Perturbation Analysis of a Prognostic DDX3X-Mediated Gene Expression Signature Identifies the Antimetastatic Potential of Chaetocin in Hepatocellular Carcinoma

ATP-dependent RNA helicase DDX3X, also known as DEAD (Asp-Glu-Ala-Asp) Box Polypeptide 3, X-Linked (DDX3X), is critical for RNA metabolism, and emerging evidence implicates ATP-dependent RNA helicase DDX3X’s participation in various cellular processes to modulate cancer progression. In this study, the clinical significance of DDX3X was addressed, and DDX3X was identified as a biomarker for poor prognosis. An exploration of transcriptomic data from 373 liver cancer patients from The Cancer Genome Atlas (TCGA) using Ingenuity Pathway Analysis (IPA) suggested an association between DDX3X expression and cancer metastasis. Lentiviral-based silencing of DDX3X in a hepatocellular carcinoma (HCC) cell line resulted in the suppression of cell migration and invasion. The molecular mechanism regarding ATP-dependent RNA helicase DDX3X in liver cancer progression had been addressed in many studies. I focused on the biological application of the DDX3X-mediated gene expression signature in cancer therapeutics. An investigation of the DDX3X-correlated expression signature via the L1000 platform of Connectivity Map (BROAD Institute) first identified a histone methyltransferase inhibitor, chaetocin, as a novel compound for alleviating metastasis in HCC. In this study, the prognostic value of DDX3X and the antimetastatic property of chaetocin are presented to shed light on the development of anti-liver cancer strategies.


Introduction
Emerging evidence indicates the critical regulatory role of DEAD (Asp-Glu-Ala-Asp) Box Polypeptide 3, X-Linked (DDX3X) in cancer progression. ATP-dependent RNA helicase DDX3X, a RNA helicase, is a DEAD-box family member, and has been reported to be involved in the splicing of pre-mRNA [1], RNA export [2], the transcription of RNA [3], and protein translation [4][5][6]. Due to its complex biological role in RNA metabolism, ATP-dependent RNA helicase DDX3X has gained increasing attention for its function in many types of cancer, and it regulates tumor progression in a complex manner. Furthermore, this complexity is further increased because ATP-dependent RNA helicase DDX3X generally exhibits its biological effects as components of multiprotein complexes [7]. The exact impacts of ATP-dependent RNA helicase DDX3X are affected by its interacting partners [8]. The overexpression of ATP-dependent RNA helicase DDX3X has been detected in hepatocellular carcinoma (HCC), and ATP-dependent RNA helicase DDX3X has been characterized as a critical gene in hepatocarcinogenesis [9]. In contrast, the loss of ATPdependent RNA helicase DDX3X was found to lead to tumorigenesis due to a reduction in the expression of DNA repair factors in a mouse model [10]. Other studies revealed the loss of DDX3X expression in hepatocellular carcinoma tissue, and increased tumor cell proliferation because of the DDX3X silencing in hepatocellular carcinomas which were infected by hepatitis virus [3,11]. Hence, the conflicting roles of ATP-dependent RNA helicase DDX3X are inconsistent within the same type of cancer.
These conflicting results emphasize the urgent need for clarification. The prognostic role of DDX3X is a critical factor in determining whether DDX3X plays an oncogenic or tumor-suppressive role. In this study, I first aimed to characterize DDX3X's role from the perspective of its prognostic significance in hepatoma patient survival. The current updated version of the liver cancer/hepatoma cohort, with more patients enrolled, was retrieved from The Cancer Genome Atlas (TCGA) database and reanalyzed. This analysis of the DDX3X-associated gene signature further suggested the potential value of studying its biological function in promoting hepatoma metastasis. Furthermore, ATP-dependent RNA helicase DDX3X has been proposed as a therapeutic target for cancers [12]. I also aimed to identify a novel use for a drug in reversing the DDX3X-mediated gene signature via a Connectivity Map (BROAD Institute) [13]. In this study, I demonstrated the prognostic value of DDX3X and the potential capability of a histone methyltransferase inhibitor, chaetocin, as a therapeutic target to shed light on the development of anti-liver cancer strategies.

TCGA Dataset Analysis
Gene expression in the TCGA liver hepatocellular carcinoma (LIHC) dataset (Dataset ID: TCGA.LIHC.sampleMap/HiSeqV2_PANCAN, version: 2017-10-13) was estimated by RNA-Seq (Illumina HiSeq), and these data were retrieved along with the associated clinical information for data analysis. The RNA-Seq read count was normalized and log2transformed. A total of 373 HCC cases were divided into a DDX3X-high group and a DDX3X-low group based on the ranking assigned by the DDX3X expression level.

