative flow plots and summary data of frequencies of TNF-, and IFN-γ-secreting cells within risky memory T cell subsets in the presence or absence of allogeneic stimulation (*, p < 0.05; data depicted are from six independent stimulator/responder pairs).

CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets exhibit high CD69 upregulation and CD4⁺ CD28⁺ T_EM exhibit high CD38 and HLA-DR upregulation following stimulation

We next sought to evaluate the expression profile of activation markers on risky memory T cell subsets. CD69, one of the earliest T cell activation markers, was significantly increased within CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets upon stimulation (Figure 2A–B). In contrast, expression of the activation markers CD38 and HLA-DR were upregulated on CD4⁺ CD28⁺ T_EM subset but not on CD4⁺CD57⁺PD-1⁻ or CD8⁺ CD28^null cells following stimulation (Figure 2C–D).

Figure 2. CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets exhibit high CD69 upregulation and CD4⁺ CD28⁺ T_EM exhibit high CD38 and HLA-DR upregulation following stimulation.

Human PBMCs from healthy donors were stimulated ex vivo with anti-CD3/CD28 beads for 3 d followed by brief restimulation in the presence or absence of PMA/Iono. PBMCs were analyzed directly ex vivo or following in vitro stimulated as described above by flow cytometry. (A) Representative flow plots of frequencies of CD69⁺ cells within risky memory T cell subsets pre-(top) and post-(bottom) stimulation. (B) Summary data of fold changes of frequencies of CD69⁺ cells within risky memory T cell subsets upon stimulation. (C) Representative flow plots of frequencies of CD38⁺ and HLA-DR⁺ cells within risky memory T cell subsets post-stimulation. (D) Summary data of fold changes of frequencies of CD38⁺ and HLA-DR⁺ cells within risky memory T cell subsets upon stimulation. (E) Representative viSNE plots of the median fluorescence intensity (MFI) of CD57, PD-1, CD28, CCR7 and CD45RA within the CD4⁺ T cell compartment. Red indicates high expression of a given marker, and blue represents low intensity. The red-rimmed population represents CD4⁺ CD28⁺ T_EM subset, and the purple-rimmed population indicates CD4⁺ CD57⁺ PD1⁻ subset. (F) Summary data of frequencies of CD57⁺, PD-1⁺ and CD28⁺ cells within risky memory T cell subsets (*, p < 0.05; n = 6 per experiment; data are representative of three independent experiments. Risky memory T cell subsets were defined by the following gating strategy: CD4⁺ CD28⁺ T_EM, CD3⁺ CD4⁺ CCR7⁻ CD45RA⁻ CD28⁺; CD8⁺ CD28^null, CD3⁺ CD8⁺ CD28⁻; CD4⁺ CD57⁺ PD1⁻, CD3⁺ CD4⁺ CD57⁺ PD1⁻; CD4⁺ CD28⁺ T_EMRA, CD3⁺ CD4⁺ CCR7⁻ CD45RA⁺ CD28⁺; CD8⁺ CD28⁺ T_EM, CD3⁺ CD8⁺ CCR7⁻ CD45RA⁻ CD28⁺; CD8⁺ CD28⁺ T_EMRA, CD3⁺ CD8⁺ CCR7⁻ CD45RA⁺ CD28⁺). G, Frequencies of CD57⁺ PD-1⁻ cells within the CD4⁺ CD28⁺ T_EM subset. H, Frequencies of CD28⁺ T_EM cells within the CD57⁺ PD-1⁻ subset.

