Proteasome stress activates YAP/TAZ through the RAP2–MAP4Ks–LATS1/2 pathway and its therapeutic implications in solid tumors

Significance

This study identifies proteasome stress as an upstream regulator of the Hippo signaling pathway. We show that proteasome stress causes ubiquitination of RAP2, leading to its inactivation, which in turn suppresses the MAP4Ks–NF2–LATS1/2 cascade and activates yes-associated protein (YAP)/transcriptional coactivator with PDZ-binding motif (TAZ). This connects proteostasis to cell growth regulation and uncovers a stress-responsive mechanism controlling the Hippo pathway. In cancer, YAP/TAZ activation during proteasome stress enhances resistance to proteasome inhibitors in solid tumors. Notably, combining YAP/TAZ inhibitors with proteasome inhibitors restores drug sensitivity and reduces tumor growth in models of diffuse-type gastric cancer. Our findings provide mechanistic insights and potential therapeutic strategies relevant to cancer biology and translational medicine.

Abstract

Tumor cells heavily depend on proteasome-mediated protein turnover, making the proteasome an attractive therapeutic target. Clinically, proteasome inhibitors are effective against hematologic cancers but show limited success with solid tumors, and the reasons for this difference are not well understood. Activation of yes-associated protein (YAP)/TAZ, the downstream effectors of the Hippo pathway, is a key mechanism behind drug resistance in cancers. Here, we demonstrate that proteasome stress acts as an upstream signal of the Hippo pathway in solid tumor cells. When the proteasome is inhibited, RAP2 undergoes ubiquitination and becomes inactive, which in turn disrupts the RAP2–MAP4Ks–NF2–LATS1/2 signaling pathway, leading to the activation of YAP/TAZ. YAP/TAZ activation promotes cell survival and resistance to proteasome inhibitors. Conversely, blocking YAP/TAZ can overcome this resistance and restore cancer cell sensitivity to these drugs. In diffuse-type gastric cancer—an aggressive solid tumor with a poor prognosis and limited treatment options—combined inhibition of the proteasome and YAP/TAZ effectively suppresses tumor growth. Therefore, this study identifies proteasome stress as an upstream signal of the Hippo pathway and provides a mechanistic basis for combination cancer therapy.

The Hippo signaling pathway controls organ size, tissue regeneration, and tumorigenesis (1–6). The core components of the Hippo pathway constitute a kinase network in which mammalian STE20-like kinases (MST1/2, Hippo kinases) and mitogen-activated protein kinase kinase kinase kinases (MAP4K1-7, Hippo-like kinases) phosphorylate and activate large tumor suppressor kinases (LATS1/2), while LATS1/2 subsequently phosphorylate and suppress the activity of yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ, also known as WWTR1). YAP and TAZ are homologous transcriptional cofactors often referred to collectively as YAP/TAZ. They are generally considered to possess overlapping functions and regulatory mechanisms. Meanwhile, accumulating evidence indicates that YAP and TAZ may exert distinct functions and are regulated by different upstream signals (7). Phosphorylated YAP/TAZ is sequestered in the cytoplasm, preventing access to downstream target genes. However, when upstream kinases are inactivated, YAP/TAZ translocates into the nucleus and, by interacting with transcription factors such as TEA domain family members (TEADs), drives the expression of genes involved in cell proliferation, differentiation, and survival (1, 2, 8–10).

Aberrant activation of YAP/TAZ has been frequently observed in various solid tumors, including gastric, liver, breast, and colorectal cancers (11–13). Elevated nuclear localization of YAP/TAZ in tumors is often correlated with aggressive clinical features and poor prognosis (11, 13, 14). In support, the activation of YAP/TAZ in mice by selective knockout of upstream components of the Hippo pathway can drive atypical hyperplasia or tumorigenesis in affected organs (6, 11, 15, 16). Based on these results, the Hippo signaling pathway is recognized as a critical regulator in tumorigenesis.

YAP/TAZ also mediates resistance to chemotherapy, targeted therapy, and immunotherapy, hence representing a target for overcoming drug resistance in cancer management (17–20). For example, in gastric cancer, increased YAP expression has been linked to resistance to the anti-HER2 antibody trastuzumab. Specifically, YAP activation was significantly upregulated in trastuzumab-resistant patient-derived xenograft models, while YAP knockdown resensitized gastric cancer cells to the treatment (17). Similarly, YAP activation drives resistance to RAF inhibitors like vemurafenib in BRAF-mutant thyroid cancer (21). Inhibition of YAP/TAZ sensitizes these resistant cells, highlighting its role as a critical mediator of resistance to targeted therapy. YAP/TAZ also contributes to immune evasion and resistance to immunotherapies. Several studies have identified PD-L1 as a transcriptional target of the YAP/TAZ–TEAD complex, leading to the upregulation of PD-L1 in various cancer types, including breast, lung, and melanoma (20, 22–24). Given the pivotal role of YAP/TAZ in driving resistance across multiple therapeutic contexts, targeting YAP/TAZ represents a promising strategy to overcome treatment failures and improve patient outcomes.

The proteasome–ubiquitin system maintains protein homeostasis by degrading misfolded and damaged proteins (25, 26). The proteasome is frequently upregulated in cancer cells to support stemness, rapid proliferation, and survival under stress conditions (26–28). Studies have demonstrated that tumor cells depend more on proteasome activity than normal cells (26, 29). Indeed, proteasome inhibitors (PIs) are regarded as promising anticancer agents, as they block the proteolytic activity of the 26S proteasome, thereby disrupting tumor cell proliferation and survival, ultimately leading to tumor cell death (30, 31). Bortezomib (BTZ, sold under the brand name Velcade), the first Food and Drug Administration (FDA)-approved proteasome inhibitor that targets chymotrypsin-like proteasomal activity by reversibly binding to the β5 subunit of the proteasome, has become a cornerstone in the frontline treatment of hematologic malignancies, including multiple myeloma and mantle cell lymphoma (30–32). Building on its clinical success, next-generation proteasome inhibitors such as carfilzomib and ixazomib, designed with enhanced potency and specificity, have received approval for use in patients with relapsed and refractory multiple myeloma (32–35).

