An Isolated TCR αβ Restricted by HLA-A*02:01/CT37 Peptide Redirecting CD8⁺ T Cells To Kill and Secrete IFN-γ in Response to Lung Adenocarcinoma Cell Lines Open Access

The presence of actionable genetic mutations in epithelial growth factor receptor (EGFR) and rearrangements in anaplastic lymphoma kinase prompt the prescription of specific tyrosine kinase inhibitors (1, 3). Precision treatment approaches are not available for mutations in the BRAF, HER2/neu, and KRAS genes (3). Moreover, 70% of patients do not carry any of these actionable mutations. Thus, we must develop innovative therapeutic approaches to address the needs of those patients. An option for those cases is the use of Ab blockers of checkpoint inhibitors (4, 5). Treatment with these Abs has shown that these tumors express immunogenic Ags (6). Some of the most immunogenic molecules (aberrantly) expressed in several tumors, including lung adenocarcinoma (ADC), are cancer/testis (CT) Ags (7–11). Hence, adoptive transfer immunotherapies targeting CT Ags may cover the needs of that important percentage of patients suffering from non–small cell lung cancer (NSCLC).

Although various lung ADCs express a diverse number of CT Ags, there is a need to carefully select those whose expression is truly restricted to cancerous and testis/ovary cells for development of adoptive transfer therapies. Following that precept, we selected the CT37 Ag. Moreover, we have identified a CT37-Ag peptide restricted by the MHC HLA-A*02:01 class I molecule and were able to pull from peripheral blood of a patient suffering from lung ADC CT37-peptide/HLA-A*02:01–restricted CD8⁺ T cells. We isolated a CD8⁺ T cell clone from the polyclonal population of CD8⁺ T cells, molecularly cloned its α and β TCR chains, developed constructs, and efficiently transduced heterologous CD8⁺ cells with these constructs, redirecting them to recognize CT37-peptide/HLA-A*02:01 complexes on the cell surface of autologous CD3-depleted PBMCs, and on HLA-A*02:01–positive lung ADC cell lines HCC2935 (carrying a wild-type TP53 gene), and the NCI-H1993 and H522 (carrying a disabling mutation in the TP53 gene). In addition, we developed a modified CD3ζ chain capable of enhancing both CD8⁺ T cell–mediated Ag-specific cytotoxic (CTL) activity and secretion of IFN-γ. Our α and β TCR/CD3ζ construct could be used to redirect CD8⁺ CTL to lyse CT37-expressing lung ADC tumor cells in an MHC class I–specific-peptide restricted fashion.

Using the CT database (http://www.cta.lncc.br) we selected those CT Ags with protein expressed only in testis/ovary and cancer cells, according to the Human Protein Atlas database (http://www.proteinatlas.org) and CT37’s GeneCard proteomics information (http://www.genecards.org). In addition, based on publicly available information, we selected those CT Ags reported to be expressed in NSCLC (7–11). An ideal CT Ag will have in silico–determined strong binding peptides (percentage rate <0.1) restricted by the most frequent MHC class I molecules in several populations, including HLA-A*02:01, HLA-A*24:02, and at least one of the following MHC class I molecules: HLA-A*03:01, HLA-A*11:01, HLA-B*07:02, HLA-C*07:01 (allelefrequencies.net). Peptides and peptide specificities were determined using the NetMHCpan Server (http://www.cbs.dtu.dk/services/NetMHCpan/).

A total of 43 patients were recruited, including 20 lung ADC patients, in 2010–2012. The rest were recruited in 2013. Patients provided informed consent under protocols approved by the Institutional Review Board of the Houston Methodist Research Institute in Houston (TX) and/or an Institutional Review Board regulated by the Peruvian Ministry of Health. Selection criteria were patients suffering from lung ADC who had undergone surgery and were under clinical observation; from this we had access to demographic information, including age and gender, and clinical data such as tobacco smoking status, computerized tomography scanning information, histological classification, with available frozen tumor tissue suitable for CT37 immunohistochemistry (IHC), mRNA, and genomic DNA extraction. We selected nine HLA-A*02:01–positive patients, with their tumors expressing CT37 mRNA. Four were freely willing to donate leukocytes for FACS analysis, and isolation and cloning of TCR α and β chains.

We used the Micro SSP HLA-A*02 Allele Specific Trays (One Λ) and 50 ng of genomic DNA, isolated from tissue or blood cell pellets, using DNeasy Blood and Tissue extraction kits (Qiagen, Valencia, CA). Available lung ADC cell lines were also genotyped.

For the assessment of CT37 (FMR1NB/NY-SAR-35) expression, we used the 7500 Fast Real-Time PCR System and assay-on-demand primers for CT37 (ID Hs00896732_m1) and the housekeeping gene pyruvate dehydrogenase β (PDHB, ID Hs00168650) (Applied Biosystems, Carlsbad, CA). To calculate the relative quantity of mRNA expression, we used the 2^{−ΔΔcycle threshold} method implemented in the software (Applied Biosystems). The data are presented as the fold change and/or log₂ of fold change in gene expression normalized to the endogenous reference gene PDHB. Total RNA was extracted from tumor tissues using RNeasy Fibrous Tissue Mini Kit RNA or from cells using Qiagen’s RNA isolation kit. cDNA was obtained from 3 μg of total RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Approximately 100 ng of cDNA was used to determine the expression levels of these genes. The HCC2935, H1993, H1299, H522, H23, and H460 lung ADC cell lines were tested as well.

Total RNA was reversed transcribed using an oligo(dT) primer and the Qiagen Omniscript RT kit. The EGFR (exons 18–22) and KRAS (exons 2 and 3) were PCR amplified using Platinum TAQ polymerase Supermix (Life Technologies, Carlsbad, CA) and primers: EGFR forward (FW) 5′-TGAAGGCTGTCCAACGAATG-3′ EGFR, reverse (RV) 5′-AGGCGTTCTCCTTTCTCCAG-3′, and KRAS FW 5′-AGGCCTGCTGAAAATGACTG-3′ and RV 5′-TGGTGAATATCTTCAAATGATTTAGT-3′. The dye terminator cycle sequencing method was used to detect gene mutations in a Beckman Coulter GenomeLab GeXP Genetic Analysis System. All lung ADC cell lines were also tested.

For IHC, lung tissue samples from the 10 lung ADC patients that underwent surgery were fixed in paraformaldehyde and embedded in paraffin. Testis samples were provided by the Department of Pathology at the Methodist Hospital. For IHC, we used heat-induced epitope retrieval in citrate buffer (Thermo Fisher Scientific, Waltham, MA), rabbit anti-human CT37 polyclonal Ab HPA011284 (Sigma-Aldrich, St. Louis, MI), and a Polymer Detection System (Lab Vision Products; Thermo Fisher Scientific, Kalamazoo, MI). We immunostained 5 μm thick sections. Negative controls were sections incubated with irrelevant normal rabbit IgG (KPL, Gaithersburg, MD). We used 1× automation buffer (pH 7.5) in all wash steps. SuperMount permanent aqueous mounting media (BioGenex, San Ramon, CA) was used for mounting immunostained tissue sections. Mounted sections were assessed at 400× total magnification and 0.8 numerical aperture of the objective lenses. Image acquisition was performed using a computerized analysis system comprising a BX41 microscope with a U-TVIX-2 and a U-CMAD3 tube and adapter attached for on-screen viewing, a C3040 4.1 megapixel digital camera, and Magnafire-SP software (Olympus America, Central Valley, PA).

