Anti-viral CD8 central memory veto cells as a new platform for CAR T cell therapy

Abstract

Central memory CD8 T cells exhibit marked veto activity enhancing engraftment in several mouse models of T cell-depleted bone marrow (TDBM) allografting. Graft-versus-host disease (GVHD) can be prevented by stimulation of mouse or human memory CD8 T cells against their cognate antigens under cytokine deprivation, in the early phase of culture followed by further expansion with IL21, IL15, and IL7. Thus, human anti-viral CD8 central memory veto T cells generated from CMV and EBV-positive donors are currently evaluated in a clinical trial at MD Anderson Cancer Centre (MDACC). Results in 15 patients indicate a low risk of GVHD. Considering that these cells could offer an attractive platform for CAR cell therapy, we evaluated methodologies for their effective transduction with 2 retroviral vectors. Initially, a vector directed against Her2 was tested and optimal transduction was attained at day 5 of culture. The transduced cells were expanded for an additional 7 days and exhibited marked anti-tumor reactivity ex-vivo while retaining their veto activity. Transduction with a vector directed at CD19 was effectively attained at days 4-5 allowing for substantial harvest of transduced cells at day 12 of culture. These Veto-CD19CAR central memory CD8 T cells exhibited marked anti-tumor reactivity in-vitro and in-vivo without GVHD, measured following transplantation into immune-deficient mice. These results strongly suggest that Veto-CAR T cells offer an attractive platform for CAR T cell therapy without gene editing for addressing the risk of GVHD or graft rejection.

Graphical Abstract

Graphical Abstract.

Significance statement.

CAR T cells represent a major breakthrough for the treatment of malignancies. Current FDA-approved cellular products are generated from the patient's own T cells, and transduced with the appropriate CAR in a time-consuming and costly procedure. Production of allogeneic off-the-shelf CAR T cells, available for immediate use is highly desirable. However, MHC disparities represent a major challenge, requiring gene editing strategies to prevent rejection and graft-vs-host disease which have the risk of inducing malignancies. We demonstrate that CD8 central memory veto T cells could potentially offer a new platform for CAR cell therapy without any need for gene editing.

Introduction

CAR-T cells have changed the treatment landscape for advanced hematological malignancies, including B cell lymphoblastic leukemia,^1,2 aggressive B-cell lymphoma,^3,4 indolent lymphoma,^5,6 and multiple myeloma.^7,8 All FDA-approved treatment modalities utilize autologous CAR T cells. The success of this approach is associated with several shortcomings including the inefficient and costly manufacturing process,^9,10 risk of malignant cell contamination¹¹ and in some instances poor quality of the harvested T-cell preparations due to previous exposure chemotherapy.^12,13

In contrast to this individualized approach of autologous CAR T cells, the use of CAR T cells from allogeneic donors can potentially address these limitations and offer off-the-shelf therapy, but issues related to MHC disparities represent a major barrier; the allogeneic cells are subject to rejection and can produce graft-vs-host disease. Attempts to address this challenge make use of gene editing, including ablation of TCRαβ to avoid GVHD, and expressing a single-chain HLA-E with minimal polymorphism fused to the B2M locus to reduce graft rejection by T and NK cells.^14-16 Alternatively, disruption of TCRα constant gene (TRAC) and CD52 gene locus, combined with the use of alemtuzumab, a pan-lymphocyte-depleting anti-CD52 monoclonal antibody, was used to minimize GVHD and graft rejection.^15,17 The latter is overcome by the selective elimination of host immune cells without ablation of the infused CAR T cells which are resistant to alemtuzumab. However, in addition to treatment-related toxicities associated in general with CAR T-cell therapy,^18,19 the immunosuppression caused by Alemtuzumab can also result in severe fungal or viral infections.^14,20,21 Furthermore, there is concern regarding the potential tumorigenicity of genetically modified T cells. Preliminary results from the phase I CALM trial, a first-in-human study of TCRαβ and CD52-depleted allogeneic CAR (UCART019) for advanced B cell leukemia, highlighted the issue of limited persistence of the CAR-T cells with median duration of 28 days in the peripheral blood and 20% of patients required a second infusion or subsequent hematopoietic stem cell transplantation (HSCT).²² Concerns for safety were also raised by a recent clinical trial using transcription activator-like effector nucleases (TALEN) for the generation of CAR T cells.²³

Another option for allogeneic CAR therapy makes use of CAR-NK cells which are devoid of GVH reactivity.²⁴ However, limited NK-cell persistence still represents a major challenge.²⁵

