The splicing regulatory factor hnRNPU is a novel transcriptional target of c‐Myc in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is the most common liver cancer with high mortality. Here, we found that hnRNPU is overexpressed in HCC tissues and is correlated with the poor prognosis of HCC patients. Besides, hnRNPU is of high significance in regulating the proliferation, apoptosis, self‐renewal, and tumorigenic potential of HCC cells. Mechanismly, c‐Myc regulates hnRNPU expression at the transcriptional level, and meanwhile, hnRNPU stabilizes the mRNA of c‐MYC. We found that the hnRNPU and c‐Myc regulatory loop exerts a synergistic effect on the proliferation and self‐renewal of HCC, and promotes the HCC progression. Taken together, hnRNPU functions as a novel transcriptional target of c‐Myc and promotes HCC progression, which may become a promising target for the treatment of c‐Myc‐driven HCC.

(Received 27 May 2020, revised 6 September 2020, accepted 7 September 2020, available online 7 November 2020) doi: 10.1002/1873-3468.13943 Edited by Angel Nebreda Hepatocellular carcinoma (HCC) is the most common liver cancer with high mortality. Here, we found that hnRNPU is overexpressed in HCC tissues and is correlated with the poor prognosis of HCC patients. Besides, hnRNPU is of high significance in regulating the proliferation, apoptosis, self-renewal, and tumorigenic potential of HCC cells. Mechanismly, c-Myc regulates hnRNPU expression at the transcriptional level, and meanwhile, hnRNPU stabilizes the mRNA of c-MYC. We found that the hnRNPU and c-Myc regulatory loop exerts a synergistic effect on the proliferation and self-renewal of HCC, and promotes the HCC progression. Taken together, hnRNPU functions as a novel transcriptional target of c-Myc and promotes HCC progression, which may become a promising target for the treatment of c-Mycdriven HCC.
Keywords: c-Myc; hepatocellular carcinoma; hnRNPU; regulatory loop; therapeutic target Hepatocellular carcinoma (HCC), the major type of liver cancer, has been demonstrated to be the second leading cause of cancer death worldwide and with a trend of increasing incidence [1]. Insufficient understanding of the precise molecular events underlying HCC progression and the resultant lack of effective therapy are main reasons for the poor prognosis of HCC [2]. Despite the inspiring progression in clinical trials of immune checkpoint inhibitors in HCC [3,4], sorafenib and lenvatinib are still the only two first-line options for advanced HCC currently [5,6]. Hence, identifying functional regulating factors and therapeutic targets in HCC is of high significance to develop effective and novel anti-HCC therapeutics.
Alternative splicing (AS) is a post-transcriptional process involving most of the eukaryotic genes. The alternative mRNA transcripts encode structurally or functionally disparate protein isoforms, so as to diversify the proteome of cells [7]. While in cancer cells, normal AS regulation is disturbed and results in cancer-specific RNA transcripts profile, which further benefit cancer cell proliferating, migrating, or escaping from cell death [8,9]. In recent studies, large-scale analyses in multiple solid tumor types have revealed that these cancer-specific splicing patterns are attributed to the aberrantly regulated splicing factors, including mutations, copy number variations, or expression alterations of the splicing regulatory genes [10]. These findings lead to increasing interests in interfering with the splicing factors in cancer development [11,12], with the aim to provide novel therapeutic strategies for cancer treatment.
With regard to HCC, aberrant AS events are also common and contribute to cancer hallmarks in liver [13,14]. Moreover, multiple AS regulators including SRSF2 [15], SRSF3 [16], hnRNPA2 [17], and PTBP3 [18] have been reported to participate in HCC progression. While the role of hnRNPU, an important splicing regulatory factor belongs to the heterogeneous nuclear ribonucleoprotein (hnRNP) family has not been well investigated in HCC yet. In current study, we found that hnRNPU was overexpressed in advanced HCC patients and predicted patients' unfavorable prognosis. To delineate the functional significance of the upregulated hnRNPU, we investigated the role of hnRNPU in the proliferation, apoptosis, and self-renewal (both in vitro and in vivo) of HCC cells. Furthermore, we explored for the upstream regulational factors and downstream effect factors involved in the function of hnRNPU. Our results demonstrated that hnRNPU might be a novel transcriptional target of c-Myc, and the stability of c-MYC mRNA is maintained by hnRNPU. Our research led to a deeper understanding of the interaction between hnRNPU and c-Myc underlying HCC development and provided potential therapeutic target in c-Myc-driven HCC.

Materials and methods
Cell culture HCC cell lines used in this study were Huh7.5.1, Hep3B, and PLC cells. The Huh7.5.1 cells were maintained in our laboratory. Hep3B cells were purchased from the National experimental cell resource sharing platform (Beijing, China). The PLC cells were purchased from ATCC (Manassas, VA, USA). The Hep3B cells were maintained in MEM with 10% FBS (Excell Biology, Shanghai, China), and other HCC cells were maintained in Dulbecco's Modified Eagle's medium (DMEM) with 10% FBS; all cell lines were cultured at 37°C with 5% CO 2 . The reagents used in this assay were listed in Table S2.