Leukocytes were obtained from donor NSCLC_7 by leukapheresis. Half of the bag (∼1 × 10⁹ leukocytes per 100 ml) was used to isolate PBMCs by Ficoll-gradient centrifugation. Then, 3 × 10⁸ CD8⁺ T cells were negatively selected using the CD8⁺ T cell isolation Miltenyi Biotec kit, four large-capacity columns, and protocols furnished by Miltenyi Biotec. Next, 5 × 10⁷ CD8⁺ T cells (the rest were frozen) were used to isolate peptide–Ag-specific cells by positive selection using biotinylated pentamers (Proimmune), streptavidin microbeads, and one large-capacity column (Miltenyi Biotec, Somerville, MA). We obtained 3 × 10⁶ peptide-specific CD8⁺ T cells. Then, 1 × 10⁶ peptide-specific CD8⁺ T cells per well were stimulated in vitro with 5 × 10⁶ YLCSGSSYFV peptide-pulsed CD3⁺ cell–depleted autologous PMCs irradiated 3000 rad per well, in 24-well Costar tissue-culture plates, and 2 ml final volume of media. We used Roswell Park Memorial Institute/RPMI 1640 medium (Life Technologies), supplemented with 10% heat-inactivated human AB serum, and 2 mM l-glutamine, 1 U/ml penicillin G, and 100 μg/ml streptomycin (Life Technologies). The CT37 peptide YLCSGSSYFV was used at 100 ng/ml final concentration. Human recombinant (hr) IL-2 (10 ng/ml; R&D Systems, Minneapolis, MN) was added 72 h after peptide stimulation and the cultures left for 10 d (to let the cells rest). Then, 48 h after IL-2 stimulation and every 48 h thereafter, 100 μl of spent culture media was retrieved from each well and replaced by 100 μl of fresh complete media per well. After two cycles of stimulation the cells were harvested; live cells were obtained by Ficoll-gradient centrifugation and analyzed by FACS for quality control. We harvested 4.5 × 10⁶ peptide-specific CD8⁺ T cells. Cultures were placed in a humidified tissue culture incubator, at 37°C and 5% C0₂. The highly purified YLCSGSSYFV and YYLCSGSSYF peptides were obtained from Creative Peptides (Shirley, NY).

HLA-A*02:01/CT37 peptide–specific CD8⁺ T cells were cloned using the limiting dilution technique. We used a limiting dilution to dispense (approximately) one HLA-A*02:01/CT37 peptide–specific CD8⁺ T cell in 100 μl of media for every three wells, with each well already containing peptide YLCSGSSYFV-pulsed 5 × 10⁵ CD3-depleted autologous PBMCs in 100 μl of media per well, in 96-well tissue culture plates from Costar. The final peptide concentration was 100 ng/ml. We added hrIL-2 (10 ng/ml; R&D Systems) 72 h after peptide stimulation. Cultures were placed in a humidified tissue culture incubator for 10 d (to let the cells rest), at 37°C and 5% C0₂.

Next, 1 × 10⁶ CD8⁺ T cells specific to the YLCSGSSYFV peptide were lysed using buffer RLT and total RNA extracted using Qiagen’s RNA isolation kit. RNAs in RNase inhibitor (Promega, http://www.promega.com) was added to the extracted total RNA. First-strand cDNA was prepared from total RNA reverse-transcribed using the SMARTer RACE kit (TaKaRa Bio, Madison, WI). Primers specific for c-terminal Cα region or Cβ regions were used together with the SMARTScribe reverse transcriptase (RT) enzyme, capable of incorporating the SMARTer II-A oligonucleotide sequence in the 5′ end of the first-strand cDNA during the RT process. This allows the amplification of TCR α- and β-chain cDNAs without a priori knowledge of the V region sequences. RT primer sequences were as follows: TCRα 5′-GCT GGA CCT CAG CCG CAG CGT CAT-3′; TCRβ1 5′-TCA GAA ATC CTT TCT CTT GAC CCA-3′; TCRβ2 5′-CTA GCC TCT GGA ATC CTT TCT CTT-3′.

The TCR α and β chains were amplified using the universal long 5‘ PCR primer II-A and 3′-specific reverse primers for Cα, or Cβ1, or Cβ2 chains (three separate reactions), and the Platinum TAQ polymerase Supermix (Life Technologies). Then, a seminested PCR was performed using the universal short 5′ PCR and each of the C-region PCR primers. The PCR products were then run in a 1.5% agarose/Tris–borate–EDTA buffer gel and the DNA bands of interest gel purified using the QIAEX II gel extraction kit (Qiagen), and cloned into the TOPO XL PCR cloning vector. TOP10 competent cells were transfected with the vectors containing the PCR products by electroporation and cultured in Luria agar plates with 100 μg/ml ampicillin for the selection of clones. Four clones for each TCR chain were selected and expanded in Luria broth supplemented with 100 μg/ml ampicillin. Plasmids containing the amplified sequences of interest were extracted from the bacteria using LPS-free Qiagen Plasmid Mini Kits and sequenced using M13 FW (–20) and M13 RV primers. The primers used to engineer the constructs using splicing by overlap extension (SOE) PCR are provided in the next section.

We engineered the constructs using the cloned TCR α- and β-chain cDNAs, primers designed to add the T2A self-cleavage sequence preceded by a Furin enzyme cleavage-recognition site in between the 3′ region of the TCR α chain and the 5′ region of the TCR β-chain, using the SOE PCR method.

The PCR products were run in a 1.5% agarose/Tris–borate–EDTA buffer gel; the DNA bands of interest were gel purified using the QIAEX II gel extraction kit (Qiagen) and cloned into TOPO XL PCR cloning vector. TOP10-competent cells were transfected with the vectors and cultured in Luria agar plates with 100 μg/ml ampicillin for the selection of clones. Four clones were expanded in Luria broth with 100 μg/ml ampicillin and sequenced. The plasmids amplified were extracted from the bacteria using LPS-free Qiagen Plasmid Mini Kits and sequenced using M13 FW (–20) and M13 RV primers. The BamH1 restriction enzyme sequence in the 5′ region and Sal1 restriction enzyme sequence in the 3′ region were PCR-added to the construct for directional cloning in the pBABEzeo amphotropic retroviral vector, which is derived from the Moloney Murine Leukemia Virus (Cell Biolabs, San Diego, CA). GigaSingles competent cells (Thermo Fisher Scientific, http://www.fishersci.com) were transfected with the vectors containing the PCR constructs and cultured in Luria agar plates with 100 μg/ml ampicillin for selection of the clones. Five clones were selected and expanded in Luria broth supplemented with 100 μg/ml ampicillin. Plasmids containing the amplified sequences of interest were extracted from the bacteria using LPS-free Qiagen Plasmid Mini Kits and sequenced using the following primers: TCRα1: 5′-ATG GCT TTG CAG AGC ACT CTG-3′, TCRα2: 5′-TCA GAA ATC CTT TCT CTT GAC-3′, Link1: 5′-GCC ATG GTC AAG AGA AAG GAT-3′, TCRβ1_1: 5′-ATG CTG CTG CTT CTG CTG CTT CTG-3′, TCRβ1_2: 5′-TCA GAA ATC CTT TCT CTT GAC CCA-3′. Thus, we call this product construct 1.

