Advertisement
Research Article
Source Data
Transparent Process
Open Access

MAPK pathway mutations in head and neck cancer affect immune microenvironments and ErbB3 signaling

Hoi-Lam Ngan, Yuchen Liu, Andrew Yuon Fong, Peony Hiu Yan Poon, Chun Kit Yeung, Sharon Suet Man Chan, Alexandria Lau, Wenying Piao, Hui Li, Jessie Sze Wing Tse, Kwok-Wai Lo, Sze Man Chan, Yu-Xiong Su, Jason Ying Kuen Chan, Chin Wang Lau, Gordon B Mills, Jennifer Rubin Grandis, Vivian Wai Yan Lui  Correspondence email
Published 7 May 2020. DOI: 10.26508/lsa.201900545

Article Figures & Data

Figures

  • Figure 1.
    Figure 1. MAPK pathway mutations in head and neck squamous cell carcinoma (HNSCC) are associated with remarkable patient survival.

    (A) Percentage of patients affected by MAPK pathway mutations and mutations of six other cancer pathways (PI3K, JAK/STAT, Notch, WNT, NF-κB, and TGFβ/Smad) in TCGA HNSCC provisional cohort (N = 510). (B) HNSCC patient outcome associations for all seven pathway mutations (i.e., pathway-mutated versus pathway WT) and HPV status. Red bars indicate favorable HNSCC overall survival (OS), whereas the blue bar indicates unfavorable OS when the pathway components are mutated (log-rank test P-values are shown). (C) Kaplan–Meier OS curves for MAPK pathway-mutated HNSCC patients versus MAPK pathway WT patients (TCGA HNSCC provisional cohort). (D) Kaplan–Meier OS curves for TP53-mutated patients with MAPK pathway-mutated versus WT HNSCC (TCGA HNSCC provisional cohort). (E) Bar graph showing the number of MAPK pathway protein components for each cancer type (total 33 cancer types) that were significantly correlated with OS. Red bars indicate associations with favorable outcomes, whereas grey bars indicate associations with unfavorable outcomes in each cancer type, when the MAPK protein component(s) is/are overexpressed (median cutoff; The Cancer Protein Atlas database). (F) Kaplan–Meier OS curves for MAPK pathway-mutated HNSCC patients versus MAPK pathway WT patients (MSK-IMPACT HNSCC cohort). (G) Kaplan–Meier survival curves showing increased OS for uterine corpus endometrial carcinoma patients with MAPK pathway mutation versus WT (TCGA uterine corpus endometrial carcinoma cohort).

  • Figure S1.
    Figure S1.

    Overall survival and the mutational burden of TCGA HNSCC tumors with respective pathway mutations and HPV status. (A, B, C, D, E, F) Kaplan–Meier curves showing overall survival of head and neck squamous cell carcinoma (HNSCC) patients with or without mutations of the remaining six key cancer pathways (PI3K, NOTCH, JAK/STAT, NF-κB, WNT, and TGF-β/Smad pathways) in the HNSCC provisional cohort (N = 508; TCGA). (G) Comparison of mutational loads of HNSCC tumors with all seven HNSCC-relevant cancer signaling pathways mutated, as well as HPV-positive tumors. Unpaired t test P-values are shown.

  • Figure S2.
    Figure S2. Mapping of mutation sites of the MAPK pathway genes based on whole-exome sequencing data of TCGA HNSCC provisional cohort.

    The frequencies of mutational events are proportional to the pin-height displayed.

  • Figure S3.
    Figure S3. Kaplan–Meier overall survival curves for HPV-negative patients with MAPK pathway mutations versus WT (TCGA HNSCC provisional cohort).
  • Figure 2.
    Figure 2. Multiple MAPK pathway mutations inhibited ErbB3 activation.

