To test our hypothesis, we generated ITAM restricted murine anti-CD19 CD28 CARs that express three copies of a single CD3ζ ITAM instead of the three different CD3ζ ITAMs as in the conventional CAR design. The CAR backbone was previously published and obtained from Addgene (Plasmid #107226) (17). It consists of the 1D3 clone anti-CD19 binding scFv domain, a CD28 transmembrane and signaling domain, followed by the CD3ζ ITAMs ( Figure 1A ). The newly designed CARs are designated as ζAAA, ζBBB and ζCCC depending on which CD3ζ ITAM sequence is used. Since we want to compare the signaling properties of the ITAM restricted CARs, we wanted to first verify the CAR surface expression levels were similar. We tested each construct for cell surface expression by retrovirally transducing primary mouse CD8+ T-cells and staining with an antibody specific for the linker sequence in the CAR construct. Using the anti-G4S antibody, we observed that the CARs are expressed on the cell surface at similar levels, making signaling studies with these CARs possible ( Figure 1B ).

To compare the signaling properties of the individual CD3ζ ITAMs, we stimulated our CAR T-cells in vitro using CD19 coated dyna-beads for different time-points and quantified the signaling differences using immunoblotting. We measured phosphorylation of proximal protein targets that have been previously shown to be activated downstream of TCR signaling. We first observed that the signaling kinetics are similar to that of the TCR with proximal signaling peaking at the 2.5min timepoint ( Figure 2A ; Supplementary Figures 1A-C ). However, at the peak proximal signaling timepoint (2.5min), we observed that the proximal signaling molecule LAT was less phosphorylated in ζBBB, while for Zap70, ζBBB and ζCCC CARs were also slightly decreased in comparison to the ζABC CAR, suggesting weaker activation ( Figures 2B-D ). To complement this observation, we tested the Ca2+ signaling response in each of the CAR-T cell constructs, which is initiated downstream of PLCγ activation. Ca2+ sensitive Indo1-AM labelled CAR-T cells were stimulated with CD19+ E-µ ALL cells and Ca2+ response kinetics was measured using flow cytometry (22). We observed that ζAAA CARs had a significantly greater median Ca2+ response compared to ζABC CARs, while ζCCC trended towards a lower response compared to ζABC CARs ( Figure 2E ). To assess the impact of ITAM restricted CARs on downstream transcriptional activity, we measured the expression of Ca2+ responsive transcription factor Nur77 (23, 24). In concordance with the Ca2+ response, we observed ζCCC CARs to have a significantly lower expression of Nur77 while ζAAA was slightly higher compared to the wild type ITAM CAR ( Figure 2F ). Together, these data indicate each ITAM sequence allows for differential CD19-CAR intracellular signaling, therefore we next wanted to test whether the alteration in signal transduction translated into altered functional outcomes.

We next performed in vitro functional assays to assess the effect of the observed signaling differences on CAR-T function. We first performed a dye-release cytotoxicity assay where we co-cultured our CAR-T cells with dye-labelled CD19+ E-µ tumor cells at different E:T ratios for 16 hours. We observed all our CAR T-cells were similarly effective in their specific cytotoxicity ( Figure 3A ). Interestingly, when we quantified the viable CD8+ CAR T-cell counts, we observed ζAAA CARs had reduced cell numbers compared to the ζABC CARs which suggests increased exhaustion or AICD ( Figure 3B ). We next performed a CD107a degranulation assay to assess lytic effector function. Each ITAM restrict CAR T cell was co-cultured with the CD19+ E-µ cells in a 1:1 E:T ratio for 6 hours in the presence of PE-conjugated CD107a antibody. Here we observed ζAAA and ζBBB CAR T-cells were similar to ζABC in terms of degranulation ( Figure 3C ). However, ζCCC CAR T cells demonstrated less staining of surface CD107a, suggesting they are less likely to degranulate upon activation ( Figure 3C ). Next, we performed an in vitro cytokine expression assay to measure effector cytokine expression of IFNγ, TNFα and IL2. CAR-T cells were co-cultured with CD19+ E-µ cells for 6 hours in the presence of Brefeldin A and monensin and expression of IFNγ, TNFα and IL2 was measured by intracellular staining. Similar to CD107a expression, we observed ζCCC CAR T-cells expressed less IFNγ and TNFα ( Figures 3D, E ). However, ζBBB showed a selective decrease in IFNγ expression while maintaining similar levels of TNFα and IL2 expression compared to ζABC ( Figures 3D, E ; Supplementary Figure 2 ). Overall, these data suggest that ζCCC CAR T-cells have a weaker activation profile upon encountering tumor antigen that may lead to less effector function but better survival over time.

