WZ35 Induces Metabolic Reprogramming of Hepatoma Cells Through ROS-Mediated Inhibition of YAP-GLUT1 Axis

Background: The signicant changes in a series of biochemical activities during the process of tumorigenesis and development, including those in glucose, lipid, and amino acid metabolism can contribute to the ability of the cancer cells to proliferate indenitely. WZ35 is a new small molecule YAP inhibitor, which can mediate YAP activity to inhibit the growth of gastric cancer, breast cancer and hepatocellular carcinoma. In this article, we explored how WZ35 can modulate YAP activity to inuence the metabolism of hepatocellular carcinoma (HCC) cells to inhibit their proliferation activity. Methods: The Gene Expression Omnibus (GEO) data, the Cancer Genome Atlas (TCGA) data, immunohistochemistry (IHC) and preclinical mouse model was utilized to detect the differential expression and vital role of YAP in HCC cells. A series of in vitro and in vivo experiments were performed to explore the antitumor activity of WZ35 and how it can target YAP activity to inhibit the tumor progression. UCSC combined JASPAR public databases were used to predicted possible binding sites of TEAD on GLUT1 promoter. Additionally, seahorse energy metabolism experiments and metabolomics analysis were utilized to explore the possible metabolic changes induced by WZ35. The clinical use of YAP and GLUT1 was veried by bioformatics and tissue microarrays of immunocytochemistry. Results: WZ35 can signicantly attenuate the proliferation and growth of HCC. In terms of mechanism(s), it was demonstrated that the suppressive effects of WZ35 on YAP were achieved by promoting ROS production and the drug can exert its signicant YAP inhibitory capacity to decrease the level of GLUT1, a glucose transporter located on the surface of cytomembrane, thereby resulting in a signicant decrease in the ability of cells to take up glucose and thus perturb the metabolism. Moreover, through bioinformatics mining, it was proposed that YAP/GLUT1 has the potential to serve as promising target for HCC therapy. Conclusions: Our ndings indicate WZ35 affecting μg/mL) or curcumin (10 μg/mL) by means of DAPI staining. Results are presented as the mean ± standard error from independent experiments performed in triplicate. *P < 0.05, **P < 0.01, student’s t test. d The levels of various apoptosis associated proteins, including Bax, Bcl-2, and Cleaved caspase-3 were detected by western blotting following the 18h treatment of WZ35, with GAPDH serving as an internal control. e Quantitation of the murine body weight and tumor volumes over time in four indicated tumor xenograft groups, as measured by utilizing electronic balance and caliper. The results are presented as the mean ± standard error from independent experiments in triplicate. *P < 0.05, **P < 0.01, ***P < 0.001, student’s t test. f Subcutaneous tumor xenograft models of WZ35-treated HCCLM3 were used. Mice were put to death by cervical dislocation and the tumors were harvested after 17 days. Optical image of each tumor and the weight curve was photographed and quantied, respectively. (n = 6 mice per group). g The representative images of Hematoxylin and eosin staining of indicated treated tumors have been shown. Scale bars, 200 μm for images. of culture was detected by hexokinase method. measured in combination with CCK-8 assay in either non glucose-treated or low glucose-treated or high glucose-treated or high glucose plus 2-DG-treated HCCLM3 cells with or without treatment of WZ35. j The contents of metabolic products in HCCLM3 cells treated with or without WZ35 (10 μg/mL). All these results have been presented as the mean ± standard error from independent experiments in triplicate. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, student’s t test. k Proposed working model showed distinct metabolic changes after drug action.

become the core determinant of human cancer malignancies [23]. It has been detected in late-stage ovarian, colon, gastric, liver, esophageal, non-small-cell lung cancers and lobular type of invasive breast cancers to foster their proliferation, progression and migration. In recent years, studies have found that YAP is an emerging node that coordinates cell nutrition and tissue homeostasis, and participates in the regulation of cell metabolism such as glycolysis, lipogenesis, and glutaminolysis [20]. Since metabolism is also closely related to cell proliferation, what is the relationship between YAP's ability to regulate cell proliferation and metabolic changes?
WZ35 is a small molecule YAP inhibitor synthesized by our team [24]. Our team has con rmed that the curcumin derivative WZ35 can signi cantly affect the growth of gastric cancer [25], hepatocellular carcinoma [26] and breast cancer [27] by targeting YAP. In this article, we further found that WZ35 could negatively regulate the metabolism of HCC cells mediated through GLUT1. However, the relationship between YAP and GLUT1 remains unclear. In an elegant study, Andrew G Cox et al used zebra sh embryos as a model to prove that YAP regulates the expression of GLUT1 and thus induces the regeneration of organs [28]. But the effect on the metabolism and regulation mechanism remains to be further explored.
In this report, we found a novel YAP inhibitor WZ35 that could inhibit the intake of glucose through ROS-YAP-GLUT1 axis. We identi ed new function of YAP and GLUT1 that were important for the quality of prognosis of HCC patients and might provide new insight into the potential mechanism of the pathogenesis of hepatocellular carcinoma. Further experiments demonstrated glucose intake inhibition inducing metabolic reprogramming and thereby suppressing the proliferation of HCC cells.

Cell culture
The human HCC cell lines including HCCLM3, HepG2 and Huh7 were obtained from the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. The cells were grown using Dulbecco's Modi ed Eagle's Medium (DMEM) containing 10% FBS (Life Technologies) and penicillin/streptomycin at 37 °C in a humidi ed 5% CO 2 incubator.