After sequencing verification of the TCRαβ construct, we engineered two more constructs. We first amplified the complete CD3ζ chain from an oligo(dT) reverse-transcribed total RNA obtained from CD3-pulled cells, PCR-amplified the CD3ζ cDNA, and cloned the gel-purified PCR product in a TOPO XL PCR cloning vector. We proceeded to the selection, expansion, and sequencing verification of the clones as explained above. Then, we PCR-removed the stop codon in the 3′ region of the TCR β-chain in the TCRαβ construct and in between this region and the 5′ region of the CD3ζ chain PCR-added a Furin enzyme recognition site and a sequence encoding the P2A self-cleaving peptide. The construct 2 was transferred to the pBABEzeo amphotropic retroviral vector, cloned and expanded in GigaSingles competent cells, and plasmids containing the amplified sequences of interest were extracted from the bacteria using LPS-free Qiagen Plasmid Mini Kits, as explained above. The construct was sequenced using the primers described above minus TCRβ1_2 and primers: TCRβ1_3: 5′-CTT CTC ACT AGG GGC GAT GTA-3′, Link2: 5′-GGT GAA AAG CGC CTT CCA CTT-3′, CD3_1: 5′-ATG AAG TGG AAG GCG CTT TTC ACC-3, CD3_2: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3. We call this product construct 2. After sequencing verification of the CD3ζ chain, we PCR-engineered a third construct producing first a CD3ζ extracellular/transmembrane domain followed by a CD28/41BB/CD3ζ chain signaling domain, and cloned the PCR product in a TOPO XL PCR cloning vector. We proceeded to selection, expansion, and sequencing verification of the clones as explained above. Then, we PCR-removed the stop codon in the 3′ region of the TCR β-chain in the TCRαβ construct and in between this region and the 5′ region of the CD3ζ/CD28/41BB/ζ-chain signaling domain PCR-added a Furin enzyme recognition site and a sequence encoding the P2A self-cleaving peptide. We cloned a gel-purified PCR product in a TOPO XL PCR cloning vector and proceeded to expansion and purification of vectors as explained above. Construct 2 was transferred to the pBABEzeo amphotropic retroviral vector, cloned, and expanded in GigaSingles competent cells, and plasmids containing the amplified sequences of interest were extracted from the bacteria using LPS-free Qiagen Plasmid Mini Kits, as explained above. The construct was sequenced using the primers described above for construct 2, and primers: J_1: 5′-AGG AGT AAG AGG AGC AGG CTC CTG-3, J_2: 5′-AAC GGG GCA GAA AGA AAC TCC TGT-3′, J_3: 5′-GGC CAC GTC TCT TGT CCA AAA-3′. We call this product construct 3.

To engineer the CD3ζ and CD3/CD28/4-1BB/CD3ζ constructs we proceeded as follows: we first amplified the CD3ζ chain using primers: CD3ζ FW: 5′-ATG AAG TGG AAG GCG CTT TTC ACC-3′ and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′. To add the P2A peptide we used the following primers: FW P2A_1: 5′-GGT GAC GTG GTT GTT AAT CCT GGT CCT ATG AAG TGG AAG GCG CTT TTC ACC-3′ (triplets coding GDVEENPGP peptide portion of P2A peptide in black) or FW P2A_2: 5′-GGC ACT AAT TTC TCG CTC CTC AAG CAA GGC GGT GAC GTG GTT GTT AAT CCT GGT CCT-3′ (triplets coding ATNFSLLKQAGDVEENPGP peptide portion of P2A peptide in black) and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′, using the product of the first amplification in the second amplification. The gel-purified final product was cloned in a TOPO XL vector and sequenced. We then proceeded to remove the stop codon from the TCRαβ construct with the following primers: FW1 TCRα: 5′-ATG AAA TCC TTG AGA GTT TTA-3′, RV1 TCRβ1: 5′-GAA ATC CTT TCT CTT GAC CCA-3′ and then added the overhang of the P2A-CD3 construct to the RV1 TCRβ1 for the SOE PCR with the following primers: FW1 TCRα: 5′-ATG AAA TCC TTG AGA GTT TTA-3′ and RV1 TCRβ1_OVERHANG: 5′-GAG GAG CGA GAA ATT AGT GCC GAA ATC CTT TCT CTT GAC CCA-3′. We then used the P2A-CD3 and the TCRαβ P2A overhang AMPLICONS in an SOE PCR using the following primers: FW1 TCRα: 5′-ATG AAA TCC TTG AGA GTT TTA-3′ and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′. We cloned the gel-purified product in a TOPO XL plasmid and PCR-sequenced the new construct.

