Sensitive bispecific chimeric T cell receptors for cancer therapy

The expression of a synthetic chimeric antigen receptor (CAR) to redirect antigen specificity of T cells is transforming the treatment of hematological malignancies and autoimmune diseases [1–7]. In cancer, durable efficacy is frequently limited by the escape of tumors that express low levels or lack the target antigen [8–12]. These clinical results emphasize the need for immune receptors that combine high sensitivity and multispecificity to improve outcomes. Current mono- and bispecific CARs do not faithfully recapitulate T cell receptor (TCR) function and require high antigen levels on tumor cells for recognition [13–17]. Here, we describe a novel synthetic chimeric TCR (ChTCR) that exhibits superior antigen sensitivity and is readily adapted for bispecific targeting. Bispecific ChTCRs mimic TCR structure, form classical immune synapses, and exhibit TCR-like proximal signaling. T cells expressing Bi-ChTCRs more effectively eliminated tumors with heterogeneous antigen expression in vivo compared to T cells expressing optimized bispecific CARs. The Bi-ChTCR architecture is resilient and can be designed to target multiple B cell lineage and multiple myeloma antigens. Our findings identify a broadly applicable approach for engineering T cells to target hematologic malignancies with heterogeneous antigen expression, thereby overcoming the most frequent mechanism of relapse after current CAR T therapies.

whereas a high frequency of cells expressing the 28z CAR activated NFAT and NFkB in the absence of antigen recognition (Extended Data Fig. 3a-c).We evaluated signaling in primary T cells expressing CD19-specific ChTCRs and CARs by first measuring calcium influx after crosslinking receptors with CD19 antigen.Split and full ChTCRs and the 28z CAR fluxed calcium with similar magnitudes, whereas calcium influx was barely detectable in T cells expressing the 4-1BBz CAR (Fig 1h).We then compared phosphorylation of LAT and Zap70 using antigen coated beads to activate CAR, ChTCR, or TCR expressing T cells.Activation of T cells expressing the split and full ChTCRs led to the rapid phosphorylation of LAT Y220 and Y171 and Zap70 Y319 at the one minute time point, matching the kinetics of the NY-ESO 1 TCR (Fig 1i, j and Extended Data Fig. 3d).In contrast, T cells expressing CARs showed less intense LAT and Zap70 phosphorylation that peaked later after 5 minutes.These data indicate that like a TCR, both ChTCR formats assemble with all CD3 chains, form TCR-like synapses, lack tonic signaling, and induce rapid antigen-specific signaling.

The CD19 full ChTCR has superior antigen sensitivity and anti-tumor efficacy
To determine the antigen sensitivity of full and split ChTCR + T cells, we generated Nalm-6 cells expressing high, medium, and low levels of CD19 and analyzed recognition by primary T cells transduced with CD19 specific ChTCRs and CARs (Fig 2a).When co-cultured with CD19 high cells, CD28z CAR T cells produced greater levels of IL-2 and IFNg compared to 4-1BBz CAR and ChTCR T cells (Fig 2b and 2c).However, when co-cultured with Nalm-6 cells expressing medium or low levels of CD19, split and full ChTCR T cells produced more IL-2 and IFN-g than either CAR format.Strikingly, T cells expressing the full ChTCR produced higher levels of IL-2 and INF-g than T cells expressing the split ChTCR (Fig 2b and 2c).There was no difference in proliferation of CAR and ChTCR T cells when co-cultured with CD19 high cells, however ChTCR T cells proliferated more than CAR T cells when co-cultured with CD19 mid and CD19 low Nalm-6 cells, and full ChTCR T cells proliferated more than split ChTCR T cells in response to the CD19 low tumor (Fig 2d).Thus, both ChTCR formats confer improved recognition of low antigen expressing tumor cells compared to CARs, and the full ChTCR was superior to the split ChTCR.To determine the antitumor activity of full ChTCR + T cells in vivo, we treated NSG mice engrafted with Raji lymphoma cells with a low dose of each of the transduced T cells (Fig 2e).T cells expressing the full ChTCR quickly eradicated tumor cells in all treated mice and improved survival compared to mice treated with split ChTCR, CD28z or 4-1BBz CAR T cells (Fig 2f-h).We further tested ChTCR + T cells for antigen sensitivity in vivo by engrafting NSG mice with Nalm-6 low cells.T cells expressing both ChTCRs exhibited superior antitumor activity early after infusion compared to T cells expressing conventional CD28z or 4-1BBz CARs, and the full ChTCR was again superior to the split ChTCR (Fig 2i, j).These data show that ChTCRs are more sensitive for recognizing antigen low tumor cells compared to conventional CARs and that the full ChTCR is the most effective receptor in vitro and in vivo.

