Phospho-mimetic CD3ε variants prevent TCR and CAR signaling

Introduction Antigen binding to the T cell antigen receptor (TCR) leads to the phosphorylation of the immunoreceptor tyrosine-based activation motifs (ITAMs) of the CD3 complex, and thereby to T cell activation. The CD3ε subunit plays a unique role in TCR activation by recruiting the kinase LCK and the adaptor protein NCK prior to ITAM phosphorylation. Here, we aimed to investigate how phosphorylation of the individual CD3ε ITAM tyrosines impacts the CD3ε signalosome. Methods We mimicked irreversible tyrosine phosphorylation by substituting glutamic acid for the tyrosine residues in the CD3ε ITAM. Results Integrating CD3ε phospho-mimetic variants into the complete TCR-CD3 complex resulted in reduced TCR signal transduction, which was partially compensated by the involvement of the other TCR-CD3 ITAMs. By using novel CD3ε phospho-mimetic Chimeric Antigen Receptor (CAR) variants, we avoided any compensatory effects of other ITAMs in the TCR-CD3 complex. We demonstrated that irreversible CD3ε phosphorylation prevented signal transduction upon CAR engagement. Mechanistically, we demonstrated that glutamic acid substitution at the N-terminal tyrosine residue of the CD3ε ITAM (Y39E) significantly reduces NCK binding to the TCR. In contrast, mutation at the C-terminal tyrosine of the CD3ε ITAM (Y50E) abolished LCK recruitment to the TCR, while increasing NCK binding. Double mutation at the C- and N-terminal tyrosines (Y39/50E) allowed ZAP70 to bind, but reduced the interaction with LCK and NCK. Conclusions The data demonstrate that the dynamic phosphorylation of the CD3ε ITAM tyrosines is essential for CD3ε to orchestrate optimal TCR and CAR signaling and highlights the key role of CD3ε signalosome to tune signal transduction.


Introduction
The T cell antigen receptor (TCR) is a multimeric protein complex composed of the antigen-binding TCRab heterodimer and the noncovalently associated CD3 proteins, including the heterodimers CD3de and CD3ge, and the zz homodimer (1,2).Each CD3 subunit contains Immunoreceptor Tyrosine-based Activation Motifs (ITAMs) that mediate signal transduction.The cytoplasmic tails of CD3e, CD3d, and CD3g each contain one ITAM, whereas z contains three ITAMs (3).Among the CD3 subunits, CD3e is unique, since in addition to the ITAM, it contains other protein-protein interaction motifs in its cytoplasmic tail.The N-terminally located Basic Rich Sequence (BRS) recruits the Src family kinase LCK via LCK's unique domain (4).Additionally, CD3e interacts with the inner leaflet of the plasma membrane through the BRS (5).The Proline Rich Sequence (PRS) recruits the non-catalytic region of tyrosine kinase adaptor protein (NCK) via NCK's first Src-homology 3 (SH3.1)domain and stabilizes the interaction of LCK with the TCR (6,7).The recently discovered Receptor Kinase (RK) motif binds LCK via its SH3 domain (8).LCK and NCK form part of the signalosome of the un-phosphorylated CD3e, but only when the TCR is in its active conformation, stabilized by antigen binding (6,8,9).
Although the structure of the ecto-and transmembrane-parts of the complete, resting TCR was resolved, a structure of all the cytoplasmic tails together remains an enigma (10).Biochemical experiments suggest that the TCR, including the cytoplasmic tails, switches between various conformational states (6,(11)(12)(13).Upon TCR engagement by agonist peptides bound to major histocompatibility complex (pMHC), the active TCR conformation is stabilized.This conformational change leads to the exposure of the ITAM, the PRS and the RK motif of CD3e (6,8,9).Thereby, LCK is recruited via its SH3 domain to the RK motif of CD3e, placing LCK in close proximity to the TCR allowing phosphorylation (8).
Dual phosphorylation of the ITAMs leads to the recruitment of the z-chain-associated protein kinase 70 (ZAP70) (14)(15)(16).A binding hierarchy of the tandem SH2 domain of ZAP70 to the ITAMs of z and CD3e has been reported.The affinity is highest to the membrane proximal ITAM of z, followed by the second ITAM of z, and finally by the ITAM of CD3e and the most distal ITAM of z (16,17).Once bound to the TCR, ZAP70 becomes phosphorylated and activated.As a result, ZAP70, along with LCK, phosphorylates downstream signaling molecules to induce TCR-controlled signaling cascades resulting in T cell activation (18).
Since NCK and LCK recruitment to the antigen-bound TCR precedes ITAM phosphorylation, we aimed here to systematically investigate how the phosphorylation of each of the individual tyrosines of the CD3e ITAM impacts the interaction with NCK, LCK and ZAP70.We mimicked tyrosine phosphorylation by mutating the tyrosines of the CD3e ITAM to the negatively charged glutamic acid.The suitability of this approach was confirmed by taking advantage of the well-described phosphorylation-sensitive binding patterns of ZAP70 and NCK to CD3e.In a step-wise approach, we demonstrated that exchanging each of the CD3e ITAM tyrosines to glutamic acid mimics irreversible phosphorylation and prevents LCK binding.We demonstrate, that in the context of the TCR complex, dynamic CD3e ITAM phosphorylation is required for optimal TCR signaling and ZAP70 phosphorylation, and that CD3e phospho-mimetic variants of Chimeric Antigen Receptors (CARs) prevent CARmediated signal transduction.
Production of lentiviruses 10 7 HEK293T cells were transfected using PEI transfection with the appropriate constructs and the packaging plasmids pCMVR8.74(gag/pol) and pMD2.G (envelope).The virus-containing supernatant was collected 24 h and 48 h after transfection.Viruscontaining supernatant was concentrated for 4 h at 10.000 xg and 6°C using a 10% sucrose gradient (supplemented with 0.5 mM EDTA).The concentrated virus was resuspended in 100 ml 0% FBS RPMI and frozen at -80°C until use.