However, despite their success in treating hematologic malignancies, proteasome inhibitors appear less effective in solid tumors. Clinical trials have generally failed to show significant therapeutic benefits in solid tumors, attributed to the pharmacokinetic limitations of these agents and the intrinsic resistance mechanisms inherent to solid tumor cells (36, 37). Addressing these resistance mechanisms could open new therapeutic avenues for proteasome inhibitors in solid tumors, potentially expanding their clinical utility and providing alternative treatment strategies for more cancer types (37, 38).

In this study, we have tested the hypothesis that proteasome stress induces YAP/TAZ activation, contributing to resistance against proteasome inhibitors in solid tumors. We have investigated whether cotargeting the proteasome and YAP/TAZ can overcome this resistance and enhance therapeutic efficacy, particularly in diffuse-type gastric cancer (DGC), a subtype with poor prognosis and limited treatment options. By elucidating the proteasome–YAP/TAZ axis in DGC, we aim to provide a therapeutic strategy for treating this aggressive cancer and offer insights into overcoming resistance mechanisms in various solid tumors.

Results

Proteasome Inhibition Activates YAP/TAZ in Solid Tumor Cells.

Analysis of the Cancer Cell Line Encyclopedia project dataset showed that YAP and TAZ expression were absent in hematopoietic and lymphoid cancer cell lines (Fig. 1A). Meanwhile, YAP, TAZ, and CYR61 (a known YAP/TAZ target gene) were undetectable at the protein level in the multiple myeloma cell lines RPMI 8226 and MM.1S, whereas they were readily detectable in various solid tumor cell lines (SI Appendix, Fig. S1A). Based on these observations, we hypothesized that the differential sensitivity to proteasome inhibitors observed between hematologic malignancies and solid tumors might be linked to their distinct YAP/TAZ expression patterns. Specifically, we proposed that proteasome inhibitor treatment could trigger prosurvival YAP/TAZ activity in solid tumors but not in hematologic malignancies (Fig. 1B).

Fig. 1.

Proteasome inhibition activates YAP/TAZ in solid tumor cells (A) Absence of YAP1 (Upper) and WWTR1 (Lower) expression in hematopoietic and lymphoid cancer cell lines. (B) Proposed model for the study hypothesis. (C) BTZ induces dephosphorylation of YAP in various solid tumor cell lines. Cells were treated with vehicle or BTZ (1 μM, 6 h). (D) BTZ induces elevated TAZ protein levels. Cells were treated with vehicle or BTZ (1 μM, 6 h). (E and F) BTZ induces nuclear localization of YAP/TAZ. AGS and HCT116 cells were treated with BTZ (1 μM, 6 h). [Scale bar, 10 μm (E) or 5 μm (F).] (G) Carfilzomib and epoxomicin induce dephosphorylation of YAP. AGS and HCT116 cells were treated with carfilzomib (1 μM), epoxomicin (1 μM), or BTZ (1 μM) for 6 h. (H) siRNA-mediated knockdown of proteasome subunits induces YAP dephosphorylation. (I) Proteasome inhibitors induce upregulation of CTGF expression. CTGF mRNA levels were measured by reverse transcription PCR (RT-PCR). (J) siRNA-mediated knockdown of proteasome subunits induces upregulation of CTGF expression. CTGF mRNA levels were measured by RT-PCR. Statistical data are presented as mean ± SD (n = 3), with statistical significance determined by Student’s t test (C and D) or one-way ANOVA (G–J). Significance levels are denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, with “ns” indicating not significant. See also SI Appendix, Fig. S1.

We treated the gastric cancer cell line AGS with the small-molecule proteasome inhibitor BTZ (marketed as Velcade) at various concentrations and durations. Treatment with 1 μM BTZ for 6 h resulted in a pronounced and consistent reduction in YAP phosphorylation, as evidenced by decreased pYAP(S127) levels and mobility shifts observed in Phos-tag gels (SI Appendix, Fig. S1 B and C). Similar treatment of multiple solid tumor cell lines also led to a significant reduction in YAP phosphorylation (Fig. 1C and SI Appendix, Table S1). In parallel, BTZ treatment led to an increase in total TAZ levels, whereas total YAP levels remained essentially unchanged across various solid tumor cell lines (Fig. 1D). The accumulation of active NFE2L1 served as an indicator of inhibited proteasome activity here (39, 40). Immunofluorescence staining revealed that BTZ induced nuclear accumulation of YAP/TAZ in AGS (human gastric cancer), HCT116 (human colon cancer), and SNU668 (human gastric cancer) cells (Fig. 1 E and F and SI Appendix, Fig. S1D). To rule out the possibility that YAP/TAZ activation resulted from off-target effects of BTZ, we further tested the impact of carfilzomib and epoxomicin, two other proteasome inhibitors, and siRNA-mediated knockdowns of various proteasome subunits, including PSMB1, PSMB3, and PSMB4. All these treatments induced proteasome stress, as evidenced by the induction of NFE2L1 (Fig. 1 G and H). Moreover, siRNA knockdown of proteasome subunits effectively reduced the mRNA expression of PSMB1, PSMB3, and PSMB4 and repressed cellular chymotrypsin-, caspase-, and trypsin-like activities (SI Appendix, Fig. S1 E and F). These treatments all resulted in a significant reduction in YAP phosphorylation (Fig. 1 G and H), accompanied by increased expression of YAP/TAZ target genes CTGF and CYR61 (Fig. 1 I and J and SI Appendix, Fig. S1G). On the contrary, BTZ failed to upregulate YAP/TAZ target genes in the multiple myeloma cell line RPMI 8226, likely due to the lack of YAP/TAZ expression (SI Appendix, Fig. S1H). These findings suggest that YAP/TAZ is activated upon proteasome inhibition in solid tumor cells.

