Acquired IFNγ resistance impairs anti-tumor immunity and gives rise to T-cell-resistant melanoma lesions

Abstract

Melanoma treatment has been revolutionized by antibody-based immunotherapies. IFNγ secretion by CD8⁺ T cells is critical for therapy efficacy having anti-proliferative and pro-apoptotic effects on tumour cells. Our study demonstrates a genetic evolution of IFNγ resistance in different melanoma patient models. Chromosomal alterations and subsequent inactivating mutations in genes of the IFNγ signalling cascade, most often JAK1 or JAK2, protect melanoma cells from anti-tumour IFNγ activity. JAK1/2 mutants further evolve into T-cell-resistant HLA class I-negative lesions with genes involved in antigen presentation silenced and no longer inducible by IFNγ. Allelic JAK1/2 losses predisposing to IFNγ resistance development are frequent in melanoma. Subclones harbouring inactivating mutations emerge under various immunotherapies but are also detectable in pre-treatment biopsies. Our data demonstrate that JAK1/2 deficiency protects melanoma from anti-tumour IFNγ activity and results in T-cell-resistant HLA class I-negative lesions. Screening for mechanisms of IFNγ resistance should be considered in therapeutic decision-making.

Figure 1. Protection from cytokine-induced cell death by acquired IFNγ resistance.

(a) Clinical history of patient Ma-Mel-36. Vertical line, time axis; left, therapeutic regimens; right, primary tumour (PT)/metastases development; arrow indicates cell line established from metastasis Ma-Mel-36; grey box, stage IV disease. (b) IFNγ-sensitive Ma-Mel-36_sens and IFNγ-resistant Ma-Mel-36_res cells sorted from IFNγ-treated (48 h) Ma-Mel-36_bulk cells based on their HLA-DR expression profile. Surface expression of indicated proteins measured by flow cytometry. Representative data from n=3 independent experiments. (c) JAK1 mutation defined by targeted sequencing on DNA from Ma-Mel-36_bulk and Ma-Mel-36_res cells. Plots of aligned sequencing reads in the location where the JAK1 c.843C>A, p.Y281* mutation was identified, arrows highlight mutation or corresponding wild-type (WT) site. Number of sequencing reads notated on the left; %, frequency of mutation in reads. (d) Melanoma cells analysed by western blot for expression of STAT1, pSTAT1, IRF1 and HLA class I heavy chains after IFNγ treatment (48 h); GAPDH, loading control. Representative data from n=2 independent experiments. (e) Ma-Mel-36_sens (S) and Ma-Mel-36_res (R) cells analysed for protein expression after IFNα and IFNγ treatment (48 h). Representative data from n=2 independent experiments. (f) SNP results given as allelic distribution of chromosome 1p shown for DNA obtained from Ma-Mel-36_bulk, Ma-Mel-36_sens, Ma-Mel-36_res and autologous Epstein-Barr virus-transformed B cells as a control. Loss of one chromosomal allele in the region 1p36.3-1p13.1 (Chr.1:854,277-116,804,754; hg19) in all Ma-Mel-36 cell populations. Dashed line indicates JAK1 location at Chr.1p31.3. (g) IFNγ release by autologous CD8⁺ T cells in the presence of IFNγ-treated (24 h) Ma-Mel-36 cell populations measured by ELISpot assay. Mean values (+s.e.m.) from n=2 measurements. (h) Real-time proliferation of Ma-Mel-36 cell populations in the presence/absence of IFNγ. Bold grey vertical lines indicate addition of IFNγ. Representative data from n=3 independent experiments. (i) IFNγ-induced (7 days) apoptosis in Ma-Mel-36 cell populations determined by AnnexinV/PI staining. Percentage of early (AnnexinV+/PI−) and late apoptotic (AnnexinV+/PI+) cells depicted. Mean values (+s.e.m.) from n=3 independent experiments. Only statistical significant differences defined by paired Student's t-test are indicated, *P<0.05.

Figure 2. JAK2 deficiency blocks HLA class I upregulation by IFNγ.

