Introduction of Genetically Modified CD3ζ Improves Proliferation and Persistence of Antigen-Specific CTLs

Endogenous or genetically modified T cells with antigen-specific T-cell receptors (TCR) represent the major approach for cytotoxic T lymphocyte (CTL) therapies. Endogenous viral-specific CTL therapy, targeting cytomegalovirus (CMV), Epstein–Barr virus, and adenovirus has shown clear efficacy in some clinical studies (1, 2). On the other hand, the generation of endogenous tumor-specific CTLs has often failed, mostly likely because of the establishment of a tumor-associated microenvironment and the inactivation of endogenous CTLs that recognize self-antigens (3). To circumvent this problem, the introduction of tumor-specific TCRs to polyclonal T cells has been attempted to generate genetically modified TCR-CTLs. However, genetically modified TCR-CTL therapies targeting malignant tumors have been generally unsuccessful in previous clinical trials (4, 5). The major reasons for this were reported to be poor expansion and short persistence of the CTLs (6). To overcome these problems, TCRs with enhanced affinity have been genetically engineered; however, trials involving these molecules have often resulted in life-threatening toxicities. For example, the clinical use of affinity-modified anti–MAGE-A3 TCR-transduced CTLs has been associated with brain damage and subsequent death. This was reportedly associated with inflammatory responses that targeted the previously unrecognized expression of a MAGE-A family member in the brain (7). Thus, this event would be categorized as an off-target effect in which the affinity-modified TCR could not distinguish subtle epitope difference among MAGE-A3 family members. Another study using affinity-modified TCRs targeting MAGE-A3 resulted in severe cardiovascular toxicities. The cardiovascular toxicity was due to the recognition of the same epitope found in an unrelated protein and recognized as an on-target toxicity (8). These severe toxicities may have arisen either from on-target or off-target effects of the genetically modified TCR-CTLs. In cases where TCR affinity has been modified, or a novel TCR is introduced into a clinical trial, we risk unexpected effects in normal human tissues and subsequent life-threatening toxicity. Such severe toxicities cannot necessarily be predicted from the results of preclinical animal studies.

In the present study, we aimed to improve the efficacy of TCR-CTL therapy without TCR-affinity modification, by focusing on the dynamics of TCR signaling. When the TCR α and β chains bind to specific antigen epitopes (the HLA/peptide complex), CD3ζ is recruited to the complex via its ionized transmembrane residues, forming a supramolecular activation cluster (SMAC). Downstream activation signals are then delivered to the CTLs by several adapter molecules, such as CD3ζ, Lck, ZAP70, and others (9–11). We enhanced the downstream activation signals resulting from TCR-epitope ligation, rather than enhancement of TCR affinity, by modifying the adapter molecule component of the complex.

In terms of intracellular signal modification, a similar approach has already been successful in the other major approach to CTL therapy, chimeric antigen receptor (CAR)-modified T-cell therapy. Among the various potential CAR-T cells, CD19-CAR-T cells are the most widely used therapy, and 70% to 90% of patients with refractory B cell malignancies have achieved complete remission with this approach (12–14). During the development of CD19-CAR-T therapy, the first-generation CD19-CAR designs used only CD3ζ as an intracellular domain (ICD), and did not proliferate or persist well in vivo. However, after the introduction of ICDs for costimulatory molecules, such as CD28 or 4-1BB, the second-generation CD19-CAR approaches have improved proliferation and persistence, and subsequent clinical efficacy (15). In these designs, the costimulatory ICDs are inserted between the transmembrane domain and the CD3ζ ICD.

In the present study, we thus designed adapter molecules based on CD3ζ, into which either the CD28 or 4-1BB ICDs are inserted, to enhance signaling after TCR-ligation to specific epitopes. We hypothesized that the adapter molecules could assemble with the TCR complex only when stimulated with the corresponding antigen, and CTLs transduced with such adapter molecules would demonstrate enhanced proliferation and persistence without altered antigen specificity. We transduced these adapter molecules into endogenous and TCR-modified T cells, and examined various T-cell functions, including proliferation, upon stimulation in vitro and in vivo. We demonstrate the potential for improvement of TCR-CTL therapy using adapter molecules consisting of CD3ζ and costimulatory molecules.

