NY-ESO-1 TCR single edited stem and central memory T cells to treat multiple myeloma without graft-versus-host disease

Although the infusion of ex vivo expanded tumor-specific T cells to melanoma patients has documented the efficacy of adoptive T-cell therapy,¹  the exploitation of this strategy to other forms of cancer remains limited by the low frequency of tumor-reactive high-avidity T cells. Indeed, tumor antigens are often overexpressed, unmodified self-antigens, subject to tolerance.²  Although neoantigens derived from oncogenic mutations can elicit T-cell responses,³  these are rarely observed in tumors with a low mutational load, such as most hematological malignancies.⁴  These limitations can be overcome by engineering high-avidity tumor-reactive T-cell receptors (TCRs), obtained from rare tumor-specific T cells, into mature lymphocytes. This is a flexible strategy, theoretically applicable in the autologous context and in the HLA-matched allogeneic setting, and that could be potentially exploited to generate third-party tumor-specificlymphocytes, in an “off-the-shelf” approach. In initial clinical trials of TCR gene transfer objective responses were observed at lower frequencies than with naturally occurring tumor-specific effectors.⁵  Indeed, TCR gene transfer suffers of some limitations intrinsic to the TCR biology that affect safety and efficacy of the cellular products. The tumor-specific TCR competes with the endogenous ones for binding to CD3, leading to mutual TCR dilution and reduced T-cell avidity and antitumor efficacy. Furthermore, because TCRs are heterodimers, the α and β chains of the endogenous TCR might mispair with the respective α and β chains of the transgenic TCR to produce new hybrid receptors, with unpredictable and potentially harmful specificities.⁶  These limitations represent major concerns both in the autologous and in the allogeneic settings, in which TCR misparing might increase the risk of graft-versus-host disease (GVHD) associated with the alloreactive T-cell repertoire.^7,8 

Several strategies have been developed to increase TCR affinity and foster correct TCR pairing.⁹  We demonstrated that combining zinc finger nuclease (ZFN)–driven disruption of the endogenous TCR α and β chain genes with lentiviral delivery of a tumor-specific TCR results in the successful and complete editing (CE) of T-cell specificity¹⁰  displaying high avidity in the absence of TCR mispairing. However, the CE approach involves 4 distinct steps of genetic manipulation and sorting, requires 40 days of culture, and leads to small numbers of tumor-redirected T cells, which need further expansion. The extensive manipulation might reduce the overall fitness of gene-modified cells, and the procedure is limited by high costs and regulatory hurdles. We designed a simplified TCR gene editing protocol, based on the disruption of only the TCR α chain followed by the simultaneous introduction of the tumor-specific TCR chain genes. Through such a single editing (SE) procedure, we can obtain high numbers of SE cells in 2 weeks. Torikai et al reported that the genetic disruption of 1 TCR chain gene, followed by the genetic transfer of a chimeric antigen receptor, allows the generation of tumor-specific T cells devoid of the original TCR repertoire,¹¹  and this approach has been recently successfully tested in 2 patients with relapsed refractory CD19^pos acute lymphoblastic leukemia as a bridge to allogeneic hematopoietic stem cell transplantation.¹²  In contrast to this strategy, the TCR SE approach could theoretically result in some level of mispairing between the tumor-specific TCR chains and the remaining endogenous one. Here we show that the associated risk of off-target reactivity is counterbalanced by the advantage obtained by the elimination of the TCR endogenous repertoire. We took advantage of an HLA-A2–restricted, NY-ESO-1_157-165–specific TCR to redirect cells toward the NY-ESO-1 cancer testis antigen,^13-16  expressed by a high proportion of tumors, including multiple myeloma (MM)^17-19  and showed that SE lymphocytes are superior to unedited T cells redirected by TCR gene transfer.