Design of a sensitive CD22 monospecific ChTCR
The TRBC chain in the CD19 specific full ChTCR was left unoccupied to allow targeting of a second antigen by fusing a scFv of different specificity to TRBC.Bispecific targeting of CD19 and CD22 with T cells has been a major interest in the clinic because CD19 and CD22 negative or low relapses occur after monospecific therapies targeting these antigens [8,37].Unfortunately, a CD19/CD22 bispecific CAR, termed the "Loop" CAR, demonstrated compromised sensitivity to each antigen compared to the respective monospecific CARs, and patients treated with the Loop CAR relapsed with CD19 low/neg CD22 + tumor cells [8].We sought to determine if the ChTCR platform could provide a more effective bispecific CD19/CD22 receptor.
We first evaluated monospecific full ChTCRs in which two different CD22 scFvs were fused to TRBC (Fig 3a).We tested the m971 scFv, which is used in the Loop CAR and targets a membrane proximal epitope, and the 9A8 scFv that targets a more membrane distal epitope (Fig 3a) [38].ChTCRs were constructed in both VHVL and VLVH orientations, expressed in TRBC and TRAC edited T cells, and compared to conventional 4-1BBz CARs constructed with the same scFvs (Fig 3a).CD8 + T cells expressing all constructs bound soluble recombinant CD22, with the m971 CAR and ChTCR showing a higher MFI of CD22 binding than observed with 9A8 (Fig 3b,c).However, when T cells were co-cultured with WT Nalm-6 cells, m971 ChTCR + T cells produced lower levels of IL-2 and INF-g, and proliferated poorly compared to T cells expressing the 9A8 ChTCR (Fig 3df; Extended Data Fig. 4a).CAR T cells constructed with each scFv were functional in these assays, with m971 CAR T cells producing higher levels of cytokines and proliferating better than 9A8 CAR T cells (Fig 3d-f).A ChTCR in which the m971 scFv was fused in VL/VH or VH/VL orientations to the TRAC chain rather than TRBC also bound soluble CD22 but did not function against CD22 + tumor cells (Extended Data Fig. 4b-f).The discrepancy between binding of soluble CD22 by m971 and 9A8 ChTCR T cells and function in response to tumor cells suggested that the epitope targeted by m971 on membrane bound CD22 is less accessible.This was supported by analysis of synapses formed by CD22-specific ChTCR T cells with lipid bilayers functionalized with CD22 and ICAM-1.T cells expressing the 9A8 ChTCR formed a synapse with CD22 localizing to the center surrounded by a ring of ICAM-1 whereas T cells expressing the m971 ChTCR showed minimal CD22 accumulation in the synapse (Fig 3g-i).We next compared antigen binding and sensitivity of T cells expressing the 9A8 full or split ChTCRs or a control 4-1BBz CAR (Extended Data Fig. 5a).Binding to CD22 was similar for the full ChTCR and 4-1BBz CAR T cells and higher than that observed with split ChTCR T cells (Extended Data Fig. 5b, c).T cells expressing the full ChTCR also showed higher TCRab expression than T cells expressing the split ChTCR (Extended Data Fig. 5d).To evaluate antigen sensitivity, we used Nalm-6 CD22 WT (11912 molecules/cell) and CD22 Low cell lines (959 molecules/cell) that were derived as target cells (Extended Data Fig. 5e).T cells expressing the full and split ChTCRs and the 4-1BBz CAR proliferated similarly after co-culture with Nalm-6 WT cells (Fig 3j), however T cells expressing the full ChTCR demonstrated greater proliferation compared to split ChTCR + and CAR + T cells after co-culture with Nalm-6 CD22 Low cells.The full ChTCR T cells also produced higher levels of IL-2 and IFN-g in response to Nalm-6 CD22 Low cells compared to split ChTCR and 4-1BBz CAR T cells (Fig 3k, l).These findings showed that the CD22 full ChTCR provided more sensitive recognition of CD22 on tumor cells than the CD22 split ChTCR and 4-1BBz CAR.This data also provided the rationale to investigate whether a ChTCR designed with CD19 and CD22 scFvs fused to each individual TCR chain could provide sensitive bispecific antigen recognition.

