Combining CD19- redirection and alloanergization to generate tumor-specific human T cells for allogeneic cell therapy of B-cell malignancies

Abstract

Allogeneic hematopoietic stem-cell transplantation can cure some patients with high-risk B-cell malignancies, but disease relapse following transplantation remains a significant problem. One approach that could be used to augment the donor T cell-mediated anti-tumor effect is the infusion of allogeneic donor-derived T cells expressing a chimeric antibody receptor (CAR) specific to the B-cell antigen CD19. However, the use of such cells might result in toxicity in the form of graft-versus-host disease mediated by CD19-specific (CD19-CAR) T cells possessing alloreactive endogenous T cell receptors. We therefore investigated whether non-alloreactive tumor-specific human T cells could be generated from peripheral blood mononuclear cells of healthy donors by the combination of CD19-redirection via CAR expression and subsequent alloanergization by allostimulation and concomitant blockade of CD28-mediated costimulation. Alloanergization of CD19-CAR T cells resulted in efficient and selective reduction of alloresponses in both CD4⁺ and CD8⁺ T cells including allospecific proliferation and cytokine secretion. Importantly, T-cell effector functions including CAR-dependent proliferation and specific target cytolysis and cytokine production were retained after alloanergization. Our data supports the application of CD19-redirection and subsequent alloanergization to generate allogeneic donor T cells for clinical use possessing increased anti-tumor activity, but limited capacity to mediate graft-versus-host disease. Therapy with such cells could potentially reduce disease relapse after allogeneic transplantation without increasing toxicity, thereby improving the outcome of patients undergoing allogeneic transplantation for high-risk B-cell malignancies.

INTRODUCTION

Disease recurrence is a major cause of mortality after allogeneic hematopoietic stem-cell transplantation (HSCT) for patients with poor risk B-lineage malignancies.(1–3) Adoptive transfer of allogeneic donor-derived T cells possessing additional anti-tumor activity has the potential to reduce relapse after allogeneic HSCT. The combination of such an approach with a strategy to selectively control alloresponses to limit toxicity from graft-versus-host disease (GvHD) might improve the outcome of allogeneic HSCT for patients with B-cell malignancies.

The introduction of a chimeric antibody receptor(4) (CAR) to redirect human T-cell specificity is one strategy to enhance desired T-cell-mediated anti-tumor activity.(5) CARs typically consist of an HLA-independent high-affinity antigen recognition domain formed from extracellular single-chain immunoglobulin variable fragments, linked to one or more cytoplasmic T-cell activation domains, including CD3-ζ. Infusion of patient-derived T cells expressing a tumor-associated antigen-specific CAR has resulted insome disease responses in early clinical trials for CD20⁺ B-cell lymphomas and GD₂⁺ neuroblastoma, but in other trials apparently limited in vivo persistence of CAR T cells restricted their therapeutic potential.(6–10)

CD19, an early cell surface B-lineage-restricted molecule, is expressed on both normal B cells and a wide range of human B-cell malignancies.(11) Therefore human CD19-specific CAR T cells have been developed to redirect a T cell-mediated anti-tumor effect.(12, 13) Second-generation CD19-CAR cells possessing modified co-stimulatory signaling domains fused to chimeric CD3-ζ, have improved in vivo persistence and antitumor efficacy in mice.(14, 15) To facilitate the clinical use of CAR⁺ T cells we and others have recently employed an augmented non-viral gene insertion strategy (the Sleeping Beauty (SB) transposon/transposase system) to introduce a second generation CD19-CAR into primary human T cells.(16–19)

CAR⁺ T cells have not yet been infused in the human allogeneic setting. Allogeneic CAR⁺ T cells could provide an additional donor-derived T cell-mediated anti-tumor effect to protect against tumor relapse after allogeneic HSCT. The use of allogeneic rather than patient-derived CAR⁺ T cells would also eliminate the risk of tumor cell contamination. Additionally, reconstituting donor-derived T cells which have not undergone CD19-redirection, but are reactive to recipient hematopoietic tissue-restricted minor histocompatibility antigens could provide protection against CD19^neg BALL precursors(20, 21) which would not be selectively targeted by CD19-CAR cells.