Pairs of previously described¹⁰  ZFNs targeting either the TCR α constant (TRAC) gene or the TRBC gene were transiently expressed on primary T cells by Ad5/F35 adenoviral vector (AdV)^10,20,21  or messenger RNA (mRNA) electroporation (see supplemental Data, available on the Blood Web site). The HLA-A2–restricted, NY-ESO-1_157-165–specific TCR genes were obtained through the Adoptive engineered T cell Trials to Achieve Cancer Killing (ATTACK) consortium²²  and cloned into self-inactivating lentiviral vectors (LVs) under the PGK promoter. For the CE procedure, NY-ESO-1–specific TCR α and β chains were cloned into separate vectors (NY-ESO-1 TCRα LV and NY-ESO-1 TCRβ LV), whereas for the SE protocol both chains were cloned into a single vector and linked with a PTV1.2A self-cleavage peptide sequence and a furin cleavage recognition site (NY-ESO-1 TCR LV). In selected experiments, an LV encoding both WT1-specific α and β TCR chains under a bidirectional promoter¹⁰  was used. LVs were packaged by an integrase-competent third-generation construct and pseudotyped by the vescicular stomatitis virus (VSV) envelope.²³ 

We cultured the T2, U266, and MM.1S cell lines in RPMI 1640 supplemented with penicillin, streptomycin, glutamine, and 10% fetal bovine serum (FBS). MR245, MSR3-LA2, and SK23 cell lines were cultured in Iscove modified Dulbecco medium supplemented with penicillin, streptomycin, and 10% FBS. We obtained human T lymphocytes from healthy donors, upon informed consent. Lymphocytes were activated by magnetic beads conjugated to antibodies to CD3 and CD28 (ClinExVivo baCD3/CD28) and cultured in Iscove modified Dulbecco medium supplemented with penicillin, streptomycin, 10% FBS, and 5 ng/mL of interleukin-7 (IL-7) and IL-15 as described.¹⁰  Medium was replaced every 3 to 4 days. The first gene transfer step was performed 48 hours after activation. For TCR SE, ZFN-AdV-transduced or mRNA-electroporated lymphocytes, sorted as CD3^neg cells, were subsequently transduced with the NY-ESO-1-TCR LV. TCR CE was performed as previously described¹⁰  (Figure 1A). Cells were expanded as described.²⁴  When indicated, we sorted Vβ13.1^pos, Vβ21.3^pos, CD3^neg, CD3^pos, and NY-ESO-1 dextramer^pos cells using selection columns and anti-fluorescein isothiocyanate (FITC), anti-CD3, or anti-phycoerythrin (PE) MACS MicroBeads, following the manufacturer’s instructions.

We used FITC-, PE-, PerCP-, APC-, PE-Cychrome 7–, APC Cychrome 7-, and Pacific Blue–conjugated antibodies directed to human CD3, CD4, CD8, CD45, CD62L, CD45RA, CD95, CD127, CD138, CD38, CD27, CD28, Vβ13.1, Vβ21.3, and HLA-A2. NY-ESO-1_157-165 dextramer PE was used following the manufacturer’s instructions. Murine leukocytes were identified by antibodies to murine CD45. To measure CD127 expression, cells were washed and cultured in the absence of cytokines for 18 hours before staining. Cells (5 × 10⁵) were incubated with antibodies for 15 minutes at 4°C and washed with phosphate-buffered saline containing 1% FBS. Samples were run through a fluorescence-activated cell sorter (FACS) Canto II flow cytometer, and data were analyzed by Flow Jo software.

In ⁵¹Cr release assay, effectors were incubated with radiolabeled target cells at different effector/target (E/T) ratios for 4 hours. Specific lysis was expressed according to the following formula: 100 × (average experimental counts per minute [cpm] − average spontaneous cpm)/(average maximum cpm − average spontaneous cpm).

Interferon-γ (IFN-γ) release Elispot was performed plating responders and stimulators for 18 hours in Multiscreen HTS 96-well filtration plates (2 × 10⁴ cells per well), at a stimulator-to-responder ratio of 1, as described.¹⁰ 

In coculture experiments, lymphocytes were cocultured with target cells at the E/T ratio of 1:5. At day 4, viable cells were counted and percentages of T cells, and targets were assessed by FACS. The elimination index was calculated as follows: 1 − (number of residual target cells in presence of NY-ESO-1 redirected T cells)/(number of residual target cells in presence of mock-transduced T cells).

The experimental protocol was approved by the Institutional Animal Care and Use Committee of San Raffaele Scientific Institute. Sublethally irradiated 10- to 12-week-old female NSG mice were first infused with U266 cells and after 30 to 40 days treated with engineered lymphocytes, cultured for a median of 22 days. Medium only and donor-matched mock-transduced lymphocytes were controls. Lymphocytes were phenotyped before infusion.

