Characterization of an Immunogenic Mutation in a Patient with Metastatic Triple-Negative Breast Cancer

Adoptive cell therapy (ACT) using tumor-infiltrating lymphocytes (TILs) has been shown to mediate durable complete responses in patients with metastatic melanoma (1, 2). Recent evidence suggests that tumor regression is mediated by T-cell recognition of unique mutated tumor antigens in melanoma (3–5). Epithelial solid cancers, which make up the vast majority of human malignancies, have far fewer somatic mutations compared with melanoma, and therefore, potentially fewer immunogenic mutations (6). Nine of 10 patients with a variety of gastrointestinal cancers were found to have one to three mutations that gave rise to mutation-reactive T cells. In one patient with cholangiocarcinoma, 26 mutations were identified, one of which, mutated ERBB2IP, generated a CD4⁺ T-cell response capable of mediating an objective tumor response (7, 8). These data demonstrate that a CD4⁺ T-cell response directed against a single mutated antigen is sufficient to mediate an objective tumor regression.

In the United States, one in eight women will develop invasive breast cancer in their lifetime. Triple-negative breast cancers (TNBCs) are the most aggressive subtype, with higher recurrence, distant metastases, and decreased survival (9). Although TILs have been identified in breast adenocarcinomas by IHC and found to correlate with patient prognosis, little is known about their reactivity or function within the tumor (10, 11). We aim to identify immunogenic mutations in patients with metastatic breast cancer with the ultimate goal of developing an ACT therapy targeting immunogenic mutations for patients with breast cancer.

Tumor 4062 (lung) was resected from a 61-year-old woman diagnosed with metastatic TNBC (NCT00001823). 4062-TILs were generated as described previously (12). Autologous CD19⁺ B cells were sorted from the patient's apheresis sample using anti-CD19 microbeads (Miltenyi Biotec) and cultured with irradiated (6,000 rad) NIH-3T3 cells stably expressing CD40L (3T3-CD40L, authenticated by FACS for CD40L) at a 1:1 ratio in B-cell media composed of IMDM media (Life Technologies) supplemented with 10% human serum and 200 U/mL IL4 (Peprotech). Peripheral blood mononuclear cells were cultured in AIM-V medium (Life Technologies) supplemented with 5% human AB serum (Valley Biomedical), 0.05 mmol/L β-mercaptoethanol, 0.1 mmol/L nonessential amino acids, 25 mmol/L HEPES, and 2 mmol/L l-glutamine (Life Technologies). Cells were maintained at 37°C and 5% CO₂. Melanoma cell lines were isolated from surgically resected metastases as described previously (13): mel624 (HLA-A2⁺/MART-1⁺) and mel2963 (HLA-DP⁺/MAGE-A3⁺). These tumor cell lines, authenticated by HLA typing and MART-1 and/or MAGE-A3 expression in the Surgery Branch (SB), were cultured in R10 medium consisting of RPMI1640 (Life Technologies) containing 10% FBS (Hyclone). Partially HLA-DR–matched allogeneic EBV-B cells (authenticated in the SB by HLA typing) were grown in R10 supplemented with 2 mmol/L l-glutamine.

Whole-exome sequencing (WES) was performed at Personal Genome Diagnostics (PGDx) and the SB, NCI (Bethesda, MD). A whole-exome library was prepared using Agilent Technologies SureSelectXT Target Enrichment System for paired-end libraries and sequenced on a NextSeq 500 desktop sequencer (Illumina). The library was prepared using 3-μg genomic DNA (gDNA) isolated from fresh tumor or ten 10-μm OCT sections following the manufacturer's protocol (Agilent Technologies). An mRNA sequencing library was prepared from fresh tumor using Illumina TruSeq RNA Library Prep Kit. In brief, poly(A) mRNA was purified from 1 μg total RNA using oligo-dT magnetic beads and fragmented in the presence of divalent ions at 94°C for 8 minutes. Fragmented RNAs were reverse-transcribed into first-strand cDNAs by random priming, and SuperScriptII followed by second-strand cDNA synthesis. Double-stranded cDNAs were end-repaired, adenylated at the 3′ ends, ligated to indexed sequencing adaptors, and amplified (15 cycles). Paired-end sequencing was carried out on a NextSeq500 sequencer using a mid-output flow cell kit (300 cycles, Illumina). Alignments were performed at SB using Novoalign MPI from Novocraft (http://www.novocraft.com/) to human genome build hg19. Duplicates were marked using Picard's MarkDuplicates tool. Indel realignment and base recalibration was carried out according to the GATK best practices workflow (https://www.broadinstitute.org/gatk/). After cleanup of data, samtools mpileup (http://samtools.sourceforge.net) was used to create pileup files, and Varscan2 (http://varscan.sourceforge.net) was used to call somatic variants. These variants were annotated using Annovar (http://annovar.openbioinformatics.org). We used the following filters to generate putative mutations for evaluation for patient 4062: variant frequency cutoff of ≥10% and ≥2 variant reads in tumor. Patient 4062 mutations are listed in Supplementary Table S1.