To engineer a CD3/CD28/4-1BB/ζ construct to be added to the TCRαβ construct we used the following primers and amplifications: CD3ζ FW (extracellular and transmembrane domains): 5′-ATG AAG TGG AAG GCG CTT TTC ACC-3′ and CD3ζ RV2: 5′-CAG GAA CAA GGC AGT GAG AAT GAC-3’ or CD3ζ/CD28RV3: 5′- CAG GAG CCT GCT CCT CTT ACT CCT CAG GAA CAA GGC AGT GAG AAT GAC -3′ to add a CD28 signaling domain overhang to be used in an SOE PCR. The CD28 signaling domain was amplified using the following primers: CD28 FW1: 5′-AGG AGT AAG AGG AGC AGG CTC CTG-3′ and CD28 RV: 5′- GCG ATA GGC TGC GAA GTC GCG TGG-3′, and then the primer CD3ζ /CD28FW2: 5′-GTC ATT CTC ACT GCC TTG TTC CTG AGG AGT AAG AGG AGC AGG CTC CTG -3′ to add a CD3ζ overhang. We linked the human CD3 ζ-chain extracellular and transmembrane domains and the CD28 intracellular domain using SOE PCR and the following primers: CD3ζ FW: 5′-ATG AAG TGG AAG GCG CTT TTC ACC-3′ and CD28 RV: 5′- GCG ATA GGC TGC GAA GTC GCG TGG-3′. The 4-1BB signaling domain was amplified using the following primers: 4-1BB FW: 5′-AAC GGG GCA GAA AGA AAC TCC TGT-3′ and 4-1BB RV1: 5′-CAG TTC ACA TCC TCC TTC TTC TTC TTT-3′, and then an overhang for the CD3ζ signaling domain was added using the primer: 4-1BB RV2: 5′-CAG TTC ACA TCC TCC TTC TTC TTC TTT CTG CGC TCC TGCTGA ACT TCA CTC-3′. The CD3ζ signaling domain was amplified using the following primers: CD3ζ FW: 5′-GAG TGA AGT TCA GCA GGA GCG CAG-3′ and CD3ζ RV: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′, and an overhang for 4-1BB added with primer: 4-1BB/CD3ζ FW2:5′- AAC GGG GCA GAA AGA AAC TCC TGT GAG TGA AGT TCA GCA GGA GCG CAG-3′. To link the 4-1BB and CD3 ζ-chain signaling domains we used SOE PCR and the following primers: 4-1BB FW: 5′-AAC GGG GCA GAA AGA AAC TCC TGT-3′ and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′. To produce the CD3 extracellular/transmembrane/CD28/4-1BB/ ζ construct we added the 4-1BB overhang to the CD28 intracellular domain (first separate reaction): CD3ζ (extracellular and transmembrane domains) FW: 5′-ATG AAG TGG AAG GCG CTT TTC ACC-3′ and RV JUNCTION: 5′-ACA GGA GTT TCT TTC TGC CCC GTT GCG ATA GGC TGC GAA GTC GCG TGG-3′. To add the 4-1BB overhang to the CD3ζ intracellular domain (second separate reaction), we used the primers: FW JUNCTION: 5′-CCA CGC GAC TTC GCA GCC TAT CGC AAC GGG GCA GAA AGA AAC TCC TGT-3′ and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′. We then used an SOE PCR to join the amplicons of the first and second reaction using the following primers: CD3ζ FW: 5′-ATG AAG TGG AAG GCG CTT TTC ACC-3′ and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′. We finally proceed to add overhangs (in black) for the P2A peptide using primers: FW P2A_1: 5′-GGT GAC GTG GTT GTT AAT CCT GGT CCT ATG AAG TGG AAG GCG CTT TTC ACC-3′ (for the P2A portion GDVEENPGP in black) and then FW P2A_2: 5′-GGC ACT AAT TTC TCG CTC CTC AAG CAA GGC GGT GAC GTG GTT GTT AAT CCT GGT CCT-3 (for the P2A portion ATNFSLLKQAGDVEENPGP in black) with primer: CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′ in two sequential reactions. The gel-purified PCR product was cloned in a TOPO XL vector and the construct sequenced. We then proceeded to remove the stop codon from the TCRαβ construct with the following primers: FW1 TCRα: 5′-ATG AAA TCC TTG AGA GTT TTA-3′ and RV1 TCRβ1: 5′-GAA ATC CTT TCT CTT GAC CCA-3′, and then added the overhang of the P2A-CD3/CD28/4-1BB/ζ construct to the RV1 TCRβ1 for the SOE PCR with the following primers: FW1 TCRα: 5′-ATG AAA TCC TTG AGA GTT TTA-3′ and RV1 TCRβ1_OVERHANG in black: 5′-GAG GAG CGA GAA ATT AGT GCC GAA ATC CTT TCT CTT GAC CCA-3′. We then used the P2A-CD3/CD28/4-1BB/ζ and the TCRαβ P2A overhang AMPLICONS in an SOE PCR using the following primers: FW1 TCRα: 5′-ATG AAA TCC TTG AGA GTT TTA-3′ and CD3ζ RV1: 5′-CGA CGA GGG GGC AGG GCC TGC ATG-3′. We cloned the gel-purified product in a TOPO XL vector and sequenced the new construct.

Amphotropic retroviral vector pBABEzeo containing constructs was transduced into PLAT-A cells (Cell Biolabs) with FUGENE 6 transfection reagent (Roche Applied Science) in Life Technologies Opti-MEM I reduced serum media. The transduced cells were grown in DMEM (Life Technologies) containing Life Technologies’ 10% heat-inactivated FBS, 4 mM l-glutamine. The medium was replaced 16 h after transduction with fresh medium containing 100 U/ml of penicillin and 100 μg/ml of streptomycin (Life Technologies). After 24 h, virus particles in the media were collected by centrifugation overnight at 8000 rpm and 4°C. The pellet was resuspended in RPMI 1640 medium. Viral titer was determined using a QuickTiter retrovirus quantitation kit (Cell Biolabs). A total of 1 × 10⁷ purified CD8⁺ T cells from HLA-A*02:01 healthy donors of PBMCs were activated with anti-CD3/CD28 Dynabeads at a ratio of three Dynabeads per CD8⁺ T cell, in the presence of 5 × 10⁷ CD3-depleted autologous PBMCs (irradiated 3000 rad). Activated CD8⁺ T cells were then transduced with 5 × 10¹⁰ retrovirus particles using the ViraDucting Retrovirus transduction kit and protocols provided (Cell Biolabs). Cultures were left for 10 d in six-well Costar tissue culture wells. After 72 h, the medium was replaced with fresh complete RPMI 1640 medium containing 10 ng/ml hrIL-2 and 200 μg/ml Zeocin (InvivoGen, San Diego, CA). Then, spent medium was replaced every 72 h with medium containing 200 μg/ml zeocin for the selection of stable cell clones. Live cells were then obtained by Ficoll-gradient centrifugation and restimulated in 24-well Costar plates at 1 × 10⁶ stable transduced cells per well with 5 × 10⁶ CD3 cell–depleted autologous PBMCs irradiated 3000 rad and pulsed with 100 ng/ml of CT37 peptide YLCSGSSYF, in 2 ml of complete RPMI 1640 containing 100 U/ml of penicillin and 100 μg/ml of streptomycin. IL-2 (10 ng/ml) was added 72 h later and 1 ml per well of spent media replaced every 72 h with fresh complete RPMI 1640 containing 100 U/ml of penicillin and 100 μg/ml of streptomycin. Cultures were left for 10 d.

Cell-surface proteins from stable retroviral transduced CD8⁺ T cells were biotinylated and isolated for immunoprecipitation and Western blot analysis using the Thermo Fisher Scientific Pierce Cell Surface Protein Isolation kit, anti-TCRβ Ab clone 8A3 (Novus Biologicals, Littleton, CO) for the immunoprecipitation of the TCR complex, SDS-PAGE, transfer to membranes, and the Immun-Star HRP kit for detection of blots, respectively. Surface membrane proteins from 1 × 10⁶ stable transduced cell were biotinylated with EZ-Link sulfosuccinimidyl-2-(biotinamido) ethyl-1, 3-dithioprpionate, a thiol-cleavage amine-reactive biotinylating reagent, and then lysed with a mild detergent (M-PER; Thermo Fisher Scientific) in the presence of Halt Protease Inhibitor Cocktail. The biotinylated surface proteins were then isolated using Sera-Mag Neutravidin-Coated Magnetic particles (GE Healthcare Bio-Sciences, Pittsburgh, PA) and resuspended in PBS with 0.01% Tween 20. The anti-TCRβ Ab clone 8A3/Dynabeads-Protein G complex was added at 1:100 final concentration and the biotinylated proteins immunoprecipitated using the Immunoprecipitation Kit Dynabeads-Protein G. Biotinylated surface membrane TCR-complex proteins were eluted from the magnetic particles using the magnet and 30 μl SDS sample buffer (62.5 mM Tris-HCL, pH 6.8, 1% SDS, 10% glycerol). The eluted proteins were separated in 10% SDS-PAGE and transferred to a polyvinylidene difluoride membrane using the Bio-Rad Trans-Blot system. Membranes were probed with 1:1000 HRP-conjugated anti-biotin Ab ab1921 (Abcam, http://www.abcam.com), the Immun-Star HRP kit, and chemiluminescence detected using a VersaDoc Imaging system (Bio-Rad Laboratories, Hercules, CA). To determine the identity of CD3 chains, immunoprecipitated proteins obtained as indicated above from cells transfected with construct 3 were blotted using rabbit anti-CD3δ EP4426, anti-CD3γ 134684, anti-CD3ε 133628, and Armenian hamster anti-CD3ζ H136-968 Abs (Abcam), 1:2000 HRP-conjugated goat anti-rabbit IgG or 1:000 HPR-conjugated rabbit anti-Armenian hamster IgG (Abcam), and the Immun-Star HRP kit, and blots were revealed by chemiluminescence using autoradiography Kodak BioMax Light film.

