NF2 and Canonical Hippo-YAP Pathway Define Distinct Tumor Subsets Characterized by Different Immune Deficiency and Treatment Implications in Human Pleural Mesothelioma

Abstract

(1) Inactivation of the tumor suppressor NF2 is believed to play a major role in the pathogenesis of malignant pleural mesothelioma (MPM) by deregulating the Hippo-YAP signaling pathway. However, NF2 has functions beyond regulation of the Hippo pathway, raising the possibility that NF2 contributes to MPM via Hippo-independent mechanisms. (2) We performed weighted gene co-expression analysis (WGCNA) in transcriptomic and proteomic datasets obtained from The Cancer Gene Atlas (TCGA) MPM cohort to identify clusters of co-expressed genes highly correlated with NF2 and phospho (p)-YAP protein, surrogate markers of active Hippo signaling and YAP inactivation. The potential targets are experimentally validated using a cell viability assay. (3) MPM tumors with NF2 loss-of-function are not associated with changes in p-YAP level nor YAP/TAZ activity score, but are characterized by a deficient B-cell receptor (BCR) signaling pathway. Conversely, MPM tumors with YAP activation display exhausted CD8 T-cell-mediated immunity together with significantly upregulated PD-L1, which is validated in an independent MPM cohort, suggesting a potential benefit of immune-checkpoint inhibitors (ICI) in this patient subset. In support of this, mutations in core Hippo signaling components including LATS2, but not NF2, are independently associated with better overall survival in response to ICI in patients. Additionally, based on cancer cell line models, we show that MPM cells with a high Hippo-YAP activity are particularly sensitive to inhibitors of BCR-ABL/SRC, stratifying a unique MPM patient subset that may benefit from BCR-ABL/SRC therapies. Furthermore, we observe that NF2 physically interacts with a considerable number of proteins that are not involved in the canonical Hippo-YAP pathway, providing a possible explanation for its Hippo-independent role in MPM. Finally, survival analyses show that YAP/TAZ scores together with p-YAP protein level, but not NF2, predict the prognosis of MPM patients. (4) NF2 loss-of-function and dysregulated Hippo-YAP pathway define distinct MPM subsets that differ in their molecular features and prognosis, which has important clinical implications for precision oncology in MPM patients.

Keywords: Hippo pathway; LATS2; NF2; YAP; immunotherapy; mesothelioma; targeted therapy.

Figure 1

Weighted gene correlation network analysis (WGCNA) reveals gene modules or clusters correlated with NF2 and p-YAP in MPM. (A) Schematic representation showing key signaling molecules regulating Hippo pathway and YAP activity. NF2 and other proteins function upstream and facilitate the activation of MST1/2, and then activate LATS1/2, which causes inhibitory phosphorylation of YAP. Hypo-phosphorylated YAP translocates to the nucleus and recruits TEAD transcription factors that regulate various biological processes. MST1/2-LATS1/2 complex acts as a core component of the Hippo signaling pathway. (B) Genetic alterations of Hippo pathway components (NF2, LATS1/2) in TCGA (The Cancer Genome Atlas) malignant pleural mesothelioma (MPM). (C,D) The protein level of NF2 and phospho (p)-YAP (S217) in MPM tumors with mutated (mut) and wild-type (WT) NF2. (E) The correlation between NF2 and p-YAP (S217) at the protein level in TCGA MPM. (F) The protein level of p-YAP (S217) in MPM tumors with mutated (mut) and wild-type (WT) LATS2. Note that the protein quantification of LATS2 is not provided in the Reverse Phase Protein Array (RPPA) dataset. (G) The correlation between YAP/TAZ score and p-YAP (S217) at the protein level in TCGA MPM. (H,I) WGCNA analysis. In (H) the upper panel shows the sample dendrogram and trait heatmap. The middle panel shows a histogram of network connectivity and the right is a log–log plot of the same histogram. The lower panel shows the gene dendrogram obtained by average linkage with hierarchical clustering. The color row underneath the dendrogram shows the module assignment determined by the Dynamic Tree Cut. Gray genes are unassigned to a module. Gene expression similarity is determined using a pair-wise weighted correlation metric, and clustered according to a topological overlap metric into modules. (I) Consensus network modules correlated with NF2 and p-YAP protein levels in MPM using the eigenmodule (the first principal component of the module). Pearson correlation coefficient along with p-value in parenthesis underneath; color-coded according to correlation coefficient (legend at right). The blue color indicates a negative correlation, while the red represents a positive correlation. WGCNA analysis of gene expression generated by unsupervised hierarchical clustering on the basis of topographical overlap followed by branch cutting reveals 13 network modules coded by different colors. Genes in the positively correlated modules (in red) indicate the abundance of these genes conferred by individual genetic events, while those in the negatively correlated ones (in blue) indicate the attenuation. Genes in the gray module are those that cannot be clustered.

