Untouched GMP-Ready Purified Engineered Immune Cells to Treat Cancer

Rising clinical data demonstrate the potency of the adoptive transfer of T cells to effectively treat cancer, when T cells are genetically engineered to express tumor-specific receptors (1–4). Although the engineering process of genetically modified T cells has been substantially improved during the last decades, to date all cellular products usually maintain a fraction of nontransduced and poorly transduced immune cells (2, 5). However, nonengineered immune cells have been reported to hamper the activity of adoptively transferred cells in an autologous situation (6, 7) due to the unwanted transfer of regulatory T cells (8) or by competing for homeostatic cytokines (9, 10). In addition, non-engineered or poorly engineered immune cells with high endogenous αβTCR expression have a clear potential to induce graft versus host disease (GVHD) in the context of allogeneic/third-party applications (11, 12).

To date, efforts to enhance purity of engineered immune cells mainly focus on positive selection either by expression of an additional transgene such as truncated CD34 (13), truncated nerve growth factor receptor (14), or direct binding of the introduced therapeutic receptor (15, 16). However, these strategies can either result in reduced expression of the desired gene of interest or additional introduction of potentially immunogenic components. In addition, the positive selection process results in so-called “touched cells” and can mediate activation-induced cell death (17, 18) directly after isolation or antibody dependent cellular cytotoxicity shortly after transfer into the recipient and may hamper long-term persistence of transferred immune cells. An obvious alternative strategy would be the isolation of “untouched engineered immune cells.”

Most recently, GMP-grade anti-αβTCR beads became available and are currently used in the context of haploidentical transplantation used by others (19) and us (NTR2463 and NTR3079). Using such selection beads in combination with the knockdown of endogenous αβTCR genes (20) should theoretically result in a population of untouched engineered immune cells with high purity and substantially reduced “off-target” effects. Despite the fact that various different strategies to knockdown endogenous αβTCR expression have been described most recently, including RNA interference (21, 22), zinc finger nucleases (ZFNs; ref. 20), or TALENS (23), clinical application of these elegant techniques is unfortunately far from translation. Off-target integration effects and the need for multiple plasmids to be expressed in one T cell are some of the reasons as to why clinical translation has noticeable obstacles. Alternative simple and cost-effective solutions are clearly still needed. Therefore, classical methods such as the interference with endogenous αβTCR chains by introducing physiological strong αβTCR competitors for the components of the CD3 complex (24–26) such as γδTCR chains (27–29), remain valid alternatives in order to substantially reduce endogenous αβTCR expression. Besides the reduction of endogenous αβTCR expression, γδTCRs are promising immune receptors to retarget αβT cells against cancer as reviewed extensively in refs. 30 and 31. γδT cells are considered as an innate-like population of immune cells recognizing molecular stress signals on infected or malignant cells. A subset of γδT cells express a TCR composed of Vγ9 and Vδ2 chains that sense accumulated nonpeptidic pyrophosphate molecules (phosphoantigens), intermediates of a deregulated mevalonate pathway of isoprenoid synthesis, via BTN3A1 (CD277) and reported by us (27–29) and others (32) to display potent cytotoxicity against a broad range of tumor species. Other intriguing properties of γδTCRs are the absence of forming potentially self-reactive mixed TCR dimers, their HLA-independency, absence of immunogenicity, which can either result in reduced therapeutic efficacy or substantial toxicity and avoidance of education and long-term tolerance when expressing innate immune receptors out of the context of an innate immune cell (30, 31).

In summary, although major progress in the quality and efficacy of clinical engineered T cells has been achieved in the last decades (1–4), a feasible and cost-effective GMP-compliant system to obtain highly purified engineered immune cells is not yet available. By utilizing a novel tumor-specific immune receptor naturally interfering with endogenous αβTCR chains (27–30), an optimized expression system, and clinical grade anti-αβTCR beads, we developed a highly pure engineered cellular product with improved antitumor activity. The production process can be readily translated to meet GMP-grade standards and the resulting cellular medicine can be applied as part of an immune intervention strategy in a broad population of cancer patients in autologous, allogeneic, and third-party situations.