In the present study, we present a novel approach for the generation of off-the-shelf allogeneic CAR-T cells with minimal genetic manipulation based on leveraging the attributes of CD8 veto T-cells. Veto activity of CD8 T-cells, first defined in 1980 by Miller et al ²⁶ is based on the ability of these cells to target host CTL-precursors directed at allo-antigens presented by the veto cells, selectively eliminating anti-donor T-cell clones from the host immune system, and thereby preventing rejection. Further studies using knock-out mice showed that the veto activity of CD8 veto T cells is mediated by a Fas-FasL mechanism of apoptosis.²⁷ However, CD8 T cells are also endowed with GVH reactivity. Thus, this approach was basically abandoned until the GVHD risk was overcome in 2003 by using anti-3rd party CTLs.²⁸ Further studies in the context of reduced conditioning demonstrated the advantage of using central CD8 memory veto cells, as opposed to veto CTLs, directed against 3rd party stimulators. The ability of central memory CD8 anti-3^rd party veto cells to enhance engraftment of T-cell-depleted mis-matched BMT without GVHD was clearly demonstrated in 2013 in the same mouse models.²⁹ More recently, considering the difficulty of finding 3^rd party human stimulators which are not cross-reactive with the intended host, we chose to interrogate in mice the possibility of using recall antigens such as ovalbumin, which suggested the potential use of viral antigens as 3rd party antigens.³⁰ In particular, it should be noted that a prerequisite for successful depletion of GVH reactivity in the generation of anti-3rd party central memory veto CD8 T cells is based on cytokine deprivation at the early phase of the culture ex-vivo. This deprivation period leads to death by neglect of anti-host T-cell clones responsible for GVH reactivity. Finally, this approach was successfully translated to the human setting by the expansion of anti-viral memory CD8 T cells from healthy donors seropositive for CMV or EBV.³¹

Memory CD8 T cells were enriched by magnetic bead depletion of CD45RA+, CD4+, and CD56 + cells. Simultaneously, CD14+ monocytes were selected from the same donor and matured into dendritic cells (mDCs). These mDCs were loaded with CMV, EBV, Adenovirus, and BKV viral peptides, irradiated, and used as stimulators. The memory CD8 T cells were cultured with the mDCs in the presence of IL-21 for 3 days, followed by the addition of IL-15 and IL-7 3 days later. After a culture of 9-12 days, a CD8 T cell population expressing predominantly a central memory phenotype and exhibiting strong veto activity was obtained and this cell preparation exhibited more than 3 log depletion of alloreactivity against host stimulators in limiting dilution analysis (LDA).³¹ A first in human phase 1-2 clinical trial in MD Anderson, assessing the safety and efficacy of such anti-viral central memory veto cells in the context of a non-myeloablative T-cell-depleted haploidentical HSCT indicates that indeed these veto cells are associated with very low risk for GVHD.³²

Considering that anti-viral central memory CD8 veto cells evade rejection without GVH reactivity, and can potentially control dangerous viral infections, we hypothesized that these veto T cells could offer an attractive platform for allogenic off-the-shelf CAR therapy.

In the present study, we demonstrate the feasibility of generating effectively human Veto-CAR CD8 central memory T cells which can offer an attractive substitute for genetically edited allogenic CAR T cells.

Methods

Antibodies and reagents

Antibodies for flow cytometry included APC conjugated anti-human (γ-specific) Fc receptor (Invitrogen, Waltham, MA, USA), allophycocyanin (APC)-conjugated anti-human CD3 (Biolegend), Brilliant Violet -conjugated anti-human CD4- (Biolegend, San Diego, CA, USA), and eFluor450 conjugated anti-human CD8- (Invitrogen). RetroNectin (Takara Bio USA Inc., San Jose, CA, USA) was used for the preparation of pre-coated transduction plates. Human IL-2 was purchased from Novartis-Pharma. For the detection of Her2CAR (anti-Her2CAR) the anti N29-biotin antibody is used (anti-N29 is an anti-Her2CAR polyclonal antibody, prepared in house).

Cell lines and culture

PG-13 (gibbon ape leukemia virus pseudotyping packaging cell line; kindly provided by Ralph Wilson, Rotterdam hospital) was cultured in DMEM supplemented with 10% fetal calf serum (FCS), 2 mM glutamine, and 1 mM sodium pyruvate. Ovarian cancer-derived cell line SKOV3 (ATCC, kindly provided by Prof Zelig Eshhar, WIZ), and Raji, originated from Burkitt’s lymphoma (kindly provided by Prof Zelig Eshhar, WIZ and ATCC, CCL-86), NALM-6-Fluc-Neo/eGFP-Puro (CL-150, Imanis Life Sciences). Human lymphocytes were cultured in RPMI-1640 (Biological Industries) supplemented with 10% FCS, and 2 mM glutamine. All media were supplemented with a mixed antibiotic solution containing penicillin (100 U/mL), streptomycin (100 μg/mL), and neomycin (10 μg/mL) (Bio-Lab). Cells were incubated in a humidified 37^°C incubator with 5% CO₂, except for PG13, which was maintained in a 7.5% CO₂ atmosphere. All cells were verified to lack mycoplasma by PCR (HyLabs). Cells were maintained in culture for no longer than 4 weeks, which corresponds to approximately 12 passages. Mononuclear cells from human healthy donors were obtained from the Israeli blood bank.

Generation of anti-viral central memory CD8 veto cells (Veto cells)

