Melanoma often recurs after a latency period of several years, presenting a T cell–edited phenotype that reflects a role for CD8+ T cells in maintaining metastatic latency. Here, we report an investigation of a patient with multiple recurrent lesions, where poorly immunogenic melanoma phenotypes were found to evolve in the presence of autologous tumor antigen–specific CD8+ T cells. Melanoma cells from two of three late recurrent metastases, developing within a 6-year latency period, lacked HLA class I expression. CD8+ T cell–resistant, HLA class I–negative tumor cells became clinically apparent 1.5 and 6 years into stage IV disease. Genome profiling by SNP arrays revealed that HLA class I loss in both metastases originated from a shared chromosome 15q alteration and independently acquired focal B2M gene deletions. A third HLA class I haplotype-deficient lesion developed in year 3 of stage IV disease that acquired resistance toward dominant CD8+ T-cell clonotypes targeting stage III tumor cells. At an early stage, melanoma cells showed a dedifferentiated c-Junhigh/MITFlow phenotype, possibly associated with immunosuppression, which contrasted with a c-Junlow/MITFhigh phenotype of T cell–edited tumor cells derived from late metastases. In summary, our work shows how tumor recurrences after long-term latency evolve toward T-cell resistance by independent genetic events, as a means for immune escape and immunotherapeutic resistance. Cancer Res; 76(15); 4347–58. ©2016 AACR.

CD8+ T lymphocytes can efficiently kill autologous melanoma cells as indicated by the remarkable clinical responses to adoptive T-cell transfer and immune checkpoint–blocking antibody therapy (1–4). Such T lymphocytes target peptide epitopes derived from different types of tumor antigens presented in the context of HLA class I surface molecules (5). Their capacity to mediate tumor rejection has also been demonstrated in different murine tumor models. However, in some mice, selective T-cell pressure led to the outgrowth of poorly immunogenic antigen loss tumor variants (6–8). This suggests tumor antigen–specific T cells are potent drivers of the so called cancer immunoediting process, comprising an elimination, equilibrium, and escape phase (9). Although different immune effectors and regulators are involved in this multimechanistic process, CD8+ T cells seem to play a major role in all phases.

The capacity of CD8+ T cells to kill tumor cells is well established. In addition, evidence supports a role in the maintenance of tumor latency: in a genetically engineered murine model for spontaneous melanoma, depletion of CD8+ T cells accelerated the outgrowth of visceral metastases from early spread of tumor cells, indicating T cells delayed the metastatic progression of disseminated tumor cells (10). Similar observations were made in a carcinogen-induced tumor model, in which transformed cells established stable masses at the carcinogen injection site characterized by equilibrium of apoptosis and proliferation. Outgrowing tumors demonstrated an immunoedited, less immunogenic phenotype. Furthermore, depletion of CD8+ T cells accelerated tumor formation (11). These and other studies demonstrate a role for CD8+ T cells in maintaining tumor latency.

To date, reports examining the immunogenicity of recurrent patient metastases to gain insight into the immunoediting activity of CD8+ T cells are rare. In 1997, Ikeda and colleagues analyzed the T-cell responses toward tumor cell lines established from two stage IV melanoma metastases, obtained in 1988 and in 1993, after a 4-year disease-free interval. Tumor cells from the second metastasis had acquired a poorly immunogenic phenotype expressing only one of six HLA class I alleles (12). Similar observations were described by Yamshchikov and colleagues for cell lines established from two consecutive metastases excised approximately 5 and 11 years after primary melanoma diagnosis. Tumor cells from the first metastasis were poorly immunogenic due to HLA haplotype loss and those from the second lesion showed a mixed HLA class I–low/-negative phenotype (13). Both studies characterized phenotypic tumor cell alterations resulting in impaired T-cell recognition.

Although immunotherapies have evolved as the potentially most promising therapeutic approaches for a wide range of advanced malignant human neoplasia, tumor recurrences as well as nonresponders still significantly impede therapeutic success. The immunoediting activity of CD8+ T cells may well contribute to disease recurrence in responders to immune checkpoint–blocking antibody therapies, which mostly occurs after a prolonged period of therapy-induced regression (3). A detailed understanding of the mechanisms governing the immunogenicity of recurrent melanoma will be critical to optimize immunotherapy approaches both in terms of decreasing the rate of tumor recurrence and if this occurs, of choosing the best rescue therapy.

Here, we follow the genetic evolution of melanoma immunogenicity in a long-term survivor who developed multiple recurrent metastases over a period of 6 years in stage IV disease. We demonstrate distinct poorly immunogenic melanoma phenotypes evolving in a single patient after long-term latency and identify varying genetic mechanisms responsible for T-cell resistance in individual tumor metastases. Overall, our data support T cell–based immunoediting of disseminated tumor cells in melanoma patients.

Patient material

Tumor tissues and peripheral blood samples from melanoma patient Ma-Mel-86 were collected after approval by the Institutional Review Board and patient-informed written consent. Blood cells, collected in April 2004, were separated on a Ficoll gradient, and peripheral blood mononuclear cells (PBMC) were cryopreserved. Tissues were mechanically dissected for generation of cell lines and cryopreservation. Melanoma cell lines were cultured in RPMI1640 medium supplemented with glutamine, 10% FCS, and penicillin/streptomycin at 37°C in a 5% CO2 atmosphere.

Cell line authentication and HLA genotyping

The QIAamp DNA Mini Kit (Qiagen) was used for isolation of genomic DNA from melanoma cell lines and autologous PBMC according to the manufacturer's instructions. For cell line authentication, genetic profiling of genomic DNA was performed at the Institute for Forensic Medicine (University Hospital Essen, Essen, Germany) using the AmpFLSTR-Profiler Plus Kit (Applied Biosystems). HLA class I genotyping on genomic DNA from cell lines and autologous PBMC was carried out at the Institute for Immunology and Genetics Kaiserslautern (Kaiserslautern, Germany).

Targeted amplicon sequencing

To define known recurrent mutations in cutaneous melanoma, a custom amplicon-based sequencing panel covering 29 genes (Supplementary Table S1) was designed, prepared, run on an Illumina MiSeq sequencer, and analyzed by CLC Cancer Research Workbench from QIAGEN as described previously (14).

SNP array analysis

SNP genotyping was performed with nonfixed PBMC and melanoma cell lines using CytoScan HD arrays and analyzed with Chromosome Analysis Suite (Affymetrix) as described previously (15). Genotypes with a P value for confidence <0.005 were chosen using R (R Development Core Team; http://www.R-project.org), translated into single base letters and IUPAC ambiguity codes, then transposed to create an alignment. Of 84,000 variable sites, 1,456 were parsimony informative. We then used the maximum parsimony criterion, minimizing the total number of evolutionary steps required to explain the relationship of the tested samples. Using MEGA (16), we calculated a maximum parsimony tree with 500 bootstrap replicates, complete deletion of missing data, and subtree pruning–regrafting search method. A maximum likelihood tree was inferred using 100 bootstrap replicates, the Kimura 2-parameter model, and partial deletion of missing data. SNP array data are available via NCBI GEO (accession number: GSE80736).

Transcriptome analyses

Total RNA was extracted from melanoma cell lines using the Qiagen RNeasy Mini Kit according to the manufacturer's protocol (Qiagen). All preparations and analyses were done in triplicates. Sequencing [RNA sequencing (RNA-Seq)] libraries were generated using the TruSeq RNA Sample Preparation Kit (Illumina) by GENterprise Genomics. RNA-Seq libraries were subjected to high-throughput sequencing on the Illumina HiSeq2000 platform of the local NGS Core facility (Biology Department, University of Mainz, Mainz, Germany). On average, 69.4 × 106 100 bp paired-end reads were generated. Sequence reads were processed (quality filtering, adapter trimming) and mapped to the annotated human genome hg19 with CLC Genomics Workbench 8.5.1 (Qiagen). For analysis of differentially expressed genes and heatmap generation, the processed replicate reads of each sample were mapped together. Gene expression values were determined as RPKM (reads per kilobase of transcript per million mapped reads) normalized read counts. Triplicates of transcriptome data are available via NCBI SRA (accession numbers: Ma-Mel-86a SRX1542616, SRX1542617, SRX1542618; Ma-Mel-86b SRX1542619 SRX1542620 SRX1542621; Ma-Mel-86c SRX1542622, SRX1542623, SRX1542624; Ma-Mel-86f SRX1542625, SRX1542626, SRX1542627).