A luciferase-expressing U266 line (U266^luc) was generated by transducing U266 with a bidirectional LV coexpressing LNGFR and Firefly luciferase (luc2) followed by FACS. U266^luc behaved similarly to the parental line in mice.

Small animal bioluminescence imaging (BLI) was performed with IVIS SpectrumCT System. Each mouse received an intraperitoneal injection of 150 mg luciferin/kg 10 minutes before BLI. The highest BLI signal was detected 15 minutes after luciferin injection. Images were obtained using the following IVIS settings: exposure time = auto, binning = 8, f = 1, and a field of view equal to 19.5 cm (field D). Dark measures were performed before BLI acquisition and subtracted from the acquired images, and no emission filters were used during BLI acquisitions. BLI image analysis was performed by placing a region of interest over the lesions and by measuring the total flux (photons/seconds) within the region of interest. Images were acquired and analyzed with Living Image 4.4.

Formalin-fixed, paraffin-embedded tissues were cut at 4-µm-thick sections, stained with hematoxylin-eosin, and underwent blind histopathological examination by an expert hematopathologist. The degree of tissue infiltration by human lymphocytes was assessed with a semiquantitative modality on a 4 scale-degree (0 to 3) and further confirmed by immunohistochemistry. For immunohistochemical stains, anti-human CD3 rabbit monoclonal antibody (clone 2GV6) was furnished by prediluted dispenser within Ventana Medical Detection System. Antigen retrieval was performed by Ventana Roche solution for 20 minutes at pH 9.

We relied on descriptive statistics or compared the data sets by 2-tailed Student t tests, 1-way analysis of variance (ANOVA). Log rank was used to compare survival curves. The statistical software packages GraphPad Prism version 7.0 and SPSS version 16.0 were used.

To develop a clinically feasible protocol of TCR gene editing, T lymphocytes from peripheral blood mononuclear cells (PBMCs) harvested from healthy donors were activated as previously reported^25-27  and redirected against NY-ESO-1 through 3 separate manipulation protocols described in Figure 1A. Transduction efficiency with the NY-ESO-1 TCR-LV was 69.4% (±1.7; range 55-86) and was similar in TR and SE procedures (73.3% and 67.1%, respectively; Figure 1B), confirming that initial disruption of 1 endogenous TCR chain gene does not impair sensitivity to further LV transduction. For the SE protocol, we compared TRAC and TRBC disruption, which proved equally feasible (supplemental Figure 1). For further studies, we focused on TRAC SE. The efficiency of ZFN-mediated knockdown of the endogenous TRAC gene was 43.1% (±4.9; range 18-66) with AdVs and 57.2% (±4.3; range 33-81) with mRNA electroporation. Further experiments were performed using mRNA electroporation. Although the first 2 steps of genetic manipulation required for CE were highly efficient, the efficiencies of the third and fourth genetic steps, performed after T-cell restimulation, were significantly lower. By the end of the procedure the mean fold changes of NY-ESO-1 TR, SE, and CE lymphocytes calculated over initial counts were 11.6, 5.6, and 0.9, respectively (Figure 1C). The simplification of the editing strategy thus ensured a production of a high number of NY-ESO-1–redirected cells in 2 weeks.

Interestingly, our protocol favors the maintenance of an early differentiated phenotype, with a pronounced prevalence of stem memory (T_SCM-CD45RA^posCD62L^posCD95^pos) and central memory (T_CM-CD45RA^negCD62L^posCD95^pos) lymphocytes (Figure 2A-B). Of note, similar distributions of memory T-cell subsets were observed in mock-transduced, TR, SE, and CE cells, indicating that the stimulation and culture protocol dominates over the genetic manipulation in determining cell differentiation fate. T_SCM and T_CM were equally represented in redirected CD4 cells, whereas T_SCM was the most prevalent subset in CD8 lymphocytes. The early differentiation phenotype of gene-modified lymphocytes was further confirmed in all T-cell subgroups by the high proportion of cells expressing CD127, a biomarker of T-cell fitness and persistence²⁸  (Figure 2C). The proportion of double-positive CD27 and CD28 T cells did not differ in mock-transduced (74% ± 5), TR (82% ± 5), and SE (80% ± 10) cells (Figure 2D), whereas it proved significantly lower in CE lymphocytes (37% ± 3), indicating that the length and complexity of the CE procedure might affect the composition of the cellular product. Interestingly, 49% of CE cells were CD28^neg but expressed CD27, a molecule associated with long persistence in adoptive transfer protocols.^29-31  This was particularly evident for CD8 T cells (supplemental Figure 2).

