DDX39B facilitates the malignant progression of hepatocellular carcinoma via activation of SREBP1-mediated de novo lipid synthesis

The detailed molecular mechanisms of aberrant lipid metabolism in HCC remain unclear. Herein, we focused on the potential role of DDX39B in aberrant lipogenesis and malignant development in HCC. DDX39B expression in HCC and para-cancer tissues was measured by immunohistochemistry. CCK-8, colony formation and Transwell assays were utilized to detect HCC cell proliferation, migration and invasion in vitro. Oil red O and Nile red staining and triglyceride and cholesterol detection were used to measure lipogenesis. Coimmunoprecipitation was used to detect interactions between DDX39B and SREBP1. Immunofluorescence assays were performed to investigate the impact of DDX39B on SREBP1 nuclear translocation. A luciferase assay was used to explore the transcriptional activity of SREBP1. The subcutaneous and orthotopic xenograft models in nude mice were generated to verify the contribution of the DDX39B/SREBP1 axis to tumor growth, lung metastasis and lipid synthesis in vivo. DDX39B is upregulated in HCC tissues and predicts a worse prognosis. Upregulated DDX39B contributes to the proliferation, metastasis and lipogenesis of HCC cells. Mechanistically, DDX39B directly interacts with SREBP1, and silencing DDX39B impairs the stabilization of the SREBP1 protein through FBXW7-mediated ubiquitination and degradation of SREBP1. Furthermore, DDX39B deficiency decreases the nuclear translocation and activation of SREBP1 and transcription of SREBP1 downstream genes, resulting in reduced lipid accumulation. Our study reveals a novel mechanism by which DDX39B facilitates the malignant progression of HCC via activation of SREBP1-mediated de novo lipogenesis, implicating DDX39B as both a potential predictor of recurrence and prognosis and a promising therapeutic target.


Introduction
Primary liver cancer, including 75%-85% hepatocellular carcinoma (HCC) and 10%-15% intrahepatic cholangiocarcinoma, has an increasing incidence and is the third leading cause of cancer death worldwide [1].More than half of HCC patients are diagnosed at an advanced stage due to the underutilization of surveillance and early detection, and the expected survival of these patients is less than one year [2,3].Over the past decade, several clinical trials have shown that combinations of multitarget tyrosine kinase inhibitors and immune checkpoint inhibitors exhibit promising efficacy for advanced stage HCC [4,5].The initiation and development of HCC are multistep and complex processes accompanied by the accumulation of genetic and epigenetic alterations, such as the most frequent driver mutations of AXIN1, 1 3 CTNNB1, TERT promoter, and TP53 [6].Moreover, metastasis and recurrence remain serious challenges to improve the prognosis of HCC patients.Thus, a deep understanding of the underlying mechanisms during HCC malignant progression will provide an opportunity to develop biomarker-driven early diagnosis and therapeutic strategies for HCC patients.
Metabolic reprogramming provides sufficient intermediates and fuels for tumor cell survival and proliferation, which is considered one of the remarkable characteristics of cancer [7].Lipids, including phospholipids, glycolipids, sterols and acylglycerols, are major components of biological membranes and are involved in intracellular signal transduction in physiological and pathological conditions.Fatty acids are an important source for lipid synthesis.Cancer cells can acquire fatty acids through the uptake of lipoproteins from endolysosomes or free fatty acids from the persistent lipolysis of adipose tissue.Moreover, the activation of de novo fatty acid synthesis, which utilizes nonlipid substrates such as glucose, glutamine and acetate, facilitates cancer cells to be more independent from externally supplied lipids [8,9].Accumulating evidence has reported that numerous genes involved in lipogenesis are upregulated in HCC cells, suggesting that abnormal lipid metabolism is a crucial factor for the development of HCC.For instance, USP30 deubiquitinated and stabilized ATP citrate lyase (ACLY, an enzyme that catalyzes the cleavage of citrate into acetyl-CoA), leading to the upregulation of ACLY, which promoted lipid synthesis to support hepatocarcinogenesis [10].The mRNA expressions of fatty acid synthase (FASN) and acetyl-CoA carboxylase (ACC) were markedly elevated in HCC samples compared with surrounding noncancerous liver tissue [11].In contrast, genetic or pharmacological restriction of FASN reduced the growth of HCC cells in both in vitro and in vivo models [12].The lncRNA LINC00958 enhanced the expression of hepatoma-derived growth factor by sponging miR-3619-5p, thereby stimulating HCC lipogenesis and progression [13].Nevertheless, the regulation of the molecular network regarding lipid metabolism in HCC initiation and development is not fully understood.
DExD-box helicase 39B (DDX39B) is an evolutionarily conserved DEAD-box helicase that was originally recognized to mediate ATP hydrolysis during pre-mRNA splicing [14].DDX39B has been demonstrated to maintain genomic stability and telomere length, while DDX39B depletion causes DNA injury and sensitizes cells to various DNA-damaging chemotherapeutic drugs, including cisplatin, camptothecin, and mitomycin C [15,16].The single nucleotide polymorphism of the DDX39B promoter is associated with several diseases, such as Alzheimer's disease, rheumatoid arthritis and malaria [17][18][19].Recent studies have expanded the function of DDX39B in tumorigenesis.For instance, DDX39B triggered the malignant progression of colorectal cancer by enhancing the protein stability and nuclear translocation of PKM2 [20].DDX39B promoted fork restarting and resistance to gemcitabine via interaction with retinol saturase in pancreatic ductal adenocarcinoma [21].In addition, the SUMOylation of DDX39B accelerated the nuclear export of circNCOR1, impeding the suppressive function of circNCOR1 on the activation of TGFβ signaling and lymph node metastasis of bladder cancer [22].Nevertheless, the underlying mechanism of DDX39B in lipogenesis and the development of HCC remains unknown.
In this study, we identify that DDX39B is upregulated in HCC tissues and cells and that HCC patients with high DDX39B expression possess worse relapse-free and overall survival than those with low DDX39B expression.DDX39B deficiency inhibits the proliferation, metastasis and lipogenesis of HCC cells in vivo and in vitro.Mechanistically, our data demonstrate that silencing DDX39B destabilizes the SREBP1 protein by enhancing FBXW7-mediated ubiquitination and degradation of SREBP1.Meanwhile, DDX39B knockdown restrains the transcriptional activity of SREBP1, leading to the decreased expression of lipogenesis-related genes and impaired lipid accumulation in HCC.Moreover, we find a positive correlation between DDX39B and SREBP1 or SREBP1 target genes (ACLY, FASN and ACACA) in HCC tissues.In summary, our results substantiate that DDX39B promotes lipogenesis reprogramming and malignant progression of HCC through stabilization and activation of SREBP1, implicating DDX39B as both a potential predictor of recurrence and prognosis and a promising therapeutic target in HCC.