Mixed lymphocyte/tumor cell culture and T-cell cloning

Mixed lymphocyte/tumor cell culture (MLTC) of selected autologous CD8+ T cells and irradiated melanoma cells was set up as described previously (15). By limiting dilution in round-bottomed 96-well plates, CD8+ T cells from MLTC were cloned and stimulated with irradiated autologous tumor cells (3 × 103 cells/well) and allogeneic EBV-transformed B lymphoblastoid cells (5 × 104 cells/well) as feeders in AIMV medium supplemented with 250 IU/mL IL2 and 10% human AB serum. Restimulations were carried out in weekly intervals.

High-throughput T-cell receptor repertoire sequencing

DNA was extracted from autologous MLTC using the QIAamp DNA Mini Kit (Qiagen) according to the manufacturer's instructions. Total DNA extracts of MLTC-86a and MLTC-86c were sent to Adaptive Biotechnologies for sequencing the T-cell receptor β chain (TCRβ) repertoire at “survey” level using their immunoSEQ assay. At Adaptive Biotechnologies, 400 ng of template DNA was used to perform multiplex PCR enrichment of the somatically rearranged TCRβ region, followed by high-throughput sequencing on an Illumina MiSeq instrument (17). The generated TCRβ sequence data were analyzed using the immunoSEQ Analyzer. Analyses included amplification bias correction (18) as well as sequencing error correction (17). Following filtering for unique sequences and discarding sequences with mutations leading to frame shifts or stop codons, the total number of unique and productive TCRβ clonotypes was used for further analyses.

Additional information is provided in Supplementary Materials and Methods.

Immunophenotype heterogeneity of consecutive melanoma metastases

Patient model Ma-Mel-86 was selected to dissect the genetic evolution of melanoma immunogenicity in the course of disease progression. This patient presented with primary melanoma in November 2001, but progression to stage III and stage IV was diagnosed already in January and September 2002, respectively. From then on, the patient developed multiple lymph node lesions and visceral metastases in brain, breast, and intestine (Fig. 1A). Accessible lymph node lesions and distant organ metastases were excised, while brain metastases regressed in response to irradiation. The patient received different types of immunotherapies, including IFNα, peptide-based vaccine (Supplementary Table S2), and tumor lysate–loaded dendritic cell vaccine, which together with surgical interventions led to a 3-year clinically disease-free interval from 2005 to 2008. In December 2008, the patient developed recurrent lesions and died from metastatic disease in April 2009.

Figure 1.

Recurrent melanoma metastases of patient Ma-Mel-86 show heterogeneous HLA class I expression. A, clinical history of melanoma patient Ma-Mel-86. Lymph node (LN) metastases Ma-Mel-86a, Ma-Mel-86b, Ma-Mel-86c, and Ma-Mel-86f were excised and cell lines established from corresponding tissue samples. Gray area, disease stage IV. B, expression of HLA class I antigen complexes on melanoma cell lines was analyzed by flow cytometry. Representative histograms from one of three independent experiments are shown. C, expression of melanoma marker HMB-45, HLA class I antigens, and T-cell marker CD3 in serial cryostat tissue sections of melanoma metastases Ma-Mel-86c and Ma-Mel-86f, determined by IHC. Red color, marker-positive cells.

Figure 1.

Recurrent melanoma metastases of patient Ma-Mel-86 show heterogeneous HLA class I expression. A, clinical history of melanoma patient Ma-Mel-86. Lymph node (LN) metastases Ma-Mel-86a, Ma-Mel-86b, Ma-Mel-86c, and Ma-Mel-86f were excised and cell lines established from corresponding tissue samples. Gray area, disease stage IV. B, expression of HLA class I antigen complexes on melanoma cell lines was analyzed by flow cytometry. Representative histograms from one of three independent experiments are shown. C, expression of melanoma marker HMB-45, HLA class I antigens, and T-cell marker CD3 in serial cryostat tissue sections of melanoma metastases Ma-Mel-86c and Ma-Mel-86f, determined by IHC. Red color, marker-positive cells.

Close modal

Tumor samples from patient Ma-Mel-86 were obtained at different time points in the course of disease, including an early-stage III lymph node metastasis (Ma-Mel-86a) excised 2 months after diagnosis and three late recurrent lymph node lesions excised in years 1.5 (Ma-Mel-86b), 3 (Ma-Mel-86c), and 6 (Ma-Mel-86f) of stage IV disease (Fig. 1A). From these metastases, cell lines were established that, at first, were studied for the surface expression of HLA class I molecules as a precondition to T-cell recognition. As shown in Fig. 1B, tumor cells from early metastasis Ma-Mel-86a and the late recurrent lesion Ma-Mel-86c expressed HLA class I molecules, whereas those from lesions Ma-Mel-86b and Ma-Mel-86f were HLA class I negative.

In line with the phenotype of the cell lines, staining of corresponding tissue sections with antibody W6/32, binding a structural epitope formed by complexes of the β2m light chain and the HLA class I heavy chains revealed HLA class I expression by tumor cells from metastases Ma-Mel-86a (19) and Ma-Mel-86c (Fig. 1C), whereas lesions Ma-Mel-86b (19) and Ma-Mel-86f (Fig. 1C) contained HLA class I–negative melanoma cells. By staining for CD3, we found T cells to be present in both metastases, Ma-Mel-86c and Ma-Mel-86f (Fig. 1C), but screening of additional tissue slices from metastasis Ma-Mel-86f suggested location of CD3+ T cells primarily in the tumor periphery (Supplementary Fig. S1).

Divergent genetic evolution of early and late metastatic melanoma cells

To demonstrate their common origin, we screened genomic DNA from the distinct melanoma cell lines by targeted sequencing for known recurrent mutations in a total of 29 genes (Supplementary Table S1). As listed in Fig. 2A, all cell lines showed the same nucleotide alterations in BRAF, CDK4, MAP2K1, PTEN, and p53. To further define their genetic relationship, a genome-wide SNP array analysis on genomic DNA from the tumor cell lines and autologous PBMC as a constitutive normal control was performed. The phylogenetic tree based on maximum parsimony showed that tumor cells from the late recurrent lesions Ma-Mel-86b, Ma-Mel-86c, and Ma-Mel-86f were closely related to each other. This grouping was present in 99% of the 500 bootstrap replicates of the analysis (Fig. 2B). Phylogenetic inference using maximum likelihood reached the same branching order and similar bootstrap support, indicating that the late tumor recurrences had separated from the Ma-Mel-86a lineage early on.

Figure 2.

Late recurrent metastases cluster in a phylogenetic subgroup distinct from the early lesion. A, mutations in melanoma cell lines defined by amplicon sequencing, known recurrent mutations among 29 genes tested. B, maximum parsimony tree showing the phylogenetic relationship of the melanoma cell lines and the autologous blood sample (reference) used as outgroup. A melanoma lineage genetically divergent from the blood sample evolved in patient Ma-Mel-86, leading to the studied cell lines. The lineage giving rise to Ma-Mel-86a cells diverged from the melanoma ancestor and accumulated specific genotypic differences. Later, the lineages of Ma-Mel-86b, Ma-Mel-86c, and Ma-Mel-86f cells diverged. Ninety-nine percent of the bootstrap replicates showed this grouping. C, expression of indicated proteins in the different melanoma cell lines determined by Western blot analysis. GAPDH served as loading control. Representative data from one of three independent experiments are shown. D, expression profile of candidate genes compared for all cell lines and plotted as heatmap. Values represent RPKM (reads per kilobase of transcript per million mapped reads) normalized read counts. E, release of IL6 and IL8 by melanoma cell lines, measured by ELISA. Data represent means (+SEM) of at least two independent experiments.