Since previous studies with the conventional T cell have suggested that weaker activation favors less exhaustion, we tested our ITAM restricted CAR T-cellsfor signs of exhaustion in an in vitro exhaustion assay (25–27). We subjected the CAR T-cells to chronic stimulation conditions by culturing them in the presence of CD19+ dyna-beads for 7 days in the presence of IL-2 and analyzed them for expression of exhaustion related inhibitory receptors ( Figure 4A ). After repeated stimulation, we observed ζCCC CAR T-cells had slightly lower co-expression of inhibitory receptors PD1 and Tim3, whereas ζBBB had significantly higher expression compared to ζABC ( Figures 4B, C , Supplementary Figures 3A–C ). Interestingly, ζCCC also exhibited higher expression of TCF1 and IL7R, both of which are associated with less differentiated more memory-like cells ( Figure 4D ). Together, these data indicate ITAM sequence can lead to differential signal transduction and functional outcomes.

MSGV-1D3-28ζ All ITAMs intact was a gift from James Kochenderfer & Steven Rosenberg (Addgene plasmid # 107226; http://n2t.net/addgene:107226; RRID: Addgene_107226).

Platinum-E (Plat-E) cells obtained from ATCC were used as transient retroviral producers. In brief, 1 million Plat-E cells were plated in 10% complete DMEM. The following day, Plat-E cells were transfected with 6µg vector plasmid using TransIT-LT1 (Mirus) transfection reagent in the presence of 25uM chloroquine. The media was replaced 6 hours later and collected at 48- and 72-hour timepoints for transduction.

On Day 0, T-cells were isolated from spleen and lymph nodes (inguinal, axillary and cervical) of 8–12-week-old donor B6 mice. Single-cell suspensions from the organs were subjected to RBC lysis followed by staining with anti-CD8 and/or anti-CD4 biotinylated antibodies. Miltenyi Streptavidin magnetic beads were used to magnetically enrich CD8+ and/or CD4+ T-cells. T-cells were activated in 24-well plates coated with 10ug/ml anti-CD3 and 2ug/ml anti-CD28 at a density of ~0.5 million cells per well. T-cells were stimulated for 24 hours before transduction.

On Day 1, activated T-cells were transduced using retroviral supernatant obtained from the Plat-E cells. Briefly, the cellular debris was removed from the Plat-E supernatant by centrifuging at 300g for 5 minutes. Polybrene (6ug/ml) and hIL-2 (50U/ml) were added to the supernatants and cells were spin-transduced at 1300g for 1 hour at 37°C. After transduction, T-cells were allowed to rest in the viral supernatant for 1 hour in the 37°C CO2 incubator before replacing media with 10% complete RPMI containing 50U/ml hIL-2. The process was repeated 24 hours later. The transduction efficiency was in the range of 30-50%.

After 48-72 hours of activation, on Day 3 the T-cells were removed from the anti-CD3 and anti-CD28 coated plates and transferred to T-cell expansion medium containing 10% complete RPMI supplemented with 10ng/ml hIL-7 and 10ng/ml hIL-15. T-cells were maintained at density of ~0.5 million cells/ml. On Day 5, CAR expressing T-cells were magnetically enriched using anti-Thy1.1 biotinylated antibodies and streptavidin magnetic beads. Post-enrichment CAR T cell populations were ~90% pure. For in vitro experiments, Day 7 - Day 10 cells were used.

CD19+ E-µ ALL tumor cell line was obtained from Dr. Marco L. Davila’s lab. Since E-µ cells cannot expand on their own, they are grown as a co-culture with NIH/3T3 cells. 3T3 cells were cultured in 5% DMEM with ciprofloxacin. One day prior to plating the E-µ cells, 3T3 cells were X-ray irradiated at 30Gy and plated at 1 million cells per 10cm plate. Next day, E-µ cells are added and expanded in a 50/50 mix of 10% complete RPMI and 10% complete IMDM.

M-450 tosylactivated dyna-beads (Invitrogen) were washed and resuspended in 0.1M sodium phosphate buffer (pH 7.8) at 4x10⁸ beads/ml and incubated overnight with 150ug of recombinant murine CD19 (R&D Systems) with gentle rotation. The beads were then washed 3 times in wash buffer (1X PBS with 0.1% BSA and 2mM EDTA pH 7.4) at 4°C for 5 min each with constant mixing. After washing, the beads were stored at 4°C in storage buffer (1X PBS with 0.02% sodium azide, 0.1% BSA and 2mM EDTA). Before use, beads were washed 3 times in complete RPMI at 4°C with constant mixing.

CAR T-cells were counted and rested in 1X PBS with calcium and magnesium for 30min at room temperature. Two million CD8+ CAR T-cells were stimulated with 6 million CD19+ dynabeads by spinning down for 1min and incubating at 37°C for the mentioned timepoints. Signaling was halted by transferring cells to ice and immediately lysing with 1% NP-40 in deionized water containing Halt Protease/Phosphatase inhibitor cocktail. Lysis was performed for 30min at 4°C with constant mixing. After spinning down debris, equal volumes of protein lysate were mixed with Laemmli’s buffer (Bio-Rad) with 2-mercaptoethanol and denatured at 95°C for 5minutes. Samples were then run in pre-cast gels (Invitrogen, NuPage Bis-Tris midi protein gels, 4-12%, 1.0mm) in MOPS SDS running buffer (Invitrogen). Immunoblotting was performed using Bio-rad’s Trans-blot Turbo semi-dry transfer machine onto a PVDF membrane. The membrane was blocked for 1 hour at room temperature using 3% BSA in tris buffered saline with 0.1% Tween20 (0.1% TBS-T). Primary antibody staining was performed overnight at 4°C in 3% BSA 0.1%TBS-T. Secondary antibody staining was performed at room temperature for 1 hour in 3% BSA 0.1%TBS-T. In between and after staining, the membrane was washed with 0.1% TBS-T 3X for 5min each. Bands were detected using chemiluminescence. Densitometric analysis was performed using Image Lab 6.1. Protein phosphorylation was quantified as a ratio of phospho band intensity and total protein band intensity. Ration was then expressed as fold change relative to unstimulated cells.

