The ROR1⁺ JeKo-1 (Mantle cell lymphoma) and ROR1^- K562 (Chronic myelogenous leukemia) Cell lines were initially obtained from ATCC; cell lines were obtained more than six months before experiments, and authentication was performed by cell banks utilizing short tandem repeat profiling. All cell lines were tested for mycoplasma contamination (MycoAlert Mycoplasma Detection Kit, LT07-318, Lonza). The number of passages was limited to 10. These cell lines were transduced with a luciferase-ZsGreen lentivirus (Addgene) and sorted with FACS Aria II (B.D. Biosciences) instrument to 100% purity. Cell lines K562 were used as controls, as indicated in the relevant figures. The cell lines were maintained in culture with RPMI1640 (Gibco) supplemented with 10% Fetal bovine serum (Gibco) and 50 U/mL penicillin/streptomycin (Gibco, Life Technologies, 15070–063). Primary cells were thawed at least 12 hours before the experiment for all functional studies and rested at 37°C. The use of recombinant DNA was approved by the Institutional Biosafety Committee (IBC HIP00000710.04).

Second-generation and third-generation ROR1 CAR constructs were synthesized de novo (GenScript). All the constructs consist of a scFv (single-chain variable fragment) against ROR1 (Clone R12) linked with an IgG4 hinge (Uniprot P01861, section 99-110) cloned into a second-generation 4-1BB and third-generation CD28 + 4-1BB, ICOS + 4-1BB, and ICOS + OX40 costimulated CAR in a third-generation lentivirus ( Table 1 ). Third-generation Lentiviral vectors were generated using transient transfection of the plasmid into 293T cells obtained from ATCC in the presence of Lipofectamine 2000 (Invitrogen), VSV-G, RSV-Rev, pMDLg-pRRE (Adgene).

Briefly, peripheral blood mononuclear cells (PBMC) are isolated from de-identified healthy donor blood apheresis samples (Vitalant). T-cell isolation from previously isolated PBMCs was performed using an 8-minutes cell isolation kit immunomagnetic negative selection (STEMCELL magnetic beads against CD19, CD16, CD15, CD14, CD34, CD56, CD123, and CD235a), obtaining a purity after isolation greater than 95% of the T-cell product. Primary cells were cultured in T-cell medium (TCM) supplemented with OpTmizer CTS, 1% penicillin-streptomycin-glutamine (Gibco), and IL-2 at a concentration of 100 U/mL (PeproTech).

Previously isolated T cells were stimulated using a Dynabeads CD3/CD28 generated in-house at a 1:3 cell: bead ratio. After 24 hours of stimulation, T cells were transduced with lentiviral particles at an MOI (multiplicity of infection) of 3.0. Beads were removed from the T cell expansion using DynaMag-50 (Invitrogen) on day 6. The expression of CAR was analyzed by flow cytometry on day 7. CARs were stained with a biotinylated ROR1 protein followed by a PR-conjugated streptavidin. On day 12, CART- Cells were harvested and cryopreserved in freezing medium made from 90% FBS (Gibco) and 10% DMSO (Millipore Sigma) for planned experiments. The CAR T cells were thawed and left resting in TCM for at least 12 hours before individual experiments.

All experiments were performed on a FACS Symphony (B.D. Biosciences), and data obtained were analyzed with FlowJo software (version 10.7.1). CAR expression was evaluated using a ROR1 biotinylated protein (ACRO Biosystem) and a streptavidin (PE-conjugated; B.D. biosciences). T cell profile and characterization were evaluated using mAb to CD45, CD3, CD4, CD8, CD62L, CCR7, PD-1, TIM-3, LAG-3, TIGIT, and the corresponding isotype controls (B.D. Biosciences). The presented results presenting CAR T cell phenotype data were performed using the following gating strategy: we excluded the doubles, by side scatter versus forward scatter, viability using LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Invitrogen), followed by the populations CD45⁺, CD3⁺, CAR⁺ Cells ( Supplementary Figure 1 ).