Patient specimens and tissue microarrays
Commercial human paired HCC tissue microarrays (LivH180Su07) were purchased from Shanghai Outdo Biotech Company (Shanghai, China).Immunohistochemistry (IHC) of HCC tissue microarrays was performed by Shanghai Outdo Biotech Company.Informed consent was obtained for experimentation with human subjects, and the study was approved by the Ethics Committee of Shanghai Outdo Biotech Company.The survival data and clinical characteristics of all samples were obtained from company websites and are listed in Table 1.

Cell cultures and reagents
The human cell lines HepG2, SK-Hep1, MHCC97H, HUH7, JHH7, LO2 and HEK-293T were purchased from the Cell Bank of the Committee on Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China).Cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco, USA) containing 10% fetal bovine serum (FBS) (Gibco, USA) and 100 units/ml penicillin and streptomycin in a humidified incubator at 37 °C containing 5% CO 2. Primary human hepatocytes (PHHs) were purchased from bioIVT (USA) and cultured in hepatocyte medium containing 10% FBS.The reagents or resources used in this study are listed in Supplementary Table S1.

Lentivirus transfection and stable cell line establishment
Lentiviral-based short hairpin RNAs (shRNAs) targeting DDX39B (target sequence sh#1: TAG ACA TCT CCT CCT ACA T; target sequence sh#2: CCG CAA GTT CAT GCA AGA T) were incorporated into the lentiviral vector pLV-EGFP: T2A: Puro-U6 from Vector Builder Company (Guangzhou, China).SREBF1 cDNA (NM_004176.5) or luciferase gene fragments were incorporated into the lentiviral vectors pLV-Hygro-CMV or pLV-Noe-CMV, respectively.The cells were infected for 48 h and subsequently screened for 2 weeks by the indicated antibiotics to obtain stable cell lines.

Quantitative real-time PCR
Total RNA was extracted using RNAiso plus reagent (Takara, Dalian, China).Reverse transcription was performed by HiScript II Q Select RT SuperMix for qPCR (Vazyme, Jiangsu, China).Afterwards, 2 × SYBR Green qPCR Master Mix (Bimake, USA) was utilized to perform real-time PCR as previously described [23].The 2 −△△Ct algorithm was used to quantify relative gene expression.Sequences of primers are listed in Supplementary Table S2.

Western blot and coimmunoprecipitation (co-IP) assays
Western blot assays were performed according to the standard procedure as described previously [24].For co-IP analysis, 3 µg of the indicated antibody was bound to 30 µl of protein A/G beads (MCE, USA) and incubated with cell lysates overnight at 4 °C.For Flag-tagged proteins, an anti-FLAG M2 bead affinity gel (Sigma, USA) was applied.After washing the beads 5 times with TBS buffer, protein samples were collected after adding 2 × SDS loading buffer and boiling for 10 min and then subjected to western blot assay.The antibodies used are listed in Supplementary Table S1.

Cell proliferation and clone formation assay
Stable cells were plated into 96-well plates at a density of 1.5 × 10 3 cells/well.Twenty-four hours later, cell viabilities were measured by adding 10% CCK-8 (Cell Counting Kit-8, Topscience, Shanghai, China) at the indicated time points.The absorbance at 450 nm was read after incubation in the dark for 2 h.For the clone formation assay, 600 indicated cells were seeded into 24-well plates, and the culture medium was changed every three days.Ten days later, the cells were washed twice with PBS, fixed with 4% paraformaldehyde and then stained with 0.5% crystal violet.

Transwell assay
Transwell chambers with or without Matrigel (Corning, USA) were used to measure the invasive and migratory abilities of HCC cells, respectively.The indicated cells (2 × 10 4 for invasion and 1.5 × 10 4 for migration) were seeded in the upper chamber cultured with DMEM, while the lower chamber was filled with DMEM supplemented with 10% FBS.The chambers were washed with PBS, fixed with 4% paraformaldehyde for 10 min and stained with crystal violet.Subsequently, nonmigrated cells were carefully removed with a cotton swab.The migrated cells were captured and counted at 100 × magnification under an inverted phase contrast microscope (Nikon, Japan).

Immunohistochemistry (IHC)
Xenograft tumor tissues were fixed in 4% paraformaldehyde, embedded in paraffin, and then sectioned into 5 µm thick sections.Immunohistochemistry assays were conducted following standard protocols as described previously [25].The primary antibodies used in IHC are listed in Supplementary Table S1.Staining intensity was scored on a scale of 0 (negative), 1 (weak), 2 (moderate), and 3 (strong).0-100% was used to depict the percentage of stained cells.The protein expressions in tissues were scored based upon proportion multiplied intensity scores.The score calculated for indicated protein expression was on a scale of 0-300%.