However, CAR⁺ T cells possess endogenous αβ T-cell receptors (TCR), and infused allogeneic CAR⁺ T cells bearing αβ TCR specific to recipient alloantigens could potentially mediate GvHD. Non-selective approaches to reducing alloreactivity after allogeneic HSCT (such as pharmacological immunosuppression) would likely reduce the ability of allogeneic CAR to expand and function in vivo. Therefore strategies have been developed to selectively reduce alloreactivity in donor T cells after allogeneic HSCT.(22–24) We and others have previously shown that one such strategy, alloanergization by allostimulation concomitant with blockade of CD28-mediated co-stimulation, effectively and selectively reduces alloreactivity of HLA-mismatched human peripheral blood mononuclear cells (PBMC).(25–27) Furthermore we have successfully applied this strategy in two prior clinical trials to selectively reduce alloreactivity of HLA-mismatched donor T cells in the setting of haploidentical bone marrow transplantation.(28, 29) A significant proportion of SB-modified CD19-CAR cells propagated on artificial CD19⁺ antigen presenting cells (aAPC) bearing the co-stimulator ligand CD86 express CD28,(18) suggesting alloanergization would be a suitable technique to selectively reduce alloreactivity in such cells.

In order to develop a clinical strategy to increase anti-tumor activity in allogeneic donor T cells while controlling alloreactivity, we investigated whether our established strategy of alloanergization could abrogate alloresponses of second-generation CD19-CAR cells without loss of viability, phenotype, and CAR-dependent T-cell effector functions.

MATERIALS AND METHODS

Plasmids

The SB transposon contains the codon optimized (CoOp) second-generation CD19RCD28 CAR, specific for human CD19, flanked by the SB inverted repeats. The ampicillin resistance gene (Amp^R) and origin of replication from the plasmid _CoOpCD19RCD28/pT-MNDU3(18) was replaced with the DNA fragment encoding the kanamycin resistance gene (Kan^R) and origin of replication (ColE1) from the pEK vector,(30) and the human elongation factor-1α (hEF-1α) promoter fragment from pVitro4 vector (InvivoGen, San Diego, CA) was swapped with MNDU3 promoter to generate CD19RCD28/pSBSO (also referred to as CD19RCD28mZ(CoOp)/pSBSO). The SB hyperactive transposase, SB11 under the control of CMV promoter from the plasmid pCMV-SB11(18) was ligated with the pEK vector fragment encoding Kan^R and ColE1 to generate pKan-CMV-SB11.

Cell Lines

CD19⁺Daudi (Burkitt Lymphoma, #CCL-213) and CD19^negK562 (erythroleukemia, #CCL-243) cells were obtained from American Type Culture Collection (Manassas, VA). CD19⁺NALM-6 (pre-B cell, #ACC128) and CD19⁺GRANTA-519 (B-cell non-Hodgkin lymphoma, #ACC342) cells were from DSMZ (Braunschweig, Germany). CD19^negLM7 (osteosarcoma) was a kind gift from Dr. Eugenie Kleinerman, M.D. Anderson Cancer Center, Houston, TX. Cell lines were maintained in HyQ RPMI 1640 (Hyclone Logan, UT) supplemented with 2 mmol/L Glutamax-1 (Invitrogen, Carlsbad, CA) and 10% heat-inactivated FCS (Hyclone) (10% RPMI). CD19⁺K562 were generated and maintained in 10% RPMI with HygroGold (Hygromycin B, 0.4mg/mL; InvivoGen) as described.(31) CD19^negU251T (glioblastoma) was a kind gift from Dr. Waldemar Debinski, Wake Forest University, NC. U251T were transfected with SB DNA plasmid (pSBSO) expressing truncated CD19 (ΔCD19/pSBSO) to generate CD19⁺U251T.(31) The U251T cell lines were maintained in 10% RPMI with G418 (0.2mg/mL; InvivoGen).

Generation of CD19-CAR cells

PBMC isolated by Ficoll-Paque (GE Healthcare, Uppsala, Sweden) density gradient centrifugation of peripheral blood obtained from healthy adult volunteer donors after informed consent from Gulf Coast Regional Center (Houston, TX) were cultured in HyQ RPMI 1640 (Hyclone, Logan, UT) supplemented with 2 mmol/L Glutamax-1 (Life Technologies-Invitrogen, Carlsbad, CA) and 10% heat-inactivated defined FCS (Hyclone). The SB transposon/transposase were electro-transferred (Amaxa/Lonza, Cologne, Germany) into T cells derived from PBMC and a population of CD19-CAR cells were numerically expanded on γ-irradiated (100 Gy) K562-artificial antigen presenting cells (aAPC) expressing CD19, 4-1BBL, CD86, CD64, and membrane-bound IL-15) as previously described, Figure 1A–B.(18, 32)

Figure 1. Generation and alloanergization of adult donor-derived CD19-CAR cells.