Patient samples
In total, 110 matched HCC tumor samples and precancerous control tissues were isolated from untreated patients undergoing surgical tumor resection at the Department of Hepatopancreatobiliary Surgery of the First A liated Hospital of Wenzhou Medical University, Wenzhou, China, between 2010 and 2014. The tumor samples were separated into two different portions, with one portion being used for routine histopathological examinations, and the rest being snap-frozen and stored at -80 °C for future studies. HCC was pathologically con rmed in all patients, with tumors staging being determined based upon the AJCC/UICC (Union for International Cancer Control/American Joint Committee on Cancer) and WHO classi cation. The Board and Ethical Committee of the First A liated Hospital of Wenzhou Medical University approved this study ID number wydw2019-0446 , and all the patients provided written informed consent.

Bioinformatics analysis
The GENE EXPRESSION OMNIBUS (GEO) database was searched using the terms "gene expression", "hepatocellular carcinoma", and "Homo sapiens". We selected the GSE14520 dataset as it met the quality control and the sample size criteria. The R software (version.4.0.3) affyPLM package was used for the regression analyses and normalization. Based on quality ltering, two healthy control microarray samples were excluded, while the remaining 41 microarrays (22 HCC samples, 19 adjacent control samples) based on the GPL571 platform (Affymetrix Human Genome U133A 2.0 Array) were retained for analysis. The RStudio Limma package was used to identify the differentially expressed genes (P < 0.05 and | logFC (Fold Change) | ≥ 1), after which probe names were converted to gene names. The heatmap.2 package was used to visualize differences in gene expression between the samples, with appropriate functions also being used to draw the boxplots, survival curves, and volcano plots in R software (version.4.0.3).We used database for annotation, visualization and integrated discovery for GO enrichment analysis.We also performed pathway enrichment analysis using pathway data obtained from the FTP service of KEGG (Kyoto Encyclopedia of Genes and Genomes). JASPAR [30] and USCS [31] were used to predicted binding sites, conservation and histone modi cation.
Cell proliferation assay A CCK-8 kit was used to measure the cellular proliferation. Brie y, HCC cells were plated in 96-well plates (8x10 3 /well) overnight, after which they were untreated or treated with curcumin (10 μg/mL), WZ35 (10 μg/mL), NAC (5 mM). For some experiments, a range of drug concentrations (0, 5, 10, 20, and 40 μg/mL) were used instead. WZ35 and curcumin were prepared using 0.03% DMSO; NAC was prepared in PBS and diluted using complete media to appropriate concentrations for use. After drug treatment, the CCK-8 kit was used to measure the viability with absorbance at 450 nm being quanti ed via microplate reader (Spectramax m5, Molecular Devices).

Colony formation assay
Initially, HCCLM3 cells were plated in 6-well plates (5,000/well) with curcumin (1 μg/mL), WZ35 (1 μg/mL), and/or NAC (5 mM). The plate incubated until individual colonies were formed containing approximately 50 cells . The cells were then washed in PBS, xed for 20 minutes with methanol, stained  for 15 minutes with crystal violet, washed with distilled water, and colonies were counted using the  ImageJ software. DAPI staining for cells Apoptosis was measured by plating HCCLM3 cells in 6-well plates overnight (5×10 5 /well). The cells were then treated using WZ35 or curcumin (both 10 μg/mL) for 18 h, after which they were washed using PBS, xed for 15 minutes in 4% paraformaldehyde (PFA), and stained for 15 minutes with 4',6-diamidino-2phenylindole (DAPI) (400 μL). The cells were thereafter imaged via microscope (NIKON, Japan) with 100 magni cation.

Western blotting
Radio immunoprecipitation assay (RIPA) lysis buffer obtained from Beyotime (Haimen, China) was used to isolate the proteins from the cells or to homogenize the tumor tissues. The samples were then spun for 15 minutes at 12,000 rpm, and equal amounts of supernatant protein (10 μg/sample) were separated via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (8, 10 or 12%) prior to transfer onto a poly (1,1-di uoroethylene) (PVDF) membrane (Millipore, Burlington, MA, USA). The blots were blocked for 90 minutes using 5% skim milk in TRIS hydrochloride & Tween (TBST), followed by an overnight incubation with the various primary antibodies at 4 °C. After washing three times in TBST (10 minutes/wash), the blots were incubated for 1 h with horseradish peroxidase (HRP)-linked secondary antibodies (1:2000), after which they were washed again and protein bands were detected using an electrochemiluminescence (ECL) reagent (Thermo Scienti c, IL, USA).

Transmission Electron Microscopy (TEM)
Curcumin-and WZ35-induced changes in the cellular structure were visualized via TEM, with DMSO being used as a control. Initially, the treated cells were pre-xed overnight in 2% glutaraldehyde, followed by a 2 h post-x in 1% osmium tetroxide. Ethanol and acetone gradients were then used to dehydrate the cells, after which the cell pellets were embedded in ethanol and acetone. Ultrathin pellet sections were then stained using 2% aqueous uranyl acetate and lead citrate, after which a TEM instrument (Wenzhou Medical University, H-7500, HITACHI, Japan) was used to image the cells with 26500 magni cation.
Flow cytometry analysis for ROS determination ROS production was quanti ed by staining the cells for 30 minutes with 10 μM 2,7-dichloro uo-rescein diacetate (DCFH-DA) in DMEM at 37 °C. The cells were then treated for 2 h with corresponding compounds (10 μg/mL), washed twice with DMEM, and DCFH-DA uorescence was quanti ed by BD Accuri TM C6 ow cytometer (BD, Franklin Lakes, NJ).
After a 18 h drug treatment, the supernatant was collected and analyzed. According to Beckman Kurt biochemical analysis system, glucose determination kit (Beckman Coulter, CA, USA) was used to detect the glucose content in the culture medium supernatant. In the presence of adenosine triphosphate (ATP) and magnesium ions, the glucose was phosphorylated by hexokinase (HK) to produce glucose-6phosphate and adenosine diphosphate (ADP). Glucose-6-phosphate dehydrogenase (G6P-DH) speci cally can oxidize glucose-6-phosphate to 6-phosphogluconic acid. Oxidized nicotinamide adenine dinucleotide (NAD + ) was also reduced to nicotinamide adenine dinucleotide (NADH). The increase of absorbance at 340 nm was proportional to the glucose concentration in the sample.