Figure 2.

Late recurrent metastases cluster in a phylogenetic subgroup distinct from the early lesion. A, mutations in melanoma cell lines defined by amplicon sequencing, known recurrent mutations among 29 genes tested. B, maximum parsimony tree showing the phylogenetic relationship of the melanoma cell lines and the autologous blood sample (reference) used as outgroup. A melanoma lineage genetically divergent from the blood sample evolved in patient Ma-Mel-86, leading to the studied cell lines. The lineage giving rise to Ma-Mel-86a cells diverged from the melanoma ancestor and accumulated specific genotypic differences. Later, the lineages of Ma-Mel-86b, Ma-Mel-86c, and Ma-Mel-86f cells diverged. Ninety-nine percent of the bootstrap replicates showed this grouping. C, expression of indicated proteins in the different melanoma cell lines determined by Western blot analysis. GAPDH served as loading control. Representative data from one of three independent experiments are shown. D, expression profile of candidate genes compared for all cell lines and plotted as heatmap. Values represent RPKM (reads per kilobase of transcript per million mapped reads) normalized read counts. E, release of IL6 and IL8 by melanoma cell lines, measured by ELISA. Data represent means (+SEM) of at least two independent experiments.

Close modal

The divergent evolution was also detectable at the level of cellular differentiation. Only late tumor recurrences expressed the lineage-specific transcriptional activator MITF and its target gene Melan-A/MART-1. In contrast, expression of receptor tyrosine kinase AXL and transcription factor c-Jun was restricted to Ma-Mel-86a cells (Fig. 2C). Recently, it was demonstrated that AXLhigh/MITFlow melanoma cells are characterized by elevated NF-κB activity (20), similar to c-Junhigh/MITFlow cells, inducing expression of proinflammatory genes (21). Consistently, analyses of transcriptome data revealed expression of cytokines IL1B, IL6, and chemokine CCL2 predominantly in Ma-Mel-86a cells (Fig. 2D). By ELISA, release of IL6 was confirmed (Fig. 2E). Thus, despite the origin from a common ancestor, early and late metastases formed genetically and phenotypically distinct groups.

Independent acquisition of B2M gene deletions by metastases Ma-Mel-86b and Ma-Mel-86f

Despite their close genetic relationship (Fig. 2B), the late recurrent melanoma cells displayed profound differences in their HLA class I phenotype. Ma-Mel-86b and Ma-Mel-86f cells, in contrast to Ma-Mel-86c, lost HLA class I surface expression due to a lack in β2m protein and corresponding mRNA expression (Fig. 3A and B). HLA class I surface expression on both cell lines could not be restored in response to type I or type II IFN treatment (data not shown) but was achieved upon transfection of Ma-Mel-86b and Ma-Mel-86f cells with a B2M expression plasmid (Supplementary Fig. S2). Previously, we demonstrated that loss of β2m expression in melanoma is associated with aberrations in chromosome 15q, to which the B2M gene maps at 15q21.1 (15, 22, 23). Indeed, by SNP array, the same large deletion on chromosome 15q encompassing the region 15q11.2 to 15q22.31 was detected in Ma-Mel-86b and Ma-Mel-86f cells, pointing to a shared chromosomal aberration associated with loss of one B2M gene (Fig. 3C and D). In addition, each of the cell lines displayed a specific small deletion, affecting the second B2M gene. In Ma-Mel-86b cells, the B2M gene and flanking sequences were lost, whereas in Ma-Mel-86f cells, the deletion affected only the B2M gene (Fig. 3E). This suggested that β2m deficiency in Ma-Mel-86b and Ma-Mel-86f cells was due to a shared chromosome 15q aberration, acquired by a common precursor, and a subsequent cell line–specific B2M gene deletion. Thus, the late recurrent Ma-Mel-86b and Ma-Mel-86f cells independently acquired their HLA class I–negative phenotype, gaining complete resistance against tumor antigen–specific CD8+ T cells.

Figure 3.

Ma-Mel-86b and Ma-Mel-86f cells independently acquired their HLA class I–negative phenotype. A, expression of β2m in melanoma cells determined by Western blot analysis. GAPDH served as loading control. Representative data from one of three independent experiments. B,B2M mRNA levels in the different cell lines quantified by qRT-PCR. Expression of B2M mRNA normalized to endogenous GAPDH mRNA, depicted relative to the expression in Ma-Mel-86a cells. Data from one of two independent experiments are shown. C and D, SNP given as copy number (C) and allelic distribution (D) of chromosome 15q are shown for DNA obtained from the melanoma cell lines and autologous PBMC (germline). Ma-Mel-86b and Ma-Mel-86f cells show a large deletion (42,982,458 bp) on chromosome 15q encompassing the region 15q11.2–15q22.31 (Chr.15: 22,752,398–65,734,856; hg19). The location of B2M at 15q21.1 (Chr.15: 45,003,675–45,011,075) is shown by the dashed line. E, higher magnification of the B2M gene region given as copy number (left) and log2 allelic display (right).

Figure 3.

Ma-Mel-86b and Ma-Mel-86f cells independently acquired their HLA class I–negative phenotype. A, expression of β2m in melanoma cells determined by Western blot analysis. GAPDH served as loading control. Representative data from one of three independent experiments. B,B2M mRNA levels in the different cell lines quantified by qRT-PCR. Expression of B2M mRNA normalized to endogenous GAPDH mRNA, depicted relative to the expression in Ma-Mel-86a cells. Data from one of two independent experiments are shown. C and D, SNP given as copy number (C) and allelic distribution (D) of chromosome 15q are shown for DNA obtained from the melanoma cell lines and autologous PBMC (germline). Ma-Mel-86b and Ma-Mel-86f cells show a large deletion (42,982,458 bp) on chromosome 15q encompassing the region 15q11.2–15q22.31 (Chr.15: 22,752,398–65,734,856; hg19). The location of B2M at 15q21.1 (Chr.15: 45,003,675–45,011,075) is shown by the dashed line. E, higher magnification of the B2M gene region given as copy number (left) and log2 allelic display (right).

Close modal

HLA class I–negative tumor cells could still be targeted by HLA class II–restricted CD4+ T cells or innate CD56+ natural killer (NK)/NKT cells. However, both HLA class I–negative cell lines lacked constitutive HLA class II expression, except for a small subpopulation in Ma-Mel-86b, and CD4+ T cells could not be detected in metastasis Ma-Mel-86f (Supplementary Fig. S3). Furthermore, CD56+ NK/NKT cells, if present, could not be distinguished from CD56-expressing Ma-Mel-86f cells. Generally, we detect very few or no CD56+ lymphocytes in HLA class I–positive (Supplementary Fig. S3) and HLA class I–negative melanoma metastases (data not shown).