    (A) Kaplan–Meier overall survival curves for head and neck squamous cell carcinoma (HNSCC) patients, whose tumors overexpressed versus underexpressed phospho-ErbB3(Y1289) (median cutoff) in TCPA HNSCC cohort (N = 344). (B) Western blot results of phospho-ErbB3(Y1289) protein levels upon ectopic expression of the MAPK1, ARAF, BRAF, HRAS, and MAP2K1 as well as MAP2K2 wild-type and mutant constructs in FaDu cells by retroviral infection (pool of at least four independent repeats). (C) Western blot results of phospho-ErbB3(Y1289) levels of HNSCC Pt.25 primary tumor cultures (carrying both HRAS p.G12S and MAPK1 p.R135K mutations), HSC-6 cell line (carrying MAPK1 p.E322K mutation), and HSC-4 (MAPK pathway WT, Cancer Cell Line Encyclopedia [CCLE]) upon MAPK inhibitor GDC-0994 treatment for 30 min. 50 µg of protein lysate was used for Pt.25 and HSC-4 samples, 50 μg of protein lysate was used, whereas for HSC-6 (because of the relatively low endogenous p-RSK levels intrinsic to this cell line), 100 μg of protein lysate was loaded for presentation of signal clarity. Bar graphs showing the quantified changes of p-ErbB3(Y1289) levels upon GDC-0994 treatment (N ≥ 4 independent experiments). (D) Negative correlation between p-ErbB3(Y1289) and p-MAPK(T202/Y204) levels in MAPK-mutant HNSCC cell lines based on the published CCLE-proteomic data (Ghandi et al, 2019) and in MAPK-mutated HNSCC patient tumors (allele frequencies [AFs] >40%) based on TCPA HNSCC RPPA cohort (Li et al, 2013, 2017a). (E) Immunohistochemical staining for p-ErbB3(Y1289) in MAPK-mutated HNSCC patient tumors with high AFs close to 40%: T40 (MAPK1 p.D321N with AF = 39.1%) and T43 (HRAS p.G12S with AF = 37.3%) versus T47 and T82 (both are MAPKWT).

    Source data are available for this figure.

  • Figure S4.
    Figure S4.

    Cumulative phospho-ErbB3(Y1289)/total-ErbB3 ratio of MAPK-mutants expressing as well as GDC-0994 inhibited FaDu cells. (A) Cumulative quantification plots of phospho-ErbB3(Y1289)/total-ErbB3 ratio upon ectopic expression of mutants and WTs of MAPK1, ARAF, BRAF, HRAS, MAP2K1, and MAP2K2 in FaDu cells (N ≥ 4 independent experiments). (B) Cumulative quantification plots of phospho-ErbB3(Y1289) and total-ErbB3 levels in HNSCC Pt.25 primary tumor cultures (carrying both HRAS p.G12S and MAPK1 p.R135K mutations), HSC-6 cell line (carrying MAPK1 p.E322K mutation) and HSC4 (MAPK pathway WT, CCLE) upon MAPK inhibitor GDC-0994 treatment for 30 min. The quantified changes of p-ErbB3(Y1289)/total-ErbB3 ratio upon GDC-0994 treatment were showed (N ≥ 4 independent experiments).

  • Figure S5.
    Figure S5. Protein–protein expression correlation graphs showing no significant correlations between expression levels of p-ErbB3(Y1289) and p-MAPK(T202/Y204) in MAPK-WT CCLE HNSCC cell lines and in MAPK-WT HNSCC patient tumors from TCPA proteome dataset.

    Spearman rank correlation coefficients, and the respective P-values are shown.

  • Figure S6.
    Figure S6.

    In vitro drug sensitivity of endogenously and ectopically MAPK mutated cell lines. (A) Head and neck squamous cell carcinoma (HNSCC) drug sensitivity comparisons for MAPK pathway-mutated versus WT HNSCC cell lines from the CCLE database (COSMIC database; Wellcome-Sanger Institute, UK). IC50 values are plotted for cisplatin, docetaxel, 5-FU, methotrexate, and cetuximab. (B) In-house carboplatin and cisplatin sensitivity dose–response curves for HNSCC cells (FaDu) expressing MAPK1-WT, MAPK1 p.D321N, and MAPK1 p.E322K mutants, as well as HRAS-WT, HRAS p.G12V, and HRAS p.C118Y mutants. Corresponding IC50 values are shown.