Primary human T cell activation, transduction, and expansion
Cryopreserved purified peripheral blood mononuclear cells (PBMCs) from healthy donors were thawed and resuspended in 10% FBS RPMI supplemented with 500 U/ml recombinant human IL-2 (PeproTech) and activated with plate-bound anti-CD3 (UCHT1, homemade) and anti-CD28.2(BioLegend) (1 mg/ml).48-72 h after activation, >99% of the cells were T cells.Activated primary human T cells were lentivirally transduced using spin infection with a virus MOI of 4 together with 5 mg/ml of protamine sulfate (Sigma) and 500 U/ml of recombinant human IL-2.8-12 days after transduction, CAR surface levels were determined by flow cytometry.After transduction, cells were grown in 10% FBS RPMI supplemented with 100 U/ml IL-2 before being used for the indicated experiments.

Cell stimulation and lysis
For flow cytometry analysis, 1.5-2x10 5 cells were resuspended in 200 ml 10% FBS RPMI.Cells were either left unstimulated, stimulated with 5 mg/ml plate-bound anti-CD3e antibody (UCHT1), stimulated with Nalm6 cells in an effector to target ratio (E:T) of 1:1, or stimulated with 10 ng/ml phorbol myristate acetate (PMA) and 250 ng/ml ionomycin.Surface levels of activation markers were assayed after 24 h of stimulation.For pull-down experiments 30x10 6 JK cells per sample were lysed in 1 ml of EMBO lysis buffer containing 20 mM Tris-HCl (pH 8), 137 mM NaCl, 2 mM EDTA, 10% glycerol, 1 mg/ml protease inhibitor cocktail (Sigma), 1 mM PMSF, 5 mM iodoacetamide, 0.5 mM sodium orthovanadate, 1 mM NaF and 0.3% Brij96V for 1 h on ice.For immunoblotting, 2x10 6 JK cells per sample were resuspended in 100 ml of 0% FBS RPMI and starved for 1 h at 37°C.Cells were then stimulated with 5 mg/ml anti-CD3e antibody (UCHT1) for 5 min at 37°C . Cells were lysed in 100 ml of EMBO lysis buffer for 30 min on ice, followed by a 20 min centrifugation at 12.000 g, then the supernatant was stored at -20°C until use.

Pull-down assay and Immunoblotting
For the PD assay, the different phospho-mimetic variants of cytoplasmic tails of human CD3e were cloned into the pRP261 vector and fused C-terminally to GST.The E.coli strain BL21 was transformed with the corresponding vectors.After 3 h induction with 1 mM IPTG, bacteria were lysed.GST-CD3e fusion proteins were purified using glutathione sepharose beads (GE Healthcare).CD3e PD assay was performed overnight at 4°C.After washing, proteins were separated by SDS-PAGE and immunoblotted in a semi-dry chamber for 1 h at 18 V.Chemiluminescence was used to image protein bands using a CCD camera (ImageQuant LAS 4000; GE Healthcare).ImageJ and ImageQuantTL software were used to determine the relative band intensity (GE Healthcare).
Enzyme-linked immunosorbent assay 2x10 5 cells incubated in 200 ml 10% FBS RPMI were either left unstimulated or stimulated with 5 mg/ml plate-bound anti-CD3e antibody (UCHT1) and 1 mg/ml soluble anti-CD28 antibody, stimulated with Nalm6 cells using a 1:1 ratio, or stimulated with 10 ng/ml PMA and 250 ng/ml ionomycin.After 24 h, the supernatants were collected and cytokine levels determined using commercial ELISA Kits (eBioscience).

Intracellular calcium influx
5x10 5 cells per sample were resuspended in 1% FBS RPMI and incubated for 30 min at 37°C with 5 mg/ml of Indo-1 and 0.5 mg/ml of pluronic F-127 (Molecular Probes).After washing, cells were kept in the dark on ice.Right before measurement, cells were prewarmed 5 min at 37°C.The Ca 2+ response was induced by addition of 0.5 mg/ml anti-CD3e (UCHT1) after recording 60 s of baseline.The change of the ratio Indo-bound versus Indo-unbound was measured with a MaxQuantX Flow Cytometer (Miltenyi Biotech).Data were normalized to the baseline with FlowJo software version 9.3.2.
In situ proximity ligation assay 1x10 5 cells per sample were starved in 0% FBS RPMI and rested on diagnostic microscope slides (Thermo Fisher Scientific) for 1 h at 37°C.Cells were left unstimulated or stimulated with 5 mg/ml anti-CD3e (OKT3) for 5 min at 37°C.Cells were then fixed in 2% PFA for 15 min, permeabilized with 0.5% saponin for 30 min and blocked.Blocked cells were stained according to the manufacturer's instructions with the Duolink kit (Olink Bioscience) with goat anti-CD3e (Everest Biotechnology) and mouse anti-LCK (3A5, Santa Cruz Biotechnology).Nuclei were stained with DAPI (Roth).A total of 5-7 images (on average 700 cells) per sample were taken at ×60 with a confocal microscope (Nikon C2) and analyzed with BlobFinder.
Bioluminescence-based cytotoxicity assay 3x10 4 primary CAR T cells were cultivated in 10% FBS RPMI supplemented with 75 mg/ml D-firefly luciferin potassium salt (Biosynth) and co-cultured with luciferase-expressing Nalm6 tumor cells using a 1:1 ratio.Bioluminescence (BLI) was measured in a luminometer (BioTek Synergy H4 Hybrid Reader) as relative light units (RLUs).Maximal cell death was defined by the RLU signals from cells treated with 1% Triton X-100 and unspecific tumor cell lysis was set by the RLU signals from target cells incubated with non-transduced T cells.Percent specific tumor cells lysis (specific killing) was calculated with the following formula: % specific tumor cell lysis = 100 × (average nontransduced T cell killing RLU − test RLU)/(average nontransduced T cell killing RLU − average maximal death RLU).

Degranulation assays
2x10 5 CAR T cells were cultured with Nalm6 cells using a 1:1 ratio in the presence of 1 ml of PE-labelled anti-CD107a antibody for 3 h.Cells were then stained with a PE-labelled anti-CD107a antibody and surface levels were assessed by flow cytometry.