CD8⁺ T cells cultured as described above were harvested and washed twice in cold incomplete RPMI 1640, then resuspended in 100 μl of cold incomplete RPMI 1640 containing 10 μg ml⁻¹ of human IgG (Sigma-Aldrich), and incubated on ice for 15 min. Cells were washed once in PBS+10% BSA (Sigma-Aldrich) and resuspended at 5 × 10⁵ cells per 100 μl of this buffer. FITC–anti-human CD8 mouse mAb HIT8a (BD Biosciences Pharmingen, bdbiosciences.com) and PE-labeled HLA-A*02:01/CT37 YLCSGSSYFV peptide pentamers or negative control PE-labeled HLA-A*02:01/NY-ESO-1 SLLMWITQV peptide pentamers (ProImmune) were added to the cells and incubated for 30 min on ice. Cells were then washed thoroughly and resuspended in cold PBS. Acquisition was done using the BD LSR II instrument and software. We acquired 100 000 events. Data were analyzed using FlowJo 7.6.4 software.

The ability of rested transduced CD8⁺ T cells (effector cells) to lyse HLA-A*02:01 autologous CD3-depleted PBMCs pulsed with 100 μg/ml of CT37 YLCSGSSYFV peptide or control CT37 YYLCSGSSYF peptide, or the HLA-A*02:01-positive HCC2935, H1993, and H522, and HLA-A*02:01-negative H1299, H23, and H460 lung ADC cell lines (target cells), was measured in [⁵¹Cr] release assays. Briefly, 1 × 10⁶ target cells were labeled for 1 h at 37°C with 200 mCi of Na₂⁵¹CrO₄ (Amersham, Arlington Heights, IL). Labeled target cells at 5 × 10³ were incubated with effector cells at E:T ratios of 1:1, 4:1, and 16:1, for 4 h at 37°C in 200 μl of complete medium. Supernatants were harvested and counted using a 1470 wizard 3″ automatic γ counter (PerkinElmer, Gaithersburg, MD). Maximum and spontaneous [⁵¹Cr] release by each target was determined by incubating 5 × 10³ labeled target cells in 2% SDS or medium, respectively, for 4 h at 37°C. Each point represented the average of triplicate wells, and percentage of specific lysis was calculated as follows: [(specific [⁵¹Cr] release − spontaneous [⁵¹Cr] release)/(maximum [⁵¹Cr] release − spontaneous [⁵¹Cr] release)] × 100. The highly purified and mass spectrometry verified peptides YLCSGSSYFV and YYLCSGSSYF were obtained from Creative Peptides. Peptides were dissolved in DMSO and diluted in incomplete RPMI 1640. We added the HLA-A2 mouse mAb BB7.2 (Novus Biologicals) to the cultures to determine whether the lysis activity was HLA-A*02:01 restricted at 1, 5, and 10 μg/ml.

Similar cold experiments were set up simultaneously. Supernatants were harvested from E:T cocultures that were left for 4 h and tested for IFN-γ secretion using high-sensitivity ELISA kits (Thermo Fisher Scientific, http://www.thermofisher.com). ELISPOT assays were conducted using the Human IFN-γ kit from Invitrogen (Thermo Fisher Scientific). Plates were conditioned according to protocols furnished by the manufacturer. In total, 8 × 10⁴ transduced CD8⁺ T cells were stimulated with 3000 rad irradiated 5 × 10³ autologous CD3-depleted PBMCs pulsed with 100 μg/ml of CT37 peptide YLCSGSSYFV or control CT37 peptide YYLCSGSSYF. Unstimulated and PHA-stimulated cells were used as negative and positive controls, respectively. Cultures were placed in 100 μl of complete RMPI 1640 per well in triplicate wells at 37°C with 5% CO₂ for 16 h. After decanting cells and medium from the wells, we proceeded according to the instructions provided. The plates were air dried, and the spots counted using a stereomicroscope. Triplicate counts were averaged to give the final count.

We used 30 pmol (final concentration) of each small interfering RNAs (siRNAs) specifically designed to silence CD8A-chain or CT37, or negative control siRNA (Thermo Fisher Scientific). This concentration of siRNA was used for 1 × 10⁶ cells plated in six-well Costar tissue culture wells, using Lipofectamine RNAiMAX reagent, following protocols furnished with the reagent (Thermo Fisher Scientific). To silence the CD8A chain, we targeted positions 1653 in exon 5 and position 1906 in exon 9. To silence CT37, we targeted position 796 in exon 5. We used FACS analysis or immunoprecipitation to assess the efficiency of silencing protocols.

The equality of more than two means or the equality of two means was tested using the Kruskal–Wallis test or the Wilcoxon rank-sum (Mann–Whitney) test, respectively. Means, SDs, and results were calculated using Stata 11 software.

Three CT Ags, CT37, VCX3a, and XAGE1b, were selected as having protein expression restricted to testis/ovary and expressed in NSCLC (http://www.proteinatlas.org, GeneCards, 7-11). CT37 was the only one with strong binding peptides restricted by the most frequent MHC class I molecules in several populations including HLA-A*02:01, HLA-A*24:02, and HLA-B*07:02 (Table I, allelefrequencies.net). Peptide specificities were determined using the NetMHCpan Server from the Technical University of Denmark (http://www.cbs.dtu.dk/services/NetMHCpan/). We selected peptides with a percentage binding rate cut-off (percentage rate) <0.1 to increase the likelihood of a peptide binding the selected MHC class I molecules in vivo. We selected the HLA-A*02:01–restricted peptide YLCSGSSYFV because it has a core amino acid sequence like that of peptide YYLCSGSSYF. This later peptide YYLCSGSSYF has a percentage rate <0.1 for the HLA-A*24:02, but not for HLA-A*02:01 (Table I). Thus, we decided to use those peptides for our studies because one is an excellent control for the other.

Patients recruited were of Hispanic ancestry. From the case review, 43 cases met the criteria and 9 were carriers of HLA-A*02:01 alleles and expressed CT37 at the mRNA level (Table II). Four of those nine cases expressed CT37 by IHC (∼44%) at the levels indicated (Table II). In total, 34 were HLA-A*02:01 negative; of them, 12 expressed CT37 at the mRNA and protein level by IHC (35.29%). Thus, 21 out of 43 cases expressed CT37 mRNA and were positive by IHC (∼48.83% of 43 cases) (data not shown). From those 21, 5 expressed moderate levels (23.8%) and 16 low levels (76.2%) of CT37 by IHC. However, we understand that a larger sample size is needed to address the issue more clearly.