Peripheral blood mononuclear cells (PBMC) were isolated from whole blood or leukapheresis of healthy volunteers by Ficoll density gradient centrifugation. The cells were cryopreserved in samples of 50 × 10⁶ cells/ml in 3 ml vials. On day -2, the cryopreserved cells were thawed and CD14+ monocytes were isolated using anti-human CD14 microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany; Ca# 130-050-201). The purified CD14+ cells were then used to generate mature dendritic cells (mDCs) as described below. On day 0, another fraction of mononuclear cells from the same donor was thawed, and cells were depleted of CD4+ T cells, CD56+ NK cells, and CD45RA+ T cells, using magnetic microbeads (Miltenyi Biotech) including anti-human CD4, (Ca# 130-045-101); anti-human CD56 (Ca# 130-050-401) and anti-human CD45RA (Ca# 130-045-901). The purified CD4-CD56-CD45RA- T-cell fraction was co-cultured with irradiated mDCs that were previously pulsed with a mixture of peptide libraries of viral proteins including CMV (HCMV pp65 GMP, Peptivator ca# 170-076-109), EBV (EBV Select, Miltenyi Biotech, Peptivator Ca# 170-076-143), Adeno virus (ADV Hexon GMP, Peptivator Ca# 170-076-106), and BKV (BKV LT GMP, Peptivator Ca# 170-076-139). These cells were cultured for the first 3 days in the presence of IL-21 (CellGenix Freiburg, Germany, Ca# 1019-50), and thereafter IL-15 (CellGenixCa# 1013-50) and IL-7 (CellGenix Ca# 1010-50) were added for additional 9 days. All reagents described above were of clinical grade.Generation of mDCs

Purified CD14 + cells were seeded (3 × 10⁶ cells/mL) in 10 cm tissue culture plates in 1% CellGro medium supplemented with 1% human serum, 1000 IU/mL IL4 (CellGenix Ca# 1003-050) and 2000 IU GM-CSF (CellGenix, Ca# 1012-050). Following 24 hours of incubation, 40 ng/mL LPS (InvitroGen, Ca# vac-3pelps) and 200 IU INF-γ (CellGenix, Ca# 1425-050) were added for an additional 18 hours to induce maturation of the dendritic cells.Veto activity assay

Veto activity assay is based on the ability of veto cells to delete CTLp, directed against antigens expressed by the veto cells themselves, but not against third-party Ags. Thus, in the assay we compare the frequency of CTLp in mixed lymphocyte reaction (MLR) cultures in which an alloreactive donor A is stimulated against irradiated stimulators originating from 2 donors, B and C in separate cultures in the presence of anti-viral Tcm cells originating from donor B only (see Supplementary Methods).

Transduction of veto cells with Her2CAR

Veto cells were transduced using pBullet N29-CD28-γ retroviral vector (cloned with Second-generation anti- Her2 CAR) as described previously.³³ Briefly, peripheral human blood lymphocytes (PBL) were isolated from the blood of healthy human donors by density gradient centrifugation on Ficoll-Paque (Axis-shield). Veto cells were generated as described above. At days 5-6 of culture, they were subjected to two consecutive retroviral transductions in RetroNectin pre-coated non-tissue culture-treated 6-well plates. After transduction, the Veto cells were cultured in the presence of the cytokine cocktail (IL-7 (30 IU/mL), IL-15 (125 IU/mL and IL21 (100IU/mL). Transduction efficiency and anti-tumor reactivity were monitored by flow cytometry 6 days after transduction.

Transduction of veto cells with CD19CAR retroviral vector

CD19CAR retroviral vector is a second-generation anti-CD19 CAR. The anti-CD19 scFv is conjugated with CD8 hinge, transmembrane domain, 4-1BB co-stimulatory, and CD3ζ activation domain (Supplementary Figure S1). Retrovirus-packed CAR construct was produced and purchased from Center for cell and Gene therapy Vector development Laboratory (Houston, TX, USA).

Details of the transduction procedure are provided in Supplementary Figure S2 . Briefly, a 6-well non-tissue culture plate was precoated with RetroNectin (Takara Bio) according to the manufacturer’s instructions. On day 4 of culture, CAR transduction was performed. Retroviral particles were added to the RetroNectin-treated plate and centrifuged at high speed 4000g for 90 minutes at 32 °C. Veto T cells were then seeded at a density of 0.5 × 10⁶ or 1.0 × 10⁶ cells per 5 mL /well, depending on the experimental conditions, supplemented with IL-7 (1200 IU/mL), IL-15 (200 IU/mL), and IL-21 (100 IU/mL) (Sartorius CellGenix GmbH Am Flughafen Freiburg, Germany) . Following 48-hour of transduction, the cells were collected and transferred to 6-well G-Rex (Wilson wolf) for further expansion till day 12 of the culture, in the presence of IL-7 (1200 IU/mL), IL-15 (200 IU/mL), and IL-21 (100 IU/mL). CD19CAR expression was measured by flow cytometry at 48 hours following transduction and the end of the culture or the day of mouse injection. Goat anti-human IgG Fcγ fragment specific Alexa Flor 647 was used (Jackson ImmunoResearch, West Grove, PA, USA) for detection analysis.

OKT3/αCD28 antibodies-activated T cells preparation

To obtain activated T cells, isolated PBMC following O.N. penning, were plated in 6-well non-tissue culture plate coated with OKT3 (1 mg/mL; Clone: OKT3)) and NA/LE anti-human CD28 antibodies (1 mg/mL; Clone: CD28.2, BD Pharmingen, Franklin Lakes, NJ, USA) at a density of 5 × 10⁶ cells per 5 mL/well. Primary human T cells were cultured in complete CTL medium containing 45% advanced RPMI-1640 media (Hyclone Laboratories, Logan, UT, USA), 45% Click’s medium (Irvine Scientific, Newport Beach, CA, USA), 5% HS (Hyclone Laboratories), 100U/mL Pen Strep (Gibco by Life Technologies, Waltham, MA, USA), and 2 mmol/ glutaMAX (Gibco by Life Technologies) supplemented with recombinant human IL-7 (10 ng/mL) and IL-15 (10 ng/mL). Following 48 hr, cells were collected from the OKT3/αCD28 plate and transferred to 6-well G-Rex plate for expansion.

