Establishing reverse chimeric antigen receptor T cells for precise targeting of immune-mediated thrombotic thrombocytopenic purpura

Abstract

Immune-mediated thrombotic thrombocytopenic purpura (iTTP) is a life-threatening autoimmune disorder caused by autoantibodies that target the metalloprotease ADAMTS13. This prevents the cleavage of ultra-large von Willebrand factor (UL-VWF), thereby forming microthrombi in the microvasculature. While current treatments are effective, ADAMTS13 relapses occur in up to 15% of the patients after rituximab administration. In this study, we propose a novel approach addressing the unmet need for treatment of patients with ADAMTS13 refractoriness and for patients requiring repeated treatment. Our strategy involves developing two targeted therapies that aim to achieve highly effective and safe B-cell depletion. For the first time, we employ adapter chimeric antigen receptor (CAR) T cells called reverse CAR (RevCAR) T cells to (i) achieve a broader and deeper B-cell depletion by targeting CD19 and (ii) selectively deplete the disease-causing anti-ADAMTS13 autoreactive B cells to increase precision and safety in iTTP treatment. Therefore, we have developed novel target modules (RevTM) against CD19+ and anti-ADAMTS13 autoreactive B cells. We show that the RevCAR T cells successfully and specifically kill target hybridomas and Nalm-6 cells in a RevTM-dependent manner and at low effector- to-target cell ratios. Moreover, the RevCAR T cells are able to eliminate the target cells even in the presence of high concentrations of ADAMTS13 autoantibodies. Overall, our findings demonstrate for the first time, a proof-of-concept that anti-ADAMTS13 autoreactive B cells can be selectively depleted by the RevCAR T-cell platform paving the way for an effective and safe therapy for iTTP.

Introduction

Immune-mediated thrombotic thrombocytopenic purpura (iTTP) is a rare and life-threatening autoimmune disease (AID) characterized by thrombotic microangiopathy. The pathophysiology of iTTP is primarily characterized by the presence of autoantibodies against ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type 1 motif, member 13), an enzyme that is important for the proteolytic cleavage of endothelial ultra-large von Willebrand factor (UL-VWF).^1-4 It contains distinct functional domains, namely; metalloprotease (M), disintegrin-like (D), thrombospondin type 1 repeats (T1-T8), a cysteine-rich region (C), spacer (S), and CUB domains (CUB1-CUB2).³ Autoantibody-induced clearance and/or inhibition of ADAMTS13⁵ and resultant accumulation of UL-VWF in the microvasculature,^6,7 causes spontaneous formation of microthrombi, microangiopathic hemolytic anemia, and severe thrombocytopenia in iTTP patients. If left untreated, these conditions can ultimately lead to fatal multiorgan failure. The existing standard-of-care therapeutic regimen involving therapeutic plasma exchange, systemic immunosuppression with corticosteroids and B-cell depleting anti-CD20 rituximab, and caplacizumab,^8-11 has substantially improved survival and acute disease control. However, these approaches do not selectively eliminate the ADAMTS13-specific autoreactive immune clones. Although many patients achieve durable clinical remission, persistence or recurrence of reduced ADAMTS13 activity is still a risk and necessitates long-term monitoring and, in some cases, repeated immunosuppression. Although exact numbers for refractory patients depend on the population under study, there have been reports of incidence of ADAMTS13 refractoriness still being as high as 12.5-17.9%.¹² This highlights the need for autoreactive B-cell receptor (BCR)-directed cellular therapies that may potentially reduce the need for ongoing, non-specific immunosuppression following initial intervention.

Targeted immunotherapies, particularly chimeric antigen receptor (CAR) T cells, which are known for their high success rates in treating B-cell malignancies,^13,14 have gained much attention in AID. α-CD19 CAR T cells have shown promising efficiency in depleting autoreactive B cells in AID.^15,16 However, as a pan-B-cell-depleting therapy, they also attack non-pathogenic B cells, causing lifelong B-cell aplasia, ultimately compromising humoral immunity. More recently, chimeric autoantigen receptor (CAAR) T cells have been developed to selectively target pathogenic B cells while sparing non-pathogenic ones in various AID, such as pemphigus vulgaris, Graves’ disease, and N-methyl-D-aspartate receptor encephalitis.^17,18 Nevertheless, all the conventional CA(A)R T approaches still pose significant risks as they lack direct control mechanisms to rapidly manage side effects, such as cytokine release syndrome (CRS) and neurotoxicity.¹⁹