Glucose metabolism test
A CCK-8 kit was used to measure the cellular proliferation. HCCLM3 cells were plated in 96-well plates (8x10 3 /well) overnight, medium was changed to sugar-free medium the next day, after which the cells were left untreated or treated with Glucose (1 mM), Glucose (25 mM), 2-DG (1 mM), WZ35 (10 μg/mL). After an 18h-drug treatment, the CCK-8 kit was used to measure the viability, with absorbance at 450 nm being quanti ed via microplate reader (Spectramax m5, Molecular Devices).

Metabolomics analysis
TCA: Targeted metabolite analysis was performed on a Xevo G2-XS QT of mass spectrometry coupled with an ACQUITY UPLC (Ultra Performance Liquid Chromatography) system and data analysis was done with Progenesis QI (all from Waters, Milford, MA, USA). The metabolites in a homogenate of 1×10 6 cells were extracted using chloroform, methanol, and water at a ratio of 8:4:3 and resuspended in 1 acetonitrile. For UPLC, the samples were injected on an HSS T3 column (100×2.1 mm, 1.8 μm) using an 8-min gradient containing ow phase A (0.1 formic acid-water) and ow phase B (0.1% formic acidacetonitrile) at a ow rate of 0.5 ml/min. For the negative ion mode, the capillary energy and sample cone were set as 2000 V and 20 V, respectively. Scan time was set at 0.1 s intervals for 60 s.

Tumor transplantation
Female BALB/c nude mice (6-8 weeks old; ~20 g) from Wenzhou Medical University Laboratory Animal Center were housed under SPF conditions. These animals were subcutaneously injected with a 100 μL volume containing 5×10 6 HCCLM3 cells. When tumors grew to 100 mm 3 , the mice were randomized into 4 groups (n=6/group) that were intraperitoneally administered saline (Mock) vehicle, curcumin (25 mg/kg), or WZ35 (25 mg/kg) daily for 17 consecutive days. Animal body weight and tumor volumes were monitored every other day, with the latter being calculated as follows: tumor volume = length × width × height/2. After 17 days, sodium pentobarbital was used to euthanize mice, and the tumor tissues were collected to make para n sections for HE staining. Our project was conducted under guidelines approved by the Experimental Animal Ethics Committee of Wenzhou Medical University ID number wydw2019-0446 .
Tissue microarrays (TMAs) and IHC staining HCC TMAs composed of 110 FFPE tissue sections (4 μm thick) were prepared the same way as previously described by Guttà C et al [32]. TMA IHC staining was conducted based on a standard approach. Brie y, the sections were depara nized using xylene, rehydrated with an ethanol gradient, treated to quench peroxidase activity, blocked with 5% normal goat serum, and incubated with anti-YAP/GLUT1 overnight at 4 °C. As a negative control, the antibody was omitted. The slides were then probed with biotin-labeled goat anti-rabbit IgG, treated with a streptavidin peroxidase solution (SABC kit, Boster, Wuhan, China), and then 3,3-diaminobenzidine (Boster, Wuhan, China) in PBS with 0.05% H 2 O 2 was used to treat the samples for 5 min at the room temperature for color development, after which the samples were counterstained using hematoxylin.

IHC staining assessment
Two pathologists blinded to the samples evaluated the clinicopathological information independently and scored IHC images of TMA samples. We utilized the Image-Pro Plus 6.0 (IPP) (Media Cybernetics, MD, USA) to calculate the integrated optical density (IOD) for the stained images, with this value corresponding to total staining per sample area [33]. Entire TMAs were scanned in order to quantify protein levels in the TMA levels. IHC staining was evaluated based on immunoreactive score (IRS) values as described in the prior studies [26], with these scores re ecting both staining intensity and the percentage of the positive cells. The percent positivity was stained as follows: 1 (≤ 10%), 2 (10-50%), 3 (50-80%), or 4 (≥ 80%). The staining intensity was scored as: 0 (negative), 1 (weak), 2 (moderate), or 3 (strong). These two scores were then multiplied together to yield the IRS value (0 -12), with these values being deemed either low (≤ 6) or high (>6).
Statistical analysis SPSS 19.0 (SPSS Inc., IL, USA) was utilized for statistical analysis. The data has been represented as the means ± SD from three independent experiments, and was compared via Student's t-tests. P < 0.05 was set as the signi cance threshold.