Ingenuity Pathway Analysis
The differential gene expression signatures in the HCC cohort were obtained after dividing the cohort into two groups with relatively high and low DDX3X levels. The gene signatures were further analyzed using Ingenuity ® Pathway Analysis (IPA; QIAGEN, Hilden, Germany; https://digitalinsights.qiagen.com/products-overview/discovery-insights-portfolio/ analysisand-visualization/qiagen-ipa/ (accessed on 1 November 2022)) according to the instructions provided. A list of relevant networks, upstream regulators and algorithmically generated mechanistic networks based on connectivity was obtained upon the comparison of the Ingenuity ® Knowledge Database with the imported dataset. The canonical pathway analysis feature of IPA was also used to identify significant diseases and functions in rank order based on the altered gene signatures.

Lentivirus-Based shRNA Production and Infection
The lentiviral shRNA constructs were purchased from Thermo Scientific (Pittsburgh, PA, USA). Lentiviruses were produced via co-transfection of 293T cells with an shRNA-expressing plasmid, an envelope plasmid (pMD.G), and a packaging plasmid (pCMV-dR8.91) using calcium phosphate (Invitrogen, Carlsbad, CA, USA). The 293T cells were incubated for 18 h, followed by the replacement of the culture medium. The viral supernatants were harvested and titered at 48 and 72 h post-transfection. The cell monolayers were infected with the indicated lentivirus in the presence of polybrene and were further selected using puromycin.

Cell Migration and Invasion Assay
In vitro migration and invasion were investigated using a Transwell assay (Millipore, Bedford, MA, USA). For the invasion assay, transwell was additionally pre-coated with 35 µL of 3× diluted matrix matrigel (BD Biosciences Pharmingen, San Diego, CA, USA) for 30 min. A total of 2 × 10 5 cells maintained in serum-free culture medium were added to the upper chamber of the device, and the lower chamber was filled with culture medium containing 10% FBS. After the indicated incubation periods, the cells remaining on the upper surface of the filter membrane were carefully removed using a cotton swab. The membrane was then fixed, stained with a 10× dilution of Giemsa solution (Merck, Darmstadt, Germany) and photographed. Cell motility was quantified by manually counting the cells in three random fields per filter membrane.

Wound Healing Assay
The wound healing assay was assessed using culture inserts (Ibidi, Martinsried, Germany). The culture inserts were transferred to plates. The cells were seeded at a density of 2 × 10 5 cells/well and were allowed to attach. After incubation, the culture inserts were removed using sterile tweezers and washed with PBS. The plates were filled with culture medium supplemented with 2% serum to induce cell migration. The cells were photographed for quantification of closure of the exposed area. The denuded area closure was calculated by (Denuded distance 0h − Denuded distance Endpoint )/Denuded distance 0h .

Western Blot Analysis
Cells were lysed using RIPA buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 0.25% sodium deoxycholate, 5 mM EDTA (pH 8.0), and 1 mM EGTA supplemented with protease and phosphatase inhibitors. After 20 min of lysis on ice, cell debris was removed via microcentrifugation and the supernatants were rapidly frozen. The protein concentration was measured via the Bradford method. In my experiments, equivalent samples containing 25-100 µg of protein were loaded onto an SDS-polyacrylamide gel, separated by electrophoresis, and electrophoretically transferred from the gel onto a polyvinylidene fluoride (PVDF) membrane (Millipore, Bedford, MA, USA). After blocking with 5% nonfat milk, the membrane was hybridized with specific primary antibodies overnight at 4 • C and subsequently incubated with a corresponding horseradish peroxidaseconjugated secondary antibody for 1 h. The relative levels of proteins on the membranes were determined using an ECL-Plus Detection Kit (PerkinElmer Life Sciences, Boston, MA, USA).

Statistical Analysis
Estimated survival rates were determined using the Kaplan-Meier method and were compared using the log-rank test. Student's t test was performed for other statistical analyses. All data are shown as the mean ± S.D. values. The p values within the following ranges were considered significant: * p < 0.05, ** p < 0.01, and *** p < 0.001. The experiments were performed at least three times and shown by the representative.