Design and function of bispecific CD19/CD22 ChTCR
We constructed a lentiviral vector that encoded the CD19-specific FMC63 scFv fused to TRAC in a VHVL orientation (Fig 1a ) and the CD22 specific 9A8 scFv fused to TRBC in a VLVH orientation (Fig 4a; Extended Data Fig. 6a).CD8 + T cells were transduced with this Bi-ChTCR followed by TCR KO using CBE.For functional comparison, T cells were transduced with CD19 monospecific 4-1BBz CAR, CD22 monospecific 4-1BBz CAR using the m971 scFv, and with the bispecific Loop CAR [37,38] (Fig 4a-b).The MFI of rCD19 and rCD22 binding was superior for T cells expressing the Bi-ChTCR compared to the Loop CAR (Fig 4b-d).
Expressing two scFvs on the same ChTCR could result in interactions between the scFvs that induce tonic signaling or affect synapse formation.When expressed in Jurkat TPR cells, Bi-ChTCR exhibited minimal antigen-independent activation as demonstrated by the frequency of NFAT + , NF-kB + and AP-1 + cells (Extended Data Fig. 6b-d).Monospecific CD19 and CD22 4-1BBz CARs also had minimal tonic signaling in Jurkat TPR cells while the Loop CAR exhibited a high level of antigen-independent activation (Extended Data Fig. 6b-d).Additionally, the Bi-ChTCR formed an organized synapse when tested on a soluble bilayer functionalized with CD19 and CD22 proteins with CD22 taking a more central position than CD19 in the cSMAC (Fig 4e).We next transduced primary T cells and tested their recognition of Nalm-6 cells that expressed endogenous levels of CD19 and CD22, Nalm-6 cells that were gene edited to express only CD19 (Nalm-6 CD22 ko ), only CD22 (Nalm-6 CD19 ko ), or neither CD19 and CD22 (Nalm-6 DKO) (Extended Data Fig. 6e).Bi-ChTCR T cells demonstrated robust proliferation and cytokine production in response to Nalm-6 WT and Nalm-6 cells expressing only CD19 or CD22 (Fig. 4f-h and Extended Data Fig. 6f).In contrast, Loop CAR T cells showed reduced functions when cocultured with CD19 ko cells compared to Nalm6 WT or CD22 ko , and CD19 and CD22 mono-specific CAR T cells only proliferated and produced cytokines when co-cultured with cells expressing their cognate antigen (Fig. 4f-h and Extended Data Fig. 6f).T cells expressing both Bi-ChTCRs lysed Nalm-6 WT, Nalm-6 CD22 ko , and Nalm-6 CD19 ko cells, while T cells expressing the Loop CAR lysed Nalm-6 CD22 ko cells but exhibited poor lysis of Nalm-6 CD19 ko target cells (Extended Data Fig. 6g).As expected, monospecific CAR T cells failed to recognize Nalm-6 cells lacking the cognate antigen (Extended