4062 WES data were used in the copy-number analysis. The data analysis was performed in the R statistical environment, version 3.3.0. The segmented copy number, cellularity, and ploidy were determined using Sequenza v2.1.2 with normal samples as references and hg19 coordinates. The cancer cell fraction of each mutation was estimated by integrating the local copy number, tumor purity (obtained from Sequenza), and variant allele frequency (VAF). For the given mutations, only mutations with tumor coverage of at least 20× were used. Mutations were defined as clonal if the 95% confidence interval overlapped 1, and subclonal, otherwise (14). All mutations with read depth greater than 3 and VAF greater than 5% were clustered using PyClone v1.3.0 Dirichlet process clustering (15). We ran PyClone with 50,000 iterations and a burn-in of 1,000. In addition, 16 metastases were obtained at autopsy. To detect whether the mutation in recombination signal binding protein for immunoglobulin kappa J region (RBPJ) was present, cDNA was generated from the RNA using Superscript III First-Strand Synthesis System (Life Technologies). gDNA was generated using a DNeasy Kit. RBPJ-specific PCR was performed using the following primers: forward 5′CCTCAGCAAGCGGATAAAAG3′ and reverse 3′GCTGTGAACTGGCATGAAAA5′, or forward 5′AGCGAAAGCACTTCATGTTG3′ and reverse 3′GACATGGAGTGGCCTGAAAT5′. Sequencing of PCR products was performed by Macrogen using the first primer set. For genomic DNA samples, RBPJ-specific PCR was performed with the primer set forward 5′CCACCTTCTGCTCCAATCAC3′ and reverse 3′TGTAGCCATCTCGGACTGTG5′, followed by sequencing with the same primer set.

Tandem minigenes were constructed as described previously (7). RNA expression plasmids encoding the minigenes (12 minigenes per TMG) were linearized with the restriction enzyme NsiI. B cells were washed once with PBS and resuspended in Opti-MEM (Life Technologies) at 1.0 × 10⁶ cells/mL. In vitro transcribed (IVT) TMG RNA (8 μg) was aliquoted to the bottom of a 2-mm gap electroporation cuvette and 100 μL of B cells were added. Electroporations were carried out using a BTX-830 square wave electroporator (150 V, 20 ms, 1 pulse). Following electroporation, cells were immediately transferred to polypropylene tubes containing B-cell media, incubated overnight at 37°C, 5% CO₂, and washed once with PBS prior to use in coculture assays.

Individual 25-mer-long long peptides were synthesized (Genscript). Autologous B cells were harvested, washed, resuspended at 1.0 × 10⁶ cells/mL in B-cell media, and incubated with 10 μg/mL individual or pooled peptide overnight at 37°C, 5% CO₂. A mutated peptide from 4051-TIL was used as a negative control (ENSG00000132837, QWKDRAETVIIGDGCVGVSLAYHLA). MHC class II epitope predictions were done using NetMHCIIpan3.1 (www.cbs.dtu.dk/services/NetMHCIIpan/).

TILs were cocultured with peptide-pulsed or electroporated B cells at a 1:5 ratio or with plate-bound OKT3 (1 μg/mL, positive control). Response to targets was measured using an IFNγ enzyme-linked immunospot assay (IFNγ ELISPOT) as described previously (7). Spots were read and quantitated on an Immunospot S6 Microanalyzer (Cellular Technology Limited).

TILs were recovered from the plates and characterized by FACS [CD3-APC-Cy7, CD4-PE, CD8-PE-Cy7, CD134-FITC (OX40), and CD137-APC (4-1BB, BD Biosciences)], and T-cell activation was assessed by upregulation of the activation markers 4-1BB and/or OX40. Finally, TILs were cocultured with autologous EBV-B cells pulsed with peptides in the presence of GolgiStop and GolgiPlug and assessed for intracellular cytokine production according to the manufacturer's protocol (BD Biosciences). TILs were surface stained with CD3-PE-Cy7, CD4-PE prior to fixation and permeabilization, then stained with IFNγ-FITC, IL2-APC, and TNFα-PerCP5.5. FACS was performed on a Canto I instrument (BD Biosciences), and data analysis was performed using FlowJo software (FlowJo LLC).

