FXR1‐Directed Alternative Splicing of MK5 Drives Hepatocellular Carcinoma Progression by Activating GSK3β Signaling

ABSTRACT

Alternative splicing plays a crucial role in the development and progression of hepatocellular carcinoma (HCC), yet the underlying regulatory mechanisms and therapeutic potential remain largely unexplored. Here, we identified the RNA‐binding protein FMR1 autosomal homolog 1 (FXR1) as a key driver of HCC pathogenesis through the regulation of alternative splicing. FXR1 was highly expressed in HCC tissues, and its elevated expression was associated with a poor prognosis. Mechanistically, high FXR1 induced the retention of exon 6 in mitogen‐activated protein kinase‐activated protein kinase 5 (MK5), generating a long, kinase‐competent isoform (termed MK5‐L). This isoform acted as an important oncogenic factor for HCC progression by phosphorylating GSK3β and subsequently activating the Wnt/β‐catenin pathway. Functional studies showed that the FXR1/MK5‐L axis is critical for HCC cell proliferation and metastasis, both in vivo and in vitro. Importantly, therapeutic intervention using an FXR1‐targeting antisense oligonucleotide (ASO) effectively suppresses tumor progression and metastasis in preclinical models by shifting splicing toward the inactive MK5‐S isoform. Overall, our study unveils a novel splicing‐mediated oncogenic pathway and establishes FXR1 and its downstream target MK5‐L as promising prognostic biomarkers and therapeutic targets for HCC.

The RNA‐binding protein FXR1 promotes hepatocellular carcinoma (HCC) progression by driving alternative splicing of MK5 to generate an oncogenic isoform, MK5‐L, which activates the Wnt/β‐catenin pathway. Targeting FXR1 with an antisense oligonucleotide suppresses tumor growth in vivo, revealing a promising therapeutic strategy for HCC.

Abbreviations

ASO

antisense oligonucleotide

c‐Myc

cellular myc

E‐cadherin

epithelial‐cadherin

FMRP

fragile X mental retardation protein

FXR1

fragile X mental retardation autosomal 1

FXR2

fragile X mental retardation autosomal 2

HCC

hepatocellular carcinoma

MAPKAPK5

mitogen‐activated protein kinase‐activated protein kinase 5

p‐GSK3β

phosphorylated‐glycogen synthase kinase 3 β

1. Introduction

Liver cancer is the sixth most common cancer and the third leading cause of cancer‐related deaths globally [1, 2]. Hepatocellular carcinoma (HCC), which accounts for about 80% of primary liver cancers, is characterized by a poor prognosis in advanced stages, largely due to an incomplete understanding of its molecular pathogenesis and a consequent lack of effective treatment strategies [3, 4]. Therefore, it is necessary to gain a deeper understanding of the molecular mechanisms underlying the development of HCC and develop more effective therapeutic strategies. Notably, alternative splicing has emerged as a significant source of potential prognostic markers and therapeutic targets in HCC [5].

Alternative splicing is a fundamental process that generates multiple distinct mature transcripts from a single precursor mRNA, thereby expanding the diversity of the transcriptome and proteome [6, 7]. This process is tightly regulated in a tissue‐specific and cell‐type‐specific manner, as well as throughout various stages of cellular differentiation [8, 9, 10]. RNA‐binding proteins (RBPs) are key regulators of alternative splicing [11], directing mechanisms such as exon inclusion/exclusion or alternative splice site selection by binding to specific sequences in pre‐mRNA [12]. In HCC, dysregulation of RBP expression represents a potential mechanism for widespread splicing alterations observed between tumor and non‐tumor tissues [13]. Supporting this notion, a comparative analysis has revealed that 231 RBP‐encoding genes were upregulated, while 55 were downregulated, in HCC tumors compared to adjacent nontumorous tissue [14].