The SUP-T1 tumor cell line was obtained from the American Type Culture Collection in 2013. HLA-transduced K562 cells (HLA-K562), Jurkat, Raji, U266, and SACHI tumor cell lines were maintained in our laboratory (16). All cell lines were routinely validated for authenticity by examining their immunophenotype by flow cytometry, and culture was limited to a maximum of 2 months prior to use. U266 and SACHI were tested for HLA-A2 expression by flow cytometry and NY-ESO-1 expression by RT-PCR. U266 was HLA-A2⁺/NY-ESO-1⁺ and SACHI was HLA-A2⁺/NY-ESO-1⁻. All cell lines were cultured in RPMI 1640 medium containing 10% FBS, 0.8 mmol/L l-glutamine, and 1% penicillin–streptomycin. Dr. Michael C. Jensen (Seattle Children's Research Institute, Seattle, WA) kindly provided the lentivirus vector encoding green fluorescent protein (GFP)-firefly luciferase (ffluc). U266-ffluc was derived by lentiviral transduction with the GFP-ffluc gene and was then sorted for expression of GFP.

The research protocols of this study were approved by the Institutional Review Board of Nagoya University Graduate School of Medicine. Peripheral blood mononuclear cells (PBMC) were obtained from healthy donors after written informed consent and were obtained in accordance with the Declaration of Helsinki.

We generated two artificial T cell–activating adapter molecules (ATAM; Supplementary Data), CD3ζ/CD28, and CD3ζ/4-1BB, and prepared unmodified CD3ζ as a control. The CD28 or 4-1BB ICDs were inserted in the middle of the CD3ζ ICD, such that they could assemble with the TCR complex when stimulated with the corresponding antigen (Fig. 1A). ATAMs were fused to a truncated version of the epidermal growth factor receptor (tEGFR) lacking the epidermal growth factor–binding and intracellular signaling domains, downstream of the self-cleaving T2A sequence (17). By inserting the T2A sequence between the ATAM and tEGFR, the two proteins were coexpressed at equimolar levels from a single transcript (18, 19). Cell surface tEGFR was detected using the biotinylated erbitux monoclonal antibody (mAb) for EGFR (Bristol-Myers Squibb; ref. 19). The ATAM transgenes were assembled by overlap extension PCR. ATAM-T2A-tEGFR was packaged into LZRS-pBMN-Z using the Hind III and Not I sites, and ATAM-encoding retrovirus was produced using the Phoenix-Ampho system (Orbigen; ref. 20). To visualize ATAM subcellular localization, we also prepared C-terminal enhanced GFP (eGFP)-fusion ATAMs. ATAM and eGFP were connected with a short glycine-linker (GGG), and ATAM-GGG-eGFP was packaged into LZRS-pBMN-Z using Hind III and Not I sites.

To verify signal transduction by ATAM, Jurkat-Dual reporter cells (InvivoGen) were transduced with NY-ESO-1 TCR and each ATAM respectively, and luciferase activity was assessed upon HLA-A2/NY-ESO-1–specific antigen peptide stimulation. As a simulator, we used peptide-pulsed, HLA-A2-transduced K562 (21). The cells were washed once, resuspended in AIM-V medium (Invitrogen), and pulsed with the NY-ESO-1–specific synthetic peptide at 5 μg/mL at room temperature for 2 hours. The cells were then washed once, irradiated (80 Gy), and used as a stimulator. ATAM⁺ Jurkat-Dual cells and peptide-pulsed stimulator were mixed at a 1:1 ratio, and incubated for 16 hours at 37°C. The culture supernatant then was assessed for luciferase activity according to the manufacturer's instructions.

To evaluate the functional alteration of antigen-specific CTLs, we established an ATAM-transduced (ATAM⁺) CMV-specific CTL model (referred to hereafter as ATAM⁺CMV-CTLs). We stimulated PBMCs with CMV pp65 peptide to induce endogenous CMV-CTLs (21, 22). On day 4 after stimulation, ATAMs were retrovirally transduced with the recombinant human fibronectin fragment (RetroNectin, Takara Bio), by centrifugation at 2100 rpm for 120 minutes at 32°C with the retroviral supernatant (23). Cells were expanded in RPMI-1640 medium containing 10% human serum, 0.8 mmol/L L-glutamine, 1% penicillin–streptomycin, and 0.5 μmol/L 2-ME (culture medium, CM), and supplemented with recombinant human IL2 to a final concentration of 50 IU/mL. On day 9 or 10 of peptide stimulation of CMV-CTL, ATAM-positive cells were enriched using immunomagnetic selection with biotin-conjugated anti-EGFR mAb and streptavidin beads according to the manufacturer's instructions (Miltenyi Biotec). Purified ATAM⁺CMV-CTLs were restimulated with anti-CD3/CD28 beads and expanded for subsequent experiments (24).