MAP4Ks and LATS1/2 Inhibition Contribute to YAP/TAZ Activation under Proteasome Stress.

The phosphorylation and subcellular localization of YAP/TAZ are primarily regulated by LATS1/2 kinase activity (6, 16, 41, 42). To investigate the mechanism of YAP activation during proteasome inhibition, we first assessed the phosphorylation levels of the hydrophobic motif (HM) and the kinase activity of LATS1/2. In various cell lines, BTZ significantly reduced the phosphorylation levels of the HM domain of LATS1/2 (Fig. 2A). We also performed in vitro kinase activity assays using immunoprecipitated LATS1 from AGS, HEK293A, and HCT116 cells treated with BTZ for various durations. The kinase assays demonstrated a marked reduction in LATS1 kinase activity, evidenced by decreased YAP phosphorylation (Fig. 2B and SI Appendix, Fig. S2 A and B). The results showed that LATS1 activity was conversely associated with proteasome activity (Fig. 2B and SI Appendix, Fig. S2 A and B). Hence, proteasome inhibition likely activates YAP/TAZ by inhibiting LATS1/2 kinase activity.

Fig. 2.

MAP4Ks and LATS1/2 inhibition contribute to YAP/TAZ activation under proteasome stress. (A) Proteasome inhibition reduces pLATS (HM) levels. Cells were treated with vehicle or BTZ (1 μM, 6 h). (B) Proteasome inhibition reduces LATS1 kinase activity. AGS cells were treated with 1 μM BTZ for 1 to 6 h. (C) Two key signaling modules regulate LATS1/2 activity. (D) Proteasome inhibition does not affect MST1 kinase activity. AGS and HEK293A cells were treated with 1 μM BTZ for 1 to 6 h. (E) Proteasome inhibition reduces MAP4K4 kinase activity. AGS and HEK293A cells were treated with 1 μM BTZ for 1 to 6 h. (F) Proteasome inhibition has no effect on MST1 kinase activity. AGS cells were treated with BTZ (1 μM) for 6 h. (G) Proteasome inhibition disrupts the MAP4K4–LATS1 interaction. HEK293A cells stably expressing FLAG-MAP4K4 were treated with BTZ (1 μM) for 6 h. (H) Proteasome inhibition disrupts the MAP4K4–NF2 interaction. HEK293A cells stably expressing FLAG-MAP4K4 were treated with BTZ (1 μM) for 6 h. (I) Proteasome inhibition disrupts the LATS1–NF2 interaction. HEK293A cells were treated with BTZ (1 μM) for 6 h. (J) Proposed model illustrating the regulation of the MAP4Ks–LATS1–YAP axis in response to proteasome inhibition. Statistical data are presented as mean ± SD (B and D–I) or mean ± SEM (A) (n = 3), with statistical significance determined by Student’s t test (A and F–I) or one-way ANOVA (B, D, and E). Significance levels are denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, with “ns” indicating not significant. See also SI Appendix, Fig. S2.

LATS1/2 activity is regulated by two key signaling modules: the MST1/2–SAV1–WWC1/2/3 axis and the MAP4Ks–NF2 axis (HPO1 and HPO2, respectively; Fig. 2C) (8, 43). To gain further mechanistic insights, we assessed the protein levels of major Hippo pathway regulators. Except for MOB1, the protein levels of most Hippo components remained stable under short-term proteasome inhibition despite significant YAP/TAZ activation, while accumulation of cleaved NFE2L1, TXNIP, and p21 confirmed proteasome impairment (SI Appendix, Fig. S2C and Fig. 2A). However, the slight increase of MOB1 could not explain the decreased LATS1/2 activity and YAP/TAZ phosphorylation. We then evaluated the kinase activities of MST1/2 and MAP4Ks, the two direct upstream kinases of LATS1/2, using in vitro kinase assays. We utilized purified His-LATS1-C3 (amino acids 981 to 1130 of LATS1) as a substrate to evaluate the kinase activity of upstream kinases MST1/2 and MAP4K4 (44–46). The results indicated that, following proteasome inhibition, MST1/2 activity was unaffected, whereas MAP4K4 kinase activity was decreased (Fig. 2 D and E and SI Appendix, Fig. S2 D–F). MST1/2 phosphorylates MOB1 at threonine 35 (T35), which increases the binding between MOB1 and LATS1/2, leading to LATS1/2 activation (47). We observed no alterations in MOB1 phosphorylation or its interaction with LATS1 in response to proteasome inhibition (SI Appendix, Fig. S2G). Moreover, the kinase activity of MST1/2 was also determined using MOB1 as a substrate in an in vitro kinase assay, and no difference was detected for MST1/2 purified from control or BTZ-treated cells (Fig. 2F and SI Appendix, Fig. S2 H and I). These results suggest that MAP4Ks, but not MST1/2, are involved in YAP/TAZ activation by proteasome inhibitors.

The activation of LATS1/2 by MAP4Ks primarily depends on NF2, which serves as an adaptor for MAP4Ks and LATS1/2 (8). So, we assessed the effect of proteasome inhibition on MAP4Ks–NF2–LATS1/2 signaling. BTZ treatment did not significantly alter the protein levels or ubiquitination of NF2, MAP4Ks, and LATS1/2, whereas treatment with the E1 inhibitor MLN7243 effectively blocked their ubiquitination (SI Appendix, Fig. S2 C and J and Fig. 2A). However, as indicated by coimmunoprecipitation assays, BTZ treatment attenuated the protein–protein interactions between LATS1 and MAP4K2/4 (Fig. 2G and SI Appendix, Fig. S2K), between NF2 and MAP4K2/4 (Fig. 2H and SI Appendix, Fig. S2L), and between NF2 and LATS1 (Fig. 2I). These results suggest the MAP4Ks–NF2–LATS1/2 signaling axis is dysregulated under proteasome stress (Fig. 2J).