Oil red O staining
HCC cells were seeded into 24-well plates precoated with sterile cover slips and cultured overnight.Then, the cells were washed with fresh PBS and fixed with 4% formaldehyde.Subsequently, the cells were stained with oil red O solution (Sangon Biotech, Shanghai, China) for 1 h at room temperature.After staining, the cells were washed with PBS three times and rinsed with 60% isopropanol for several seconds to remove nonspecific dye.The nuclei were then stained with hematoxylin staining solution (Beyotime, Shanghai, China).Frozen sections from mouse xenograft tumors and livers of the orthotopic transplantation model were also stained with oil red O solutions as described above.Average cellular staining intensity was analyzed and quantified by ImageJ software as previously described [26], and average intensity was normalized to cell number.

Triglyceride and cholesterol detection
Cellular and tumor triglyceride and cholesterol contents were measured by a tissue triglyceride (TG) content assay kit and a tissue total cholesterol (TC) content assay kit from Applygen Technologies Inc. (Beijing, China) according to the manufacturer's recommended protocols.The concentrations of proteins were detected using a BCA Protein Assay Kit (Sangon Biotech, Shanghai, China), and the cellular contents of TG and TC were normalized to the protein concentration.

Nile red staining
HCC cells were cultured in 24-well plates, washed with fresh PBS, fixed with 4% paraformaldehyde, and then stained with Nile red dye (MCE, USA) at a final concentration of 10 µg/ml.DAPI (Sangon Biotech, Shanghai, China) was used to counterstain the nucleus.The images were captured using a fluorescence microscope (DM4B, Leica, Germany).The cellular fluorescence intensity was analyzed and quantified by ImageJ software as previously described [27].The average fluorescence intensity was normalized to the cell number.

D-Glucose-13 C 6 -labeled metabolic flux analysis
DDX39B-deficiency HepG2 cells were cultured in complete medium containing D-glucose-13 C 6 (3.5 g/L) for 24 h.HepG2 cells were collected and lipids were extracted according to the protocols as previous described [28].Detection and metabolic flux analysis were performed by APExBIO Technology LLC. (Shanghai China).

Luciferase assay
The DNA fragment of the FASN promoter containing SREBP1 binding sites was cloned into the pGL3-basic plasmid and termed pGL3-FASN-promoter.A total of 2 × 10 4 stable HepG2 cells were seeded into 24-well plates, and then pGL3-FASN-promoter containing firefly luciferase and pRL-TK containing Renilla luciferase were cotransfected into HepG2 cells by Lipofectamine 3000 (Invitrogen, USA).After transfection for 48 h, the cells were lysed, and luciferase activities were determined following the manufacturer's protocol (Dual-Luciferase Reporter Assay System, Promega, USA).Renilla luciferase activities were used for internal normalization.

Chromatin immunoprecipitation (ChIP)
ChIP was performed using a SimpleChIP® Enzymatic Chromatin IP Kit (CST, USA) according to the manufacturer's instructions.In brief, HCC cells cultured in 15 cm dishes were cross-linked, quenched, washed and collected in PBS.
Cell pellets in the tube were then lysed, and the resulting extracts containing chromatin were sonicated into fragments.Diluted chromatin immunoprecipitation lysates were incubated with the indicated antibodies as well as protein G magnetic beads at 4 °C overnight.qPCR was used to analyze the bound DNA, and the primers are listed in Supplementary Table S2.

Animal experiments
All experimental studies were performed with approval from the Institutional Animal Care and Use Committee of West China Hospital Sichuan University.Six-week-old male nude mice were purchased from HFK BIOSCIENCE (Beijing, China) and randomly divided into four groups.All mice were fed standard rodent forage and maintained under specific-pathogen-free conditions at an ambient temperature of 23 °C and 12 h/12 h light/dark cycle.In xenograft studies, 6 × 10 6 HepG2 stable cells or 4 × 10 6 MHCC97H stable cells resuspended in PBS were injected subcutaneously into nude mice (six mice per group).Tumor volumes were measured every 3 days with a caliper and calculated using the formula: V (mm 3 ) = 1/2 × (length × width 2 ).All mice were sacrificed at 22 days after injection.Subsequently, the tumors were completely detached from the mice, and then volumes and weights were measured.To further mimic in situ growth and metastasis, 6 × 10 6 HepG2 cells or 4 × 10 6 MHCC97H stable cells stably labeled with luciferase were injected into the liver parenchyma of nude mice (five mice per group).Forty days later live luciferase imaging was performed using IVIS® Lumina III (PerkinElmer, USA).After sacrificing the mice, liver and lung samples were isolated and subjected to ex vivo imaging, H&E and oil red O staining.

Statistical analysis
All experiments were performed independently with at least three biological replicates.GraphPad Prism software (GraphPad, CA, USA) and the SPSS software package (SPSS, IL, USA) were employed to analyze the statistics.Student's t test was used to calculate differences between the two groups, and one-way ANOVA was employed to assess multiple groups.Correlations of gene or protein expression were determined with the Pearson coefficient.P values < 0.05 were considered statistically significant.* indicates P value < 0.05, ** indicates P value < 0.01, *** indicates P value < 0.001.