Daudi (CCL-213) and Phoenix-Ampho (CRL-3213) cells were obtained from the American Type Culture Collection at the initiation of the study in 2010 [authentication by short tandem repeat (STR) profiling/karyotyping/isoenzyme analysis]. OPM2, RPMI8226/S, OPM2-Luciferase (OPM2-Luc), and RPMI8226/S-luc (RPMI-Luc) were kindly provided by Anton Martens in 2012 (University Medical Center Utrecht, Utrecht, the Netherlands) and Daudi-Luciferase (Daudi-Luc) by Genmab in 2013 (Utrecht, the Netherlands). The EBV-transformed lymphoblastoid cell lines (EBV-LCL) were a kind gift from Tuna Mutis in 2010 (University Medical Center Utrecht) and not authenticated since receipt. All cells were passaged for a maximum of 2 months, after which new seed stocks were thawed for experimental use. All cell lines were routinely verified by growth rate, morphology, and/or flow cytometry and tested negative for mycoplasma using MycoAlert Mycoplasma Kit. Phoenix-ampho cells were cultured in DMEM+1%Pen/Strep (Invitrogen) +10% FCS (Bodinco), all other cell lines in RPMI+1%Pen/Strep+10%FCS. Peripheral blood mononuclear cell (PBMCs) were isolated from buffy coats obtained from the Sanquin Blood Bank (Amsterdam, the Netherlands) or from the Institute for Transfusion Medicine and Immunohematology, Frankfurt, Germany. PBMC samples from acute myeloid leukemia (AML) patients were a kind gift from Matthias Theobald (Mainz, Germany) and from the University Medical Center Utrecht Biobank and were collected according to GCP and Helsinki regulations.

The highly tumor reactive γ9δ2TCR chain genes were obtained via combinatorial-γδTCR chain exchange (28). The TCRγ chain was derived from γδT cell clone G115 (33) and TCRδ chain from clone 5 (28), codon optimized (Geneart Life Technologies) and cloned into the retroviral vector pBullet as single TCR chain vectors containing either γ-chain-IRES-neomycine or δ-chain-IRES-puromycine. In addition, four different transgene cassettes containing both γ and δTCR chains were designed by exchanging two different 2A peptide linker sequences, F2A and T2A (34), and the order of TCR chains (δ-F2A-γ, γ-F2A-δ, δ-T2A-γ, γ-T2A-δ; Fig. 1A). These γδTCR cassettes were cloned into the optimized retroviral vector pMP71 (kindly provided by Miriam Heemskerk, Leiden University Medical Center, Leiden, the Netherlands) to express both TCR chains simultaneously. A nonsense murine αβTCR, consisting of the α chain derived from the MDM2/HLA-A2 TCR (35) and the β chain from the p53/HLA-A2 TCR (36), was used as control TCR in both the pBullet and the pMP71 retroviral vector system. Also, truncated nerve growth factor receptor in pMP71 was used as control in retroviral transduction experiments (pMP71:ΔNGFR; kindly provided by Miriam Heemskerk).

γδTCRs were transduced into αβT cells as previously described (27). In brief, packaging cells (phoenix-ampho) were transfected using FugeneHD reagent (Promega) with helper constructs gag-pol (pHIT60), env (pCOLT-GALV) (35) and two retroviral constructs (pBullet) containing either γ-chain-IRES-neomycine or δ-chain-IRES-puromycine, or one retroviral construct (pMP71) containing both γδTCR chains separated by a 2A sequence. Human PBMC were preactivated with αCD3 (30 ng/mL; Orthoclone OKT3; Janssen-Cilag) and IL2 (50 IU/mL; Proleukin, Novartis) and transduced twice with viral supernatant within 48 hours in the presence of 50 IU/mL IL2 and 6 μg/mL polybrene (Sigma-Aldrich). Transduced T cells were expanded by stimulation with anti-CD3/CD28 Dynabeads (0.5 × 10⁶ beads/10⁶ cells; Life Technologies) and IL2 (50 IU/mL) and in case of pBullet retroviral system selected with 800 μg/mL Geneticin (Gibco) and 5 μg/mL puromycin (Sigma-Aldrich). Next, TCR-transduced T cells were expanded based on a previously described rapid expansion protocol (REP; ref. 27).