To facilitate controllability, minimize side effects, and allow for precise targeting of autoreactive B cells in iTTP patients, we here present for the first time a novel switchable adapter CAR system developed based on the reverse CAR (RevCAR) T-cell platform,^20-22 which recently entered first clinical trials for treatment of CD123⁺ hematological malignancies (clinicaltrials gov. Identifier: NCT05949125). This platform is based on the interaction between the RevCAR on T cells and its corresponding adapter molecule, the reverse target module (RevTM) (Figure 1A). Extracellularly, the RevCAR contains a short peptide epitope (E5B9) derived from the human La protein^23-25 and is fused to intracellular CD28/ CD3ξ signaling domains (Online Supplementary Figure S1A). RevCAR T-cell activity is finely regulated by the RevTM, a tunable bispecific adapter that simultaneously binds to the E5B9 epitope on RevCAR T cells and a surface entity on target cells. We here employ the RevCAR system to investigate and compare the efficacy of two different retargeting strategies in Nalm-6 and hybridoma cell models. First, we designed a novel CD19-specific RevTM that allows broader, deeper and temporally controlled pan B-cell depletion. This approach was founded on the established capacity of CD19-directed CAR T cells in achieving effective deep tissue penetration and robust B-cell depletion in malignant settings and other AID. Second, to specifically deplete autoreactive B cells in iTTP via the RevCAR system, a set of ADAMTS13-derived RevTM was developed. This would enable redirection of RevCAR T cells exclusively against disease-causing polyclonal α-ADAMTS13 autoreactive B cells. Both strategies have the potential to also be applied in other B-cell-driven AID. Moreover, this would help illustrate the flexibility of the system in scenarios where specific or broader B-cell targeting may be clinically indicated even beyond iTTP.

Methods

Reverse chimeric antigen receptor T-cell generation

Human peripheral blood mononuclear cells (PBMC) were isolated from healthy donor buffy coats (German Red Cross, Dresden) by Pancoll density centrifugation. Pan-T cells were obtained by negative magnetic selection using the Pan T Cell Isolation Kit (Miltenyi Biotec) and cultured for 3 days in RPMI_complete medium supplemented with interleukin (IL)-2 (50 U/mL). T cells were activated with T-Cell TransAct™ and transduced with lentiviral vectors to generate RevCAR T cells as previously described.²¹ During transduction, cells were maintained in TexMACS medium (Miltenyi Biotec) supplemented with IL-2, -7, and -15 (Miltenyi Biotec). Prior to experiments, RevCAR T cells were transferred to cytokine-deprived RPMI_complete medium for 24 hours (h).

Purification and characterization of reverse target modules

For RevTM purification, 2×10⁶ RevTM-producing 3T3 cells were seeded in a T175 flask (Greiner Bio-One). On day 4, supernatants were harvested and centrifuged at 360×g for 3 minutes (min) at 4°C, then 10,000xg for 20 min at 4°C. Cleared supernatants were purified using a Strep-Tactin®XT 4Flow® column (IBA-Lifesciences) according to the manufacturer’s instructions. Elution fractions were dialyzed in 1×phosphate-buffered saline (PBS).

Cytotoxicity assay

To assess RevCAR T-cell cytotoxicity, Nalm-6 or hybridoma cells were labeled with eFluor™ 670 following the manufacturer’s protocol. Cells were washed with ice-cold PBS, stained for 15 min at 37°C, quenched with cold medium, and washed to remove excess dye. Labeled target cells (2×10⁴) were co-cultured with RevCAR T cells ± RevTM in 96-well plates (200 μL). Target cells alone served as a minimum control. Unless specified, the effector:target (E:T) ratio was 2.5:1 and the RevTM were used at saturating concentrations of 25 nM for CUB and MDTCS RevTM; and 5 nM for fL and α-CD19 RevTM. After incubation, 20 μL was diluted 1:5 in DAPI solution and eFluor™ 670⁺ cells were quantified using a MACSQuant VYB Analyzer. The following equation was used to calculate the percent lysis of target cells:

For autoantibody assays, 5×10⁴ RevCAR T cells and 2×10⁴ target cells were co-cultured ± 25 nM CUB or MDTCS RevTM and 1.6-20 μg/mL autoantibody in a final volume of 200 μL.

