We generated a novel CD19CAR (CAT) with a lower affinity than FMC63, the binder utilised in many clinical studies. CAT CAR T cells showed increased proliferation/cytotoxicity in vitro and enhanced proliferative capacity and anti-tumor activity than FMC63 CAR T cells in a xenograft model. In a clinical study (CARPALL, NCT02443831), 12/14 patients with relapsed/refractory pediatric BALL obtained molecular remission after CAT CAR T cell therapy. CAR T cell expansion compared favourably with published data on other CD19CARs and persistence was demonstrated in 11 of 14 patients at last follow-up. Toxicity was low with no severe cytokine release syndrome. At a median follow up of 14 months, 5/14 patients (37%) remain in molecular CR with circulating CAR T cells.

Mature T cell cancers are typically aggressive, treatment resistant and associated with poor prognosis. Clinical application of immunotherapeutic approaches has been limited by a lack of target antigens that discriminate malignant from healthy (normal) T cells. Unlike B cell depletion, pan-T cell aplasia is prohibitively toxic. We report a new targeting strategy based on the mutually exclusive expression of T cell receptor β-chain constant domains 1 and 2 (TRBC1 and TRBC2). We identify an antibody with unique TRBC1 specificity and use it to demonstrate that normal and virus-specific T cell populations contain both TRBC1 and TRBC2 compartments, whereas malignancies are restricted to only one. As proof of concept for anti-TRBC immunotherapy, we developed anti-TRBC1 chimeric antigen receptor (CAR) T cells, which recognized and killed normal and malignant TRBC1, but not TRBC2, T cells in vitro and in a disseminated mouse model of leukemia. Unlike nonselective approaches targeting the entire T cell population, TRBC-targeted immunotherapy could eradicate a T cell malignancy while preserving sufficient normal T cells to maintain cellular immunity.

T cell acute lymphoblastic leukemia (T-ALL) is an aggressive malignancy of immature T lymphocytes, associated with higher rates of induction failure in comparison to B-ALL. The potent immunotherapeutic approaches applied in B-ALL, which have revolutionized the treatment paradigm, have proven more challenging in T-ALL, largely due to a lack of target antigens expressed on malignant but not healthy T cells. Unlike B cell depletion, T cell aplasia is highly toxic. Here, we demonstrate that the chemokine receptor CCR9 is expressed in >70% of cases of T-ALL, including >85% or relapsed/ refractory disease, and only on a small fraction (<5%) of normal T cells. Using cell line models and patient-derived xenografts, we show chimeric antigen receptor (CAR)-T cells targeting CCR9 are resistant to fratricide and have potent anti-leukemic activity both in vitro and in vivo, even at low target antigen density. We propose anti-CCR9 CAR-T cells could be a highly effective treatment strategy for T-ALL, avoiding T cell aplasia and the need for genome engineering that complicate other approaches.

Chimeric antigen receptor (CAR) T cells are a promising form of cancer immunotherapy, although they are often associated with severe toxicities. Here, we present a split-CAR design incorporating separate antigen recognition and intracellular signaling domains. These exploit the binding between the tetracycline repressor protein and a small peptide sequence (TIP) to spontaneously assemble as a functional CAR. Addition of the FDA-approved, small molecule antibiotic minocycline, acts as an “off-switch” by displacing the signaling domain and down-tuning CAR T activity. Here we describe the optimization of this split-CAR approach to generate a CAR in which cytotoxicity, cytokine secretion and proliferation can be inhibited in a dose-dependent and reversible manner. Inhibition is effective during on-going CAR T cell activation and inhibits activation and tumor control in vivo. This work shows how optimization of split-CAR structure affects function and adds a novel design allowing easy CAR inhibition through an FDA-approved small molecule.

Peripheral T cell lymphomas are typically aggressive with a poor prognosis. Unlike other hematologic malignancies, the lack of target antigens to discriminate healthy from malignant cells has limited the efficacy of immunotherapeutic approaches. The T cell receptor expresses one of two highly homologous chains [T cell receptor β-chain constant (TRBC) domains 1 and 2] in a mutually exclusive manner, making it a promising target. We previously described an antibody with unique TRBC1 specificity (Jovi-1). Here we demonstrate specificity redirection by rational design using structure-guided computational biology to generate a TRBC2-specific antibody (KFN). This permitted the generation of paired reagents (chimeric antigen receptor-T cell) specific for TRBC1 and TRBC2, with preclinical evidence to support their efficacy in T cell malignancies.