Data Fig. 6g).It was conceivable that signaling or sensitivity of the Bi-ChTCR for each single antigen would be compromised by the presence of two different scFvs in close proximity.We measured LAT phosphorylation after co-culturing T cells expressing individual monospecific and Bi ChTCRs, or the Loop CAR with CD19 or CD22 positive Nalm-6 cells.LAT phosphorylation was comparable in intensity and kinetics between the Bi-ChTCR and the monospecific CD19 or CD22 full ChTCRs, and greater than that observed with the Loop CAR (Fig. 5a-b).To evaluate antigen sensitivity, we compared T cell recognition of Nalm-6 CD22 ko cells expressing high and low levels of CD19, and of Nalm-6 CD19 ko cells expressing high and low levels of CD22 (Fig 5c).Strikingly, T cells expressing the CD19/CD22 Bi-ChTCR exhibited strong proliferation (Fig 5d and g) and cytokine production against both CD22 ko CD19 low and CD22 low CD19 ko tumor cells (Fig 5e, f, h and i) that was equivalent to mono-specific CD19 and CD22 full ChTCRs and superior to the Loop CAR.We next modeled in vivo therapy of NSG mice engrafted with a mixture of Nalm-6 WT (CD19 + /CD22 + ), Nalm-6 CD19 ko /CD22 + , and Nalm-6 CD19 + /CD22 ko tumor cells (Fig. 5j).Mice treated with bi-ChTCR T cells showed improved tumor clearance and survival compared to mice treated with monospecific or bispecific CAR T cells (Fig. 5k-m).We harvested tumor cells at euthanasia due to tumor progression to determine the expression of CD19 and CD22 on tumor cells (Fig. 5n-o).In control untreated mice, tumor cells were predominantly CD19 + CD22 + with a smaller frequency of CD19 + CD22 -and CD19 -CD22 + cells than in the initial tumor inoculum, illustrating a proliferative advantage for Nalm-6 WT tumor cells in vivo.Mice that received CD19 CAR T cells or Loop CAR T cells relapsed with predominantly CD19 -CD22 + tumor cells and minor populations of CD19 low CD22 + and CD19 -CD22 -tumor cells.In mice that received CD22 CAR T cells, the persisting tumor cells were predominantly CD19 + CD22 -with a small frequency of CD19 + CD22 +/low and CD19 -CD22 -cells.Only one mouse progressed and was euthanized from the group that received Bi-ChTCR T cells and in this case the tumor cells were predominantly CD19 -CD22 -with a small fraction of CD19 -CD22 Low cells.These data show that heterogeneity in antigen expression and antigen density limit antitumor activity after mono-and bispecific CAR T cell therapy, and these barriers to efficacy are overcome by sensitive Bi-ChTCR T cells.