Flow cytometry analysis of veto T cells and CD19CAR-veto T cells

Flow cytometry was performed on day 0 (co-culturing day), days 4, 5, or 7 (transduction day), and day 12 (end of cell line culture) to assess the phenotype and confirm the composition of the cell line populations. The following antibodies were used for the analysis: Ghost Dye Violet 510 (Tonbo Biosciences, Seattle, WA, USA), and fluorochrome-conjugated antibodies anti-CD3 mouse anti-human FITC (clone SK7), anti-CD3 mouse anti-human APC-H7 (clone SK7), anti-CD4 mouse anti-human APC-H7 (clone SK3), anti-CD8 mouse anti-human Pacific Blue (clone RPA-T8), anti-CD45 mouse anti-human V500-C (clone 2D1), anti-CD45RO mouse anti-human PE (clone UCHL1), anti-CD56 mouse anti-human PE-Cy7 (clone B159), anti-CD62L mouse anti-human APC (clone DREG-56), anti-CD62L mouse anti-human PE-Cy7 (clone DREG-56), anti-CD11c mouse anti-human PE (clone 3.9), anti-CD14 mouse anti-human PE-Cy7 (clone MφP9), anti-CD80 mouse anti-human FITC (clone L307.4), anti-CD83 mouse anti-human APC (clone HB15e), and their respective isotype controls (all BD Biosciences, Franklin Lakes, NJ, USA).

Flow cytometry analysis

Flow cytometry analysis was performed using a FACS-CantoII and LSR Fortessa x-20 (BD Biosciences, Franklin Lakes, NJ, USA) and analyzed on BD-FACS Diva and flowJo Software.

In-vitro assessment of Veto-CAR T cells reaction to their target

Veto-CAR T cells and their control non-transduced (UT) cells, were incubated for 24 hours with target cells (SKOV3 or Raji) at a 1:2 target: effector (T: E) ratio. Cell-free growth medium was collected and analyzed for -γ secretion by ELISA using a human IFN-γ ELISA kit, according to the manufacturer’s instructions (R&D systems, Minneapolis, MN, USA).

Killing assay using CountBright absolute counting beads

This method was used to test specific killing in vitro of CD19+ tumor cells by veto Tcm cells transduced with anti-CD19CAR. Briefly, transduced, and non-transduced veto cells were incubated with CD19 + Raji cells in 24 well flat bottom plate for 72 hours in 37 °C 5% CO₂. Thereafter, the cells were harvested and analyzed using CountBright absolute counting beads for flow cytometry (ThermoFisher Scientific, Waltham, MA, USA, Ca# C36950). Harvested cells were washed and stained for expression of CD19 with PE conjugated anti-CD19 antibody (BD Biosciences, Ca# C55541) and Ghost dye UV450 viability reagent (Tonbo Biosciences, Ca# 13-0868T100). Thereafter, samples were washed, supernatant was discarded and 1 mL PBS was added to the dry pellet to make sure the samples volume is the same. Thereafter, counting beads were added (50 µL) and the sample was analyzed by flow cytometer gating on the counting beads by using forward scatter threshold sufficiently low. A total of 5000 bead events was collected to assure a statistically significant determination of sample volume.

Calculation of cell concentration

The concentration of sample as cells/μL was defined by calculating, A/B × C/D, where: A = number of cell events, B = number of bead events, C = assigned bead count of the lot (beads/50 μL), D = volume of sample (Ll).

Mouse xenograft model for evaluation of target killing by veto-CD19CAR T cells

To assess the ability of Veto-CD19CAR T cells to eliminate CD19+ malignant cells in vivo, NALM-6 pre-ALL CD19+ cells expressing Firefly Luciferase (NALM-6-luc) were administrated by IV injection into the tail vain of NSG mice (NOD-SCID IL2Rgamma; Jackson Laboratory, Bar Harbor, ME, USA). Three days later CD19CAR-Veto T cells were administered by IV injection and tumor progression was evaluated by imaging of the bioluminescent reporter as previously described.³⁴ Briefly, tumor burden was monitored by recording luminescence from whole body imaging using Lumina x5 IVIS Imaging system every 3-5 days following intraperitoneal injection of 150 mg/kg body weight of d-Luciferin (PerkinElmer, Waltham, MA, USA). Photons emitted from luciferase-expressing cells were quantified using the Living Image software. Mice were euthanized after the tumor burden reached luminescence level of 10⁸ photons/s or after displaying signs of high tumor burden. Mice were maintained under conditions approved by the Institutional Animal Care and Use Committee (IACUC) at MD Anderson Cancer Center (protocol # 2430).

Mouse xenograft model for evaluation of GVH activity

To evaluate the GVH activity exerted by Veto -CD19CAR T cells, NSG mice (NOD-SCID IL2Rγ; Jackson Laboratory, Bar Harbor, ME, USA) underwent sub-lethal total body irradiation (TBI) at a dose of 3 Gy (day -1), using a Precision X-ray irradiator (X-RAD SmART). Following 24 hours (day 0), 15 × 10⁶ CD19CAR-Veto T cells were administered (i.v.) to irradiated mice in comparison to additional 2 mouse groups, one infused with 15 × 10⁶ OKT3/αCD28 activated T cells serving as positive control and the other not infused with cells, serving as irradiation (3GY) control. Post-injection, the mice were monitored for body weight twice a week and the appearance of GVHD symptoms including weight loss (> 15%), hunched posture, ruffled fur, reduced mobility, and survival.