Introduction: The CARPALL study (NCT02443831) employed a novel CD19CAR (CAT-41BBz CAR) with a faster off rate than the Kymriah FMC63-41BBz CAR (CAT 3.1x10-3s-1, FMC 6.8 x 10-5s-1), with equivalent on-rate (CAT 2.2 x 105, FMC 2.1 x 105). We herein report updated outcomes and CAR T cell persistence with an additional 6 months follow up from a submitted manuscript (Ghorashian et al., Nat Med, submitted) Methods: Patients aged <25 years with high risk, relapsed CD19+ B-ALL were eligible on this multi-centre, open label, non-randomised phase I study of autologous CAT-41BBz CAR T cells. Patients were followed to a data cut-off of 07/18/2019. CAT-41BBz CAR T cells were generated by magnetic bead activation of leucapheresed PBMCs, lentiviral transduction, followed by bioreactor expansion and magnetic bead removal prior to cryopreservation. All patients received lymphodepletion (fludarabine + cyclophosphamide) followed by 1x106/kg CAR T cells. Presence of CAR T cells in the blood and bone marrow (BM) was assessed (flow cytometry and qPCR) monthly for 6 months, then 6 weekly to 1 year and then 3 monthly. BM MRD was assessed (IgH qPCR, flow cytometry) at the same time-points up to 2 years to establish durability of responses as a stand-alone therapy. Primary end-points were incidence of grade 3-5 toxicity and the proportion of patients achieving molecular remission. Results: Of 17 patients recruited, 14 were treated due to manufacturing failure in 3 patients.The median age was 9 years (range 1-19 years). All patients had advanced ALL with a median of 4 prior therapy lines. 10 of 14 patients (71%) had relapsed post allogeneic SCT. Prior to lymphodepletion, 4 patients had >5% BM disease, 6 had disease between 5x10-2and 1x10-5, 4 were BM MRD negative having had recurrent isolated CNS disease. Median transduction efficiency was 31% (range 16.5 to 96.4%). 12/14 treated patients received the anticipated dose of 1x106CAR T cells/kg (2 received 0.9x106/kg). Considering all evaluable patients, (n=14 for CAR T cell persistence by qPCR, n=13 by flow) the geometric mean of Cmax was 128 912/µg DNA and of the area under the curve between D0 and D28 was 1,721,355 copies/ µg DNA (Table 1). At the point of maximal expansion, a median of 35% of circulating T cells were CAR+. Median half-life was 34 days (range 3-102). CAR T cells continued to be detectable by qPCR in 11 of 14 (79%) patients at last assessment and by flow cytometry up to 30 months post infusion in 8 of 13(61%). Median duration of CAR T persistence by flow was 261 days (range 7-917). 3 patients failed to have persistence of CAR T cells beyond 1 month. T cell mediated anti-CAR specific cytotoxic activity was detected in 2/2 evaluable patients. Updated persistence data will be presented at the meeting Cytokine release syndrome (CRS) occurred in 13 (93%, grade 1 n=9, grade 2 n=4). None developed ≥grade 3 CRS, had CRS-related ICU admission, or received Tocilizumab. CRS was associated with modest elevations of IL-6, IFN-γand IL-10. Grade 2 neurotoxicity was observed in 3 patients and resolved spontaneously. One patient had grade 4 leucoencephalopathy presumed due to chemotherapy as well as grade 5 sepsis. Ten patients (71%) had grade 3-4 cytopenia persisting beyond day 28 or recurring afterthis. 12/14 (86%) patients achieved molecular complete or continuing complete remission at a median of 30 days post infusion (range 30-90 days, Table 2). At a median follow-up of 20.3 months, 4/14 (29%) evaluable patients remain MRD negative. 5 relapsed with CD19-disease, 1 with CD19+ disease. The median duration of EFS (based on death or morphological relapse) has not been reached, 12 month EFS = 52%, OS = 70% (Figures 1, 2 and Table 3). Conclusion: We noted excellent CAR T cell expansion and persistence in a ALL cohort treated with the fast off-rate CAT-41BBz CAR despite their lower BM disease at treatment compared to other studies. The kinetics documented for all evaluable patients showed a 5-fold greater CAR T cell expansion and 2-fold longer half-life than responders in published series utilising tisagenlecleucel in a similar ALL cohort (Mueller et al., Blood 2017). Patients had a favourable toxicity profile with no severe (grade 3-4) CRS and equivalent disease outcomes to the ELIANA study despite having similarly advanced disease (Maude et al., NEJM 2018292). These data suggest long lived CAR T cell persistence supports stand-alone therapy for ALL with durable responses. Disclosures Ghorashian: Celgene: Honoraria; novartis: Honoraria; UCLB: Patents & Royalties: UCLB. Kramer:UCLB: Patents & Royalties. Ciocarlie:Servier: Other: Financial Support. Farzaneh:Autolus Ltd: Equity Ownership, Research Funding. Pule:Autolus: Employment, Equity Ownership, Patents & Royalties. Amrolia:UCLB: Patents & Royalties.