DDX39B is upregulated in HCC and predicts worse prognosis
To investigate the functional role of DDX39B in HCC, we first analyzed the level of DDX39B mRNA in HCC tissues based on the TCGA database.Higher mRNA expression of DDX39B was observed in HCC tissues than in normal liver tissues (Fig. 1A).Next, we evaluated the protein expression of DDX39B in HCC and paired para-cancer tissues.We found that DDX39B protein levels were significantly increased in HCC tissues compared with matched para-cancer tissues (Fig. 1B-C).A consistent result was validated by the Clinical Proteomic Tumor Analysis Consortium database (Fig. 1D).We further examined the mRNA and protein levels of DDX39B in HCC and nontumorigenic liver cells.Our data showed that both the mRNA and protein expressions of DDX39B were elevated in five HCC cell lines (SK-Hep1, HepG2, MHCC97H, Huh7 and JHH-7) compared with the normal liver cells LO2 and PHHs (Fig. 1E and F).Remarkably, HCC patients with high DDX39B expression (above the median) exhibited poorer overall survival (Fig. 1G) and relapse-free survival (Fig. 1H) than those with low DDX39B expression (below the median).Similar results were observed from the GEPIA (Gene Expression Profiling Interactive Analysis) database (Fig. 1I and J).Increased DDX39B protein expression was positively associated with recurrence and advanced pathological grade in HCC patients (Table 1).Multivariate analysis revealed that DDX39B was an independent prognostic factor related to poor overall survival and relapse-free survival in HCC patients (Table 2).These findings reveal that DDX39B is upregulated in HCC tissues and predicts worse prognosis in HCC patients.

DDX39B suppression inhibits the proliferation, migration and invasion of HCC cells in vitro
To explore the underlying effects of DDX39B in HCC development, two DDX39B-specific short hairpin RNAs (shDDX39B#1 and shDDX39B#2) were delivered by lentivirus into three HCC cell lines (HepG2, SK-Hep1 and MHCC97H) with high DDX39B expression.The silencing efficiency of DDX39B was confirmed by qPCR and western blot analysis (Fig. 2A-B and Supplementary Fig. S1A-S1B).Knockdown of DDX39B significantly reduced the viability of HepG2, SK-Hep1 and MHCC97H cells (Fig. 2C and Supplementary Fig. S1C).Consistently, decreased colony numbers were observed in DDX39B-deficient HCC cells compared with control cells (Fig. 2D and Supplementary Fig. S1D).
Transwell assays revealed that DDX39B silencing diminished the migratory and invasive abilities of HCC cells (Fig. 2E and Supplementary Fig. S1E).These results reveal that DDX39B positively regulates the growth and motility of HCC cells.

DDX39B deficiency attenuates lipid accumulation in HCC cells
To clarify the potential downstream signaling of DDX39B in HCC, DDX39B-silenced HepG2 cells were used to perform transcriptome sequencing.The top 25 differentially expressed genes (Supplementary Table S3) were enriched in several biological pathways, including cholesterol biosynthetic processes (ACLY, INSIG1 and LSS) and fatty acid biosynthetic processes (ACLY and FASN), according to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis (Fig. 3A).Because dysregulation of lipid metabolism contributes to the malignant progression of HCC, we subsequently determined the effect of DDX39B on lipid biosynthesis in HCC cells.qPCR analysis demonstrated decreased mRNA levels of ACLY, FASN, LSS and INSIG1 in DDX39Bdeficient HCC cells (Fig. 3B and Supplementary Fig. S2A), which is in agreement with the results of RNA sequencing.Oil red O staining revealed that DDX39B silencing reduced the accumulation of neutral lipids in HepG2, SK-Hep1 and MHCC97H cells (Fig. 3C and Supplementary Fig. S2B).Similar results were obtained from Nile red staining in HCC cells (Fig. 3D and Supplementary Fig. S2C).Moreover, knockdown of DDX39B resulted in decreased contents of cellular triglycerides (Fig. 3E and Supplementary Fig. S2D) and cholesterols (Fig. 3F and Supplementary Fig. S2E) in HCC cells.
To further unveil the role of DDX39B in de novo lipogenesis process, we preformed metabolic flux analysis to monitor 13 C-labeled unsaturated and saturated fatty acids derived from D-glucose-13 C 6 .As shown in Fig. 3G, diminished contents of newly synthetic fatty acids were observed in DDX39Binsufficient HepG2 cells compared with shNC HepG2 cells.Taken together, our results uncover that DDX39B enhances lipid biosynthesis in HCC cells.