Ethics

Human studies were conducted according to the principles expressed in the Declaration of Helsinki and approved by the IRB under protocol 0185-14-TLV for obtaining blood donation of normal human lymphocytes at Tel Aviv Sourasky Medical Center, Israel, Protocol 9812-12-SMC for obtaining blood donation of normal human lymphocytes at Sheba Medical Center, Israel. We also purchased, fresh HemaPrime leukopak from peripheral blood that was collected from healthy donors under an IRB approved protocol of Charles River company (Northridge, CA, USA).

Results

Transduction of anti-viral central memory CD8 veto cells with a retroviral CAR vector targeting Her2

To define the feasibility of transducing human anti-viral central memory CD8 Veto cells with CAR vectors, we initially assessed CAR transduction efficacy using CAR directed against Her2 and subsequently we evaluated the potential feasibility of effective transduction with CD19CAR which is clinically more applicable presently.

Thus, we first evaluated transduction efficiency with an Her2 (N29) CAR vector at various time points during the culture period employed for Veto cell production. In these experiments we used a pBullet retroviral vector comprising ScFv N29 previously described by Globerson et al³⁵ and Deshet-Unger, et al³⁶. In this vector ScFv N29 is fused with CD28 costimulatory and FcγR activating ITAM motif, along with a GFP reporter gene. The procedure for the generation of anti-viral veto cells, was as previously described by Bachar-Lustig et al³¹ Briefly, memory CD8 T cells enriched by magnetic beads depletion of CD45RA+, CD4+, and CD56+ cells were cocultured with irradiated viral peptide loaded mDCs that were previously generated from CD14+ monocytes from the same donor and loaded with CMV, EBV, Adeno, and BK viral peptides. Co-culture was initiated on day 0 in T-cell growth medium supplemented with IL-21 for 3 days and thereafter IL-15 and IL-7 were added. As shown schematically in Figure 1, transduction on 2 consecutive days was initially performed in calibration experiments on days 3-4, days 5-6, and days 7-8. The final transduced veto cell preparation was harvested 6 days after transduction day, and analyzed by flow cytometr for transduction levels.

Figure 1:

Timeline of anti-viral central memory CD8 Veto cell generation and transduction with a gammaretroviral Her-2CAR vector.³⁷ Cells were transduced at days 3-4, 5-6 and 7-8 of culture and tested 6 days later by cytofluorimetry for the expression of the vector.

As shown in Figure 2, assessment of transduction efficiency showed that optimal outcome was attained following transduction on days 5-6. Thus, we continued to evaluate the functionality of VETO-Her2-CAR T cells transduced on days 5-6 and harvested on day 12 of culture.

Figure 2:

Transduction of anti-viral central memory CD8 veto cells with a retroviral CAR vector targeting Her2 + (N29) at different timepoints of culture. Cells were Transduced on day 3-4, 5-6 and 7-8 and 6 days after transduction the expression of N29-Her2-CAR was evaluated by cytofluorimetry using anti-N29-biotin antibody. (A, D,G)) Percentage of CD3 + CD8 + T cells on day 12 of culture. (B,E,H) Percentage of CD45RO + CD62L + T cells on day 12 of culture. (C,F,I)Percnetage of transduced cells on day 12 of culture.

As shown in Figure 3, these cells exhibited marked specific response in-vitro to the Her2 antigen (Figure 3A) and retained marked veto activity as illustrated by their ability to deplete specifically alloreactive T cells in MLR directed against stimulators from the donor of the veto cells, compared to reduced inhibitory activity in MLR against third-party stimulators (Figure 3B-C).Transduction of anti-viral central memory CD8 Veto cells with a retroviral anti-CD19CAR vector

Figure 3:

Specific recognition of target cells and veto activity of Veto-N29/αHER2 CAR T cells transduced on days 5-6 and harvested on day 12 of culture. (A) Specific stimulation against Her2 + SKOV3 cells assessed by ELISA of INF-γ secretion compared to response of untransduced veto cells (UT). Avrages ± SD of stimulation in four different experiements are depicted. (B-C) Veto activity of N29/αHER2 CAR T cells. The veto assay comprised two stages. First- a 5 days bulk MLR culture against stimulators obtained from the veto cell donor or against third-party stimulators in the presence or absence of cells tested for veto activity (Veto: responder ratio of 5:1). Second- limit dilution analysis (LDA) using additional 7 days of culture against the original stimulators used in the first bulk culture. Results of LDA analysis shown at (B) represent specific cultures in which veto cells and stimulator cells are from the same donor (upper dashed line) and control non-specific culture in which veto and stimulator are from different donors (lower dashed line). Level of alloreactive T cell clones upon addition of veto cells in specific and non-specific cultures was calculated from the slope of the linear regression shown in (B) and presented as the minimal number of responder cells allowing detection of one alloreactive cell (C). Results depict a representative experiment out of two independent experiments.