We hypothesized that there might be potential overlap between these risky memory T cell subsets. To investigate this possibility, we utilized viSNE analysis from the Cytobank platform (Cytobank.org). viSNE is an algorithm which employs t-stochastic neighbor embedding (t-SNE) to reduce high-dimensional cytometry data down to a two-dimensional map for ease of visualization, where the distance between cells corresponds to their marker profile similarity. Color is then used as a third dimension to interactively visualize features of these cells (36). When the profiles of the gated CD4⁺ T cells were visualized by viSNE, the resulting viSNE maps clearly distinguished CD4⁺ CD28⁺ T_EM vs. CD4⁺ CD57⁺ PD1⁻ populations in space (depicted as red-rimmed and the purple-rimmed islands, respectively, in Figure 2E). Interestingly, results demonstrated minimal overlap between the two populations. Summary data of CD57, PD-1 and CD28 expression on the three memory T cell subsets revealed that CD4⁺ CD28⁺ T_EM subset exhibited the lowest expression of CD57 (less than 10%), and the highest expression of PD-1 (roughly 40%) among all the risky memory T cell subsets (Figure 2F). We also found that CD4⁺ CD57⁺ PD1⁻ subset was composed predominantly of T_EM and T_EMRA subsets (Figure 2E) and the CD28 expression on CD4⁺ CD57⁺ PD1⁻ subset was extremely variable (from 3% to 97%) (Figure 2F). In directly analyzing the overlap between the subsets, we found that only ~0.6% of CD4⁺ CD4⁺ CD28⁺ T_EM were CD57⁺ PD-1⁻ cells. Likewise, ~14% of CD4⁺ CD57⁺ PD-1⁻ cells were CD4⁺ CD28⁺ T_EM (Figure 2G–H). Thus, CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets exhibit minimal overlap, and thus they may play distinct and non-redundant roles in mediating belatacept-resistant rejection.

TIGIT expression is maintained on “risky” memory T cell subsets in vivo and in vitro in the presence of belatacept

Given the finding that these T cell populations may be non-overlapping and therefore play non-redundant roles in mediating belatacept-resistant rejection, we aimed to identify coinhibitory receptors expressed on all three subsets that could potentially be targeted to limit their response following transplantation. As depicted in Figure 3A–B, all three memory T cell subsets expressed the coinhibitory molecule TIGIT. While TIGIT expression increased on all subsets following stimulation (Supplemental Figure 3A), the CD8⁺ CD28^null subset exhibited the highest level of TIGIT expression both pre- and post- anti-CD3 stimulation as compared to the CD4⁺ CD28⁺ T_EM or CD4⁺ CD57⁺ PD1⁻ subsets. The expression of TIGIT on these three risky memory T cell subsets in comparison to naïve, central memory, and effector memory CD4⁺ and CD8⁺ T cell subsets is shown in Supplemental Figure 3B–C. In contrast, the CD8⁺ CD28^null subset showed significantly lower expression of the costimulatory counter-receptor DNAM in both pre- and post-stimulation samples relative to the other two subsets (Figure 3C).

Figure 3. TIGIT expression is maintained on “risky” memory T cell subsets in vivo and in vitro in the presence of belatacept.

(A) Schematic of experimental design. Peripheral blood T cells from healthy donors were stimulated ex vivo with anti-CD3 in the presence or absence of belatacept. Risky memory T cell subsets were defined as in Figure 1. Representative flow plots and summary data of frequencies of TIGIT⁺ cells (B) and DNAM⁺ cells (C) within risky memory T cell subsets pre- and post-stimulation. TIGIT (D) and DNAM (E) expressions were assessed on risky memory T cell subsets in the setting of belatacept (*, p < 0.05; n = 6 per experiment; data are representative of three independent experiments). (F) Schematic of experimental design. Peripheral T cells from renal transplant patients treated with belatacept as described in (8) were analyzed directly at baseline or following transplantation. (G) TIGIT expression was assessed on risky memory T cell subsets in vivo belatacept-treated transplant patients (*, p < 0.05; n = 5-7 per experiment; data are representative of four independent experiments).

We next investigated TIGIT expression on memory T cell subsets following stimulation in the setting of belatacept. Following polyclonal stimulation with plate-bound anti-CD3 and CD80/CD86-expressing APC in the presence of belatacept, TIGIT and DNAM were both still expressed on all three risky memory T cell subsets (Figure 3D–E, Supplemental Figure 3D). In the presence of belatacept, there were no statistically significant differences in TIGIT expression between the three subsets. Intriguingly, belatacept resulted in a statistically significant increase in TIGIT expression on FOXP3⁺ Treg (Supplemental Figure 3E). These data therefore suggest that TIGIT could be a therapeutic target on all three risky memory T cell subsets in the context of belatacept therapy.