Figure 1.Reverse chimeric antigen receptor cell production and reverse target modules binding capacity. (A) Reverse chimeric antigen receptor (RevCAR) T-cell redirection strategy with the 4 candidate reverse target modules (RevTM) to the right. (B, left) RevCAR T-cell transduction efficiency as assessed via the reporter marker enhanced green fluorescent protein (EGFP) (green) overlaid with untransduced control (grey). (B, right) Average percent EGFP of all donor T cells used in the assays. (C, left) RevCAR staining to visualize correlation of RevCAR expression with EGFP. RevCAR-positive histogram peak (red), untransduced control (grey). (C, right) Quantification of RevCAR expression as receptor density per T cell. (D) RevTM binding assessment to RevCAR. (E) RevTM binding to target hybridomas (α-S, top and α-CUB, bottom). (F, G) RevTM binding to target Nalm-6 cells expressing patient-derived II-1 (α-S) (F) and z1-201 (α-CUB) (G) single-chain variable fragment. Overlaid histograms are shown. RMFI: relative median fluorescence intensity (RMFI) was calculated by subtracting the median fluorescense intensity (MFI) of the secondary antibody control from the MFI of the stained sample. Ab: antibody; fL: full-length

Cytokine-release assay

Cell-free supernatants of cytotoxicity assays (see above) were collected after 24 hours (h) to determine the concentration of cytokines released into co-culture supernatants. OptEIA human IL-2, tumor necrosis factor (TNF), interferon y (IFN-y), granulocyte-macrophage colony-stimulating factor (GM-CSF) enzyme-linked immunosorbant assay (ELISA) kits were used according to the manufacturer’s instructions (BD BioSciences). The samples were diluted 1:10 before analysis. Assays were read out using the Infinite M200 pro plate reader and Magellan™ software (TECAN).

Intracellular staining

After 24 h co-incubation of 5×10⁴ RevCAR T cells with 2×10⁴ target cells ± 25nM RevTM, RevCAR T cells were harvested and stained for intracellular granzyme B and perforin as previously described.^26-28 Briefly, cells were first stained extracellularly with a-human CD8 PE-Vio770 (Miltenyi Biotec) and Zombie Red™ (BioLegend), then fixed and permeabilized using the Inside Stain kit (Miltenyi Biotec) according to manufacturer instructions. During permeabilization, 5 µL FcR Blocking Reagent was added to prevent non-specific binding. Cells were then incubated for 30 min with α-human granzyme B-PE (1:20) and α-human Perforin-VioBlue™ antibodies (1:10). Data acquisition was performed on a MACSQuant10 Analyzer and analyzed using MACSQuantify and FlowJo™.

Ethics statement

Ethical approval for research involving human T cells was obtained from the local ethics committee of the Medical Faculty Carl Gustav Carus, Technische Universität Dresden, Germany (EK138042014).

Results

Design of novel reverse target modules for flexible and safe targeting of autoreactive B cells in immunemediated thrombotic thrombocytopenic purpura

To generate targeted therapy for iTTP with increased safety profile, we developed two distinct adapter CAR approaches based on the RevCAR system,²¹,²⁹ by creating RevTM with distinct specificity (Online Supplementary Figure S1B). First, to activate RevCAR T cells for temporally controlled pan-B-cell depletion, an α-CD19 RevTM was generated by fusing the deimmunized single-chain variable fragment (scFv) of the α-CD19 monoclonal antibody (mAb) (clone HD37) to the humanized scFv of the α-La mAb (clone 5B9) via short peptide linkers. Second, a more specific approach was implemented to eliminate only the disease-causing autoreactive B cells. To this end, RevTM were engineered to contain either the full-length ADAMTS13 autoantigen (fL RevTM) (Figure 1A) or highly immunogenic ADAMTS13 sub-domains (MDTCS RevTM and CUB RevTM containing MDTCS or CUB1-2 domains, respectively).^30,31 Since ADAMTS13 has proteolytic activity, an E225Q mutation was introduced into the M domain³¹ to ensure that ADAMTS13 and its fragments within the RevTM are enzymatically inactive. All RevTM are endowed with an N-terminal immunoglobulin (Ig) κ leader and C-terminal purification tags (His, Strep). To produce RevTM, stable eukaryotic expression cell lines were established using 3T3 fibroblast cells. As shown in Online Supplementary Figure S1C, D, the α-CD19 RevTM as well as fL RevTM, MDTCS RevTM and CUB RevTM were successfully purified from cell culture supernatants in sufficient quality and quantity.

Reverse target modules bind successfully to both reverse chimeric antigen receptor T cells and target cells

To characterize the novel α-CD19, fL, MDTCS and CUB RevTM, we first analyzed their binding properties to both RevCAR T cells and target cells.