  • Figure 3.
    Figure 3. Transcriptomic analyses reveal prominent immune signatures of MAPK-mutant HNSCC tumors.

    (A) A volcano plot showing differential RNA expressions between MAPK pathway-mutated versus WT HNSCC tumors. (B) Distinct gene expression patterns of MAPK pathway-mutated tumors versus MAPK-WT tumors (based on RNA-seq dataset of TCGA HNSCC cohort), with 130 protein-coding differentially expressed gene with log2 fold-change > 0.5 and false discovery rate (FDR) < 0.05 shown in the respective heat maps. (C) Gene Set Enrichment Analysis for MAPK pathway-mutated (versus MAPK-WT) HNSCC tumors demonstrating enrichment of immune-related gene sets in four of eight enrichment functional gene sets. (D) Comparison of expression levels of PRF1, GZMA, GZMB, GZMH, and GZMK mRNA in MAPK-mutated versus WT HNSCC tumors (based on RNA-seq dataset of TCGA HNSCC provisional cohort).

  • Figure S7.
    Figure S7. The Gene-Concept network plot shows the linkages between differentially expressed genes for MAPK pathway-mutated (versus MAPK-WT) HNSCC tumors and GSEA enriched biological concepts.
  • Figure S8.
    Figure S8.

    Immune-related transcriptome signatures of HPV negative HNSCCS. (A) Bar graphs showing significant increase in expression levels of PRF1, GZMA, GZMB, GZMK, GZMH, and GZMM in HPV-negative MAPK-mutated versus MAPK-WT HNSCC tumors (TCGA Provisional cohort). (B) Clustering of tumor infiltrating immune cell types and immune signature scores (CYT, T-effector signature and IFN-γ functionality score) in TCGA HPV-negative HNSCC tumors with MAPK pathway mutational status.

  • Figure 4.
    Figure 4. MAPK-mutant HNSCC patient tumors were CD8+ T-cell inflamed with immunoreactive cytolytic signatures.

    (A) A bubble plot showing the degree of statistical significance for HNSCC tumor infiltration levels of B cells, CD4+ T-cells, CD8+ T-cells, dendritic cells, macrophages, and neutrophils (by TIMER analyses (Li et al, 2016, 2017b), and the CYT score, T-effector score and IFN-γ score for HNSCC tumors bearing respective pathway mutations (versus respective WT tumors). Bubbles are highlighted in red outline when P < 0.05 with calculated positive correlations for increase in the respective TIL or immune score, and bubbles are highlighted in orange outline when P < 0.05 with calculated negative correlations indicating decrease in the respective TIL or immune score when a pathway is mutated. (B) Results for TIMER analysis for MAPK pathway-mutated versus WT HNSCC tumors for all HNSCC tumors (upper panel), or for human papillomavirus (HPV)-negative HNSCC only (lower panel). MAPK pathway-mutated tumor showed most significant increases in CD8+ T-cell infiltrations in all HNSCC, as well as in HPV-negative HNSCC (lower panel). P-values were shown for each immune cell type (unpaired t test). (C) Comparisons of CYT score, T-effector signature score, and IFN-γ functionality score between MAPK-mutated versus WT HNSCC tumors (for all HNSCC in upper panel; and for HPV-negative HNSCC only in the lower panel).

  • Figure S9.
    Figure S9.

    TIMER analyses of six oncogenic pathways. (A, B, C, D, E, F) Results of TIMER analyses for HNSCC tumors with the remaining six cancer pathway mutations (PI3K, JAK/STAT, NF-κB, NOTCH, WNT, and TGF-β/Smad in [A, B, C, D, E, F] versus respective WT tumors).