Proteasome Stress Induces RAP2 Ubiquitination to Repress MAP4Ks.

The activity of MAP4Ks is regulated by multiple mechanisms, including thousand-and-one amino acid protein kinase 1/2/3 (TAOK1/2/3), striatin-interacting phosphatase and kinase (STRIPAK) complex, and ras-related protein 2 (RAP2) (SI Appendix, Fig. S3A). We then dissected whether these mechanisms were responsible for MAP4Ks inhibition under proteasome stress. TAOK1/2/3 functions upstream of MST1/2 and MAP4Ks to activate LATS1/2, while also acting in parallel to directly activate LATS1/2 (48, 49). However, the in vitro kinase activity of TAOK1/2/3 was not changed in the presence of BTZ (Fig. 3A and SI Appendix, Fig. S3 B and C). Hence, proteasome inhibition does not modulate YAP activity via TAOK1/2/3.

Fig. 3.

Proteasome stress induces RAP2 ubiquitination to repress MAP4Ks. (A) Proteasome inhibition does not affect TAOK1 kinase activity. HEK293A cells were transfected with FLAG-TAOK1 and treated with BTZ (1 μM) for 1 to 6 h. (B) Proteasome inhibition does not affect PP2Ac phosphatase activity. HEK293A cells were transfected with FLAG-PPP2CA and treated with BTZ (1 μM, 6 h). (C) Proteasome inhibition reduces RAP2 activity. HEK293A and AGS cells were treated with vehicle or BTZ (1 μM, 6 h). (D) Proteasome inhibition enhances RAP2A ubiquitination. HEK293A cells were cotransfected with FLAG-RAP2A and HA-UB, followed by treatment with BTZ (1 μM, 6 h) or MLN7243 (1 μM, overnight). (E) Schematic illustration of the KR mutation sites in RAP2A. (F) RAP2A ubiquitination is regulated by K117, K148, and K150 residues. HEK293A cells were cotransfected with FLAG-RAP2A mutants and HA-UB. (G) The 11KR, 7KR, 5KR, and 3KR mutations enhance RAP2A activity. HEK293A cells were transfected with FLAG-RAP2A mutants. (H) Ubiquitination at K117, K148, and K150 regulates RAP2A activity. HEK293A cells were transfected with FLAG-RAP2A mutants. (I) RAP2A is mainly modified by K63-linked ubiquitin chains, which accumulate upon proteasome inhibition. HEK293A cells were cotransfected with FLAG-RAP2A and HA-UB, then treated with BTZ (1 μM, 6 h) or MLN7243 (1 μM, overnight). (J) NEDD4 promotes K63-linked ubiquitination of RAP2A. HEK293A cells were cotransfected with FLAG-RAP2A, HA-UB, and MYC-NEDD4. (K) NEDD4 overexpression suppresses RAP2 activity. HEK293A cells were transfected with MYC-NEDD4. (L) A proposed model illustrating the regulation of the RAP2–MAP4Ks–LATS1–YAP axis in response to proteasome inhibition. Statistical data are presented as mean ± SD (n = 3), with statistical significance determined by Student’s t test (B and C) or one-way ANOVA (A, G, H, and K). Significance levels are denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.005, and ****P < 0.001, with “ns” indicating not significant. See also SI Appendix, Fig. S3.

STRIPAK is a protein phosphatase 2A (PP2A) complex that regulates the Hippo signaling pathway by dephosphorylating and thereby modulating the kinase activity of MAP4Ks and MST1/2 (50, 51) (SI Appendix, Fig. S3A). We assessed protein components within the STRIPAK complex–PP2Ac, STRIP1, SLMAP, and STRN1/3–and found no significant alterations in their protein levels (SI Appendix, Fig. S3D). However, the knockout of STRIP1/2 resulted in a marked increase in YAP phosphorylation, thereby inhibiting the ability of the proteasome inhibitor to dephosphorylate YAP. When a functional STRIP1 was put back into these knockout cells, the ability of BTZ on YAP phosphorylation was effectively restored (SI Appendix, Fig. S3E). Consistent results were observed for the subcellular localization of YAP in immunofluorescence assays (SI Appendix, Fig. S3 F and G). In the absence of STRIPAK activity, the hyperactivation of MAP4Ks and MST1/2 may mask the effect of proteasome inhibitors on MAP4Ks, LATS1/2, and YAP/TAZ. We then investigated whether proteasome inhibitors regulated the activity of STRIPAK by performing in vitro phosphatase assays. The results revealed that BTZ treatment did not significantly alter total PP2A activity or STRIP1- and STRN-associated PP2A activity (Fig. 3B and SI Appendix, Fig. S3 H and I). These findings indicate that, although STRIPAK is required for proteasome inhibitors to regulate YAP/TAZ, its activity is not directly targeted by the proteasome.