Results
YAP upregulation in HCC corresponds to poorer patient outcomes YAP, as an oncogenic gene, plays an important part in the occurrence and development of cancer [34][35].
To understand the possible role of YAP in HCC, we searched the Gene Expression Omnibus (GEO) datasets. YAP1 is regarded the most common subtypes of YAP, which plays a signi cant role and exhibited the highest content among YAPs. We began by analyzing YAP1 expression levels in the GSE14520 HCC gene expression dataset. The differential expression of YAP1 was visualized in the volcano map and Boxplots ( Fig. 1a,b), revealing the upregulation of YAP1 in HCC (log FC > 2, P << 0.001). Heat map similarly supported this HCC-speci c upregulation of YAP1 ( In order to con rm these bioinformatics ndings, we next extracted fresh protein from the tissue samples of 18 HCC patients and corresponding adjacent controls. The results of Western blotting con rmed that YAP expression was signi cantly higher in the tumor samples relative to control samples (Fig. 1c), which was consistent with our in silico ndings. The results of IHC staining of YAP in TMAs composed of 110 patient-derived HCC samples was analyzed to assess the relationship between YAP expression and the different clinicopathological ndings of HCC patients. Apparently, YAP expression was highly signi cant and a positive association with tumor histological TNM stage (P < 0.001) was noted (Fig. 1d,e). Moreover, the expression of YAP in well-differentiated (G1) HCC tissues was signi cantly lower as compared with the tissues that were moderately (G2 ) (P = 0.003) and poorly ( G3 ) (P < 0.001) differentiated, and the levels of YAP were signi cantly higher in G3 samples in contrast to G2 samples (P < 0.001) (Fig. S1c). These ndings indicated that an upregulation of YAP can coincide with the progression of HCC. We next examined the relationship between levels of YAP protein and overall survival (OS) of HCC patients via Kaplan-Meier analyses, and found that the high expression of YAP was closely related to poorer OS, while low related to better. (Fig. 1f) (P < 0.01). We then conducted univariate and multivariate analyses to identify the possible relationship between the expression of YAP and clinicopathological features of other HCC patients (Table1). In univariate analyses, the levels of YAP, tumor nodules, and venous invasion were all found to serve as signi cant prognostic factors, while in multivariate analyses, the expression of YAP was found to be independently associated with the prognoses of HCC patients. Together, these results indicated that YAP protein was overexpressed in HCC and may serve as an important target for therapeutic intervention.
WZ35 inhibited YAP from entering the nucleus and thereby affecting its function In order to explore the functional role of YAP in HCC, we rst knocked down or over-expressed this protein in HCCLM3 for various assays (Fig. 2a). YAP knockdown signi cantly reduced the colony formation rates as compared to the control cells, whereas the opposite facilitated this process (Fig. 2b,c). Moreover, consistent results were observed in a CCK-8 assay as well (Fig. 2d). These ndings clearly suggested that the proliferation of HCC cells was regulated by YAP.
After shuttling from cytoplam into the nucleus, YAP can regulate gene transcription and ultimately promote proliferation, apoptosis and morphogenesis [36]. We have previously shown that the curcuminderived WZ35 compound (Fig. 2e) can cause a signi cant downregulation in the protein levels of YAP. To explore the localization of YAP's in HCC cells, we performed immuno uorescence assay in WZ35-treated HCCLM3 cells via confocal microscopy. It was observed that the green signals were signi cantly diminished following WZ35 treatment, particularly in the nuclei of HCC cells, with DAPI (blue) used for nuclear visualization (Fig. 2f), thereby indicating that WZ35 was able to retard the nuclear translocation of YAP and its subsequent binding with TEA domain (TEAD). Consistent with this nding, WZ35 signi cantly increased the phosphorylation of YAP at serine 127 residue (Fig. 2g). It has been shown that YAP can regulate the transcription primarily via TEAD family transcription factors, thereby affecting cellular proliferation. In a non-phosphorylated state, YAP undergoes nuclear translocation and functions as a potential TEAD transcriptional activator, whereas S127 phosphorylation of YAP results in its interaction with 14-3-3 protein and consequent cytoplasmic sequestration [37]. Overall, not only can WZ35 down-regulate YAP signaling, but can effectively hinder its translocation and ability to promote the viability of HCC cells.

WZ35 showed superior anti-cancer activity under in vitro and in vivo settings
We next attempted to elucidate the relative inhibitory activities of Curcumin and WZ35 in vitro. In a CCK-8 assay, both the compounds caused a dose-dependent inhibition of the tumor cell proliferation, with WZ35 being clearly superior to curcumin (Fig. 3a). Moreover, similar outcomes were observed in the colony formation assays (Fig. 3b). Next, to identify whether inhibition of viability was also accompanied with apoptosis or cell cycle arrest, ow cytometry analysis was performed. Representative cell images indicated that WZ35 substantially induced G2/M cell cycle arrest (Fig. S2a), thus suppress the growth of HCCLM3 cells, and signi cantly promoted apoptosis (Fig. S2b). To elucidate whether the apoptosis of HCCLM3 cells was correlated with changes in the morphology of the nuclei, we utilized DAPI staining to visualize the changes, and the results revealed that WZ35 treatment induced nuclear shriveling and abnormal nuclear morphology at higher rates as compared to curcumin (Fig. 3c). The levels of various apoptosis-related protein levels were also assessed via Western blotting, with treated cells exhibiting an upregulated levels of pro-apoptotic Bax and cleaved caspase-3 as well as a decrease in expression of anti-apoptotic Bcl-2 (Fig. 3d). These ndings clearly suggested that WZ35 was superior to curcumin in inhibiting HCC cells growth in vitro.
In order to extend our range of ndings to in vivo settings, nude mice bearing subcutaneous HCCLM3 tumor xenografts were used. We found that both curcumin and WZ35 treatment did not signi cantly alter murine body weight, thereby suggesting no treatment-associated toxicity during the study period (Fig. 3e). While both the compounds were able to inhibit HCC tumor growth, WZ35 was found to be superior in suppressing this growth over a 17-day study period (Fig. 3e,f). According to the result of tumor histological assessment, it was apparent that WZ35-treated tumors revealed lower cellularity and higher tumor necrosis rates as compared with the vehicle and curcumin treatment (Fig. 3g). In conclusion, these results suggest that WZ35 may exhibit preferable in vivo e cacy by markedly suppressing HCC tumor growth as compared to curcumin.