The Fragile X proteins (FXPs) are a small subfamily of RBPs containing three members: FMRP (encoded by FMR1), FXR1P (encoded by FXR1), and FXR2P (encoded by FXR2) [15]. These proteins share a remarkable structural similarity, with approximately 60% amino acid sequence identity [16]. Of the three, FXR1 is particularly noteworthy as it interacts with diverse proteins, coding RNAs, and non‐coding RNAs (ncRNAs) to form ribonucleoprotein complexes. These complexes play a crucial role in regulating RNA splicing, translation, localization, stability, polyadenylation, and degradation [17]. Importantly, emerging evidence suggests that among the FXRPs, FXR1 exerts the most potent oncogenic functions across multiple cancer types [18]. For instance, in non‐small cell lung cancer, FXR1 promotes tumor cell growth in vitro and in vivo by forming complexes with partners such as ECT2 and PRKCI [19]. Similarly, in oral cancer, FXR1 binds to miR301a‐3p to facilitate the degradation of p21 mRNA, thereby promoting tumor progression [20]. However, very little is known about the role of FXR1 in the progression of HCC.

In this study, we found that FXR1 is highly expressed in HCC and that elevated FXR1 expression correlates with poorer prognosis for patients. Functional investigations, comprising in vitro and in vivo assays, showed that FXR1 promotes the malignant phenotypes of HCC cells. Mechanistically, we demonstrated that FXR1 regulates the alternative splicing of mitogen‐activated protein kinase‐activated protein kinase 5 (MK5), leading to preferential production of the long isoform MK5‐L, which in turn activates GSK3β and promotes the malignant progression of HCC. In conclusion, our findings identify FXR1 as a potential prognostic biomarker and a promising therapeutic target for HCC.

2. Material and Methods

2.1. Antibodies

A list of the antibodies used is provided in Table S1.

2.2. RNA Isolation and PCR Analysis

Total RNA was extracted with TRIzol reagent (Invitrogen), and cDNA was synthesized using the PrimeScript RT Reagent Kit (TaKaRa). Quantitative PCR (qPCR) was performed with SYBR Green Master Mix (Bio‐Rad) on a Bio‐Rad CFX96 system, with data analyzed via the 2^−ΔΔCT method. Semiquantitative RT‐PCR was used to amplify specific spliced isoforms, with primers listed in Table S2.

2.3. Immunohistochemistry

Deparaffinized liver sections are stained with primary and secondary antibodies, visualized with DAB, and counterstained with hematoxylin for analysis. The final degree of immunostaining was evaluated based on the extent and intensity of staining. For this analysis, adjacent non‐tumor tissue sections were defined as those taken ≥ 2 cm from the tumor margin and pathologically confirmed to be of hepatocyte origin without tumor infiltration.

2.4. CRISPR/Cas9‐Mediated Gene Knockout

Using the E‐CRISP tool, an sgRNA targeting FXR1 (sgFXR1: 5′‐GAGCTGACGGTGGAGGTTCG‐3′) was designed and cloned into a CRISPR lentiviral expression vector provided by Professor Xue Ding at Tsinghua University in Beijing, China, which can express the Cas9 protein. Lentiviral packaging was then performed, and stable knockout cell lines were selected [21].

2.5. Adenoviruses Production

The full‐length cDNA of FXR1 was inserted into the pAdTrack‐TO4 plasmid provided by Dr. T‐C He at the University of Chicago. The recombinant adenovirus AdFXR1 (with a FLAG tag) was constructed using the AdEasy system. The AdEasy system is widely used for packaging and constructing recombinant adenoviruses due to its high infectivity advantage [22, 23]. AdGFP, which expresses only green fluorescent protein, served as a control.

2.6. Others

Other materials and methods are described in the Doc S1.

3. Results

3.1. FXR1 Is Upregulated in HCC and Indicates a Poor Prognosis

Given the crucial role of alternative splicing in HCC progression, we first analyzed a set of alternative splicing events from the TCGA‐LIHC dataset by comparing the percent‐spliced‐in (PSI) values of different genes between normal liver tissues and HCC tissues. This analysis compared the splicing events between HCC tissues and their adjacent non‐tumor tissues (Figure S1A). We identified approximately 2000 alternative splicing events in normal tissue and HCC tissue. For further analysis, we filtered these events, retaining only those detected in at least 60% of the samples with |ΔPSI| > 10% (where PSI is per cent spliced in). Statistical significance was required, with p value < 0.05 and FDR < 0.05.

