Engineering synthetic suppressor T cells that execute locally targeted immunoprotective programs

Abstract

Immune homeostasis requires a balance of inflammatory and suppressive activities. To design cells potentially useful for local immune suppression, we engineered conventional CD4⁺ T cells with synthetic Notch (synNotch) receptors driving antigen-triggered production of anti-inflammatory payloads. Screening a diverse library of suppression programs, we observed the strongest suppression of cytotoxic T cell attack by the production of both anti-inflammatory factors (interleukin-10, transforming growth factor-β1, programmed death ligand 1) and sinks for proinflammatory cytokines (interleukin-2 receptor subunit CD25). Engineered cells with bespoke regulatory programs protected tissues from immune attack without systemic suppression. Synthetic suppressor T cells protected transplanted beta cell organoids from cytotoxic T cells. They also protected specific tissues from unwanted chimeric antigen receptor (CAR) T cell cross-reaction. Synthetic suppressor T cells are a customizable platform to potentially treat autoimmune diseases, organ rejection, and CAR T cell toxicities with spatial precision.

Fig. 1.. Engineering synthetic suppressor T cells that drive antigen-induced production of immune suppressive payloads.

(A) Design of synthetic suppressor T cells that inducibly produce anti-inflammatory payloads. These are human conventional CD4⁺ T cells engineered to express a synNotch receptor that triggers the expression of a custom suppressive payload upon target antigen binding. Three-cell coculture was used to assess the ability of engineered suppressor cells to block CAR T cell proliferation and target cell killing in vitro. TF, transcription factor. (B) Synthetic suppressor T cells reduced proliferation of CAR T cells in vitro. As described in (A), fold proliferation of CD4⁺ and CD8⁺ CAR T cell over 72 hours in the presence of synthetic suppressor T cells with synNotch-induced individual payloads is shown. Fold change normalized to the 0 hour time point (n = 3 replicates, error bars = standard error). Dashed line indicates no payload suppressor T cell control. Statistical significance was tested using a two-tailed Student’s t test comparing to no suppressor T cell control (*P < 0.05). (C) Combinations of synNotch-induced payloads drove stronger suppression of CAR T cells in vitro. The fold proliferation of K562 target cells (Her2⁺, CD19⁺) and CAR T cells over 72 hours is shown for both cocultures of suppressor T cells and target cells with CD4⁺ or CD8⁺ CAR T cells. Each point indicates a pairwise combination of payloads from the library in (A) induced by anti-CD19 synNotch suppressor cells (mean, n = 3 replicates). Fold change normalized to the 0 hour time point. Gray point indicates the no-payload suppressor T cell control. See fig. S2, A and B, for data in (C) as a heatmap of combinatorial payloads. See fig. S3, A and B, for similar analysis with polyclonal T cells stimulated through their endogenous TCR.

Fig. 2.. Combinatorial induction of both CD25 and TGFβ1 by the same suppressor cell leads to more effective suppression of CAR T cells in vitro.

(A) Synthetic suppressor T cells that act as a source for inhibitory cytokines and a sink for inflammatory cytokines drove stronger suppression of CAR T cells in vitro. Synthetic suppressor T cells that induced a combination of TGFβ1 and CD25 were more potent at suppressing CD8⁺ CAR T cell activity compared with each individual payload alone. Cell counts are normalized to the 0 hour time point (n = 3 replicates, error bars = standard error, filled markers indicate two-tailed t test, P < 0.05, comparison to no-suppressor cell control). (B) Synthetic suppressor T cells depleted IL-2 produced by activated CD4⁺ T cells in vitro. Human CD4⁺ T cells activated by anti-CD3/CD28 beads for 24 hours were cocultured with synthetic suppressor T cells activated with synNotch activating beads (anti-Myc beads). The IL-2 levels in the supernatant were measured by ELISA (t = 48 hours, n = 3 replicates, error bars = standard error, two-tailed t test comparing TGFβ1 and CD25 to each payload alone, *P < 0.05). (C) Synthetic suppressor T cells required both TGFβ1 and CD25 to be produced by the same cell for effective suppression in vitro. Separation of TGFβ1 and CD25 into two separate cells led to weaker suppression of CD8⁺ CAR T cell killing (reduced target-cell proliferation) than a one-cell system where both payloads are produced by the same suppressor T cell in vitro (n = 3 replicates, error bars = standard error, filled markers indicate two-tailed t test, P < 0.05, comparison to no-suppressor cell control). (D) CD25 drives increased TGFβ1 production by synthetic suppressor T cells in vitro. Suppressor cells that induced a combination of TGFβ1 and CD25 led to more TGFβ1 accumulation than suppressor cells inducing TGFβ1 alone. Suppressor cells were activated in vitro with synNotch activation beads (anti-Myc beads). TGFβ1 levels were measured by ELISA of supernatant (t = 72 hours, n = 3 replicates, error bars = standard error, two-tailed t test between TGF β1 circuit with and without CD25, *P < 0.05). (E) CD25 can enhance suppressor cell activity by two mechanisms. CD25 depletes IL-2 from the local microenvironment and drives preferential proliferation of suppressor cells. An increase in suppressor cell number can yield higher TGFβ1 accumulation.