CD8+ CAR T-cells were labelled with the Ca²⁺ sensitive ratiometric Indo-1AM dye for detection of Ca²⁺ signaling. Briefly, the CAR-T cells were washed in 1X PBS and incubated with 5uM Indo-1 AM dye in PBS for 30min in 37°C incubator. After incubation, the cells were washed twice and resuspended in Ca2+ signaling buffer (Hank’s Balanced Salt Solution Cat# 55037C with calcium, magnesium and no phenol red, plus 10% FCS and 25mM HEPES). CAR-T cells were then stained for Thy1.1 for 10 minutes at room temperature. After washing and prior to flow cytometric analysis, the CAR T-cells were stored on ice in 5ml flow tubes at 1 million cells/500ul. The E-µ cells were labelled with 5uM Cell Proliferation Dye eFluor™ 670 (eBioscience™) in 1X PBS like the CAR-T cells and stored on ice at 2 million cells/500ul. Before running each sample on the flow cytometer, the cells were warmed in a 37°C water-bath for 15minutes.

Before adding the E-µ cells, the CAR-T cells were run on flow cytometer for 1min to collect baseline Ca²⁺ bound/Ca²⁺ free ratio. Flow rate is maintained constant at ~1000 events/sec across samples. At 1 min, the tubes were taken off with recording still ongoing, and 2 million E-u cells were added, mixed at allowed to form T:E conjugates by centrifugation at 300g for 25sec. At the 2 min timepoint, the tube was put back into run mode and data was recorded for 10 minutes to capture Ca²⁺ response. After 10 minutes, the tube was taken off the cytometer and ionomycin (~1ug/ml) was added to check maximum Ca²⁺ response.

For measurement of Ca²⁺ response, we gated on Thy1.1+ ef647+ conjugates and created a kinetics curve for median of Ca²⁺ bound/Ca²⁺ free ratio derived parameter. Gates were set for the peak response (120sec to 500sec) and area under curve was calculated and plotted relative to the ζABC CAR.

E-µ cells were labelled with 5uM eFluor™ 670 dye in 1X PBS. Thy1.1+ CAR T-cells were co-cultured with 50,000 dye labelled E-µ cells at the mentioned E:T ratios in a 96-well flat-bottomed plate for ~15 hours at 37°C. Dye+ E-µ cell numbers were quantified using flow cytometry and %specific lysis was calculated as

A total of 100,000 Thy1.1+ CAR T-cells were co-cultured with 200,000 E-µ cells in 10% complete RPMI at 37°C for 6 hours in 96 well round-bottomed plates. Brefeldin A (5ug/ml) and Monensin (2uM) were added at the beginning of incubation to prevent secretion of cytokines. Cytokine expression was analyzed by intra-cellular staining and flow cytometry.

A total of 100,000 Thy1.1+ CAR T-cells were co-cultured with 200,000 E-µ cells in 10% RPMI at 37°C in 96-well flat-bottom plates. Phycoerythrin (PE) labelled anti-CD107a (LAMP-1) antibody was added to each well at 1ug/ml final concentration at the beginning of the incubation to label CD107a exposed on the cell membrane as a result of T cell degranulation.

A total of 100,000 Thy1.1+ CD8+ CAR T-cells (Day 7 post-activation) were incubated with 400,000 CD19 coated dyna-beads on Day 0 in flat-bottomed 96-well plates in 10% complete RPMI containing 50U/ml IL-2. CAR-T cells were re-stimulated with 400,000 CD19 coated beads at Day 3 and Day 6. Media was replaced every 1.5 days with fresh RPMI containing 50U/ml IL-2. On Day 7, beads were removed, and cells were analyzed by flow cytometry.

Single-cell suspensions were stained in buffer containing PBS, 3% v/v FCS and 0.05% w/v sodium azide. Surface staining was performed on ice for 20 minutes after Fc receptor blocking with FcBlock (Biolegend) for 10 minutes. Live/Dead staining was performed using Zombie Red (Biolegend) in 1X PBS for 30 min on ice. For intra-cellular staining, cells were fixed with FoxP3 Fix/Perm reagent (eBioscience) and then stained under permeabilizing conditions overnight at 4°C in Permeabilization Buffer (eBioscience). Flow cytometry data were acquired on a BD LSRFortessa (Becton Dickinson) flow cytometer and analyzed using the FlowJo 10 software (FlowJo LLC).