HLA restriction mapping was performed by incubating peptide-pulsed B cells with pan anti-HLA-I W6/32 (Abcam), pan anti-HLA-II IVA12 (SB), anti-HLA-DP (Leinco), anti-HLA-DQ (Beckman Coulter), or anti-HLA-DR (SB) antibodies at 10 μg/1.0 × 10⁵ cells in 100 μL for 2 hours on ice. HLA-DR restriction was confirmed using allogeneic EBV-B cells partially matched at the HLA-DR locus. In an attempt to define the minimal epitope, truncated mutated peptides were synthesized (BioSynthesis) and pulsed onto autologous EBV-B cells at 10 μg/mL. T-cell avidity was assessed by peptide titration using EBV-B cells pulsed with serial dilutions of mutated or wild-type peptide (1–10,000 ng/mL). TIL recognition was assessed by coculture and IFNγ ELISPOT.

MutRBPJ-stimulated 4062-TILs (F9) were sorted on the basis of expression of CD3⁺, CD4⁺, and 4-1BB⁺ using a FacsJazz Cell Sorter (BD Biosciences). T-cell receptor (TCR) deep sequencing was performed on the sorted cells by Adaptive Biotechnologies. Four TCRs were synthesized (Genscript) using combinations of the two most dominant alpha and beta sequences based on CDR3 frequency. Alternatively, 4-1BB⁺–sorted cells were subjected to single-cell RNA sequencing (RNA-seq) on the Fluidigm C1, according to the manufacturer's protocol and paired TCR-alpha and beta sequences were identified and synthesized (Genscript). Synthesized TCRs were then cloned into our pMSGV1 gammaretroviral vector backbone for screening.

Patient 4062, a patient with metastatic TNBC, had a lung metastasis resected, and 72 nonsynonymous point mutations were identified after WES of the lesion, 43 of which were expressed on the basis of RNA-seq data (Supplementary Table S1). 4062-TILs were predominantly CD4⁺ T cells, ranging from 66%–94% (Supplementary Table S2). 4062-TILs were screened and ELISPOT reactivity observed in fragment 9 (F9; 756, PP3 vs. 37, irrelevant peptide) from 4062-TILs to a mutated antigen present in PP3 (Fig. 1A) as well as upregulation of 4-1BB (19.6%, PP3 vs. 0.53%, irrelevant peptide) in the CD4⁺ T-cell population (Fig. 1B). Similar findings were observed following coculture of 4062-TILs (F9) with B cells electroporated with TMG3 mRNA, which corresponds to the same mutated peptides present in PP3 (Supplementary Fig. S1), suggesting that the exogenous mutated antigen can be processed and cross-presented in the context of MHC class II. When screened against the 12 individual peptides in PP3, 4062-TILs (F9) recognized a peptide (RLRSQTVSTRYLLVEGGNFHASSQQ) encoding a mutation in RBPJ (RBPJ^{c.A611T/p.H204L}) by ELISPOT (Fig. 1C) and upregulation of both 4-1BB and OX40 in the CD4⁺ T-cell population (Fig. 1D). 4062-TIL (F9) reactivity was specific for RBPJ^{c.A611T/p.H204L} (Fig. 2A) and polyfunctional based on CD4⁺ T cytokine secretion (IL2, IFNγ, and TNFα, Fig. 2B).

62-TIL (F9) recognition of RBPJ^{c.A611T/p.H204L} was blocked by both a pan anti-MHC class II and anti-HLA-DR antibody (Fig. 3A; Supplementary Fig. S2). HLA-DRB1*1501–restricted based on recognition of allogeneic EBV-B cells partially matched at the HLA-DR locus (Fig. 3B). We identified a core sequence of the potential minimal epitope, RYLLVEGGN (Fig. 3C), suggesting the minimal epitope is likely located within a 15 amino acid sequence, RYLLVEGGNFHASSQ, predicted to be the third highest binding HLA-DRB1*1501 epitope [rank (17%), affinity (257 nmol/L), Fig. 3C]. In an attempt to isolate a reactive TCR, 4062-TILs (F9) were stimulated overnight and the top 7% of the CD4⁺/41BB⁺ T cells were sorted and sent for TCR deep sequencing (Fig. 4A). No dominant TCR clonotype was identified, suggesting that the frequency of the RBPJ^A611T-reactive cells was extremely low at the time of the sort. Four TCRs containing the two most dominant TCRα [Fig. 4B, TCRAV26 (16.4%) and TCRAV05 (14.5%)] and TCRβ [Fig. 4C, TCRBV20 (41.1%) and TCRVB07 (26.2%)] were synthesized (TCRAV26/TCRBV20, TCRAV26/TCRBV07, TCRAV05/TCRBV20 and TCRAV05/TCRBV07) and cloned into pMSGV1 backbone. Only one low-avidity TCR, TCRAV05/TCRBV07, recognized RBPJ^A611T in a dose-dependent manner based on ELISPOT and upregulation of 4-1BB (Fig. 4D and E, respectively). The low T-cell avidity was not based on poor TCR expression, as the TCRAV05/TCRBV07 TCR T-cell transduction efficiency was 70.0% ± 3.2% (range, 63.7%–73.9%, data not shown). In a separate experiment, 4-1BB–sorted T cells were sorted to single cells and analyzed by RNA-seq; however, no additional TCRs reactive against RBPJ^A611T were identified (data not shown).