After sorting for dextramer^pos cells (with purity ≥97%), and expansion through a previously described protocol,²⁴  we quantified NY-ESO-1 TCR surface expression upon staining with a TCR Vβ13.1 family–specific antibody and with a PE-conjugated HLA-A2-NY-ESO-1_157-165 dextramer. Notably, although no differences emerged from the expression of TCR Vβ13.1 among TR, SE, and CE cells (Figure 2E), dextramer binding was significantly higher in edited T cells than in the TR counterpart (Figure 2F; supplemental Figure 3). These differences also emerged when separately analyzing CD4 and CD8 subsets, although the dextramer relative fluorescence intensity was significantly higher in CD8 than in CD4 cells (P < .05), because of higher stability of the TCR-major histocompatibility complex class I (MHC-I) binding in the presence of CD8.³²  Interestingly, the average vector copy numbers were 10 and 4 for TR and SE cells, respectively (supplemental Figure 4), indicating that the lower level of TCR expression on TR cells is not because of a lower vector copy number but possibly to competition of the NY-ESO-1 TCR with the endogenous TCR repertoire and/or mispaired TCRs. The similar levels of NY-ESO-1 TCR expression in CD4 and CD8 SE and CE lymphocytes suggest that the exogenous β chain properly pairs with the exogenous α chain with similar efficiency in SE and CE cells, and with more efficiency than in TR cells, thus suggesting that some degree of mispairing between endogenous and exogenous TCR chains occurs in conventionally TR cells. Noticeably, with this set of experiments we cannot rule out residual mispairing within the endogenous β chains and the exogenous α chain in SE cells. We then investigated efficacy and specificity of redirected T cells.

We challenged TCR-redirected T cells and mock-transduced lymphocytes with T2 cells pulsed with decreasing concentrations of the NY-ESO-1_157-165 peptide or with the irrelevant WT1_126-134 HLA-A2 restricted peptide, the U266 (HLA-A2^pos, NY-ESO-1^pos), and the MM.1S (HLA-A2^neg, NY-ESO-1^neg) MM cell lines by IFN-γ Elispot, ⁵¹Cr release, and 4-day coculture.

A high frequency of TR, SE, and CE cells produced IFN-γ in the presence of NY-ESO-1_157-165 pulsed T2 cells and of U266, naturally processing the tumor antigen (Figure 3A-B), with a trend in favor of TCR edited cells. Expression of both restriction element and antigen were required for recognition from redirected T cells (supplemental Figure 5).

NY-ESO-1-TCR–expressing T cells efficiently and specifically killed HLA-A2^pos, NY-ESO-1^pos targets (Figure 3C-D), again with a trend in favor of edited cells as compared with TR cells.

Finally, all redirected T cells proliferated extensively when cocultured with U266 (Figure 3E) and proved able to efficiently kill nearly 100% of U266 cells with a calculated mean elimination index of 0.99 at an E/T ratio of 1:10 (Figure 3F-G). Overall, the coculture assay reveals the ability of T lymphocytes, in the absence of cytokines, to eliminate cancer cells in a 4-day time span, also at low E/T ratios (supplemental Figure 6), mimicking the unfavorable clinical setting. The same results were obtained upon coculture of redirected lymphocytes with NY-ESO-1_157-165 peptide-pulsed T2 cells. Cytokine release and killing were extremely specific as shown by the negligible recognition of WT1_126-134 pulsed T2 cells and of the NY-ESO-1^neg MM.1S cells.