(A) Electroporation of human T cells with SB DNA plasmids and propagation on CD19-specific K562-derived aAPC. After electroporation, T cells were co-cultured with γ-irradiated K562 (genetically modified to co-express CD19, CD64, CD86, 4-1BBL and surface membrane-bound IL-15) with addition of soluble IL-2 every alternate weekday, resulting in expansion of stably transfected CAR⁺ T cells to numbers suitable for use in adoptive cell therapy trials. (B) Schematic of the SB DNA plasmids. _CoOpCD19RCD28/pSBSO (Transposon): EF-1α promoter, human elongation factor-1α promoter; _CoOpCD19RCD28, codon-optimized CD19RCD28 CAR; IR, SB-inverted/direct repeats; bGh pAn, polyadenylation signal from bovine growth hormone; Kan^R, kanamycin resistance gene. pKan-CMV-SB11 (Transposase): SB11, SB-transposase SB11; CMV promoter, CMV enhancer/promoter; SV40pAn, polyadenylation signals from SV40. (C) Alloanergization of human CD19-CAR cells by allostimulation with co-stimulatory blockade. T cells are co-cultured with γ-irradiated allostimulator PBMC in the presence of antibody-mediated blockade of CD28-mediated co-stimulation. T cells possessing alloreactive endogenous αβ TCR receive Signal 1 (alloantigenic stimulus), but not Signal 2 (CD28-mediated co-stimulation). This triggers intracellular events rendering the alloreactive T-cell hyporesponsive (anergic) to subsequent alloantigenic challenge, even in the presence of CD28-mediated co-stimulation.

Alloanergization of CD19-CAR cells and measurement of secondary alloresponses

Equal numbers of CD19-CAR cells and γ-irradiated (3.5 Gy) first-party allostimulator PBMC (isolated from healthy unrelated adult volunteers after consent on an IRB-approved protocol) were co-cultured in culture media (RPMI, penicillin/streptomycin, 10% human AB serum, Sigma-Aldrich (SA), St. Louis, MO) with or without humanized monoclonal anti-B7.1 (clone h1F1) and anti-B7.2 (h3D1) antibodies (10μg/10⁶ cells, Wyeth, Madison NJ) as described(26) and outlined in Figure 1C. After 72 hours, co-cultures with anti-B7 antibodies (“alloanergized”) and without anti-B7 antibodies (“non-anergized”) were washed and resuspended. Secondary alloresponses were measured after restimulation with γ-irradiated first- or third-party allostimulator PBMC or soluble CD3 and CD28 antibodies (10μg/mL, Beckman Coulter, Fullerton, CA). Proliferation was determined by thymidine incorporation as described.(26)

Alloanergization efficiency (AE) was calculated as

$$\text{AE}=100-\{{100}^{\ast }(\text{Secondary}\text{Alloproliferation}(\text{alloanergized}\text{cells})/\text{Secondary}\text{Alloproliferation}\text{(non}-\text{anergized}\text{cells)}\}.$$

Additionally, CD19-CAR cells were labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE, Invitrogen, Carlsbad, CA) prior to restimulation.(26) Allospecific precursor frequency was calculated as previously described,(33) using Flowjo V4 software. Cytokine responses after allorestimulation were also measured using intracellular cytokine flow cytometry (ICC).(34) Positive controls were stimulated with Staphylococcal Enterotoxin B (SEB, 10μg/mL, Sigma-Aldrich). Stimulus-specific responses were calculated by subtracting values for unstimulated cells from values for stimulated cells.

CD19-depletion of allostimulator PBMC

Allostimulator PBMC were depleted of CD19⁺ B cells by labeling with conjugated anti-CD19 paramagnetic microbeads (Miltenyi Biotec, Auburn, CA) and passage through a LD column and midiMACS magnetic device before irradiation. PBMC were analyzed by flow cytometry to assess efficiency of depletion.

Flow cytometry

Unless stated, antibodies were from Beckman Coulter. For alloresponses assessment, cells were stained with anti-CD3 (clone UCHT1), -CD4 (13B8.2), -CD8 (SFCI21Thy2D3), -CD14 (M5E2, Becton Dickinson (BD), San Jose, CA) and -CD19 (RMO52) antibodies conjugated to Fluorescein Isothiocyanate (FITC), Phycoerythrin (PE), Energy Coupled Dye (ECD), PE-Cy5 and/or PE-Cy7. Viability was assessed by 7-Amino Actinomycin D (7-AAD, BD). For ICC, cells were stained for surface molecules, fixed, permeabilized and then stained with IFN-γ-FITC (4S.B3, BD) and TNF-α-PE-Cy7 (MAb11, BD). Events were acquired on a Cytomics FC500 flow cytometer and analyzed using Flowjo version 4. For detection of cell-surface expression of CD19-CAR, Goat F(ab′)₂ anti-human IgG (γ) -PE (Invitrogen #H10104) or -FITC (Jackson ImmunoResearch Laboratories Inc., West Grove, PA #109-096-170) were used (1/20 dilution) with anti-CD4 (RPA-T4, BD), -CD8 (SK1, BD) and -CD3 (SK7, BD). Anti-CD27 (M-T271), -CD28 (L293), -CD45RO (UCHL1), -CD62L (Dreg 56) and -CD95 (DX2) were used for memory cell phenotyping (all BD). Nonspecific antibody binding was blocked using 2% FBS buffer. For CAR-dependent IFN-γ secretion, 10⁵ CD19-CAR cells were incubated for 4–6 hours with 0.5×10⁶ stimulator cells or phorbol 12-myristate 13-acetate (PMA, 5ng/mL and ionomycin, 500ng/mL) in 200μL culture medium with Golgi Plug (BD), fixed, permeabilized and stained for intracellular IFN-γ (B27, BD). Events were acquired on a FACSCalibur flow cytometer and analyzed using CellQuest version 3.3 (both BD).