Stimulation of distinct T-cell repertoires by Ma-Mel-86a and Ma-Mel-86c cells

Of the four metastases, we detected HLA class I antigen expression only on Ma-Mel-86a and Ma-Mel-86c cells (Fig. 1B). The latency period of more than 3 years between the outgrowth of both metastases as well as their divergent genetic evolution led us to ask to what extent, differences in their T cell–stimulatory capacity could be observed. To this end, autologous MLTCs were set up, in which peripheral blood CD8+ T cells of the patient, obtained in 2004, were stimulated twice with irradiated Ma-Mel-86a or Ma-Mel-86c cells for enrichment of tumor-reactive T cells. By screening for T-cell responses to the different tumor cell lines, we observed that around 14% of CD8+ T cells from MLTC with Ma-Mel-86a cells (MLTC-86a) released IFNγ in the presence of Ma-Mel-86a cells, but only 1.5% of these T cells reacted toward Ma-Mel-86c cells (Fig. 4A). On the other hand, approximately 13% of the CD8+ T cells from MLTC-86c responded toward Ma-Mel-86c melanoma cells, but none of these T cells reacted toward Ma-Mel-86a cells (Fig. 4B). The same pattern of T-cell reactivity was observed in three independent MLTC experiments (Fig. 4C), indicating that each of the melanoma cell lines stimulated the outgrowth of a specific T-cell repertoire with very limited to no cross-reactivity toward autologous tumor cells from distinct metastases.

Figure 4.

Ma-Mel-86a and Ma-Mel-86c cells stimulate their specific T-cell repertoire. A and B, in MLTC, CD8+ T cells were stimulated with Ma-Mel-86a (MLTC-86a) or Ma-Mel-86c (MLTC-86c) cells. After two stimulations, T-cell reactivity toward the indicated target cells was determined by intracellular staining for IFNγ. A, first dot plot (left), spontaneous IFNγ production by CD8+ T cells from MLTC-86a; second and third dot plot, production of IFNγ in response to the different target cells indicated above. B, first dot plot (left), spontaneous IFNγ production by CD8+ T cells from MLTC-86c; second to third dot plot, production of IFNγ in response to the different target cells indicated above. Representative results from one of three independent experiments are shown; numbers in dot plots indicate percentage of IFNγ+ CD8+ T cells. C, results from three independent experiments depicted as mean (+SEM) of percent IFNγ+ cells in CD8+CD3+ cells. *, P < 0.05. D and E, results from high-throughput T-cell receptor β chain sequencing. D, clonal T-cell receptor β repertoire of MLTC-86a and MLTC-86c. E, selected intermediate to low frequency clones shared between both MLTC.

Figure 4.

Ma-Mel-86a and Ma-Mel-86c cells stimulate their specific T-cell repertoire. A and B, in MLTC, CD8+ T cells were stimulated with Ma-Mel-86a (MLTC-86a) or Ma-Mel-86c (MLTC-86c) cells. After two stimulations, T-cell reactivity toward the indicated target cells was determined by intracellular staining for IFNγ. A, first dot plot (left), spontaneous IFNγ production by CD8+ T cells from MLTC-86a; second and third dot plot, production of IFNγ in response to the different target cells indicated above. B, first dot plot (left), spontaneous IFNγ production by CD8+ T cells from MLTC-86c; second to third dot plot, production of IFNγ in response to the different target cells indicated above. Representative results from one of three independent experiments are shown; numbers in dot plots indicate percentage of IFNγ+ CD8+ T cells. C, results from three independent experiments depicted as mean (+SEM) of percent IFNγ+ cells in CD8+CD3+ cells. *, P < 0.05. D and E, results from high-throughput T-cell receptor β chain sequencing. D, clonal T-cell receptor β repertoire of MLTC-86a and MLTC-86c. E, selected intermediate to low frequency clones shared between both MLTC.

Close modal

Thus, we studied the TCR repertoire usage of MLTC-86a and MLTC-86c. High-throughput TCRβ sequencing revealed 850,400 and 1,055,825 sequencing reads, representing an estimated number of 64,853 and 78,245 total haploid nuclear genome copies. After discarding 11.8% and 6.3% of the sequences due to errors, resulting in premature stop codons and open reading frame shifts, as well as filtering for unique sequence haplotypes, a total number of 515 and 336 unique and productive TCRβ clonotypes were detected in MLTC-86a and MLTC-86c, respectively (data not shown). Both repertoires showed a high clonality score of 0.46 and 0.59. In MLTC-86a, one clone was present at a frequency of 28.1%; this clone, however, was not detectable in MLTC-86c. In MLTC-86c, two different clonotypes expanded to approximately 20% each, again neither of the clonotypes was present in MLTC-86a (Fig. 4D). In addition to those high frequency–specific TCRβ clonotypes, a number of shared TCRβ clonotypes expanded to very low frequencies in both MLTCs (Fig. 4E), indicating that some subdominant T-cell epitopes remained present in both Ma-Mel-86a or Ma-Mel-86c cells.

Amplification of differentiation antigen-specific CD8+ T cells by Ma-Mel-86c cells

The amplification of cell line–specific T-cell repertoires led us to ask for the underlying mechanisms. We first studied melanoma cells for the surface expression of inhibitory PD-L1, dampening the proliferation and effector function of PD1+ T cells (24). Comparable PD-L1 expression levels were detected on Ma-Mel-86a and Ma-Mel-86c cells, despite Ma-Mel-86a cells having elevated levels of the PD-L1 transcriptional activators c-Jun and pSTAT3 (Fig. 2C and Supplementary Fig. S4; refs. 25, 26). This suggested that differences in the strength of inhibitory signaling via the PD-L1/PD1 axis did not account for the amplification of different T-cell repertoires (27).

CD8+ T-cell responses in melanoma patients are frequently directed toward melanoma differentiation antigens (MDA), including gp100, Melan-A/MART-1, TRP1, TRP2, and tyrosinase. We detected MDA protein expression in Ma-Mel-86c but not Ma-Mel-86a cells (Fig. 5A), consistent with the results obtained by transcriptome analyses (Fig. 2D). This led us to assume that the pool of Ma-Mel-86c–reactive T cells contained MDA-specific CD8+ T cells not responding to Ma-Mel-86a cells due to its stable dedifferentiation. To analyze this, we transiently transfected Ma-Mel-86a cells with MDA-encoding expression plasmids and used the transfectants as stimulators for bulk T cells from MLTC-86c. As shown in Fig. 5B, Ma-Mel-86a cells transiently expressing tyrosinase were recognized by a large proportion of the T cells from MLTC-86c in contrast to Ma-Mel-86a control cells. The same result was obtained also with a tyrosinase-specific CD8+ T-cell clone established from MLTC-86c (Fig. 5C). Thus, the T-cell repertoire amplified by Ma-Mel-86c cells was dominated by MDA-specific CD8+ T cells, not responding to dedifferentiated Ma-Mel-86a cells.

Figure 5.

Ma-Mel-86c cells amplify MDA-reactive CD8+ T cells. A, melanoma cells Ma-Mel-86a and Ma-Mel-86c analyzed for MDA expression (Melan-A/MART-1, gp100, TRP2, tyrosinase) by Western blot analysis. GAPDH served as loading control. Representative data from one of three independent experiments are shown. B, Ma-Mel-86a cells, transiently transfected with a control (ctrl) vector or expression plasmids encoding Melan-A, TRP1, TRP2, or tyrosinase (Tyro), cocultured with T cells from MLTC-86c. After a 4-hour coincubation, T cells were stained for intracellular IFNγ. Numbers in dot plots indicate percentage of IFNγ+ CD8+ T cells. Representative data from one of three independent experiments are shown. C, activation of an autologous tyrosinase-specific CD8+ T-cell clone (Tc) by different target cells determined by IFNγ ELISpot. Each dot represents a cytokine-releasing T cell. Representative data from one of two independent experiments depicted. D, expression of tyrosinase and Melan-A in serial cryostat tissue sections of metastasis Ma-Mel-86c determined by IHC. Red, marker-positive cells. E and F, Ma-Mel-86c cells treated with IFNγ (50 U/mL) in a 2-day interval for 7 days. E, expression of tyrosinase in IFNγ-treated and control Ma-Mel-86c cells, determined by Western blot anaylsis. GAPDH served as loading control. Representative data from one of three independent experiments are shown. F, activation of tyrosinase-specific CD8+ T-cell clones in the presence of IFNγ-treated or control Ma-Mel-86c cells, determined by IFNγ ELISpot assay. Mean values (+SEM) of triplicates from one of four independent experiments depicted. *, P < 0.05. G, coexpression of CD8 (brown) and PD-L1 (red) in serial cryostat tissue section of metastasis Ma-Mel-86c determined by IHC.