αβT cells were transduced with pMP71: γ-T2A-δ and incubated with a biotin-labeled anti-αβTCR antibody (clone BW242/412; Miltenyi Biotec) followed by incubation with an anti-biotin antibody coupled to magnetic beads (anti-biotin MicroBeads; Miltenyi Biotec). Next, the cell suspension was applied onto an LD column and αβTCR-positive (αβTCR⁺) T cells were depleted by MACS cell separation according to the manufacture's protocol (Miltenyi Biotec). After depletion, γδTCR⁺ T cells were expanded using T-cell REP.

Antibodies used for flow cytometry included: pan-γδTCR-PE (clone IMMU510), pan-αβTCR-PE-Cy5 (clone IP26A; both Beckman Coulter), mouse TCRβ-chain-PE (clone H57-597), CD4-FITC (clone RPA-T4; both BD Biosciences), CD8-PerCP-Cy5.5 (clone RPA-T8; Biolegend), CD20 (Rituximab; Roche), and Goat-anti-Human-IgG-PE (Jackson ImmunoResearch Laboratories). All samples were analyzed on a FACSCanto II using FACSdiva software (BD Biosciences).

⁵¹Chromium-release assay for cell-mediated cytotoxicity was previously described (28). In brief, target cells were labeled overnight with 100 μCu ⁵¹Cr and incubated for 4 to 5 h with transduced T cells in five effector-to-target ratios (E:T) between 30:1 and 0.3:1. Percentage of specific lysis was calculated as follows: (experimental cpm − basal cpm)/(maximal cpm − basal cpm) × 100 with maximal lysis determined in the presence of 5% triton and basal lysis in the absence of effector cells.

IFNγ ELISPOT was performed using anti-hu IFNγ mAb1-D1K (I) and mAb7-B6-1 (II) (Mabtech) following the manufacturer's recommended procedure. Target and effector cells (E:T 1:3) were incubated for 24 hours in the presence of pamidronate (10 μmol/L; Calbiochem) where indicated.

IFNγ ELISA was performed using ELISA-ready-go! Kit (eBioscience) following manufacturer's instructions. Effector and target cells (E:T 1:1) were incubated for 24 hours in the presence of pamidronate when indicated.

TCR surface expression assay was performed with T cells that were rested for 21 days and starved of fresh medium and IL2 supplement for 6 days. T cells were either stimulated in a 24 wells plate with anti-CD3/CD28 Dynabeads at a bead-to-cell ratio of 1:1 in a final volume of 2 mL medium or with EBV-LCLs in a 96 wells plate at an E:T ratio of 1:3 in a final volume of 200 μL medium supplemented with 30 IU/mL IL2. After a stimulation period of 4, 24, and 48 hours at 37°C and 5% CO₂, cells were stained with mAbs against the introduced and/or endogenous TCR and analyzed by flow cytometry. TCR cell surface expression was calculated as follows: [mean fluorescence intensity (MFI) of T cells stimulated with beads/MFI of unstimulated T cells] × 100 and indicated as % TCR of control.

The RAG2^−/−/γc^−/−-BALB/C mice, originally obtained from AMCAS b.v., were bred and housed in the specific pathogen-free (SPF) breeding unit of the Central Animal Facility of Utrecht University. Experiments were conducted according to Institutional Guidelines after acquiring permission from the local Ethical Committee and in accordance with current Dutch laws on Animal Experimentation. For the experiments, female mice from 8 to 12 weeks of age were used. 10⁷ γδTCR or Mock TCR–transduced T cells were intravenously injected together with 0.5 × 10⁶ Daudi-Luc or 5 × 10⁶ OPM2-Luc cells. Mice received 0.6 × 10⁶ IU of IL2 in IFA subcutaneously on day 1 and every 21 days till the end of the experiment. Pamidronate (10 mg/kg body weight) was applied in the indicated groups at day 1 intravenously and every 21 days until the end of the experiment. For the OPM2 tumor rechallenge experiment, mice treated with γδTCR T cells that remained tumor free for more than 120 days were rechallenged with 5 × 10⁶ OPM2-Luc cells without prior irradiation. Naïve and non-irradiated mice were used as control for tumor outgrowth. Tumors were visualized in vivo by bioluminescent imaging. Mice were anesthetized by isoflurane before they received an intraperitoneal injection (100 μL) of 25 mg/mL Beetle Luciferin (Promega). Bioluminescence images were acquired by using a third-generation cooled GaAs intensified charge-coupled device camera, controlled by the Photo Vision software and analyzed with M³Vision software (all from Photon Imager; Biospace Laboratory).