Anti-human and anti-mouse antibodies were purchased from BioLegend, eBioscience, or B.D. Biosciences. BD FACS lyse buffer (B.D. Biosciences) was used to lyse mouse red blood cells (JeKo-1 xenograft) peripheral blood samples (day −1 and 8 of ROR1 CART-cell infusion) before staining and flow cytometric analysis. Cells from in vitro culture (in vitro antigen-specific degranulation/cytokine production assays and antigen-specific proliferation assay). Before staining, cells were washed twice in a flow buffer [PBS supplemented with 1% FBS and 1% sodium azide (Ricca Chemical)] and stained at room temperature unless specified for the antibody. For cell number quantitation, Countbright beads (Invitrogen) were used according to the manufacturer’s instructions (Invitrogen). In all analyses, the population of interest was gated based on forward versus side scatter characteristics, followed by singlet gating, and live cells were gated following staining with LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Invitrogen).

Surface expression of the CAR was detected by staining with a ROR1 biotinylated protein. In brief, an aliquot of the CAR-T cells (e.g., 50,000 T cells) was first washed and then resuspended in 50 μL of a flow buffer. Cells were stained with 1 μL of ROR1 biotinylated protein and 0.3 μL of LIVE/DEAD Fixable Aqua Dead Cell Stain Kit to exclude dead cells and incubated in the dark for 15 minutes at room temperature. After incubation, cells are washed by adding 150 μL of a flow buffer and centrifuged at 650 × g for 3 minutes at 4°C. Then 8 μL of streptavidin-PE and 50 μL of flow buffer were added, and cells were incubated for 30 minutes at 4°C. After surface staining, cells are fixed and permeabilized by adding 100 μL of a fixation medium and incubated for 15 minutes at room temperature. Cells are then washed with 100 μL of a flow buffer. Fixed/permeabilized cells were resuspended in 50 μL of a permeabilizing buffer. Cells were washed and resuspended in 200 μL of a flow buffer and acquired on a flow cytometer. Flow cytometry was performed on a Symphony (B.D. Biosciences) five-laser cytometer ( Supplementary Table 1 ). Analyses were performed using FlowJo X10.0.7r2 software.

IFN-γ and CD107a production were evaluated using flow cytometry, according to the manufacturer’s instructions. Briefly, 0.5x10⁶ CAR T cells were seeded for (IFN-γ 12 hrs, CD107a 4 hrs) with 5x10⁵ target cells (Jeko-1 or K562). To create stringent testing conditions similar to physiological environment we tested an E:T ratio of 1:1, in triplicate wells on 96-well round bottom plates as reported in previous studies (45–47). Negative and positive controls were represented by CAR T cells that remained unstimulated (medium only) or treated with 40 ng/mL of PMA and 4 mg/mL of ionomycin (Sigma-Aldrich). After the stimulation, the corresponding antigens were measured using FACS in the Flow Cytometry core at Mayo Clinic AZ.

We developed a Luciferase-based rechallenge cytotoxicity assay. The target Luciferase⁺ Jeko-1 and K562 cells were incubated with effector CAR T-cells transduced with different constructs as indicated. Each day the E:T ratios were adjusted to generate seven different E:T ratios per construct. Effector cells remain unchanged from day one until day five to assess their ability to induce cytotoxicity during this rechallenge experiment. On the other hand, fresh new target cells were added as needed to adjust the E:T ratios. Each day and for each construct, we measured the cytotoxicity generated by the seven different E:T ratios, and with these data, we obtained the logarithmic trendline and calculated the E:T ratio that induced cytotoxicity in 50% of the target cells (EC50). The EC50 E:T ratio was used as a proxy for each cellular product’s potency and ability to induce repetitive cytotoxicity and expand over time (persistence). CAR T-cells transduced with constructs that induce low EC50 E:T ratios at five days are expected to be fitter and have a better persistence profile, while those with high EC50 E:T ratios are more prone to exhaustion. The killing was calculated by BLI on a Xenogen IVIS-200 Spectrum camera (catalog no. 124262, PerkinElmer) as a measure of residual live cells. Ten minutes before imaging, samples were treated with 1 mL D-luciferin (30 mg/mL, Gold Biotechnology) per 100 mL sample volume.