The identification and isolation of mutation-reactive lymphocytes were conducted with the intent to treat patient 4062; however, the patient's disease progressed rapidly, and she expired prior to treatment. At autopsy, the RBPJ mutation was detected in all 16 sites of metastases (Supplementary Table S3). We were able to isolate mRNA of sufficient quality for RT-PCR analysis in eight of 16 samples, all of which expressed the mutation at the mRNA level. In addition, analysis of deep sequencing data obtained from the original resected metastasis indicated that 36 of 72 (50%) of the called mutations were clonal, which included RBPJ^{c.A611T/p.H204L}, six of 72 (8.3%) were subclonal, and 30 of 72 (41.7%) were unable to be determined due to the number of reads being below the cut-off value (Supplementary Table S1; Supplementary Fig. S3).

We demonstrated that TILs isolated from the autologous T-cell repertoire of a patient with TNBC recognized an immunogenic nonsynonymous mutation in RBPJ (RBPJ^{c.A611T/p.H204L}). The mutation reactivity was mediated by HLA-DRB1*15:01–restricted CD4⁺ T cells (Fig. 1). 4062-TIL (F9) reactivity was specific and highly avid against the mutated 25-mer peptide (Fig. 2A) and demonstrated a polyfunctional cytokine profile secreting multiple effector cytokines, including IFNγ, IL2, and TNFα (Fig. 2B). Given the specificity of mutation-reactive 4062-TILs (F9), it was unexpected that we were unable to isolate an avid TCR (Fig. 4). It is likely that the long-term expansion of this culture required to achieve cell numbers sufficient for analysis resulted in a decrease in the frequency of the desired TCR. For TCR gene therapy, the inability to isolate a highly avid TCR points to a potential limitation of this therapeutic approach; however, one could modify the TCR complementarity determining regions (CDRs) to potentially enhance the TCR avidity; however, there are risks associated with such modifications (16).

RBPJ is a ubiquitously expressed nuclear protein that, in the absence of Notch, binds DNA promoter sites and acts as a transcriptional repressor (17, 18). Thus, a mutated RBPJ could potentially dysregulate Notch signaling as a constitutive transcriptional activator or repressor, both of which could play a role in tumor growth. Data suggest the loss of RBPJ function can promote tumorigenesis (19, 20). The functional consequence of an RBPJ mutation (RBPJ^{c.A611T/p.H204L}) is not known; however, this mutation was present in all sites of metastasis, and our data suggest that mutated RBPJ is likely a clonal mutation (Supplementary Fig. S3; Supplementary Table S3). Ideally, identification of multiple immunogenic mutations, of both MHC class I and II restriction would likely increase the probability of developing a successful tumor treatment and reduce the need to target clonal mutations. Taken together, these data suggest that there is therapeutic potential for adoptive cell therapy in metastatic breast cancers using mutation-specific TILs.

No potential conflicts of interest were disclosed.

Conception and design: Y. Assadipour, S.A. Rosenberg, S.L. Goff, S.A. Feldman

Development of methodology: Y. Assadipour, N. Zacharakis, H. Xu, L. Lu, Z. Zheng, Y.-C. Lu, P.F. Robbins, S.L. Goff, S.A. Feldman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Assadipour, J.S. Crystal, T.D. Prickett, R.P.T. Somerville, M.A. Black, H. Chinnasamy, I. Kriley, L. Lu, S.L. Goff, S.A. Feldman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Assadipour, N. Zacharakis, J.S. Crystal, J.J. Gartner, L. Jia, L. Lu, P.F. Robbins, S.A. Rosenberg, S.L. Goff, S.A. Feldman

Writing, review, and/or revision of the manuscript: Y. Assadipour, N. Zacharakis, J.S. Crystal, T.D. Prickett, J.J. Gartner, I. Kriley, S.A. Rosenberg, S.L. Goff, S.A. Feldman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.S. Crystal, H. Xu, M.A. Black, H. Chinnasamy, L. Lu, S.A. Rosenberg, S.L. Goff, S.A. Feldman

Study supervision: S.L. Goff, S.A. Feldman

The authors would like to thank Drs. Eric Tran and Alena Gros for technical discussions. In addition, the authors would like to thank Arnold Mixon and Shawn Farid for assistance with flow cytometry.

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.