Following the establishment of transduction feasibility as shown above, we further explored the possibility of using the veto cell platform for generating Veto-CD19CAR cells. To produce Veto-CD19CAR cells, a second generation anti-CD19 4-1BB ζ CAR gammaretroviral vector was used (Supplementary Figure S1-S2). Veto cell generation was as described above except that the final veto cell preparation was harvested on 12 of culture. On days 4, 5, or 7 samples of 1-2 × 10⁵ cells were transduced using 1.5 ml supernatant of CD19 retroviral vector on a RetroNectin coated 24 well plate. After 48 hours of incubation at 37 °C 5% CO₂, the transduced veto T-cell cultures were transferred to Grex10 and expanded in T-cell medium supplemented with IL21, IL15, and IL7 cytokines until the end of the culture on day 12. As shown in Figure 4A-C optimal transduction (68.1%) was attained on day 4 of culture, leading at day 12 to a very homogenous cell preparation comprising 99.8% CD3 + CD8+ T cells of which 89.3% exhibited a CD45RO + CD62L + central memory phenotype. Notably, upon transduction on day 4 of culture in 6-wells plates, in 3 independent experiments the average transduction percentage was 54.1 ± 15.4 % and recovery of transduced Veto-CD19CAR cells at day 12 of culture was (4 ± 2.3) × 10⁶ cells per 1 × 10⁶ Veto cells used for transduction on day 4. Thus, as shown in Figure 4D, a final yield of 6.5 ± 1.2 × 10⁶ Veto-CD19CAR cells could be obtained from an average input of 1 × 10⁸ mononuclear.

Figure 4:

Transduction of anti-viral central memory CD8 veto cells with a gammaretroviral anti-CD19CAR vector at different time point of culture. Transduction of 2 × 10⁵ veto cells was performed on days 4, 5 or 7 of cultureby cytofluorimetry on the end of the culture on day 12 (A-C; right). (A-C; left). Percentage of CD3 + CD8 + T cells on day 12 of culture. (A- C; middle) Percentage of CD45RO + CD62L + T cells on day 12 of culture. (D) Transduction efficacy and cell recovery of Veto-CD19CAR T cells following transduction on day 4 of culture. Recovery was calculated in three indepnednt experiments per 1 × 10⁸ MNC in the initiation of the procedure used for the isolation of CD4-CD56-CD45RA- cells.

Anti-tumor reactivity of Veto-CD19CAR T cells

Anti-tumor reactivity of Veto-CD19CAR T cells was initially evaluated using ELISA for INF-γ secretion. As shown in Figure 5A-C, depicting results of a typical experiment, 93.7% of the cells at the end of the culture were CD3 + CD8 + T cells (Figure 5A) and 66.5% of the cells expressed CD19CAR (Figure 5B), based on the gating data shown in Supplementary Figure S3A and retained their CD45RO + CD62L + central memory phenotype (Figure 5C). Notably, Marked INF-γ secretion was found following 24 hours in culture against CD19 + Raji cells (51.2 ± 5.4 × 10³ picogram/ml) compared to a very low level (0.081 ± 0.1 × 10³ pg/ml) upon incubation with non-transduced Veto cells (Figure 5D). Likewise, a CountBright killing assay using FACS analysis of surviving CD19 + Raji cells upon MLR culture of 72 hours with Veto-CD19CAR T cells at an effector-to-target cell ratio (E: T) of 1: 4, revealed robust killing (81 ± 3%) compared to 0% killing upon incubation with non-transduced veto cells (Figure 5E-5H).

Figure 5.

Veto-CD19CAR T cells display potent activation and anti-tumor killing. Cytofluorimetry analysis of transduced Veto T cells showing the level of CD3 + CD8 + T cells (A), CD19CAR + (B), and CD45RO + CD62L + (C) cells. (D) INF-γ secretion analyzed by ELISA reveals strong specific activation when Veto- CD19CAR-Veto T cells were exposed to CD19 + Raji target cells for 24 hours in contrast to background levels secreted upon incubation with non-transduced veto cells. (E-G) FACS analysis using CountBright™ absolute counting beads, revealing marked reduction in the remaining Raji cells upon incubation with -Veto-CD19CAR T cells, compared to low killing with non-transduced control or in culture of Raji alone. (H) Killing summary showing marked killing with transduced Veto cells (right) compared to no killing with non-transduced veto cells (middle) as calculated from the absolute CD19 + cells/µL of each culture. Statistical significance was calculated using T-test. P < 0·05 indicates statistical significance. In the figure four asterisks (****) denoted P ≤ 0.0001. The detailed gating strategies employed in the cytofluorimetry analysis are shown in Supplementary Figure S3. Countbright killing results shown in (D) represent four replicates from a representative experiment, out of three independent experiments, and INF-γ secretion results represent four replicates of a representative experiment, out of two experiments performed.

To assess the anti-tumor potential of Veto-CD19CAR T cells in vivo, NOD.Cg-Prkdc scid IL2rg tm1Wjl (NSG) mice were infused with 0.25 × 10⁶ NALM-6-Luc + CD19 + cells. Three days later, mice were transplanted with 3 × 10⁶ Veto-CD19CAR or non-transduced Veto cells (Figure 6A). Notably, as shown in a typical experiment (Figure 6B-D) significant reduction of tumor burden was observed in mice receiving Veto-CD19CAR T cells, compared to recipients of non-transduced veto cells or mice receiving tumor cells alone. This significant anti-tumor reactivity of Veto-CD19CAR T cells is clearly reflected by the bioluminescence images tested at different time points (Figure 6B) as well as by the analysis of radiance (Figure 6C-D). Thus, as shown in Figure 6B-D, the reduction of tumor load started to be statistically significant on day 8 and continued to be significant until day 24, the last day of monitoring. This reduction in tumor load also translated into a survival benefit, with Veto-CD19CAR T cell-treated mice exhibiting significant extended longevity, P < 0.0001 (Median survival time (MST) = 41 days) compared to the two control groups (MST = 27 and MST = 29 respectively, Figure 6E).