WZ35 inhibits HCC growth in a ROS-dependent manner
To further ascertain the cellular mechanism related to drug-associated apoptosis, we examined the impact of WZ35 on intracellular morphology in HCCLM3 cells via TEM. As shown, apparent mitochondrial dysregulation, scattering, and swelling were observed in the treated cells (Fig. 4a), clearly suggesting that this compound had substantially disrupted mitochondrial homeostasis. Mitochondria are the major intracellular sources of ROS generation owing to electron leakage from the electron transport chain (ETC). We therefore analyzed ROS levels via ow cytometry in WZ35-treated HCCLM3 cells, revealing that this treatment led to a more than 2-fold increase in intracellular ROS levels (Fig. 4b). At normal levels, ROS play an important role in regulating cellular signaling and proliferation, whereas their excess levels can cause deleterious oxidative damage [38]. Hence, an increased ROS generation observed in WZ35-treated cells was likely to contribute to observed reduction in the cellular proliferation.
Thereafter, we utilized N-acetyl-l-cysteine (NAC), an inhibitor of ROS, to pretreat HCCLM3 cells for 2 h before WZ35 treatment, after which the viability of these co-treated cells was assessed. NAC pretreatment signi cantly neutralized WZ35-induced HCCLM3 cells proliferation inhibition (Fig. 4c), and correspondingly facilitated the colony forming abilities of WZ35-treated cells (Fig. 4d) while markedly attenuating their apoptotic death (Fig. 4e). At the molecular level, the impact of NAC pretreatment on WZ35-induced changes in the levels of YAP in HCCLM3 cells was analyzed via Western blotting, and the ndings revealed that this antioxidant can signi cantly counteract changes in the expression of YAP caused by WZ35 (Fig. 4f).
All these results suggested that ROS generation was essential in order for WZ35 to suppress growth and proliferation, with the expression of YAP being closely linked to increased ROS production in treated HCC cells.
WZ35 antitumor activity depends upon inhibition of the YAP protein and allows it to regulate the expression of GLUT1 As we elucidated the direct link between ROS and YAP above, we next proceeded to identify whether YAP was the target affected by WZ35 in drug-treated HCCLM3 cells and its anti-cancer activity. Firstly, YAP was knocked down and over-expressed in HCCLM3 cells. The results of various assays indicated that WZ35 and downregulation of YAP caused a concurrent inhibitory effect on cell viability (Fig. 5c) and clonogenicity (Fig. 5b, Fig. S3b), therefore facilitating their death (Fig. 5a, Fig. S3a), whereas WZ35 with co-treatment of upregulation of YAP counteracted the impact of WZ35 on HCCLM3 cells Moreover, an upregulation or downregulation of YAP only was positively or negatively correlated with proliferation and colony forming abilities of HCCLM3 cells.
To further explore the role of YAP in anti-cancer activity, data was downloaded from TCGA database and divided into two groups according to the level of YAP1 expression. We identi ed differential genes between the two groups by limma R package. Enrichment analysis of these genes based on Gene Ontology (GO) annotation system was conducted. The GO annotation system uses a controlled and hierarchical vocabulary to assign function to genes or gene products in any organism. As displayed in Fig. S3c: the genes corelated with YAP1 expression had enriched GO terms related to energy metabolism especially ATP function. According to pathway information (Fig. S3d) which is pivotal in de niting YAP function, the corelated genes were mainly enriched in many known oncogenic pathway for HCC. Among them, we noticed that the expression of YAP1 was possitively associated with central carbon metabolism. In the traditional sense, central carbon metabolism, including embden meyerhof pathway (EMP), pentose phosphate pathway (PPP) and tricarboxylic acid cycle (TCA) pathway, is the main source of energy required by organisms. As shown in Fig. 5d,e and Fig. S3j, high expression of YAP1 was signi cantly associated with central carbon metabolism genes such as SLC2A1, PFKP, G6PD, etc. And SLC2A1 had the most signi cant correlation. SLC2A1 is the o cial name of GLUT1 in NCBI. GLUT1 is the main glucose transporter in hepatoma carcinoma cells [9] and plays a vital role in glycolysis. Consistent with above, bioinformatics analysis of GEO database further proved that YAP expression may be closely related to cellular metabolism especially the glycolysis process in hepatoma carcinoma (Fig. S3e,f). GLUT1 was signi cantly up-regulated in tumor samples versus normal ones (Fig. S3g,h) and high levels of it was associated with a shorter survival in hepatocellular carcinoma patients (Fig. S3i). Western blotting assays unfolded that WZ35 could decrease the expression of YAP as well as GLUT1, and that knockdown of YAP by shRNA enhanced the suppressive effect of WZ35 on GLUT1 (Fig. 5f), whereas overexpression of YAP neutralized it (Fig. S3k). As per the JASPAR [30] database 5 potential TEAD binding sites were present on the GLUT1 promoter (Fig. 5g). UCSC Genome Browser revealed that GLUT1 promoter is highly conserved among mammals and an increased level of histone methylation correlated with H3K4Me3 at the promoter, which indicated active transcription (Fig. S4a).
