YAP Activity is Not Associated with Survival of Uveal Melanoma Patients and Cell Lines

Recent experimental studies have demonstrated an essential role for the Hippo-Yes-associated protein (YAP) pathway in GNAQ/GNA11-induced tumorigenesis in uveal melanoma (UM). However, the association between YAP activity and clinical outcomes remains elusive. We investigated possible associations between YAP activity and clinicopathological features including survival outcomes in patients with UM using The Cancer Genome Atlas (TCGA) cohort and our local cohort. We estimated YAP activity by mRNA expression levels, Gene Set Variation Analysis (GSVA) for the TCGA cohort, and immunohistochemical YAP staining for the local cohort. In the TCGA cohort, most clinicopathological features including tumor stage, mitotic counts, mutation of genes, and tumor sizes did not significantly differ between low and high YAP activity groups. In the local cohort, YAP nuclear-positive staining was observed in 30 (42%) of 72 patients with primary UM. UM-specific survival was not significantly different between tumors with low and high YAP activities. Unlike mesothelioma cells harboring a mutation of negative regulators of YAP, the survival of multiple UM cell lines was not significantly reduced by YAP/TAZ depletion. Our results suggest that the effect of YAP on development, growth, and invasion of UM in actual patients is less than previously demonstrated in experimental studies.


Results
Study population. For the TCGA cohort, all patients had UM with choroid or ciliary body involvement. Among them, two patients had UM involving the choroid, ciliary body, and iris. The mean RNA-seq by Expectation Maximization (RSEM)-normalized YAP mRNA levels were 1430.5 ± 362.4 and 2719.4 ± 700.8 for the low expression and high expression groups, respectively (P < 0.001). The mean Gene Set Variant Analysis (GSVA) scores for the YAP signature were 0.162 ± 0.115 and 0.168 ± 0.124 in the low YAP signature and high YAP signature groups, respectively (P < 0.001; Fig. 1a). When compared with the YAP-downregulated group, the YAP signature gene sets were significantly upregulated in the high YAP signature group (normalized enrichment score: 2.30; P < 0.001; Fig. 1b and Supplementary Fig. 1). The clinicopathological characteristics of the TCGA cohort are listed in Table 1.
Regarding the local cohort, from January 2001 to December 2012, surgical resections of primary UM were performed in 74 patients; the resulting formaldehyde-fixed and paraffin-embedded tissues were available from 72 patients. Among these 72 patients, 61 and 11 were treated with enucleation and sclerouvectomy with brachytherapy, respectively. When immunohistochemical (IHC) staining for YAP was performed on these tissues, YAP nuclear-negative staining was observed in 42 patients, whereas YAP nuclear-positive staining was observed in 30 patients (Fig. 1c,d). The mean ages were 51.3 ± 14.6 and 53.6 ± 16.1 years in the YAP nuclear-negative and YAP nuclear-positive groups, respectively (P = 0.53). Twenty-four and 14 patients in the YAP nuclear-negative and YAP nuclear-positive groups, respectively, were male (P = 0.47).
Largest basal diameter and thickness of tumors. In the TCGA cohort, based on the YAP mRNA expression levels, the largest basal diameters and thicknesses were not significantly different between the YAP  Table 1). When comparing the lower 25% and upper 25%, the largest tumor basal diameters and thicknesses also did not significantly differ (P = 0.36 and P = 0.74, respectively). When comparing the low YAP signature and high YAP signature groups, there were no significant differences in tumor basal diameter and thickness (P = 0.13 and P = 0.79, respectively; Fig. 2c,d and Table 1). No significant differences were observed in tumor diameter and thickness between the lower 25% and upper 25% of the GSVA scores for the YAP signature (P = 0.16 and P = 0.49, respectively).
In the local cohort, when the YAP nuclear-negative and YAP nuclear-positive groups were compared, the largest basal diameter was significantly smaller in the YAP nuclear-positive group (12.5 ± 2.9 mm vs. 10.9 ± 3.2 mm; P = 0.038; Fig. 2e), whereas the tumor thicknesses were not significantly different between groups (9.5 ± 0.6 mm vs. 9.7 ± 0.5 mm; P = 0.81; Fig. 2f).

Metastasis.