Patient NSCLC_7, who was willing to donate leukocytes, had tumor cells expressing moderate levels of the CT37 Ag by IHC. CT37 expression was clearly observed in the cytoplasm and cell membrane of malignant cells (Fig. 1A). In addition, ∼8% of their circulating CD8⁺ T cells were carrying a TCR specific for HLA-A*02:01/CT37 peptide YLCSGSSYFV, as determined by FACS analysis (Fig. 1B). This patient was a former smoker of more than one pack per d of tobacco cigarettes (20 cigarettes/pack), with a computerized tomography scan showing a solid solitary ground-glass nodule of 2.8 cm located in the upper right pulmonary lobe of the lung. This patient had a stage 1 disease (T1bN0M0), with lung ADC carrying an EGFR driver mutation in exon 21 (L858R, Table II). This patient remains cancer free 6 y after lobectomy (at the article submission date). Two more patients, NSCLC_5 and NSCLC_9, were also willing to donate leukocytes, had tumor cells expressing low levels of CT37 Ag by IHC analysis, and had ∼3 and 4% circulating CD8⁺ T cells, respectively, but were within 5 y of follow-up of surgery.

Because weak IHC staining in healthy kidney and urinary bladder tissues is mentioned in the Human Protein Atlas using one of two Abs, in this case using the rabbit anti-human CT37 polyclonal IgG purified Ab HPA011284 (http://www.proteinatlas.org), we double-checked this in de novo processed frozen tissues from five randomly selected necropsy samples, five kidneys, and five urinary bladders. We observed only background staining like that in negative controls exposed to rabbit polyclonal isotype Ab control (Supplemental Fig. 1).

Given that the patient NSCLC_7 was freely willing to donate leukocytes, we decided to isolate their Ag-specific CD8⁺ T cells. After two in vitro stimulations we obtained a 98% pure population of polyclonal CD8⁺ T cells expressing a TCR restricted by HLA-A*02:01 MHC class I/CT37 peptide YLCSGSSYFV complexes (Supplemental Fig. 2A). Using the limiting dilution method, we were able to isolate two cell clones from 300 HLA-A*02:01/CT37 YLCSGSSYFV peptide-specific polyclonal CD8⁺ T cells. One clone was expanded, reaching a final number of 1.6 × 10⁶. We amplified, molecularly cloned, expanded, and sequenced the cDNA encoding the TCR α- and β-chains from that CD8⁺ T cell clone. All plasmids sequenced had the same TCR α or the same TCR β-chains cDNA sequences and they were in frame. The TCR α- and TCR β-chains were 813 and 909 nucleotides long (Supplemental Fig. 2B), and their deduced V and J rearrangements were TRAV2S1_AJ24S1 and TRBV22S1_BJ2S1, respectively, indicating monoclonality (Supplemental Fig. 2C, 2D).

The features of the constructs produced are depicted in Fig. 2A. A typical retroviral transduction efficiency of 60% is shown in the Supplemental Fig. 2. In vitro expansion of stable transduced cells yielded 93% pure CD8⁺ T cells (Supplemental Fig. 2E). Using FACS analysis we observed high expression of our TCRαβ constructs on transduced CD8⁺ T cells (Fig. 2B). The expression of TCRαβ constructs was higher for constructs 2 and 3, carrying the CD3ζ chain or the engineered CD3ζ chain (Fig. 2B). We immunoprecipitated the TCRαβ complexes from transduced CD8⁺ T cells whose membrane proteins were biotinylated, and observed that all TCRαβ constructs coprecipitated with CD3δ, CD3ε, CD3γ, and CD3ζ chains, indicating that the engineering strategy used did not hamper the ability of these chains to polymerize in vivo (Fig. 2C). Western blot analysis with specific Abs corroborated the identity of those CD3 chains (Fig. 2D).

To determine the functional avidity of the TCR cloned for HLA-A*02:01/CT37 peptide YLCSGSSYF complexes, we first calculated the peptide concentration needed to obtain 50% of maximal lysis (EC₅₀) using a serial dilution of the peptide YLCSGSSYFV (Fig. 3A, left) and the [⁵¹Cr]-release cytotoxic assay as explained in 2Materials and Methods. We then proceeded to extrapolate the HLA-A*02:01–specific peptide dissociation constant value (K_off), using existing plots of EC₅₀ values obtained from the interactions of human CD8⁺ T cell clones specific for HLA-A*02:01/NY-ESO-1 SLLMWITQC peptide or the antimelanoma CD8⁺ T cell clone recognizing HLA-A*02:01/Melan-A^MART-1 ELAGIGILTV peptide versus monomeric TCR/HLA-A*02:01 peptide complexes K_off values obtained from dissociation kinetic measurements (12). Finding the K_off is key to determining functional avidity because a lower K_off is correlated with higher functional avidity for HLA-A*02:01–specific peptide (higher activation and calcium mobilization) (12).

To experimentally obtain EC₅₀ values, we used a 16:1 E:T ratio of CD8⁺ T cells isolated from six HLA-A*02:01–positive healthy donors, which were transduced with construct 1, and their autologous CD3-cells depleted PBMCs pulsed with indicated concentrations of the CT37 YYLCSGSSYF peptide (Fig. 3A, left). The mean values obtained from the six experiments were used to calculate the EC₅₀. The EC₅₀ was extrapolated using a plot of mean percentage specific lysis versus peptide concentration used (Fig. 3A, right). To find the EC₅₀ we used y = 49.85. We converted nanograms per milliliter into moles per liter based on the m.w. of CT37 peptide YLCSGSSYF (m.w. = 1125.28). The extrapolated EC₅₀ value was 0.005 ng/ml (4.44 × 10⁻¹² mol/l). The K_off values extrapolated for this EC₅₀ are <1 × 10⁻¹ s⁻¹, placing our TCR among those elite clones exhibiting the lowest K_off values and highest functional avidity (12).

Given the fact that TCRs exhibiting such dissociation constants can stably interact with HLA-specific peptide complexes in the absence of CD8 complexes, we silenced the expression of the CD8A-chain in an aliquot of our CD8⁺ cell clone NSCLC_7 (from which we isolated our TCR). We used FACS analysis to detect any change in the percentage of cells stained with HLA-A*02:01/CT37 YYLCSGSSYF peptide complexes (Fig. 3B). We did not observe a significant change in the percentage of cells stained with HLA-A*02:01/CT37 YYLCSGSSYF peptide complexes between CD8A silenced and control unsilenced NSCLC_7 CD8⁺ T cell clone (Fig. 3B). Thus, we concluded that our cloned TCR exhibit a high functional avidity for HLA-A*02:01/CT37 YYLCSGSSYF peptide complexes.

To determine whether expressed TCRαβ constructs were functional and highly restricted by the appropriate HLA/peptide complexes, we tested whether transduced CD8⁺ T cells were able to lyse HLA-A*02:01 autologous cells pulsed with CT37 peptide YLCSGSSYFV or control peptide using [⁵¹Cr] release assays. We observed that CD8⁺ T cells transduced with all our constructs were able to lyse autologous targets pulsed with CT37 peptide YLCSGSSYFV, but not those pulsed with control CT37-YYLCSGSSYF peptide (Fig. 4A). At E:T ratios of 4:1 and 16:1, the lysis activities of CD8⁺ T cells transduced with constructs 2 and 3 were significantly higher than those of cells transduced with construct 1, with the highest being that of the CD8⁺ T cells transduced with construct 3 (Fig. 4A). Blocking the HLA-A*02:01 molecule with anti–HLA-A2 Ab abrogated the lysis activity of CD8⁺ T cells transduced with construct 1, 2, and 3 in a dose-dependent manner (Fig. 4B).