DDX39B stabilized the SREBP1 protein by preventing FBXW7-mediated ubiquitination and degradation of SREBP1
To elucidate the molecular mechanism of DDX39B-mediated lipogenesis in HCC cells, we performed immunoprecipitation with an anti-DDX39B antibody and mass spectrometry analysis to characterize the specific binding proteins of DDX39B (Fig. 4A).We identified various proteins that interacted with DDX39B, including Vimentin, Myosin-9, Plectin, Prelamin-A/C and SREBP1 (Supplementary Table S4).Considering that SREBP1 is a definite transcription factor that switches on de novo lipogenesis in cancer cells, we focused on SREBP1 for further study.Reciprocal immunoprecipitation analysis confirmed that endogenous DDX39B was able to bind to both the precursor and nuclear forms of the SREBP1 protein in HepG2, SK-Hep1 and MHCC97H cells (Fig. 4B-C and Supplementary Fig. S3A).A consistent result was obtained in HEK-293T cells cotransfected with Flag-DDX39B and Myc-SREBP1 plasmids (Supplementary Fig. S3B).Moreover, immunofluorescence staining validated the colocalization of endogenous DDX39B and SREBP1 in HCC cells (Fig. 4D and Supplementary Fig. S3C).To further investigate the association of DDX39B with SREBP1 in living HCC cells, a bimolecular fluorescence complementation assay was performed as described in our previous study [20].As shown in Fig. 4E, reconstituted fluorophore signals were observed in HEK-293T and HepG2 cells when cotransfected with VN173-SREBP1 and VC155-DDX39B plasmids, supporting that DDX39B directly interacted with SREBP1.His pull-down assay also demonstrated a direct interaction between DDX39B and nSREBP1 in vitro (Fig. 4F).We subsequently assessed the impact of DDX39B on SREBP1 expression in HCC cells.We found that DDX39B was unable to alter the SREBP1 mRNA level in HepG2, SK-Hep1 and MHCC97H cells (Fig. 4G and Supplementary Fig. S3D).In contrast, downregulation of DDX39B decreased the expression of precursor and mature forms of SREBP1 proteins in HCC cells (Fig. 4H and Supplementary Fig. S3E), indicating that DDX39B may influence the stability of the SREBP1 protein.Cycloheximide chase experiments revealed that DDX39B deficiency accelerated the degradation of the precursor and nuclear forms of the SREBP1 protein in HepG2, SK-Hep1 and MHCC97H cells (Fig. 4I and Supplementary Fig. S3F).Notably, DDX39B knockdownmediated reduced expression of SREBP1 protein was largely blocked by the proteasome inhibitor MG132, while the lysosome inhibitor chloroquine failed to prevent the degradation of SREBP1 protein (Fig. 4J).Moreover, DDX39B suppression increased the polyubiquitination of the SREBP1 protein in HepG2 and SK-Hep1 cells (Fig. 4K).These observations indicated that DDX39B interacted with SREBP1 and enhanced SREBP1 protein expression via restriction of SREBP1 ubiquitination and degradation.A previous study demonstrated that FBXW7 serves as a specific E3 ubiquitin ligase of SREBP1 [29].Therefore, we next examined whether DDX39B affected the interactions between SREBP1 and FBXW7 in HCC cells.
Although no significant changes in FBXW7 protein expression were observed in DDX39B-deficient HepG2 and SK-Hep1 cells, DDX39B silencing strengthened the association of SREBP1 and FBXW7 (Fig. 4L).Additionally, exogenous reintroduction of FBXW7 greatly augmented the polyubiquitination of the SREBP1 protein in HCC cells, which was partially interrupted by DDX39B overexpression (Fig. 4M).Furthermore, DDX39B protein levels were positively correlated with SREBP1 expression in clinical samples of HCC (Fig. 4N).Altogether, these data imply that DDX39B suppression alleviated its association with the SREBP1 protein, leading to the increased interaction between SREBP1 and FBXW7 and subsequently facilitating FBXW7-mediated ubiquitination and degradation of SREBP1 in HCC cells.

DDX39B silencing mitigates the nuclear translocation and transcriptional activity of SREBP1
Given the evidence that nuclear translocation of SREBP1 is essential for its transcription factor function to transactivate the transcription of lipid metabolism-related genes [30], we next wondered whether DDX39B contributed to the nuclear accumulation of SREBP1 in HCC cells.Immunofluorescence assays showed that both the cytoplasmic and nuclear distribution of SREBP1 was decreased in the absence of DDX39B (Fig. 5A and Supplementary Fig. S4A).As FASN is a well-known target gene of SREBP1, a fragment of the FASN promoter containing a sterol regulatory element was cloned into a luciferase reporter vector to evaluate the impact of DDX39B on the transcriptional activity of SREBP1 in HCC cells.As shown in Fig. 5B, DDX39B silencing diminished the basic luciferase activity of the FASN promoter.Importantly, ectopic expression of SREBP1 markedly reversed the DDX39B suppression-mediated reduction in the transcriptional activity of the FASN promoter (Fig. 5B).ChIP assays verified a physiological association between the DDX39B protein and a DNA fragment of the FASN promoter in HepG2 and SK-Hep1 cells (Fig. 5C).In agreement with the above results, the mRNA and protein levels of SREBP1 target genes (including ACLY, FASN, ACACA, and SCD1) were considerably decreased in DDX39B-deficient HCC cells (Fig. 5D-E and Supplementary Fig. S4B-S4C).Moreover, DDX39B levels were positively associated with ACLY, FASN, and ACACA expression in HCC patients from the TCGA database (Fig. 5F).These data indicate that DDX39B downregulation inhibited the transcriptional activity of SREBP1 and the expression of SREBP1 target lipogenic enzymes by attenuating the nuclear accumulation of SREBP1.
1245 DDX39B facilitates the malignant progression of hepatocellular carcinoma via activation… 1 3

SREBP1 is crucial for DDX39B-mediated proliferation, migration, invasion and lipid synthesis in HCC cells
To further explore whether SREBP1 is involved in DDX39B-triggered growth and metastasis, we reintroduced SREBP1 by using lentivirus infection in DDX39B-silenced HCC cells.The overexpression of SREBP1 and knockdown of DDX39B in HepG2, SK-Hep1 and MHCC97H cells were confirmed by qPCR and western blot assays (Fig. 6A-B and Supplementary Fig. S5A-S5B).SREBP1 overexpression effectively restored the decreased expression of lipogenic enzymes (ACLY, FASN, ACACA and SCD1) caused by DDX39B suppression in HCC cells (Fig. 6A-B and Supplementary Fig. S5A-S5B).Moreover, DDX39B deficiency resulted in reduced cell viability and colony formation of HCC cells, which was reversed by upregulation of SREBP1 (Fig. 6C-D and Supplementary Fig. S5C-S5D).Consistently, enforced SREBP1 expression rescued the decreased migration and invasion of HCC cells in the absence of DDX39B (Fig. 6E and Supplementary Fig. S5E).Similar results were obtained by oil red O (Fig. 6F and Supplementary Fig. S5F) and Nile red staining (Fig. 6G and Supplementary Fig. S5G).In addition, DDX39B-deficient HCC cells expressing SREBP1 displayed elevated levels of intracellular triglycerides and cholesterol compared with control cells (Fig. 6H-I and Supplementary Fig. S5H-S5I).Collectively, our results reveal that SREBP1 is indispensable for DDX39B-induced growth, metastasis and lipid accumulation in HCC cells.

