Expression of the Hippo transducer TAZ in association with WNT pathway mutations impacts survival outcomes in advanced gastric cancer patients treated with first-line chemotherapy

Over the past two decades, a wave of studies in flies elucidated the central role of the Hippo pathway in organ development [1]. Ablation of a set of genes including Warts (wts), Hippo (hpo), Salvador (sav) and Mob as tumor suppressor (mats) led to a remarkable tissue overgrowth, a process tied to increased cellular proliferation and reduced apoptosis [2‐11]. These alterations were phenocopied upon the forced over-expression of the transcriptional co-activator Yorkie (yki) [12]. Complemented by functional and biochemical evidence, studies in Drosophila deciphered the functional architecture of the “Salvador–Warts–Hippo” (SWH) pathway, and have been instrumental for characterizing the Hippo pathway in mammals. Indeed, the use of conditional knockout alleles and inducible transgenic mice revealed that manipulation of Hippo pathway components resulted in tissue overgrowth and tumorigenesis [13, 14]. Functionally, Hippo is organized into a core regulatory module and a transcriptional module. The first is composed by the kinases sterile 20-like kinase 1 and 2 (MST1 and MST2; Hpo in Drosophila) and large tumor suppressor 1 and 2 (LATS1 and LATS2, Wts in Drosophila), together with the adaptor proteins Salvador homolog 1 (SAV1; Sav in Drosophila) and MOB kinase activator 1A and 1B (MOB1A and MOB1B; Mats in Drosophila). The latter encompasses the transcriptional cofactors yes-associated protein and its paralog transcriptional co-activator with PDZ-binding motif (YAP and TAZ, respectively; Yki in Drosophila), along with their transcriptional partners TEA domain-containing sequence-specific transcription factors (TEAD1-4; Scalloped in Drosophila) [1]. The core module orchestrates a phosphorylation cascade that results in the inhibition of YAP/TAZ, promoting their nuclear exclusion, cytoplasmic retention and proteasomal degradation [14‐18]. When inactivated, or in the presence of stimuli that bypass its function, YAP/TAZ accumulate into the nucleus, interact with their transcriptional partners and ultimately promote the transcription of target genes. Given that loss-of-function of Hippo kinases and adaptors fuelled tumor formation in animal models, and a similar outcome was observed upon the forced expression of Hippo transducers, Hippo was designated as a tumor suppressive signaling deputed to inhibit the oncogenic proteins YAP and TAZ [1].

Hippo signaling lies at the centerpiece of an intricate molecular network [19, 20]. Indeed, a number of regulatory branches modulate its activity, spanning from cell polarity and cell adhesion factors to kinases acting upstream the regulatory module, mechanical forces (mechanotransduction), G-protein-coupled receptors (GPCRs) and metabolic routes [1]. An emerging level of regulation refers to the cooperation between Hippo and the Wnt pathway [1]. Central in the regulation of the Wnt signaling is the β-catenin destruction complex [21]. This is composed by a set of proteins that, in the absence of Wnt ligand stimulation, retains β-catenin in the cytoplasm and enables its degradation, thus preventing β-catenin nuclear translocation and transcription of target genes [21]. The crosstalk between Hippo and Wnt prevalently takes place at the level of β-catenin regulation [22, 23]. Two not mutually exclusive models have been proposed that functionally concatenate these two pathways. The first envisions the incorporation of YAP/TAZ in the β-catenin destruction complex [22]. When the Wnt pathway is in the off state, YAP/TAZ participate in β-catenin degradation, whereas stimulation by Wnt ligands disassembles the complex promoting nuclear accumulation of both YAP/TAZ and β-catenin [22]. The second model proposes that Adenomatosis Polyposis Coli (APC), a central component of the β-catenin destruction complex, serves as a scaffold protein whose correct function is instrumental for the activation of Hippo kinases and consequent inhibition of YAP/TAZ [23]. Consistently, loss of APC disables Hippo-mediated control of YAP/TAZ [23].

Functional in vitro and in vivo studies linked aberrant activation of YAP/TAZ to the progression of gastric cancer (GC) [24], and the inhibition of the YAP/TAZ–TEAD interaction achieved with a Vgl-like-4—(VGLL4) mimicking peptide severely impaired GC cell survival [25]. Moreover, the comprehensive characterization of GC carried out by The Cancer Genome Atlas (TCGA) network revealed oncogenic mutations in central Wnt pathway components, including CTNNB1 (β-catenin), APC and FBXW7 (F-box/WD repeat domain-containing 7), an antagonist of the Wnt signaling that targets β-catenin for degradation [26]. On this ground, we hypothesized that the Hippo–Wnt pathway crosstalk may be active in GC, conferring more aggressive molecular traits that translate into adverse survival outcomes. To test this hypothesis, tissue samples from 86 GC patients treated with first-line chemotherapy, either in prospective phase II trials or in routine clinical practice [27‐30], were retrospectively evaluated by immunohistochemistry (IHC) for assessing the expression of YAP and TAZ. Immunohistochemical characterization was integrated with targeted DNA next-generation sequencing (NGS) analysis of CTNNB1, APC and FBXW7.