To weigh the alloreactive potential of engineered lymphocytes, TCR-expressing cells and controls were stimulated with allogeneic irradiated PBMCs from 8 (4 HLA-A2^neg and 4 HLA-A2^pos) healthy donors in separate mixed lymphocyte cultures. After 2 cycles of stimulation, effectors were tested against PHA activated T cells from the same donors, autologous PHA lines, and NY-ESO-1_157-165 pulsed HLA-A2^pos (10 µM) irradiated cells in an IFN-γ Elispot (Figure 4A) and in a ⁵¹Cr release (Figure 4B) assays. No responses were observed against the autologous cells (not shown). SE and CE lymphocytes proved significantly less alloreactive than untransduced and TR counterparts in both functional assays, while they persisted in recognizing NY-ESO-1^pos cells (Figure 4C-D). We further verified the alloreactivity of the SE protocol by performing TCR gene transfer, α SE and β SE with a TCR specific for the HLA-A0201 restricted WT1_126-134 peptide.¹⁰  As shown in supplemental Figure 7, TR and SE cells proved equally able to recognize WT1^pos targets. In accordance to the results obtained with NY-ESO-1–redirected T cells, α and β SE lymphocytes were significantly less alloreactive than TR cells. These results indicate that the abatement of alloreactivity observed with the editing process is not confined to a single TCR. To better investigate the basis of residual alloreactivity in SE T cells, we analyzed the expression of TCR Vβ families in the 4 cellular products. As expected, the expression of Vβ TCRs other than Vβ21.3 was consistently lower in α and β SE T cells than in TR cells. Interestingly, although in α SE T cells a residual expression was observed, in β SE cells this expression was almost abrogated. These results suggest that the potential residual alloreactivity of SE cells is not because of inefficient disruption of the endogenous repertoire, but instead to mispairing between the remaining TCR chain with the tumor-specific one. The ZFN-mediated disruption of one individual endogenous TCR chain not only eliminates the endogenous TCR repertoire, but also appears sufficient to significantly abate, although not abrogate, the risk of chain mispairing, responsible for unwanted unspecific reactivity. Overall, based on the encouraging results obtained in vitro with SE cells and considering the higher feasibility of the SE protocol compared with the CE one, we proceeded to the in vivo validation of the SE approach and compared its efficacy/safety profile with that of TCR gene transfer.

We developed an experimental humanized-mouse preclinical model of MM (supplemental Figure 8). NY-ESO-1 TR and SE T cells and mock-transduced lymphocytes were IV infused in NSG mice 4 weeks after irradiation and tumor IV injection. One week after infusion, TR and SE cells expanded, reaching a peak at ∼7 days after injection (41% ± 9 and 23% ± 3 in TR and SE treated mice, respectively; P < .05) (Figure 5A). The initial T-cell expansion was associated with high amounts of human IFN-γ in animal sera, reaching peaks 1 day after T-cell infusion of 1480 pg/mL and 2003 pg/mL for TR and SE injected mice, respectively (P < .01; Figure 5B). Although mock-transduced and transferred T cells displayed a second wave of expansion and IFN-γ secretion, timely correlated with body weight loss, ruffled fur and hunchback, all animals infused with SE cells displayed a stable and low chimerism, low serum IFN-γ levels and a body weight similar to untreated littermates (Figure 5C).

Animal organs were analyzed at euthanization to reveal the presence of residual myeloma and/or pathological signs of GVHD.

Although mice receiving mock-transduced T cells displayed variable levels of myeloma infiltration in the bone marrow, all mice treated with NY-ESO-1 redirected T cells were completely tumor-free (Figure 5D). When we repeated the experiment with luciferase labeled U266, we observed that NY-ESO-1 redirected T cells act in a very short timeframe. Although 1 week after lymphocyte infusion untreated and mock-transduced treated animals showed variable levels of luminescence, indicative of the persistence of malignant cells, TR and SE infused mice were completely signal free (Figure 5E). Thus, early waves of IFN-γ secretion and T-cell expansion observed within the first week after adoptive transfer of TR and SE cells correlate with MM debulking. A stable complete remission was confirmed for 10 out of 11 animals treated with NY-ESO-1–redirected lymphocytes for up to 8 weeks (Figure 5F). One mouse, previously injected with SE cells, displayed an increase of total photon flux in the upper abdomen, starting from week 4 after treatment, in the absence of clinical signs. To investigate whether this observation was because of tumor immune evasion, as already described with U266 cells,³³  we performed abdominal ultrasound, CT scan, and magnetic resonance. None of these techniques revealed abnormalities (not shown), and at necroscopy, no malignant cells could be detected either in bone marrow or in the abdominal cavity.