CAR-T cell therapy against CD19 has changed the treatment landscape in relapsed/refractory (r/r) B-ALL. R/r T-ALL has a dismal prognosis, with an unmet need for effective targeted therapies. Several unique challenges mean that CAR-T cell therapy has yet to be successfully translated to T-ALL. Most strategies have targeted pan-T cell antigens (CD7, CD5) but these are limited by T cell aplasia and fratricide, requiring elimination of CAR-T antigen surface expression during manufacture. An ideal target would be exclusively or largely confined to the malignant T cell component but published examples of these (CD1a and TRBC1) are expressed in only minor T-ALL subsets. We previously showed that CD21 is expressed in a NOTCH-dependent manner in T-ALL (Leukemia. 2013, 27:650) and have developed it as a potential immunotherapy target, being primarily expressed on normal B cells, with minimal expression on mature T cells. 70% of human T-ALL cell lines (9/16) expressed surface CD21 by flow cytometry (FACS), with a median antigen density in positive lines of 2545/cell. In primary T-ALL, 57% of presentation samples (n=58) expressed CD21 (median antigen density 1168/cell). 45% of relapse (n=11) and 20% of primary refractory cases (n=30) expressed CD21, with a similar antigen density to presentation samples. CD21 positivity varied by maturational stage, with highest expression in cortical T-ALL (80% of cases) followed by pre-T (72%), mature (67%), ETP (25%) and pro-T (17%). Healthy donor blood (n=14) showed CD21 expression limited to B cells and a low proportion (11%) of T cells (10-fold lower intensity v B cells, 316 antigens/cell). T cell CD21 expression was not up-regulated upon activation with CD3/CD28 antibodies (n=6) and was not associated with markers of differentiation/exhaustion. To target CD21, DNA gene-gun vaccination of rats with a plasmid encoding full-length CD21, followed by phage display was performed and multiple anti-CD21 scFvs isolated. These were cloned into 4-1BBζ CARs and expressed in primary T cells but failed to kill or secrete cytokines in response to CD21+ SupT1 cells. CD21 is a bulky molecule, with 15/16 sushi repeats in the extracellular domain. All isolated scFvs were found to bind membrane-distal domains. We hypothesized that ineffective signalling due to inadequate synapse formation was responsible for poor performance of anti-CD21 CAR-T, and that binders to membrane-proximal epitopes would signal more efficiently. We re-vaccinated rats with the first 5 sushi repeats of CD21 and generated a library of binders which bound CD21 at this membrane-proximal region. Multiple candidate binders expressed as CARs were functional, with cytotoxicity and interferon-γ secretion in response to CD21+ target cells. However, non-specific background cytokine secretion was seen against CD21 negative cells, and no IL-2 secretion was seen. Re-cloning binders into a fragment antigen binding (Fab)-CAR architecture yielded constructs capable of specific cytotoxicity, IFN-γ and IL2 secretion against a CD21+ cell line but not its CD21 negative counterpart (n=6). Our lead anti-CD21 candidate CAR specifically proliferated in vitro, without fratricide or premature exhaustion/ differentiation, and was active against low-density CD21-positive cell lines (n=3) and primary cells from 2 T-ALL patients. Improved functionality of Fab v scFv-based CAR was not driven by higher affinity binding or CAR surface expression. We tested anti-CD21 CAR in murine models of T-ALL. NSG mice were injected with SupT1-luciferase cells and treated with aCD19 or aCD21 CAR-T on day +5. At 2 weeks post treatment, markedly lower disease burden was seen in CD21 CAR-T v CD19 recipients by bioluminescence imaging (median radiance 71700 v 790000 p=0.0079). Further, we injected primary T-ALL blasts in another cohort, treating with aCD19 or aCD21 CAR-T on D+20. Serial bleeds from day 27 post CAR-T showed tumour control in aCD21 CAR treated mice (p=0.024) with an overall survival advantage (median OS 44 days vs undefined, HR = 19.8, p = 0.0069, n=4/group). In summary, we propose CD21 as a novel target for CAR-T cell therapy in T-ALL. Its expression is largely restricted to the malignant T cell compartment, overcoming issues with fratricide and on-target off-tumour effects seen in many T-ALL CAR-T strategies to date. Despite the complexity of the target, we have successfully generated an aCD21 CAR that is functional both in vitro and in vivo. Disclosures Maciocia: Autolus: Current equity holder in publicly-traded company. Onuoha: Autolus: Ended employment in the past 24 months. Khwaja: Pfizer: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Novartis: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Jazz Pharmaceuticals: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Astellas: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Abbvie: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau. Maciocia: Autolus: Current equity holder in publicly-traded company, Research Funding. Pule: Autolus: Current Employment, Current equity holder in publicly-traded company.