LCK binding to recombinant CD3ϵ is abolished by mimicking irreversible phosphorylation of Y50E
It has been previously shown that LCK binds with its SH3 domain to the un-phosphorylated cytoplasmic tail of CD3e (8).It was demonstrated using phosphorylated peptides that this binding was reduced when the CD3e ITAM was doubly phosphorylated.Molecular modeling predicted that the OH group of the second tyrosine (Y50, numbered based on the sequence shown in Figure 1A) of the CD3e ITAM forms a hydrogen bond with a serine from LCK (8).This interaction should be abolished upon Y50 phosphorylation, explaining why the SH3(LCK) domain failed to bind the phosphorylated cytoplasmic tail of CD3e.We aimed here to experimentally confirm this hypothesis and to systematically assay the role of each CD3e ITAM tyrosines for LCK binding.
To mimic phosphorylation, we mutated each CD3e ITAM tyrosine (Y), individually or both simultaneously, to glutamic acid (E) (Figure 1A).We generated GST-fusion proteins containing the different phospho-mimetic variants of the cytoplasmic tail of CD3e (WT, Y39E, Y50E, Y39/50E) and used them for pull-down assays with the lysate of Jurkat (JK) T cells (Figure 1B).Using phospho-peptides, it has been demonstrated that ZAP70 binds with its tandem SH2 domain to doubly phosphorylated ITAMs (8,19).In our experiments, ZAP70 bound significantly better to the CD3e Y39/50E variant than to WT CD3e or the single mutants (Figures 1B, C).NCK recruitment to the TCR occurs via the interaction between the SH3.1 domain of NCK and the PRS of CD3e ( 6), and phosphorylation of Y39 interrupts this interaction as demonstrated using phosphorylated peptides (20,21).As expected, NCK bound to CD3e WT and CD3e Y50E, but not to the Y39E or Y39/50E variants in our experiments (Figures 1B, D).Binding to CD3e Y50E was enhanced compared to CD3e WT, being in line with data suggesting that the SH2 domain of NCK binds to phosphorylated Y50 (21).Hence, we show here that mutation of tyrosine to glutamic acid is able to mimic tyrosine phosphorylation of the CD3e ITAM, confirmed by the specific binding pattern of ZAP70 and NCK.LCK binding to CD3e was reduced by 80% for CD3e Y50E and Y39/50E variants compared to CD3e WT, confirming our previous hypothesis that phosphorylation of Y50 abolishes LCK binding to CD3e (Figures 1B, E).Altogether, these results show that LCK binding to the cytoplasmic tail of CD3e is phosphorylationsensitive, suggesting that Y50 acts as a molecular switch regulating the interaction of the full-length LCK with CD3e.

Phospho-mimetic variants of the CD3e ITAM tyrosines reduce T cell activation
Next, we aimed to investigate whether mimicking irreversible phosphorylation of the CD3e ITAM tyrosines impacts signal transduction in the context of the complete TCR.To this end, we expressed four variants of CD3e (WT, Y39E, Y50E, Y39/50E) in Jurkat (JK) T cells with a CRISPR/Cas9 knock-out for CD3e (JK eKO) (Figure 2A).TCR and CD28 cell surface levels were equal in all generated cell lines (Supplementary Figure 1).
The phospho-mimetic CD3e variants did not transduce activation signals in the absence of stimulus, demonstrated by the lack of CD69, CD25 and 4-1BB up-regulation, as well as IL-2 secretion under tonic conditions (Figures 2B-D).Upon stimulation with an anti-CD3 antibody for 24 hours, JK T cells expressing CD3e WT up-regulated CD69, CD25 and 4-1BB (Figures 2B, C).The up-regulation of these activation markers was reduced by 10-40% in JK T cells expressing the phospho-mimetic CD3e variants.IL-2 production in JK T cells expressing the phospho-mimetic CD3e variants was strongly reduced (60%) when compared to their WT counterpart (Figure 2D).These results suggest that irreversible phosphorylation of the CD3e ITAM could impact optimal T cell activation.However, it also suggests some redundancy between the ten ITAMs of the TCR-CD3 complex, because in none of the phospho-mimetic variants signaling was completely abolished.The multitude of ITAMs might contribute to signal amplification and/or diversity.Nevertheless, evidences support that some degree of redundancy between the CD3 ITAMs is tolerated during T cell development (22).
We next tested for intracellular Ca 2+ mobilization upon TCR triggering.The CD3e Y50E variant slightly, but consistently, reduced the percentage of Ca 2+ responding cells (Figure 2E).This result is in line with our previous study showing that reducing LCK recruitment by mutating the RK motif, leads to a slight reduction of Ca 2+ responding cells (8).Expression of the Y39E and the Y39/50E variants resulted in a stronger reduction of the percentage of responding cells when compared to the CD3e WT (Figure 2E), suggesting that abolishing NCK recruitment, leads to a stronger reduction of Ca 2+ responding cells as previously proposed by mutating the PRS motif (8).
Next, we tested whether the phospho-mimetic CD3e variants are able to activate the T cell transcription factors NFkB and NFAT.Therefore, we used JK-derived reporter cells, in which GFP is expressed under the control of NFkB or NFAT (23) (Figure 2F), after performing a CRISPR/Cas9 knock out of CD3e to prevent signaling through the endogenous TCR.The efficiency of the eKO was confirmed by flow cytometry and western blot analysis (Supplementary Figures 2A-C).Then we transduced the reporter cell lines with the respective CD3e variants and confirmed restored TCR surface levels by flow cytometry (Supplementary Figures 2D-M).Activation for 24 h with an anti-CD3 antibody resulted in activation of NFkB in the JK NFkB CD3e WT cells, indicated by the increase of GFP + cells.Expression of the CD3e Y50E and Y39/50E variants reduced this activation by up to 50% (Figure 2G).In the JK NFAT reporter cells, GFP expression was reduced upon expression of the CD3e Y39E and Y39/50E variants compared to WT, while the CD3e Y50E variant did not affect NFAT activation in line with our Ca 2+ influx results (Figures 2E, G).
To deeper investigate the mechanisms behind the reduced signal transduction of the phospho-mimetic CD3e variants, we performed an in situ proximity ligation assay (PLA) between the TCR (CD3e) and LCK.To this end, the JK eKO cells reconstituted with the CD3e WT, Y39E, Y50E or Y39/50E construct were either left unstimulated or stimulated with an anti-CD3 antibody.Upon stimulation, the proximity of the TCR and LCK strongly increased in the CD3e WT expressing cells (Figures 3A, B).The proximity between the TCR and LCK was slightly increased by the Y39E variant, while the Y50E variant reduced TCR-LCK proximity by 60%.The TCR-LCK proximity for the Y39/50E variant was reduced by 30% compared to the WT, which might be explained by binding of LCK to other phosphorylated ITAM motif, beyond CD3e, with its SH2 domain (Figures 3A, B).This pattern was also observed in our previous study, where reduced NCK recruitment by a PRS depletion did not affect TCR-LCK proximity, while reducing LCK recruitment by mutating the RK motif, strongly reduced TCR-LCK proximity compared to CD3e WT (8).LCK activity is regulated by the phosphorylation status of its regulatory tyrosines Y394 and Y505 (24).Phosphorylation of these tyrosine residues was unaffected by the phospho-mimetic CD3e variants (Supplementary Figure 3).We next studied the phosphorylation status of ZAP70.LCK phosphorylates ZAP70 at Y315 and Y319 to unlock the autoinhibited conformation of ZAP70.This stabilizes its binding to phosphorylated ITAMs and facilitates catalytic activity and downstream signaling to activate T cells (25).JK eKO cells expressing CD3e WT or the phospho-mimetic CD3e variants were left unstimulated or stimulated with an anti-CD3 antibody.Total cell lysates were analysed for ZAP70 phosphorylation at Y319.JK eKO cells were used as negative control.In JK eKO cells reconstituted with CD3e WT, Y319 phosphorylation of ZAP70 increased upon TCR triggering (Figures 3C, D).In contrast, the Y319 phosphorylation of ZAP70 upon stimulation was significantly decreased by 30-50% in cells expressing the phospho-mimetic CD3e variants (Figures 3C, D).
Altogether, mimicking the irreversible phosphorylation of the CD3e ITAM reduces phosphorylation of ZAP70, and thereby prevents efficient downstream signaling and T cell activation.We suggest that each CD3e phospho-variant prevents optimal TCR activation by different means.Y39E allows the binding of LCK, but the premature recruitment of CSK ( 26) might prevent the catalytic activity of LCK.Y50E can bind NCK but not LCK and therefore is unable to transduced effective signaling.Y39/50E can bind ZAP70, however, in the absence of LCK binding, ZAP70 might not be efficiently activated.Importantly, these data suggest a specific and non-redundant role for the CD3e ITAM in signal initiation, but also highlights some degree of compensation by any of the additional ITAMs present in the TCR-CD3 complex.