In the local cohort, 19/42 (45%) and 19/30 (63%) patients developed metastases in the YAP nuclear-negative and YAP nuclear-positive groups, respectively. Liver metastasis was observed in 17 and 18 patients in the YAP nuclear-negative and -positive groups, respectively. Extrahepatic metastasis occurred in two patients in the YAP nuclear-negative group and one patient in the YAP nuclear-positive group. Details of the metastatic sites are listed in Supplementary Table 1. www.nature.com/scientificreports www.nature.com/scientificreports/ comparison of iHc staining patterns between primary and metastatic UM tumors. There were 16 metastatic UM tumors (liver metastasis); nine (56%) showed a YAP nuclear-positive pattern using IHC staining. The YAP IHC staining patterns did not significantly differ between primary (n = 72) and metastatic (n = 16) UM tumors (P = 0.41).
YAP/TAZ knockdown exhibited smaller effects on survival in multiple UM cell lines (92.1, MP41, MP46, and MP65). Compared with RPE1, MSTO-211H, and H2373 cell lines, 92.1, MP41, MP46, and MP65 cell lines did not exhibit significant reduction in cell survival at 72 hours after siRNA transfection (Supplementary Fig. 4). Similar results were observed at 144 hours after transfection (Fig. 4). These results suggested that the effect of YAP on cell survival was much smaller in UM cell lines than in RPE1 cell line and mesothelioma cell lines.

Discussion
Cancer genome sequencing efforts have identified frequently mutated genes in UM, including GNAQ, GNA11, BAP1, SF3B1, and EIF1AX 7,25 . Mutations of these genes are reportedly associated with the development or prognosis of UM 7,26,27 . Remarkably, GNAQ and GNA11 are mutated in most UM tumors (~93%) with hotspot mutations (Q209P/L) 7 . Recent studies have highlighted the roles of GNAQ and GNA11 mutations in UM development. GNAQ and GNA11 mutations result in constitutive activation of oncogenic Gq/G11 subunits, leading to UM tumorigenesis by sequential activation of YAP 10,11,23 .
The Hippo-YAP signaling pathway is a key determinant of organ size, stem cell homeostasis, and cellular differentiation 14,16 . Two Hippo pathway transducers, YAP and its paralog TAZ, are transcriptional coactivators with nuclear-cytoplasmic distributions that are mainly controlled by Hippo signaling 28 . LATS1/2, MST1/2, and NF2 are the major upstream kinases of YAP, which cause YAP cytoplasmic retention and degradation by phosphorylating YAP 14 . Nuclear YAP (the active form of YAP) induces the expression of cell proliferative and anti-apoptotic genes, mainly by interacting with TEAD family transcription factors 14 . In experimental models using the 92.1 UM cell line, inhibition of YAP by short hairpin RNA or verteporfin suppressed the growth of UM 10,11 . Previous studies have reported that high YAP activity is associated with poor prognoses in various cancers 14,22 . In this study, we investigated the association between YAP activity and clinicopathological characteristics using the TCGA cohort and a local cohort. Nucleocytoplasmic shuttling of YAP is rapidly changed by various cellular cues; YAP phosphorylated by Hippo kinases shows cytoplasmic retention, followed by rapid degradation by the proteasome 15 . Therefore, YAP mRNA levels may not reflect YAP activity. To overcome this limitation, we also estimated YAP activity by calculating the enrichment score for the YAP conserved signature genes by GSVA for each tumor sample. In the TCGA cohort, epithelioid cell type and marked pigmentation were associated with high YAP activity. However, the cancer stages, mitotic counts, and gene mutation profiles did not differ between groups. Consistently, the YAP mRNA expression levels and GSVA scores for YAP signatures were not significantly associated with tumor size and prognosis.
We further validated the clinical outcomes in the local cohort. With respect to YAP, because subcellular localization reflects activity, it is possible to estimate YAP activity by using IHC staining 28 . YAP nuclear staining, which indicated active YAP, was observed in only 30 (42%) of patients with UM; moreover, YAP IHC staining patterns were not significantly different between primary and metastatic tumors. Tumor size and prognosis were also not significantly different between the YAP nuclear-negative and YAP nuclear-positive groups.