Figure 2

NF2 loss-of-function is characterized by a deficient B-cell receptor (BCR) signaling pathway. (A–C) The top 5 significantly enriched Gene Ontology (GO; biological process (BP)), Kyoto Encyclopedia of Genes and Genomes (KEGG) (B), and Reactome (C) pathways based on genes in the MEpink module. Cnetplot in (C, lower panel) list genes in the enriched Reactome pathways. (D) STRING (https://string-db.org/ accessed on 21 December 2020) protein interaction map based on the top 30 hub genes in the MEpink module. (E) Correlation (Pearson r) between gene expression of CD20 (also known as MS4A1), a B-lymphocyte-specific membrane protein, and NF2 across TCGA MPM cohort. (F) Kaplan–Meier curves for MPM patients grouped by the plasma B-cell infiltrative signature and NF2 gene expression. The high vs. low groups are based on the median value of the indicated gene expression. Here, multiple covariates, e.g., tumor stage and purity, patients’ age and gender were included for adjustment. (G) Volcano plot showing the correlation between the AUC (Area the Under Curve) value of drugs and the NF2 protein level. Blue dots indicate the significantly (p-value < 0.05) negatively correlated drugs while the red indicates the positively correlated ones. Here, a positive correlation indicates the association of a larger (more resistant) AUC value with higher gene expression and vice versa.

Figure 3

MPM tumors with YAP activation display exhaustion of T-cell-mediated immune response. (A,B) The top 10 significantly enriched Gene Ontology (GO; biological process [BP]) (A) and Reactome (B) pathways based on genes in the MEred module. (C,D) Correlation (Pearson r) between PD-L1 protein and p-YAP (C, left) or between PD-L1 protein and NF2 protein (D, left). Violin plots show the association between PD-L1 and the genetic alterations of LATS2 (C, right) or NF2 (D, right). (E) Association between the mutations in LATS1/2 (n = 92) or other key regulatory components (LATS1/2, MST1, YAP1; n = 111) of the Hippo pathway with overall survival in cancer patients after immune checkpoint blockade (anti-PD-1/PD-L1, or anti-CTLA4, or combination treatment). Of note, there is no significant difference in the treatment regimens between the wildtype (WT) and mutated (Mut) subgroups. Data were downloaded from cBioPortal https://www.cbioportal.org/ accessed on 21 December 2020. (F,G) Volcano plots showing the proteins (F) and drugs (G) that are correlated with p-YAP across TCGA MPM samples and solid cancer cell lines, respectively. Blue dots indicate the significantly (p-value < 0.05) negatively correlated proteins/drugs while the red the positively correlated ones. In (G) the AUC (Area Under Curve), value of drugs was used to indicate the drug effect, with a positive correlation representing the association of a larger AUC value (more resistance) with a higher p-YAP level and vice versa. (H) The median inhibitory concentration (IC50) values of a panel of MPM cell lines were treated with Dasatinib (96 h). MPM cells seeded in triplicate at 96-well plates were drugged 24 h later, over a 12-point concentration range (two-fold dilution). DMSO-treated cells were used as control. IC50 was determined using GraphPad Prism 7. N = 3 biological replicates. The lower panel shows the gene annotations of the indicated MPM cell lines.

Figure 4

Dysregulation of NF2 and Hippo-YAP exhibits different infiltrative immune signatures in MPM. (A) Tumor-infiltrating immune cell profiles across the TCGA pan-cancer cohort were shown. The number of patients shown in parentheses. Data were downloaded from TIMER (version 2.0), a comprehensive resource for systematic analysis of immune infiltrates across diverse cancer types (http://timer.comp-genomics.org/) (See the methods). (B) Percentage of immune subtype models (C1–C6) across the TCGA MPM cohort, in which the reverse-phase protein array (RPPA) data were used. The genes contained in each signature were evaluated using model-based clustering by p the “mclust” R package. Each sample was finally to be grouped based on its predominance with the C1–C6 signature. The immune subtype models were based on Thorsson V et al. Immunity. 2018 (See the methods). (C) Correlative analysis of the hub genes-based signature score in the red module (Figure 1I) with YAP/TAZ target score and NF2 gene expression based on an independent MPM dataset (GSE29354).

Figure 5

Association of LATS1/2 mutations with TMB and MSI in MPM. (A,B) Tumor mutational burden (TMB; A) and microsatellite instability (MSI; B) across TCGA pan-cancer cohort. (C) Association of LATS1/2 mutations with TMB and MSI in MPM.

Figure 6

A pleiotropic role of NF2. (A) The common (N = 93) physical interactors of NF2 based on 3 curated public databases (BioGRID, HitPredict, and APID). (B–D) Top 10 significantly enriched Reactome (B), Kyoto Encyclopedia of Genes and Genomes (KEGG) (C), and Gene Ontology (GO; biological process [BP]) (D) pathways based on the common interactors (N = 93) of NF2 in (A). (E) Venn plot showing the gene changes in mouse brain cells with Yap1 overexpression and Nf2 knockout. Data were downloaded from GSE48078. (F) Association of NF2, phospho-YAP (S127), YAP/TAZ score with overall survival in MPM patients. (G) Univariate (A) and multivariate (B) survival analysis showing that YAP/TAZ score and the protein level of p-YAP (S127) but not NF2 predict prognosis of MPM patients.