Next, we performed unbiased de novo motif analysis of these alternative splicing events using the XSTREME tool. A top‐ranked enriched motif was similar to the known FXR1 binding motif (E‐value = 9.39e‐05) (Figure 1A). FXR1 has been reported to regulate the variable splicing process of mRNA during Epithelial‐Mesenchymal Transition (EMT) [24]. Notably, our analysis of the TCGA cohort showed that 12% of HCC patients exhibited genomic amplification of FXR1 (Figure 1B), and this amplification correlated with its high expression. Consistent with this, interrogation of independent datasets from TCGA and GEO databases confirmed that FXR1 mRNA levels were significantly upregulated in HCC samples compared to normal liver tissues (Figure 1C–E). We validated these findings using qRT‐PCR and western blotting on clinical specimens, which confirmed that both FXR1 mRNA and protein were markedly increased in HCC tissues (Figure 1F–H). Immunohistochemistry (IHC) staining of 20 pairs of HCC and paracarcinoma tissue sections further corroborated the predominant overexpression of FXR1 protein in HCC tissues (Figure 1I). To assess the clinical relevance of FXR1 upregulation, we analyzed patient survival data from public databases. In the TCGA cohort, high FXR1 expression was significantly associated with poorer overall survival (Figure 1J). Furthermore, both univariate and multivariate Cox regression analyses confirmed that FXR1 expression level was an independent prognostic factor for overall survival in HCC patients (Figure S1B). Collectively, these results suggest that FXR1 is a potential prognostic biomarker and a candidate therapeutic target for HCC.

FIGURE 1.

FXR1 is upregulated in HCC and indicates a poor prognosis. (A) De novo motif analysis of differentially spliced event in TCGA‐LIHC (HCC vs. normal liver) using the XSTREME Tool. The identified motif exhibits strong similarity to the known FXR1 binding motif (E‐value = 9.39e‐05). (B) Genetic alterations of FXR1 in Liver cancer in the cohort from CBioPortal (n = 366). (C–E) FXR1 mRNA expression levels were analyzed in normal tissues and HCC tissues at different clinical stages in the TCGA database and the GEO database. (F, G) Analysis of FXR1 protein expression levels in paired human liver cancer and adjacent non‐tumor tissues (n = 36). (H) Analysis of FXR1 mRNA expression levels in paired human liver cancer and adjacent non‐tumor tissues (n = 24). (I) Representative IHC staining for FXR1 in HCC paraffin sections. Scores (0–3) were calculated according to the percentage of stained cells and intensity. Scale bar = 50 μm. (J) Survival analysis based on the Cancer Genome Atlas (TCGA)‐HCC database was used to assess the prognostic survival rate of HCC patients for FXR1. The log‐rank test was used for analyzing differences in prognosis. p‐values were derived from one‐way ANOVA, followed by Tukey's test in (D), unpaired two‐tailed Student's t test in (C, E), and paired two‐tailed Student's t test in (G–I). Data are represented as the mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001.

3.2. FXR1 Promotes the Proliferation and Migration of HCC Cells

To establish the causal role of FXR1 in hepatocellular carcinoma (HCC), we first detected its endogenous expression across multiple HCC cell lines (Figure 2A) and selected cells with low or high expression for further study. We generated FXR1‐knockout (KO) cells and corresponding rescue models, while also constructing adenovirus‐mediated FXR1‐overexpressing cells (Figure 2B,C). Functional assays showed that FXR1 knockout significantly inhibited cell proliferation and migration compared to controls, phenotypes that were fully restored in the rescue groups (Figure 2D,F,H,J and Figure S2A,C,E). Conversely, FXR1 overexpression promoted both proliferation and migration (Figure 2E,G,I,K and Figure S2B,D,F). These in vitro data demonstrate that FXR1 is both necessary and sufficient for HCC cell proliferation and migration.

FIGURE 2.

FXR1 promotes the proliferation and migration of HCC cells. (A) Endogenous FXR1 protein levels in human HCC cell lines. (B) Validation of FXR1 knockout (KO) and rescue (Res) by Western blot. (C) Validation of FXR1 overexpression by Western blot. (D, E) Cell growth curves under indicated conditions. (F, G) Colony formation assays. FXR1 loss reduces clonogenicity, which is restored upon re‐expression (F), while overexpression promotes it (G). (H, I) EdU assays measuring DNA synthesis. (J, K) Transwell migration assays. FXR1 knockout inhibits migration, and re‐expression reverses this effect (J); overexpression enhances migration (K). p‐values were derived from two‐way ANOVA, followed by Tukey's test in (D) and one‐way ANOVA, followed by Tukey's test in (F, H, J) and unpaired two‐tailed Student's t test in (E, G, I, K). Data are represented as the mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001.

