Antigen Experienced T Cells from Peripheral Blood Recognize p53 Neoantigens

This article is featured in Highlights of This Issue, p. 1201

Adoptive cell therapy (ACT) using autologous tumor-infiltrating lymphocytes (TIL) mediated durable, complete cancer regressions in patients with melanoma, breast, colon, cervical, and bile duct cancers (1–6). Collectively, these responses were likely based on recognition of unique, patient-specific mutated neoantigens through the T-cell receptor (TCR; refs. 3–5, 7). TP53 is the most frequently mutated gene across all cancers and encodes the tumor suppressor p53 protein (8). Approximately 30% of TP53 mutations are shared “hotspots” in unrelated individuals (9). However, TP53-targeted therapies have not demonstrated efficacy beyond in vitro models and mutant TP53 immunotherapies are not currently available (9–11). We previously evaluated the immunogenicity of the 12 most common hotspot TP53 mutations according to the Catalog of Somatic Mutations in Cancer (COSMIC) database by measuring T-cell responses of autologous TILs. Approximately 1 in 4 of our patients with metastatic epithelial cancers seen in our clinic expressed one of these twelve TP53 mutations, and 40% of patients expressing a TP53 hotspot mutation had TIL recognizing an autologous p53 neoepitope (12–14). Thus, TP53 appears to be immunogenic when mutated.

Whether similar T-cell responses to p53 neoantigens exist in the peripheral blood T-cell repertoire remains largely unknown. Preliminary studies showed evidence of p53 neoantigen responses in peripheral blood lymphocytes (PBL) using either in vivo peptide vaccination in a small number of patients (15) or in vitro stimulation (IVS) with predicted a p53 peptide in a patient with squamous cell carcinoma of the head and neck (16). Major advantages of IVS are that it can increase antigen-specific precursor pools from PBLs for research or therapy and is agnostic to the human leukocyte antigen (HLA) haplotype of the patient, obviating the need for identifying candidate epitopes using HLA prediction algorithms (17–19). This approach has been leveraged for clinical translation using IVS of bulk PBL with Wilms tumor-1 (WT-1) and NY-ESO-1 cancer germline antigens (20, 21). It has been demonstrated that the naïve T cells (CD62L⁺CD45RO⁻), which are present within the bulk PBL, can have depressed effector function and proliferation, low avidity TCRs, and were unlikely to have previously experienced naturally occurring processed and presented peptides in vivo relative to antigen-experienced T cells (effector memory: CD62L⁻CD45RO⁺, central memory: CD62L⁺CD45RO⁺, and effector: CD62L⁻CD45RO⁻; refs. 22, 23). The antigen-experienced T-cell populations were recently shown to contain all or most of the mutated neoantigen-reactive T cells in PBLs, including private neoantigens and shared KRAS mutations, following IVS with patient-specific tandem minigenes (TMG) and peptide pools (24). Thus, we hypothesized that antigen-experienced subsets of PBLs would contain TP53 mutation-reactive T cells in the circulation of patients with known intratumoral T-cell responses to TP53.

To test this hypothesis, antigen-experienced T cells from PBLs of patients with metastatic epithelial cancers expressing a TP53 hotspot mutation and TIL screening results were subjected to a modified IVS protocol. Antigen-experienced CD4⁺ and CD8⁺ T cells were given IVS with either a single p53 neoantigen long peptide (LP) corresponding to the TP53 mutation expressed in the autologous tumor or a TP53-TMG that was used for TIL screening which contain multiple hotspot TP53 mutations. After 12 days of growth, a coculture with autologous antigen-presenting cells either pulsed with mutant p53-LP or electroporated with mutated TP53-TMG was performed, and the following day activated T cells were sorted by 41BB and/or OX40 expression. The sorted T cells were given a rapid expansion protocol (REP) and were screened for p53 neoantigen responses. Collectively, this strategy identified T cells reactive against TP53 hotspot mutations, suggesting that PBLs may be a viable source of generating p53 neoantigen–targeted cancer immunotherapies.