At the end of the experiment, although no differences were observed between TR and SE treated mice in terms of CD3^pos marrow infiltration (Figure 6A), splenic infiltration by TR cells (50.5 ± 9) was higher than that of SE cells (23.1 ± 5.5), and similar to that of mock-transduced lymphocytes (48.5 ± 9.4) (Figure 6B). Such different behavior is consistent with a difference in the ability of TCR redirected cells to induce off-target toxicity in this experimental model. Accordingly, TR cells, despite efficiently clearing MM, induced a high rate of xenogeneic GVHD (74.3% of treated animals), resulting in lower overall and event free survival than those obtained with SE cells (Figure 6C-D). At the end of the experiments, all mice treated with SE cells were alive and well (overall survival: 26.7% vs 100% in TR and SE infused animals, respectively; P < .005). All untreated mice had to be killed before the end of the experiment because of disease progression. All animals infused with mock-transduced T cells required premature euthanization, because of either MM progression (3/15 animals presented extramedullary localizations), or xenogeneic GVHD, with 27% of animals showing signs of both MM and GVHD.

At euthanization, several organs (skin, tongue, heart, lung, kidney, liver, and gut) were collected, fixed in formalin, and blindly analyzed by immunohistochemistry for human T-cell infiltration. The number of organs showing T-cell infiltration was significantly higher in TR (mean 4) than in SE (mean 1) treated mice (Figure 6E, left panel). More importantly, the global pathological score, corresponding to the degree of T-cell infiltration for each organ, was also significantly higher in animals infused with nonedited T cells than in animals receiving SE cells (means of 6.2 and 1.9 for TR and SE, respectively; Figure 6E right panel). Figure 7 shows the differences in organ infiltration observed by hematoxylin-eosin staining and hCD3 immunostaining in representative samples.

Overall these results indicate that the potent antimyeloma activity of NY-ESO-1-TR lymphocytes is limited by a high alloreactive potential in vivo. Conversely, NY-ESO-1 SE cells exhibited high tumor specificity, resulting in optimal antitumor activity, in the absence of GVHD.

Cancer immunotherapy and in particular chimeric antigen receptor T-cell therapy have produced a recent string of impressive clinical results. Gene transfer tools have been critical for reaching this phase of clinical development. Technological advancements today allow not only to add a gene and function to a target cell, but, thanks to the genome editing technology, to completely and permanently substitute 1 or more biological functions in a cell of choice. Coupling genome editing to cancer immunotherapy offers a unique opportunity to treat cancer patients with safe, effective and long-lasting therapeutic approaches. However, for a rapid and wide clinical translation, increased technological complexity should not impact on the manufacturing feasibility.

The complete TCR gene editing approach was designed to fully redirect T-cell specificity and significantly improved the efficacy and safety profile of TCR gene transfer.¹⁰  However, this was achieved through a 40-day manipulation protocol, requiring 2 steps of cell activation and 4 sequential steps of genetic engineering.

To improve the clinical feasibility of TCR gene editing, we describe here a single TCR gene editing approach, based on the disruption of the endogenous TCR α chain gene followed by the transfer of a tumor-specific TCR, in a single round of T-cell activation. This approach generates a large numbers of SE T cells in 2 weeks and is sufficient to abrogate the endogenous TCR repertoire. We show that this simplified highly feasible approach retains the advantages of the original TCR CE protocol, in vitro and in vivo.