Differences were analyzed using indicated statistical tests in GraphPad Prism (GraphPad Software Inc.).

The first goal of the study was to obtain the strongest antitumor reactivity and maximal interference with endogenous αβTCRs (24), by no other means than selectively introducing optimally a strong competitor for components of the CD3 complex such as a novel and highly tumor-reactive γ9δ2TCR (γG11/δ5), obtained by combinatorial-γδTCR-chain exchange (28). General consensus is lacking, regarding the optimal TCR transgene cassette for adoptive cell therapy with TCR-engineered T cells (37–41). Therefore, we designed four different γδTCR transgene cassettes (Fig. 1A) to test both the influence of γδTCR-chain orientation as described for αβTCR transgene cassettes (37) as well as a particular 2A ribosomal skipping sequence, F2A versus T2A (34). All genes were codon optimized, cloned into the clinical-grade retroviral vector pMP71 (42), and introduced into αβT cells from multiple donors in order to test general applicability. Placing the γ chain 5′ and the δ chain 3′ of the F2A sequence (γ-F2A-δ) increased transduction efficiency more then 2-fold, depicted as fold increase relative to transduction efficiency of the δ-F2A-γ cassette and corresponded with significantly higher TCR expression (Fig. 1B, left). However, the T2A sequence significantly enhanced transduction efficiency and MFI for both γδTCR chain orientations (Fig. 1B, right). Flow cytometric plots from a representative experiment with the four γδTCR transgene cassettes are shown in Supplementary Fig. S1. To determine whether the increased γδTCR expression was associated with increased specific cytolytic capacity, transductants were cocultured with Daudi (Burkitt's lymphoma) or OPM2 (multiple myeloma) cells (Fig. 1C). T cells transduced with δ-F2A-γ did not induce significant lysis of tumor cells and an opposed orientation of γ- and δ-chain could only minimally increase lysis at the highest effector-to-target (E:T) ratio. In contrast, T cells transduced either with δ-T2A-γ or γ-T2A-δ efficiently lysed target cells. We used T cells transduced with both T2A containing transgene cassettes to examine whether the γδTCR chain-orientation influenced tumor-specific release of effector cytokines (Fig. 1D). Although lytic activity was not significantly enhanced, γ-T2A-δ transductants produced significant higher amounts of IFNγ compared with T cells transduced with δ-T2A-γ and argued for the use of the γ-T2A-δ as our clinical candidate transgene cassette. Given these results, it is worthwhile to individually optimize transgene cassettes for each particular clinical candidate TCR in order to achieve the most efficient competition with endogenous αβTCR chains as well as the strongest expression and activity of introduced receptors.