  • Figure 5.
    Figure 5. CD8+ T-cell–inflamed and cytolytic features of MAPK-mutated head and neck squamous cell carcinoma (HNSCC) were recapitulated in immunocompetent models with massive apoptosis in situ.

    (A) Nonhierarchical clustering of tumor infiltrating immune cell types and immune signature scores (CYT, T-effector signature and IFN-γ functionality score) in TCGA HNSCC tumors with MAPK pathway mutational status and HPV status. (B) High infiltrating levels of CD8+ T-cells are associated with markedly improved HNSCC patient survival compared with patients with low CD8+ T-cell infiltration levels (top and bottom 20% cutoffs). (C) Immunohistochemical staining of MAPK pathway-mutated HNSCC tumors (T40 with MAPK1 p.D321N hotspot mutation) showing increased expressions of the CD8 marker (for CD8+ T infiltration). These tumors also expressed higher levels of CD11c marker (for dendritic cell infiltration) and neutrophil elastase marker (for neutrophil infiltration) when compared with MAPK-WT tumors (T69). Scale bars are shown. (D) Increased expressions of CD8 marker (for CD8+ T infiltration) were detected in mHRAS p.G12V mutant, mMAPK1 p.D319N mutant (corresponding to MAPK1 p.D321N mutation in human MAPK1), and mMAPK1 p.E320K mutant (corresponding to MAPK1 p.E322K mutation in human MAPK1), and in their respective mouse WT xenografts by immunohistochemistry on day 6 (for mMAPK1 WT versus mMAPK1 p.E320K pair) and on day 11 (for mHRAS WT versus mHRAS p.G12V & mMAPK1 WT versus mMAPK1 p.D319N pairs) after tumor cell inoculation. (E) Dramatic increases of apoptotic cells were also observed in MAPK pathway-mutated tumors labeled with TUNEL and corresponding DAPI staining on day 6 after tumor cell inoculation. (F, G, H) Kaplan–Meier overall survival (OS) curves for HNSCC patients with (F) higher IFN-γ functionality score, (G) higher T effector signature score, and (H) higher CYT score and versus patients with respective lower IFN-γ functionality, T-effector signature and CYT score in TCGA HNSCC (N = 522) RNA-seq cohort (top 20% and bottom 20% cutoffs). (I) Kaplan–Meier OS curves for MAPK pathway mutated patients with high IFN-γ functionality score and high T-effector signature score and high CYT score (all top 20% cutoffs), demonstrating long OS (median OS not reached).

  • Figure S10.
    Figure S10.

    Immunohistochemistry staining of tumor infiltrating lymphocytes in in-house HNSCC tumors. (A) Immunohistochemical staining of two additional pairs of head and neck squamous cell carcinoma (HNSCC) tumors (MAPK pathway-mutated versus WT), showing increases in the levels of intratumoral infiltrations of CD8+ T-cells, dendritic cells, and neutrophils. T25 (MAPK-mutated HNSCC tumor with bearing HRAS p.G12S and MAPK1 p.R135K) versus T75 (MAPK-WT) and T06 (MAPK-mutated HNSCC tumors with HRAS p.G13D mutation) versus T39 (MAPK-WT). Scale bars of 100 μm are shown. (B) The corresponding isotype control stainings of MAPK-mutated T40, T25, and T06 as well as MAPK-WT T69, T75, and T39 are shown. Scale bars of 100 μm are shown.