Fig. 3.. Synthetic suppressor cells block CAR T cell killing in vivo in locally targeted manner.

(A) Two-tumor mouse model was used to assess local immune suppression. Two tumors were injected subcutaneously into immunocompromised NSG mice, such that the right flank had a dual-antigen tumor (Her2⁺ CD19⁺ K562 tumor) and the left flank had a single-antigen tumor (Her2⁺ K562 tumor). Anti-Her2 CAR T cells and anti-CD19 synNotch suppressor T cells were injected intravenously. Tumor volumes were measured by calipers. (B) Synthetic suppressor T cells can block CAR T cell killing locally without systemic suppression. Suppressor T cells (anti-CD19 synNotch→TGFβ1+CD25) are effective at blocking CAR T cell killing of the dual-antigen tumor (CD19⁺) without compromising killing of the single antigen tumor (CD19⁻). Suppressor cells producing each payload alone were not sufficient to protect the dual-antigen tumor from CAR T cell killing. Tumor measurements shown as time after T cell injection (n = 5 replicates, solid line = mean, shading = standard error, two-tailed t test, *P < 0.001 on day 28). Dashed gray line indicates tumor growth with no T cell injection. See fig. S7 for tumor growth curves for individual mice. (C) Synthetic suppressor T cells reduced CAR T cell proliferation in dual-antigen tumor in vivo. Flow profiling of isolated tumors at day 14 showed reduced accumulation of both CD4⁺ and CD8⁺ CAR T cells (GFP⁺) and an increased accumulation of suppressor cells (BFP⁺) in the dual-antigen tumor. Cell counts normalized to tumor weight after isolation (n = 3 replicates, error bars = standard error, two-tailed t test, *P < 0.05). (D) Multicellular NOT gate tumor-killing circuit combining CAR T cells and synthetic suppressor T cells drove robust local suppression. Multicellular NOT gate circuit leads to more-robust local suppression than iCAR NOT circuit in two-tumor model in vivo (25). iCAR NOT gate circuit (anti-Her2 CAR + anti-CD19 PD-1 iCAR) fails to block killing of the dual-antigen tumor. In the multicellular NOT gate tumor-killing circuit, anti-Her2 CAR T cells recognize and kill both tumors, whereas anti-CD19 synthetic suppressor T cells block killing in the CD19⁺ dual-antigen tumor (n = 5 replicates, solid line = mean, shading = standard error, two-tailed t test, *P < 0.001 day 21). Dashed gray line indicates tumor growth with no T cell injection. Additional replicates shown in fig. S9A, and tumor growth curves for individual mice shown in fig. S9B.

Fig. 4.. Synthetic suppressor cells protect beta cells from T cell–mediated destruction in vitro.

(A) eBC organoids were generated from hPSCs. eBC organoids were differentiated from hPSCs as previously described (31). eBC organoids were engineered to express model antigen CD19 by lentiviral transduction on day 19 of differentiation. eBC organoids were HLA-A2⁺ and expressed GFP under the control of the insulin promoter. Confocal microscopy (maximum projection) of an eBC is shown on day 23 of differentiation. Coculture with T cells was performed on day 26 of differentiation. Scale bar, 100 μm. (B) Cytotoxic T cells can kill eBC organoids. Human anti-HLA-A2 CAR CD8⁺ T cells cocultured with HLA-A2⁺ eBC organoids effectively killed eBC organoids in vitro. Confocal microscopy (maximum projection) showed eBC organoid destruction mediated by CAR T cells in vitro after 48 hours (n = 3 replicates, error bars = standard error). (C) Synthetic suppressor T cells protected beta cells from cytotoxic T cell killing. eBC organoids were cocultured with T cells as in (B). Anti-HLA-A2 CAR T cell killing of eBCs was blocked by synthetic suppressor T cells (anti-CD19 synNotch→TGFβ1+CD25 circuit) but not by no-payload control cells (anti-CD19 synNotch→mCherry). Dashed lines indicate CAR-only control (blue) and no T cell control (gray) (n = 3 replicates, error bars = standard error, two-tailed t test, *P < 0.001 at 70 hours comparing control cells to suppressor cells). Confocal microscopy (maximum projection) shows protection of an eBC organoid with suppressor T cells. Caspase 3/7 dye was used to label apoptotic cells and imaged (maximum projection) at the 48-hour time point. (D) Synthetic suppressor T cells self-organized around cytotoxic T cells during suppression in vitro. Suppressor T cells spatially self-organized around individual activated CAR T cells during suppression (t = 48 hours), blocking the formation of CAR T cell clustering that is normally observed in target killing in the absence of suppression. Scale bars, 100 μm (zoomed-out image) and 25 μm (zoomed-in image).