Bi-ChTCR specific for multiple myeloma antigens
To determine whether the Bi-ChTCR architecture could be used to target multiple myeloma, we designed BCMA/SLAMF7 Bi-ChTCRs.Multiple permutations were tested including fusing a BCMA-targeting scFv to TRAC and a SLAMF7-specific scFv to TRBC, both in VH/VL orientations (format #1), switching the pairing of scFv and TCR chains (format 2), and placing either the two scFvs in VL/VH orientation on each TCR chain (format 3) or the BCMA scFv in VH/VL and SLAMF7 scFv in VL/VH (format 4) (Extended Data Fig. 7a).All 4 Bi-ChTCRs were expressed in TCR ko Jurkat cells at similar levels as measured by binding to rBCMA and SLAMF7, and assembled with CD3 as demonstrated by restored cell surface expression of CD3e (Extended Data Fig. 7b-e).These results illustrate the resilience of the ChTCR architecture for bispecific targeting of tumor antigens.We proceeded with a detailed analysis of format 1 since the VH/VL orientation provided a direct comparison to previously designed BCMA and SLAMF7 CARs (Extended Data Fig. 8a, Fig. 6a).The BCMA/SLAMF7 Bi-ChTCR and monospecific BCMA and SLAMF7 CARs were expressed in primary CD8+ T cells with simultaneous CBE of endogenous TRAC and TRBC and SLAMF7 to eliminate endogenous TCR expression and to avoid fratricide since SLAMF7 is expressed on some T cells [39] (Extended Data Fig. 8b-e).Gene edited SLAMF7 and BCMA/SLAMF7 Bi-ChTCR T cells expanded in culture and had the same viability as BCMA CAR T cells.Binding of soluble BCMA and SLAMF7 was significantly higher for T cells transduced with the monospecific CARs compared to the Bi-ChTCR (Fig 6b-d).Functional studies showed that BCMA/SLAMF7 Bi-ChTCR T cells recognized the MM cell line INA-6 expressing both BCMA and SLAMF7, and INA-6 engineered to express only a single antigen, whereas monospecific CAR T cells only recognized INA-6 cells that expressed their cognate antigen (Fig. 6e-g).Bi-ChTCR T cells also eliminated a mixture of heterogenous Nalm-6 target cells that were BCMA + SLAMF7 ko and BCMA ko SLAMF7 + in vitro, unlike monospecific CAR T cells (Extended Data Fig. 8g).Bi-ChTCR T cells exhibited more rapid and intense phosphorylation of Zap70 and LAT compared to monospecific BCMA and SLAMF7 CAR T cells (Fig. 6h and i).Importantly, despite the lower level of receptor expression, BCMA/SLAMF7 Bi-ChTCRs exhibited superior sensitivity for each antigen compared to monospecific CAR T cells, as demonstrated by higher levels of IFN-g secretion when cultured with a range of concentrations of plate-bound antigen (Extended Data Fig. 8h-i).Before analyzing the in vivo function of T cells expressing the BCMA/SLAMF7 Bi-ChTCR, we compared expression and signaling to a previously described BCMA/SLAMF7 bispecific CAR in primary T cells [40].We observed superior binding of BCMA and SLAMF7 to the Bi-ChTCR compared to the BCMA/SLAMF7 CAR (Extended Data Fig. 9a-c), and T cells expressing Bi-ChTCR exhibited more rapid and intense Zap70 phosphorylation than bispecific CAR T cells after stimulation with bead-coated BCMA and SLAMF7 alone, or together (Extended Data Fig. 9d,e).
We then asked whether BCMA/SLAMF7 Bi-ChTCR + T cells could eliminate a tumor inoculum comprised of Nalm-6 cells that were heterogeneous for BCMA and SLAMF7 expression in NSG mice (Fig. 6j).Bi-ChTCR + T cells rapidly eliminated tumor in all mice, while the bispecific CAR + T cells showed a moderate improvement in tumor control and survival over each of the monospecific CAR + T cell products (Fig. 6k-m).Bi-ChTCR T cells also demonstrated superior expansion in the blood during the period of tumor eradication compared to all other treatment groups, despite the absence of a costimulatory domain the receptor construct (Fig. 6n).We rechallenged mice that were tumor free after the administration of Bi-ChTCR with the same mixture of Nalm-6 cells.Control mice quickly developed tumors, whereas all mice from the Bi-ChTCR treated group were protected from tumor challenge (Fig. 6o-p).Collectively, the data demonstrates that Bi-ChTCRs can be designed to sensitively target pairs of B cell lineage antigens relevant to therapy of leukemia, lymphoma and multiple myeloma.