WZ35 might mediate the inhibition of glucose intake to reduce proliferation GLUT1 is a transporter of glucose, which is very important for transportation of glucose in HCC cells [10], and glucose is a major nutrient source for the survival and proliferation of cancer cells because it feeds directly into glycolysis and then also into several other metabolic pathways [39][40]. As it has been demonstrated above that WZ35 could inhibit the expression of glucose transporter GLUT1, the metabolism changes in HCCLM3 cells were detected next. Statistically, after action of curcumin or WZ35, both two curves of the total acidi cation rate moved down, and the descending degree of the latter outweighed that of the former (Fig. 6a), so did the curves of OCR (Fig. 6d). The glycolytic capacity (Fig.  6b), maximal glycolytic capacity (Fig. 6c) and glycolytic reserve were signi cantly inhibited by curcumin and WZ35. Consistent with the aforementioned results, basal respiration (Fig. 6e), maximal respiration (Fig. 6f) and ATP production (Fig. 6g) were signi cantly restrained as well. We then detected the absorption of glucose was via glucose determination kit (hexokinase method). Corresponding to the above, WZ35 was found to signi cantly attenuate the ability to absorb glucose from outside the cells obviously (Fig. 6h). Additionally, by means of CCK-8 assay (Fig. 6i), the absorption of glucose was correlated to the proliferative activity of cells. The cells were cultured in different kinds of mediums with different concentrations of glucose with or without the retardant of glycolysis '2-DG'. According to the results, it was found that the hindrance of glucose absorption by WZ35 plays a vital role in reducing the proliferation activity of the HCC cells. When we arti cially retarded the glycolysis process, the ability of WZ35 to inhibit cell proliferation was signi cantly reduced, thereby suggesting that WZ35 can potentially serve as an effective blocker of glucose absorption, which could possibly exert its anti-proliferative effects by hindering the process of cell glycolysis. To explore the subsequent effects of drug-induced glucose intake on energy metabolism, metabolomics was used to detect the levels of the various intracellular metabolites. As displayed in g. 6j, treatment with WZ35 enhanced the ratio of NAD + to NADH, signifying the blockade of oxidative phosphorylation. The content of purine metabolites, such as adenosine monophosphate, adenosine triphosphate and guanosine diphosphate, etc. were signi cantly attenuated in WZ35-treated HCCLM3 cells. Since syntheses of nucleotides was limited to support the process of proliferation, the ability of WZ35 to inhibit cellular proliferation can be explained. In summary, WZ35 restricted the uptake of glucose in HCCLM3 cells, thus leading to inhibition of glycolysis and oxidative phosphorylation. Finally, the subsequent disorder in the purine metabolism could signi cantly affect the proliferation ability of HCCLM3 cells. The general metablosim changes is shown in g. 6k.
YAP combined GLUT1 may serve as a valuable prognostic and/or diagnostic biomarker in HCC We further explored clinical application of YAP and GLUT1 in above tissue chip in 53 HCC samples.
Consistent with YAP, the majority of these HCC patients were GLUT1-positive (Fig. 7a,b) (P = 0.0208). Boxplot showed that GLUT1 expression was positively correlated with TNM stage (Fig. 7c). An elevated GLUT1 expression was found to be signi cantly associated with tumor TNM stage, thereby suggesting that GLUT1 upregulation could coincide with HCC progression. Kaplan-Meier survival analysis showed that high level of GLUT1 was associated with a shorter survival in HCC (Fig. 7d) and the areas under curve (AUC) in ROC curve of GLUT1 were 0.7 (at 1 year), 0.77 (at 3 years) and 0.85 (at 5 years) (Fig. 7e). Since biological information mining has found that YAP and GLUT1 might display a good predictive function in predicting the prognosis of HCC patients (Fig. S5a,d), to further substantiate the survival signi cance of YAP and GLUT1, the correlation between YAP/GLUT1 high expression and clinical features of HCC patients was analyzed in above 53 clinical cases. The expression of YAP was found to be proportional to GLUT1 (Fig. S5e). High integrated expression levels of YAP and GLUT1 were negatively correlated with survival in HCC patients (Fig. 7f). A nomogram was made according to the result of multivariate logistic proportional hazards analysis which considers Stage I and II as the early stage, Stage III and IV as the late stage. Through xing the points associated with YAP and GLUT1, the risk of later stage for an individual patient could be calculated (Fig. 7g). The total score was 0-200, and YAP and GLUT1 were calculated and merged. A high score indicated a high risk of late stage HCC. A ROC calibration plot was made to assess the internal validity for the constructed nomogram (Fig. 7h).The AUC for combination of YAP and GLUT1 was 0.889, which was higher than GLUT1 alone (Fig. S5f). Considering these data, we can infer the potential link between YAP and GLUT1 and the integration between them may potentially serve as a valuable prognostic biomarker in HCC patients.