Fig. 5.. Synthetic suppressor cells locally protect hPSC-derived beta cell transplants from T cell–mediated killing in vivo.

(A) Transplant rejection was modeled by cytotoxic T cell rejection of transplanted eBC organoids under the kidney capsule of immunocompromised NSG mice. Fourteen days after transplantation, T cells were coinjected intravenously. eBC organoids express luciferase, allowing for noninvasive imaging of transplant survival. (B) Synthetic suppressor T cells blocked cytotoxic T cell killing of eBC organoid transplants. Bioluminescence imaging was used to track eBC organoid survival. Human anti-HLA-A2 CAR T cells alone cleared transplants within 2 weeks. However, transplants remained intact when synthetic suppressor T cells (anti-CD19 synNotch→TGFβ1+CD25 circuit) were coinjected along with CAR T cells. (C) Synthetic suppressor T cells protect eBC organoid transplants with synNotch priming antigen (CD19⁺). Survival of CD19⁺ eBC organoid transplants as in (B) is assessed by noninvasive imaging (n = 6 to 8 replicates, two-tailed t test, *P < 0.001 comparing CAR T cell condition with and without suppressor T cells). Increased survival of eBC organoid transplants was observed with synthetic suppressor T cells (anti-CD19 synNotch→TGFβ1+CD25 circuit), but all transplants were cleared by anti-HLA-A2 CAR T cells alone. Dashed line indicates no–T cell control (n = 3 replicates, mean). (D) Synthetic suppressor T cells did not protect eBC organoid transplants that lack the synNotch priming antigen. Survival of CD19⁻ eBC organoid transplants is assessed as in (B). No survival advantage was observed in the presence or absence of suppressor T cells in all cases (n = 5 replicates, two-tailed t test, *P < 0.001 comparing CAR T cell condition with and without suppressor T cells). Dashed line indicates no–T cell control (n = 3 replicates, mean). (E) Transplanted eBC organoids (CD19⁺) maintain their structure in the presence of synthetic suppressor T cells but are cleared by CAR T cells alone. eBC organoids were transplanted as in (B). Anti-human CD19 staining was used to identify transplanted eBC organoids in isolated mouse kidneys from transplanted mice 5 days after T cell injection. Staining shows survival of transplants in no–T cell control and CAR T cell in the presence of suppressor cells. Minimal human CD19 staining was observed in the CAR T cell–only condition. Scale bars, 100 μm (zoomed-in images) and 500 μm (zoomed-out image). See fig. S12A for anti-human CD19 and insulin staining of adjacent tissue section. (F) Transplanted eBC organoids retain endocrine function after synthetic suppressor T cell protection. Glucose challenge test was performed on NSG mice with eBC organoid transplants 21 days after injection of T cells (35 days after transplantation). Human C-peptide during fasting conditions and 30 min after intraperitoneal glucose injection (n = 3 or 4 replicates, error bars = standard error) was measured by ELISA of blood serum. Glucose challenge showed that eBC organoids in mice injected with synthetic suppressor T cells remain functional and can secrete human C-peptide after glucose stimulation. P = 0.0018, two-tailed t test between CAR T cells with and without suppressor cells after glucose injection.

Fig. 6.. Potential application of synthetic suppressor cells for local immune protection.

(A) Synthetic suppressor T cells could act as NOT gates to block off-target CAR T cell toxicity in cross-reactive tissues without blocking on-target tumor killing. Suppressor T cells could be directed to off-target tissue (nontumor) using a healthy tissue–specific synNotch to block cytotoxic T cell activity. (B) Synthetic suppressor T cells could recognize allogeneic transplants and locally suppress rejection by host immune cells. Local recognition of transplants by suppressor T cells could remodel the transplant microenvironment to improve transplant survival without systemic immunosuppression. (C) Synthetic suppressor T cells could locally block autoimmune destruction of tissues (e.g., type 1 diabetes, multiple sclerosis). Suppressor T cells that are directed to protect a target tissue using a tissue-specific synNotch could act locally to prevent or treat autoimmunity.