Phospho-mimetic variants of a CD3econtaining chimeric antigen receptor impair T cell activation
We next designed minimalistic TCRs in the form of CD3econtaining CARs to solely investigate the CD3e ITAM and to avoid compensation by other ITAMs, as it is the case in the context of the complex.To this end, we engineered CAR constructs by exchanging the z to the CD3e cytoplasmic tail in the FDA-approved second-generation CAR containing the co-stimulatory domain of 4-1BB (CD137) and targeting CD19.We have previously shown that a CD3e containing CAR outperformed the FDA-approved z CARs in a preclinical model (27).In addition to CD3e WT, we have here generated CARs containing the phospho-mimetic variants of CD3e.These CARs will be named from now on 19BBe WT, 19BBe Y39E, 19BBe Y50E and 19BBe Y39/50E (Figure 4A).All constructs were expressed on the surface of JK T cells, which neither express CD4 nor CD8 (8), and all express CD28 similarly, as verified by flow cytometry (Supplementary Figure 4).
Strong tonic signaling is a major limitation of CARs that can lead to early CAR T cell exhaustion and to excessive cytokine secretion (28).We did not observe increased tonic signaling when the phospho-mimetic 19BBe variants were compared to the WT neither by CD69, CD25 or 4-1BB up-regulation, nor by IL-2 secretion (Figures 4B-D).Next, all JK 19BBe variants were coincubated at a 1:1 ratio with the CD19 + B cell precursor leukaemia cell line Nalm6 for 24 hours.Cells expressing the 19BBe WT CAR were effectively activated, as shown by up-regulating CD69, CD25 and 4-1BB, as well as by IL-2 secretion (Figures 4B-D).In contrast, all phospho-mimetic CAR variants failed to transmit activation signals as seen by lack of up-regulation of CD69, CD25 and 4-1BB as well as IL-2 secretion above background (Figures 4B-D).
To deeper investigate the signaling defects observed by the phospho-mimetic 19BBe variants, we transduced the JK NFkB eKO and JK NFAT eKO reporter cell lines with a mock construct or the CAR variants.All 19BBe CARs were expressed on the cell surface of the reporter cell lines (Supplementary Figure 5).It is well known that the tumor necrosis factor receptor (TNFR) family, such as 4-1BB, activates the NFkB pathway (29).Indeed, all 19BBe variants activated NFkB as seen by the increase in GFP + cells in the absence of ligand compared to the mock-expressing control cells ( Figures 4E, F).In contrast, no tonic activation of NFAT was observed ( Figures 4E, F).Upon co-incubation with CD19 + Nalm6 cells, 19BBe WT CARs significantly activated the transcription factors NFkB and NFAT (Figures 4E, F).In comparison with the 19BBe WT CAR, the three phospho-mimetic CAR variants (Y39E, Y50E, Y39/50E) failed to further increase 4-1BB-mediated NFkB activation (Figures 4E, F).Likewise, all phospho-mimetic 19BBe variants prevented activation of the transcription factor NFAT upon ligand binding to the CARs (Figures 4E, F).Taken together, these results indicate that mimicking constitutive tyrosine phosphorylation of the CD3e ITAM prevents signal transduction by CARs and thus, T cell activation in JK cells.