Written, informed consent was granted from all study participants. This study was approved by the Investigational Review Board in accordance with an assurance filed with and approved by the U.S. Department of Health and Human Services at the NCI (Bethesda, MD) and was registered at https://clinicaltrials.gov under NCT01174121. The study was conducted in accordance with the U.S. Common Rule. Patients were chosen on basis of availability of pretreatment leukapheresis and TIL screening results and had received prior therapies (surgery, chemotherapy, radiotherapy) per standard of care (12, 14). Ficoll–Hypaque was used to isolate PBLs from leukapheresis and cells were cryopreserved for further use. All patients studied had a confirmed TP53 mutation by whole-exome sequencing in line with clinical protocols as described previously (12).

CEM/C1, HCC1395, SKMEL5, Saos2, and KLE tumor cell lines were purchased from ATCC. TYKNU tumor cell line was acquired from Japanese Collection of Research Bioresources Cell Bank. Mycoplasma testing and cell authenticity were not independently performed by our group as we relied on the commercial vendor's testing. Saos2-R175H and TC#4266 cell lines were generated and authenticated as described previously (12). All cell lines were acquired within the past 3 years and used within 2 to 10 passages from receipt from the commercial vendor and within 1 to 5 passages from the time of thaw. Tumor cell lines were grown for at least a week from cryopreserved stocks before coculture.

Fluorescently labeled antibodies used in flow cytometry are detailed in Supplementary Table S1. Analytic flow cytometry was performed on FACSCanto II (BD Biosciences) with analysis by FlowJo software (TreeStar). All cells were gated by lymphocytes and live cells by exclusion of cells stained with propidium iodide (PI). Cells were sorted for IVS and 41BB/OX40 on a FACS Aria II (BD Biosciences). 41BB⁺ and/or OX40⁺ cells were sorted separately through CD3⁺CD4⁺CD8⁻ (CD4) and CD3⁺CD4⁻CD8⁺ (CD8) gates. Cocultures were sorted by SH800S sorted (Sony Biotechnology) for single-cell PCR to identify TCR genes.

The wild-type (TMG-WT-TP53) and mutated TP53 (TMG-MUT-TP53) tandem minigene constructs were generated for TIL screening as described previously (12). In brief, each TP53 hotspot mutation (R175H, Y220C, G245S, G245D, R248L, R248Q, R248W, R249S, R273C, R273H, R273L, and R282W) was composed into a minigene with mutated codon in the middle and 12 normal codons upstream and downstream and the minigenes were concatenated into a TMG. A similar sequence corresponding to wild-type sequences was also derived. TMGs were synthesized as DNA and cloned in frame to a LAMP signal sequence and a DC-LAMP localization sequence then in vitro transcribed to mRNA using mMESSAGE mMACHINE T7 Ultra Kit according to the manufacturer's instructions (Thermo Fisher Scientific). In addition, wild-type and mutated peptides were synthesized for p53^R175H, p53^Y220C p53^R248W, p53^R248Q, and p53^R273H and purified to >95% by high-performance liquid chromatography (Genscript). All peptides were reconstituted in DMSO.

Monocyte-derived immature dendritic cells (DC) were generated by adherence method (25). Briefly, PBLs were plated in AIM-V media (Life Technologies) containing DNase (Genentech Inc.) and incubated for 1.5 to 2 hours at 37°C. Nonadherent cells were removed and used fresh or cryopreserved. Adherent cells were washed in AIM-V, incubated for 1 hour at 37°C, and media were exchanged to DC media [RPMI1640, 2 mmol/L l-glutamine, with 5% human serum, 100 U/mL penicillin, 100 μg/mL streptomycin, amphotericin B, 800 IU/mL granulocyte-macrophage colony-stimulating factor (GM-CSF; Leukine) and 200 U/mL IL4 (Peprotech)]. Cells were fed every 2 to 3 days with cytokines and harvested on days 5 to 6.