Discussion
The outgrowth of tumor cells that have downregulated or lost the target antigen is a major mechanism for the failure of CAR T cell therapies [6,8,11,18,41,42].Improving efficacy requires the development of synthetic receptors that are highly sensitive and capable of recognizing multiple tumor antigens.Bispecific CARs have been designed for this purpose but often exhibit reduced sensitivity for each individual antigen, enabling escape of tumor cells with low antigen levels [8,43].Here, we describe the design and function of chimeric TCRs that simultaneously target two tumor antigens with high sensitivity.The concept of linking an antigen binding domain to the TCR predates the design of current CARs, but was ineffective due to mispairing of ChTCR chains with endogenous TCR chains, which compromised the cell surface expression of the ChTCRs [31,44,45].We employed base editing to reproducibly and efficiently disrupt endogenous TRAC and TRBC expression.Base editing may be a safer alternative to conventional CRISPR for multiplexed gene editing as it does not induce double-strand DNA breaks that can lead to chromosome losses, translocations and/or recombination events [46,47].Base editing can also be used to disrupt expression of multiple genes in T cells without loss of editing efficiency, as shown by the simultaneous knockout of TRAC, TRBC, and SLAMF7 during generation of Bi-ChTCR T cells.The absence of endogenous TRAC and TRBC chains is required for optimal ChTCR expression to facilitate the assembly of all CD3 subunits with the ChTCR and provide diversity in ITAM sequences, which is crucial for optimal T cell activation [20,48].Recent studies have incorporated signaling portions from CD3e, d or g subunits into CARs to manipulate CAR T cell function.CD3e, in particular, tuned down T cell cytokine production, extended T cell persistence and prevented dysfunction in vivo [20,49,50].However, the antigen sensitivity of these CD3e modified CARs was not evaluated and tuning of T cell activation might reduce antigen sensitivity.The importance of recapitulating the TCR structure is illustrated by our synapse studies, which revealed that ChTCRs, unlike CARs, form a TCR-like bull's eye synapse with ligand-functionalized lipid bilayers [51].Immune synapses play a pivotal role in regulating T cell activation, from signal initiation, propagation and termination [52].This regulation is likely to be preserved in ChTCR T cells that assemble with all CD3 chains and may be important in achieving optimal antitumor responses after ACT.The full ChTCR format, in which an scFv is linked to a single TCR constant chain, demonstrated superior antigen sensitivity in vitro and antitumor activity in vivo compared to the split ChTCR format [30,31].We leveraged the superior monospecific full ChTCR format to design bispecific ChTCRs that target CD19 and CD22 or BCMA and SLAMF7.Given the absence of rules for designing optimized ChTCRs, we constructed monospecific CD22 ChTCRs with two CD22 scFvs that had similar binding affinities (m971 and 9A8) but were specific for membrane-proximal and membrane-distal epitopes, respectively [38,53].Surprisingly, while T cells expressing ChTCRs using each scFv bound soluble CD22 protein, only T cells expressing the 9A8 ChTCR formed an organized immune synapse and recognized CD22 positive tumor cells.In contrast, the m971 scFv specific for a membrane proximal epitope in CD22 was superior for CAR T cells [54].This further highlights differences between CARs and ChTCRs that may be related to distinct synaptic distance requirements between the two receptor classes [55,56].Unlike CARs, ChCTRs do not possess a hinge domain between the TCR constant chains and the scFv, which could limit the flexibility necessary to engage the membrane-proximal epitope of a bulky protein like CD22.These findings underscore the need to define rules for optimal ChTCR design, informed by functional screening of scFv binders of known epitope specificity to facilitate structural analysis.T cells expressing Bi-ChTCRs targeting CD19 and CD22 or BCMA and SLAMF7 recognized target cells expressing either one or both target proteins and exhibited superior sensitivity for tumor cells with low antigen levels compared to T cells expressing monospecific and bispecific CARs.In vivo models that mimic therapy of tumors with heterogeneous antigen expression showed improved efficacy of Bi-ChTCR T cells compared to bispecific CAR T cells targeting the same antigens [37,40].Bispecific targeting with CARs can also be achieved with a bicistronic vector or by the infusion of two mono-specific CAR products either at the same time or sequentially.These strategies weren't formally compared here but would not be predicted to provide better antigen sensitivity, and in the case of two products, add manufacturing complexity and cost [43,57].Surprisingly the superior in vivo efficacy of the Bi-ChTCR T cells was observed in the absence of providing additional costimulation either in trans or intrinsic to the ChTCR [30,31,58].Whether providing