As shown in Figure 1B and C, human T cells could be successfully transduced with the RevCAR at a high efficiency of ~60%, resulting in an average expression of 5,700 RevCAR per cell. RevTM binding towards RevCAR T cells was assessed using α-His Phycoerythrin (PE) antibody (Ab). Flow cytometric analysis demonstrated comparable binding to EGFP⁺ RevCAR T cells for all novel RevTM (Figure 1D).

To further investigate the binding of the novel RevTM to α-ADAMTS13 B-cell receptor (BCR)⁺ cells and to subsequently test their functional properties, two cell models were chosen: (i) our in-house available hybridoma cell lines that naturally express α-human ADAMTS13 BCR specific against S (mAb clone 15D1)³² or CUB domains (mAb clone 20D2)³³ and (ii) the CD19⁺ human B-cell line Nalm-6, engineered to overexpress patient-derived α-human ADAMTS13 scFv on its surface (termed surrogate decoy receptor [SDR]) (Online Supplementary Figure S2A, C). SDR were designed using the scFv derived from the patient α-ADAMTS13 mAb II-1 and z1-201,^34-37 that are specific for the S and CUB domains of the ADAMTS13 autoantigen, respectively. To create the α-ADAMTS13 SDR, the patient-derived scFv were fused to the CD28 transmembrane domain for anchoring them on the cell surface (Online Supplementary Figure S2C). Extra-cellularly, the receptor is equipped with the E7B6 tag for flow cytometric detection. Nalm-6 cells were engineered to stably overexpress the α-S (Nalm-6 α-S) and α-CUB SDR (Nalm-6 α-CUB) via lentiviral transduction.

As confirmed by flow cytometry, the majority of both hybridomas and engineered Nalm-6 cells express α-human ADAMTS13 BCR and SDR, respectively, on their surface. The receptor densities range from ~10,000 to ~200,000 molecules per cell (Online Supplementary Figure S2), which is consistent with the average number of BCR found on human B cells (15,000 to 100,000).^38,39

As summarized in Figure 1E-G, the α-CD19 RevTM and all ADAMTS13 RevTM showed specific binding to their respective target cells. In comparison to the MDTCS and CUB RevTM, the fL RevTM demonstrated considerably lower binding, particularly toward hybridoma cells, as indicated by reduced median fluorescense intensity (MFI) values.

Reverse target module successfully redirect reverse chimeric antigen receptor T cells to kill CD19+ and α-ADAMTS13 B-cell receptor/surrogate decoy receptor+ target cells

Having validated the binding properties of the novel RevTM, we assessed their cytotoxic potential in the context of the RevCAR system.

First, in order to check the therapeutic window, killing capacity of the different RevCAR systems was evaluated using serial dilutions of RevTM. Thus, titrating the RevTM concentrations would be useful in determining the redirection capabilities of the RevTM even at lower concentrations. Briefly, eFluor 670-stained target cells were co-cultured with RevCAR T cells with or without RevTM for 24 h. Post-incubation, the number of target cells were enumerated via flow cytometry (Figure 2). As shown in Figure 2A, all RevTM showed concentration-dependent redirection of RevCAR T cells against Nalm-6 target cell lines. The half-maximal effective concentration (EC₅₀) was in the low picomolar range for α-CD19, MDTCS and CUB RevTM, and in the low nanomolar range for the fL RevTM. At 5 nM all RevTM were in the saturated plateau phase, showing maximum killing of 60-80% after 24 h (Figure 2A).

Figure 2.Reverse target modules activate reverse chimeric antigen receptor T cells specifically and in a concentration-dependent manner against different target cells. Reverse target modules (RevTM) redirection efficiency against Nalm-6 α-S- and α-CUB-expressing cells (A) and α-S and α-CUB hybridomas (B, C) at different RevTM concentrations. Co-cultures were set up at effector:target (E:T) ratios of 2.5:1 by staining target cells with eFluor 670 and incubating with reverse chimeric antigen receptor (RevCAR) T cells and RevTM. Co-cultures were analyzed by assessing the target cell number via flow cytometry after 24 hours of co-incubation. Non-linear regression curve fit model was used to generate the half-maximal effective concentration (EC₅₀) curve and interpolate the EC₅₀ value. Mean with standard error of the mean (SEM) error bars are represented. fL: full-length; w/o: without.

Compared to Nalm-6 cell lines, after using hybridomas in co-culture assays, EC₅₀ values obtained were 100- to 10,000-fold higher for CUB and MDTCS RevTM (Figure 2B). However, no significant killing was observed for fL RevTM (Figure 2C), which might be attributed to its lower binding capability toward hybridomas than toward SDR⁺ Nalm-6 cells (Figure 1E-G).