To perform the antigen experienced sort, pretreatment PBLs were processed as described previously (24). Cryopreserved apheresis samples were thawed, washed, set to 5–10 × 10⁶ cells/mL with AIM-V media containing DNase, and 1.75–2 × 10⁸ viable cells were plated per T175 flasks (Corning Inc.) and incubated at 37°C, 5% CO₂ for 90 minutes. After 90 minutes, the nonadherent monocyte-depleted PBLs were harvested, centrifuged, and incubated overnight at 37°C and 5% CO₂ in 50/50 media [AIM-V media, RPMI1640 media (Lonza), 5% human AB serum, 100 U/mL penicillin and 100 μg/mL streptomycin (Life Technologies), 2 mmol/L l-glutamine (Life Technologies), 10 μg/mL gentamicin (Quality Biological Inc.), 12.5 mmol/L HEPES (Life Technologies)]. Adherent monocytes were differentiated into immature DCs as described above. After resting the nonadherent cells overnight, the cells were harvested, and 1–2 × 10⁸ cells were resuspended in 50 μL of staining buffer (PBS, 0.5% BSA, 2 mmol/L EDTA) with CD3, CD8, CD4, CD62L, CD45RO antibodies. Cells were incubated for 30 minutes at 4°C and washed twice before acquisition. To determine the sorting population, gating was performed on live cells (propidium iodide negative), single cells, CD3⁺ T cells then antigen experienced cells (CD62L⁻CD45RO⁺, CD62L⁺CD45RO⁺, CD62L⁻CD45RO⁻), which were further subdivided into CD4⁺ or CD8⁺. The CD8⁺ and CD4⁺ antigen experienced memory T cells were sorted separately, collected, counted, and resuspended in 50/50 media containing concentration of 60 ng/mL IL21. Autologous DCs were electroporated 18 to 24 hours in advance with TMG-MUT-TP53 or pulsed for 2 to 4 hours the day of the FACS sort with patient-specific mutant p53-LP. Target cells (DCs) were washed in 50/50 media twice and resuspended in 50/50 media with no cytokines. IVS was performed with a 1:3 to 1:6 ratio (DC:T cell) coculture in a final concentration of 30 ng/mL IL21 after the addition of the DCs. Following 14 days of growth and 3 feedings with IL21 and IL2 (aldesleukin) as described previously (24), autologous DCs were again electroporated or pulsed with TMG-MUT-TP53 or mutated LP, respectively, and cocultured with IVS cultures for 18 to 24 hours at 37°C and 5% CO₂. In parallel, IVS cultures were cocultured with DCs electroporated with irrelevant TMG or pulsed with DMSO for negative controls during the 41BB/OX40 enrichment sort. Following coculture, the cells were harvested and resuspended in 50 μL of staining buffer containing CD3, CD4, CD8, 41BB, and OX40 antibodies, incubated for 30 minutes at 4°C, and washed twice before acquisition. The sorted 41BB⁺/OX40⁺ enriched T cells were expanded by REP using irradiated PBL feeders, 30 ng/mL OKT3 antibody (Miltenyi Biotec) and 3,000 IU/mL IL2 in 50/50 media. The REP cultures were fed 3 times and were tested for reactivity or cryopreserved on day 14.

Screening of IVS and enriched T cells was accomplished through same strategy used to screen TIL fragments (12). Briefly, autologous DCs were electroporated with TMG (10⁵ cells/well) and rested overnight or pulsed with peptide or DMSO (8 × 10⁴ cells/well) for 2 to 4 hours. Target cells were washed twice and resuspended in 50/50 media and cocultured with 2 × 10⁴ T cells in IFNγ ELISPOT plates (EMD Millipore). Phorbol 12-myristate 13-acetate (PMA) and ionomycin (Thermo Fisher Scientific) were used as a positive control and media only was a negative control. The cocultured cells were removed, stained, and analyzed by flow cytometry as above, while the ELISPOT plate was processed according to the manufacturer's instructions. Tumor cells were cocultured at 1:1 ratio with T cells (2 × 10⁵ total cells) overnight in round-bottom 96 well plates. Following harvesting of coculture supernatant to assess IFNγ secretion by ELISA (Thermo Fisher Scientific), the cocultured cells were and analyzed by flow cytometry.

Similar method previously described was used to identify minimal peptides and determine HLA restrictions (12). In short, NetMHC peptide binding affinity algorithm (v3.4) was used to predict neoepitopes for HLA class-I alleles (26). Candidate 9–11 amino acids were cocultured as described above. To investigate the CD4⁺ minimal neoantigens, 15 amino acid peptides overlapping 14 amino acids were cocultured as described above. To determine the HLA restrictions, COS7 tumor cells were plated at 2.5 × 10⁴ cells/well in RPMI1640, 2 mmol/L l-glutamine and 10% FBS in flat-bottom 96-well plates and incubated overnight at 37°C. Patient-specific individual HLA class-I alleles (300 ng/well) or both HLA class-II α and β chains (150 ng/well each) in DNA plasmids (pcDNA3.1) were transfected with Lipofectamine2000 according to the manufacturer's instructions (Thermo Fisher Scientific). When TMGs (100 ng/well) were cotransfected with HLA, the concentration of HLAs reduced to 150 ng/well for class-I and to 100 ng/well for class-II. The wild-type TMGs used for these experiments were the TMG-MUT-TP53 reverted to wild type only at the position of interest, for example, R175H. Following 24-hour incubation, transfection media were removed, peptides or DMSO were pulsed for 2 to 4 hours in 50/50 media when applicable, wells were washed twice with 50/50 media, and 10⁵ T cells were incubated overnight. Coculture supernatants were analyzed for IFNγ secretion by ELISA and cells were stained for upregulation of 41BB and analyzed by flow cytometry.