3.3. FXR1‐Mediated Splicing Induces MK5 Exon 6 Retention in HCC Cells

Given our initial finding that splicing‐altered genes in HCC harbor homologous to the FXR1 binding site, we hypothesized that FXR1 contributes to HCC progression by regulating alternative splicing (AS). To test this, we performed RNA sequencing on FXR1‐overexpressing PLC/PRF/5 cells and AdGFP control cells to identify FXR1‐regulated AS events (Figure 3A). Our analysis identified a total of 1751 FXR1‐regulated AS events, including 1126 skipped exons (SE), 127 alternative 5′ splice sites (A5SS), 151 alternative 3′ splice sites (A3SS), 314 mutually exclusive exons (MXE), and 33 intron retention (IR) (Figure 3B).

FIGURE 3.

FXR1 regulates MK5 pre‐mRNA splicing. (A) Workflow for RNA‐seq and subsequent mRNA splicing analysis of control cells and FXR1‐overexpressing PLC/PRF/5 cells. This figure was drawn by BioRender. (B) Pre‐mRNA alternative splicing events regulated by FXR1 were analyzed by RNA splicing analysis. A3SS, alternative 3′ splicing site; SE, skipping exon; RI, retention intron; MXE, mutually exclusive exon; A5SS, alternative 5′ splicing site. (C) Different splicing variants of FXR1 exhibit different exon skipping events in control (blue) and FXR1‐overexpressing cells (red). Schematic diagram of exon 6 junction in aligned RNA sequencing data; supported junction reads are marked between exons. (D) Sequence diagram of MK5‐L and MK5‐S (confirmed exon 6), with exon 6 containing the kinase site. (E) The PCR products from MHCC‐97H cells infected with FXR1 sgRNA or negative control sgRNA lentivirus and PLC/PRF/5 cells infected with AdFXR1 or negative control AdGFP adenovirus were analyzed by agarose gel electrophoresis to detect MK5 PSI (PSI = splice_in/(splice_in+splice_out)) (n = 3 independent experiments). (F) IF and RNA FISH detection of FXR1 and MK5 mRNA in PLC/PRF/5 cells. IF/FISH shows colocalization of MK5 mRNA (red), FXR1 protein (green), and DAPI (blue)‐stained cell nuclei. Scale bar = 25 μm. (G) RIP‐qPCR validation of FXR1 interaction with MK5 mRNA in PLC/PRF/5 cells and MHCC‐97H cells overexpressing FXR1 (n = 3 independent experiments). (H) Survival analysis of HCC patients from the TCGA cohort stratified by MK5‐L expression. The optimal cutoff (PSI = 0.904) for dichotomization was determined using maximally selected rank statistics. The association with overall survival was evaluated by the log‐rank test. p‐values were derived from unpaired two‐tailed Student's t test. Data are represented as mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001.

Next, to identify key DEGs involved in the variable splicing process, we conducted a KEGG pathway enrichment analysis on the genes differentially spliced after FXR1 overexpression (Figure S3A). Given that the MAPK pathway is a crucial regulator of fundamental cellular processes, including proliferation and migration [25], we focused our investigation on six candidate genes from this pathway (FDR < 0.05). A subsequent correlation analysis of FXR1 with these six DEGs in the TCGA database revealed that MAPKAPK5 (MK5) showed the highest correlation with FXR1 (Figure S3B). Therefore, we selected MK5 for further study.

A strong correlation was found between FXR1 expression levels and the magnitude of the splicing event in MK5‐L, and Sashimi plot revealed that MK5 undergoes differential splicing at exon 6 (Figure 3C). Notably, exon 6 encodes a critical portion of the MK5 kinase domain (Figure 3D). We confirmed this finding using RT‐PCR assay with isoform‐specific primers: FXR1 overexpression increased the abundance of the long isoform containing exon 6 (MK5‐L), while FXR1 knockout favored the expression of the short isoform lacking exon 6 (MK5‐S) (Figure 3E). We also performed RNA fluorescence in situ hybridization experiments and found strong co‐localization of MK5 mRNA with FXR1 protein in PLC/PRF/5 cells (Figure 3F). Furthermore, RIP‐qPCR revealed that FXR1 binding to MK5 mRNA was significantly increased in FXR1‐overexpressing hepatocellular carcinoma cells compared to the control group (Figure 3G). More importantly, analysis of the sequencing data from the OncoSplicing database revealed that HCC patients with high MK5‐L expression exhibit a poorer overall prognosis (Figure 3H). These findings collectively suggest that FXR1 promotes the expression of the oncogenic MK5‐L isoform by inducing the retention of exon 6.