We adopted the NY-ESO-1–specific TCR as a model of tumor-specific TCR-gene insertion. Briefly, the construct was developed based on the 1G4 TCR that targets the NY-ESO-1 peptide 157-165 (SLLMWITQC) bound to HLA-A02:01 (25, 26). In addition, the TCR-α and -β chains were sterically stabilized by the additional interchain disulfide-bonding (26). CD8⁺ T cells were purified with immunomagnetic beads (Miltenyi Biotec) from healthy donor PBMCs, activated with anti-CD3/CD28 beads at the ratio of 1:1 (Invitrogen), and cultured in CM supplemented with 50 IU/mL of human IL2. On days 3 and 4, expanded CD8⁺ T cells were transduced by centrifugation on recombinant human fibronectin (RetroNectin) preloaded with the retroviral vector as described before. The retroviral supernatant from ATAM-tEGFR and NY-ESO-1 TCR-eGFP-packaged Phoenix-Ampho cells were mixed at a 1:1 ratio, and loaded onto RetroNectin to transduce beads-stimulated T cells. After 3 to 5 days from transduction, NY-ESO-1 TCR⁺ and ATAM⁺ cells were sorted by FACSAria II, targeting the eGFP⁺ EGFR⁺ fraction. Purified ATAM⁺NY-ESO-1 TCR-CTLs were restimulated with anti-CD3/CD28 beads at the ratio of 1:1 and expanded for subsequent experiments (24).

All samples were analyzed by flow cytometry on the FACSAria II and FACSCanto II (BD Biosciences), and analyzed using FlowJo (Tree Star). All fluorescent dye-conjugated antibodies to the following proteins were purchased from BD Biosciences. [Streptavidin-PE, -APC, and -BV421, CD8-APC (clone RPA-T8), IFNγ-FITC (25723.11), IL2-APC (5344.111), CD45RA-FITC (L48), and CD62L-PE-CF594 (DREG-56).]

GFP-fusion ATAMs were retrovirally introduced into SUP-T1 cells (GFP-ATAM⁺ SUP-T1) using the same method described in the generation of ATAM⁺CMV-CTLs. GFP-ATAM⁺ SUP-T1 cells were mounted on slides with Smear Gell according to the manufacturer's instructions (GenoStaff). After fixation with 4% formaldehyde, SlowFade Diamond Antifade Mountant (Thermo Fisher Scientific) was applied with DAPI counterstaining, followed by the addition of a coverslip. Confocal imaging experiments were performed on an inverted confocal laser microscope system (Nikon TiE-A1R, Nikon Instech Co., Ltd.). Image processing and analysis were performed using NIS Elements Viewer 4.20 software (Nikon Instech Co., Ltd.).

For the ⁵¹Cr release assay, target cells were labeled for 2 hours with ⁵¹Cr (PerkinElmer), washed twice, dispensed at 2 |\times $| 10³ cells per well into triplicate cultures in 96-well round-bottom plates, and incubated for 4 hours at 37°C with ATAM⁺CTLs at various effector-to-target (E:T) ratios. Percentage of specific lysis was calculated using a standard formula [(experimental − spontaneous release)/(maximum load − spontaneous release) × 100%] and expressed as the mean of the triplicate samples. For the coculture assay, U266 cells were labeled with CellTrace violet (Thermo Fisher Scientific), washed, and plated with control or CD3ζ/4-1BB–transduced NY-ESO-1 TCR CTLs at various E:T ratios, without IL2 supplementation. From 72 to 168 hours of incubation, effector cells were stained with anti-CD8 and analyzed with flow cytometry every 24 hours, assessing the percentage of CTLs and U266 cells within the live cell gates.

ATAM⁺CMV-CTLs, and HLA-K562 cells pulsed with or without CMV peptide, were mixed at a 1:1 ratio in the presence of brefeldin A (Sigma-Aldrich) and then fixed and permeabilized with Cell Fixation/Permeabilization Kits (BD Biosciences) for the intracellular cytokine assay (21). After fixation, CTLs were stained with CD8 mAb to separate them from stimulator cells. Anti-IFNγ or anti-IL2 were used to detect intracellular cytokines. For the cytokine secretion assay, ATAM⁺CMV-CTLs, and HLA-K562 cells pulsed with or without CMV peptide, were mixed at a 1:1 ratio and incubated for 16 hours (21). Culture supernatants were collected and IL2 measured by ELISA according to the manufacturer's instructions (BD Biosciences).