TCRB survey sequencing was performed from genomic DNA by Adaptive Biotechnologies. A minimum 5 × 10⁴ cells were sent for sequencing. Analysis of productive TCR rearrangements was performed using ImmunoSEQ Analyzer 3.0 (Adaptive Biotechnologies).

TCRs were identified by sorting cocultures of T cells and DCs expressing p53 neoantigens into single wells followed by single cell reverse transcriptase polymerase chain reaction (RT-PCR) of TCR genes similar to previous studies (27, 28). The PCR products were kept separate for TCRα and TCRβ and were analyzed by Sanger sequencing. These partial TCR sequences were analyzed with IMGT/V-Quest (http://www.imgt.org/IMGT) and IGBLAST (https://www.ncbi.nlm.nih.gov/igblast) websites, which identified the CDR3 and J or D/J regions and inferred the TRAV and TRBV family. The human full-length variable sequences were fused to murine constant chains as was done in other studies (27, 29). The murinized TCRα and TCRβ genes were linked with a RAKR-SGSG sequence and P2A ribosomal slip sequence to result in stoichiometric expression of the TCR chains in a single cistron. This sequence was synthesized and cloned into MSGV1 vector for generation of transient retroviral supernatants.

PBL donors were adjusted to 3 × 10⁶ cells/ml in 50/50 media supplemented 50 ng/mL soluble OKT3 and 300 IU/ml IL-2 and were activated on low adherence plates for two days prior to retroviral transduction. The pMSGV1 plasmid encoding mutation-specific TCR (1.5 μg/well) and the envelope-encoding plasmid RD114 (0.75 μg/well) were co-transfected into 10⁶ HEK293GP cells/well of a 6-well poly-D-lysine-coated plate using Lipofectamine2000 (Life Technologies). Retroviral supernatants were collected two days after transfection, diluted 1:1 with DMEM media, and centrifuged onto non-tissue culture-treated 6-well plate coated with Retronectin (10 μg/well, Takara) at 2,000 × g for 2 hours at 32°C. Supernatants were aspirated and 2 × 10⁶ stimulated T cells at 5 × 10⁵ cells/mL were added to each well in 50/50 media with 300 IU/mL IL2. The T cells were centrifuged onto the retronectin coated plates for 10 minutes at 300 × g. The media were exchanged 3 to 4 days later with 300 IU/mL IL2 and transduced cells were assayed 10 to 14 days posttransduction.

Antigen experienced T cells from PBLs were evaluated in 7 colon, 1 rectal, and 1 ovarian cancer patients with a TP53-mutated tumor (Table 1). Patients 4217 (p53^R175H), 4213 (p53^R248Q), 4257 (p53^R248W), and 4254 (p53^R273H) did not have TIL responses to p53 neoantigens. In contrast, TILs were reactive to the autologous TP53 mutation in patients 4141 (p53^R175H), 4285 (p53^R175H), 4149 (p53^Y220C), 4266 (p53^R248W), and 4273 (p53^R248W; refs. 12, 14). Cryopreserved aphereses (prior to any ACT) were used to sort antigen experienced CD4⁺ or CD8⁺ T cells from PBLs (Fig. 1A, left). Sorted CD4⁺ or CD8⁺ T cells were in vitro stimulated with DCs expressing mutated TP53-TMG mRNA (TP53-TMG-IVS) or pulsed with patient-specific mutated p53-LP (p53-LP-IVS). After 12 days of culture in the presence of IL21 and IL2, in vitro stimulated CD4⁺ and CD8⁺ memory cells were cocultured with DCs electroporated with mutated TP53-TMG mRNA, in the case of TP53-TMG-IVS, or cocultured with patient-specific mutated p53-LP, in the case of p53-LP-IVS, and sorted the following day based on expression of T-cell activation markers 41BB and/or OX40 (Fig. 1A, middle). The cell yields after TP53-TMG-IVS and p53-LP-IVS were comparable for both CD4⁺ and CD8⁺ T cells ranging from 5 × 10⁵ to 3.6 × 10⁷ cells from 3 × 10⁵ to 1.6 × 10⁷ starting T cells (Table 2). A portion of CD4⁺41BB⁺/OX40⁺ and CD8⁺41BB⁺/OX40⁺ T cells were sorted from all populations for completeness and symmetry of the experiment, which ranged from 2 × 10² to 1.7 × 10⁵ cells and 0.1% to 6.9% from the parent CD4 or CD8 gate (Table 2). The sorted T cells underwent a REP and were analyzed after of 14 days rapid expansion. The final cell yields ranged from 7 × 10⁶ to 4.2 × 10⁸ T cells, which were likely influenced by the input cell numbers into the REP (Table 2).