DDX3X Expression Is
Significantly Correlated with Poor Outcomes in Liver Cancer, and the Significantly Differential Gene Signature Was Identified after Comparison between Patient Groups with High and Low DDX3X Levels I first considered DDX3X's clinical significance in cancer patient cohorts. A comprehensive pan-cancer study integrating cancer patients' clinical data with RNA expression profile has been completed and released from the database: Human Protein Atlas (HPA) [14][15][16][17][18] and Kaplan-Meier plotter [19]. The prognostic data of DDX3X in different cancer types is listed and shown in Table 1. DDX3X expression predicted better survival rate in colorectal cancer, urothelial cancer, lung cancer and gastric cancer. However, DDX3X appeared to be an unfavorable prognostic biomarker in liver, pancreatic, breast and ovarian cancer. Among those cancer types, the up-to-date version of transcriptomic and clinical data of 373 liver cancer patients were analyzed. A scatter plot of the indicated expression patterns in the 373 patients is shown in Figure 1A.  Survival data were collected from databases The Human Protein Atlas, TCGA and Kaplan-Meier plotter. N.S.: no significance. The patients were divided into two groups with high or low DDX3X expression. The patients in the DDX3X-high group had a poor overall survival rate (p = 0.024, Figure 1B). Data from Cox regression analyses also indicated a trend toward DDX3X being useful as a prognostic factor ( Table 2). These results suggest DDX3X's prognostic value and indicate that further investigation regarding the molecular mechanisms altered by its overexpression is warranted. In addition, a total of 2890 differentially expressed gene targets in the DDX3X-high group were selected and shown after hierarchical clustering analysis ( Figure 1C and Supplementary Data Table S1).

Knowledge-Based Analysis of Gene Signatures Reveals the Potential for Triggering Cancer Metastasis and Progression
The gene signature identified in the DDX3X-high group was analyzed with knowledgebased IPA. The results obtained with the canonical pathway module of IPA identified the significant pathways ranked by overlap (log p value), and the activation status of each signaling was determined by the transformed Z score (Activation Z score, Figure 1D). The similarity of the high DDX3X gene signature to those related to the activation of colorectal cancer metastasis and planar cell polarity (PCP) signaling was identified ( Figure 1D,E). The inhibition of PTEN signaling was observed (Z score: −2.921), suggesting the activation of Akt ( Figure 1D).
In addition, PCP pathway activation can lead to increased cancer proliferation and metastasis [20,21]. In Figure 2A, classic PCP signaling pathway was illustrated after IPA analysis of the gene signature observed in liver cancer patients with high DDX3X levels. Statistically highly expressed molecules including WNT9A, WNT2, WNT7B, ROR2, CTHRC1, DAMM1 and ATF2 were shown ( Figure 2B). In a clinical setting, most of the aforementioned molecules were associated with poor overall survival in the liver cancer cohort (Figure 3). Moreover, the gene signature in the high DDX3X group was investigated by the analysis match module of IPA, which combined the curated and publicly available datasets. A significant similarity with those datasets annotated with the increased cell movement, migration of cells, cell movement of tumor cell lines and invasion of cells was detected (Table 3 and Figure 4). Hence, the transcriptomic data indicate the possibility that high DDX3X expression in liver cancer patients might trigger cancer progression, especially via metastasis, consistent with the observation of poor clinical outcomes ( Figure 1B).

Knockdown of DDX3X Expression Inhibits Cell Migration and Invasion in a Liver Cancer Cell Line
Therefore, I focused on ATP-dependent RNA helicase DDX3X's biological effect on modulating cancer cell migration. The relative cell migration levels were determined in an HCC cell line panel including HA 22T/VGH, HA 59T/VGH, C3A (HepG2/C3A) and HepG2 cells ( Figure 5A). HA 22T/VGH exhibited significant malignant metastatic behavior ( Figure 5B). ATP-dependent RNA helicase DDX3X expression was further stably silenced via the lentiviral transduction of two specific shRNAs in HA 22T/VGH cells ( Figure 5C). Decreases in both cell migration and invasion were found after DDX3X silencing (p < 0.001, Figure 5D,E). Hence, roles of ATP-dependent RNA helicase DDX3X in regulating cancer cell migration and invasion are proposed.

Knockdown of DDX3X Expression Inhibits Cell Migration and Invasion in a Liver Cancer Cell Line
Therefore, I focused on ATP-dependent RNA helicase DDX3X's biological effect on modulating cancer cell migration. The relative cell migration levels were determined in an HCC cell line panel including HA 22T/VGH, HA 59T/VGH, C3A (HepG2/C3A) and HepG2 cells ( Figure 5A). HA 22T/VGH exhibited significant malignant metastatic behavior ( Figure 5B). ATP-dependent RNA helicase DDX3X expression was further stably silenced via the lentiviral transduction of two specific shRNAs in HA 22T/VGH cells ( Figure  5C). Decreases in both cell migration and invasion were found after DDX3X silencing (p < 0.001, Figure 5D,E). Hence, roles of ATP-dependent RNA helicase DDX3X in regulating cancer cell migration and invasion are proposed.  Relative ATP-dependent RNA helicase DDX3X levels were evaluated by Western blotting. NS: nonsilencing control (scrambled RNA). (D,E) The migration and invasion ability of the indicated HA 22T/VGH cells were investigated in Transwell devices. *** p < 0.001.