costimulation to Bi-ChTCR T cells would further improve efficacy remains to be evaluated.It is notable that mono and Bi-ChTCR + T cells secrete lower levels of cytokines compared to conventional CD28z CAR T cells in response to high antigen expressing tumor cells.High cytokine levels correlate with the severity of cytokine release syndrome and neurotoxicities observed with CAR T cell therapy [59].It is therefore reasonable to expect a better toxicity profile with ChTCRs.ChTCR + T cells produce higher levels of cytokines and proliferate better against low antigen tumor cells than CAR T cells, enabling the antitumor response to be sustained and eradicate antigen low tumor cells.Collectively, these data identify a new approach for sensitive and potent recognition of two target antigens with a single engineered T cell product that holds promise for reducing antigen escape and relapse in B cell malignancies and multiple myeloma.This strategy may also be applicable to solid tumors where tumor heterogeneity is prevalent and antigen levels may be lower than in B-cell malignancies.However, unlike B cell lineage antigens, many targets in solid tumors are also expressed on normal epithelial tissues, and receptors that are too sensitive may cause on target off tumor toxicity.Careful selection of target antigens with well-defined tumorrestricted expression profile will be essential to apply Bi-ChTCRs in solid tumors.Split ChTCR Full ChTCR       h.δ  and polybrene (Millipore, TR-1003-G) at a final concentration 4.4µg/mL.T cells were spinoculated at 800g, 32° C for 90 min, after overnight incubation beads removed prior to gene editing.Cytidine base editing was performed to knock-out expression of endogenous TRAC, and TRBC, both, and SLAMF7 in some experiments.1x10 6 T cells were resuspended in P3 buffer (Lonza, V4XP-3032), mixed with 1µg concentration of RNA guide and 1.5µg of CBE BE4max mRNA (Addgene plasmid 112093) (Aldevron), and electroporated using the Lonza 4D device (Lonza) [9].Sequences for sgRNAs are described in Table S1.T cells were cultured in CTL supplemented with IL-2 (150IU/mL), IL-7 (5ng/ml), IL-15 (5ng/mL) initially, and then maintained in CTL supplemented with IL-2 (50IU/mL) for one week before being used for assays.For assays requiring larger cell numbers (western blot analysis and Calcium flux assays), flow sorted CAR and ChTCR specific T cells were expanded using OKT3 (30ng/mL) or PHA-L (500x) (Thermo Fisher Scientific, 00-4977-93), g-irradiated lymphoblastoid cell line (LCL) (8000 rad), and girradiated PBMCs at a LCL to T cell ratio (100:1) and PBMC to T cell ratio (600:1).IL-2 was added 24-hours after co-culture, and OKT3 or PHA-L was washed out on day 4. Cultures were fed with CTL supplemented with IL-2 (50IU/mL) and rested without IL-2 addition before use in assays.
Generation and fluorescent labeling of extracellular protein domains and density quantification 12X His-tagged CD19 (N138Q) and ICAM-1 extracellular domains were produced using the "Daedalus" mammalian expression system [10].Briefly, HEK293F cell were transduced with lentivirus expressing CD19(N138Q)-12x His tag or ICAM-1-12x his tag constructs.Proteins were captured from expression culture supernatant by HisTrap FF crude (Cytiva, 11000458) Niaffinity chromatography and polished by Superdex 200 (Cytiva, 28-9909-44) size exclusion chromatography.Purified proteins were flash frozen in liquid nitrogen in 1X PBS and stored at -80 °C.Extracellular domain of CD22-10x His tag was commercially available (Acrobiosystems, CD2-H52H8).For labeling, 100ug of His-tagged extracellular proteins were concentrated to 1mg/mL using Amicon Ultra Centrifugal Filters (Milipore, UFC500396), and pH adjusted to 8.3 by addition of NaHCO3.Proteins were incubated with Alexa Fluor 488 TFP ester, Alexa Fluor 647 NHS ester or Alexa Fluor 555 NHS ester (ICAM_AF488, CD19_AF647, CD22_AF555, CD22_AF647) at room temperature for 15 minutes.Non-reactive dye was removed by sizeexclusion chromatography, followed by repeat buffer exchange with Amicon Ultra Centrifugal Filters.Aliquots of protein were frozen and stored at -80° until use.Molecular density of extracellular domains in the soluble lipid bilayer (SLB) was determined using SLB-coated silica beads as previously described [11].Briefly, silica microspheres (Bangs Laboratories Inc, SS05003) equaling the surface area of a single 96-well chamber were washed and resuspended in PBS.Beads were incubated with lipids (as described in section "soluble lipid bilayer generation and TIRF imaging") and washed following the same steps as bilayer preparation on 96-well plate.His-tagged proteins were serially diluted and incubated with SLB-coated silica beads for 30 min with gentle shaking.Beads were run on a flow cytometry machine (BD FACSCelesta) along with Quantum Alexa Fluor 488 MESF or Quantum Alexa Fluor 647 MESF (Bangs Laboratories Inc, 647A/488A).Degree of labeling of proteins was determined by 280 and fluorescent absorption, and MESF standards were used to determine the absolute molecular density of proteins on silica beads.