The DDX39B/SREBP1 axis modulates tumor growth, metastasis and lipid biosynthesis in HCC cells in vivo
To further verify whether the DDX39B/SREBP1 axis contributes to the malignant development of HCC in vivo, the indicated HepG2 and MHCC97H cells were injected into nude mice to establish a subcutaneous xenograft model.The volume and weight of subcutaneous tumors were pronouncedly inhibited in the DDX39B-deficient group compared with the shNC group, whereas reintroduction of SREBP1 significantly accelerated the tumor growth of HepG2 and MHCC97H cells in the absence of DDX39B (Fig. 7A-C and Supplementary Fig. S6A-6C).DDX39B knockdown impaired lipid accumulation in subcutaneous tumors, while overexpression of SREBP1 reversed this phenomenon, as determined by oil red O staining (Fig. 7D and Supplementary Fig. S6D) and triglyceride and cholesterol measurements (Fig. 7E-F).Additionally, DDX39B suppression diminished the expression of the proliferation marker Ki67 and lipogenic enzymes (FASN, SCD1 and ACLY) in subcutaneous tumors that were restored by SREBP1 overexpression (Fig. 7G).Subsequently, we established an orthotopic liver cancer xenograft model to evaluate the effect of DDX39B/ SREBP1 signaling on tumor growth and metastasis using the indicated HepG2 and MHCC97H cells.We observed weaker bioluminescence signals in the shDDX39B group than in the shNC group (Fig. 8A-D and Supplementary Fig. S6E-S6H).Notably, the reintroduction of SREBP1 facilitated tumor growth in situ compared with DDX39B-silenced HepG2 and MHCC97H cells (Fig. 8A-D, Supplementary Fig. S6E-S6H).Moreover, upregulated SREBP1 effectively rescued the decreased lung metastatic nodules caused by DDX39B silencing (Fig. 8C and E, Supplementary Fig. S6G and S6I).Consistent results were obtained in H&E staining of liver and lung tissues (Fig. 8F-G and Supplementary Fig. S6J-S6K).
We also examined lipid accumulation in liver tissues through oil red O staining.Our results showed that DDX39B suppression resulted in a decreased content of neutral lipids; however, enforced SREBP1 expression promoted intracellular lipid accumulation in DDX39B-deficient cells (Fig. 8H and Supplementary Fig. S6L).Importantly, the concentrations of AST and ALT in the serum of mice were decreased in the shDDX39B group compared with the shNC group, whereas overexpression of SREBP1 partially restored their content in the absence of DDX39B (Fig. 8I-J and Supplementary Fig. S6M).Collectively, these data imply that DDX39B regulates the growth, metastasis and lipid accumulation of HCC cells via SREBP1.