Phospho-mimetic variants of CD3e prevent CAR activation in primary human T cells
We next investigated impact of the phospho-mimetic 19BBe variants in a more physiological and translational-orientated system by expressing them in primary human peripheral blood T cells (Figure 5A).All CARs were expressed at similar levels as assayed by flow cytometry (Supplementary Figure 6).Tonic signaling was comparable between the phospho-mimetic 19BBe variants and WT CAR as shown by neither CD69, CD25 nor 4-1BB up-regulation under basal conditions (Figures 5B-D).Triggering of the 19BBe WT CAR by co-incubation of primary CAR T cells with CD19 + Nalm6 cells efficiently induced T cell activation, with a significant increase in the proportion of cells expressing CD69, CD25 and 4-1BB (Figures 5B-D).In marked contrast, all phospho-mimetic 19BBe variants failed to optimally induce T cell activation (Figures 5B-D).Stimulation with PMA and ionomycin, bypassing the membrane proximal signaling events of the CAR, resulted in maximal CD69 upregulation in all 19BBe CAR variants (Figure 5B).This control demonstrates that all CAR T cells have the same capacity to upregulate CD69, excluding intrinsic differences in T cell differentiation induced by the individual CAR constructs.These results substantiate our hypothesis that irreversible phosphorylation of 19BBe CARs impairs signal transduction upon CAR triggering.

Reduced and delayed cytotoxicity as well as cytokine secretion by primary CAR T cells expressing CD3e phospho-mimetic CAR variants
CAR T cells mediate their anti-tumoral effects through directly killing tumor cells, as well as by the release of cytokines to sensitize the tumor stroma.Thus, we investigated the cytotoxic activity of primary CAR T cells expressing our novel 19BBe CARs.CAR T cells mediate cytotoxicity mainly through the perforin and granzyme axis.Therefore, we first tested for the degranulation of primary CAR T cells upon encounter with tumor cells.This process results in the fusion of the granule membrane with the plasma membrane of the CAR T cell, causing the surface exposure of proteins that are present on the membrane of the lytic granules, such as CD107a.The protein CD107a can be detected using flow cytometry.19BBe WT CAR expressing primary T cells efficiently degranulated upon 3 hours of co-incubation with CD19 + Nalm6 cells (Figures 6A, B).However, primary T cells expressing the phospho-mimetic variants of 19BBe CARs (Y39E, Y50E, Y39/ 50E) failed to degranulate upon co-incubation with target cells (Figures 6A, B).Stimulation of the endogenous TCR with an anti-CD3 antibody was used as positive control and resulted in equal levels of degranulation in primary CAR T cells independent of the expressed 19BBe variants (Figures 6A, B).To account for the cytotoxicity of these constructs in broader time window and to also consider alternative cytotoxic axes such as Fas/FasL, we performed a luciferase-based killing assay targeting CD19 + Nalm6 cells by measuring specific killing at several time points (Figure 6C).The killing of tumor cells by the primary T cells expressing the phospho-mimetic CARs (Y39E, Y50E, Y39/50E) was significantly reduced and delayed compared to CAR T cells expressing the 19BBe WT CAR.For instance, the 19BBe WT CAR T cells killed 100% of CD19-expressing target cells after 12 hours of co-incubation.In marked contrast, 19BBe Y39E, Y50E and Y39/50 CAR T cells killed only 20% of Nalm6 cells at this time point.Only after 52 hours of co-incubation, all CAR T cells expressing the three phosphomimetic CARs killed 100% of the target cells (Figure 6C).Next, we investigated the ability to secrete the proinflammatory cytokines IFNg and TNF by primary CAR T cells expressing our novel 19BBe CARs.Primary CAR T cells were coincubated with CD19 + Nalm6 cells for 24 hours, and cytokine secretion was assayed by ELISA.Primary human T cells expressing the 19BBe WT CAR efficiently secreted IFNg and TNF upon CAR triggering (Figures 6D, E).In contrast, all three phospho-mimetic variants (Y39E, Y50E, Y39/50E) failed to transmit signals resulting in IFNg or TNF production (Figures 6D, E).However, stimulation with PMA and ionomycin demonstrated equal potency to secrete IFNg and TNF for all cells, independent of the expressed CAR construct indicating that all CAR T cells have the potential to be activated (Figures 6D, E).Altogether, primary T cells expressing 19BBe WT CARs are functional to eliminate tumor cells.However, mimicking irreversible phosphorylation of the 19BBe CAR impairs signal transduction, reducing and delaying cytotoxicity as well as preventing cytokine secretion.