To assess the in vivo expression patterns of TIGIT and DNAM in renal transplant patients treated with belatacept, peripheral T cells from renal transplant recipients treated with belatacept (as described in (8)) were analyzed directly at baseline or following transplantation (Figure 3F–G). Consistent with the in vitro data, all three memory T cell subsets isolated ex vivo from belatacept-treated renal transplant recipients expressed TIGIT and DNAM. These results suggest that while expression patterns of TIGIT and DNAM may indicate the CD8⁺ CD28^null subset is more exhausted and/or senescent relative to the other two subsets, TIGIT is still maintained on all the three risky memory T cell subsets in vivo and in vitro in the presence of belatacept.

Agonistic αTIGIT induces apoptosis in all three “risky” memory T cell subsets both in the presence and absence of belatacept

We next sought to determine whether agonism of the TIGIT pathway could inhibit responsiveness of these “risky” memory T cell subsets associated with belatacept-resistant rejection. Responder PBMCs were stimulated ex vivo with allogeneic antigens and treated with a TIGIT-specific agonistic monoclonal antibody (clone: A15153G) or isotype control. Relative to isotype control, there were no differences in the frequency of Ki-67⁺ proliferating cells among any of the three subsets in the presence or absence of belatacept (Supplemental Figure 4A), indicating that TIGIT agonism does not inhibit proliferation of these memory T cell subsets. Similarly, TIGIT agonism failed to modulate TNF, IFN-γ and IL-2 production on risky memory T cell subsets (Supplemental Figure 4B–D). However, as depicted in Figure 4, significant increases in apoptosis as measured by Caspase 3/7⁺ 7-AAD⁺ double positive cells were observed in the presence of agonistic αTIGIT in all the three risky memory T cell subsets, indicating that TIGIT agonism results in increased cell death among all three risky memory T cell subsets associated with belatacept-resistant rejection. Indeed, TIGIT agonism in the presence of belatacept also resulted in markedly increased frequencies of active Caspase 3/7⁺ 7-AAD⁺ cells among all three risky memory T cell subsets (Figure 4A–B). These data therefore demonstrate that TIGIT agonism results in the attenuation of risky memory T cell responses by increasing apoptotic cell death.

Figure 4. Agonistic αTIGIT induces apoptosis in all three “risky” memory T cell subsets both in the presence and absence of belatacept.

Peripheral blood T cells from healthy donors were unstimulated or stimulated ex vivo with allogeneic antigens and treated with agonistic αTIGIT or isotype control in the presence or absence of belatacept. Risky memory T cell subsets were defined as in Figure 1. Representative flow plots (A) and summary data (B) of frequencies of Caspase3/7⁺ 7-AAD⁺ apoptotic cells within risky memory T cell subsets (**, p < 0.01; *, p < 0.05; data depicted are from eight independent stimulator/responder pairs; data are representative of two independent experiments).

TIGIT-mediated memory T cell apoptosis requires FOXP3⁺ Treg

As shown above, TIGIT agonism resulted in increased apoptosis in belatacept-resistant CD4⁺ and CD8⁺ memory T cell responses. Because these cultures were performed using PBMC that contained T_reg, and because Treg contain high frequencies of TIG IT⁺ cells (Supplemental Figure 5A), it was unclear whether the effect of TIGIT agonism on memory T cell apoptosis was a cell-autonomous effect, or due to TIGIT agonism-mediated activation of FOXP3⁺ T_reg in these cultures (27, 32). To determine the ability of TIGIT agonism to induce risky memory T cell apoptosis in a cell-autonomous manner, CD25⁻ CD4⁺ T_conv and CD8⁺ T cells were sorted such that all cultures were devoid of FOXP3⁺ Treg, and stimulated ex vivo with anti-CD3/CD28 beads in the presence or absence of agonistic αTIGIT (Figure 5A). Analysis of Caspase 3/7 and 7-AAD staining revealed that TIGIT agonism was ineffective at inducing apoptosis in sorted risky memory T cell subsets in the absence of FOXP3⁺ regulatory T cells (Figure 5B–C), compared to the effect of TIGIT agonism when FOXP3⁺ Treg were present (Figure 4). Moreover, “adding back” increasing numbers of FOXP3⁺ Treg to purified CD4⁺ or CD8⁺ memory T cell populations resulted in increased frequencies of apoptotic cells in cultures treated with the anti-TIGIT agonist, but not cultures treated with isotype control (Supplemental Figure 5B–E). These data therefore demonstrate that although all memory T cell subsets express TIGIT, the presence of FOXP3⁺ Treg is required for the TIGIT agonist to mediate its suppressive effect on these cells.