Previous studies have established that RAP2 (RAP2A/B/C) regulates MAP4Ks activity in the Hippo pathway (52) (SI Appendix, Fig. S3A). We then investigated whether RAP2 was regulated by proteasome by using a RalGDS-RBD pull-down assay to evaluate RAP2 activity directly (53, 54). RAP2 activity was markedly decreased following BTZ treatment, although the total protein levels of RAP2 were unchanged (Fig. 3C). Proteasome inhibition typically results in the accumulation of ubiquitinated proteins, and RAP2 activity is known to be influenced by its ubiquitination status (55, 56). We hypothesized that proteasome inhibition might reduce RAP2 activity by inducing its ubiquitination. Indeed, we detected a robust increase in the ubiquitination level of RAP2A in the presence of BTZ (Fig. 3D). Ubiquitination predominantly occurs on lysine (K) residues. RAP2A contains 11 K residues, three of which (K117, K148, and K150) are modified by ubiquitination and are highly conserved across RAP2A, RAP2B, and RAP2C (55, 56). To further elucidate the ubiquitination of RAP2A in response to proteasome stress, we mutated these three K residues into arginine (3KR) (Fig. 3E). Notably, ubiquitination signals were nearly abolished in the RAP2A 3KR mutant, and additional K to R conversions (11KR, 7KR, and 5KR) showed no further effect (Fig. 3 E and F). However, single-residue mutations at K117, K148, or K150 had a minimal impact on RAP2A ubiquitination (Fig. 3F). Consistently, the activity of RAP2A was elevated in the 11KR, 7KR, 5KR, and 3KR mutants, whereas the increase was marginal in K117R, K148R, or K150R mutants (Fig. 3 G and H). These findings indicate that ubiquitination at K117, K148, and K150 under proteasome stress cooperatively regulates RAP2A activity. K48- and K63-linked ubiquitin chains are the two most abundant ubiquitination (57). Immunoprecipitation showed that RAP2 is primarily modified by K63- rather than K48-linked ubiquitination. Proteasome inhibition further promotes the accumulation of K63-linked ubiquitinated RAP2 (Fig. 3I). However, the RAP2A 3KR mutant abolished K63-linked ubiquitination, whereas single mutations at K117, K148, or K150 had minimal effects (SI Appendix, Fig. S3J). Previous studies have suggested NEDD4 and Cullin5 as potential E3 ligases for RAP2 (55, 56). We found that NEDD4, but not Cullin5, promotes K63-linked ubiquitination of RAP2A and suppresses RAP2 activity upon overexpression (Fig. 3 J and K and SI Appendix, Fig. S3 K and L). Based on these findings, we propose the following mechanism for YAP/TAZ activation under proteasome stress: Proteasome stress leads to enhanced ubiquitination and decreased activity of RAP2, which in turn suppresses the MAP4Ks–NF2–LATS1/2 signaling axis, ultimately leading to the activation of YAP/TAZ (Fig. 3L).

The Proteasome–YAP/TAZ Axis Is Conserved in Various Solid Tumors.

We then analyzed whether a connection exists between proteasome and YAP/TAZ activities in solid tumor specimens. Previously, we reported that the YAP/TAZ activity signature had the most robust prognostic value in stomach cancer among different cancer types (58). The Cancer Genome Atlas (TCGA) stomach cancer transcriptome analysis revealed a correlation between proteasome subunit genes and Hippo pathway target genes (Fig. 4A). Specifically, the expressions of PSMD14 or PSMG1 were found to be negatively correlated with the YAP/TAZ target genes CTGF, AMOTL2, or CYR61 (Fig. 4B and SI Appendix, Fig. S4A). Similar trends were observed in other solid tumors, such as colon and esophageal cancers (SI Appendix, Fig. S4 B–E). Notably, when incorporating TCGA molecular classification into our analysis, genomically stable (GS) subtype gastric cancer emerged as a subgroup characterized by lower expression of proteasome subunit genes and higher expression of YAP/TAZ target genes (Fig. 4A).

Fig. 4.

The proteasome–YAP/TAZ axis is conserved in gastric cancer. (A) The expression of proteasome subunits tends to negatively correlate with YAP target genes. CIN: Chromosomal Instability, EBV: Epstein–Barr Virus Positive, MSI: Microsatellite Instability, GS: Genomically Stable. (B) The expression of PSMD14 is negatively correlated with CTGF, AMOTL2, or CYR61. Spearman’s correlation rho and P values are shown. (C) BTZ treatment led to an increase in UB-K48 protein levels. AGS cells were treated with vehicle or BTZ (1 μM, 6 h). (D) IHC staining of UB-K48 in PDOX tissue sections from BTZ-treated mice (1 mg/kg, every other day for 6 d). (E) BTZ treatment increased the H-score of UB-K48 staining in PDOX tissue sections. IHC staining intensity was categorized as negative, 1+ (weak), 2+ (moderate), and 3+ (strong), with H-score calculated by combining intensity and the percentage of positive cells. Stronger staining corresponds to higher H-scores. H-scores were quantified using QuPath. Statistical data are presented as mean ± SD (n = 6), with significance determined by Student’s t test. (F) IHC staining of UB-K48 and YAP in DGC tissue microarray. (G) IHC results from (F) reveal a positive correlation between UB-K48 and YAP in DGC. Each dot represents the UB-K48 and YAP H-scores from an individual sample (n = 128). Pearson’s correlation rho and P values are shown. See also SI Appendix, Fig. S4.

The GS subtype of gastric cancer aligns closely with DGC as classified by the Lauren classification, characterized by high malignancy, poor prognosis, and limited treatment options (59, 60). K48-Linkage-Specific Ubiquitin (UB-K48) is a key modification involved in proteasomal protein degradation, and its accumulation serves as a hallmark of impaired proteasome activity (61). Western blot analysis revealed a robust increase in total UB-K48 levels in AGS cells following BTZ treatment, indicating that UB-K48 accumulation can serve as a marker of proteasome inhibition (Fig. 4C). In the DGC patient-derived organoid xenograft (PDOX) model, immunohistochemical (IHC) analysis revealed that BTZ-treated tumors exhibited higher H-scores for UB-K48 staining (Fig. 4 D and E). IHC staining of tissue microarrays from 128 DGC samples demonstrated a positive correlation between the H-scores of UB-K48 and YAP. As the accumulation of UB-K48 conjugates reflects impaired proteasome activity, these findings indicate a negative correlation between proteasome activity and YAP/TAZ activity in DGC samples (Fig. 4 F and G). These results demonstrate that the regulation of YAP/TAZ by proteasome is conserved in solid tumors and may hold particular clinical significance in the GS gastric cancer or DGC.