An analytic screen was performed on the cultures after IVS, 41BB/OX40 enrichment and REP where reactivities were evaluated by cell surface marker 41BB upregulation via flow cytometry and IFNγ secretion using an ELISPOT assay (Fig. 1A, right). Peripheral blood T cells were not reactive to p53 neoantigens in patients 4213, 4217, 4254, and 4257, which corroborated the TIL screening results (Table 1). These patients may not have had an immunogenic combination of HLA and p53 neoepitope. In contrast, TP53 mutation–specific T cells were identified in antigen experienced T cells from PBL of 5 patients (4141 and 4285: p53^R175H, 4149: p53^Y220C, 4266 and 4273: p53^R248W) who had intratumoral T-cell responses to mutated TP53 by TILs (Fig. 1B; Table 1; ref. 12). TP53-TMG-IVS resulted in p53 neoantigen-specific T cells in 4141-CD8, 4285-CD4, 4149-CD4, and 4266-CD8 cultures (Fig. 1B, left), and p53-LP-IVS resulted in p53 neoantigen-specific T cells in 4285-CD4 and 4273-CD4 cultures (Fig. 1B, right). The specificity of the responses to mutated TP53 (Fig. 1B, closed shapes) was exemplified by the lack of response to the wild-type counterpart (Fig. 1B, open shapes). The highest frequency of TP53 mutation-reactive cells was 78% in the 4141-CD8 TP53-TMG-IVS culture against p53^R175H (Fig. 1C). The 4285-CD4 TP53-TMG-IVS culture exemplified a positive IFNγ secretion screening result as T cells were reactive to the mutant TMG-MUT-TP53 and p53^R175H LP but not against wild-type TMG-WT-TP53, irrelevant TMG, DMSO (peptide vehicle), or wild-type p53^R175 LP (Fig. 1D). Selected cultures were deemed reactive based on 41BB upregulation and/or IFNγ secretion and were studied further (Supplementary Table S2). The 4285-CD4 TP53-TMG-IVS culture showed specific recognition of the cognate p53^R175H LP with peptide concentrations down to 10 ng/mL (Fig. 1E). Thus, through IVS and 41BB/OX40 enrichment, highly specific CD8⁺ and CD4⁺ T cells targeting public TP53 mutations could be identified.

We wanted to further characterize the TCR diversity for each patient's TP53 mutation reactive T-cell population and determine whether the IVS and 41BB/OX40 enrichment altered the T-cell repertoire. To accomplish this, we performed TCRB deep sequencing (27) and measured productive T-cell clonotype frequencies based on unique CDR3B sequences and overall sample clonality, which is a normalized measurement of the population diversity where more oligoclonal samples approach 1 (30, 31). TCRB clonality significantly increased in TP53-TMG-IVS and p53-LP-IVS cultures generated from all patient PBL samples relative to the pre-IVS PBL (Fig. 2A). Similarly, the most frequent unique TCRB clonotype in each TP53-TMG-IVS and p53-LP-IVS sample was of higher frequency than in the pre-IVS PBL (Fig. 2B). The ranking of p53 neoantigen-reactive TCRB sequences ranged from 1 to 167 in the final p53-LP-IVS or TP53-TMG-IVS cultures but were not detected or ranked 5,020 in the original PBL (Supplementary Table S3). The increased clonality or maximum TCRB clonotype frequencies were not restricted to cultures with T-cell responses to p53 neoantigens (Fig. 2A and B, asterisks denoted positive cultures). This suggested that there was T-cell repertoire skewing through the IVS and 41BB/OX40 enrichment but assessments of top frequency and clonality were not sufficient to predict a culture's response to mutated TP53.