  • Figure S11.
    Figure S11. Kaplan–Meier overall survival curves for HNSCC patients with higher CYT score versus patients lower CYT score in TCGA HNSCC RNA-seq cohort (N = 522) with risk tables. Sensitivity analysis of significant level across different cutoffs (from quintile [top 20% versus bottom 20%], quartile [top 25% versus bottom 25%], and tertile [top 33% versus bottom 33%] to median [50%]) are shown.
  • Figure S12.
    Figure S12. Kaplan–Meier overall survival curves for HNSCC patients with higher IFN-γ score versus patients lower IFN-γ score in TCGA HNSCC RNA-seq cohort (N = 522) with risk tables. Sensitivity analysis of significant level across different cutoffs [from quintile (top 20% versus bottom 20%), quartile (top 25% versus bottom 25%), and tertile (top 33% versus bottom 33%) to median (50%)] are shown.
  • Figure S13.
    Figure S13. Kaplan–Meier overall survival curves for HNSCC patients with higher T-effector signature score versus patients lower T-effector signature in TCGA HNSCC RNA-seq cohort (N = 522) with risk tables. Sensitivity analysis of significant level across different cutoffs (from quintile [top 20% versus bottom 20%], quartile [top 25% versus bottom 25%], and tertile [top 33% versus bottom 33%] to median [50%]) are shown.
  • Figure 6.
    Figure 6. ErbB3 inhibition and CD8+ T-cell immunoactivation as plausible mechanisms contributive to improved survival of MAPK pathway–mutated HNSCC patients.

    (A) Oncoprints (20% and 50% cutoffs respectively) showing that MAPK pathway-mutated patients with low p-ErbB3 protein expression, and high IFN-γ functionality, T-effector signature and CYT score were not significantly overlapping (P = n.s.). The upper panel shows the oncoprint with deep blue color (Embedded Image) denoting those with bottom 20% of pErbB3 level (i.e., p-ErbB3 down-regulation), and levels above that as noncolored white bars (Embedded Image), whereas individuals without available RPPA data on p-ErbB3 are denoted by grey bars (Embedded Image). Similarly, those with top 20% immune scores, IFN-γ score, CYT score and T-effector are indicated by deep pink (Embedded Image), light pink (Embedded Image) and orange (Embedded Image), respectively, while non-colored white bars (Embedded Image) denote patients with immune scores lower than the top 20%. In the lower panel, the same color coding is adopted, but the colored bars refer to patients with a median (i.e., 50% cutoffs), that is, lower 50% for p-ErbB3 level and top 50% for IFN-γ, CYT, and T-effector scores. (B) A schematic summarizing two plausible mechanisms for markedly improved clinical outcomes in HNSCC tumors with MAPK aberrations.

  • Figure 7.
    Figure 7. MAPK pathway mutations may predict patient outcomes with immune checkpoint inhibitors independent of tumor mutational burden (TMB).

    (A) MAPK pathway mutations (10 genes) are associated with good clinical outcome from an independent cohort of Samstein et al (2019) in advanced or metastatic pan-cancer patients (N = 1,662) treated with PD1/PD-L1 or CTLA4 inhibitors. (B) A pie chart showing 34% patients with somatic MAPK pathway mutations in both TMB-high and TMB-low groups of patients in the pan-cancer dataset. High-TMB was previously defined as top 20% cutoff (i.e., TMB value ≥10.3 for head and neck squamous cell carcinoma [HNSCC]), whereas low-TMB represented the remaining 80% of patients (i.e., TMB value < 10.3 for HNSCC) per original publication by Samstein et al (2019). (C) Overall survival (OS) curves of four subgroups of patients: TMB-high with MAPK mutations, TMB-high with MAPK-WT, TMB-low with MAPK mutations, and TMB-low with MAPK-WT in the Samstein study. (D) Kaplan–Meier OS curves for MAPK pathway-mutated HNSCC patients versus MAPK pathway WT patients (MSS ICI pan-cancer cohort; N = 249). (E) Kaplan–Meier OS curves for MAPK pathway-mutated HNSCC patients versus MAPK pathway WT patients (study by Samstein et al (2019); HNSCC Ooal subsite cohort; N = 47). (F) Table of Fisher’s exact test showing association of HNSCC subtypes with the immune class as defined by Chen et al (2019). (G) In HNSCC patients with distant metastases (lung, liver, heart, brain, and bone), MAPK pathway mutations are associated with better OS upon PD1/PD-L1 inhibitor treatment (P = 0.0489; based on databases from Samstein et al (2019)). (H, I) The corresponding oncoprints showing no significant overlap between (H) patients with MAPK pathway mutations and high tumor mutational burden in this HNSCC-oral cancer cohort (TMB score 20% cutoff within HNSCC histology [N = 139] in the study by Samstein et al (2019)) and (I) no significant overlap for patients with MAPK pathway mutations and high tumor mutational burden in this HNSCC distant metastasis cohort (TMB score 20% cutoff within HNSCC histology).