ATAM⁺CMV-CTLs, or ATAM⁺NY-ESO-1 TCR-CTLs, were stimulated with γ-irradiated HLA-K562 cells pulsed with the specific peptide at a 1:1 ratio. For simulator, we prepared peptide-pulsed, HLA-transduced K562 as described previously (21). ATAM⁺ CMV-CTLs or ATAM⁺ NY-ESO-1 TCR-CTLs were mixed with the peptide-pulsed stimulator at a 1:1 ratio and incubated at 37°C. The live cell number was assessed using standard trypan-blue dye exclusion. For the single-stimulation assay, CTLs were cultured with 10 ng/mL of IL7 and IL15 twice weekly. For the repeated stimulation assay, CMV-CTLs were stimulated with their specific antigen on days 0, 7, 14, and 21. IL2 (50 IU/mL) was supplied every 3 days. To assess the CTL apoptotic status after antigen stimulation, ATAM⁺CMV-CTLs were stained with annexin V and DAPI. The phenotypic change of ATAM⁺CMV-CTLs was analyzed by dual staining with CD45RA and CD62L.

The murine experimental procedures were approved by the Institutional Animal Care and Use Committee of Nagoya University Graduate School of Medicine. Seven-week-old male NOD/Shi scid IL2Rγ knockout (NOG) mice were sub-lethally irradiated with 1.5 Gy using X-ray irradiation equipment (Hitachi Medical Corporation). The next day, mice were inoculated with 5 × 10⁶ U266-ffluc cells by intravenous (i.v.) injection. Twenty-one days from U266-ffluc injection, 2 × 10⁶ untransduced CD8⁺ T cells, tEGFR⁺NY-ESO-1-TCR or CD3ζ/4-1BB⁺NY-ESO-1-TCR CTLs, were transferred to the mice by i.v. injection. For bioluminescence-based measurement of tumor growth, animals were injected intraperitoneally with 10 μL/g body weight of 15 mg/mL firefly D-luciferin in PBS. Tumor progression was monitored using the IVIS Spectrum System (Caliper Life Sciences) every 7 days.

All experimental data are presented as mean ± SEM. Data were analyzed using paired Student t tests to evaluate two-tailed statistical differences when comparing two groups. Differences among three or more groups were evaluated with one-way ANOVA followed by Bonferroni test. Two-way ANOVA followed by Bonferroni multiple comparison test was used to assess differences between multiple treatment groups over different time points. Statistical analysis was performed on GraphPad Prism 7 software (GraphPad Software).

To transduce ATAMs into live cells, we designed cDNAs that encode ATAMs and coexpressing tEGFR (Fig. 1A and Supplementary Data). By using retrovirus vectors, ATAMs were then transduced into CD8⁺ T cells three days after CD3/28 beads stimulation, and tEGFR selection was performed on day 7. Transduction efficiencies were approximately 18%–35%. After immunomagnetic selection, the tEGFR positive cells were purified up to 76%–89% (Fig. 1B). Western blots showed that each ATAM⁺CD8⁺ T cells express CD3ζ-like molecules corresponding to the expected ATAM size and different to endogenous CD3ζ (Fig. 1C). To clearly show independent ATAM expression in the absence of endogenous CD3ζ, we transduced ATAMs into the Raji cell line and assessed by flow cytometry. The mAb to CD3ζ is specific for the CD3ζ transmembrane domain, which is contained within the ATAMs in equal amounts. ATAM transduction into Raji cells resulted in an increased CD3ζ signal (Fig. 1D). To further investigate the subcellular localization of ATAMs, SUP-T1 cells were transduced with GFP-tagged ATAMs. Fluorescent confocal microscopy revealed that ATAMs localized to the cell membrane and cytoplasm. This indicated that the transduced ATAMs were transported through cytoplasm and expressed on the cell surface (Fig. 1E). To confirm signal transduction upon antigen-stimulation, we investigated alterations in NF-κB signaling using Jurkat-Dual reporter cells transduced with the different constructs. Luciferase activities after HLA-A2/NY-ESO-1–specific antigen stimulation were assessed, and the differences from untransduced Jurkat-Dual cells are shown. NF-κB signals increased upon stimulation in all groups, compared with untransduced cells. Transduction of CD3ζ or CD3ζ/CD28 increased NF-κB signaling, with peak activity observed at 24 hours, whereas CD3ζ/4-1BB transduction resulted in smaller, but constant, increases in NF-κB signaling (Fig. 1F).