Chaetocin Reverses the High DDX3X Expression-Mediated Gene Signature and Suppresses Liver Cancer Cell Migration
I further performed in silico data analysis to identify suitable compounds for inhibiting HCC progression. The updates of the Connectivity Map containing over 1.3 million L1000 profiles have been released [13]. I compared the DDX3X perturbation analysis with those data in the L1000 platform of the Connectivity Map to screen potential therapeutic compounds that could reverse the gene expression signature detected in the HCC patient group with high DDX3X expression. Candidate compounds were listed according to the negative connectivity scores determined after perturbation analysis ( Figure 6A). Mitomycin-c, gemcitabine and chaetocin were the top three candidates that appeared to possess the ability to reverse the gene signature observed in the DDX3X-high group. Both mitomycin-c and gemcitabine have been proven to reduce cancer cell metastasis [22,23]. Chaetocin is a fungal metabolite isolated from Chaetomium species fungi and shows various pharmacological and biological functions, including the ability to inhibit histone lysine methyltransferase activity [24]. Chaetocin was found to repress the self-renewal of bladder cancer stem cells [25] and to act as a sensitizer of apoptosis in glioblastoma [26]. However, the biological impact of chaetocin on cancer metastasis remains unknown. Chaetocin was first selected for the next tests of therapeutic capabilities to suppress HCC cell migration in this study. As shown in Figure 6B, chaetocin treatment resulted in a dose-dependent reduction in HA 22T/VGH cell proliferation. In addition, a sublethal concentration of 200 nM chaetocin was used for the treatment of HA 22T/VGH cells. The experimental results further showed chaetocin's inhibitory effect on cancer cell migration ( Figure 6C,D).

Chaetocin Reverses the High DDX3X Expression-Mediated Gene Signature and Suppresses Liver Cancer Cell Migration
I further performed in silico data analysis to identify suitable compounds for inhibiting HCC progression. The updates of the Connectivity Map containing over 1.3 million L1000 profiles have been released [13]. I compared the DDX3X perturbation analysis with those data in the L1000 platform of the Connectivity Map to screen potential therapeutic compounds that could reverse the gene expression signature detected in the HCC patient group with high DDX3X expression. Candidate compounds were listed according to the negative connectivity scores determined after perturbation analysis ( Figure 6A). Mitomycin-c, gemcitabine and chaetocin were the top three candidates that appeared to possess the ability to reverse the gene signature observed in the DDX3X-high group. Both mitomycin-c and gemcitabine have been proven to reduce cancer cell metastasis [22,23]. Chaetocin is a fungal metabolite isolated from Chaetomium species fungi and shows various pharmacological and biological functions, including the ability to inhibit histone lysine methyltransferase activity [24]. Chaetocin was found to repress the self-renewal of bladder cancer stem cells [25] and to act as a sensitizer of apoptosis in glioblastoma [26]. However, the biological impact of chaetocin on cancer metastasis remains unknown. Chaetocin was first selected for the next tests of therapeutic capabilities to suppress HCC cell migration in this study. As shown in Figure 6B, chaetocin treatment resulted in a dose-dependent reduction in HA 22T/VGH cell proliferation. In addition, a sublethal concentration of 200 nM chaetocin was used for the treatment of HA 22T/VGH cells. The experimental results further showed chaetocin's inhibitory effect on cancer cell migration ( Figure 6C,D).

Conclusions
In this study, DDX3X was characterized as a poor prognostic biomarker for an HCC patient cohort, and correlations of high DDX3X level with cancer metastasis and the activation of PCP signaling in HCC cells were discovered. DDX3X knockdown in the HCC cell line results in the repression of cells' migration and invasion ability. Furthermore, my study demonstrates for the first time a high-throughput drug screening-based characterization of chaetocin and reveals its potential antimetastatic effect, especially in HCC cells with high DDX3X expression levels, as illustrated in Figure 7.

Conclusions
In this study, DDX3X was characterized as a poor prognostic biomarker for an HCC patient cohort, and correlations of high DDX3X level with cancer metastasis and the activation of PCP signaling in HCC cells were discovered. DDX3X knockdown in the HCC cell line results in the repression of cells' migration and invasion ability. Furthermore, my study demonstrates for the first time a high-throughput drug screening-based characterization of chaetocin and reveals its potential antimetastatic effect, especially in HCC cells with high DDX3X expression levels, as illustrated in Figure 7.