3.4. FXR1/MK5‐L Axis Promotes Tumorigenicity and Metastasis In Vitro and In Vivo

Given that FXR1 is highly expressed in HCC and that elevated expression of MK5‐L isoform is closely associated with poor prognosis in patients, we hypothesized that a significant proportion of HCC tumors may predominantly express longer MK5‐L RNA and protein isoforms. To test this, we first examined the expression levels of MK5‐L in HCC patient samples and found that higher MK5‐L RNA and protein levels were significantly higher in tumor tissues compared to paired normal tissue samples (Figure 4A,B). To test the functional role of each MK5 isoform, we analyzed the oncogenic potential of MK5‐L and MK5‐S by overexpressing them in FXR1‐knockout HCC cells. The expression levels of MK5‐L and MK5‐S in each group were measured by qPCR and Western blot (Figure S4A,B). Our in vitro assays revealed that only the overexpression of MK5‐L, but not MK5‐S, was sufficient to rescue cellular invasion capacity, wound healing ability, and colony formation capability in these cells (Figure 4C–E and Figure S4C–E). These findings demonstrate that the oncogenic function of FXR1 in hepatocellular carcinoma is mediated, at least in part, through its regulation of MK5 alternative splicing, specifically by promoting the expression of the pro‐tumorigenic MK5‐L isoform. Next, to validate the carcinogenic effects of FXR1 and MK5‐L in vivo, we established both subcutaneous tumorigenesis and tail vein pulmonary metastasis models using nude mice. As expected, FXR1 knockout significantly inhibited tumor growth, as evidenced by decreased tumor volume and weight. Critically, this effect was rescued by MK5‐L, but not by MK5‐S (Figure 4F–H). To explore the role of FXR1 in HCC metastasis, MHCC‐97H cells were injected into BALB/c nude mice via the tail vein to assess lung colonization. The number of lung tumors from FXR1‐knockout MHCC‐97H cells was significantly lower than in the control group. Again, the metastatic potential was restored specifically upon reintroduction of MK5‐L (Figure 4I,J). Collectively, these findings indicate that the FXR1/MK5‐L axis acts as a driver of hepatocellular carcinoma in vitro and in vivo.

FIGURE 4.

FXR1/MK5‐L axis promotes tumorigenicity and metastasis in vitro and in vivo. (A) Representative images of MK5‐L and MK5‐S were obtained in liver cancer tissue and corresponding adjacent non‐tumor tissue using RT‐PCR technology (left figure). Analysis of PSI values (MK5‐L/(MK5‐L+MK5‐S)) in normal tissue and corresponding liver cancer tissue (right figure). (B) Western blot analysis of MK5‐L and MK5‐S expression in eight pairs of adjacent and HCC tissues from patients. (C) Representative images and quantitative results of clonogenic assays of PLC/PRF/5 cells (n = 3 independent experiments). (D) Representative images and quantitative analysis of transwell assays of PLC/PRF/5 cells (n = 3 independent experiments). (E) Representative images and quantitative results of wound‐healing assays of PLC/PRF/5 cells (n = 3 independent experiments). (F–H) MHCC‐97H cells were processed as described above and injected subcutaneously into nude mice (n = 6 per group). (F) Representative images of xenograft tumors are shown. Tumor weight (G) and volume (H) were measured and calculated. (I, J) Representative images of lung metastases (upper panel) and H&E staining (lower panel) are shown (I) and were quantitatively analyzed (J). Scale bar = 1000 μm. p‐values were derived from one‐way ANOVA, followed by Tukey's test in (C–J) and paired two‐tailed Student's t test in (A). Data are represented as mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001.