To evaluate the functional features of ATAM⁺CTLs, we adopted CMV-specific CTLs as an antigen-specific CTL model and transduced ATAMs into CMV-specific CTLs to generate an ATAM⁺CMV-CTL. All subsequent assays were performed between days 12 and 14 (Fig. 2A). Transduction efficiency was 15%–30%, and tEGFR⁺ cells were observed predominantly in the CMV-pp65 tetramer⁺ cell fraction. After EGFR selection, the proportion of tEGFR⁺ cells increased to 70%–90% and was similar for each construct (Supplemental Fig. S1A and S1B).

To evaluate the potential effect of ATAM transduction on the immediate response to specific antigens, the cytotoxicity of ATAM⁺CMV-CTLs was investigated with standard ⁵¹Cr release assays (Fig. 2B and C). Target cells were HLA-K562 pulsed with or without specific CMV-pp65 peptide. The cytotoxicity of control CMV-CTLs and each type of ATAM⁺CMV-CTL was equivalent and dependent on the E:T ratio. In the intracellular cytokine assay against HLA-K562 cells pulsed with CMV peptide, similar proportions of responders were positive for IFNγ (Fig. 2D and E) and IL2 (Fig. 2F) with both control and ATAM⁺CMV-CTLs. IL2 secreted by ATAM⁺CMV-CTLs was determined in the supernatant using ELISA. CMV-CTLs were cocultured with HLA-K562 cells pulsed with or without CMV-pp65 peptide. No significant difference in IL2 secretion against positive targets was observed with control or ATAM⁺CMV-CTLs. Without specific target, no ATAM⁺CMV-CTL secreted detectable IL2 (Fig. 2G).

A major problem in past clinical trials of adoptive T-cell therapy has been the poor expansion and short persistence of CTLs. Because we saw no differences in the immediate response of ATAM⁺CMV-CTLs, we investigated the survival, phenotypic change, and expansion of CTLs after antigen stimulation. Control or ATAM⁺CMV-CTLs were produced (Fig. 2A; Supplementary Fig. S1), and after selection, each type of CMV-CTL was stimulated using HLA-K562 cells pulsed with CMV-pp65 peptide at a 1:1 ratio. First, we analyzed cell survival 24 hours after stimulation, using annexin V/DAPI staining (Fig. 3A and B). The control cells (tEGFR-transduced) had a significantly lower survival compared with the CD3ζ or the ATAM⁺ cells, whereas the CD3ζ and the ATAM⁺ groups had no significant differences in survival (Fig. 3A and B). The CD3ζ group showed a tendency toward better survival, but it was not statistically significant compared with other ATAMs. These data indicate that ATAM⁺CMV-CTLs were more resistant to activation-induced cell death than were control CMV-CTLs. Because ATAM transduction may influence cell fate after antigen exposure, we next evaluated phenotypic changes after stimulation (Fig. 3C and D). The expression of CD45RA and CD62L was analyzed before, and 7 days after, second antigen stimulation in total. It would be interesting if we could obtain phenotype and exhaustion marker alteration during repeated stimulation, however, we only could assess the phenotype before and after the second stimulation, because of the availability of the cultured cells. Before stimulation, most CTLs showed a CD45RA⁻CD62L⁻ effector memory T cell (T_EM) phenotype. Although surface phenotype was not changed in control CMV-CTLs, that of ATAM⁺CMV-CTLs was significantly skewed toward a CD45RA⁻CD62L⁺ central memory T cell (T_CM) phenotype following stimulation (Fig. 3C and D). Thus, ATAM⁺CMV-CTLs might expand and persist after stimulation. To assess the proliferation capacity after stimulation, cells were exposed to a single-course of antigen stimulation in vitro and cell numbers recorded weekly. CD3ζ/4-1BB–transduced CMV-CTLs showed significantly greater expansion compared with CD3ζ-transduced CMV-CTLs (Fig. 3E). We did not observe significant differences between the CD3ζ and CD3ζ/CD28 groups, or the CD3ζ/CD28 and CD3ζ/4-1BB groups. Whereas proliferation was better following ATAM transduction, we confirmed self-limiting growth after antigen stimulation.