Figure 5.

Ma-Mel-86c cells amplify MDA-reactive CD8+ T cells. A, melanoma cells Ma-Mel-86a and Ma-Mel-86c analyzed for MDA expression (Melan-A/MART-1, gp100, TRP2, tyrosinase) by Western blot analysis. GAPDH served as loading control. Representative data from one of three independent experiments are shown. B, Ma-Mel-86a cells, transiently transfected with a control (ctrl) vector or expression plasmids encoding Melan-A, TRP1, TRP2, or tyrosinase (Tyro), cocultured with T cells from MLTC-86c. After a 4-hour coincubation, T cells were stained for intracellular IFNγ. Numbers in dot plots indicate percentage of IFNγ+ CD8+ T cells. Representative data from one of three independent experiments are shown. C, activation of an autologous tyrosinase-specific CD8+ T-cell clone (Tc) by different target cells determined by IFNγ ELISpot. Each dot represents a cytokine-releasing T cell. Representative data from one of two independent experiments depicted. D, expression of tyrosinase and Melan-A in serial cryostat tissue sections of metastasis Ma-Mel-86c determined by IHC. Red, marker-positive cells. E and F, Ma-Mel-86c cells treated with IFNγ (50 U/mL) in a 2-day interval for 7 days. E, expression of tyrosinase in IFNγ-treated and control Ma-Mel-86c cells, determined by Western blot anaylsis. GAPDH served as loading control. Representative data from one of three independent experiments are shown. F, activation of tyrosinase-specific CD8+ T-cell clones in the presence of IFNγ-treated or control Ma-Mel-86c cells, determined by IFNγ ELISpot assay. Mean values (+SEM) of triplicates from one of four independent experiments depicted. *, P < 0.05. G, coexpression of CD8 (brown) and PD-L1 (red) in serial cryostat tissue section of metastasis Ma-Mel-86c determined by IHC.

Close modal

Tyrosinase-specific CD8+ T cells had the capability of killing Ma-Mel-86c cells (Supplementary Fig. S5), but low in vivo antigen expression could have limited their antitumor activity (Fig. 5D). Interestingly, we observed a strong decrease of tyrosinase expression in IFNγ-treated Ma-Mel-86c cells, abrogating antigen-specific T-cell recognition (Fig. 5E and F). This mechanism could also have applied to metastasis Ma-Mel-86c, as PD-L1–positive tumor cells were detected primarily in the vicinity of CD8+ T cells, indicating IFNγ-mediated PD-L1 upregulation by activated T cells, as described previously (Fig. 5G; ref. 28).

HLA haplotype loss in Ma-Mel-86c cells leads to resistance against CD8+ T cells attacking early disease stage tumor cells

CD8+ T cells from MLTC-86a killed Ma-Mel-86a cells but ignored melanoma cells from metastasis Ma-Mel-86c (Fig. 4A and C and Supplementary Fig. S6). Screening Ma-Mel-86c cells for mechanisms protecting from recognition by the MLTC-86a–specific T-cell repertoire, we asked for differences in the intensity and pattern of HLA class I expression. As shown in Fig. 6A, HLA class I expression levels on Ma-Mel-86c cells were lower compared with Ma-Mel-86a cells. Screening SNP array data for molecular alterations in the HLA class I region on chromosome 6p22.1–6p21.3, we detected a large deletion on one chromosome 6 in Ma-Mel-86c cells ranging from 6p25.3 to 6p21.1, expected to be associated with an HLA haplotype loss (Fig. 6B). Fittingly, HLA genotyping on DNA from both cell lines and autologous PBMC revealed loss of the HLA-A*24:02, HLA-B*15:01, HLA-Cw*03:03 haplotype in Ma-Mel-86c cells, whereas all HLA alleles were present in Ma-Mel-86a cells (Fig. 6C). On the basis of this result, we concluded that the haplotype loss protected Ma-Mel-86c cells from recognition by CD8+ T cells of MLTC-86a. Indeed, T cells from MLTC-86a became activated in the presence of Ma-Mel-86c cells transiently reexpressing the HLA-A*24:02 and HLA-B*15:01 alleles (Fig. 6D). Notably, none of the tested MLTC T cells recognized vaccine peptides applied in 2002 (Supplementary Table S2; data not shown).

Figure 6.

HLA haplotype loss protects Ma-Mel-86c cells from Ma-Mel-86a–reactive T cells. A, HLA class I expression levels on Ma-Mel-86a and Ma-Mel-86c cells determined by flow cytometry. Representative data from one of two independent experiments are shown. B, SNP results of chromosome 6p shown for DNA obtained from Ma-Mel-86a, Ma-Mel-86c, and autologous PBMC (germline). Top, allelic display; bottom, magnification of the HLA gene region. Ma-Mel-86c cells show a large deletion (40,998,052 bp) on chromosome 6 encompassing the region 6p25.3 to 6p21.1 (Chr.6: 156,974–41,155,026; hg19), including the HLA-A, HLA-B, and HLA-C genes. Dashed lines, location of HLA genes. C, HLA class I genotype of Ma-Mel-86a and Ma-Mel-86c cells. D, reactivity of T cells from MLTC-86a toward Ma-Mel-86a cells as well as Ma-Mel-86c cells transfected with expression plasmids encoding HLA-A*24:02 (pHLA-A*24), HLA-B*15:01 (pHLA-B*15), HLA-C*03:03 (pHLA-C*03), or an empty control vector (pcDNA) determined by IFNγ ELISpot assay. Representative data from one of two independent experiments are shown.

Figure 6.

HLA haplotype loss protects Ma-Mel-86c cells from Ma-Mel-86a–reactive T cells. A, HLA class I expression levels on Ma-Mel-86a and Ma-Mel-86c cells determined by flow cytometry. Representative data from one of two independent experiments are shown. B, SNP results of chromosome 6p shown for DNA obtained from Ma-Mel-86a, Ma-Mel-86c, and autologous PBMC (germline). Top, allelic display; bottom, magnification of the HLA gene region. Ma-Mel-86c cells show a large deletion (40,998,052 bp) on chromosome 6 encompassing the region 6p25.3 to 6p21.1 (Chr.6: 156,974–41,155,026; hg19), including the HLA-A, HLA-B, and HLA-C genes. Dashed lines, location of HLA genes. C, HLA class I genotype of Ma-Mel-86a and Ma-Mel-86c cells. D, reactivity of T cells from MLTC-86a toward Ma-Mel-86a cells as well as Ma-Mel-86c cells transfected with expression plasmids encoding HLA-A*24:02 (pHLA-A*24), HLA-B*15:01 (pHLA-B*15), HLA-C*03:03 (pHLA-C*03), or an empty control vector (pcDNA) determined by IFNγ ELISpot assay. Representative data from one of two independent experiments are shown.

Close modal

Overall, we demonstrated that lesions growing out after long-term latency in patient Ma-Mel-86 were protected from effector functions of tumor antigen–specific T cells by HLA haplotype loss or total HLA class I loss, the latter independently acquired by two different melanoma metastases.

The mechanisms keeping tumor cells in latency are poorly understood but are most likely diverse, including a limited blood supply or an equilibrium of apoptosis and proliferation as observed in different mouse tumor models (29). CD8+ T cells seem to contribute to equilibrium maintenance by directly killing tumor cells and releasing antiproliferative cytokines (10, 11, 30). Thus, to switch from latency to proliferation, tumor cells either need to blunt T-cell activity by establishing an immunosuppressive microenvironment or have to acquire genetic/epigenetic alterations allowing escape from direct T-cell recognition (9).