3.5. MK5‐L Phosphorylates GSK3β and Activates the Wnt/β‐Catenin Pathway

To further explore the mechanism by which MK5‐L promotes HCC progression, we investigated its downstream signaling. We note a previous report that MK5 can upregulate GSK3β phosphorylation at Ser9 and increase β‐catenin levels [26]. Given that exon 6, exclusive to the MK5‐L isoform, encodes the kinase active site of MK5, we hypothesized that MK5‐L directly interacts with and phosphorylates GSK3β. To test this, we performed endogenous and exogenous immunoprecipitation assays in PLC/PRF/5 cells and MHCC‐97H cells. Our results showed that both MK5‐L and MK5‐S isoforms can interact with GSK3β (Figure 5A–C). Furthermore, immunofluorescence experiments confirmed that MK5 co‐localizes with GSK3β in HCC cells (Figure 5D). Having established their interaction, we next sought to determine whether MK5‐L directly phosphorylates GSK3β. In vitro kinase assays revealed that MK5‐L, but not MK5‐S, directly phosphorylates GSK3β (Figure 5E). Next, we performed Gene Set Enrichment Analysis (GSEA) of FXR1 and revealed a significant positive correlation between FXR1 expression and the WNT signaling pathway (Figure S5A). Consistent with this, knockout of FXR1 downregulated key WNT pathway components (p‐GSK3β, β‐catenin, Cyclin D1, and c‐Myc) and upregulated E‐cadherin (Figure S5B). Conversely, FXR1 overexpression produced the opposite effect (Figure S5C). Crucially, the functional consequence was isoform‐specific. Overexpression of MK5‐L, but not MK5‐S, led to increased levels of p‐GSK3β and β‐catenin in PLC/PRF/5 and MHCC‐97H cells (Figure 5F). Furthermore, in FXR1‐knockout cells, the expression of MK5‐L—but not MK5‐S—rescued the expression of downstream WNT targets (p‐GSK3β, β‐catenin, Cyclin D1, c‐Myc) and suppressed E‐cadherin (Figure 5G). Moreover, IHC assays confirmed that FXR1 knockout reduced the protein levels of p‐GSK3β, β‐catenin, c‐MYC, and Cyclin D1. This suppression was specifically rescued by MK5‐L overexpression (Figure S5D). Thus, FXR1‐driven exon 6 retention converts MK5 into an active kinase that phosphorylates and inactivates GSK3β, thereby unleashing β‐catenin‐dependent transcriptional programs critical for HCC progression.

FIGURE 5.

MK5‐L phosphorylates GSK3β and activates the Wnt/β‐catenin pathway. (A) Endogenous Co‐immunoprecipitation analysis was performed to investigate the interaction between FXR1 and MK5 in PLC/PRF/5 and MHCC‐97H cells. (B) Co‐immunoprecipitation (Co‐IP) experiments were performed on PLC/PRF/5 cells transfected with Flag‐MK5‐L, Flag‐MK5‐S, and HA‐GSK3β using anti‐HA or anti‐Flag antibodies. IgG was used as a control. (C) Co‐immunoprecipitation (Co‐IP) experiments were performed on MHCC‐97H cells transfected with Flag‐MK5‐L, Flag‐MK5‐S, and HA‐GSK3β using anti‐HA or anti‐Flag antibodies. IgG was used as a control. (D) Representative immunofluorescence image for MK5 and GSK3β in PLC/PRF/5 and MHCC‐97H cells. Scale bar = 100 μm. (E) In vitro kinase assay. Workflow of the in vitro kinase assay (Left). Immunoblot analysis showing phosphorylation of GSK3β by MK5‐L (Right). (F) Western blot analysis of phosphorylated GSK3β (p‐GSK3β), GSK3β, β‐catenin, CyclinD1, c‐Myc, E‐cadherin and MK5 in hepatoma cells infected with AdGFP, AdMK5‐L, or AdMK5‐S. β‐actin was used as a loading control. (G) Expression of phosphorylated GSK3β (p‐GSK3β), GSK3β, β‐catenin, CyclinD1, c‐Myc, E‐cadherin and MK5 were detected by western blot in PLC/PRF/5 and MHCC‐97H cells transfected with indicated plasmid.