To investigate the persistence of ATAM⁺CMV-CTLs, we stimulated ATAM⁺ cells repetitively (Fig. 3F). ATAM⁺CMV-CTLs were stimulated weekly using irradiated HLA-K562 cells pulsed with CMV-pp65 peptide (days 0, 7, 14, and 21), and the resulting proliferation assessed. CD3ζ/4-1BB–transduced CMV-CTLs continued to proliferate until day 21 after stimulation (Fig. 3F). CD3ζ/4-1BB–transduced CMV-CTLs showed significantly higher total cell counts compared with the CD3ζ-transduced CMV-CTLs. We observed a significant difference between the CD3ζ and CD3ζ/4-1BB groups, but not between the CD3ζ and CD3ζ/CD28 groups. ATAM⁺CMV-CTLs expansion, including that of CD3ζ/4-1BB, was limited, unlike the unlimited expansion of a neoplastic cell line.

To further investigate the effect of ATAM transduction in the context of the antitumor effect of TCR-CTL therapy, we generated ATAM⁺NY-ESO-1 TCR-CTLs (Fig. 4A). ATAMs and NY-ESO-1 TCR were simultaneously transduced into CD8⁺ T cells on days 3 and 4 of culture, sorted by FACSAria II, and restimulated for the downstream experiments. The transduction efficiency of the eGFP⁺ EGFR⁺ fraction was 3% to 22%. After cell sorting, the eGFP⁺ EGFR⁺ fraction significantly increased. The eGFP⁺ EGFR⁺ fractions of purified cells were similar for each construct (Fig. 4B). We investigated in vitro expansion with these eGFP⁺ EGFR⁺ T cells after a single course of antigen stimulation at a 1:1 ratio with HLA-K562 cells (pulsed with NY-ESO-1 specific peptide). CD3ζ/4-1BB–transduced NY-ESO-1 TCR-CTLs showed significantly better expansion than CD3ζ or CD3ζ/CD28–transduced NY-ESO-1 TCR-CTLs (Fig. 4C). The cytotoxic activities of ATAM⁺NY-ESO-1 TCR-CTLs were investigated with standard ⁵¹Cr release assay to confirm the immediate CTL response, using the HLA-A2⁺/NY-ESO-1⁺ U266 and HLA-A2⁺/NY-ESO-1⁻ SACHI tumor cell lines as target cells (Fig. 4D and E). The cytotoxic activities of control NY-ESO-1 TCR-CTLs and ATAM⁺NY-ESO-1 TCR-CTLs were all similar, and dependent on the E:T ratio. Taken together, the most promising construct among the ATAMs was CD3ζ/4-1BB, given the improved NY-ESO-1 TCR-CTL persistence without altering the immediate response to specific targets.

To confirm this hypothesis, we performed coculture assays. The targets were U266 tumor cells and the effector cells were control or CD3ζ/4-1BB–transduced NY-ESO-1 TCR-CTLs. These were cocultured at various E:T ratios and incubated for a total of 168 hours, assessing the percentage of CTLs and U266 cells by flow cytometry every 24 hours. Under short-term incubation up to 72 hours, the control and CD3ζ/4-1BB–transduced NY-ESO-1 TCR-CTLs showed similar E:T cell proportions under both E:T ratio conditions (1:1 and 1:8). However, when we prolonged the incubation time up to 144 hours, the CD3ζ/4-1BB–transduced NY-ESO-1 TCR-CTLs showed a higher E:T cell proportion than controls, particularly in the E:T ratio of 1:8 (Fig. 4F and G). Collectively, these data indicated that CD3ζ/4-1BB transduction could improve NY-ESO-1 TCR-CTL persistence and antitumor effects in vitro.

To assess the possible improvement of antitumor effect by CD3ζ/4-1BB transduction, we generated a U266-xenografted NOG mouse model (Fig. 5A). The tumor burden of each mouse was then evaluated by weekly bioluminescent imaging (Fig. 5B and C). Before CTL transfer, tumor flux was similar among each group, but the NY-ESO-1 TCR-CTL injection was found to suppress tumor burden, as calculated by the sum of the entire body signal intensity. Comparing with the NY-ESO-1 TCR untransduced control, tEGFR-transduced and CD3ζ/4-1BB–transduced NY-ESO-1 CTLs demonstrated better tumor suppression, and the effect was observed for a prolonged period. However, because the growth of the U266 cell line is relatively slow and tumor-engrafted mice remained alive for a long period, mice treated with CD3ζ/4-1BB–transduced NY-ESO-1 CTLs did not show a survival benefit in this model.