Discussion
HCC is the third deadliest and sixth most prevalent cancer type globally in 2020 [1]. However, the onset of HCC is insidious, and once discovered, it is mostly diagnosed in an advanced stage. Additionally, coupled with the lack of effective treatments, the mortality rate of hepatocellular carcinoma is relatively high.
Therefore, identi cation of e cient treatment methods and treatment targets for HCC has become an urgent clinical problem [41]. The genesis and development of tumor cells is a complex process, in which signi cant changes in a series of biochemical processes such as glucose, amino acid, nucleotide and lipid metabolism can impart unlimited proliferation potential on the tumor cells [42]. The discovery of the Warburg effect in the late 1920s not only emphasized the important role of carbohydrate metabolism in the cancer cells [43], but also opened a new era in the study of tumorigenesis and development of novel therapies from the perspective of metabolism. Therefore, in-depth exploration of various regulatory mechanisms from these metabolic perspectives can potentially serve as an important starting point for the development of effective treatment modalities for HCC.
WZ35 is a small molecule YAP inhibitor independently developed by our team. We have already demonstrated that WZ35 could target YAP to suppress the growth of a variety of tumors, including hepatocellular carcinoma, gastric cancer, and breast cancer [24][25][26][27]. Among them, we found by accident that WZ35 could inhibit glycolysis [25]. In this article, we further discovered that WZ35 can block glucose absorption, just as the glucose inhibitor 2-DG (Fig. 6i). According to the desperate demanding of cancer cells for glucose, some scholars suggested intermittent fasting might diminish the progression of tumours [44]. Clinical data showed that fasting can help prevent DNA and immune cells damage caused by chemotherapy [45], promote the regeneration of normal tissues. It is reported that combining fasting with other therapies can improve the therapeutic effect, prevent drug resistance and reduce side effects [46]. However, the lack of nutrition caused by fasting makes the overall anti-cancer effect unsatisfactory.
For example, the reduction of glutamine due to glucose de ciency can alleviate immune in ltration in the tumor microenvironment [47]. Because some trace elements and essential amino acids cannot be synthesized in the human body, fasting might induce more diseases such as anemia, neurocognitive de cits, ect [48]. But through CCK-8 assay, we demonstrated that the glucose blocking ability of WZ35 is also the main source for its anti-cancer ability (Fig. 6i). In other words, WZ35 has the same anti-cancer effect as intermittent fasting, but it cleverly avoids the serious side effects caused by nutritional de ciency. Combining the results of in vivo experiments and in vitro experiments, we conclude that WZ35 is a superior anti-cancer candidate drug. What's more, due to its glucose blocking ability, it is also expected to be used as a candidate adjuvant drug for diabetes. WZ35 may reduce the damage caused by the high concentration of glucose in ltration in the tissue uid through reducing the sensitivity of somatic cells to glucose, which is expected to realize that diabetic patients do not need to go on a painful diet, and this function remains to be explored. But how WZ35 blocks glucose absorption inhibition? How does the blockade of glucose absorption affect the metabolism of hepatoma carcinoma cells? These are the exact issues we will explain below.
Here, we identi ed a brand new signal axis, ROS-YAP-GLUT1 axis. WZ35 can restrain the glucose intake of hepatoma carcinoma cells through this axis and inhibit the proliferation of HCC. Through a series of experiments such as DAPI staining and Western blot analysis, we proved that WZ35 can promote ROS production to stimulate the translocation of YAP into the nucleus and reduce the total amount of YAP.
YAP is a pivotal protein involved in the Hippo pathway. Many studies have found that YAP can regulate various important physiological processes such as organ regeneration [49], atherosclerosis [50], ect. In our study, we identi ed a new function of YAP that can affect the expression of GLUT1 in HCC cells. GLUT1 (Glucose transporter 1) is the most highly conserved and widely found glucose transporter in different cancers [10]. Glucose is primarily transported into cells by GLUT1 [12]. This can explain why WZ35 can regulate glucose uptake. Then, we explored the possible mechanism of YAP regulating GLUT1. YAP, as a transcription regulator, can regulate TEAD-directed trascriptional function when it localizes in the nuclear [51]. We found 5 possible binding sites of TEAD in GLUT1 promoter region and four of them were found an increased level of histone modi cations associated with H3K4me3. H3K4me3 is possitively associated with gene expression and gathers around TSS [52]. The discovery of increased H3K4me3 further supports the reliability of these binding sites and we guessed the existence of H3K4me3 might also increase the possibility of combination between TEAD and GLUT1 promoter. So we concluded that the mechanism of YAP regulating GLUT1 may be that YAP enters into the nuclear and directs TEAD to bind with GLUT1 promoter and exercises activation function. As bioinformatics analysis and use of the clinical samples preved, the high expression of YAP and GLUT1 was closely related to the low survival of HCC patients. Multi-variate Cox proportional hazards analysis of clinical samples testi ed that the level of YAP and GLUT1 has important reference signi cance for predicting the stages of disease progression in HCC patients. We then put forward that YAP and GLUT1 have the potential to serve as novel biomarkers for the diagnosis and treatment of HCC and have an enormous potential in prognosis as well. The ndings above might provide possible targets for the development of cancer targeted drugs and novel ideas for the therapy and aid in realization of the goal of both early detection and treatment of HCC.
We next explored how the inhibition of glucose uptake leads to changes in the cellular metabolic network, and its association with cell proliferation. Increased glucose uptake is the main characteristic of glucose metabolism in the tumor cells [53]. Glycolysis and oxidative phosphorylation are both important parts of the glucose metabolism. Seahorse energy metabolism experiments showed the lack of glucose caused by WZ35 restrained glycolysis and oxidative phosphorylation (Fig. 6a,g). The cancer cells are primarily dependent on glycolysis for energy production. Many intermediates of glycolysis can be transferred to various intermediate pathways to produce different biosynthetic precursors. Among them, 5-phosphate ribose, the structural component of nucleotides, is partially oxidized from 6-phosphate glucose in the rst branch pathway, the pentose phosphate pathway (PPP) [54]. Additionally, a number of studies have found that the level of oxidative phosphorylation is closely related to the proliferation ability of the tumor cells. The intervention in the process of oxidative phosphorylation has a signi cant impact on the proliferation and metastasis of cancer both under in vivo and in vitro settings [55][56]. Purine metabolism is a type of nucleotide metabolism. Gaglio D et.al found that generation of the nucleotide bases can rescue the proliferation of glutamine-de cient cells [57][58]. Metabolomics analysis showed the purine metabolism of HCC cells was restrained by WZ35. Combining the above evidence, we speculated that the inhibition of glucose causes glycolysis and oxidative phosphorylation reduction and then leads to an impaired purine metabolism and ultimatly decreases the proliferation activity of HCC cells (Fig. 6k). This nding demonstrates targeting cancer cell metabolism, especially glycometabolism, might be a potential method for HCC treatment.