Discussion
The malignant development of HCC is a complicated biological process, including hyperproliferation of tumor cells in situ, remodeling and degradation of the extracellular matrix, stromal and vascular invasion, resistance to anoikis and distant metastasis [31,32].Identifying novel and effective biomarkers will improve current therapeutic strategies for HCC and prolong the survival of patients with HCC.Herein, we showed that DDX39B was upregulated in HCC tissues compared with paired para-cancer tissues.Elevated DDX39B levels were positively associated with pathological grade and recurrence in HCC patients.Knockdown of DDX39B repressed the proliferation and metastasis of HCC cells in vitro and in vivo.We performed transcriptome sequencing to investigate the potential downstream signaling of DDX39B in HCC.TFRC, which encodes a cell surface receptor required for intracellular iron uptake and maintenance of iron homeostasis, was decreased in the absence of DDX39B.Consistent with our observations, a previous study showed that TFRC was upregulated in human HCC tissue and contributed to the initiation of liver carcinogenesis [33].We also found upregulated expression of CDKN1B and PDCD4 in DDX39B-deficient HepG2 cells.CDKN1B, which induces cell cycle arrest at G1 phase by binding to and preventing the activation of the cyclin E/CDK2 or cyclin D/CDK4 complex, exhibits a tumorsuppressive function in hepatocarcinogenesis and is negatively associated with the prognosis of HCC patients [34].Moreover, PDCD4 inhibited the proliferative and metastatic potential of HCC cells via the repression of MTA1 gene expression [35].These studies combined with our findings indicate an important function of DDX39B in HCC development.SREBPs (SREBP1 and SREBP2) have a basic helixloop-helix-leucine zipper structure and belong to the nuclear transcription factor family.Newly synthesized SREBPs, as precursor proteins, are retained in the endoplasmic reticulum by the SREBP cleavage-activating protein (SCAP)/insulininduced gene (Insig) complex.When intracellular lipid/cholesterol levels are insufficient, SCAP disassociates from Insig and assists the translocation of SREBPs to the Golgi apparatus, where SREBPs are activated by a two-step proteolytic process.Subsequently, the active form of SREBPs enters the nucleus and binds to promoter sequences containing sterol response elements or E-box motifs to promote transcription of target genes [36,37].Generally, SREBP1 activates fatty acid and cholesterol biosynthesis, while SREBP-2 is relatively specific to orchestrate cholesterol uptake and synthesis [30,38].Transgenic mice overexpressing a mature form of SREBP1 develop a progressive and massive fatty liver accompanied by obvious elevations in the transcription of genes involved in fatty acid and cholesterol synthesis, such as FASN, SCD1, HMGCS, and HMGCR [39].Interestingly, the liver tissue from transgenic mice with constitutive activation of SREBP2 exhibited a preferential induction of cholesterol biosynthesis-related enzymes and a lesser increase in fatty acid synthesis-related genes [40].These observations suggest a functional overlap between SREBP1 and SREBP2.In the present study, we identified SREBP1, but not SREBP2, as a potential binding protein of DDX39B via immunoprecipitation and mass spectrometry analysis.We also demonstrated that DDX39B directly bound to and stabilized both the precursor and nuclear forms of SREBP1.The decreased expression of SREBP1 protein caused by DDX39B deficiency was considerably impeded in the presence of the proteasome inhibitor MG132 but not the lysosome inhibitor chloroquine, suggesting that the degradation of SREBP1 protein is modulated by the ubiquitin-proteasome pathway.Consistently, DDX39B disrupted the association of precursor and nuclear forms of SREBP1 with its E3 ubiquitin ligase FBXW7, leading to the nuclear accumulation of SREBP1 protein and subsequent transcription of SREBP1 target genes involved in lipogenesis.In line with our observations, recent studies have reported that knockdown of FBXW7 enhanced the expression of premature and mature SREBP1 [41,42], suggesting that FBXW7 could modulate both the full-length and nuclear forms of SREBP1.Moreover, DDX39B mRNA expression was positively correlated with the mRNA levels of the SREBP1 target genes ACLY, ACACA and FASN in HCC tissues from the TCGA database.We also found downregulation of LSS and INSIG1 in DDX39B-insufficient HCC cells.In agreement with our data, increased expression of LSS and INSIG1 was observed in the livers of SREBP1 transgenic mice [43], although the detailed molecular mechanism has not been disclosed.Our previous study demonstrated that DDX39B augmented aerobic glycolysis in colorectal cancer by enhancing the protein kinase and cotranscription factor functions of nuclear PKM2 [20].Importantly, PKM2 has been illustrated to interact with and phosphorylate the nuclear form of SREBP1, which facilitates lipogenic gene expression, lipid biosynthesis and cell proliferation in HCC [44].These studies further supported a critical role of DDX39B in the regulation of the crosstalk between glucose and lipid metabolism in cancer cells.Hence, the development of small molecule inhibitors targeting DDX39B will contribute to providing an effective therapeutic strategy for HCC.
DDX39A (also known as URH49, which is encoded by the DDX39A gene located at 19p13.12) is a close homolog of DDX39B (also known as UAP56, which is encoded by the DDX39B gene located at 6p21.33).The DDX39A protein shares 90% identity and 96% similarity of amino acid sequence with DDX39B, suggesting similar or redundant functions of these two proteins.For instance, suppression of either DDX39A or DDX39B caused a speckled pattern of poly(A) + RNA accumulation in the nucleus [45].Both DDX39A and DDX39B were able to promote the generation of the androgen receptor splice variant AR-V7 in prostate cancer cells [46].However, a previous study reported that DDX39A participated in the assembly of the AREX complex that contributed to mRNA export, while DDX39B formed the TREX complex that was involved in the integration of diverse mRNA processing steps [47].As a result, DDX39A inhibition led to chromosome arm resolution defects and collapse of cytokinesis, whereas DDX39B silencing triggered sister chromatid cohesion defects and mitotic delay.Moreover, depletion of DDX39A or DDX39B provoked the enrichment of short or long circRNAs in the nucleus, respectively [48].These studies indicate that DDX39A and DDX39B exert nonoverlapping functions in distinct cellular processes.Zhang and colleagues recently confirmed that DDX39A assisted HCC growth and metastasis through activation of the Wnt/β-catenin pathway [49].Notably, constitutively active SREBP1 has been shown to activate Wnt/β-catenin signaling in esophageal squamous cell carcinoma [50].Based on our finding that DDX39B increased SREBP1 expression and accelerated the nuclear translocation of SREBP1 in HCC cells, we inferred a crosstalk between DDX39A and DDX39B in the regulation of HCC progression.
To the best of our knowledge, no studies have reported the prognostic value of DDX39B for HCC patients.In the present study, our data showed that upregulated expression of DDX39B protein correlated with reduced overall survival and relapse-free survival in HCC patients.Similar results were obtained according to DDX39B mRNA levels from the TCGA database.Although potentially curative therapies (such as surgical resection and liver transplantation) provide better clinical outcomes for early-stage HCC, more than half of HCC patients present with tumor recurrence within five years [51].Adjuvant or neoadjuvant therapy offers clinical benefit for several cancers, including breast cancer, colon cancer and melanoma [52][53][54].However, many systemic treatments, such as hepatic-artery chemotherapy and the tyrosine kinase inhibitor sorafenib, have failed to improve relapse-free or overall survival in HCC [55,56].Thus, the identification of HCC patients with a high risk of recurrence will allow clinicians to provide appropriate strategies to improve the prognosis after resection of HCC.Here, we revealed that DDX39B was an independent factor for predicting poor overall survival and relapse-free survival in patients with HCC.Given the functions of DDX39B in HCC proliferation, invasion, and metastasis, we considered DDX39B to be a potential and promising predictor for monitoring the aggressive development and recurrence of HCC.Recent evidence has reported that the small-molecule inhibitor CCT018159 selectively suppresses the ATPase activity of DDX39B and appears to have a strong antiviral function against Kaposi's sarcoma-associated herpesvirus [57], which encourages us to evaluate the inhibitory effect of CCT018159 on HCC in the future.
In conclusion, our present study provided the first evidence that elevated DDX39B expression was observed in HCC tissues compared with paired para-cancer tissues and correlated with worse overall and relapse-free survival in patients with HCC.We demonstrated that DDX39B directly interacted with and enhanced the stability of the SREBP1 protein by restraining FBXW7-mediated ubiquitination and degradation of SREBP1 in HCC cells, leading to nuclear translocation and activation of SREBP1 and subsequent transcription of lipogenesis-related genes.Enforced expression of SREBP1 effectively restored the decreased growth, metastasis and lipid accumulation caused by DDX39B deficiency in HCC cells.Taken together, our findings highlight the crucial clinical significance of DDX39B in the recurrence and prognosis of HCC and unveil a novel mechanism by which DDX39B contributes to the malignant progression of HCC via activation of SREBP1-mediated de novo lipid synthesis.Disruption of the DDX39B/SREBP1 signaling axis may be a feasible therapeutic strategy for HCC.
GYF, KQM and WQJ prepared Table 1 and Table 2; FTY and LK wrote the manuscript; LK and LP revised the manuscript.LP supervised this work.All authors reviewed the manuscript.