Measurement of cytotoxicity

The cytolytic activity of CD19-CAR cells was determined by 4 hour chromium release assay.(13) CD19-CAR cells were incubated with 5 × 10^{3 51}Cr-labeled target cells in V-bottomed 96-well plates (Costar, Cambridge, MA). The percentage of specific cytolysis was calculated from the release of ⁵¹Cr, as described, using a TopCount NXT (Perkin-Elmer).

Statistics

Statistical analysis was performed with Graph Pad Prism v4 (San Diego, CA). A p value of <0.05 was used to reject the Null Hypothesis. Data are presented as mean (+/− SD) unless otherwise stated. P values are for 2 tailed paired t tests.

RESULTS

Proliferative alloresponses of CD19-CAR cells were specifically reduced after alloanergization

We screened CD19-CAR T-cell lines from 6 adult donors for secondary (recall) alloproliferative responses after allostimulation (priming) and subsequent allorestimulation with γ-irradiated PBMC from 18 different unrelated adult volunteers. Secondary alloproliferative responses were detectable in all CD19-CAR T-cell lines. We next examined the efficacy and specificity of alloanergization in reducing secondary alloresponses. Viability of CD19-CAR cells was similar before (87%+/−7%, CD4⁺ and 92%+/−4%, CD8⁺) and after alloanergization (83%+/−11%, CD4⁺ and 90%+/−5%, CD8⁺). Alloanergized CD19-CAR cells were hyporesponsive to first-party allorestimulation. This was not due to a change in kinetics of the alloproliferative response, Figure 2A. Reduction of peak first-party-specific alloproliferation after alloanergization was seen in all CD19-CAR cell lines, whereas there was no significant change in third-party-specific alloproliferation or mitogen-stimulated proliferation, demonstrating that hyporesponsiveness was specific to alloantigens used during alloanergization, Figure 2B. The median efficiency of alloanergization was 82% (range 33–96%) with a median 5.4-fold (range 1.5–26) reduction in first-party alloproliferative responses. Third-party and mitogen-stimulated proliferation were not reduced, Figure 2C. These data demonstrate that alloanergization specifically reduced alloproliferation of CD19-CAR cells, consistent with our previous data for alloanergization of non-genetically modified human PBMC.(26, 35)

Figure 2. Alloproliferation of CD19-CAR cells was specifically reduced by alloanergization.

(A) Mean values (+/− SD) for proliferation ([³H]-thymidine incorporation) of non-anergized and alloanergized CD19-CAR cells after allorestimulation for 18 stimulator-responder pairs. (B) Peak proliferation in non-anergized and alloanergized CD19-CAR cells after restimulation with first- or third-party allostimulators or mitogenic CD3 and CD28 antibodies. Symbols represent means of triplicate values for unique stimulator-responder pairs. Horizontal bars depict mean values. (C) Fold reduction in proliferation of alloanergized CD19-CAR cells (compared to non-anergized cells) after restimulation with allostimulators or mitogen is shown as box-and-whisker plots. Horizontal bars represent medians, boxes interquartile range and whiskers minimum and maximum values. (D) Efficiency of alloanergization of CD19-CAR cells using unsorted PBMC or CD19⁺ B cell-depleted PBMC as allostimulators from allostimulators Results depict 3 separate experiments. Horizontal bars and adjacent numbers depict median values.