Figure 3.Reverse chimeric antigen receptor T cells retain cytotoxic ability even at low effector:target ratios. Co-culture assays were setup by staining the Nalm-6 (A, B) and hybridoma (C, D) target cells with eFluor 670 and adding reverse chimeric antigen receptor (RevCAR) T cells and reverse target modules (RevTM) at a saturated concentration (25 nM for CUB and MDTCS RevTM and 5 nM for a-CD19 and full-length [fL] RevTM). Percentage lysis was read out after 24 hours of co-culture. Data was generated from independent experiments using T cells isolated from 3 donors. Paired t test, one-tailed; *P<0.05; **P<0.01; ***P<0.001. was performed by comparing all conditions with the respective condition without RevTM. Mean with standard error of the mean (SEM) error bars are represented. fL: full-length; NS: not significant; w/o: without.

Second, we also attained information about the RevCAR T-cell functionality under different E:T ratios. This is of relevance in secondary lymphoid organs e.g., spleen, where autoreactive B-cell numbers are high⁴⁰ and RevCAR T cells may be overpowered (Figure 3). All the RevTM efficiently redirected RevCAR T cells to kill Nalm-6 cells at different E:T ratios (Figure 3A, B). As expected, the α-CD19 RevTM and fL RevTM effectively engaged RevCAR T cells to kill both the Nalm-6 α-S and the Nalm-6 α-CUB cell lines. In contrast, MDTCS RevTM specifically killed the Nalm-6 α-S line while the CUB RevTM killed the Nalm-6 α-CUB cell line. Even at a limiting E:T ratio of 1:5, the α-CD19, MDTCS and CUB-based systems achieved a high specific killing of up to 60%, while the fL RevTM achieved similar efficacy only at 1:1 E:T ratio. These findings are consistent with the hybridoma results (Figure 3C, D). CUB and MDTCS RevTM were able to efficiently target the α-CUB and α-S hybridomas, respectively. Vice versa, when MDTCS and CUB RevTM were used against target-negative hybridomas, no target cell lysis was observed, thereby underscoring the target specificity of the system (Online Supplementary Figure S4). However, in line with the results obtained in Figure 2, no killing of α-CUB and α-S hybridomas occurred with the fL RevTM (Figure 3C, D).

Reverse target modules specifically stimulate reverse chimeric antigen receptor T cells to upregulate and release various effector molecules

We next assessed the expression of various effector molecules expressed by the CAR T cells representing their activation, proliferative capacity, and cytotoxic potential. In particular, we examined the cytokine profile for IL-2, TNF, IFN-γ, as known immunomodulatory cytokines, and GM-CSF as a stimulator of innate immune components. Additionally, we also assessed perforin, and granzyme B levels as important drivers of cytotoxicity.

For this, we first investigated supernatants from cytotoxicity co-culture assays collected after 24 h. As shown in Figure 4A-D, RevCAR T cells were stimulated to secrete different pro-inflammatory cytokines (TNF, IFN-y, GM-CSF) and the growth-promoting cytokine IL-2 in the presence of α-CD19, CUB or MDTCS RevTM and their respective target hybridoma and Nalm-6 cells (Figure 4A-D). Upon fL RevTM-mediated redirection of RevCAR T cells against a-S and α-CUB Nalm-6 cells, only minimal or no cytokines were detectable in the co-culture supernatant which is in line with their lower ability to kill the target cells (Figures 2 and 3; Online Supplementary Figure S3A). Second, redirected RevCAR T cells were analyzed via flow cytometry for the upregulation of intracellular granzyme B and perforin. The cells were also stained with α-CD8 antibody to determine potential differences in granzyme B and perforin expression among CD8-positive and CD8-negative cells. Both the pan-B cell depleting α-CD19 RevTM, and the ADAMTS13-specific RevTM (CUB, MDTCS, and fL RevTM) stimulated RevCAR T cells to specifically upregulate granzyme B and perforin. This effect was only observed upon cross-linkage with their corresponding target cell lines (Figure 5; Online Supplementary Figure S3B). Interestingly, both CD8-positive and CD8-negative RevCAR T cells expressed cytotoxic molecules to a comparable extent, suggesting that TM-dependent RevCAR T-cell activation induces the same cytotoxic effector mechanisms in both T-cell subtypes, and can also stimulate CD8-negative T cells to exert a cytotoxic function (Figure 5).