Tonic signaling evaluation using CD3z Pho(Tyr142) on ROR1 CAR constructs was performed. Different CAR T cells were expanded for 12 days with T cell media +100 U/ml of IL-2. The samples were electrophoresed using 4-12% Bis-Tris Protein Gel (Cat.NP0322BOX). Resolved proteins were transferred onto a nitrocellulose membrane (Cat.IB23001) by iBlot^® 2 Dry Blotting System (Cat. IB21001). The blot was probed with Phospho-CD3z (Tyr1 42) Polyclonal Antibody (Cat.PA5-37512, 1:1000 dilution) and Goat anti-Rabbit IgG (H+ L) Supercional™’ Recombinant Secondary Antibody, HRP (Cat. A27036, 1:20,000) using the Bright™M FL 1500 (Cat. A44115). Chemiluminescent detection was performed using SuperSignal™M West Alto Ultimate Sensitivity Substrate (Cat. A38556). The fact that CD3z was not present in the untransduced, and fresh T-cells confirms that the antibody is specific to native CD3z and CAR CD3z on controls.

The Mayo Clinic Hospital Institutional Animal Care and Use Committee approved all animal experiments. NOD scid gamma (NSG) mice were purchased from The Jackson Laboratory and housed in the Mayo Clinic, Arizona vivarium, under specific pathogen-free conditions in microisolator cages and were provided ad libitum access to autoclaved food and acidified water.

Male and female, 8 to 12-week-old, non-obese diabetic/severe combined immunodeficient bearing a targeted mutation in the IL2 receptor gamma chain gene (NSG) mice were purchased from the Jackson Laboratory (catalog no. 005557) and maintained within the Mayo Clinic Department of Comparative Medicine under an Institutional Animal Care and Use Committee–approved protocol (IACUC A00005886-21). The ROR1+luciferase+ MCL JeKo-1 cell line was used to establish a systemic ROR1+ tumor model. 5 × 10⁵ cells were resuspended in PBS and injected via the tail vein, 1 week before T-cell infusion. The tumor burden was assessed weekly by BLI using a Xenogen IVIS-200 Spectrum camera to confirm the engraftment of JeKo-1 cells and visual inspection. The mice were randomized according to the tumor burden assessed by IVIS to receive 3 × 10⁶ of (i) UTD, (ii) ROR1.BBζ, (iii) ROR1.IOζ or (iv) CD19.BBζ CAR T-cells (Cells infused were normalized by CAR% of each group, except for the UTD group). One week after T-cell infusion, mice were imaged with BLI, as described below. Mice were euthanized after the completion of the experiment.

A Student’s t-test was used to compare two value sets, while we used one-way ANOVA when three groups were involved. Histograms represent mean values ± standard deviations. P < 0.05, P < 0.01, or P < 0.001 were indicated by *, ** or ***, respectively. All statistics were performed using GraphPad Prism version 8.05 for Windows (GraphPad Software, www.graphpad.com). Statistical tests are described in detail in the representative figure legends.

We evaluated and optimized different scFv sequences derived from anti-ROR1 antibodies using a 2G-CAR construct design with 41BB-ζ ICD ( Figure 1A ) (48). First, we studied the effects of the cloning position of the variable heavy (VH) and variable light (VL) domains in the lentiviral plasmid vector and the effects of the interspace linker (GGGGS)n length in between the VH and VL domains - (short linker n=5 vs. long linker n=15). Specifically, we evaluated differences in the CARs’ ability to induce CAR expression after T-cell transduction, T-cell expansion, cytotoxicity, T-cell activation, or degranulation.

We observed that longer linkers induced lower levels of CAR expression, but these differences were not statistically significant ( Figure 1B ). Also, CAR constructs with long linkers, particularly the VL-long linker-VH (LLH) sequence, induced higher levels of T-cell expansion, reaching a 25-fold increase at day ten compared to the other constructs ( Figure 1C ).