We then identified TCRs from the p53 neoantigen-reactive IVS populations for potential therapeutic and research use and to track the TCRB clonotypes during the culture period to evaluate the extent of TP53 mutation-reactive T-cell enrichment. The TCRs were identified following a co-culture of reactive IVS cultures with the cognate p53 neoantigen (TMG or LP), sorting 41BB⁺ T cells and single-cell RT-PCR of TCR alpha and beta genes, similar to previous studies (27). TCR pairs were reconstructed, cloned into retroviral vectors and transduced into donor PBL. A total of 11 TCRs targeting p53^R175H (patients 4141 and 4285) and p53^R248W (patients 4266 and 4273) neoantigen were identified (Supplementary Table S3). We were unable to determine the TCR for patient 4149 due to limited availability of T cells. The same p53 neoantigen-reactive TCRs from TILs (12) were identified in PBLs after IVS and 41BB/OX40 enrichment in patients 4141, 4266 and 4273 targeting p53^R175H/HLA-A*02:01, p53^R248W/HLA-A*68:01 and p53^R248W/HLA-DPB1*02:01, respectively. No additional TP53 mutation-reactive T cells were identified in these patients. In contrast, 7 unique p53^R175H neoantigen-specific TCRs from patient 4285 were identified from PBL, which were not present in the TIL study, and the TCR derived from TILs was not found in the PBL populations. The functional avidities of the PBL-derived TCRs (4285-PBL-TCR) were comparable to the TIL-derived TCR (4285-TIL-TCR) with recognition of p53^R175H LP to 10 ng/mL and no response to the wild type p53^R175 LP (Fig. 2C). Tracking of p53^R175H neoantigen-specific TCRB clonotypes from patients 4141 and 4285 demonstrated exponential expansion after IVS and 41BB/OX40 enrichment compared to the starting PBL (Fig. 2D, left). Moreover, the CDR3B sequences from patient 4285 were below the limit of detection (<0.001% from 2 × 10⁵ reads) in bulk PBL but were of sufficient frequencies following IVS and 41BB/OX40 enrichment, including four TCRs in the top 10 total CDR3B clonotypes (Supplementary Table S3). The p53^R248W neoantigen-specific TCRs from patient 4266 were also below the limit of detection in the PBLs but were 2.6% (4266-PBL-TCR3) and 7.5% (4266-PBL-TCR2) in the 4266-CD8 TP53-TMG-IVS culture (Fig. 2D, right). The 4273-PBL-TCR with specificity to p53^R248W neoantigen enriched from 0.002% to 0.017% in the 4273-CD4 p53-LP-IVS culture (Fig. 2D, right). Collectively, the data demonstrated that PBL can be source of public TP53 mutation-reactive TCRs identical or comparable to intratumoral TCRs.

The recognized minimal p53 neoepitopes and corresponding HLA restrictions were then assessed. HLA mapping was accomplished by transfecting DNA plasmids corresponding to each of the patient's individual HLA alleles into COS7 monkey cell line (lacking HLA) and pulsing peptides or cotransfecting with TMG, similar to previous reports (12, 14). The 4141-CD8 TP53-TMG-IVS culture was specific for the p53^R175H HMTEVVRHC neoepitope in the context of HLA-A*02:01, a highly frequent HLA in the U.S. population (32), as measured by 41BB expression (Fig. 3A). HLA-A*68:01 restricted the p53^R248W neoepitope SSCMGGMNWR recognized by the 4266-CD8 TP53-TMG-IVS culture as measured by IFNγ secretion (Fig. 3B). This was expected as the TCRs from TIL in patients 4141 and 4266 were found in these IVS cultures and had already established minimal neoepitopes and HLA restriction elements (12). Similarly, the 4273-PBL-TCR was found in the 4273-CD4 p53-LP-IVS culture and had a known p53^R248W and HLA-DPB1*02:01 combination from the TIL studies. Even though the TCRs were different between PBL and TIL for patient 4285, the 4285-CD4 TP53-TMG-IVS culture was specific for the same p53^R175H and HLA-DBR1*13:01 combination found in the TIL (Fig. 3C). Furthermore, all 4285-PBL-TCRs were specific for p53^R175H and HLA-DBR1*13:01 (data not shown) and were all present in the 4285-CD4 TP53-TMG-IVS culture by TCRB sequencing (Supplementary Table S3). Fifteen amino acid p53^R175H peptides overlapping 14 amino acids were pulsed on DCs from patient 4285 and cocultured with TCR transduced T cells, and the core peptide EVVRHCPHHER was determined to be the common sequence recognized by 4285-PBL-TCRs in the context of HLA-DRB1*13:01 (Fig. 3D). In sum, we found the same TCRs from TILs in PBLs that recognized the same HLA and minimal p53 neoepitopes in 3 cases and in one case found additional TCRs with the same p53 neoantigen specificity as the intratumoral T cells.