Our study on melanoma recurrences suggests that tumor antigen–specific CD8+ T cells modify tumor cell immunogenicity by selective enrichment of genetically altered poorly immunogenic variants. We found melanoma cells forming overt metastases in patient Ma-Mel-86 after a latency period of approximately 1.5, 3, and 6 years in stage IV disease to be T cell–resistant, albeit at varying degrees. Of the three late recurrent lesions, Ma-Mel-86b and Ma-Mel-86f, excised in years 1.5 and 6, respectively, were completely T cell–resistant due to the loss of β2m expression. On the basis of genome profiling by SNP arrays, we determined the chronology of genetic alterations leading to β2m deficiency as follows: a common ancestor of Ma-Mel-86b and Ma-Mel-86f cells acquired a large deletion on one chromosome 15q encompassing the B2M gene. From this ancestor, the two cell lines diverged and independently acquired additional small deletions affecting the second B2M allele. Recently, we detected a similar chronology of genetic alterations in two additional patient models (15), identifying chromosome 15 alterations as early genetic events predisposing to total HLA class I loss in melanoma.

A remarkable fraction of melanoma cells shows constitutive HLA class II expression (31), sensitizing to CD4+ T cells, of which a subgroup has perforin/granzyme–mediated cytotoxic activity against tumor cells (32, 33). In addition, CD4+ T cells can halt tumor growth and even eradicate established tumors by release of antiproliferative and proapoptotic cytokines (IFNγ/TNFα) acting on tumor and stroma cells (34, 35). HLA class I–negative Ma-Mel-86 melanoma cells lacked constitutive but still showed IFNγ-inducible HLA class II expression (19). NK cells targeting HLA class I–negative tumor cells could be a source of IFNγ, however rarely infiltrate melanoma lesions. In this regard, strategies mobilizing NK cells into the tumor, such as inhibitor treatment of BRAFV600E-mutant melanoma (36, 37), local irradiation (38), or localized virotherapy (39), are of potential interest. Such treatments may also attract CD4+ and CD8+ T cells that even in the absence of HLA class I/II expression on tumor cells could eradicate lesions by recognition of cross-presented tumor antigen on stroma cells (40–42).

In contrast to metastases Ma-Mel-86b and Ma-Mel-86f, tumor cells derived from lesion Ma-Mel-86c, excised in year 3 of stage IV disease, still presented HLA class I antigens. Accordingly, Ma-Mel-86c cells were capable of stimulating autologous CD8+ T cells, similar to HLA class I–positive Ma-Mel-86a cells obtained from an early stage III metastasis. However, in short-term cocultures with autologous CD8+ T cells, each of the HLA class I–positive melanoma cell lines amplified a specific T-cell repertoire with distinct dominant T-cell clones. T lymphocytes stimulated twice with Ma-Mel-86c cells were dominated by MDA-specific T cells that did not cross-react to Ma-Mel-86a cells, which completely lacked MDA expression. In contrast, T cells stimulated twice with Ma-Mel-86a cells were dominated by a T-cell repertoire recognizing their cognate antigens in the context of the HLA-A*24:02, HLA-B*15:01, HLA-C*03:03 haplotype. Because of an aberration in chromosome 6p, this haplotype was lost in Ma-Mel-86c cells, protecting them from Ma-Mel-86a–reactive T cells. Despite the haplotype loss, Ma-Mel-86c cells were still recognized by autologous tyrosinase-specific CD8+ T cells. Weak expression of the differentiation antigen in metastasis Ma-Mel-86c, potentially induced by T cell–derived IFNγ, might have limited T-cell efficacy in vivo. Furthermore, expression of this self-antigen in the thymus generates a tolerized specific T-cell repertoire that might not be capable of preventing metastasis progression (43). In contrast, T cells recognizing mutant neoantigens can effectively mediate tumor regression as demonstrated in different mouse tumor models but can also select for an enrichment of immune escape variants (6–8, 44). The potential contribution of neoantigens to immunoediting in the Ma-Mel-86 melanoma model remains to be determined.

A broader relevance of chromosome 6p alterations in melanoma immune escape is indicated by different studies detecting HLA haplotype loss in late recurrent melanoma cells from a long-term survivor and in metastases from patients after immunotherapy (13, 45, 46). Recently, Rooney and colleagues found mutations in genes involved in antigen presentation (B2M, HLA-A, HLA-B, HLA-C) to be enriched in tumors with immune-cytolytic activity (47). On the basis of all these data, we assume that continuous CD8+ T-cell activity kept the precursors of Ma-Mel-86b, Ma-Mel-86c, and Ma-Mel-86f cells in latency and that only upon acquisition of HLA alterations, overt metastases formation occurred, although an additional contribution of immunosuppressive mechanisms is still possible.

The establishment of an immunosuppressive microenvironment might have supported the outgrowth of the stage III lesion Ma-Mel-86a. Our data demonstrated that these early metastatic cells differed from late recurrent tumor cells in terms of genetic alterations, as defined by SNP array and their c-Junhigh/MITFlow phenotype. This phenotype has been associated with elevated NF-κB activity, which induces expression of proinflammatory genes recruiting myeloid cells, and shares similarity to the AXLhigh/MITFlow melanoma phenotype showing high invasive capacity (20, 21, 48). We could not study the immune cell composition of metastasis Ma-Mel-86a due to the lack of corresponding tissue but detected expression of cytokines IL1B, IL6, and chemokine CCL2 only in Ma-Mel-86a cells. CCL2 is an attractant of macrophages and myeloid-derived suppressor cells generating an immunosuppressive microenvironment (49). This is in line with a recent report by Hugo and colleagues showing an association of the AXLhigh/MITFlow phenotype with resistance to checkpoint-blocking anti-PD1 antibody therapy (3, 4, 50). Approximately 30% of patients objectively respond to therapy with anti-PD1 antibodies, including a remarkable number of long-term responders. However, some patients show disease recurrence after a long period of therapy-induced regression (3). Although the underlying mechanisms are so far unknown, it is expected that some tumor cells might have acquired specific genetic alterations that directly interfere with T-cell recognition.

In summary, the comparative analysis of tumor cells derived from an early metastasis and multiple later recurrences has allowed us to identify genetic alterations associated with evolving T-cell resistance during metastatic latency. Our data suggest that under the selective pressure of tumor antigen–specific T cells, individual immunoevasive subclones can emerge, generating a tumor heterogeneity counteracting complete tumor eradication by immunotherapy. New technologies directed toward identifying mechanisms of genetic and functional T-cell resistance in longitudinally collected tumor biopsies could potentially enable early appropriate adaptation of treatment regimens, resulting in higher numbers of long-term therapy responders.

J.C. Becker has received speakers bureau honoraria from Amgen and EMD and is a consultant/advisory board member for Amgen, Bristol-Myers Squibb, and EMD. No potential conflicts of interest were disclosed by the other authors.

Conception and design: F. Zhao, J.C. Becker, D. Schadendorf, K. Griewank, A. Paschen

Development of methodology: A. Sucker, S. Horn, M. Stiller, J.C. Becker, T. Wölfel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Zhao, A. Sucker, C. Heeke, N. Bielefeld, B. Schrörs, G. Gaudernack, J.C. Becker, D. Schadendorf

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Zhao, A. Sucker, S. Horn, C. Heeke, B. Schrörs, A. Bicker, M. Stiller, J.C. Becker, V. Lennerz, D. Schadendorf, K. Griewank, A. Paschen

Writing, review, and/or revision of the manuscript: F. Zhao, S. Horn, A. Roesch, G. Gaudernack, M. Stiller, J.C. Becker, V. Lennerz, T. Wölfel, D. Schadendorf, K. Griewank, A. Paschen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Schrörs, M. Lindemann, T. Wölfel, D. Schadendorf

The authors thank Dr. Klein-Hitpass for performing the SNP array analyses.