Introduction Mature T cell lymphomas are aggressive, treatment resistant cancers that are associated with poor prognosis. Clinical application of immunotherapeutic approaches has been limited by a lack of target antigens that discriminate malignant from healthy T cells. Unlike B cell depletion, pan-T cell aplasia is prohibitively toxic. Previously we reported a targeting strategy based on the mutually exclusive expression of T cell receptor beta-chain constant domains 1 and 2 (TRBC1 and TRBC2). We identified an antibody with unique TRBC1 specificity and demonstrated that anti TRBC1 chimeric antigen receptor (CAR) T cells can ablate cells expressing TRBC1 TCRs while sparing those expressing TRBC2 TCRs. A phase I clinical study investigating the efficacy of our TRBC1 CAR is ongoing. T cell malignancies are clonal, and the ratio of TRBC2 to TRBC1 expressing lymphoma cases is predicted to be approximately 2:1. To treat all cases of T cell lymphoma, a CAR that targets TRBC2 is needed. TRBC1 and 2 are highly homologous. Structural studies suggest that amino acid inversions at positions 4 and 5 of the constant beta chain provide an accessible discriminating portion between the two proteins. Given the structural similarities between TRBC1 and TRBC2 and our characterized binder against TRBC1, we explored generating antibodies with specificity towards TRBC2 via a structure guided computational biology approach; engineering the previously identified TRBC1 antibody and reversing its specificity such that it recognised TRBC2. Results The crystal structure of the TRBC1 specific monoclonal antibody was solved in complex with a TRBC1-TCR to 2.4Å, Figure 1. Through computational biology and protein engineering we rationally designed a mutant version of TRBC1 binder that was specific for TRBC2 and had a 1000 fold decreased affinity towards TRBC1. Flow cytometry analysis of the TRBC2 specific antibody demonstrated the ability to bind to T-cells expressing TRBC2 TCRs. We further showed that the engineered antibody retained favourable biophysical characteristics with high stability (Fab Tm > 65oC) and low aggregation propensity (>99% monomer). We used the engineered monoclonal antibody to generate a 2nd generation anti-TRBC2 CAR. We demonstrated that our anti-TRBC2 CAR showed specificity, cytokine release and cytotoxicity in 72hr co-cultures against TRBC2+ cell lines but not TRBC1+ cell lines or cell lines that did not express TCR on the surface. Anti-TRBC2 CAR T-cells also demonstrated proliferative capacity in long-term co-culture assays. Conclusions We have utilised structural biology and rational protein design to generate CAR-T cells capable of specifically targeting TRBC2. The combination of TRBC1 and 2 targeting CAR-T cell products with a patient stratification companion diagnostic assay offers a therapeutic strategy for the treatment of a wide range of, otherwise untreatable, T-cell lymphomas. Figure 1. A. Structural interface between TCR Beta and Fab fragment of TRBC1 specific antibody. B and C. CDR fold of TRBC1 binder and 90o rotation. Disclosures Onuoha: Autolus Ltd: Employment, Equity Ownership, Patents & Royalties. Ferrari:Autolus Ltd: Employment, Equity Ownership. Bulek:Autolus Ltd: Employment, Equity Ownership, Patents & Royalties. Bughda:Autolus Ltd: Employment, Equity Ownership, Patents & Royalties. Manzoor:Autolus Ltd: Employment, Equity Ownership, Patents & Royalties. Srivastava:Autolus Ltd: Employment, Equity Ownership. Ma:Autolus Ltd: Employment. Karattil:Autolus Ltd: Employment, Equity Ownership, Patents & Royalties. Kinna:Autolus Ltd: Employment, Equity Ownership. Thomas:Autolus Ltd: Employment, Equity Ownership, Patents & Royalties. Cordoba:Autolus Ltd: Employment; Autolus Ltd: Patents & Royalties; Autolus Ltd: Equity Ownership. Maciocia:Autolus: Equity Ownership, Patents & Royalties: UCLB. Pule:Autolus Ltd: Employment, Equity Ownership, Other: Salary contribution paid for by Autolus, Research Funding; University College London: Patents & Royalties: Patent with rights to Royalty share through UCL.