Cell stimulation and western blot
Beads for T cell stimulation were prepared as previously described [14].2x10

Cytokine measurement
T cell cytokine release was determined by measuring cytokine concentration in the supernatant after 18-24 hours of co-culture of ChTCR and CAR T cells with target cells or with plate-bound antigen.For target cell simulation, T cells and target cells were co-cultured at a 1:1 E:T ratio.After antigen stimulation, supernatant was harvested and cytokine concentration was determined by ELISA according to kit manufacture protocol: IL-2 (BioLegend, 431816) IFN-g (BioLegend, 430116).For plate-bound antigen stimulation, 96-well plates were coated with avidin (10µg/mL) overnight and incubated with PBS + 3% BSA to block non-specific protein binding.Avidin-coated plates were then coated with biotinylated extracellular protein domains for one hour at specific concentrations.50,000 T cells were resuspended in 50µL CTL and transferred to antigen coated plates for incubation.

Cell proliferation assay
ChTCR and CAR T cells were harvested and washed with warm PBS.Cell Trace Violet (CTV) Cell Proliferation dye (Invitrogen, C34557) was resuspended in 200µL DMSO.T cells were resuspended in 1 mL PBS and incubated with 2µL CTV for 10 minutes at 37° C with periodic mixing.1mL FBS was added to absorbed unbound dye, cells were washed and resuspended in CTL.T cells were co-cultured with Nalm-6 target cells expressing appropriate target antigen at 1:2 E:T ratio for 72 hours.Cells were harvested, washed with PBS, and stained with anti-CD8-APC antibody (Biolegend, 344722) before acquisition by flow cytometry.

Cytotoxicity assay
Target tumor cells were incubated with Cr 51 overnight, washed, resuspended in culture media, and plated with effector T cells to achieve indicated E:T ratios.Plates were briefly centrifuged (100rpm for 1 minute), then incubated for 4 hours.After incubation, 30µL of supernatant was harvested, transferred to LumaPlates (Revvity, 6006633), and plates were dried overnight.
Plates were read by scintillation counter and percent specific lysis was calculated using the standard formula.

NSG mouse tumor model
6-8-week-old, female NOD/SCID/gc -/-mice were purchased from Jackson Laboratory or bread in-house.For the Raji model, mice were engrafted with 0.5 million Raji/GFP-ffluc intravenously by tail vain injection.For the Nalm-6 models, mice were engrafted with 0.5 million Nalm-6 WT GFP-ffluc or 1 million Nalm-6 GFP-ffluc cells that expressed low levels of CD19 antigens by tail vain injection.For experiments with bi-specific receptors, a heterogenous mixture of Nalm-6 WT and Nalm-6 cells expressing only a single antigen were engrafted.Antigen expression in all tumor lines was checked by flow cytometry prior to injection.Mice were injected intracenously 7 days (Raji/GFP-ffluc) or 4 days (Nalm-6/GFP-ffluc) after tumor inoculation with ChTCR of CAR modified CD8 + and CD4 + T cells at a 1:1 ratio or with PBS.Cell numbers were normalized based on total number of receptor positive cells in the total cell population, as determined by flow cytometry prior to infusion.Mice were followed by bioluminescence imaging after intraperitoneal injection of luciferin substrate using the Xenogen IVIS Imaging System (Caliper Life Sciences) and for survival.Living Image Software V4.7.3 (Caliper Life Sciences) was used to analyze luciferase activity and photon flux within regions of interest that encompassed the entire body of each individual mouse.Blood was obtained from mice at various timepoints, single-cell suspensions from peripheral blood were prepared by lysing red blood cells using ammonium-chloride-potassium (AKC) lysing buffer (Quality Biological, 118-156-101).Single-cell suspensions were stained with the following antibody panel for flow cytometry analysis; Nalm6-GFP, anti-CD45-PE (Biolegend, 304008), anti-CD8-BUV805 (BD, 612889), anti-CD4-cflour R840 (Cytec, R7-20165), rBCMA-biotin (Accro Biosystems, BCA-H82E4), rSLAMF7-biotin (Accro Biosystems, SL7-H82E0), streptavidin-APC (Invitrogen, 17-4317-82).

Figure 1 :
Figure 1:Chimeric TCRs expressed in T cells reproduce canonical TCR structure, synapse formation and proximal signaling

Figure 1 .
Figure 1.Chimeric TCRs expressed in T cells reproduce canonical TCR structure, synapse formation and proximal signaling

Figure 2 .Figure 2 .
Figure 2. T cells expressing the CD19-specific Full ChTCR recognize CD19 low tumor cells and have superior anti-tumor effect in vivo

Figure 5 .Figure 5 .
Figure 5. T cells expressing the CD19/CD22 Bi-ChTCR have exquisite sensitivity for both antigens and potent anti-tumor activity.