This study was supported by the “Hiege Stiftung gegen Hautkrebs” (A. Paschen), “Stiftung Rheinland-Pfalz für Innovation” (961-386261/1049 to T. Wölfel), and Deutsche Krebshilfe “Translational Oncology” (111546 to A. Paschen, D. Schadendorf, and T. Wölfel).

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.

1.
Rosenberg
SA
,
Yang
JC
,
Sherry
RM
,
Kammula
US
,
Hughes
MS
,
Phan
GQ
, et al
Durable complete responses in heavily pretreated patients with metastatic melanoma using t-cell transfer immunotherapy
.
Clin Cancer Res
2011
;
17
:
4550
7
.
2.
Hodi
FS
,
O'Day
SJ
,
McDermott
DF
,
Weber
RW
,
Sosman
JA
,
Haanen
JB
, et al
Improved survival with ipilimumab in patients with metastatic melanoma
.
N Engl J Med
2010
;
363
:
711
23
.
3.
Topalian
SL
,
Sznol
M
,
McDermott
DF
,
Kluger
HM
,
Carvajal
RD
,
Sharfman
WH
, et al
Survival, durable tumor remission, and long-term safety in patients with advanced melanoma receiving nivolumab
.
J Clin Oncol
2014
;
32
:
1020
30
.
4.
Larkin
J
,
Chiarion-Sileni
V
,
Gonzalez
R
,
Grob
JJ
,
Cowey
CL
,
Lao
CD
, et al
Combined nivolumab and ipilimumab or monotherapy in untreated melanoma
.
N Engl J Med
2015
;
373
:
23
34
.
5.
Coulie
PG
,
Van den Eynde
BJ
,
van der Bruggen
P
,
Boon
T
. 
Tumour antigens recognized by T lymphocytes: at the core of cancer immunotherapy
.
Nat Rev Cancer
2014
;
14
:
135
46
.
6.
DuPage
M
,
Cheung
AF
,
Mazumdar
C
,
Winslow
MM
,
Bronson
R
,
Schmidt
LM
, et al
Endogenous T cell responses to antigens expressed in lung adenocarcinomas delay malignant tumor progression
.
Cancer Cell
2011
;
19
:
72
85
.
7.
DuPage
M
,
Mazumdar
C
,
Schmidt
LM
,
Cheung
AF
,
Jacks
T
. 
Expression of tumour-specific antigens underlies cancer immunoediting
.
Nature
2012
;
482
:
405
9
.
8.
Matsushita
H
,
Vesely
MD
,
Koboldt
DC
,
Rickert
CG
,
Uppaluri
R
,
Magrini
VJ
, et al
Cancer exome analysis reveals a T-cell-dependent mechanism of cancer immunoediting
.
Nature
2012
;
482
:
400
4
.
9.
Mittal
D
,
Gubin
MM
,
Schreiber
RD
,
Smyth
MJ
. 
New insights into cancer immunoediting and its three component phases–elimination, equilibrium and escape
.
Curr Opin Immunol
2014
;
27
:
16
25
.
10.
Eyles
J
,
Puaux
AL
,
Wang
X
,
Toh
B
,
Prakash
C
,
Hong
M
, et al
Tumor cells disseminate early, but immunosurveillance limits metastatic outgrowth, in a mouse model of melanoma
.
J Clin Invest
2010
;
120
:
2030
9
.
11.
Koebel
CM
,
Vermi
W
,
Swann
JB
,
Zerafa
N
,
Rodig
SJ
,
Old
LJ
, et al
Adaptive immunity maintains occult cancer in an equilibrium state
.
Nature
2007
;
450
:
903
7
.
12.
Ikeda
H
,
Lethe
B
,
Lehmann
F
,
van Baren
N
,
Baurain
JF
,
de Smet
C
, et al
Characterization of an antigen that is recognized on a melanoma showing partial HLA loss by CTL expressing an NK inhibitory receptor
.
Immunity
1997
;
6
:
199
208
.
13.
Yamshchikov
GV
,
Mullins
DW
,
Chang
CC
,
Ogino
T
,
Thompson
L
,
Presley
J
, et al
Sequential immune escape and shifting of T cell responses in a long-term survivor of melanoma
.
J Immunol
2005
;
174
:
6863
71
.
14.
van de Nes
J
,
Gessi
M
,
Sucker
A
,
Möller
I
,
Stiller
M
,
Horn
S
, et al
Targeted next generation sequencing reveals unique mutation profile of primary melanocytic tumors of the central nervous system
.
J Neurooncol
2016
;
127
:
435
44
.
15.
Sucker
A
,
Zhao
F
,
Real
B
,
Heeke
C
,
Bielefeld
N
,
Maβen
S
, et al
Genetic evolution of T-cell resistance in the course of melanoma progression
.
Clin Cancer Res
2014
;
20
:
6593
604
.
16.
Tamura
K
,
Stecher
G
,
Peterson
D
,
Filipski
A
,
Kumar
S
. 
MEGA6: Molecular evolutionary genetics analysis version 6.0
.
Mol Biol Evol
2013
;
30
:
2725
9
.
17.
Robins
HS
,
Campregher
PV
,
Srivastava
SK
,
Wacher
A
,
Turtle
CJ
,
Kahsai
O
, et al
Comprehensive assessment of T-cell receptor beta-chain diversity in alphabeta T cells
.
Blood
2009
;
114
:
4099
107
.
18.
Carlson
CS
,
Emerson
RO
,
Sherwood
AM
,
Desmarais
C
,
Chung
MW
,
Parsons
JM
, et al
Using synthetic templates to design an unbiased multiplex PCR assay
.
Nat Commun
2013
;
4
:
2680
.
19.
Schwinn
N
,
Vokhminova
D
,
Sucker
A
,
Textor
S
,
Striegel
S
,
Moll
I
, et al
Interferon-gamma down-regulates NKG2D ligand expression and impairs the NKG2D-mediated cytolysis of MHC class I-deficient melanoma by natural killer cells
.
Int J Cancer
2009
;
124
:
1594
604
.
20.
Konieczkowski
DJ
,
Johannessen
CM
,
Abudayyeh
O
,
Kim
JW
,
Cooper
ZA
,
Piris
A
, et al
A melanoma cell state distinction influences sensitivity to MAPK pathway inhibitors
.
Cancer Discov
2014
;
4
:
816
27
.
21.
Riesenberg
S
,
Groetchen
A
,
Siddaway
R
,
Bald
T
,
Reinhardt
J
,
Smorra
D
, et al
MITF and c-Jun antagonism interconnects melanoma dedifferentiation with pro-inflammatory cytokine responsiveness and myeloid cell recruitment
.
Nat Commun
2015
;
6
:
8755
.
22.
Paschen
A
,
Arens
N
,
Sucker
A
,
Greulich-Bode
KM
,
Fonsatti
E
,
Gloghini
A
, et al
The coincidence of chromosome 15 aberrations and beta2-microglobulin gene mutations is causative for the total loss of human leukocyte antigen class I expression in melanoma
.
Clin Cancer Res
2006
;
12
:
3297
305
.
23.
Paschen
A
,
Mendez
RM
,
Jimenez
P
,
Sucker
A
,
Ruiz-Cabello
F
,
Song
M
, et al
Complete loss of HLA class I antigen expression on melanoma cells: a result of successive mutational events
.
Int J Cancer
2003
;
103
:
759
67
.
24.
Wei
F
,
Zhong
S
,
Ma
Z
,
Kong
H
,
Medvec
A
,
Ahmed
R
, et al
Strength of PD-1 signaling differentially affects T-cell effector functions
.
Proc Natl Acad Sci U S A
2013
;
110
:
E2480
9
.
25.
Marzec
M
,
Zhang
Q
,
Goradia
A
,
Raghunath
PN
,
Liu
X
,
Paessler
M
, et al