3.6. Therapeutic Targeting of FXR1 With ASO Suppresses HCC Progression

Antisense oligonucleotide (ASO) offers a clinically translatable strategy to block RNA‐binding proteins whose surfaces lack druggable pockets [27]. We therefore designed three ASOs specifically targeting FXR1 and evaluated their efficacy in PLC/PRF/5 and MHCC‐97H cells (Figure 6A). Transfection experiments identified ASO2 as the most potent candidate, achieving significant suppression of FXR1 expression (Figure S6A). Functionally, ASO2 treatment shifted the alternative splicing of MK5 toward the exclusion of exon 6, reducing the abundance of the oncogenic MK5‐L (Figure 6B). To further elucidate whether the effects of ASO2 were mediated through the MK5 splicing switch, we performed rescue experiments with four groups: control, ASO2‐treated, ASO2‐treated + MK5‐L rescue, and ASO2‐treated + MK5‐S rescue. Consistently, the downregulation of key WNT pathway effector proteins induced by ASO2 was specifically reversed by MK5‐L rescue, but not by MK5‐S rescue (Figure 6C). Similarly, the inhibitory effects of ASO2 on cell migration and colony formation were also rescued by MK5‐L, but not by MK5‐S (Figure 6D,E). Importantly, the therapeutic potential of ASO2 was confirmed in vivo. Administration of ASO2 markedly inhibited FXR1‐induced tumor growth in a xenograft model (Figure 6F–H). Additionally, in the experimental lung metastasis model, application of ASO2 significantly reduced the number of metastatic nodules in mouse lungs (Figure S6B,C). Similarly, immunohistochemical analysis also showed that ASO2 treatment inhibited the levels of p‐GSK3β, β‐catenin, c‐Myc, Cyclin D1 (Figure S6D). Collectively, these results demonstrate the efficacy of an ASO‐based approach in targeting the FXR1/MK5‐L axis and validate FXR1 as a promising therapeutic target for HCC.

FIGURE 6.

ASO‐mediated FXR1 knockdown suppresses the proliferation and metastasis of HCC cells. (A) ASO‐mediated degradation of target mRNA through the Rnase H1 pathway. This figure was drawn by BioRender. (B) The PSI of MK5 was analyzed using agarose gel electrophoresis of PCR products in PLC/PRF/5 cells and MHCC‐97H cells transfected with antisense oligonucleotide targeting FXR1. PSI was calculated by gel densitometry using Image J. (C) Western blot analysis of PLC/PRF/5 and MHCC‐97H cells treated with antisense oligonucleotide (ASO) or ASO, followed by re‐expression of MK5‐L or MK5‐S. (D, E) Effects of the indicated treatments on the migration and clonogenic ability of HCC cells (n = 3 independent experiments). ASO‐mediated suppression is specifically rescued by MK5‐L, but not by MK5‐S. (F) ASO injection to a subcutaneous tumor model in mice (n = 6 per group). (G, H) Analysis of tumor weight and tumor volume. p‐values were derived from one‐way ANOVA, followed by Tukey's test in (D, E) and unpaired two‐tailed Student's t test in (G, H). Data are represented as mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001.

4. Discussion

Alternative splicing of pre‐mRNA is a fundamental and highly regulated mechanism in eukaryotic gene expression [28]. In the liver, over 80% of protein‐coding genes were found to produce four or more distinct transcripts through alternative splicing [14]. Increasing evidence implicates dysregulated splicing as a key driver in the progression of HCC [29, 30, 31, 32]. RBPs are central regulators of AS that interact with RNA to form ribonucleoprotein complexes. This interaction determines the maturation and fate of their target RNA substrates and regulates various aspects of gene expression, including pre‐mRNA splicing [11]. Here, our research reveals the important role of the RNA‐binding protein FXR1 in promoting liver cancer progression by regulating alternative splicing.

The N‐terminus of the FXR1 protein consists of a tandem agent‐like domain, a non‐classical nuclear localization signal (NLS), a three protein K homology (KH), nuclear export signal (NES), and c‐terminal RGG box [33]. The presence of KH and RGG domains classifies FXR1 as a canonical RBP, enabling its multifaceted involvement in RNA processing, including stability, subcellular localization, translation, and alternative splicing [18]. Previous studies have shown that FXR1 is upregulated in head and neck cancer and lung squamous cell carcinoma [19, 34]. However, the mechanism by which FXR1 regulates tumorigenesis is highly complex and varies substantially across different cancer types. The significance of FXR1 in HCC remains poorly characterized. Our work addresses this gap by systematically profiling FXR1‐regulated AS events through RNA‐seq, revealing a predominant impact on exon skipping. This finding extends previous reports of FXR1‐mediated mRNA alternative splicing during EMT [24] by establishing its critical role in HCC, specifically through the modulation of MK5 splicing.