TP53 mutation–reactive T cells were evaluated for the capacity to recognize naturally processed and presented antigen expressed on the tumor cell surface. Saos2-R175H osteosarcoma tumor cell line (HLA-A*02:01 and overexpressing full length p53^R175H) and TC#4266 human xenograft tumor cell line (p53^R248W and HLA-A*68:01:02 colon cancer) were co-cultured overnight with CD8⁺ T cells from the 4141-CD8 TP53-TMG-IVS and 4266-CD8 TP53-TMG-IVS cultures, which upregulated 41BB in response to Saos2-R175H and TC#4266 cell lines, respectively, with minimal activation against the cross-matched cell line (Fig. 3E). To test whether p53 neoantigen-reactive TCRs identified from peripheral blood would be of value as gene-modified cell therapy, T cells transduced with the p53^R175H/HLA-A*02:01-specific 4141-PBL-TCR were cocultured with allogeneic cell lines expressing p53^R175H, HLA-A*02:01 or both. TYKNU, KLE and Saos2-R175H (p53^R175H, HLA-A*02:01) were recognized by 4141-PBL-TCR as measured by specific upregulation of 41BB (Fig. 3F) and secretion of IFNγ into coculture supernatants (Fig. 3G). In contrast, 4141-PBL-TCR did not recognize CEM/C1 (p53^R175H, HLA-A*02:01^negative), HCC1395 (p53^R175H, HLA-A*02:01^negative), SKMEL5 (p53^WT, HLA-A*02:01) and Saos2 cell lines (13), suggesting that the TCR was specific to cell lines expressing p53^R175H and HLA-A*02:01 and that sufficient p53 neoepitope was on the tumor cell surface to trigger TCR-engineered T-cell recognition. Mock-transduced T cells were not reactive to any of the cell lines, further supporting the specificity of the 4141-PBL-TCR. Thus, TP53 mutation-specific T cells and TCRs from PBL could recognize autologous and allogeneic tumor cells with naturally processed and presented p53 neoepitopes.

The intratumoral (TIL) and antigen experienced T responses from PBLs were congruent in 9 patients, including identical TCRs in 3 cases. This suggested that TP53 mutation-reactive T cells circulate in peripheral blood, which can provide insight into the intratumoral T-cell response and can be a source of p53 neoantigen-reactive T cells and TCRs for ACT. The initial amount of blood necessary to find the reactive cells will likely depend upon the relative precursor frequencies, which in this study were less than one in 10⁵ PBLs in most cases. Thus, leukapheresis or a large volume venipuncture may be needed to start with 10⁶ antigen experienced CD4⁺ or CD8⁺ T cells, but a thorough study may be needed to determine the limits of detection for this technique. The ability to translate the findings from this study to clinical application may depend on the frequency and quality of intratumoral T-cell responses to TP53 mutations that make their way to the peripheral circulation.

Both p53-LP-IVS and TP53-TMG-IVS were able to expand p53 neoantigen-reactive T cells. The TMG approach may be more advantageous as it covers 12 mutations at once, shows the T cells an intracellularly processed and presented neoepitope and resulted in more positive cultures than the p53-LP-IVS (Fig. 1B). This TP53-focused strategy could be expanded to other TP53 mutations, some of which are shared in unrelated people, to interrogate T cells from both TIL and PBL in their capacity to recognize p53 neoantigens in a more comprehensive manner. Indeed, patients with gastrointestinal tumors have demonstrated T-cell responses by TILs to nonhotspot TP53 mutations (33), suggesting that mutated TP53 is likely immunogenic outside of the most frequent genetic changes. In our previous studies, almost all neoantigens with a verified T-cell response were unique to the individual patient with the exception of KRAS and TP53, which have immunogenic mutations recognized by unrelated people with the appropriate HLA restriction element (3, 5–7, 12, 14, 25, 29, 33–37). This indicates that KRAS and TP53 could be high value targets for ACT.