Although retroviral transduction efficiencies have been greatly increased over the past decades and clinical αβTCR-engineered T-cell products contain up to 60% to 85% TCR⁺ T cells, the interpatient variation remains significant (2, 43). To overcome this problem and obtain a pure γδTCR-engineered T-cell product for clinical application, we introduce a new procedure for the depletion of nontransduced or poorly transduced T cells, taking advantage of the observation that upon introduction of a γδTCR, the endogenous αβTCR expression is decreased (refs. 27 and 29; and Supplementary Fig. S1 and Fig. 2A). This procedure is based on a monoclonal antibody against the αβTCR complex (clone BW242/412; Miltenyi Biotec), currently used to deplete αβT cells from stem cell grafts resulting in a mean 4 logs αβT-cell depletion (NTR2463, NTR3079, and ref. 19). In short, αβT cells were transduced with pMP71:γ-T2A-δ and incubated with the biotin-labeled anti-αβTCR antibody followed by incubation with an anti-biotin antibody coupled to magnetic beads for the depletion of αβTCR⁺ T cells by MACS cell separation. After depletion, γδTCR T cells were expanded utilizing a previously described T-cell expansion protocol (44). This procedure resulted in near complete depletion of single αβTCR⁺ T cells (from 51% to 0.4%) and a dramatic increase in γδTCR single-positive T cells (from 20% to 77%; Fig. 2A). Importantly, the αβTCR/γδTCR double-positive T cells that remained were characterized by relative low surface expression of the endogenous αβTCR. This phenotype was stable until day 5 after stimulation, when T cells were highly activated and proliferative. However, the phenotype changed at day 9 towards a major population of αβTCR/γδTCR double-positive T-cells (68%) and a decreased percentage of γδTCR single-positive cells (25%) when T cells reside in a more resting phase (Fig. 2A).

The reappearance of endogenous αβTCR chains in more resting cells allowed us to test whether early αβTCR T-cell depletion still provides a functional advantage when interference with endogenous TCR chains is far from complete. The functionality of the purified engineered immune cell product was therefore tested 10 days after selection and expansion and compared with a bulk engineered cell product without αβTCR depletion. Indeed, αβTCR T-cell depletion significantly increased specific lysis of Daudi cells (P < 0.01; Fig. 2B) as well as IFNγ production in response to three different tumor cell lines (P < 0.001; Fig. 2C). These results demonstrate that αβTCR T-cell depletion increased the antitumor potential of the engineered cell product.

In order to test whether highly purified engineered immune cells in a resting state with suboptimal expression of introduced γδTCR chains still provide a therapeutic advantage compared with primary γδT cells with optimal γδTCR expression but polyclonal γ9δ2TCR usage, antitumor activity of primary bulk γ9δ2T cells and GMP-grade T cells engineered to express an optimized γ9δ2TCR were tested at day 10 against a panel of primary leukemic cells from AML patients. Treatment of leukemic cells with pamidronate to block the mevalonate pathway downstream to induce accumulation of isopentenyl pyrophosphate resulted in IFNγ secretion by T cells in response to 9 of 16 AML samples. In five of eight tested samples, the optimized γδTCR-engineered T cells produced significantly enhanced levels of IFNγ compared with primary polyclonal γ9δ2T cells isolated from a healthy donor (Fig. 2D). This result suggests that transfer of optimized engineered γδTCR T cells, as part of a clinical immune intervention strategy, is preferred over a polyclonal γδT cell product. Together, we propose here a depletion procedure using a GMP-grade antibody for the enrichment of untouched γδTCR engineered T cells resulting in a highly tumor-reactive clinical cell product even in the absence of a complete interference with the expression of endogenous αβTCR chains.

The enrichment procedure that resulted in a pure γδTCR-transduced T-cell population with low or absent expression of αβTCR may reduce the allo-reactive potential of the T-cell product in an allogeneic setting. However, this advantage could theoretically be counterbalanced by the latter upregulation of endogenous αβTCRs in resting engineered immune cells (Fig. 2A). To further simulate a resting T-cell following in vivo transfer with substantial reoccurring expression of endogenous αβTCR chains, we pushed the system to a greater extend by using engineered T cells that lacked stimulus for more than 20 days and were starved of IL2 for 6 days. Mock (ΔNGFR transduced), γδTCR-engineered, and γδTCR-engineered αβTCR-depleted T cells were tested against a panel of 13 mismatched EBV-LCL cell lines or healthy donor–derived PBMCs in an IFNγ ELISPOT assay (Fig. 3A and B; see Fig. 3C, top, for γδTCR vs, endogenous αβTCR expression in rested T cells). Although Mock T cells produced IFNγ in response to 9 out of 13 EBV-LCL cell lines, allo-reactivity of γδTCR-engineered bulk T cells was greatly reduced (significant reduction for eight of nine EBV-LCL lines) and more importantly even completely abolished in the γδTCR-engineered αβTCR-depleted T-cell population. The reduced allo-reactivity of γδTCR-engineered T cells was even more apparent when the different T-cell populations were tested against a panel of 20 different healthy donor–derived PBMCs (Fig. 3B). No allo-reactivity was detected in the γδTCR-transduced T-cell populations, but Mock T cells produced IFNγ in response to 9 out of 10 PBMC donor combinations. Importantly, although allo-reactivity was abolished, both γδTCR-engineered populations maintained their antitumor reactivity (Fig. 3A and B).