(a) Clinical history of patient Ma-Mel-54. Vertical line, time axis; left, therapeutic regimens; right, primary tumour (PT)/metastases development; arrows indicate cell lines established from metastases Ma-Mel-54a and Ma-Mel-54b; grey box, stage IV disease. (b) Mutations defined by targeted sequencing on DNA from melanoma cells and autologous blood cells as wild-type (WT) control (ctrl). Plots of aligned sequencing reads in the location where the JAK2 c.2876A>C, p.Q959P mutation was identified. WT sequences shown on the bottom, arrows highlight mutation sites. Number of sequencing reads notated on the left; %, frequency of mutations in reads. (c) Lysates from Ma-Mel-54a, Ma-Mel-54b and control cells (ctrl) analysed by western blot for JAK2 expression. (d) Lysates from IFNγ-treated (48 h) Ma-Mel-54a and Ma-Mel-54b cells analysed for expression of STAT1, pSTAT1, IRF1 and HLA class I heavy chains. (c,d) GAPDH, loading control. Representative data from n=2 independent experiments. (e) Expression of CD54, HLA class I and PD-L1 on IFNγ-treated (48 h) melanoma cells, measured by flow cytometry. Representative data from n=3 independent experiments. (f) SNP results given as allelic distribution of chromosome 9p shown for DNA obtained from Ma-Mel-54a, Ma-Mel-54b and autologous peripheral blood cells as normal control (germline). Loss of one chromosomal allele in region 9p24.3–p13.2 (Chr.9:203,861-37,578,327; hg19) present in both cell lines. Dashed line indicates JAK2 location at Chr.9p24.1. (g) Lysates from IFNγ-treated (48 h) Ma-Mel-54a-JAK2 transfectants analysed by western blot for expression of indicated proteins. (h) Real-time cell proliferation in the presence or absence of IFNγ. Bold grey vertical lines indicate addition of IFNγ. (g,h) Representative data from n=2 independent experiments. (i) Ma-Mel-54a and Ma-Mel-54b cells analysed for specific mRNA expression by quantitative reverse transcription–PCR. (j) Ma-Mel-54a-JAK2 transfectants analysed for specific mRNA expression by quantitative reverse transcription–PCR in the presence or absence of IFNγ (48 h). (i,j) Relative expression levels given as means (+s.e.m.) from n=2 independent experiments. (k) HLA class I expression on IFNγ-treated (48 h) Ma-Mel-54a cells and Ma-Mel-54a-JAK2 transfectants, measured by flow cytometry. Representative data from n=2 independent experiments.

Figure 3. Chromosome 1 alterations predispose to JAK1 deficiency.

(a) Clinical history of patient Ma-Mel-61. Vertical line, time axis; left, therapeutic regimens; right, metastases development; arrows indicate cell lines established from metastases Ma-Mel-61a, Ma-Mel-61b, Ma-Mel-61c, Ma-Mel-61e, Ma-Mel-61g and Ma-Mel-61h; grey box, stage IV disease. (b) Surface expression of CD54, HLA class I, and PD-L1 on IFNγ-treated (48 h) Ma-Mel-61b and Ma-Mel-61g cells, measured by flow cytometry. Representative data from n=3 independent experiments. (c) Mutation defined by targeted sequencing on DNA from Ma-Mel-61g cells. Plots of aligned sequencing reads in the location where the JAK1 c.1798G>T, p.G600W mutation was identified. Arrow highlights mutation site in Ma-Mel-61g or corresponding wild-type site (WT) in Ma-Mel-61b cells. Number of sequencing reads notated on the left; %, frequency of mutation in reads. (d) Lysates from Ma-Mel-61b and Ma-Mel-61g cells analysed by western blot for JAK1 expression; GAPDH, loading control. Representative data from n=3 independent experiments. (e) Lysates from IFNγ-treated (48 h) melanoma cells analysed by western blot for expression of STAT1, pSTAT1, IRF1 and HLA class I heavy chains; GAPDH, loading control. Representative data from n=3 independent experiments. (f) Lysates from IFNγ-treated (48 h) JAK1-transfected Ma-Mel-61g cells analysed for expression of the indicated proteins. Representative data from n=3 independent experiments. (g) IFNγ release by autologous CD8⁺ T cells in the presence of melanoma cells, measured by ELISpot assay. Means and s.e.m. (error bars) from n=4 independent measurement. Statistical significant differences defined by paired Student's t-test are indicated, *P<0.05. (h) Ma-Mel-61g and Ma-Mel-61g-JAK1 cells subjected to impedance-based real-time measurement of proliferation in the presence or absence of IFNγ. Addition of IFNγ indicated by bold grey vertical lines. Representative data from n=3 independent experiments. (i) SNP results given as allelic distribution of chromosome 1p shown for DNA obtained from the different Ma-Mel-61 cell lines and autologous blood cells as normal control (germline). Loss of one chromosomal allele in the region 1p34.3–1p12 (Chr.1:40,061,699-118,932,325; hg19) present in all cell lines. Dashed line indicates location of JAK1 at Chr.1p31.3.