Discussion
The TCR-CD3 complex contains ten ITAMs and although these ITAMs share a conserved YxxL/I-X 6-8 -YxxL/I motif, the exact amino acid sequence of each of them is distinct, resulting in different binding affinities to signaling molecules.In contrast, other receptors of the immune system have significantly fewer ITAMs.For instance, the B cell receptor has only two ITAMs.The multitude of TCR ITAMs likely contributes to signal amplification, as suggested by murine models, which show that reducing the number of ITAMs below seven per TCR-CD3 complex leads to an autoimmune disorder, as a result of impaired TCR function during establishment of central tolerance in the thymus (22).Alternative evidence suggests that the different ITAMs may also have non-redundant functions.Namely, experiments in which all ITAMs of the TCR were exchanged by a given ITAM, support that ITAM diversity is relevant for signal transduction and T cell development even when the total number of ITAMs is conserved (30).
Upon TCR binding to its ligand (pMHC), ITAMs are phosphorylated by LCK (18, 19, 31), allowing the recruitment of SH2 domain-containing proteins.However, how LCK is recruited to the TCR-CD3 complex, to phosphorylate the ITAMs, is still not fully understood.Several models have been proposed to mechanistically explain this crucial step.For instance, the formation of a tri-molecular complex of the TCR with pMHC and the co-receptors CD4 or CD8, bringing co-receptor-associated LCK close to the TCR to initiate phosphorylation.However, most recent evidences support that the formation of this tri-molecular complex is a consequence of ITAM phosphorylation.In this scenario, the driving force of the tri-molecular complex is the co- receptor-bound LCK that interacts via its SH2 domain with the phosphorylated ITAMs (32)(33)(34)(35)(36).In another model, LCK randomly encounters TCRs by (37).Furthermore, ITAM phosphorylation is controlled by conformational changes within the TCR-CD3 complex regulating the accessibility of the ITAMs (9,12,(38)(39)(40).Additionally, stabilization of the active TCR conformation exposes the PRS motif, which overlaps with the Nterminal half of the CD3e ITAM, leading to the recruitment of NCK via its SH3.1 domain (6).We have recently identified the RK motif (RKxQRxxxY), which overlaps with the C-terminal half of the CD3e ITAM (8).Upon ligand-binding to the TCR, the RK motif is exposed and recruits LCK by a non-canonical interaction that involves the SH3 domain of LCK and the un-phosphorylated CD3e ITAM.Therefore, ITAMs might have functions prior to their phosphorylation and the evidence suggests that the CD3e ITAM might function as a molecular switch, tightly regulating the CD3e signalosome upon TCR engagement.Herein, we aimed to investigate how phosphorylation of each of the tyrosines of the CD3e ITAM impacts TCR signal initiation.Our working hypothesis was that protein interactions with the unphosphorylated and mono-phosphorylated CD3e ITAM constitute a crucial step to initiate TCR downstream signals.To this end, we embraced an approach to mimic irreversible phosphorylation of the CD3e ITAM by exchanging its tyrosines to glutamic acid.Our results validate the suitability of this strategy.The tandem SH2 domains of ZAP70 have been reported to interact with the double-phosphorylated ITAM of CD3e (15), while they hardly bind to mono-phosphorylated ITAMs (16).Interestingly, the ITAM of CD3e has the lowest affinity to ZAP70 among the TCR ITAMs (41).As expected, ZAP70 bound better to the doublemutated CD3e ITAM variant (Y39/50E), than to the other CD3e variants or eWT.Un-phosphorylated CD3e recruits NCK via its SH3.1 domain to the PRS motif.This interaction is further stabilized by binding of the NCK SH2 domain to the phosphorylated second tyrosine (Y50) of CD3e (21), while phosphorylation of the first tyrosine of this ITAM (Y39) disrupts NCK binding to CD3e (20).Indeed, we confirmed these results by showing that NCK bound to the un-phosphorylated CD3e WT ITAM and this binding was enhanced by the Y50E mutation.Upon ligand-binding to the TCR, the RK motif is exposed and recruits LCK via its SH3 domain to the un-phosphorylated CD3e ITAM.Molecular modeling predicted that phosphorylation of Y50 will prevent this interaction, and experimental evidence showed that LCK preferentially binds to non-phosphorylated CD3e over double-phosphorylated CD3e (8).Here, we have systematically assessed the role of the CD3e ITAM tyrosines for LCK binding with our phospho-mimetic variants.In our pull-down experiments with CD3e-GST fusion proteins, we demonstrated that LCK bound to the un-phosphorylated CD3e WT and the phospho-mimetic CD3e Y39E, while the variants Y50E and Y39/50E abolished LCK binding.Using our established TCR-LCK in situ PLA, we confirmed only residual binding (<30%) of LCK to the CD3e ITAM when mono-phosphorylated at the C-terminal tyrosine (Y50E).
The impact of these variants in the context of the TCR-CD3 was clear but modest, since in the TCR-CD3 complex, eight additional ITAMs might compensate for these mutations (22).By expressing eWT or the phospho-mimetic CD3e variants in a JK T cell line lacking endogenous CD3e, we demonstrated that all CD3e variants restored TCR assembly and expression at the T cell surface.Interestingly, the phospho-mimetic variants did not induce signaling in the absence of TCR engagement that would result in up-regulation of activation markers like CD69 or in IL-2 secretion, suggesting that the TCR-CD3 complex prevents spurious signals despite mimicking the phosphorylation of the CD3e ITAM and the recruitment of ZAP70.Upon TCR engagement, the up-regulation of CD69, CD25 and 4-1BB was decreased by 10-40%, IL-2 secretion by cells expressing the phospho-mimetic variants was 60% reduced compared to CD3e WT.All phospho-mimetic CD3e variants reduced ZAP70 phosphorylation by 30-50%, preventing optimal downstream signaling.These results highlight that reversible CD3e phosphorylation is key to ensure optimal T cell activation.Previous reports have shown that phenylalanine substitution at Y39 abolished signal transduction, including