The use of CD19-depleted allostimulators resulted in a further reduction of residual alloresponses after alloanergization of CD19-CAR cells

Proliferation of CD19-CAR cells after stimulation with allogeneic PBMC containing CD19⁺ cells could result from either alloantigen-specific stimulation via endogenous TCR or from direct CD19-mediated stimulation via CAR. The latter could provide a confounding factor in our proliferation assays. Therefore we sought to determine whether the presence of CD19⁺ cells within allostimulator PBMC affected the residual proliferation of alloanergized CD19-CAR cells. We depleted CD19⁺ cells from allostimulator PBMC prior to γ-irradiation, resulting in a median 500-fold reduction in CD19⁺ cells. Proliferative responses after allorestimulation of non-anergized CAR⁺ T cells were retained using CD19-depleted allostimulator PBMC, consistent with retention of allostimulatory capacity. However, the use of CD19-depleted allostimulator PBMC resulted in a lower residual proliferation after allorestimulation of alloanergized CAR⁺ T cells compared with the use of unsorted allostimulator PBMC, suggesting that a direct stimulatory effect mediated by CD19⁺ cells within allostimulator PBMC contributed to the residual proliferation after alloanergization. As a result, measured efficiency of alloanergization was significantly higher (median 93%, range 87–94%) using CD19-depleted allostimulator PBMC when compared to unsorted allostimulator PBMC (median 72%, range 55–77%, p=0.04), Figure 2D.

Alloanergization reduced alloproliferation in both CD4⁺ and CD8⁺ CD19-CAR cells

As CD19-CAR cells contained both CD4⁺ and CD8⁺ T cells, we next determined whether alloanergization reduced alloproliferation in both cellular subsets. We labeled CD19-CAR cells with CFSE prior to allorestimulation, and measured proliferation by CFSE dilution. After six days of allorestimulation 21% (+/−6.2%) and 15% (+/−6.1%) of non-anergized CD4⁺ and CD8⁺ CD19-CAR cells had proliferated. This represented a median CD4⁺ and CD8⁺ alloprecursor frequency of 1.5% (range 1.0–1.8%) and 1.4% (range 1.3–1.9%), respectively. In contrast, the mean percentage of CD4⁺ and CD8⁺ T cells proliferating after allorestimulation of alloanergized CD19-CAR cells was significantly lower at 6.9% (+/−4.7%) and 4.1% (+/−3.6%) respectively. Importantly, proliferation of both CD4⁺ and CD8⁺ CD19-CAR cells after mitogenic stimulation was unaffected by alloanergization (TABLE 1). Thus, CD19-CAR cells contain both alloproliferative CD4⁺ and CD8⁺ T cells, and alloproliferation in both these subsets was specifically reduced after alloanergization.

Table 1.

Percent^* of CD4⁺ and CD8⁺ subsets of non-alloanergized and alloanergizecd CD19-CAR⁺ T cells proliferating after restimulation with alloantigens or mitogenic CD3 and CD28 antibodies

	Allorestimulation		Mitogen Restimulation
	CD4+	CD8+	CD4+	CD8+
Non- alloanergized cells	21 (+/− 6.2)	15 (+/− 6.1)	53 (+/− 5.1)	48 (+/− 3.6)
Alloanergized cells	6.9 (+/− 4.7)	4.1 (+/− 3.6)	61 (+/− 8.4)	59 (+/− 9.7)
p value†	0.01	0.02	0.30	0.22

^*+/− sd,

^†two-tailed paired t test comparing values for non-alloanergized and alloanergized cells

Alloanergization of CD19-CAR cells reduced allospecific cytokine production

Alloreactive human CD4⁺ and CD8⁺ T cells secrete pro-inflammatory cytokines (predominantly IFN-γ and TNF-α) after HLA-mismatched allostimulation.(34) Therefore, we used ICC to examine the impact of alloanergization upon allospecific cytokine responses of CD19-CAR cells. Allospecific cytokine⁺ T cells were detected within CD19-CAR T-cell populations with cell frequencies of 3.8% (+/−2.7%, CD4⁺IFN-γ⁺), 0.8% (+/−0.7%, CD8⁺IFN-γ⁺), 2.5% (+/−1.4%, CD4⁺TNF-α⁺) and 0.8% (+/−0.7%, CD8⁺TNF-α⁺). Allospecific cytokine⁺ T-cell frequencies were reduced in alloanergized CD19-CAR cells to cell frequencies of 0.4% (+/−0.2%, CD4⁺IFN-γ⁺), 0.15% (+/−0.1%, CD8⁺IFN-γ⁺), 0.20% (+/−0.1%, CD4⁺TNF-α⁺) and 0.17% (+/−0.02% CD8⁺ TNF-α⁺). This represented median fold reductions of 11 (range 3–17, CD4⁺IFN-γ⁺), 15 (range 5–25, CD4⁺TNF-α⁺), 2.3 (range 1.7–19, CD8⁺IFN-γ⁺) and 2.6 (range 1.8–10, CD8⁺TNF-α⁺). In contrast, CD19-specific cytokine responses after stimulation with CD19⁺ Daudi and U251T targets were only modestly reduced after alloanergization. CD19-specific IFN-γ⁺ cell frequencies were 49% (non-anergized) and 26% (anergized) (Daudi) and 50% (non-anergized) and 38% (alloanergized) (CD19⁺U251T) representing a 1.9- and 1.3-fold reduction respectively, Figure 3. This demonstrated that alloanergization preferentially reduced allospecific cytokine responses within CD19-CAR cells whilst maintaining the majority (but not all) of CD19-specific responses.