Cotargeting Proteasome and YAP/TAZ Suppresses Solid Tumor Growth.

YAP/TAZ activation in cancer is linked to drug resistance and relapse (11, 14, 18). The proteasome stress-induced activation of YAP/TAZ may contribute to resistance to proteasome inhibitors in solid tumors. Hence, combining proteasome inhibitors with YAP/TAZ inhibitors represents a therapeutic strategy for treating solid tumors. We first evaluated the effects of this combination therapy in DGC cells (SNU668) using various YAP/TAZ inhibitors. Cotreatment with Verteporfin (VP, a small molecule disrupting YAP–TEAD interaction) boosted the inhibition of BTZ on cancer cell colony formation (62) (Fig. 5A). Similar to VP, additional YAP/TAZ inhibitors VT107 (a pan-TEAD autoacylation inhibitor) and VT103 (a TEAD1 palmitoylation inhibitor) showed similar effects (63) (Fig. 5 B and C). Moreover, BTZ and the PP2A inhibitor LB100 (which indirectly causes YAP/TAZ inactivation) also revealed a synergistic effect (SI Appendix, Fig. S5A).

Fig. 5.

Cotargeting proteasome and YAP/TAZ suppresses DGC cell growth. (A–C) Coinhibition of proteasome and YAP–TEAD suppresses SNU668 growth, shown in representative images. (D) Schematic of the in vivo combination treatment of BTZ and VT107. (E) Body weight of tumor-bearing mice during the drug treatment process. (F) Images of SNU668 tumor xenografts 14 d posttreatment. (Scale bar, 1 cm.) (G) Growth kinetics of SNU668 tumor xenograft in BALB/c nude mice treated as indicated. (H) Tumor weight in (F) 14 d posttreatment. (I) Coinhibition of proteasome and YAP–TEAD reduces tumor proliferation and enhances apoptosis. IHC staining of Ki67 and Cleaved Caspase-3. (Scale bar, 20 μm.) (J and K) Quantification of stained tumor sections in panel (I). Statistical data are presented as mean ± SD (A–C, E, J, and K) or mean ± SEM (G and H), with statistical significance determined by one-way ANOVA (A–C, H, J, and K) or two-way ANOVA (G). Significance levels are denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, with “ns” indicating not significant. See also SI Appendix, Fig. S5.

We also investigated the combination therapy in additional cancer cell lines. In the AGS cell lines, cotreatment with VP significantly induced cell death, as evidenced by a decrease in the IC50 of BTZ in CCK-8 assays (SI Appendix, Fig. S5B). Furthermore, the combination of VP and BTZ also effectively inhibited the colony formation of various cancer cells, including AGS, HCT116, and MKN28 (SI Appendix, Fig. S5 C–E). Hence, combining a proteasome inhibitor with a YAP/TAZ inhibitor can synergistically induce cytotoxicity in cancer cells derived from solid tumors, including DGC.

Due to its unique histological features and molecular profile, DGC typically responds poorly to conventional chemotherapy and targeted therapies, resulting in limited treatment options and a dismal prognosis for patients (59, 60). Given the lack of effective therapeutic strategies, there is an urgent clinical need to explore novel treatment approaches that can improve the survival outcomes of patients with this challenging malignancy. In this context, innovative therapeutic strategies targeting specific molecular pathways hold significant potential for enhancing treatment efficacy. We then specifically chose DGC as the model to evaluate the effect of concurrent inhibition of proteasome and YAP/TAZ in the following work.

We further tested the combination therapy in mice using the SNU668 xenograft tumor model. Once the subcutaneous tumors reached 40 to 120 mm³, mice were randomly assigned to different treatment groups. BTZ was administered intraperitoneally every other day, while VT107 was given daily via oral gavage. Mice were killed for analysis 14 d posttreatment (Fig. 5D). Notably, the combination of BTZ and VT107 significantly reduced tumor volume and weight, without affecting body weight, indicating this treatment was both safe and effective (Fig. 5 E–H). IHC staining and quantitative analysis revealed that the combination therapy significantly decreased the number of Ki67-positive cells and increased Cleaved-Caspase3-positive cells (Fig. 5 I–K). These findings indicate that this combination therapy suppresses tumor cell proliferation and promotes apoptosis.

Combination Therapy for DGC in PDO and PDOX Models.

PDOs have emerged as valuable preclinical models. We established two PDOs (PDO#1 and PDO#3) from residual tumor tissues of two gastric cancer patients who underwent preoperative chemotherapy (clinical characteristics summarized in SI Appendix, Table S2). These PDOs may exhibit chemotherapy-resistant properties inherent to their primary tumors. PDO#1, derived from a mixed-type gastric carcinoma, displayed a glandular morphology. PDO#3, originating from a DGC patient, exhibited a classic solid morphology characteristic of DGC (64, 65) (Fig. 6A). In these organoids, BTZ treatment resulted in a progressive reduction in YAP phosphorylation, with an optimal effect observed after 6 h and more prolonged treatment (Fig. 6B). BTZ activated YAP/TAZ in both PDOs, as indicated by decreased YAP phosphorylation and upregulation of YAP/TAZ target gene expression (Fig. 6 C–E). Hence, in a three-dimensional (3D) PDO system, inhibition of the proteasome leads to YAP/TAZ activation, consistent with results obtained from two-dimensional (2D) cultured cancer cells.

Fig. 6.