Conclusions
In conclusion, our work rst demonstrated that a potent small molecule YAP inhibitor WZ35 can restrain the proliferation of HCC through ROS-YAP-GLUT1 axis mediated glucose intake inhibition. YAP and GLUT1 has important reference signi cance for predicting the stages of disease progression in HCC patients and have the potential to serve as novel biomarkers for the diagnosis and treatment of HCC.
Targeting the metabolic process of hepatoma carcinoma cells may be a hopeful way in tumor treatment. Our ndings present novel insights into the pathology of HCC and potential targets for new therapeutic strategies. In the next step, we plan to further clarify the mechanism by which YAP regulates GLUT1, determine the speci c binding site of TEAD and GLUT1 promoter, and explore the role of epigenetic changes such as histone methylation modi cation in activating the transcription of GLUT1. Procedures involving animals and their care was performed in accordance with the Declaration of Helsinki and approved by Wenzhou Medical University Ethics Committee.

Consent for publication
Informed consent was obtained from all individual participants included in the study.

Availability of data and materials
The datasets analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Competing interests
The authors declare no competing interests. The graphic was created with BioRender.com.   presented as the mean ± standard error from independent experiments in triplicate. ***P < 0.001, student's t test. c,d,e HCCLM3 cells with or without NAC (5 mM) pretreatment were incubated with or without WZ35 (10 μg/mL) to assess the cellular proliferation and apoptosis via CCK-8 assays (c), colony formation assays (d) and DAPI staining assays (e) with representative images and/or quanti cations have been shown. All these results are presented as the mean ± standard error from independent experiments in triplicate. **P < 0.01, ***P < 0.001, student's t test. f Western blotting analysis of the YAP protein level of HCCLM3 cells treated with WZ35 or NAC.

Figure 5
WZ35 antitumor activity depends upon inhibition of the YAP protein and allows it to regulate the expression of GLUT1 a,b,c DAPI staining assays a colony formation assays (b) and CCK-8 assays (c) were conducted to measure apoptosis and cellular viability in control or WZ35-treated (10 μg/mL) HCCLM3 cells with or without the pretreatment of knockdown or overexpression via shRNA or pcDNA/peGFP vectors. d Genes in central carbon metabolism pathway of YAP high expression group on KEGG graph rendered by pathview. Gene expression levels are indicated as signi cantly higher (red), unchanged (gray), or lower (green). e Barchart visualized signi cantly related genes in central carbon metabolism with higher YAP1 expression including 10 genes with the highest logFC and 4 genes with the lowest logFC from datasets GSE97098, GSE77314 and GSE153783. f Western blotting analysis of the YAP and GLUT1 protein level of HCCLM3 cells treated with WZ35 and plasmid. g Predicted TEAD binding motif site sequence in the promoter region of GLUT1 host gene chromosome 1 from the database JASPAR. concentration of culture medium was detected by hexokinase method. i The absorption of glucose was measured in combination with CCK-8 assay in either non glucose-treated or low glucose-treated or high glucose-treated or high glucose plus 2-DG-treated HCCLM3 cells with or without treatment of WZ35. j The contents of metabolic products in HCCLM3 cells treated with or without WZ35 (10 μg/mL). All these results have been presented as the mean ± standard error from independent experiments in triplicate. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, student's t test. k Proposed working model showed distinct metabolic changes after drug action.

Figure 7
YAP combined GLUT1 may serve as a valuable prognostic and/or diagnostic biomarker in HCC a,b GLUT1 IHC staining in HCC patient samples. Images of cells were used to evaluate GLUT1 expression between the normal and the tumor. c GLUT1 IHC staining in HCC patient samples. Representative GLUT1stained images of TNM stage I, II, III and IV HCC tumor tissues were quanti ed. d Kaplan-Meier approach visualized the relationship between GLUT1 expression and OS in HCC patients (n = 53, P < 0.001) e Timedependent ROC curves analyzing the potential value of expression levels of GLUT1. f Kaplan-Meier curve for the disease free survival and overall survival have been illustrated for the subgroups YAP high/GLUT1 low, YAP high/GLUT1 high, YAP low/GLUT1 low and YAP low/GLUT1 high. g Nomogram for predicting the stage according to the expression levels of GLUT1 and YAP. Instruction: Each characteristic was located on the corresponding variable axis, and a vertical line was drawn upwards to the points axis to determine the speci c point value. This process was repeated and the total points value was tallied and located on the linear pretector axis. A vertical line was drawn down to the rank to obtain the probability of later stage h ROC calibration plots of the nomogram for combined YAP and GLUT1 The AUC was 0.889.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementryFigure.pdf