Figure 3. Alloanergization of CD19-CAR cells resulted in reduced allospecific cytokine production.

(A) Cytokine secretion by non-anergized and alloanergized CD19-CAR cells after restimulation with allostimulators or SEB. Flow cytometer dot plots are shown depicting intracellular cytokine production in CD4⁺ and CD8⁺ cells, gated on CD3⁺ events excluding irradiated stimulator cells. Boxed regions represent cytokine⁺ events and numbers represent frequency of cytokine⁺ events expressed as a percentage of CD4⁺ or CD8⁺ cells. Results are shown for a representative experiment (of 3). (B) Frequencies of cytokine⁺ cells in non-anergized and alloanergized CD19-CAR cells after restimulation with allostimulators, SEB or CD19⁺ target cells.

Phenotypic characteristics of CD19-CAR cells after alloanergization

Although the proportion of CD8⁺CD19-CAR cells expressing surface CAR was not affected by alloanergization, the proportion of CD4⁺CD19-CAR cells expressing surface CAR was temporarily reduced by up to 50%. The majority of CD19-CAR cells were CD45RO⁺CD27^neg memory cells both before and after alloanergization. Using co-expression patterns of CD28 and Fas (CD95) to distinguish CD28^negCD95⁺ effector memory T cells (T_EM) from CD28⁺CD95⁺ central memory (T_CM) cells,(36) we were able to identify similar proportions of T_CM (24.2% and 37.2%) and T_EM (75.8% and 62.8%) cells before and after alloanergization of CD19-CAR cells. Using co-expression patterns of CD45RO and CD62L(37), CD19-CAR cells also contained similar proportions of T_CM cells (31.0% and 45.7%) and T_EM (69.0% and 53.8%) before and after alloanergization, Figure 4A. This is consistent with CAR transgene expression in T_CM cells and preservation of these cells after alloanergization.

Figure 4. Characterization of CD19-CAR cells before and after alloanergization.

(A) Representative flow cytometry dot plots showing cell surface expression of CD19-CAR on CD4⁺ and CD8⁺ T cells and proportions of CD19-CAR cells with a memory T cell (CD27^negCD45RA⁺) and central memory (T_CM, defined as CD28⁺CD95⁺ or CD62L⁺CD45RO⁺) before and after alloanergization. Numbers represent percentages of cells in each quadrant. (B) Killing of CD19⁺ target cells (Granta 519; NALM-6; HLA class I^neg Daudi; HLA class I/II^neg K562 cells transfected to express truncated CD19) in a 4 hour ⁵¹Cr release assay by non-anergized and alloanergized CD19-CAR cell effectors. Background lysis of CD19^neg (parental K562 and LM7) cells is also shown. Results of mean (points) ± SD (bars) of specific lysis of triplicate wells are shown. (C) Killing of cultured first-party allogeneic targets by CD19-CAR cell effectors in a 4 hour ⁵¹Cr release assay. Results from one representative experiment (of 2) are shown.

CD19-specific cytolytic function of CD19-CAR cells was preserved after alloanergization

CD19-CAR cells were evaluated for redirected killing before and after alloanergization in a 4 hour ⁵¹chromium release assay. CD19-CAR cells effectively lysed CD19⁺ B-cell lines before and after alloanergization (Daudi, before, 59.4%; after, 56.7%, NALM-6, before, 29%; after, 26.7% at an effector: target ratio of 20:1). Retention of CD19-specificity was demonstrated by the 1.8-fold (before alloanergization) and 2.2-fold (after alloanergization) increased killing of CD19⁺ targets when compared to CD19^negK562 targets at an effector: target ration of 20:1. In comparison, alloanergization resulted in a 5-fold reduction in lysis of cultured first-party allogeneic target cells at an effector: target ratio of 20:1, Figure 4B.

