Fig 2: MAP3K8 controls proliferation, migration and invasion of ovarian cancer cells and tumour growth in mouse xenograft models.(a) Left: growth curves of SKOV3 and IGROV-1 ovarian cancer cell lines (OCCLs) treated either with MAP3K8 kinase inhibitor (KI) or with DMSO (vehicle medium for KI), for the indicated times (n=3). Right: bar plots show the doubling times of SKOV3 and IGROV-1 OCCL either untreated (DMSO) or treated with MAP3K8 kinase inhibitor (KI) (n=3). (b) Left: bar plots represent the percentage of SKOV3 and IGROV-1 OCCL in G1, S and G2 phases of the cell cycle, in control condition (DMSO) or following treatment with MAP3K8 KI, for the indicated times. For each cell line, the T0 time point represents the state of the cells just before any treatment (n=3). Right: representative cell-cycle distribution of SKOV3 and IGROV-1 OCCL without (DMSO) or with KI treatment. (c) Bar plots represent cell migration (left panels) or invasion (right panels) of SKOV3 and IGROV-1 OCCL either treated with MAP3K8 KI or with the vehicle medium for KI (DMSO) (n?3). (d) Left up: representative western blot showing MAP3K8 protein levels in stable cell lines expressing non-targeting shRNA (shCtrl) or two different MAP3K8-targeting shRNA (shMAP3K8_1 and shMAP3K8_2), derived from SKOV3 and IGROV-1 OCCL, as indicated. Actin is used as an internal control for protein loading. Left down: bar plot showing the ratio of MAP3K8/Actin protein levels, as assessed by densitometry analysis of western blots (as shown above) and expressed as percentage of shCtrl (n=3). Right, Growth curve of stable cell lines (shCtrl, shMAP3K8_1 and shMAP3K8_2) derived from SKOV3 and IGROV-1 OCCL, for the indicated times (n=3). (e) Tumour growth curves over time from one representative experiment of xenografted stable cell lines expressing shCtrl, shMAP3K8_1 or shMAP3K8_2 derived from SKOV3 and IGROV-1 OCCL (N?10 tumours per group). For all panels, data are shown as mean±s.e.m. P values are based on a one-sample t-test (c) and on the Student's t-test (a,b,d,e). *P=0.05; **P=0.005 and ***P=0.0005. n stands for the number of replicated independent experiments.

Fig 3: MAP3K8 protein level correlates with its kinase activity and MEK activation in human HGSC.(a) Representative western blots showing MAP3K8, P-MEK, MEK, P-ERK and ERK protein levels from human HGSC protein extracts enriched in epithelial cells. On a total number of 108 HGSC analysed, representative results on 54 patients are shown. Actin is used as an internal control for protein loading. (b) Correlation plot between MAP3K8 protein levels evaluated by western blots (as shown in a) and histological scoring in epithelial cells (as shown in Fig. 1a,b) from 108 HGSC tumour samples. (c) Correlation plots between P-MEK/MEK and P-ERK/ERK (left); MAP3K8/actin and P-MEK/MEK (middle); MAP3K8/actin and P-ERK/ERK (right) in human HGSC. Values have been assessed by densitometry analysis of western blots (as those shown in a). (d) Western blots showing MAP3K8 phosphorylation on threonine 290 (T290) or on serine 400 (S400), as well as P-MEK and MEK protein, before (input) or after Myc-Tag-specific antibody immunoprecipitation (IP:tag) from protein extracts of SKOV3 cells transfected with empty vector (Ctrl) or increasing amounts of myc-tagged MAP3K8 expression vector (tagged MAP3K8). The Myc-tag is used as an internal control for MAP3K8 protein levels (tag). (e) Correlation plot between P-MAP3K8 (T290) and MAP3K8 protein levels (left panel) and between P-MAP3K8 (T290) protein levels and P-MEK/MEK ratio (right panel), as assessed by densitometry analysis of western blots (as shown in d) (n=4). (f) Western blots showing P-MEK, MEK, P-ERK, ERK, P-p90RSK and p90RSK protein levels in mouse xenografted tumours (N?6 tumours per group) derived from shCtrl, shMAP3K8_1 or shMAP3K8_2 SKOV3 stable cell lines (tumour growth described above, Fig. 2e). Actin is used as an internal control for protein loading. (g) Scatter plot of P-MEK/MEK (left), P-ERK/ERK (middle) and P-p90RSK/p90RSK (right) ratios in xenograft tumours derived from shCtrl, shMAP3K8_1 and shMAP3K8_2 SKOV3 stable cell lines, as assessed by densitometry analysis of the western blots shown in f. P values are based on the Student's t-test. Data are shown as mean±s.e.m. (N?6 tumours per group). For all correlation plots, correlation coefficients s and P values are based on Spearman's rank correlation test. n stands for the number of replicated independent experiments. a.u., arbitrary unit.