One possible explanation for the reduced allo-reactivity in γδTCR-transduced T cells despite their significantly recovered endogenous αβTCR may be the preferential hit of memory T cells and not allo-reactive naïve T cells in the transduction procedure. However, this explanation could be excluded because CD45RA-depleted (memory) and CD45RO depleted (naïve) T cells were transduced with identical transduction efficiencies (data not shown). Meanwhile, reduced surface expression of the endogenous αβTCR in γδTCR-transduced T cells may partially account for reduced allo-reactivity. In unstimulated resting cells, the total amount of αβTCR, as indicated by the MFI, in γδTCR engineered cells was reduced for bulk cells by 4% (MFI: 12546) and for αβTCR-depleted cells by 14% (MFI: 11334) when compared with ΔNGFR-transduced cells (MFI: 13117; Fig. 3C, top). Finally, a preferential recovery of introduced as compared with endogenous TCR chains upon antigen encounter may also contribute to this effect. Indeed, upon stimulation of resting T cells with anti-CD3/CD28 beads, the expression of the retrovirally introduced γδTCR decreased less profoundly, was restored within 24 hours and reached higher levels than before T-cell stimulation within 48 hours (Fig. 3C, bottom, and Supplementary Fig. S2A–S2D). This quick restoration and high level of TCR surface expression following T-cell reactivation was not restricted to γδTCRs, thus most likely a property of the expression system used here, because expression kinetics of a retrovirally introduced αβTCR were comparable (Supplementary Fig. S2). These results suggest that although the endogenous αβTCR is reexpressed in resting engineered T cells, reactivation preferentially decreases endogenous αβTCR expression (Fig. 3C, bottom). To formally address this hypothesis in a more physiologically relevant setting, we cocultured resting γδTCR-engineered αβTCR-depleted T cells with an HLA-mismatched EBV-LCL that induced an allo-reactive IFNγ response by ΔNGFR-transduced cells (LCL1 from Fig. 3A). Co-culturing these cells resulted in a downregulated endogenous αβTCR expression, which reached a minimal level after 48 hours, whereas the γδTCR expression remained largely stable over time (Supplementary Fig. S2E). Thus, we confirm that in γδTCR-engineered T cells T-cell reactivity, such as allo-reactivity, is prevented and the expression of retrovirally introduced tumor-specific TCRs is favored and antitumor reactivity maintained. These data question the need of a complete elimination of endogenous αβTCR chains in purified engineered immune cells when expressing strong αβTCR competitors with an optimal expression system.

We evaluated the clinical potency of the optimized γδTCR T-cell product in relation to γδTCR-engineered T cells produced with the extensively used pBullet retroviral transduction and antibiotic selection system (45), referred to as pB:γδTCR T cells. Following transduction of peripheral blood αβT cells and selection with antibiotics (pBullet) or enrichment with αβTCR depletion beads (pMP71) and subsequent T-cell expansion, both immune products were evaluated. Interestingly, not only the percentage γδTCR⁺ T cells was higher for the αβTCR-depleted γδTCR T-cell product, but also the number of γδTCR complexes per cell increased more than 2-fold compared with pB:γδTCR T cells as measured by MFI (Fig. 4A). In addition, lysis of three tested tumor cell lines was significantly enhanced by αβTCR-depleted γδTCR T cells transduced with the pMP71:γ-T2A-δ vector cassette as compared with pB:γδ TCR T cells (Fig. 4B). To test if the superior antitumor activity of αβTCR-depleted γδTCR T cells is reflected in vivo, we used a humanized mouse tumor model for adoptive transfer of γδTCR-engineered T cells. Irradiated RAG2^−/−γ_c^−/− double-knockout mice were injected with Luciferase-positive Daudi tumor cells and either with γδTCR or Mock TCR-engineered T cells and tumor growth was evaluated by bioluminescence imaging. Both γδTCR-engineered T-cell products significantly inhibited tumor growth compared with Mock TCR T cells, but αβTCR-depleted γδTCR T cells further delayed tumor outgrowth and significantly increased survival compared with pB:γδTCR T cells (Fig. 4C).