In the present study, we examined the expression of the Hippo transducers YAP/TAZ together with mutations in central components of the Wnt pathway in a relatively large series of advanced GC patients treated with chemotherapy in the first-line setting. Approximately half of the patients examined were treated in the context of prospective phase II trials [27‐30]. This study, which is hypothesis-generating by nature, capitalizes on a growing body of evidence that converge on assigning to the Hippo–Wnt pathway cooperation a central role in three intertwined processes, namely organ development, tissue repair after injuries and tumorigenesis [22, 23]. Collectively, our results indicate that: (i) a subset of GC is characterized by a signature denoting deregulation of both Hippo and Wnt, (ii) the coexistence of nuclear TAZ expression and pathogenic Wnt pathway mutations seems to be predictive of shorter PFS, and then reduced efficacy of first-line chemotherapy, and (iii) the TAZ^pos/WNT^mut signature may also represent an adverse prognostic factor. To our knowledge, this is the first report striving to address the clinical significance of the Hippo–Wnt crosstalk in GC. Earlier studies suggested that YAP/TAZ are often expressed in GC, which is consistent with our data [24, 36‐38]. Nevertheless, studies reported so far have described small-sized case series without a clear focus on therapeutic outcomes (e.g. by pooling data concerning patients with various disease stages and prognosis), or have been conducted in specific disease entities which are not necessarily representative for the overall category of advanced GC (e.g. signet ring cell carcinoma, gastroesophageal junction cancers) [24, 36‐38].

In our opinion, our findings raised a number of points that may streamline the identification of Hippo/Wnt-related predictive factors in GC. First, the molecular characterization of GC delineated four distinct molecular subtypes: chromosomal instability (CIN), microsatellite instability (MSI), genomically stable (GS) and Epstein–Barr virus (EBV)-positive [26]. Mutations in Wnt pathway components were observed across all non-hypermutated subtypes. Conversely, hallmarks of GS–GC are RHOA and CDH1 mutations, together with CLDN18‐ARHGAP26 fusions. All these alterations suggest genetically-driven deregulation of the Hippo pathway. Indeed, Rho GTPases are involved in the activation of YAP/TAZ and in the inhibition of Hippo kinases via two distinct mechanisms: (i) stimulation by soluble factors that act through G-protein-coupled receptors (GPCRs) and Rho GTPases [39‐43], and (ii) mechanical cues, such as extracellular matrix stiffness and changes in cell geometry, attachment status and density, that regulate YAP/TAZ through Rho GTPases and the remodeling of the F-actin cytoskeleton [44‐46]. Next, CDH1 encodes for the cell–cell adhesion molecule E-cadherin, the central component of adherens junctions. E-cadherin is an established positive regulator of MST1/2 activity, whereas the E-cadherin-associated protein α-catenin sequesters YAP/TAZ in the cytoplasm, hindering their nuclear translocation [47‐49]. Consistently, disruption of the E-cadherin–catenin complex at the cell–cell junction fuels YAP/TAZ activation [47‐49]. Finally, the CLDN18‐ARHGAP26 fusion implies defects in CLDN18 and ARHGAP26. CLDN18 encodes for Claudin 18, a component of tight junctions (TJs) [50]. TJ proteins promote activation of Hippo kinases and/or sequester YAP/TAZ in the cytoplasm, whereas ARHGAP26 encodes for the Rho-Type GTPase-Activating Protein 26 [51‐57]. These observations suggest that GS–GC is characterized by multiple defects in cell–cell adhesion mechanisms that, in turn, can propel YAP/TAZ activation. Different considerations apply to EBV-related GC. Experimental models of liver and cervical tumors are beginning to shed light on the connection between viral proteins and YAP/TAZ. For instance, the hepatitis B virus X protein (HBx) up-regulates YAP promoting the growth of hepatoma cells, whereas in hepatocellular carcinoma cell lines the transcriptional activator PreS2 up-regulates TAZ via the suppression of miRNA-338-3p [58, 59]. Likewise, in cervical cancer cells the HPV E6 protein protects YAP from proteasome-dependent degradation in a process that ignites cancer cell proliferation [60]. Remarkably, a distinctive feature of EBV-associated GC is the extreme DNA hypermethylation, and both MOB1B and WWTR1 (the gene encoding for TAZ) present frequent promoter hypermethylation [26]. Thus, more tailored investigations are needed in the future, which specifically take into account the molecular classification of GC and the underlying molecular portraits characterizing the different subtypes.

Another aspect that deserves mention is that the activity of Hippo and Wnt is modulated by negative feedback loops. Indeed, the YAP/TAZ–TEADs and β-catenin-TCF/LEF complexes also promote the transcription of negative pathway regulators [61, 62]. For instance, the YAP/TAZ–TEADs complex controls the activity of Hippo kinases by inducing the expression of LATS2, and mediates the transcription of neurofibromin 2 (NF2, also known as Merlin), an established positive regulator of LATS1/2 kinase activity [61]. Aware of these mechanisms, our original experimental workflow envisioned targeted RNA sequencing for evaluating two signatures denoting the activation of YAP/TAZ and Wnt. The logic behind this was to carry out an extensive characterization at three different levels (protein, transcript and gene), which would have enabled us to investigate negative feedback loops. Even though this task was halted owing to excessive RNA degradation in the majority of samples, transcript-level analysis will be further pursued in future studies from our research team.

Finally, Hippo and Wnt are two pieces of a wider cross-regulation process involving multiple signaling pathways [i.e. Hedgehog, Notch and Bone Morphogenetic Protein (BMP)], whose activity is central in organ development, tissues homeostasis, stem cell fate and tumorigenesis [19, 20]. Albeit these pathways are not targeted by genetic events in GC [26], the evaluation of biomarkers functioning as readout for their activation may add further granularity, allowing the evaluation of co-regulated signaling avenues.

Our data pointed to the combined activation of two oncogenic avenues, YAP/TAZ and Wnt, as potential biomarkers for predicting the efficacy of chemotherapy in GC patients. Considering the intricate molecular network that co-regulates YAP/TAZ and Wnt, we believe that more comprehensive, subtype-restricted, pathway analyses will be instrumental to gain a better understanding on the Hippo–Wnt pathway crosstalk and its clinical implications.