Figure 6:

Anti-tumor activity of Veto-CD19CAR T cells. (A) Schematic representation of the NSG mouse model used to evaluate anti-tumor activity. On day 0 Mice were infused with 0.25 × 10⁶ NALM-6-Luc + CD19 + cells (Tu). 3 × 10⁶ Veto-CD19CAR T cells or non-transduced veto cells were infused on day 3. Tumor burden was evaluated at different time points by Luciferase bioluminescence in these two groups compared to mice receiving tumor cells only. (B) Imaging of tumor burden at different time points (C) Average Radiance of mice treated with Veto-CD19CAR T cells (T) compared to mice treated with non-transduced (NT)Veto cells or mice receiving tumor only. Each data point represents the mean ± SD of radiance. (D) Statistical significance of the differences in Radiance between the three groups at different time points. Significance was calculated using one way ANOVA followed by multiple comparisons. (E) Survival of mice receiving infusion of Veto-CD19CAR T cells) compared to mice treated with non-transduced Veto cells or mice receiving tumor cells only. Median survival distributions of mice in each group were tested using the log- rank statistic. Statistical analysis was performed using GraphPad Prism software (Graphpad Software, San Diego, CA, USA). P < .05 indicates statistical significance. In the figure, this is presented as follows: one asterisk (*) for P ≤ .05, two asterisks (**) for P ≤ .01, three asterisks (***) for P ≤ .001, and 4 asterisks (****) for P ≤ .0001.

GVH reactivity of Veto-CD19CAR T cells compared to T cells expanded by polyclonal activation

To Examine GVH reactivity of Veto-CD19CAR T cells, NSG recipient mice conditioned with 3Gy TBI were used. In this experiment the GVH reactivity of Veto-CD19CAR T cells was compared to that exhibited by OKT3/αCD28-activated T cells (ATC) genrally used for transduction with CAR vecotrs. Thus, the mice were intravenously (i.v.) administered with 15x10⁶ Veto-CD19CAR T cells or with 15x10⁶ ATC. A 3rd group that was only irradiated and not transplanted served as control group. As can be seen in Figure 7, no weight loss could be found in the control group or in mice receiving the Veto-CD19CAR T cells (Figure 7A) and all mice in these groups survived (Figure 7B). In contrast, marked weight loss (Figure 7A) was found in the group receiving ATC, exhibiting hunched posture, decreased mobility, and ruffled fur, symptoms consistent with GVHD, and all these mice died by day 61 (Figure 7B).

Figure 7:

GVH reactivity of Veto-CD19CAR T cells compared to that exhibited by ATC. NSG mice conditioned with 3Gy TBI were infused with 15 × 10⁶ Veto-CD19CAR T cells or ATC. (A) Weight loss of mice in the two groups (mean ± SD) at different time points after transplantation. While there is no weight loss in the group receiving Veto-CD19CAR cells, marked weight loss was found in the group receiving ATC and statistical significance calculated by one way ANOVA at each time point is depicted for this group (B) Survival of the three groups at different time points. Statistical significance was calculated using Log-rank.

Discussion

During the past decade several approaches have been developed to address the challenges associated with allogeneic off-the-shelf CAR T-cell therapy. All are making use of extensive gene editing including TCR and MHC knockdown to reduce the risk for GVHD and graft rejection, respectively.

We have previously demonstrated in murine models that fully mis-matched anti-3rd party allogenic veto Tcm are markedly depleted of GVH reactivity and can avoid graft rejection following reduced intensity conditioning by virtue of their veto activity.^29,30,38-40 Thus, we hypothesized that these cells can potentially offer an attractive platform for CAR cell therapy without any need for gene editing other than that required for the transduction with a CAR vector. More recently, generation of human anti-3rd party veto Tcm was achieved, using GMP grade reagents, by stimulation and expansion against viral antigens under cytokine deprivation in the early phase of culture with further addition of IL21, IL7 and IL15 after the third day for a total culture time of 12 days. This methodology afforded large numbers of highly homogenous CD8 + CD45RO + CD62L + Tcm Veto cells, exhibiting marked veto activity and more than 3 log depletion of alloreactive T cell clones as measured by limiting dilution analysis. Anti-viral activity tested by intracellular expression of INF-γ and TNF-α, showed an average of 38.8 ± 19.6% positive cells on 6 hours of stimulation against the viral peptide mixture used for stimulation.³² Basically, In the early phase of the culture, selective depletion of alloreactive T cells under cytokine depravation is likely mediated through death-by-neglect of non-stimulated cells (conditions which are favorable to anti-viral clones which are activated and are not dependent for survival on provision of cytokines). However, further dilution of non-stimulated alloreactive T-cell clones continues upon addition of IL15 and IL7 under the prevailing stimulation against viral antigens, associated with dominant selective expansion of anti-viral T-cell clones.