Figure 4.Reverse target modules specifically stimulate reverse chimeric antigen receptor T cells to secrete various cytokines. Sandwich enzyme-linked immunosorbent assay (ELISA) was performed on the co-culture supernatants to quantify different proinflammatory and proliferative cytokines like interleukin-2 (IL-2), tumor necrosis factor (TNF), interferon γ (IFN-γ) and granulocyte monocyte colony-stimulating factor (GM-CSF). Reverse chimeric antigen receptor (RevCAR) cells specifically release these cytokines only in the presence of the respective RevTM and target Nalm-6 (A, B) and hybridoma cells (C, D), albeit at different levels. Data was generated from independent experiments using T cells isolated from 3 donors. Paired t test, one-tailed; *P<0.05;***P<0.001. Was performed by comparing all conditions with the respective condition without target moldules (TM). Mean with standard error of the mean (SEM) error bars are represented. NS: not significant; w/o: without.

The presence of autoantibodies does not hamper the cytotoxic potential of the reverse chimeric antigen receptor system

iTTP is predominated by the presence of α-ADAMTS13 autoantibodies in patient plasma. To elucidate whether these circulating autoantibodies interfere with the performance of the RevCAR system, we set up cytotoxicity assays in the absence or presence of α-ADAMTS13 Abs that compete with the respective TM using either II-1/ z1-201 patient-derived antibodies or 15D1/ 20D2 mAb for Nalm-6 or hybridoma co-cultures, respectively. As described in Pos et al.,⁴¹ the α-ADAMTS13 autoantibody titers were evaluated in a cohort of 48 iTTP patients. This assessment employed both the Technozyme inhibitor ELISA and an in-house ELISA, revealing α-ADAMTS13 autoantibody titers of up to 590 IU/mL and up to 19.5 μg/mL, respectively. Therefore, we applied a concentration range of 1.5 μg/ mL to 20 μg/ mL to ensure a suitable high titer of these antibodies.⁴¹ As shown in Figure 6, RevCAR T cells redirected by either the MDTCS or the CUB RevTM, achieved approximately 80% killing of Nalm-6 or hybridoma cells expressing either a-S or a-CUB SDR/ BCR, respectively within 24 h. Thus, the addition of competing autoantibodies across all tested conditions did not influence the RevCAR system as they maintained their cytotoxic potential. There was only a slight inhibition observed at the highest Ab concentration of 20 ng/µL, particularly in assays using the hybridomas. On a molecular level, at 20 ng/μL of autoantibody concentration, the molar ratio of RevTM to autoantibody is 1:5. This infers that the cytotoxic potential of the system is hampered only by ~20% even when the antibody molecules are five times more than the RevTM (Figure 6B). Thus, high concentrations of clinically relevant antibodies do not significantly impair the cytotoxic effects of the RevCAR system.

Discussion

AID are chronic diseases, in which the immune system attacks the body´s own cells leading to tissue, organ damage and even death. Current treatment options, including for iTTP, primarily focus on symptom management, e.g., general immunosuppression, rather than addressing the underlying cause, thereby compromising both cellular and humoral immunity. This is accompanied by relapse and refractoriness,^42,43 for which the treatment is particularly challenging and clinical guidelines are scarce. In iTTP, ADAMTS13 refractoriness has been observed in up to 15% of patients following rituximab administration, and even 60% of patients experience ADAMTS13 relapses, necessitating repeated rituximab administrations over the course of several years in many cases. Therefore, it remains a substantial unmet clinical need for personalized curative treatment strategies.⁴⁴

The targeted elimination of the B-cell repertoire, encompassing autoreactive B cells, via a-CD19 CAR T cells, has shown highly encouraging outcomes in patients with ulcerative colitis, severe myositis and systemic sclerosis, as these patients have entered long-term remissions without AID symptoms.^45,46 This development represents a significant advancement, as it provides AID patients with access to potential curative treatments and opens new avenues for their care and management.