the phosphorylation at the CD3e C-terminal tyrosine, and ZAP70 association (42) pointing out the importance of the OH group of tyrosines for molecular interactions.Our data suggest different thresholds for the up-regulation of activation markers and the secretion of IL-2, indicating different degrees of compensation for the harmful effects of the phospho-mimetic CD3e variant by the additional TCR ITAMs.In all, our results suggest that a reduction of 50% in signal transduction seems to be still enough to induce transcription of activation markers, but fails to induce cytokine secretion.Indeed, IL-2 expression upon TCR engagement is binary (all-or-none), while the upregulation of activation markers is graded and can be induced by suboptimal signals.Experimental evidences and mathematical modeling have suggested that TCR triggering results in digital NFAT activation, while NFkB activation is graded.Subsequently, NFAT translocates into the nucleus in an all-or-none fashion, transforming the strength of TCR-stimulation into the number of nuclei positive for activated NFAT and IL-2 transcription (43).Taken together, CD3e phospho-mimetic variants are thus unable to efficiently transduce TCR-stimulation into downstream activation signals.
Due to the compensatory effect of the additional TCR-CD3 ITAMs, we decided to take advantage of CARs as minimal TCR constructs.In fact, very early studies, assessing the minimal requirements for T cell activation, set the path for engineering these synthetic receptors, demonstrating that the cytoplasmic tail of z, with its three ITAMs, is sufficient to drive T cell effector functions (44).Indeed, we and others have recently demonstrated that mechanistic principles discovered in the TCR are instrumental to improve the performance of CARs against cancer in pre-clinical models (8,26).On the one hand, reduced tonic signaling and increased anti-tumor efficacy was achieved by introducing the RK motif of CD3e into a secondgeneration FDA-approved 19BBz CAR.Mechanistically, LCK recruitment to the CAR and its phosphorylation was better harmonized with CAR engagement (8).On the other hand, incorporating the CD3e cytoplasmic tail in a second-generation FDA-approved CD28-based CAR enhanced persistence and decreased toxicity.Recruitment of CSK to the mono-phosphorylated CD3e ITAM restrained excessive T cell activation and cytokine secretion (26).Still, our knowledge of the differences and similarities of how CARs and TCRs transmit activation signals is quite limited.When CARs and TCRs were stimulated with the same antigen concentrations, the TCR was more efficiently phosphorylated, and ZAP70 better recruited (45).Similarly, the adaptor protein LAT and PLCg were only weakly phosphorylated by CAR stimulation when compared to TCR stimulation (46).Here, we have generated secondgeneration CARs containing only the cytoplasmic tail of CD3e to avoid compensatory effects by additional ITAMs of the TCR-CD3 complex.
Our CARs contain the signaling domain of 4-1BB, which activates the NFkB pathway.Indeed, we saw some degree of NFkB activation in the absence of ligand upon CAR expression independently of the CD3e variant used.CAR Tonic signaling was reported to play a crucial role in regulating in vivo CAR T cell fitness and its anti-tumor function (47,48).The phospho-mimetic CAR variants did not increased tonic signaling compared to the WT CAR, demonstrated by the lack of up-regulation of the activation marker CD69, CD25 and 4-1BB or the secretion of IL-2 in the basal conditions.Upon co-incubation with CD19-expressing tumor cells, the activation of NFkB and NFAT, the upregulation of CD69, CD25 and 4-1BB, and the secretion of IFNg and TNF were severely impaired by the phospho-mimicking CAR variants compared to the WT CAR.The data provide evidence that for optimal T cell activation by CARs, dynamic CD3e ITAM phosphorylation is essential.Likewise, CAR-induced T cell degranulation and tumor killing were strongly reduced in CAR T cells expressing the phospho-mimicking CAR variants.Substitution of Y39 to phenylalanine also diminished CAR T cell activation in a previous study (27), again supporting that dynamic CD3e ITAM phosphorylation is key for the functioning of these CAR T cells.The mechanism how each phospho-mimetic variant reduced T cell activation differs.Y39E prevents NCK recruitment, Y50E prevents LCK recruitment, and Y39E/Y50E allows ZAP70 binding that cannot be optimally activated due to the absence of LCK.These results highlight the importance of the CD3e signalosome to initiate T cell activation.The fact that the primary T cells expressing the phosphomimetic CAR variants showed decreased secretion of proinflammatory cytokines and slower killing might be beneficial in some therapeutic settings.Therefore, we think that it is worthwhile to test these constructs in pre-clinical models that not only assess for anti-tumor effects, but in humanized models that can also assay CAR T cell-associated toxicities.The current notion is that excessive CAR T cell activation leads to T cell dysfunction or exhaustion, to shorter survival due to activation-induced apoptosis, and to harmful cytokine secretion inducing a life-threating cytokine release syndrome.Nevertheless, these studies are out of the scope of this work.
In all, our data support the notion that the CD3ϵ ITAM might has evolved to facilitate initiation of TCR signaling prior to its phosphorylation unravelling a phosphorylation-independent function of an ITAM.Thus, the CD3e ITAM is a unique and multifunctional switch fine-tuning TCR downstream signaling and T cell activation.institutional requirements.The human samples used in this study were acquired from a by-product of routine care or industry.Written informed for participation was not required from the participants or the participants' legal guardians/next of kin in accordance with the national legislation and institutional requirements.Ethical approval was not required for the studies on animals in accordance with the local legislation and institutional requirements because only commercially available established cell lines were used.