CD19-specific proliferation of CD19-CAR cells after alloanergization

We next compared the capacity for CAR-dependent proliferation of non-anergized and alloanergized CD19-CAR cells using CD19⁺ aAPC without or with CD64 (FcγRI)-loaded OKT3 (to provide an additional antigen-independent proliferative signal). Numbers of non-anergized CD19-CAR cells were expanded by 3–4 logs over 21 days on both CD19⁺ aAPC and OKT3-loaded CD19⁺ aAPC. Alloanergized CD19-CAR cells retained the capacity to expand on both CD19⁺ aAPC and OKT3-loaded CD19⁺ aAPC, with a 2–2.5 log expansion over 21 days. Although CD19-dependent expansion was 1 to 2 logs less than that seen in non-anergized CD19-CAR cells, the retention of capacity to proliferate in a CAR-dependent manner implies that alloanergization did not substantially interfere with the ability of the CAR to provide and sustain a proliferative signal. In contrast, alloanergized CD19-CAR cells could not be expanded by repeat stimulation with first-party allostimulators, Figure 5A. Furthermore, expansion of alloanergized CD19-CAR cells on OKT3-loaded CD19⁺ aAPC restored the proportion of CD4⁺CD19-CAR cells expressing surface CAR to similar levels seen in non-alloanergized cells, Figure 5B. Importantly, alloanergized CD19-CAR cells remained hyporesponsive to allostimulation after in vitro expansion on CD19⁺ aAPC, Figure 5C. Finally, to confirm that CD19-expanded alloanergized CAR⁺ cells retained effector function, we again examined CD19-specific IFN-γ production. In vitro expanded alloanergized CD19-CAR cells retained up to 70% of their capacity to produce IFN-γ after contact with cell-surface CD19 when compared with expanded non-anergized CD19-CAR cells. Intracellular cytokine staining demonstrated a 3-fold increase in IFN-γ production when alloanergized CD19-CAR cells were stimulated with a CD19⁺ B-cell line (Daudi). IFN-γ production was 1.6 fold greater when alloanergized cells were stimulated with CD19⁺ transfected U251T glioma cells in comparison to CD19^negU251T cells, demonstrating CD19-specificity of IFN-γ production, Figure 5D. These data are consistent with retention of capacity of expanded alloanergized CD19-CAR cells to be activated via the introduced CD19-CAR.

Figure 5. CD19-dependent expansion of alloanergized CD19-CAR cells.

(A) Expansion of non-anergized and alloanergized CD19-CAR cells on CD19⁺ aAPC, OKT3-loaded CD19⁺ aAPC and first-party allostimulators (B) Expression of surface CAR on non-anergized and alloanergized CD4⁺CD19-CAR cells after 21 days of expansion on OKT3-loaded CD19⁺ aAPC. Numbers represent percentages of cells in each quadrant. (C) First-party alloproliferation ([³H]-thymidine incorporation) of non-anergized, alloanergized and alloanergized then aAPC-expanded CD19-CAR cells after first-party allostimulation. (D) CD19-specific intracellular IFN-γ secretion of aAPC-expanded CD19-CAR cells before and after alloanergization following incubation with CD19⁻ and CD19⁺ stimulator cells. Events are gated on CD3⁺ lymphocytes. Numbers represent percentages of cells in each quadrant.

DISCUSSION

Disease relapse remains a major cause of treatment failure after allogeneic HSCT, especially in patients with advanced B-cell malignancies. A significant unmet need, therefore, is a clinically applicable strategy to enhance the anti-tumor effect of allogeneic donor T cells. In the present study, we have developed such a strategy, by combining the approaches of redirection of human donor T cells to the B-cell antigen CD19 and alloanergization to reduce the potential of such cells to mediate GvHD. We show that the strategy of alloanergization effectively reduces alloresponses without adversely impacting CD19-specifc effector functions of CD19-CAR cells. This combined approach could be used to reduce relapse without increasing toxicity after allogeneic HSCT for patients with B-lineage leukemias and lymphomas.

A major concern with the infusion of allogeneic CAR⁺ T cells is their potential to mediate toxicity in the form of GvHD. CAR⁺ T cells can be activated by pathogen-specific antigens via their endogenous αβ TCR indicating that these receptors and their associated intracellular pathways remain functionally intact.(7, 30, 38) We have previously shown that CD19-CAR cells generated using the SB technology retain broad endogenous αβ TCR Vβ distribution,(18) in apparent contrast to some other strategies used to enrich antigen-specific T cells utilizing repetitive antigenic stimulation.(39) It has also been shown that murine and human folate-binding protein-specific CAR⁺ T cells can be activated via endogenous αβ TCR by stimulation with alloantigens, supporting the potential of such cells to mediate alloresponses.(40) In our current study we detected CD4⁺ alloprecursor frequencies within human CD19-CAR cells at comparable levels to those detected by CFSE dye dilution in unmanipulated human CD4⁺ T cells by Martins et al. (1.1%), who in common with our strategy, used single donor allostimulator PBMC.(34) This suggests that alloproliferative CD4⁺ T cells occur at a similar frequency within human CD19-CAR cells and unmanipulated T cells.