All the ROR1-CAR constructs we tested induced similar cytotoxicity and IFN-γ production levels specifically driven by ROR1 expression on the target cells ( Figure 1D, E ). However, the level of degranulation measured by CD107a expression was higher on the T-cells transduced with long linker constructs (HLL, LLH) ( Figure 1F ).

The above data suggest that the cloning positioning of VH or VL was not relevant to the CAR T-cell parameters that we measured, that induced cytotoxicity was similar in all the 2G constructs regardless of VH-VL domain orientation or linker length, and that long interspace linkers between VH-VL domains are critical for the optimal degranulation and expansion of the CAR T-cells. Because of these findings, we selected the LLH scFv design to proceed with the 3G-CAR cloning, including the ICOS and OX40 ICD. CARs containing the 4-1BB and CD28 were linked to the CD28 TM domain, while CARs with a membrane-proximal ICOS ICD had the ICOS TM domain (unless otherwise indicated), and all the CARs constructs used an IgG4 hinge ( Figure 2A ).

We evaluated three 3G-CAR construct designs [A. ROR1-ICOS/4-1BB/CD3ζ (ROR1-IBζ); B. ROR1-ICOS/OX40/CD3ζ (ROR1-IOζ); C. ROR1-CD28/4-1BB/CD3ζ (ROR1-28Bζ)]; and compared with a 2G-CAR control [ROR1-4-1BB/CD3ζ (ROR1-Bζ)]. There were no differences in the level of CAR protein expression after lentivirus transduction of T-cells ( Supplementary Figure 2 ). 3G-CAR T-cells show a statistically significant slower expansion growth than 2G-CAR T-cells ( Figure 2B ). CAR T-cell persistence was evaluated in vitro using a model of repetitive target cell challenge developed in our laboratory (49) ( Figure 2D ). On day 1, all the 3G-CAR constructs showed similar E: T ratios suggesting equal cytotoxic potency, and there was a trend for increased cytotoxicity with lower E: T ratios for the cells transduced with the 2G-CAR construct. This trend became statistically significant on Day 2, and by day 5, the cytotoxicity of 2G-CAR T-cells was lost with E:T ratios that were very high and comparable with the untransduced T-cells (UTD). The T-cells transduced with 3G-constructs behave differently; they showed relatively low cytotoxicity during the first 48 hr of culture, and gradually the E:T ratios decreased, and their cytotoxicity was significantly increased by Day 5 compared with the T cells transduced with the 2G-CAR construct or UTD cells ( Figure 2D ). Moreover, after 5 days of in vitro rechallenge with ROR1⁺ tumor cell lines, all T cells transduced with the 3G-constructs secreted significantly higher IFN-γ levels than the 2G-CAR construct and UTD control ( Figure 2E ).

In addition, we studied the phenotypic changes of T cells transduced with different constructs, particularly the T-cell central memory (T_CM) and effector memory (T_EM) phenotype ratio (T_CM/T_EM) that has been associated with CAR T-cell persistence (50). After 12 days of in vitro expansion, all the constructs that we tested exhibited at least a threefold increase in CD4 and CD8 T-cell central memory (T_CM) compared with effector memory (T_EM) phenotype ( Supplementary Table 3 ). However, this elevated “baseline” T_CM predominance changed after the cells were evaluated using our five days in vitro rechallenge assay with ROR1⁺ tumor cell lines. For example, the 3G ROR1-IOζ CAR T-cells continue to express the highest CD4⁺ T_CM levels, with CD8⁺ T_CM comparable to the Bζ and IBζ cells and higher than the 28Bζ ones. Regarding the T_EM phenotype, the ROR1-IOζ CAR T-cells showed the lowest percentage of T_EM cells (CD4⁺ and CD8⁺). The calculated T_CM/T_EM ratios showed a > 4-fold increase in CD4⁺ T-cells transduced with the ROR1-IOζ CAR compared to the other constructs. There were no substantial differences in the T_CM/T_EM ratios among CD8⁺ T cells transduced with any of the constructs ( Figure 2F ; Supplementary Table 3 ).