Both the levels of IFN-γ secretion and the frequency of IFN-γ–secreting CD8⁺ T cells were significantly higher in CD8⁺ T cells transduced with constructs 2 or 3, as compared with CD8⁺ T cells transduced with construct 1, when exposed to autologous cells pulsed with CT37 peptide YLCSGSSYFV (Fig. 4C, 4D), with the highest observed for CD8⁺ T cells transduced with construct 3. We could not detect IFN-γ secretion or frequency of IFN-γ–secreting CD8⁺ T cells higher than background in cultures where PBMCs were pulsed with control CT37 YYLCSGSSYF peptide (data not shown). Notably, we also observed in silico that both the CT37 HLA-A*02:01 peptide YLCSGSSYFV and the CT37 HLA-A*24:02 peptide YYLCSGSSYF are predicted to be produced by the proteasome (Supplemental Fig. 4). Nontransduced CD8⁺ T cells did not secrete IFN-γ when exposed to target cells pulsed with appropriate peptide. HLA-A*02:01 restriction was evident in cultures where PBMCs were pulsed with CT37 YLCSGSSYFV peptide but processed in the presence of an anti–HLA-A02 Ab. The anti–HLA-A02 Ab blocked IFN-γ secretion in a dose-dependent manner (Kruskal–Wallis tests for equality of IFN-γ secretion levels in cultures containing 0, 1, 5, or 10 μg/ml of anti–HLA-A02 Ab: p = 0.002 for construct 1, p = 0.001 for construct 2, and p = 0.001 for construct 3; Fig. 4C).

We next examined whether CD8⁺ T cells transduced with our constructs will exhibit the same functions when exposed to lung ADC cell lines. We genotyped available (American Type Culture Collection) lung ADC cell lines for detection of those carrying the HLA-A*02:01 allele. We first tested the HLA-A*02:01–positive lines HCC2935 and H1299, and the HLA-A*02:01–negative line H1299 as control. HCC2935 was the only line carrying one of the cancer-driver mutations tested. The line carries an allele with an EGFR deletion of 15 nucleotides in exon 19 (ΔE746-A750) and a wild-type EGFR allele. Regardless of the cancer-driver mutation that these lines carried, we observed that all of them expressed CT37 mRNA and protein (Fig. 5Ai, ii).

CD8⁺ T cells transduced with construct 3 significantly lysed HCC2935 and H1993 in a HLA-A*02:01–restricted fashion, as the anti–HLA-A2 Ab abrogated this lytic activity (Fig. 5Bi, ii). Remarkably, CD8⁺ T cells transduced with any of our constructs were unable to lyse HLA-A*02:01–negative H1299 line (Fig. 5Biii). We also observed that CD8⁺ T cells transduced with construct 3 secreted the highest levels of IFN-γ when cultured in the presence of the HLA-A*0201–positive HCC2935 and H1993 cell lines, but not when cultured in the presence of the HLA-A*02:01–negative H1299 line. Notably, the supernatants tested were obtained from cultures that were left for the 4 h incubation period and under conditions like those used to perform the [⁵¹Cr] release assay (Fig. 6A). Only background levels of IFN-γ were observed when CD8⁺ T cells transduced with any of the three constructs were exposed to the H1299 line (Fig. 6A). In concordance, CD8⁺ T cells transduced with construct 3 showed the highest frequency of IFN-γ–secreting cells (Fig. 6B). Of note, we did not see significant differences in the frequency of IFN-γ–secreting cells between CD8⁺ T cells transduced with construct 1 and those transduced with construct 2 when exposed to HLA-A*02:01–positive HCC2935 and H1993 cell lines (Fig. 6B). Only the background frequency of cells was observed when CD8⁺ T cells transduced with any of the three constructs were exposed to the HLA-A*02:01–negative H1299 line (Fig. 6B). Nontransduced CD8⁺ T cells did not lyse or secrete IFN-γ when exposed to target lung ADC cells. We conducted these assays with an additional set of CT37-expressing and HLA-A*02:01–positive H522 and HLA*02:01-negative H23 and H460 lung ADC cell lines, and obtained similar results (Supplemental Fig. 3).

To ensure Ag specificity in CTL activity against HLA-A*02:01/CT37 peptide complexes, we silenced the expression of CT37 in CT37-expressing and HLA-A*02:01–positive cell lines HCC2935, H1993, and H522, and tested the effect of this using CD8⁺ T cells transduced with construct 1 in [⁵¹Cr] release assays (Fig. 7). Silencing of CT37 expression abrogated CTL activity.

Our studies in patients suffering from lung ADC combined with our ex vivo and in vitro experiments led us to the identification and isolation of a CT37 peptide-specific CD8⁺ T cell clone, and the molecular cloning of unique TCR α- and β-chains from this clone through: 1) the ex vivo analysis of tissues using CT37 IHC; 2) the in-silico identification of the CT37 YLCSGSSYFV peptide followed by the identification of a carrier of relatively high percentage of polyclonal CD8⁺ T cells restricted by HLA-A*02:01/CT37 peptide complexes in peripheral blood; and 3) our in vitro model using autologous CD3-depleted PBMCs. All TCR α- and β-chains isolated from that cellular clone were TRAV2S1_AJ24S1 and TRBV22S1_BJ2S1, respectively, supporting the fact that we obtained a CD8⁺ T cell clone. This TCR α and β complex exhibited high functional avidity for HLA-A*02:01/CT37 YLCSGSSYFV–peptide complexes.

The isolation of those unique TCR α- and β-chains allowed us to engineer constructs that when transferred to allogeneic CD8⁺ T cells redirect them to express those TCR chains and lysis, with high specificity and functional avidity, CD3-depleted autologous PBMCs pulsed with the CT37 YLCSGSSYFV peptide, but not those PBMCs pulsed with control peptide. Moreover, TCR-transfected CD8⁺ T cells lysed and secreted IFN-γ when exposed only to HLA-A*02:01–positive lung ADC cell lines expressing CT37 Ag, but not HLA-A mismatched or HLA-A*02:01–positive lung ADC cell lines with a silenced CT37 Ag. Of note, IFN-γ is a key stimulator of the proteasome and MHC class I molecule expression (13–15).

As mentioned, CT37 was the only CT Ag among those selected that had in silico strong binding peptides restricted by the most frequent MHC class I molecules including: HLA-A*02:01, HLA-A*24:02, and HLA-B*07:02 alleles (Table I, allelefrequencies.net). Moreover, the CT37 encoding gene is in the X chromosome (Xq27.3-q28 chromosome region; http://www.genecards.org), where most CT Ags are located (http://www.cta.Incc.br). Our study also demonstrates that under appropriate parameters, in silico approaches are useful to identify peptides and MHC class I specificities suitable for development of adoptive transfer preclinical platforms. Stringent criteria for selection of peptides (a percentage rate <0.1), may have increased the likelihood of finding a peptide most likely binding in vivo the selected MHC class I molecule.

The fact that patient NSCLC_7 had clinical features of a long-term cancer-free survivor after appropriate clinical intervention, and coincidentally had relatively large percentages of circulating CD8⁺ T cells carrying TCRs recognizing HLA-A*02:01/CT37 Ag–peptide complexes may prompt the identification of similar cases for isolation of unique TCRs and for the development of timely immunotherapy interventions.