In this study, we present for the first time, a controllable adaptor-based RevCAR T-cell platform for deep, selective and safe elimination of both pan B cells, as well as of the disease-causing a-ADAMTS13 autoreactive B cells in iTTP For this we designed RevTM against CD19 allowing general B-cell depletion, and RevTM containing parts of or the entire ADAMTS13 protein. Our in vitro data confirm the highly potent and sensitive pan (a-CD19) or ADAMTS13-specific elimination of various target B-cell lines via the RevCAR system at low E:T ratios and low RevTM concentrations, for both approaches. Upon cross-linkage with target cells via a-CD19 or ADAMTS13 RevTM (MDTCS, CUB), RevCAR T cells further showed marked upregulation of cytotoxic effector molecules like granzyme B and perforin while also releasing various effector cytokines, including TNF, IFN-y, IL-2 and GM-CSF. This underscores that RevCAR T cells are activated in a strict RevTM- and target-dependent manner. Despite its high activity against Nalm-6 cells, the fL RevTM demonstrated reduced binding and cytotoxic activity against the target hybridoma cells. As the underlying cause of this observation was not immediately apparent, we evaluated its conformational state and found that it predominantly adopts an open conformation (data not shown). Since this structural orientation may permit exposure of both cryptic and non-cryptic epitopes of ADAMTS13, thereby enabling engagement of a broader range of autoreactive B-cell clones, the mechanistic basis for the reduced efficacy still remains unclear. Further characterization of the fL RevTM, including e.g., assessment of post-translational modifications will be required to clarify the observed differences. Overall, through a comparative analysis of the various ADAMTS13 RevTM, we identified MDTCS and CUB RevTM, encompassing the most immunogenic domains of the ADAMTS13 protein,^1,3 as the lead candidates in terms of their cytotoxic potency, while α-CD19 RevTM is a promising candidate for a less specific but temporal pan B-cell reset via the RevCAR system.

Figure 5.Reverse target modules specifically stimulate reverse chimeric antigen receptor T cells to express cytotoxic effector molecules. Reverse chimeric antigen receptor (RevCAR) T cells were stained intracellularly for granzyme B and perforin 24 hours after co-culture. Effector molecules were analyzed in both CD8-positive and CD8-negative RevCAR T cells in co-cultures against Nalm-6 α-S (A, B), Nalm-6 α-CUB (C, D), hybridoma α-S (E, F) and hybridoma α-CUB (G, H). Raw data histogram plots for 1 representative donor are shown on the left of each sub-panel. Data was generated from independent experiments using T cells isolated from 3 donors. Paired t test, one-tailed; *P<0.05; **P<0.01; ***P<0.001 was performed by comparing all conditions with the respective condition without RevTM. NS: not significant; w/o: without.

Figure 6.Reverse chimeric antigen receptor adaptor system retains cytotoxic ability even in the presence of high concentrations of autoantibodies. Co-culture assay conditions: reverse target modules (RevTM) concentration - 25 nM, effector:target (E:T) ratio - 2.5:1, co-culture time - 24 hours and increasing concentrations of autoantibodies against both Nalm-6 (A) and hybridoma (B) target cell lines. Data was generated from independent experiments using T cells isolated from 3 donors. Statistical analysis was performed by comparing all conditions with condition without RevTM. Paired t test; two-tailed; *P<0.05; **P<0.01; ***P<0.001. Ab: antibody; mAb: monoclonal antibody; w/o: without.

In contrast to the CD19-specific RevCAR approach, circulating autoantibodies in iTTP patients may clinically hinder the successful application of ADAMTS13-based RevTM and negatively affect treatment efficacy due to their binding to RevTM. We have addressed such concerns, and based on the results from the competition assays performed, we consider the risk to be rather low. Our data underline that both MDTCS and CUB RevCAR systems are only moderately hampered in the presence of high concentrations of competing autoantibodies. Even at an autoantibody concentration of 20 μg/mL, RevTM were still found to be highly effective in redirecting RevCAR T cells. According to Pos et al.,⁴¹ this concentration corresponds to the highest levels of α-ADAMTS13 autoantibody titers observed in a cohort of 48 iTTP patients. Unlike conventional CA(A)R approaches, where potential drops in treatment efficiency can only be compensated by infusing higher CA(A)R T cell numbers, in the RevCAR system it can be achieved by simply increasing the RevTM dose. Furthermore, the risk to fix complement to RevTM-autoantibody complexes or mediate antibody-dependent cellular cytotoxicity in iTTP is low, as the most predominant isotype in iTTP relapse is IgG4.^47-49