Figure 5. TIGIT-mediated memory T cell apoptosis requires FOXP3⁺ Treg.

CD8⁺ T cells and CD4⁺ T_conv cells were sorted and stimulated ex vivo with anti-CD3/CD28 beads in the presence of agonistic αTIGIT or isotype control for 3 d. (A) Representative flow plots of the gating strategy to identify sorted CD8⁺ T cells (top) and CD4⁺ T_conv cells (bottom, CD3⁺CD4⁺Foxp3⁻). (B) Representative flow plots and (C) Summary data of frequencies of Caspase3/7⁺ 7-AAD⁺ apoptotic cells within risky memory T cell subsets in the presence of agonistic αTIGIT or isotype control following ex vivo stimulation (*, p < 0.05; n = 4-5 per experiment; data are representative of three independent experiments).

Discussion

Identifying strategies to overcome belatacept-resistant rejection may allow more renal transplant recipients to benefit from the improved toxicity profile of belatacept relative to standard CNI-based immunosuppression (1, 2, 5). In the present study, we show that TIGIT is expressed on three risky memory T cell subsets that have been previously implicated in belatacept-resistant rejection (15, 18, 19), and that TIGIT agonism functions to inhibit human risky memory T cell responses in the presence of belatacept by inducing apoptotic cell death. The finding that TIGIT coinhibition may inhibit memory T cell alloreactivity by exclusively increasing apoptosis (as opposed to limiting cellular proliferation or effector function) is somewhat unique among T cell coinhibitory pathways. For example, CTLA-4 coinhibitory signaling has been identified to regulate cell cycle progression, but does not effectively impact cell death (37–39). Engagement of PD-1 signaling can block T cell proliferation, cytokine production and cytolytic function, and also impair T cell survival (40, 41). PD-1 coinhibition is also known to inhibit cytokine production to a greater extent than cell proliferation. Here, our study provides insight into the biologic role of TIGIT agonism in modulating risky memory T cell response and demonstrates that TIGIT signaling regulates risky memory T cell apoptotic cell death, but does not differentially affect the proliferation and effector function of risky memory T cell subsets.

Our results further demonstrate that the ability of TIGIT agonism to induce risky memory T cell apoptosis requires FOXP3⁺ Treg. There are two main paradigms to explain these findings. First, it is possible that the TIGIT agonist induces a negative signal directly into the memory T cell but requires some factor from FOXP3⁺ Treg, such as IL-10, TGF-beta, or even the removal of CD28-mediated costimulation, to push memory T cells to undergo apoptosis. The second possibility is that the TIGIT expressed on memory T cells themselves is not at all required for apoptosis, but that agonism of TIGIT on FOXP3⁺ Treg results in Treg-mediated induction of apoptosis of memory T cells. The main differentiating factor between these two scenarios is the requirement for TIGIT to be expressed on the memory T cell in order for the TIGIT agonist to mediate the apoptosis-inducing effect. Evidence supporting both paradigms exists in the literature. For example, Chauvin et al. found that TIGIT was upregulated on CD8⁺ tumor-infiltrating lymphocytes (TIL), and in vitro culture experiments suggested that TIGIT blockade was functioning directly on effector/ memory T cells in a cell intrinsic manner, and not through the activation or stabilization of Foxp3⁺ Treg (33). Joller et al. reported similar findings in in vitro cultures of CNS-specific memory T cells (42) and Josefsson et al. showed that TIGIT expression correlated with dysfunctional TCR signaling in effector/memory CD8⁺ T cells in human anti-tumor responses. On the other hand, a myriad of studies have identified an indirect role for TIGIT in inhibiting effector T cell responses, via an effect on a Treg intermediate. For instance, TIGIT agonism has been shown to augment Foxp3⁺ Treg suppressive function (27), and TIGIT⁺ T cells can induce immunoregulatory DC via ligation of CD155 (24, 30, 43). Conditional knockout studies in mice, or knockdown studies of human T cells, will be required to definitively determine the cell autonomous role of TIGIT signaling on belatacept-resistant memory T cell subsets.