To improve TCR-CTL therapies, we sought to modify the intracellular signaling of TCR-CTLs after epitope ligation, without altering TCR-specificity. In the present study, we generated two adapter molecules, which we named ATAMs, and used the unmodified CD3ζ molecule as a control. The ATAMs were based on the CD3ζ backbone, but with a CD28 or 4-1BB ICD inserted just below the transmembrane domain. We hypothesized that ATAMs could assemble to the TCR complex upon antigen stimulation and mediate enhanced activation signals within the SMAC. We successfully transduced ATAMs into T cells, and showed that this improved T-cell expansion after specific antigen stimulation, without any change in cytokine production or cytotoxic activity in vitro. Antigen stimulation of ATAM⁺CMV-CTLs caused a major phenotype change from an effector memory to a central memory cell type. Among the ATAMs we designed, CD3ζ/4-1BB generated the best improvement in CTL expansion and persistence after repetitive antigen stimulation. In the NY-ESO-1 TCR-CTL model, we further demonstrated that CD3ζ/4-1BB transduction not only improved in vitro persistence, but also in vivo antitumor effects, without modifying TCR affinity.

The relationship between long-term persistence of transferred T cells and the clinical response of genetically modified TCR-CTL therapy has been at times controversial (4, 5, 27–29). However, it is generally well established that persistence of genetically modified CTLs is correlated with tumor regression in T-cell transfer therapy (29, 30). The persistence provided by CD3ζ/4-1BB transduction in the present study was demonstrated with repeated stimulation and coculture assays with low doses of effector cells. The CD3ζ/4-1BB–transduced NY-ESO-1 TCR-CTLs also showed superior tumor regression. In comparison with previous reports, the T-cell numbers injected in mice was significantly lower in this study (31). This lower dose of T cells more accurately reproduces the actual tumor environment (32, 33). CD3ζ/4-1BB transduction thus made the CTLs more resistant to antigen stimulation, and the low dose CD3ζ/4-1BB–transduced-CTLs profoundly suppressed antigen-positive tumor growth in vivo. Taken together, these data suggest that CD3ζ/4-1BB transduction has the potential to improve the efficacy of TCR-CTL therapy.

The most attractive aspect of our ATAM strategy is safety, when compared with previous approaches to TCR-CTL modification. ATAM transduction did not cause any alteration of responses against target cells with or without HLA/specific peptide. No unexpected reactions were observed in the absence of target HLA/peptide complexes, in either the CMV-CTL or the NY-ESO-1 TCR-CTL models. In contrast, TCR affinity modification carries a risk of unexpected reactions to normal human tissues and subsequent life-threatening toxicity. Although it is still a concern that the transduction of a novel gene might induce the neoplastic transformation of primary T cells, our experiments showed that ATAM⁺ CTLs did not expand indefinitely, and instead underwent apoptosis after a period of time (34, 35). Furthermore, excessive cytokine production was not associated with the ATAM transduction. In CD19-CAR-T therapy on the other hand, which has potent antileukemia activity, some patients experienced severe cytokine release syndromes (12, 36). To prove the potential efficacy of ATAM, we eventually need to directly compare affinity-modified TCRs with the ATAM⁺ CTLs in vivo.

ATAMs have the advantage of a broad potential utility in combination with other TCR-CTLs. A wide variety of tumor-specific antigens, such as cancer–testis antigens (e.g., MAGE-A, SSX2, NY-ESO-1, and XAGE) and neoantigens are now recognized (37–40). However, compared with the high response ratio of CD19-CAR-T therapy, the efficacy of genetically modified TCR-CTL therapies targeting such antigens has not been satisfactory in previous clinical trials (4, 5, 12). To improve this efficacy, we thus introduced ATAMs to enhance intracellular signaling following antigen recognition. In the context of CAR-T modification, some research efforts have been devoted to control intracellular signaling synthetically (41–43). Such synthetic signal strategies might also therefore be applicable to TCR-CTL approaches to perfectly control T-cell signaling in the near future.