Fig 4: MAP3K8 exhibits pro-tumorigenic properties in HGSC and is a predictive marker for treatment with MEK inhibitor.Among epithelial ovarian cancers, the serous histological subtype is the most abundant and is subdivided into two subgroups according to grade (low versus high grade). Seventy percent cases of low-grade ovarian carcinomas exhibited KRAS or BRAF mutations and subsequent MEK activation, which led to the use of MEK inhibitors as a new line of treatment in recent clinical trials with some relevant efficiency. In contrast, KRAS or BRAF mutations remain extremely rare in HGSC (less than 1%). Yet, we demonstrate here that the MEK pathway is constitutively activated in half cases of HGSC. MAP3K8 (also referred as TPL-2 or COT) accumulation in HGSC mediates this constitutive MEK/ERK pathway activation, MAP3K8 being the other MEK kinase, besides RAF. MAP3K8 contributes to tumour growth via p90RSK-dependent regulation of cyclin D1 and FAK, key regulators of the G1/S transition and adhesion dynamics, respectively. Therefore, our study demonstrates for the first time that the use of MEK inhibitors should not be restricted to low-grade ovarian carcinomas, as they are currently in clinical trials, but could be extended to patients suffering from HGSC characterized by MAP3K8 accumulation.

Fig 5: Correlations of PDE4D expression and TPL2 phosphorylation with human MIBC development. (a) H&E and IHC staining of PDE4D in the representative bladder tumor tissues and the adjacent normal bladder tissues. (Scale bar: 200 µm). (b) Statistical data of PDE4D staining in the bladder tumor tissues and adjacent normal bladder tissues. (c) Because all specimens had the same positive-staining scores of PDE4D, we used the staining intensity score to replace the staining index. All the specimens were segregated into two groups based on their staining index (lower expression < staining index 2; higher expression = staining index 2) and compared to observe the variations. (d) Kaplan-Meier survival curves based on the PDE4D expression levels to demonstrate the prognostic importance of PDE4D. (e) H&E and IHC staining of p-TPL2 in the representative bladder tumor tissues and adjacent normal bladder tissues. (Scale bar: 200 µm). (f) Statistical data of p-TPL2 staining in the bladder tumor tissues and adjacent normal bladder tissues. (g) All specimens were segregated into two groups based on their staining index (high: = 4 and low: < 4) and compared to observe the variations. (h) Kaplan-Meier survival curves based on the p-TPL2 levels to demonstrate the prognostic importance of pTPL2. (i, j) The data derived from TCGA database were analyzed and PDE4D mRNA levels were significantly down-regulated in the bladder tumor compared to the bladder normal tissue (i) and correlated with the poor prognosis (j). Error bars indicate mean ± SEM. Statistical significances of differences between experimental groups were evaluated using the Wilcoxon signed rank test (c and g), unpaired Wilcoxon test (i), and log-rank test (d, h, and j). P < 0.01 was considered as statistically significant value