In a second tumor model of multiple myeloma, the antitumor activity of the optimized αβTCR-depleted γδTCR-engineered cell product was assessed. Irradiated Rag2^−/−γ_c^−/− double-knockout mice were injected with Luciferase-positive OPM2 cells and γδTCR or Mock TCR-engineered T cells and tumor growth was evaluated by bioluminescence imaging. Interestingly, tumor growth was completely prevented by clinical-grade γδTCR T cells in four of seven mice (Supplementary Fig. S3). One hundred and twenty days after first tumor and T-cell injections, tumor free mice were rechallenged with a second injection of tumor cells without prior irradiation and nonirradiated naïve mice were used as control for tumor outgrowth. Selectively rechallenged mice remained tumor free, indicating that adoptive transfer of engineered immune cells can mediate long-term tumor protection in vivo (Fig. 4D). In summary, these data underscore the potency of a cost-effective purified engineered immune cell product for clinical application taking advantage of an optimal expression system, strong competitors for endogenous TCR chains such as a γδTCR, and GMP-grade anti-αβTCR beads.

Adoptive cell therapy using gene-engineered lymphocytes has moved towards a feasible and effective treatment modality for cancer as demonstrated by exciting clinical results during the last decade (1–5). However, large-scale clinical implementation of this promising immunotherapy still awaits some critical hurdles to be taken such as the generation of a purified engineered immune product. We provide evidence that the combination of an optimized and individualized expression system with strong αβTCR competitors and clinical anti-αβTCR beads is a highly efficient system, even if interference with endogenous αβTCR expression is far from complete. Thus, we present to our best knowledge the first enrichment procedure using a clinical-grade anti-αβTCR mAb for the production of untouched gene-engineered immune cells suitable for autologous, allogeneic, or third-party immune intervention platforms, which can be valuable for expression-receptor formats interfering with endogenous αβTCR chains.

Optimal gene expression is crucial when interference with αβTCR chains is not mediated via siRNA (21), ZFN (20), or TALENS (23). Therefore, an individualized testing of existing expression systems was essential when utilizing a natural occurring strong competitor for the CD3 complex (27). Although virus-derived 2A “ribosomal skipping sequences” are widely used to obtain equimolar expression of introduced genes, there are a handful of different 2A sequences available. A T2A sequence was superior to the F2A and we can unfortunately not provide a reasonable explanation for this observation. Our data revealed a preferred position of the TCRδ chain downstream of the 2A cleavage element and this is in line with an optimal αβTCR gene cassettes described by Leisegang and colleagues (37).

Our selection procedure utilizes a GMP-grade clinical bead, which is currently employed in the stem cell transplantation field by others (19) and us (NTR2463 and NTR3079) in order to deplete αβT cells from the donor. This procedure resulted in removal of nontransduced and poorly transduced bystander cells, improved γδTCR surface expression levels, and resulted in almost 100% γδTCR-engineered cells in the end product, which was translated into an increased antitumor function both in vitro and in vivo. Interestingly, when compared with a nonengineered polyclonal primary γδT-cell population, selected γδTCR-engineered T cells not only recognize an increased number of primary AML patient samples but also clearly produce higher amounts of IFNγ, in case the tumor cells are recognized. This encouraging feature of our engineered immune cells is likely due to the choice for a high avidity γδTCR in contrast to a polyclonal usage of γδTCRs in a primary γδT-cell population in combination with high γδTCR surface expression as a result of the selection procedure.