The functional differences between the ATAMs we tested must reflect the particular combination of intracellular costimulatory molecules in each case. The inclusion of either the CD28 or 4-1BB ICD in the CMV-CTL model improved CTL expansion after the first stimulation. In particular, CD3ζ/4-1BB–transduced CMV-CTLs maintained a significantly better expansion after repetitive stimulation. The in vitro study of NYESO-1 TCR-CTLs showed that, among the ATAMs tested, only CD3ζ/4-1BB could improve expansion. Second-generation CAR-T therapies with CD28 and 4-1BB domains (CD28-CARs and 4-1BB-CARs, respectively) have different impacts on CAR-T cell functionality and clinical effects. In patients with chronic lymphocytic leukemia, the clinical efficacy of 4-1BB-CAR-T therapy has been superior to that seen with CD28-CARs (44, 45). One of the reasons was the long-term in vivo persistence of 4-1BB CAR-T cells (44). The endogenous 4-1BB signal implicated in prolonging the survival of memory CD8 T cells involves tumor necrosis factor receptor-associated factors (46). Even though endogenous activity is not fully recapitulated in CAR-T cells, it has been reported that 4-1BB-CARs increase T_CM cells and improve persistence (47). On the other hand, CD28-CARs have been associated with greater functionality soon after antigen-exposure (48). In the present study, we did not observe such functional changes in CD3ζ/CD28-transduced CTLs, but the benefits shown by CD3ζ/4-1BB–transduced CTLs were similar to those reported for 4-1BB-CARs.

Our study has some limitations that must be overcome to enable clinical application of the findings. ATAM transduction efficacy must be improved; the ATAM transduction efficacy was lower than expected, particularly during the generation of ATAM⁺NY-ESO-1 TCR-CTLs, meaning that we needed to restimulate and culture before transfer. A previous study reported that patients with clinical responses received T cells that had been subjected to shorter periods of in vitro culture (4). Although efficient transduction of both TCR and ATAM genes is difficult, improvement in the generation method for ATAM⁺ CTLs, without repeated stimulations, is ideally required. It would also be of interest to develop all-in-one versions of TCR-ATAM that are already fused to the costimulatory domains to the TCRs.

In conclusion, ATAM⁺ TCR-CTLs exhibited improved proliferation and persistence in vitro and in vivo. In combination with various tumor-specific TCRs, this strategy of controlling the intracellular signaling of TCR-T cells with ATAM transduction may be a genetic-modification approach to improving the efficacy of TCR-T therapy.

H. Kiyoi reports receiving commercial research grants from Chugai Pharmaceutical Co. Ltd., Bristol-Myers Squibb, Kyowa Hakko Kirin Co. Ltd., Zenyaku Kogyo Co. Ltd., FUJIFILM Corporation, Nippon Boehringer Ingelheim Co. Ltd., Astellas Pharma Inc., and Celgene Corporation, has received honoraria from speakers bureau of Bristol-Myers Squibb and Pfizer, and is a consultant/advisory board member for Astellas Pharma Inc. and Daiichi Sankyo Co. Ltd. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Miyao, S. Terakura

Development of methodology: K. Miyao, S. Terakura, K. Watanabe

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Miyao, S. Terakura, S. Okuno, J. Julamanee, K. Watanabe, H. Hamana, H. Kishi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Miyao, S. Terakura, S. Okuno, K. Watanabe, H. Kishi, D. Koyama, T. Goto, T. Nishida

Writing, review, and/or revision of the manuscript: K. Miyao, S. Terakura, K. Watanabe, R. Sakemura, M. Murata, H. Kiyoi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Hamana, R. Sakemura

Study supervision: S. Terakura, M. Murata, H. Kiyoi

This work was supported by grants from the Foundation for the Promotion of Cancer Research (Tokyo, Japan; to S. Terakura), the Japan Society for the Promotion of Science (JSPS) KAKENHI (24790969 and 15k09497 to S. Terakura), Practical Research for Innovative Cancer Control (15ck0106067h0002 and 17ck0106291h0001 to S. Terakura), and Practical Research Project for Allergic Diseases and Immunology (15ek0510010h0003 to M. Murata). J. Julamanee was supported by a grant from the Research Foundation of Prince of Songkla University (Grant No. MOE. 0521.1.0601(2)/6058).

The authors would like to thank the Division of Experimental Animals and the Division of Medical Research Engineering, Nagoya University Graduate School of Medicine for their technical assistance.

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.