Based on these results demonstrating a novel approach for depleting alloreactive T-cell clones from preparations of anti-viral CD8 veto cells, a first in human, phase 1-2 clinical trial testing the safety and efficacy of these veto Tcm was launched in MD Anderson Cancer Center. In this study, elderly patients with hematological malignancies receive haploidentical megadose T-cell depleted HSCT plus veto Tcm from the same donor, following very mild non-myeloablative conditioning. Preliminary results in 15 patients strongly indicate that indeed administration of Tcm is safe with very low risk for GVHD.³²

Based, on these encouraging results we interrogated in the present study the feasibly of transducing these human veto Tcm with retroviral CAR constructs.

Initially we found that marked transduction can be attained with a vector directed against Her2 and that the transduced Her2CAR-Veto cells retained their veto activity, as measured ex-vivo by limiting dilution analysis, and exhibited in-vitro marked specific killing of target cells comparable to that exhibited by regular CAR T cells.

Following these initial studies, we found similar efficacy of transduction using a CD19CAR vector and demonstrated marked specific killing of target tumor cells both in-vitro and in-vivo. Notably, evaluation of GVH reactivity in immune deficient recipient mice clearly demonstrated that Veto-CAR T cells do not induce any symptoms of GHVD in contrast to control T cells activated by OKT3/αCD28 antibodies (generally used for generating CAR T cells).

Taken together, these results suggest that anti-viral CD8 central memory veto cells can offer an attractive platform for off-the-shelf CAR T cells. In particular, the demonstration that the transduced cells are markedly depleted of GVH reactivity represents a major attribute, sparing the need of TCR knockdown or some other form of gene editing, used by all alternative approaches aiming at avoiding GVHD. Likewise, the ability of Veto cells to overcome T-cell mediated³¹ or NK cell⁴¹ mediated rejection, strongly indicate that using this platform could allow avoiding the need for yet another form of gene editing, aiming at increasing persistence of allogeneic CAR T or NK cells. These include different manipulations of MHC expression^42,43 or knockout of CD52 combined with the use of Alemtuzumab, a pan-lymphocyte-depleting anti-CD52 monoclonal antibody.^15,17

However, our results also indicate that the advantage associated with reduced risk of GVHD affects overall anti-tumor reactivity, which is mediated both by alloreactivity and by CAR mediated killing. Thus, using a similar 4-1BB vector, Mamonkin et al.³⁴ were able to demonstrate more effective tumor killing, prolonging survival in the same mouse model from 19 days to 29 days upon infusion of only 1 million conventional CAR T cells, compared to 3 million cells used in our study. Similar reduced anti-tumor reactivity was also demonstrated using gene editing of TCR in CAR T cells to reduce GVH reactivity.⁴⁴ In that study, a degree of survival prolongation similar to our results required 7 million CAR T cells. Notably, considering that in our clinical trial, we were able to escalate the number of infused non-transduced veto cells up to 10 million per kg without inducing GVHD, and considering that infusions of conventional CAR T cells are generally limited to 1 million cells per kg, we envision that it will be possible in the future to infuse 5-10 million VETO-CAR T cells per kg without causing GVHD, and thereby compensate for the overall reduced anti-tumor reactivity. Notably, as shown in Figure 4D, a final yield of 6.5 ± 1.2 × 10⁶ Veto-CD19CAR cells could be obtained from an average input of 1 × 10⁸ mononuclear cells at the beginning of the isolation procedure, indicating that for treating an adult patient weighing 70Kg with 5-10 × 10⁶ veto-CAR cells per kg, approximately 5.5-11 × 10⁹ mononuclear donor cells are required. Thus, one leukapheresis yielding 1 × 10¹⁰ cells could potentially afford a sufficient number of Veto-CD19CAR cells.

In conclusion, our results demonstrate the feasibility of attaining Veto-CAR T cells with minimal risk for GVHD. Considering the demonstration in different mouse models that veto cells can markedly overcome rejection of BM allografts when administered together with T cell depleted BM^29,30,38-40 and their marked persistence in MHC disparate recipient mice even in the absence of BM transplantation,⁴⁵ further clinical trials testing the safety, persistence and efficacy of Veto-CAR T cells, are warranted.

Author contributions

Wei-Hsin Liu, Anat Globerson Levin, Assaf Lask, and Esther Bachar Lustig designed, performed, and organized most of the experiments, analyzed and interpreted the data, and co-wrote the manuscript. Galit Horn, Tova Waks, Bar Nathansohn Levi, Irit Milman Krentsis, Xiaohua Su, Einav Shoshan, assisted in performing experiments and participated in discussions. Maksim Mamonkin provided the CD19CAR vectors and assisted in the design of the transduction protocol. Richard E. Champlin participated in editing the manuscript, Yair Reisner arranged financial support, designed, coordinated, and conducted the study, including analysis and interpretation of data, co-wrote the manuscript and provided final approval of the manuscript.

Funding

This work was supported in part by Cell Source Inc, by the Cancer Prevention and Research Institute of Texas (CPRITRR170008) and a staff appreciation and recognition reward (STARs award from the University of Texas system).

Conflicts of interest

Y.R. is a consultant and a share holder in Cell Source Inc. WHL; AGL; AL; GH; TW; BNL; EBL and YR are inventors in US patnet application #20230321235.

Data availability:

All data published in this article will be made available upon request.