Figure 4. IFNγ-resistant melanoma evolves into a T-cell-resistant lesion.

(a) Mutation defined by targeted sequencing on DNA from Ma-Mel-61h cells and autologous blood cells as wild-type (WT) control (ctrl). Plots of aligned sequencing reads in the location where the JAK1 c.1798G>T, p.G600W mutation was identified. WT sequence shown on the bottom, arrow highlights mutation or corresponding wild-type (WT) site. Number of sequencing reads notated on the left; %, frequency of mutations in reads. (b) Lysates from IFNγ-treated (48 h) Ma-Mel-61b, Ma-Mel-61g and Ma-Mel-61h cells analysed by western blot for expression of IRF1 and HLA class I heavy chains; GAPDH, loading control. Representative data from n=3 independent experiments. (c) HLA class I and CD54 surface expression on IFNγ-treated (48 h) Ma-Mel-61g and Ma-Mel-61h cells, measured by flow cytometry. Representative data from n=3 independent experiments. (d) Immunohistochemical staining of serial cryostat tissue sections from metastasis Ma-Mel-61g for melanoma marker HMB45 and HLA class I. (e) Ma-Mel-61h and Ma-Mel-61g cells, transfected with expression plasmids encoding wild-type JAK1 or mutant JAK1-G600W, analysed for specific mRNA expression by quantitative reverse transcription–PCR in the presence of absence or IFNγ (48 h). Ma-Mel-61b cells served as a control (ctrl). Relative expression levels given as means (+s.e.m.) from n=2 independent experiments. (f) HLA class I surface expression on IFNγ-treated (48 h) Ma-Mel-61h-JAK1 and Ma-Mel-61h-JAK1-G600W transfectants, measured by flow cytometry. Representative data from n=2 independent experiments. (g) Ma-Mel-61h and Ma-Mel-61g cells, transfected with expression plasmids encoding wild-type JAK1 (WT) or mutant JAK1-G600W (M), analysed for recognition by autologous CD8⁺ T cells in the presence or absence of IFNγ (48 h). T-cell activation measured as IFNγ release by ELISpot assay. Representative data from n=2 independent experiments.

Figure 5. Mutations in melanoma tissue samples and cell lines.

(a) Genetic alterations (homozygous deletions, single-nucleotide variants [SNV], small insertions and deletions [Indel]) in components of the type II IFN signalling pathway in TCGA melanoma samples (n=287, accessed 31/12/2016). (b) Mutations in JAK1, JAK2 and BRAF in the TCGA melanoma dataset stratified by tissue origin (primary tumours vs metastases). y axis, indicates expression of corresponding mRNA. Each circle represents a sample, samples harbouring SNV or Indels are shown in red (n=287, accessed 30/12/2016). (c–e) Recurrent mutations of type II IFN signalling pathway genes in published datasets. (f) SNP array data from 59 melanoma cell lines show LOH at the JAK1 (Chr. 1p31.3) and JAK2 (Chr. 9p24.1) locus.

Figure 6. Alterations in IFNγ pathway signalling impact on patient survival.

(a) Truncating mutations, homozygous deletions and low protein levels of IFNGR1, IFNGR2, JAK1, JAK2, STAT1 and IRF1 define a subset of melanoma patients in the TCGA data set with decreased survival. Of 479 patients with data on aberrations, 468 had available survival data. Log-rank P value shown. (b) Kaplan–Meier survival curves for IFNG, STAT1 and IRF1 expressing TCGA melanomas, P values shown from Log-rank tests. Multivariate Pvalues corrected for age: IFNG P=1.79e-05, n=90, 47 events; STAT1 P=3.82e-05, n=90, 33 events; IRF1 P=2.52e-05, n=90, 44 events.