The tumor antigen targeted in this work is the cancer-testis antigen NY-ESO-1, an attractive target for tumor immunotherapy because of its expression in high percentages of common tumors including high-risk MM, breast, lung, liver, esophago-gastric, prostate and ovarian cancers.^17-19  The advantage of T-cell based cancer immunotherapy, when compared with alternative treatments, largely lies on the ability of T lymphocytes to persist long-term, promote tumor immune-surveillance and expand upon tumor recurrence. These features cluster within early differentiated (T_SCM and T_CM) lymphocytes.^26,34-39  By coupling a T-cell manipulation protocol designed to preserve cellular fitness and functional properties of early differentiated T cells^25-27  with TCR SE, we generated high numbers of NY-ESO-1–redirected T_SCM and T_CM. The editing process resulted in superior levels of NY-ESO-1 TCR expression compared with the TCR transfer protocol. This observation is particularly relevant considering that the overall avidity of a tumor-specific T cell depends not only on the affinity of the TCR for its HLA-peptide complex, but also on the level of TCR expression. The TCR that we employed is codon optimized and affinity enhanced.^14,15,22,40  Overall such structural modifications ensured high tumor recognition in vitro and in vivo by both TCR gene transferred and TCR edited lymphocytes. Still, the aforementioned technological advances did not appear sufficient to abrogate alloreactivity of TR lymphocytes, that recognized and killed allogeneic targets in vitro and in vivo. Alloreactivity and GVHD are highly complex phenomena, difficult to model in vitro and in vivo.⁴¹  We compared the alloreactive potential of our different cellular products through mixed lymphocyte reactions and in a humanized mouse model. Although none of these models is fully representative of the human scenario, results were univocal. Transferred lymphocytes proved significantly more alloreactive than edited cells in vitro. Noticeably, similar results were observed by using a different tumor-specific TCR, indicating that the reduced alloreactivity observed with the editing process is not confined to a single TCR.

In our in vivo model, the detrimental alloreactive effect mediated by TR and mock-transduced lymphocytes, led to T-cell infiltrations and GVHD-like lesions in several tissues. Conversely, the full abrogation of the endogenous TCR repertoire in SE tumor redirected cells prevented off-target reactivity, while promoting a potent antitumor reaction. Overall, the rate of xenogeneic GVHD related mortality was 74.3% in mice treated with transferred lymphocytes while absent in mice receiving edited cells. Interestingly, in a clinical trial based on the infusion of NY-ESO-1 transferred lymphocytes to patients affected by MM, the development of skin and gut GVHD was recently reported in 3/20 patients,⁴²  thus indicating that, even with this optimized TCR, toxicity is a threat also in the autologous platform.

We propose TCR SE as an innovative and feasible new approach for redirecting T-cell specificity. We expect to implement it in 3 clinical settings: autologous, matched allogeneic, third-party off-the-shelf, depending on the disease type and patient characteristics. The reduction of TCR mispairing is a valuable gain for all clinical platforms. On the other hand, the complete abrogation of the entire TCR endogenous repertoire allows to safely exploit TCR edited cells in the allogeneic context and represents a first step toward the generation of off-the-shelf cellular products.

The authors thank all of Bonini’s and Naldini’s laboratories for useful discussion and Laura Falcone and Margherita Norelli for support and discussion.

This work was supported by European Union Seventh Framework Programme (SUPERSIST, ATTACK) and by the Italian Association for Cancer Research, Italian Ministry of University and Research (MIUR). E.R. was supported by a fellowship by Associazione Italiana per la Ricerca sul Cancro cofunded by the European Union.

Contribution: S.M. designed the study, conducted laboratory experiments, analyzed and interpreted data, and wrote the manuscript; P.G. and G.S. designed and performed genetic manipulations of the T cells and participated in the design of the study, data discussion, and interpretation of results; Z.M. and E.L. conducted laboratory experiments and analyzed and interpreted data; E.R. conducted experiments with the WT1-specific TCR; B.C. participated in the in vivo experiments; M.R. processed pathological samples; M.P. performed histopathological examination; G.E. and B.G. constructed the U266-Luc cell line; A.S. performed bioluminescence acquisition and analysis; N.C., G.O., and A.M. participated in data discussion and interpretation of results on T-cell activation and phenotype; M.C. participated in coculture experiments and data discussion; A.L., E.P., A.B., L.V., F.C., and L.N. participated in data discussion and interpretation of results; A.R. and M.C.H. designed and implemented the ZFN pairs used in the study; and C.B. designed and supervised the study and wrote the manuscript.

Conflict-of-interest disclosure: A.R. and M.C.H. are employees of Sangamo Biosciences. C.B. had a research contract with Molmed S.p.A. The remaining authors declare no competing financial interests.

Correspondence: Chiara Bonini, Experimental Hematology Unit, Ospedale San Raffaele Scientific Institute, via Olgettina 60, Milano, Italy; e-mail: bonini.chiara@hsr.it.