  • Figure S14.
    Figure S14.

    OS prediction by TMB in HNSCC immunotherapy cohorts. (A) Kaplan–Meier overall survival curves for tumor mutational burden (TMB)-high versus TMB-low patients (study by Samstein et al (2019); HNSCC Oral subsite cohort; N = 47). (B) In HNSCC patients with distant metastases (lung, liver, heart, brain, and bone), TMB-high (defined by TMB value >10.3 for head and neck cancer from Samstein et al (2019)) is also associated with improved overall survival (P = 0.1969, a trend; right panel).

  • Figure S15.
    Figure S15. Kaplan–Meier overall survival curves for MAPK pathway-mutated HNSCC patients versus MAPK pathway wild-type (WT) patients based on the Samstein et al (2019) cohort (HNSCC oropharynx cohort; N = 32).

Tables

  • Table showing the sequenced locations and the aminoacids convered of the partially sequenced genes in our custom-designed MAPK pathway gene panel.

    GeneChromosomeLocus positionAmino acid position
    ARAFChrX47426003-47426252186-233
    BRAFChr7140434532-140434640, 140439596-140439815, 140449044-140449156, 140449186-140449273, 140453021-140453215, 140453948-140454071, 140476652-140476772, 140476828-140476946, 140477788-140477937, 140481362-140481448454-526, 545-631, 642-667, 679-722
    HRASChr11533782-533882, 534220-5343066-34, 58-91
    KRASChr1225368427-25368555, 25378527-25378605, 25378635-25378758, 25380260-25380368, 25398189-253983104-66, 98-121, 132-173
    MAPK1Chr2222123469-22123664, 22127153-22127280, 22142514-22142695, 22142879-22143107, 22153230-22153505, 22160105-22160333, 22161922-22162142, 22221511-2222164429-362
    MAP2K1Chr156667947-66679845, 66727308-66727523, 66727541-66727651, 66728983-66729286, 66735605-66735722, 66736897-66737080, 66774071-66774222, 66777239-66777548, 66779529-66779647, 66781535-66781633, 66782015-66782132, 66782829-6678296121-356
    MAP2K2Chr194090254-4090351, 4090679-4090789, 4094494-4094602, 4095288-4095521, 4097192-4097295, 4097316-4097449, 4099378-4099495, 4101127-4101237, 4101257-4101387, 4102338-4102451, 4110387-4110512, 4110586-4110696, 4117321-4117447, 4117464-411756652-85, 92-124, 149-183, 191-198, 236-246, 308-315, 323-350, 365-373, 483-493
Back to top
Download PDF
Email Article
Citation Tools
Survival-favoring biology of MAPK-mutant HNSCC
Hoi-Lam Ngan, Yuchen Liu, Andrew Yuon Fong, Peony Hiu Yan Poon, Chun Kit Yeung, Sharon Suet Man Chan, Alexandria Lau, Wenying Piao, Hui Li, Jessie Sze Wing Tse, Kwok-Wai Lo, Sze Man Chan, Yu-Xiong Su, Jason Ying Kuen Chan, Chin Wang Lau, Gordon B Mills, Jennifer Rubin Grandis, Vivian Wai Yan Lui
Life Science Alliance May 2020, 3 (6) e201900545; DOI: 10.26508/lsa.201900545
Twitter logo Facebook logo Mendeley logo
Issue Cover

In this Issue

Volume 3, No. 6
June 2020
Advertisement

Jump to section

Subjects