Targeting TP53 mutations with T cells may be an efficacious strategy because of the importance of the mutated TP53 to the tumor. The loss of HLA is a limiting aspect to any ACT strategy, the extent of which is largely unknown and likely dependent upon the target or HLA/p53 neoepitope combination. In contrast to HLA loss, antigen loss is less likely in the case of mutated TP53. Tumors expressing TP53 missense mutations typically have loss-of-heterozygosity of the wild type allele thereby limiting the chances of antigen loss after ACT (12, 38). Furthermore, some TP53 mutations can have gain-of-function activity, further emphasizing their importance to tumor survival and fitness (39). TP53 is expressed at high levels in cells but the p53 protein is degraded by MDM2 in normal cells (40). Mutations in TP53 can interrupt the p53 degradation process and lead to accumulation in the cytosol (41) indicating that there will likely be high levels of neoepitope available for T-cell recognition.

This study lays the foundation for the generation of cell therapy directly using PBL after IVS and 41BB/OX40 enrichment or by genetic modification with TCRs. Furthermore, TCRs identified from IVS and 41BB/OX40 enrichment could be used for any unrelated donors with matching TP53 mutation and HLA expression in an off-the-shelf setting. Animal models of mutated TP53 xenografts can potentially be established to evaluate the pre-clinical efficacy of p53 neoantigen-reactive TCR gene–engineered T cells in vivo though these experiments of human cells in the mouse are often difficult to interpret. An all-in-one noninvasive strategy could be achieved by combining IVS and 41BB/OX40 enrichment with circulating tumor DNA detection of TP53 mutations in patient plasma or serum, which has been shown to strongly correlate with genetic features of the tumor (42–45) and is a promising diagnostic and prognostic tool (44, 46–48). Identification of TP53 mutations from circulating tumor DNA using a liquid biopsy could make PBL the sole source for neoantigen and T cells. This strategy can likely benefit patients ineligible for surgery and deepen the scope of targeting the most commonly mutated gene in cancer.

N.P. Restifo is an employee/paid consultant for Lyell Biopharma and holds ownership interest (including patents) in Lyell Immunopharma. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Malekzadeh, R. Yossef, M. Jafferji, P.F. Robbins, N.P. Restifo, S.A. Rosenberg, D.C. Deniger

Development of methodology: P. Malekzadeh, R. Yossef, G. Cafri, P.F. Robbins, S. Ray, N.P. Restifo, D.C. Deniger

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Malekzadeh, G. Cafri, B.C. Paria, F.J. Lowery, M. Jafferji, M.L. Good, A. Sachs, A.R. Copeland, S.P. Kim, S. Kivitz, M.R. Parkhurst, S. Ray, L. Xi, Z. Yu, N.P. Restifo, R.P.T. Somerville, S.A. Rosenberg, D.C. Deniger

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Malekzadeh, F.J. Lowery, M. Jafferji, A.R. Copeland, S.P. Kim, S. Kivitz, P.F. Robbins, S.A. Rosenberg, D.C. Deniger

Writing, review, and/or revision of the manuscript: P. Malekzadeh, R. Yossef, F.J. Lowery, M. Jafferji, M.L. Good, A.R. Copeland, S.P. Kim, S. Kivitz, P.F. Robbins, M. Raffeld, N.P. Restifo, S.A. Rosenberg, D.C. Deniger

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Malekzadeh, A. Sachs, L. Xi, R.P.T. Somerville, S.A. Rosenberg, D.C. Deniger

Study supervision: N.P. Restifo, S.A. Rosenberg, D.C. Deniger

We thank the TIL lab (NCI Surgery Branch) for their efforts processing the apheresis, and Arnold Mixon and Shawn Farid for their assistance with FACS. This work was supported by an award to S. Rosenberg by the Intramural Research Program of the NIH at the Center for Cancer Research, NCI.

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.