Fig. 1
Fig. 1 Upregulated DDX39B correlates with poor prognosis in human HCC.(A) The relative mRNA level of DDX39B in HCC and normal samples from The Cancer Genome Atlas (TCGA) database.(B-C) The protein expression of DDX39B in HCC and paired paracancer tissues was measured by immunohistochemistry staining.(B) Representative images are shown, and (C) the relative staining intensities of DDX39B are shown.(D) The level of DDX39B protein in HCC and normal samples from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) database.(E) qRT-PCR analysis of DDX39B mRNA expression in LO2, PHH, SK-Hep1, HepG2,

Fig. 2 Fig. 3
Fig. 2 DDX39B knockdown inhibits HCC proliferation, migration and invasion in vitro.(A and B) qRT-PCR (A) and western blotting (B) were used to measure the interference efficiency of DDX39B by two specific shRNAs in HCC cells.(C and D) CCK-8 (C) and colony formation (D) assays were utilized to measure the viability and

Fig. 4
Fig. 4 DDX39B binds to and impedes the degradation of SREBP1 protein.(A) Flow diagram of IP-MS in HepG2 cells.(B and C) Immunoprecipitation assays were used to determine the endogenous interactions between DDX39B and SREBP1 in HCC cells.(D) The localization of DDX39B (green) and SREBP1 (red) was assessed in HCC cells by confocal microscopy.DAPI (blue) shows the nucleus.Scale bars: 50 μm.(E) HEK-293T and HepG2 cells were transfected with pBiFC-VC155-DDX39B and/or pBiFC-VN173-SREBP1, and the fluorescence signals were visualized.(F) A His pull-down assay was carried out to evaluate the direct interaction between DDX39B and SREBP1.(G and H) The mRNA (G) and protein (H) levels of SREBP1 in DDX39B-deficient HCC cells were detected by qRT-PCR and western blotting.(I) The effect of DDX39B on the degradation of SREBP1 protein in HCC cells was assessed by cycloheximide chase experiment.(J) The impact of MG132 or chloroquine on DDX39B-mediated SREBP1 protein levels in HCC cells.(K) Ubiquitination of the SREBP1 protein in DDX39B-deficient HCC cells was assessed by immunoprecipitation assay.(L) The effect of DDX39B on the association of the SREBP1 protein with its E3 ligase FBXW7 in HCC cells was tested by immunoprecipitation assay.(M) DDX39B-overexpressing HCC cells were cotransfected with Myc-FBXW7 and Flag-ub plasmids.The ubiquitination of SREBP1 protein was detected by immunoprecipitation assay.(N) Representative images of the expression of DDX39B and SREBP1 in HCC tumor tissues.Scale bars: 50 μm.The correlation between DDX39B and SREBP1 protein expression was analyzed in HCC tissues.LE, long exposure.P, precursor; N, mature/nuclear form.*** p < 0.001, ns represents not significant ◂

Fig. 5 Fig. 6
Fig. 5 DDX39B deficiency impairs the nuclear translocation and transcriptional activity of SREBP1.(A) The subcellular localization of SREBP1 in DDX39B-knockdown HepG2 and SK-Hep1 cells was visualized by immunofluorescence assay.Scale bars: 50 μm.(B) A luciferase assay was conducted to assess the effect of DDX39B silencing and ectopic SREBP1 expression on the luciferase activity of the FASN promoter in HCC cells.(C) The binding of DDX39B protein to the FASN promoter containing a sterol regulatory element was examined by ChIP assay.(D and E) The mRNA (D) and protein (E) levels of ACLY, FASN, ACACA and SCD1 were detected by qRT-PCR and western blotting in HCC cells.(F) The correlation between the mRNA levels of DDX39B and ACLY, FASN, ACACA or SCD1 from the TCGA database.**p < 0.01, ***p < 0.001

Fig. 7
Fig. 7 The DDX39B/SREBP1 axis modulates HCC cell growth and lipid accumulation in a subcutaneous xenograft model.(A) Images of resected subcutaneous tumors from the indicated HepG2 cell-infected groups in nude mice.(B and C) Tumor sizes (B) and tumor weights (C) were measured.(D) The formations of neutral lipids were stained with oil red O from frozen sections of subcutaneous tumors.Scale bars: 20 μm.(E and F) The levels of triglycerides (E) and cholesterols (F) in subcutaneous tumors were measured by a tissue triglyceride assay kit or a tissue total cholesterol kit, respectively.(G) The expression of DDX39B, SREBP1, KI67, FASN, SCD1 and ACLY in subcutaneous tumors was detected by IHC, and the scores of staining intensities are shown below.Scale bars: 20 μm.**p < 0.01, ***p < 0.001

Fig. 8
Fig. 8 The DDX39B/SREBP1 axis modulates tumor growth, metastasis and lipid metabolism in an orthotopic xenograft model.(A and B) The indicated HepG2 cells were injected into the liver parenchyma.Forty days later, the bioluminescent signal of the orthotopic tumor was captured and (A) quantified (B).(C) Liver and lung tissues were isolated, and representative bioluminescent images of the indicated tissues were obtained.(D and E) The bioluminescent signals of

Table 2
Univariate and multivariate analyses of OS and RFS in HCC patients Statistical significance was analyzed using the Cox proportional hazards model.*P < 0.05, **P < 0.01, ***P < 0.001