MK5 is a protein kinase with several potential downstream phosphorylation substrates, including HSP27, Foxo1, and Rheb [35]. It has been shown to participate in diverse cellular processes, including metabolism [36], autophagy [37] and the rearrangement of the actin cytoskeleton [38]. MK5 expression correlates with metastatic potential and poor prognosis in patients with lung cancer and breast cancer [39]. Furthermore, MK5 promotes oncogenesis by stabilizing and enhancing the nuclear localization of YAP/TAZ [40]. Here, we report for the first time the existence of two MK5 isoforms in liver tissue, MK5‐L and MK5‐S, which arise from alternative splicing of exon 6. Notably, exon 6 encodes the kinase active domain; its inclusion in MK5‐L, but not in MK5‐S, underlies the isoform‐specific functional differences we observed. Analysis of the OncoSplicing data indicates that high MK5‐L is associated with poorer survival in patients with HCC. Functional assays confirmed that only MK5‐L promotes the malignant phenotypes, consistent with its intact kinase activity. Although both isoforms bind GSK3β, only MK5‐L phosphorylates it. We propose that while the GSK3β interaction interface lies outside exon 6, the catalytic activity conferred by this exon is essential for phosphorylation. The precise structural determinants of this interaction warrant further investigation.

Antisense oligonucleotide (ASO) technology has emerged as a powerful therapeutic strategy for modulating gene expression by targeting RNA, showing promise for a range of diseases [27]. ASOs are under clinical investigation for oncology applications. For example, ASOs targeting WBP11 alter MCM7 splicing to suppress the proliferation and metastasis of ovarian cancer cells [41], and those that switch PKM splicing from the cancer‐associated PKM2 subtype to the PKM1 subtype inhibit the Warburg effect and tumor growth [42]. In line with this strategy, we designed an ASO targeting FXR1, which effectively suppressed FXR1 expression, shifted MK5 splicing toward the MK5‐S isoform, thereby inhibiting the malignant progression of HCC. These results position FXR1 as a promising therapeutic target and support the further development of ASO‐based strategies for HCC, pending improvements in the specificity and delivery efficiency.

In summary, this study highlights the functional importance of FXR1 in mediating HCC progression by regulating MK5 splicing. Increased FXR1 expression promotes the production of the long MK5 isoform, which facilitates HCC cell proliferation and metastasis by activating GSK3β (Figure 7). Collectively, our findings identify FXR1 and MK5‐L as potential prognostic biomarkers and reveal a splicing‐dependent mechanism that may be therapeutically targeted for HCC treatment.

FIGURE 7.

Schematic diagram of the mechanism by which FXR1 mediates the progression of hepatocellular carcinoma.

Author Contributions

Yutong Li: conceptualization, investigation, methodology, writing – original draft. Jin Xiang: data curation, investigation, methodology. Bin Cheng: data curation, validation. Chenhao Li: methodology, formal analysis. Kai Wang: conceptualization, project administration, supervision. Ni Tang: conceptualization, project administration, supervision. Luyi Huang: conceptualization, project administration, supervision, resources.

Funding

This study was supported by the National Natural Science Foundation of China (82303236, J.X.; 82304288, L.H.).

Ethics Statement

This study was approved by the Institutional Review Board of The Second Affiliated Hospital of Chongqing Medical University (No. 299/2025) for human research, and by the Institutional Animal Care and Use Committee of Chongqing Medical University (IACUC‐CQMU‐2024‐0978) for animal research. Informed consent was obtained from all human participants.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Doc S1. Materials and Methods.

Figure S1: Splicing alterations in HCC and prognostic evaluation of FXR1.

Figure S2: FXR1 promotes the proliferation and migration of hepatocellular carcinoma cells.

Figure S3: FXR1 exhibits strong correlation with the MK5 gene in the MAPK pathway.

Figure S4: MK5‐L promotes HCC cells proliferation and invasion in vitro.

Figure S5: FXR1 participates in the regulation of the Wnt/β‐catenin pathway.

Figure S6: ASO‐mediated FXR1 knockdown suppresses the proliferation and metastasis of HCC cells.

Table S1: Antibodies are used in this study.

Table S2: Primer sequences are used in this study.