A combination therapy for DGC in the PDO model. (A) Representative images showing the typical morphology of PDO#1 and PDO#3. (Scale bar, 400 μm.) (B) Time-course BTZ treatment induces YAP activation in the PDO model. PDO#3 was treated with 1 μM BTZ for 2 to 10 h followed by immunoblotting. (C) Proteasome inhibition induces YAP dephosphorylation in two distinct PDOs. PDOs were treated with vehicle or BTZ (1 μM, 6 h). (D and E) Proteasome inhibition upregulates YAP/TAZ target genes in two PDOs. PDOs were treated with BTZ (1 μM, 6 h), followed by RT-PCR. (F) Schematic representation of the 3D PDO drug assay utilizing BTZ and VT107. (G and H) Coinhibition of proteasome and YAP–TEAD suppresses the growth of diffuse-type gastric PDOs. (Scale bar, 400 μm.) Statistical data are presented as mean ± SD (n = 3), with statistical significance determined by Student’s t test (C–E) or one-way ANOVA (B). Significance levels are denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, with “ns” indicating not significant. See also SI Appendix, Figs. S6 and S7.

We then tested the effect of the combination therapy (BTZ and VT107) on these two PDOs. PDOs were subcultured, and after 2 d, mini organoids (2,000 µm² for PDO#1, 300 µm² for PDO#3) were randomly assigned into four groups. These organoids were treated with Ctrl, BTZ, VT107, or BTZ+VT107 over 8 d (refresh drug-containing medium every 2 d) with microscopic imaging conducted on day 0, day 4, and day 8 (Fig. 6F). On both day 4 and day 8, the combination of BTZ and VT107 significantly reduced the size and total area of organoids compared to treatment with either BTZ or VT107 alone (Fig. 6 G and H and SI Appendix, Figs. S6 and S7). At the endpoint, organoids exceeding 8,000 μm² (PDO#1) or 6,000 μm² (PDO#3) were hardly spotted in the presence of both BTZ and VT107 (Fig. 6 G and H and SI Appendix, Fig. S6 D and H). Additionally, a marked increase in cell debris was observed following combined BTZ and VT107 treatment, indicating that cotreatment significantly enhanced apoptosis in PDO cells (Fig. 6 G and H). These findings demonstrate that coinhibition of the proteasome and YAP–TEAD effectively impairs organoid growth and promotes apoptosis in DGC PDOs.

PDOXs are robust in vivo preclinical cancer models, effectively preserving the histopathological characteristics and drug sensitivities of the original tumor (65, 66). PDO#3 was injected subcutaneously into nude mice to establish a PDOX model of DGC. VT104 is an enantiomer analog of VT107 with improved oral pharmacokinetics and enhanced therapeutic efficacy (63). Once tumors reached a predetermined size, mice were randomly assigned to receive Ctrl, BTZ, VT104, or BTZ+VT104 treatments (Fig. 7A). In this DGC model, mice treated with the combination of BTZ and VT104 exhibited reduced tumor volume and weight compared to those treated with BTZ or VT104 alone, with no significant changes in body weight (Fig. 7 B–E). Interestingly, analysis of YAP IHC in PDOX tissue sections revealed stronger nuclear localization and higher staining intensity of YAP in BTZ-treated samples (Fig. 7 F and G). Furthermore, IHC staining of PDOX sections indicated that the combination therapy decreased the number of Ki67-positive cells and increased Cleaved Caspase-3-positive cells (Fig. 7 H–J). These findings suggest that the concurrent inhibition of proteasome and YAP/TAZ effectively suppresses tumor growth in DGC PDOX models by reducing cell proliferation and promoting apoptosis.

Fig. 7.

A combination therapy for DGC in the PDOX model. (A) Schematic of the in vivo combination therapy for the DGC PDOX model. (B) Body weight of tumor-bearing mice during the BTZ and VT104 treatment process. (C) Images of PDOX tumors 6 d after treatment. (Scale bar, 1 cm.) (D) Tumor weight in (C) 6 d posttreatment. (E) PDOX growth kinetics in BALB/c nude mice treated as indicated. (F) IHC staining of YAP in PDOX tissue sections. (Scale bar, 20 μm.) (G) Quantification of stained tumor sections in panel (F). (H) The BTZ and VT104 combination significantly reduces tumor proliferation and enhances apoptosis. Representative IHC images of Ki67 and Cleaved Caspase-3 are shown. (Scale bar, 20 μm.) (I and J) Quantification of stained tumor sections in panel (H). Statistical data are presented as mean ± SD (n = 5), with statistical significance determined by Student’s t test (G), one-way ANOVA (D, I, and J), or two-way ANOVA (E). Significance levels are denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, with “ns” indicating not significant.

Discussion

This study has identified a molecular mechanism by which proteasome inhibitors activate YAP/TAZ in solid tumor cells, contributing to resistance against proteasome-targeted therapies. These findings indicate that proteasome stress serves as an upstream signal for the Hippo pathway, and the dysregulation of the Hippo pathway may contribute to drug resistance and disease progression. Based on this mechanistic insight, we have developed a potential combination therapy for solid tumors, including aggressive DGC, by concurrently inhibiting proteasome and YAP/TAZ.

The identification of upstream signals is crucial to understanding how the Hippo pathway is regulated in various contexts, particularly in cancer (11, 67). It has been shown previously that cellular stress cues, such as hypoxia, heat shock, energy deprivation, and endoplasmic reticulum (ER) stress, lead to dysregulated YAP/TAZ activity and tissue homeostasis (68–71). Here, we report that the Hippo pathway is sensitive to proteasome stress. The mechanistic basis for YAP/TAZ activation in response to proteasome inhibition involves impaired enzymatic activity and disruption of protein–protein interactions within the Hippo signaling network. First, RAP2 undergoes ubiquitination and enzymatic inactivation upon proteasome inhibition, resulting in decreased kinase activity of MAP4Ks and LATS1/2, as well as reduced phosphorylation of YAP/TAZ (SI Appendix, Fig. S8). Second, proteasome inhibition disrupts the interactions among several Hippo pathway components, including MAP4Ks, NF2, and LATS1/2 (SI Appendix, Fig. S8). It remains unclear whether these two mechanisms are related, and additional mechanisms may also participate in regulating the Hippo pathway by proteasome stress.