Oncogenic kinase NPM/ALK induces through STAT3 expression of immunosuppressive protein CD274 (PD-L1, B7-H1)
.
Proc Natl Acad Sci U S A
2008
;
105
:
20852
7
.
26.
Jiang
X
,
Zhou
J
,
Giobbie-Hurder
A
,
Wargo
J
,
Hodi
FS
. 
The activation of MAPK in melanoma cells resistant to BRAF inhibition promotes PD-L1 expression that is reversible by MEK and PI3K inhibition
.
Clin Cancer Res
2013
;
19
:
598
609
.
27.
Simon
S
,
Vignard
V
,
Florenceau
L
,
Dreno
B
,
Khammari
A
,
Lang
F
, et al
PD-1 expression conditions T cell avidity within an antigen-specific repertoire
.
Oncoimmunology
2016
;
5
:
e1104448
.
28.
Taube
JM
,
Anders
RA
,
Young
GD
,
Xu
H
,
Sharma
R
,
McMiller
TL
, et al
Colocalization of inflammatory response with B7-h1 expression in human melanocytic lesions supports an adaptive resistance mechanism of immune escape
.
Sci Transl Med
2012
;
4
:
127ra137
.
29.
Paez
D
,
Labonte
MJ
,
Bohanes
P
,
Zhang
W
,
Benhanim
L
,
Ning
Y
, et al
Cancer dormancy: a model of early dissemination and late cancer recurrence
.
Clin Cancer Res
2012
;
18
:
645
53
.
30.
Matsushita
H
,
Hosoi
A
,
Ueha
S
,
Abe
J
,
Fujieda
N
,
Tomura
M
, et al
Cytotoxic T lymphocytes block tumor growth both by lytic activity and IFNgamma-dependent cell-cycle arrest
.
Cancer Immunol Res
2015
;
3
:
26
36
.
31.
Donia
M
,
Andersen
R
,
Kjeldsen
JW
,
Fagone
P
,
Munir
S
,
Nicoletti
F
, et al
Aberrant expression of MHC class II in melanoma attracts inflammatory tumor-specific CD4+ T- cells, which dampen CD8+ T-cell antitumor reactivity
.
Cancer Res
2015
;
75
:
3747
59
.
32.
Kitano
S
,
Tsuji
T
,
Liu
C
,
Hirschhorn-Cymerman
D
,
Kyi
C
,
Mu
Z
, et al
Enhancement of tumor-reactive cytotoxic CD4+ T cell responses after ipilimumab treatment in four advanced melanoma patients
.
Cancer Immunol Res
2013
;
1
:
235
44
.
33.
Quezada
SA
,
Simpson
TR
,
Peggs
KS
,
Merghoub
T
,
Vider
J
,
Fan
X
, et al
Tumor-reactive CD4(+) T cells develop cytotoxic activity and eradicate large established melanoma after transfer into lymphopenic hosts
.
J Exp Med
2010
;
207
:
637
50
.
34.
Qin
Z
,
Blankenstein
T
. 
CD4+ T cell–mediated tumor rejection involves inhibition of angiogenesis that is dependent on IFN gamma receptor expression by nonhematopoietic cells
.
Immunity
2000
;
12
:
677
86
.
35.
Braumuller
H
,
Wieder
T
,
Brenner
E
,
Aßmann
S
,
Hahn
M
,
Alkhaled
M
, et al
T-helper-1-cell cytokines drive cancer into senescence
.
Nature
2013
;
494
:
361
365
.
36.
Knight
DA
,
Ngiow
SF
,
Li
M
,
Parmenter
T
,
Mok
S
,
Cass
A
, et al
Host immunity contributes to the anti-melanoma activity of BRAF inhibitors
.
J Clin Invest
2013
;
123
:
1371
81
.
37.
de Andrade
L Ferrari
,
Ngiow
SF
,
Stannard
K
,
Rusakiewicz
S
,
Kalimutho
M
,
Khanna
KK
, et al
Natural killer cells are essential for the ability of BRAF inhibitors to control BRAFV600E-mutant metastatic melanoma
.
Cancer Res
2014
;
74
:
7298
308
.
38.
Ni
J
,
Miller
M
,
Stojanovic
A
,
Garbi
N
,
Cerwenka
A
. 
Sustained effector function of IL-12/15/18-preactivated NK cells against established tumors
.
J Exp Med
2012
;
209
:
2351
65
.
39.
Zamarin
D
,
Holmgaard
RB
,
Subudhi
SK
,
Park
JS
,
Mansour
M
,
Palese
P
, et al
Localized oncolytic virotherapy overcomes systemic tumor resistance to immune checkpoint blockade immunotherapy
.
Sci Transl Med
2014
;
6
:
226ra232
.
40.
Spiotto
MT
,
Rowley
DA
,
Schreiber
H
. 
Bystander elimination of antigen loss variants in established tumors
.
Nat Med
2004
;
10
:
294
8
.
41.
Zhang
B
,
Bowerman
NA
,
Salama
JK
,
Schmidt
H
,
Spiotto
MT
,
Schietinger
A
, et al
Induced sensitization of tumor stroma leads to eradication of established cancer by T cells
.
J Exp Med
2007
;
204
:
49
55
.
42.
Zhang
B
,
Karrison
T
,
Rowley
DA
,
Schreiber
H
. 
IFN-gamma- and TNF-dependent bystander eradication of antigen-loss variants in established mouse cancers
.
J Clin Invest
2008
;
118
:
1398
404
.
43.
Gotter
J
,
Brors
B
,
Hergenhahn
M
,
Kyewski
B
. 
Medullary epithelial cells of the human thymus express a highly diverse selection of tissue-specific genes colocalized in chromosomal clusters
.
J Exp Med
2004
;
199
:
155
66
.
44.
Duan
F
,
Lin
Y
,
Liu
C
,
Engelhorn
ME
,
Cohen
AD
,
Curran
M
, et al
Immune rejection of mouse tumors expressing mutated self
.
Cancer Res
2009
;
69
:
3545
53
.
45.
Mendez
R
,
Serrano
A
,
Jager
E
,
Maleno
I
,
Ruiz-Cabello
F
,
Knuth
A
, et al
Analysis of HLA class I expression in different metastases from two melanoma patients undergoing peptide immunotherapy
.
Tissue Antigens
2001
;
57
:
508
19
.
46.
Mendez
R
,
Ruiz-Cabello
F
,
Rodriguez
T
,
Del Campo
A
,
Paschen
A
,
Schadendorf
D
, et al
Identification of different tumor escape mechanisms in several metastases from a melanoma patient undergoing immunotherapy
.
Cancer Immunol Immunother
2007
;
56
:
88
94
.
47.
Rooney
MS
,
Shukla
SA
,
Wu
CJ
,
Getz
G
,
Hacohen
N
. 
Molecular and genetic properties of tumors associated with local immune cytolytic activity
.
Cell
2015
;
160
:
48
61
.
48.
Sensi
M
,
Catani
M
,
Castellano
G
,
Nicolini
G
,
Alciato
F
,
Tragni
G
, et al
Human cutaneous melanomas lacking MITF and melanocyte differentiation antigens express a functional Axl receptor kinase
.
J Invest Dermatol
2011
;
131
:
2448
57
.
49.
Zhang
J
,
Patel
L
,
Pienta
KJ
. 
Targeting chemokine (C-C motif) ligand 2 (CCL2) as an example of translation of cancer molecular biology to the clinic
.
Prog Mol Biol Transl Sci
2010
;
95
:
31
53
.
50.
Hugo
W
,
Zaretsky
JM
,
Sun
L
,
Song
C
,
Moreno
BH
,
Hu-Lieskovan
S
, et al
Genomic and transcriptomic features of response to anti-PD-1 therapy in metastatic melanoma
.
Cell
2016
;
165
:
35
44
.