Alloanergization of CD19-CAR cells effectively and specifically reduced proliferative and cytokine alloresponses in these cells. When the confounding effect of CD19-driven proliferation was removed (by using CD19-depleted allostimulators in both the alloanergization step and in assays to detect residual alloreactivity), the efficiency of alloanergization of CD19-CAR cells was over 90%. This was consistent with both our previous published data and other effective strategies used to reduce alloreactivity of HLA-mismatched PBMC.(23, 26) Because our strategy of alloanergization only directly affects CD28⁺ T cells, the strategy could theoretically be less effective at reducing alloresponses in CD8⁺CD19-CAR cells than in CD4⁺CD19-CAR cells, as human CD8⁺ T cells typically contain a lower proportion of CD28⁺ cells compared to human CD4⁺ T cells. However, alloanergization reduced alloproliferative responses effectively in both CD4⁺ and CD8⁺CD19-CAR cells. This may reflect an indirect effect on CD8⁺ T cells consequent to alloanergization of CD4⁺ T cells, consistent with reports that alloproliferative CD4⁺ T cells are required for alloreactive CD8⁺ T cells to proliferate.(34, 41) The concern remains that CD4⁺CD28^neg T_EM-mediated alloresponses may persist after alloanergization. However clinically significant GvHD mediated by such cells is likely to be limited. T_EM cells are less potent mediators of proliferative alloresponses in vitro than T_CM or naïve human T cells,(42) and human CD28^neg T cells typically have shortened telomeres compared to CD28⁺ T cells (43) predicting a restricted lifespan in vivo after infusion.(44)

Alloanergization did not reduce the proportion of CD8⁺CD19-CAR cells expressing surface CAR, and CD19-specific redirected target cell cytolysis was preserved, demonstrating retained functionality and specificity. However, alloanergization resulted in modest reductions in the proportion of CD4⁺CD19-CAR cells expressing surface CAR, frequencies of CD19-specific IFN-γ⁺ cells and capacity for CD19-CAR cells to expand in vitro after stimulation on CD19⁺ aAPC. These findings suggest that the in vivo efficacy of alloanergized CD19-CAR cells could be reduced in comparison to non-anergized cells. However, more than half of CD4⁺CD19-CAR expressing cell surface CAR and 50–70% of the proportion of CD19-CAR cells secreting IFN-γ⁺ after CD19 stimulation were retained after alloanergization, and CD19-specific proliferation was still demonstrable consistent with retention of significant capacity for CD19-specific expansion and target cell lysis. Furthermore, it is likely that homeostatic expansion resulting from the lymphopenic environment created after allogeneic HSCT would augment expansion and persistence of infused alloanergized donor CD19-CAR cells. This is supported by the data from in vitro expansion of alloanergized CD19-CAR cells on CD19-aAPC loaded with OKT3 which restored the proportion of CD4⁺CD19-CAR cells expressing surface CAR to levels seen in non-alloanergized cells without loss of allospecific hyporesponsiveness, suggesting that CD19-specific function and alloanergy might be maintained after expansion. Further studies using an immunodeficient tumor-bearing mouse model could be used to compare the in vivo persistence and anti-tumor efficacy of non-anergized and alloanergized human CD19-CAR cells.

Allogeneic CD19-CAR cells could be used to augment anti-tumor effects in a variety of allogeneic HSCT settings. These data support the application of a clinical strategy in which CD19-CAR cells are generated from allogeneic donors and subsequently anergized to recipient alloantigens prior to infusion after allogeneic HSCT. The broad endogenous TCR Vβ subfamily distribution retained by CD19-CAR cells generated using the SB system(18) suggests that these cells could also contribute to pathogen-specific immunity after allogeneic HSCT via their endogenous αβ TCR. Therefore allogeneic HSCT approaches utilizing T cell-depleted hematopoietic stem cell sources or umbilical cord blood cells, both of which are associated with delayed immune reconstitution and increased infectious complications,(45–47) would be particularly suitable platforms for the use of allogeneic donor CD19-CAR cells. Although we are developing approaches to limit off-target effects, one consequence of allogeneic CD19-CAR T cell therapy might be destruction of healthy donor-derived CD19⁺ B cells. Intravenous immunoglobulin could be used to correct clinically-significant hypogammaglobulinemia in such an event. Another potential limitation to our approach is that repeated in vivo stimulation may contribute to replicative senescence of CD19-CAR cells via telomere erosion (48) and preclude their long-term persistence. In this case, repeat infusions of alloanergized CD19-CAR cells could provide a persistent anti-tumor effect.

In summary, we describe the successful application of alloanergization to selectively reduce alloreactivity in human CD19-specific T cells without significant impairment of CAR-dependent effector functions. Non-viral gene transfer of CAR, propagation on aAPC, and induction of alloanergy all employ methods currently individually in use in phase 1/2 clinical trials. These approaches could therefore be readily applied in combination at a clinical scale to generate donor-derived T cells engineered to contain enhanced anti-tumor activity, but reduced alloreactivity, suitable for use after allogeneic HSCT to reduce disease relapse while limiting toxicity from GvHD.