The substantial upregulation of endogenous αβTCR chains in resting engineered immune cells allowed us to thoroughly investigate pitfalls, if interference with endogenous αβTCR chains is not complete. Even after 3 weeks, tumor control remained preserved in engineered immune cells in vitro and in vivo and allo-reactivity could not be observed. This is most likely due to (i) a still reduced surface expression of αβTCR when compared with nonengineered cells and (ii) faster and higher upregulation of introduced TCR surface expression after antigen encounter. This observation is in line with previous data showing that in contrast to the endogenous TCR, introduced TCR expression under the control of an optimized viral promoter is restored, and increased shortly after antigen-specific stimulation of either the endogenous or introduced TCR (46). The observed downmodulation of the endogenous αβTCR upon introducing a tumor-specific immune receptor paves the way towards the broader applicability of our strategy: the combination of siRNA (22), ZFNs (20), or TALENS (23) to knockdown endogenous TCRs with the introduction of other innate receptors followed by the depletion with anti-αβTCR beads. This would yield cellular end products with pure populations of engineered cells that do not express their endogenous receptor and significantly increase the efficacy and safety both in an autologous and allogeneic scenario. However, our data stress also that a complete long-term interference with endogenous αβTCR chains is most likely not necessary for safety and efficacy. Residual expression of endogenous TCR chains might be even needed for an optimal long-term homeostatic proliferation (47).

Surprisingly, we also observed a relatively low allo-reactivity in nonengineered bystander cells, questioning whether our read-out systems have been sensitive enough. We excluded a preferential transduction of naïve T cells and subsequent downregulation of endogenous αβTCRs in naïve T cells, which have been reported as more potent GVHD mediators than memory T cells (48, 49). However, our observation is also in line with the recent clinical observation that the transfer of nonselected CAR-engineered donor T cells with intact endogenous αβTCR chains in human did not associate with substantial GVHD (50) and might reflect a rather reduced αβTCR expression in such extensively cultured and rapidly expanded cells prior infusion. In line with this observation and in contrast to reports by others (11, 20), we did not observe substantial GVHD in different humanized mouse models after adoptive transfer of engineered immune cells despite the fact that some models have employed similar mouse strains (refs. 11 and 20; T. Straetemans and J. Kuball; unpublished data).

All together, we provide a novel, feasible, and cost-effective strategy that is ready to be translated into a GMP-grade procedure to generate untouched engineered immune cells by taking advantage of an individualized expression system in combination with a strong competitor for endogenous αβTCR chains and clinical-grade αβTCR beads. We also demonstrate that the interference with endogenous αβTCR chains does not necessarily need to be complete in terms of long-term competition in order to obtain a safer and more efficient product. We conclude that the developed depletion strategy is therefore applicable to nearly any engineered immune product interfering temporarily or permanently with endogenous αβTCR chains.

J. Kuball reports receiving commercial research grants from Miltenyi and is listed as an inventor on a patent, owned by University Medical Center Utrecht, on the use of antibodies for enrichment of engineered T cells with exogenous immune receptors and antibodies for use in depletion of engineered T cells. No potential conflicts of interest were disclosed by the other authors.

Conception and design: T. Straetemans, C. Gründer, J. Kuball

Development of methodology: T. Straetemans, C. Gründer, I. Slaper-Cortenbach, J. Kuball

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Straetemans, S. Heijhuurs, S. Hol, J. Kuball

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Straetemans, C. Gründer, S. Heijhuurs, S. Hol, J. Kuball

Writing, review, and/or revision of the manuscript: T. Straetemans, C. Gründer, I. Slaper-Cortenbach, H. Bönig, Z. Sebestyen, J. Kuball

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Hol, H. Bönig

Study supervision: T. Straetemans, J. Kuball

The authors thank Kirsten Scholten for assistance in preparation of experiments.

Funding for this study was provided by ZonMW 43400003 and VIDI-ZonMW 917.11.337, KWF UU 2013-6426 and UU 2014-6790, and Vrienden van het UMCU to J. Kuball. Research support was provided by Miltenyi Biotec.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.