Data presented here also depict a direct comparison of the phenotypic and functional profiles of the three memory T cell subsets known to be associated with belatacept-resistant rejection. These analyses show that the CD4⁺ CD28⁺ T_EM subset was on par with CD4⁺ CD57⁺ PD1⁻ subset in terms of proliferative capacity, activation (CD69) and pro-inflammatory cytokine production (TNF, IFN-γ, IL-2) upon stimulation. Likewise, similar expression patterns of TIGIT and DNAM were noted on CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets. Thus, we conclude that the CD4⁺ CD28⁺ T_EM and CD4⁺ CD57⁺ PD1⁻ subsets are more similar to each other than either is to the CD8⁺ CD28^null subset. While these subsets may be functionally similar, viSNE analysis showed that the cell populations themselves are non-overlapping, raising the possibility that they could play non-redundant roles in mediating belatacept-resistant rejection. Importantly, while CD4⁺ CD57⁺ PD1⁻ cells readily underwent apoptosis following allogeneic stimulation, CD4⁺ CD28⁺ T_EM cells were relatively resistant to apoptosis. These data suggest that increasing apoptosis of CD4⁺ CD28⁺ T_EM cells, potentially through the TIGIT pathway, could be an effective method to control their response.

To our knowledge, there is only one report comparing these three signatures of risky memory T cell subsets in terms of their association with belatacept-resistant rejection (44). As part of a clinical trial, pretransplant frequencies of these risky memory T cell subsets in belatacept-treated renal transplant patients (n=20) were analyzed. Only the pre-transplant frequency of CD8⁺ CD28⁺⁺ T_EMRA trended toward an increase in patients who underwent belatacept-resistant rejection as compared to those that remained stable, suggesting that this putative pre-transplant cellular biomarker warrants further investigation. Moreover, continuing to amass more subjects to understand the range of prevalence of these risky memory T cell subsets in both transplant candidates and normal controls is an important goal. In addition, it will be important to understand if these frequencies are dynamic over time in both healthy controls and transplant patients, in order to determine the stability of this potential biomarker. While this study examined the association between belatacept-resistant rejection and CD8⁺ CD28⁺ T_EMRA instead of CD4⁺ CD28⁺ T_EMRA, it is worth noting that we also showed that TIGIT agonism induced apoptotic cell death of CD8⁺ CD28⁺ both T_EMRA and T_EM subsets (Supplemental Figure 4). While we have certainly identified differences between the three human memory T cell subsets associated with belatacept-rejection, it is also interesting that all three subsets behaved similarly in response to TIGIT-mediated inhibition: that is, TIGIT coinhibition induced cell death but did not inhibit proliferation or effector function in all three subsets, and in all three subsets were FOXP3⁺ Treg required for the apoptosis-inducing effects of TIGIT coinhibition. These results demonstrate that despite their phenotypic and functional differences, TIGIT coinhibition exacts a similar impact on all memory T cell subsets associated with belatacept-resistant rejection, suggesting that therapeutic targeting of this pathway would have a rather homogeneous and predictable effect on belatacept-resistant memory T cell responses.

In conclusion, agonism of the TIGIT coinhibitory pathway regulates the apoptosis of risky memory T cell subsets associated with belatacept-resistant rejection. These data increase our fundamental understanding of the role of TIGIT signaling on immunity to allografts, and offer novel insight to inform and direct the translation of TIGIT agonism as a potential new therapeutic target in clinical transplantation.

Abbreviations

APC

antigen-presenting cell

DC

dendritic cell

DNAM

DNAX accessory molecule

FBS

fetal bovine serum

IFN-γ

interferon-γ

IL-2

interleukin-2

IL-17

interleukin-17

PBMC

peripheral blood mononuclear cell

PMA

phorbol 12-myristate 13-acetate

T_CM

central memory T cell

T_EM

effector memory T cell

T_EMRA

effector memory RA T cell

TIGIT

T cell immunoreceptor with Ig and ITIM domains

TNF

tumor necrosis factor

T_reg

regulatory T cell

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon request.