Similar to all adaptor and conventional CAR T approaches, potential clinical application of the RevCAR platform in AID, including iTTP, would involve transient lymphodepleting preconditioning to facilitate engraftment and expansion of the infused RevCAR T cells. As known from CD19-directed CAR T-cell therapy, preconditioning has been generally well tolerated in patients with selected autoimmune indications.¹⁵ However, additional studies will be necessary to further optimize conditioning regimens and carefully balance benefit-risk ratios. Compared to conventional CAR, a central advantage of the RevCAR system lies in its modular design with its universal RevCAR T cells and short-lived, switchable adaptor molecules (RevTM) providing an intrinsic safety mechanism. The principle of applying adaptor CAR therapies has been clinically explored in our recent phase I study of the related adaptor platform “UniCAR”.^50,51 After infusing a single dose of CAR T cells, UniCAR T-cell activation can be regulated through controlled administration of the target-specific TM. Similarly, a clinical application of the RevCAR system in iTTP would involve initial RevCAR T-cell administration followed by continuous RevTM infusion for several weeks during the initial phase. Moreover, in the event of a (ADAMTS13) relapse, therapy could be reinitiated upon repeated RevTM dosing. To facilitate management of acute toxicities of CAR T cells in AID, like cytokine-release syndrome, as well as long-term consequences such as sustained B-cell aplasia, RevTM administration could be discontinued to immediately terminate RevCAR-mediated signaling. Especially for autoimmune patients who may already have a compromised immune system, its inherent control mechanism represents a significant advantage over conventional CA(A)R approaches. The feasibility of such a rapid and temporal control switch has been successfully proven in the related adaptor CAR approach “UniCAR”,^50,51 where UniCAR T cells remained detectable for up to 6 months. They also demonstrated re-expansion upon adaptor re-infusion after treatment-free intervals.^50,51

In addition, the modular architecture of the RevCAR system permits redirection of the same CAR T-cell product toward multiple targets through exchange of RevTM specificity, thereby expanding therapeutic flexibility. Clinically, the simultaneous or subsequent combination of MDTCS and CUB RevTM would be a highly promising strategy for treatment of iTTP patients, as they together address the majority of disease-relevant epitopes,^1,3 and thus would enable RevCAR T cells to efficiently eliminate nearly all α-ADAMTS13 autoreactive B cells in patients. Alternatively, or in case of relapse, the herein presented α-CD19 RevTM can be used for pan B-cell reset via the RevCAR system. While bispecific antibodies can also provide conditional T-cell engagement, these molecules typically rely on CD3-mediated TCR signaling, which may be downregulated upon T-cell activation.^52,53 In contrast, RevCAR provide their own activation and co-stimulatory domains, ensuring sustained functionality across all T-cell subtypes. In conclusion, the present study establishes proof-of-principle for the adaptor RevCAR T-cell platform in iTTP, representing the first application of such a system to selectively target autoreactive B cells in the disease. Thus, it offers iTTP patients a prospective curative treatment option and also poses significant potential for successful application in other AID.

Footnotes

• Received January 12, 2026
• Accepted April 7, 2026

Correspondence

^#CA and AF contributed equally as senior authors

A. Feldmann a.feldmann@hzdr.de

C. Arndt c.arndt@hzdr.de

Disclosures

MB has filed patents related to the RevCAR system and the anti-La monoclonal antibodies. All other authors have no conflicts of interest to disclose.

Contributions

Methodology, data curation, validation, formal analysis, visualization, writing - original draft by RSG. Resources, writing - reviewing and editing by SRP. Formal analysis, resources, writing - reviewing and editing by LRL. Resources, writing - reviewing and editing by MD and AvJ-T. Supervision, funding acquisition, resources, writing - reviewing and editing by MB, JV, and KV. Conceptualization, methodology, formal analysis, supervision, funding acquisition, project administration, resources, writing - original draft by CA. Conceptualization, methodology, supervision, funding acquisition, project administration, resources, writing -reviewing and editing by AF.

Funding

This work was supported by funds of the EU MSCA project TOLERATE by the European Union under grant agreement number 101072729 and funded by the European Union under grant agreement number 101115159 (Horizon EIC 2022 Pathfinder challenges 01 project B-specific). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or European Research Executive Agency or the European Innovation Council and SMEs Executive Agency (EISMEA). Neither the European Union nor the granting authority can be held responsible for them.

Acknowledgments

We would like to thank the students Eugenia Crespo, Haidy A. Saleh Hassan, Karla Elizabeth Gonzalez Soto, Tabea Bartsch, Katharina Hoellen, Fabienne I. Weinberg, Lamice Shurafa, Hugo Boutier, and Nathalia Jones Cifuentes for their precious help. We warmly acknowledge Annegret Riedel, Kim Weiße, Inge Pareyn, Anne-Sophie Delmote and Paul Kaijen for their excellent technical work and support. We also thank Dr. Jurie Tashkandi, Tim Postmus, and PhD. Zsolt Szeles for their valuable inputs. We also thank Torsten Tonn (German Red Cross Blood Donation Service North East, Dresden, Germany) for providing the buffy coats of healthy donors.