Weak IHC staining in healthy kidney and urinary bladder tissues is mentioned in the Human Protein Atlas, using one of two Abs, in this case using the rabbit anti-human CT37 polyclonal IgG purified Ab HPA011284 (http://www.proteinatlas.org). As mentioned previously, we did not observe this in healthy tissue regions of de novo processed frozen tissues from five randomly selected cancer cases, five kidneys, and five urinary bladders. We observed only background staining in healthy tissue regions like that observed in negative controls treated with a rabbit polyclonal isotype Ab (see the Supplemental Fig. 1). In addition, when testing lung ADC cell lines for CT37 protein expression we lysed these cells and immunoprecipitated the protein lysate using the HPA011284 (binding CT37 epitopes in the region comprising aa 98–187) followed by Western blot analysis using a rabbit polyclonal AP51690PU-N (binding CT37 epitopes in the region comprising aa 1–130) and observed a strong specific band of ∼29 kDa, and a weaker band of ∼50 kDa. Because these Abs are targeted against highly specific CT37 amino acid sequences as we determined it running a blast search for proteins carrying similar amino acid sequences (blast.ncbi.nlm.nih.gov/Blast.cgi), a possible explanation is that the band of ∼50 kDa is a dimer of CT37 molecules. Arguing against this, CT37 is predicted to have a molecular mass of 29.24 (web.expasy.org). Thus, those polyclonal Abs may harbor an Ab recognizing a nonlinear (conformational) epitope exhibiting a nonspecific cross-reactivity. This feasible explanation may explain the weak IHC staining in healthy kidney and urinary bladder tissues mentioned in the Human Protein Atlas when using the HPA011284 polyclonal rabbit Ab.

Our study agrees with the concept that some CT Ags are superb targets for the development of immune-cell–based immunotherapies (7–11). Indeed, protein expression of CT37 Ag is likely restricted to testis/ovary and malignant cells, as determined by proteomics (CT37 GeneCard), and by IHC (this study). In addition, lung ADC cells express CT37 [(8); this study]. However, as these features are key to decreasing the chance of a patient suffering on-target off-tumor deleterious lysis of healthy tissue, we will conduct more extensive studies (with a larger sample of normal tissues) to corroborate our findings. We think that the production and validation of mAbs specific against CT37 will be necessary to conduct those studies.

Taken together, our IHC results, the ex vivo pulling of CD8⁺ T cells recognizing HLA-A*02:01/CT37 YLCSGSSYFV complexes from patient NSCLC_7, the isolation of the NSCLC_7 CD8⁺ T cell clone bearing a αβTCR complex with high specificity, and high functional avidity for HLA-A*02:01/CT37 YLCSGSSYFV complexes in vitro, we think that our cloned TCR α- and β-chains and constructs are suitable for preclinical in vivo assessment of their antitumor activity, using animal models. If our adoptive transfer studies in animal models proves significant efficacy and we seek to move toward first-in-human studies, we must thoroughly investigate CT37 expression in normal human tissues to avoid potential hazards in targeting this Ag in human studies. We understand that species difference could conceal normal tissue toxicities that might occur in human patients.

We used the self-cleavage T2A polypeptide linkers in between the TCR α- and β-chains to increase the chance that these two chains will be dimerized among them. This may decrease the likelihood of creating new and potentially deleterious specificities by dimerization with native TCR α- and β-chains (16). Of the three constructs produced, construct 3 is the most promising because CD8⁺ T cells transduced with this construct exhibited the highest lytic, IFN-γ secretion, and frequency of IFN-γ–secreting cells. Indeed, the addition of the engineered CD3 chain containing the CD28/4-1BB/CD3ζ signaling domains to the TCR α and β construct (construct 3), conferred the highest lysis and IFN-γ secretion capacity to CD8⁺ T cells exposed to HLA-A*02:01–positive lung ADC lines expressing CT37 Ag. This indicates that TCR signaling should have profited by addition of the engineered CD3ζ chain containing the costimulatory CD28 and 4-1BB fragments (17, 18). We also observed that those levels of IFN-γ increased the expression of HLA-A*02:01 and of key immunoproteasome molecules (Supplemental Fig. 4).

To engineer constructs 2 and 3, we used a self-cleavage P2A polypeptide linker in between the TCRαβ construct and the CD3ζ chain to induce the expression of similar relative quantities of TCRαβ and CD3ζ chains (16), avoiding the ubiquitination and destruction of the TCR α- and β-chains that do not find a CD3 chain available for polymerization (19). The molecular engineering strategies used did not hamper the ability of those chains to polymerize among them and with host cells encoded CD3δ, CD3ε, and CD3γ chains, nor did they affect the functions of the TCRαβ/CD3 complexes. Notably, CD8⁺ T cells transduced with construct 3 were able to express the highest lysis and IFN-γ secretion capacities as early as 4 h after exposure to lung ADC cell lines, and in a HLA-A*02:01–restricted fashion. We considered it likely that those HLA-A*02:01 molecules were loaded with the CT37 peptide YLCSGSSYFV because it is among the peptides predicted to be generated by the proteasome, as we determined in silico. This is encouraging because some lung ADC tumors escape immune surveillance by downmodulating the proteasome activity and expression of MHC class I molecules, and IFN-γ is known to enhance both tumor cell proteasome activity and expression of MHC class I molecules (13–15). MHC class I restriction is highly specific and sensitive to minor changes in the sequences of peptides as we could see with two CT37 peptides containing similar core sequences YLCSGSSYFV and YYLCSGSSYF but complexing with HLA-A*02:01 and HLA-A*24:02, respectively, one of them specifically distinguished by our TCRαβ constructs.

We transduced purified CD8⁺ T cells to test primarily CTL activity and IFN-γ secretion by this subset of cells, and secondarily to avoid providing HLA-A*02:01/CT37–peptide specificities to Foxp3⁺ CD4⁺ T cell regulatory cells or RORγt⁺ CD4⁺ TH17 cells. The latter subsets do not carry cell-membrane specific markers, so cannot be physically excluded from the pool of CD4⁺ T cells. Their presence among transduced populations could have confounded our analysis. Moreover, current reports show deleterious effects of these T cell populations, interfering with the immune response or promoting the proliferation of cancerous cells, respectively (20–29).

We conclude that our findings will allow the launch of the development of a platform to conduct preclinical studies assessing and comparing in vivo whether the adoptive transfer of CD8⁺ T cells transduced with our constructs will be able to recognize lung ADC cells expressing the HLA-A*02:01/CT37–Ag complexes, to be activated, and to express efficient CTL-activity and IFN-γ production, as they did in vitro. In addition, if in vivo studies in animal models are successful, first-in-human studies will not be conducted until CT37 expression in normal human tissues is thoroughly ruled out. This will help avoid the potential hazards of targeting this Ag in human studies.

We thank patients and healthy donors for kind cooperation. We are also grateful to the Methodist Hospital for the recruitment funds provided to P.O.F.-V.

The authors are filing a patent covering their rights on the structure of molecular constructs, nucleotide and amino acid sequences of these constructs, and on their potential clinical applications.