Although YAP activities measured by several methods were not associated with the prognoses of patients with UM, YAP may be a therapeutic target for UM if suppression of YAP activity significantly affects the survival of UM cell lines. To confirm this possibility, we investigated whether siRNA-mediated YAP knockdown affected the survival of UM cell lines. YAP knockdown slightly reduced the survival of 92.1, MP41, MP46, and MP65 UM cell lines. However, these effects were much smaller than the effects observed for the two mesothelioma cell types, MSTO-211H (LATS2 mutation) and H2373 (NF2 mutation), which harbor mutations in negative regulators of YAP 29 . In the TCGA mesothelioma cohort, 35 (42%) of 83 patients had mutations in MST1/2, LAST1/2, or NF (negative upstream regulators of YAP). In another analysis, a high YAP signature was associated with poor prognoses in the TCGA mesothelioma cohort (HR: 2.675; 95% CI: 1.63-4.389; P < 0.001) 22 .
In conclusion, our results suggested that the effects of YAP on development, growth, and invasion of UM in affected patients were much less than the effects observed in experimental studies. However, YAP may contribute to the failure of anti-cancer therapies in patients with UM. It has been shown that YAP is important for acquisition of resistance to targeted therapy in tumors with KRAS and BRAF mutations, and that YAP can induce immune evasion by increasing the expression of PD-L1 [16][17][18][19]30 . YAP activation by the GNAQ and GNA11 mutations may explain the failure of MEK1/2 inhibitors or immunotherapy in patients with UM. Additional studies are therefore needed to determine the exact role of YAP in the treatment of UM.
Methods the tcGA cohort and data acquisition. mRNA expression (RNA Seq V2, RSEM) data were obtained from the TCGA dataset for 80 patients with UM via Firebrowse (Broad Institute, Cambridge, MA, USA). RSEM is a tool for quantifying transcript abundances from RNA-seq data 31 . The TCGA provides RSEM-normalized data for RNA-seq; therefore, we used these data. Clinicopathological characteristics and tumor size data were also downloaded from Firebrowse. Updated follow-up survival data were downloaded using R package TCGA biolinks (http://bioconductor.org/packages/release/bioc/html/TCGAbiolinks.html).
the local cohort. This study analyzed a consecutive cohort of patients with UM who underwent surgical resection of primary tumors (enucleation or sclerouvectomy, 72 patients) or metastatic tumors (liver metastasis, 16 patients) at Severance Hospital (Yonsei University College of Medicine, Seoul, Republic of Korea) between January 2001 and December 2012. The medical records of patients who met the following inclusion criteria were retrospectively reviewed: (1) pathological confirmation; (2) presence of formaldehyde-fixed, paraffin embedded tissue blocks with adequate remaining tissue for additional immunohistochemical IHC staining of slides; and (3) available clinical follow-up data. This research adhered to the tenets of the Declaration of Helsinki. The Institutional Review Board/Ethics Committee approved this retrospective study (Severance Hospital, Yonsei University Health System, IRB no. 4-2016-0300), and informed consent for the surgical procedure was obtained from all participants before surgery.
Gene set enrichment analysis (GSeA) and GSVA analysis. GSEA is a popular framework for condensing information from gene expression profiles into a pathway or signature summary 32 . GSVA is a method that estimates variation of pathway activity over a sample population 33 . We used the GSEA and GSVA methods to estimate the YAP activity from RNA-seq data. GSVA was performed using the R package GSVA. A list of genes for the YAP signature was determined using MSigDB (CORDENONSI_YAP_CONSERVED_SIGNATURE, Broad Institute) 34 . The normalized count data calculated by expectation maximization analysis were incorporated as a matrix in R 31 . The enrichment score was then calculated by GSVA for each tumor sample in the matrix data using the following arguments: kcdf = "Poisson", min.sz = 5, max_sz = 500, mx.diff = TRUE. GSEA was performed using the C6 MSigDB gene set database to test whether the YAP signature was enriched in the high YAP signature group, compared with the low YAP signature group 32 . For detailed information, see: http://software.broadinstitute.org/gsea/index.jsp and Supplementary Table 2. iHc staining for YAp. A representative area of each tumor on the formalin-fixed, paraffin-embedded block was punched out and a 5-mm tissue core was placed into a recipient block. Two tissue cores were extracted to minimize extraction bias. To generate a tissue microarray, each tissue core was assigned. Tissue microarray blocks were used to obtain 5-μm-thick sections. IHC was performed with an anti-YAP antibody (dilution: 1:200; sc-101199; Santa Cruz Biotechnology, Dallas, TX, USA) in an automated immunohistochemical staining instrument (Ventana BenchMark XT; Ventana Medical System, Oro Valley, AZ, USA), in accordance with