2x10 5
cells were transduced with a virus multiplicity of infection (MOI) of 4 for 72 h.Cells were then washed and transduction efficiency was assessed by flow cytometry 5-7 days after transduction.

Figures
Figures and figure legends provide statistical parameters such as the actual value of n, precision measurements (mean ± SD), and statistical significance.Differences were considered significant when the p values were < 0.05.Data sets were tested for normality using the Shapiro-Wilk test.Statistical significance in each figure was calculated as indicated in the figure legend.Only statistically significant differences are shown.Statistical analysis was performed in GraphPad PRISM 10.

1
FIGURE 1 CD3e phospho-mimetic variants bind distinct signaling molecules.(A) Sequence of the human CD3e-fusion proteins used in the study.(B) Pull-down (PD) assay in Jurkat (JK) T cell lysates using beads coupled to the human cytoplasmic CD3e tail fused to GST and followed by immunoblotting with the indicated antibodies.(C-E) Quantification of 4-5 independent experiments done as in (B).ZAP70 (C), NCK (D) and LCK (E) signals were normalized to GST for each sample.One sample t test was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.Each dot represents one independent experiment.Ns, non-significant; *P < 0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

2
FIGURE 2 Reduced activation of JK T cells expressing phospho-mimetic CD3e variants.(A) Schematic depiction of the human CD3e (hCD3e) variants used in this study.(B, C) JK eKO cells expressing the respective hCD3e variants were left unstimulated or stimulated with 5 mg/ml anti-CD3 antibody for 24 (h) Representative histograms (B) and the percentage of CD69 + , CD25 + or 4-1BB + cells (C) was assessed by flow cytometry.4-5 independent experiments are pooled.(D) JK eKO cells expressing the indicated hCD3e variants were left unstimulated or stimulated with 5 mg/ml anti-CD3 and 1 mg/ml anti-CD28 antibody for 24 (h) The relative IL-2 secretion was assessed by ELISA.5 independent experiments were performed and each normalized to the stimulated eWT sample.(E) Percentage of Ca 2+ fluxing cells assessed by flow cytometry upon stimulation with 0.5 mg/ml anti-CD3 antibody.One representative experiment out of three independently performed experiments is shown.(F) JK eKO reporter cells, in which GFP expression is regulated by the responsive elements of NFkB, or NFAT, were transduced with a mock vector or the respective hCD3e variants.(G) JK NFkB/NFAT cells were left unstimulated or stimulated with 5 mg/ml anti-CD3 antibody for 24 (h) Percentage of GFP + cells was assessed by flow cytometry.5 independent experiments are pooled.Two-way ANOVA with Dunnett's multiple comparisons was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.Each dot represents one independent experiment.Ns, non-significant, *P < 0.5, **P < 0.01, ****P < 0.0001.
FIGURE 3 Phospho-mimetic variants of CD3e define TCR-LCK proximity and diminish ZAP70 phosphorylation in JK cells.(A) JK eKO cells expressing the respective phospho-mimetic hCD3e variant were either left unstimulated or stimulated with 5 µg/ml anti-CD3 antibody for 5 min at 37°C.A proximity ligation assay (PLA) was performed between the TCR (CD3e) and LCK.A red dot indicates a proximity closer than 80 nm.Images from one representative experiment are shown.Cell nuclei are stained with DAPI.(B) Quantification of four independent PLA experiments.Dots per cell are normalized to the stimulated WT for each experiment.(C) Total cell lysates were subjected to immunoblotting with anti-pZAP70 (Y319), anti-ZAP70, and anti-GAPDH antibodies.JK eKO cells were used as negative control.(D) The quantification of 5 independent experiments normalized to the stimulated WT is shown.Two-way ANOVA with Dunnett's multiple comparisons test was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.Each dot represents one independent experiment.Ns, non-significant, *P < 0.5, ***P < 0.001, ****P < 0.0001.

4
FIGURE 4 Reduced activation and signaling capacity of JK T cells expressing phospho-mimetic CAR variants.(A) Schematic depiction of the CD19-targeting CAR constructs (19BBe) used in this study.(B-D) JK cells expressing the indicate 19BBe constructs were left unstimulated or stimulated with CD19 + Nalm6 tumor cells (effector to target ration (E:T) 1:1) for 24 (h) Representative histograms (B) and the percentage of CD69 + , CD25 + or 4-1BB + cells (C) are shown from five independent experiments.The relative IL-2 secretion (D) is shown for four independent experiments each normalized to the stimulated WT sample.(E, F) JK NFkB/NFAT eKO reporter cells were transduced with a mock construct or the respective 19BBe variants.Cells were left unstimulated or stimulated with CD19 + Nalm6 tumor cells for 24 (h) Representative histograms (E) and the percentage of GFP + cells (F) are shown.Four independent experiments were pooled.Two-way ANOVA with Dunnett's multiple comparisons test was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.Each dot represents one independent experiment.Ns, non-significant, ***P < 0.001, ****P < 0.0001.

5
FIGURE 5 Phospho-mimetic CAR variants prevent T cell activation in primary human CAR T cells.(A) Schematics shows how PBMCs were activated with 1 mg/ ml anti-CD3 and anti-CD28, and then lentivirally transduced to generate the respective 19BBe CAR T cells.(B-D) Primary CAR T cells were either left unstimulated, stimulated with CD19 + Nalm6 tumor cells (E:T 1:1), or with 10 ng/ml PMA and 250 ng/ml ionomycin for 24 (h) The percentage of CD69 + (B), CD25 + (C) and 4-1BB (CD137) + (D) cells were assessed by flow cytometry.Three different healthy donors were used.Two-way ANOVA with Dunnett's multiple comparisons was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.Each dot represents one healthy donor.****P < 0.0001.

6
FIGURE 6 Reduced cytotoxicity and cytokine secretion of primary T cells expressing phospho-mimetic CAR variants.(A, B) Primary CAR T cells expressing the respective 19BBe CAR constructs were either left unstimulated, stimulated with CD19 + Nalm6 tumor cells (E:T 1:1) or stimulated with 5 mg/ml anti-CD3 antibody for 3 (h) Representative histograms for CD107a (A) and the percentage of CD107a + cells (B) are shown.Two-way ANOVA was performed after Shapiro-Wilk test for normality.Each dot represents one healthy donor.(C) Time course showing the specific tumor cell lysis of 19BBe CAR T cell variants upon co-incubation with CD19 + Nalm6 tumor cells.(D, E) Primary CAR T cells were either left unstimulated, stimulated with CD19 + Nalm6 tumor cells (E:T 1:1) or stimulated with 10 ng/ml PMA and 250 ng/ml ionomycin for 24 h and IFNg (D) as well as TNF (E) secretion was assessed by ELISA.Each experiment was normalized to the respective Nalm6-stimulated 19BBe WT sample.Two-way ANOVA with Dunnett's multiple comparisons was performed after Shapiro-Wilk test for normality.3 different healthy donors were used.Each dot represents one healthy donor.Mean values ± SD are indicated.****P < 0.0001.

6
CAR surface levels in JK NFkB eKO reporter cells lentivirally transduced with the indicated 19BBe CAR (B) Percentage of BFP + cells.(C) MFI of BFP in BFP + cells.(D) Percentage of CAR + cells.(E) MFI of CAR in CAR + cells.(F) Representative dot plots showing BFP levels and CAR surface levels in JK NFAT eKO reporter cells lentivirally transduced with the indicated 19BBe CAR constructs.(G) Percentage of BFP + cells.(H) MFI of BFP in BFP + cells.(I) Percentage of CAR + cells.(J) MFI of CAR in CAR + cells.One-way ANOVA with Dunnett's multiple comparisons test was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.Each dot represents one independent experiment.*P < 0.5, **P < 0.01, ****P < 0.0001.SUPPLEMENTARY FIGURE Generation of primary human T cells expressing phospho-mimetic 19BBe CAR variants.(A) Representative dot plots showing GFP levels and CAR surface levels in primary human T cells lentivirally transduced with the indicated 19BBe CAR constructs.Untransduced (UTD) as well as Mock transduced primary T cells were used as negative controls.(B) Percentage of GFP + cells.(C) MFI of GFP in GFP + cells.(D) Percentage of CAR + cells.(E) MFI of CAR in CAR + cells.Each dot represents one healthy donor.One-way ANOVA test was performed after Shapiro-Wilk test for normality.Mean values ± SD are indicated.*P < 0.5, **P < 0.01, ****P < 0.0001.