We evaluated the tonic signaling of the T cells transduced with different CAR constructs before engagement with the specific target antigen ( Figure 2C ). CAR T cells were expanded for 12 days using T cell media +100 U/ml of IL-2. The baseline CD3ζ CAR chimeric and the native CD3ζ proteins were evaluated for tyrosine phosphorylation (Ty142) by Western blot using specific monoclonal antibodies. This basal CD3ζ signaling helped us to evaluate the tonic phosphorylation of the ICD and potential correlations with cytotoxicity, expansion, and phenotype. CAR T-cells transduced with the 3G ROR1-IOζ showed the lowest phospho-CD3ζ-Ty142 adjusted expression ratios suggesting that this construct induced the lowest ICD tonic signaling activation. On the other hand, T-cells transduced with the 2G ROR1-Bζ showed the highest CD3ζ-Ty142 phosphorylation. Higher phospho-CD3ζ-Ty142 tonic signaling correlated with CAR constructs that induced rapid expansion, cytotoxicity, and exhaustion ( Figure 2 ; Supplementary Figure 4 ).

We used the Jeko-1 (ROR1⁺) NHL model in NOD-SCID mice for in vivo testing (51). The 3G-ROR1-IOζ construct was selected for these experiments based on the previous results that suggested that this construct could induce better T-cell fitness with a favorable persistent phenotype. We compared the activity in vivo of T cells transduced with the 3G-ROR1-IOζ construct vs. T cells expressing the 2G-ROR1-Bζ and CD19-Bζ constructs ( Figure 3A ). All of the Jeko-1 tumor-bearing mice were injected with anti-CD19 CAR T-cells, and the 3G-ROR1-IOζ showed excellent response with no clinical evidence of progression ( Supplementary Figure 3 ). Three mice injected with the 2G-ROR1-Bζ showed progression and required euthanasia. As expected, the mice treated with untransduced T-cells (UTD) had rapid progression, and all of them died or required euthanasia by Day 28 post-T-cell injection ( Figure 3B ). Kaplan-Meier survival analysis showed that all the mice injected with ROR1-IOζ and CD19-Bζ CAR T-cells survived the 80-day observation study. The median Overall Survival (mOS) of the mice treated with UTD T-cells or ROR1-Bζ CAR T-cells was 20 and 65 days, respectively P< 0.05 ( Figure 3C ). The treated mice developed xenogenic graft-versus-host disease (GVHD), as previously reported on extended CAR T-cell in vivo experiments by various groups (52). However, the clinical GVHD manifestations observed in the mice receiving the 2G-ROR1-Bζ T-cells were the most severe.

We measured circulating anti-ROR1 CAR T-cells in peripheral blood samples collected weekly from Day 0 to Day 70, and we did not observe differences in the median number of cells identified by flow cytometry. For example, median CAR T-cells on Day 70 were 9.8 and 14.8 CAR T cells/μl of blood on ROR1-IOζ and ROR1-Bζ, respectively ( Figure 4A ). We analyze T_CM, T_EM phenotypic profiles (Day 7, Day 28, and Day 70), and exhaustion markers (Day 70) using the same samples. The cells were gated on CD45⁺CD3⁺CAR⁺CD4⁺ for this specific analysis. At Day 70, the circulating CAR T-cells from mice treated with the ROR1-IOζ construct showed a significantly higher number of T_CM and lower number of T_EM cells than ROR1-Bζ ( Figure 4B ). In addition, the CAR T-cells from the group of mice treated with the 3G-ROR1-IOζ T-cells, have low expression of exhaustion markers, including PD1, LAG3, TIGIT, and TIM-3 ( Figure 4C ). CD8⁺ T-cells expressing central memory phenotype corresponded to 82% of the total population by Day 70 in the mice treated with 3G-ROR1-IOζ CAR T-cells compared to only 12% on those receiving the 2G-ROR1-Bζ similar distribution observed in the 2G-CD19-Bζ ( Figure 5 ).