Previous studies in hepatocellular carcinoma (HCC) and gastrointestinal stromal tumors (GIST) cell lines reported that BTZ inhibits YAP activity, which appears to contradict our findings. We propose that this discrepancy may reflect differences in tumor type and treatment duration: While the HCC study applied BTZ for 48 h and the GIST study for 72 h, our experiment focused on an earlier response (less than 6 h) (72, 73). Long-term treatment affects the fitness and attachment of cancer cells, which may contribute to the inhibition of YAP/TAZ. These observations underscore the importance of tumor context and treatment durations, and highlight the need for further investigation across different tumor models and time points.

The activation of YAP/TAZ by proteasome stress may contribute to the progression of various diseases. For example, YAP/TAZ activation is frequently observed in solid tumor specimens (11, 12). However, genetic alterations of Hippo pathway genes are relatively rare, and the mechanism behind YAP/TAZ activation in tumors remains unclear (2, 11, 15). Our results suggest that proteasome stress may promote YAP/TAZ activation in tumor cells, which are often overloaded with misfolded or damaged proteins under conditions of hypoxia, oxidative stress, and high protein synthesis rates (27, 28, 69, 74, 75). Notably, reduced proteasome activity could be observed in a subset of solid tumor cells with cancer stem cell–like traits (76, 77). Meanwhile, YAP/TAZ were known to induce the expression of stemness-associated genes (11, 78, 79). Hence, YAP/TAZ activation by proteasome stress may contribute to the maintenance of cancer stem cells. Moreover, proteasome stress is widely present in degenerative diseases, and YAP/TAZ plays a critical role in tissue regeneration (3–5, 26, 80). It would be interesting to explore whether YAP/TAZ activation plays a role in the pathogenesis of different degenerative diseases.

This study has also investigated the potential of repurposing proteasome inhibitors for treating solid tumors. While proteasome inhibitors, such as BTZ, have successfully treated hematologic malignancies, their efficacy in solid tumors has been limited due to an unknown mechanism. Our findings reveal that the activation of YAP/TAZ upon proteasome inhibition is a key mediator of drug resistance in solid tumors, and concurrent inhibition of proteasome and YAP/TAZ effectively prevents the progression of tumors, including DGC, in vitro and in vivo (SI Appendix, Fig. S8). Given the limited treatment options and poor prognosis associated with DGC, this combination therapy could provide a broad avenue for clinical intervention. Moreover, this combination therapy may offer treatment options for a broader range of refractory solid tumors. Similarly, a study has shown that tazemetostat, an FDA-approved EZH2 inhibitor used for the treatment of the hematologic malignancy follicular lymphoma, can also be effective in solid tumor non–small cell lung cancer when combined with YAP/TAZ inhibition (81). This suggests that cotargeting YAP/TAZ may provide a strategy to repurpose drugs originally approved for hematologic malignancies for the treatment of solid tumors.

Mutant p53, one of the most frequent alterations in human cancers, has been shown to acquire oncogenic gain-of-function properties through the activation of the proteasome machinery, thereby disrupting protein homeostasis and promoting resistance to proteasome inhibitors (82). In parallel, YAP has been reported to physically interact with mutant p53, co-occupying the promoters of critical cell cycle regulators such as cyclin A/B, thereby amplifying their proproliferative transcriptional programs (83). These findings suggest that tumors harboring mutant p53 may be particularly sensitive to the combination of proteasome and YAP/TAZ inhibitors.

Despite the promising results, several limitations of this study must be acknowledged. First, while our experiments establish a link between proteasome inhibition and YAP/TAZ activation, further studies are required to uncover the detailed molecular mechanism. Additionally, our research primarily focused on DGC, thus, the generalizability of our findings to other solid tumor types should be validated in future studies. Finally, although we demonstrated the efficacy of combining proteasome and YAP/TAZ inhibitors in preclinical models, the safety and feasibility of such a strategy in patients remain to be determined. Clinical trials will be necessary to evaluate the potential benefits and risks of this combination therapy.

Materials and Methods

All animal experiments were approved by the Animal Ethics Committee of Shanghai Medical College, Fudan University. Human gastric cancer PDOs were obtained from Zhongshan Hospital with informed consent and ethics approval (B2021-449, Y2024-401). All samples were anonymized to ensure confidentiality.

Detailed protocols for mouse models, cell lines and cell culture, human gastric cancer PDOs culture, 3D PDO drug assay, plasmids transfection, lentivirus production, RNA interference, proteasome activity assay, immunoprecipitation, in vitro kinase assay, in vitro phosphatase assay, recombinant protein production, GST-RalGDS-RBD pull-down assay, ubiquitination assay, immunoblotting, immunohistochemistry, immunofluorescence, RNA extraction, reverse transcription, real-time PCR, plate clone formation assay, and statistics can be found in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Previously published data were used for this work. The YAP1/TAZ expression data used in Fig. 1A were obtained from the Cancer Cell Line Encyclopedia project (https://data.broadinstitute.org/ccle/CCLE_DepMap_18Q2_RNAseq_RPKM_20180502.gct) (84). The transcriptome data used in Fig. 4 A and B and SI Appendix, Fig. S4 A–E were obtained from TCGA (https://toil-xena-hub.s3.us-east-1.amazonaws.com/download/tcga_RSEM_gene_tpm.gz) (85). All other data are included in the manuscript and/or SI Appendix.