Discussion
In this study, I characterized DDX3X's clinical significance in predicting poor hepatoma patient survival based on RNA expression data. The experimental findings indicated that ATP-dependent RNA helicase DDX3X might potentially possess functions in regulating liver cancer progression. Hence, the next urgent need is to investigate the clinical significance including the associations with patient survival rate, TNM status and clinicopathological features at a protein level in a patient cohort of HCC.
In this research study, I first observed the potentially therapeutic application of chaetocin for hepatocellular carcinoma in terms of precision medicine; that is, especially for those cancer patients displaying high DDX3X expression levels. Actually, ATP-dependent RNA helicase DDX3X appeared to be modulated by several compounds and cytokines via direct and/or indirect interactions. In liver hepatocellular cells, previous findings indicated that 5-HT treatment could augment 5-HT receptor 7-mediated DDX3X promoter activity as well as the induction of an innate immunity to abolish hepatitis B virus (HBV) infection [27]. Another research finding in hepatocellular HepG2 cells demonstrated that the addition of tazemetostat, SP2509, decitabine and trichostatin A led to the downregulation of DDX3X RNA [28]. Inhibiting ATP-dependent RNA helicase DDX3X's ATP binding domain with the small molecule RK-33 was one of choices, and the effect was discovered to synergize with radiotherapy to reduce tumor cells' proliferation in vitro and in vivo [29]. In a cancer-related study, RK-33 treatment was indicated to inactivate the WNT signaling axis and induce cell cycle G1 phase arrest, leading to cell apoptosis in lung cancer [30]. In addition, 1,3,4thiadiazole is another ATPase activity inhibitor for ATP-dependent RNA helicase DDX3X, synthesized for blocking HIV-1-mediated effects [31]. The design of ATP-competitive inhibitors reported that FE15 and FE109 appeared to block helicase and the ATPase function of ATP-dependent RNA helicase DDX3X, along with having the capability to reduce the HIV viral load in peripheral blood mononuclear cells [32]. Furthermore, doxorubicin is considered to be one of the ATP-dependent RNA helicase DDX3X inhibitors via an in silico molecular docking approach which showed a potential interaction with common amino acid residues Tyr200/Thr201 and unique amino acid residue Thr198 via doxorubicin. The realistic function was also corroborated in oral squamous cell carcinoma H357 cells. DDX3X downregulation and the decreases in ATP hydrolysis, inorganic phosphate release and cancer proliferation were reported [33]. High throughput virtual screening was performed to identify ATP-dependent RNA helicase DDX3X inhibitors, in which Ketorolac salt had been characterized as a bioactive compound for its ability to bind with ATP-dependent RNA helicase DDX3X. The interaction further appeared to suppress the tumor growth in oral cancer [34]. In contrast to the aforementioned strategies, I first explored the drug candidates, focusing on the basis of comprehensively reversing DDX3X-dependent downstream gene signatures, and characterized chaetocin as one of potential targets for therapeutics in liver cancer. Nevertheless, several molecules related to upstream regulation and the direct interaction of ATP-dependent RNA helicase DDX3X might be noticed. AGR2 is a member of the protein disulfide isomerase family, which acts as a proto-oncogene. AGR2 was found to trigger cancer metastasis in animal models, and the interaction of ATP-dependent RNA helicase DDX3X with AGR2 has been characterized at protein level [35]. In addition, fibronectin 1 was uncovered to upregulate the expression of human DDX3X at an RNA level in cultured HUVEC cells in a microarray analysis [36]. Ginsenoside Rg3 stimulus-enhanced DDX3X upregulation and the Akt-p53-dependent DDX3X promoter transactivation were detected, along with the activation of innate immune response via the TBK1-IKKε-IRF3 pathway [37].
Chaetocin was identified based on the comparison of gene signatures derived from those cancer patients displaying high DDX3X levels with the datasets in the Connectiv-ityMap database (negative connectivity score, Figure 6A). The same gene signature was simultaneously analyzed via IPA. A significant signaling axis, the PCP pathway, was characterized according to the activation Z score ( Figure 1D), which is a pivotal stimulus in triggering cancer progression [20,21]. Therefore, the PCP signaling axis might serve as a potential route inhibited by chaetocin to reverse DDX3X-mediated effects because of the overlap of similar downstream molecules.
Data Availability Statement: All necessary data have been provided in this manuscript and Supplementary Data.