Western blot
The western blot (WB) assay was performed as previously described [19]. Briefly, HCC cell pellets were collected and lysed in RIPA lysis buffer with protease inhibitors for 30 min at 4°C. The cell lysates were centrifuged at 12 000 g for 15 min. The supernatants of lysates were separated by SDS/PAGE and transferred to PVDF membranes (Millipore, Billerica, MA, USA). The blots were blocked by 5% BSA/TBST for 1 h and then incubated with primary antibodies at 4°C overnight. Next, the blots were blocked with HRP-conjugated secondary antibodies for 1 h at room temperature. Immunoreactivity was analyzed by Immobilon Western Chemiluminescent HRP Substrate (Bio-Rad, Hercules, CA, USA). The antibodies used in this assay were listed in Table S3.

Colony and tumorsphere formation assay
The stable transfected HCC cells were seeded on 12-well plates (2000 cells per well) or six-well plates (4000 cells per well). After culturing for two weeks, images of clonal subpopulation formed in each group were captured and analyzed. The tumorsphere formation assay was performed as previously described [19]. The reagents used in this assay were listed in Table S2.

Animal studies
Mice were purchased from the Vital River Laboratories (Beijing, China) and manipulated according to protocols approved by the Beijing Medical Experimental Animal Care Commission. For limiting dilution assay, 10 6 /10 5 /10 4 stable transfected scramble and hnRNPU-knockdown PLC cells were subcutaneously injected into left/right side of the flanks of NOD/SCID mice (n = 6), respectively. About 4 weeks later, mice were sacrificed and data were collected.

Stable transfected HCC construction
The vector pLV[Exp]-EGFP-T2A-Puro-EF1A-HNRNPU was purchased from Vector Builder (Guangzhou, China, #VB171103-1072syx). The vector for c-Myc overexpression is pCDH-RFP-c-Myc (Addgene, Watertown, MA, USA, #102626). The plasmid for hnRNPU mRNA interference was constructed by inserting the short hairpin RNA sequences into the pSicoR-GFP lentiviral vector (Jacks Lab, MIT Center for Cancer Research, Cambridge, MA, USA). The lentivirus packaging and cell transfection were performed as previously described [19]. Three days after transfection, the stable transfected HCC cells (GFP/RFP positive) were enriched by FACS. The primers used in this assay were listed in Table S1.

Quantitative RT-PCR
Total RNA of HCC cells was isolated by TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Reverse transcription and quantitative RT-PCR (qRT-PCR) in this study were performed as previously described [19]. The quantification of gene expression was determined based on the DDC t method and normalized to the reference gene GAPDH. The primers used in this assay were listed in Table S1.

Immunohistochemistry
Briefly, the fixed tissues were embedded and sectioned at 4 lm thickness, and antigen retrieval was performed by a pressure cooker in 0.01 M citrate buffer for 15-20 min. The primary antibodies of hnRNPU and c-Myc were incubated at 4°C overnight, followed incubation with HRP-conjugated secondary antibodies. Then, immune-detection was performed using DAB kits according to the manufacturer's instructions. The antibodies used in this study were listed in Table S3.

Dual-luciferase assay
The 293T cells were seeded in 24-well plates and were divided into five groups with four-well repeats. The details are as follows: negative control (NC) group cells were cotransfected with 300 ng of pCDH-NC vectors and 400 ng of pGL3-control (Promega, Madison, WI, USA) vectors, the rest four groups were cotransfected with 300 ng of pCDH-c-Myc vectors (Addgene, #102626) and 400 ng of pGL3-hnRNPU/hnRNPA1-promoter vectors. After 36-h transfection, the last three groups were treated with reagents (DMSO\10058-F4\10074-G5, respectively) for 9 h. The pRL-Renilla-luciferase vector (Promega) was used to normalize the transfection efficiency in each groups. The luciferase activity was detected by the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's protocol.

Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) was performed using an EZ-ChIP TM ChIP Kit (Millipore), according to the manufacturer's instructions. Briefly, 10 7 cells were fixed with 1% formaldehyde and then lysed and sonicated by six cycles of 10 s on/20 s off in 30% AMPL with a Cole-Parmer CPX 130 Ultrasonic Processor (Cole-Parmer, Vernon Hills, IL, USA). The cell lysates were precleared by protein A/G agarose beads at 4°C for 1 h. And c-Myc antibody or rabbit isotype IgG was added to the precleared cell lysates and rotated at 4°C overnight. Protein A/G agarose beads were then incubated with the protein/DNA complexes for 1 h at 4°C with rotation. The cross-links between the protein and DNA fragments were then reversed to free the c-Myc-binding DNA fragments which were then analyzed by PCR. The reagents and kits used in this assay were listed in Table S2.

Cell viability analysis
For cell viability analysis, 2 9 10 3 HCC cells were seeded per well in 96-well plates and cultured DMEM with 10% FBS. The proliferation of each group was detected by CCK-8 at 24/48/72 h after plating. For the doxorubicin (Dox) and 5-fluorouracil (5-FU) chemoresistance assay, 3 9 10 3 -4 9 10 3 cells per well were seeded in 96-well plates. After 24 h, the medium was replaced with medium containing the indicated concentrations of the chemotherapeutic agents. The absorbance at 450 nm was measured at the indicated time point by SpectraMax M2e (Molecular Devices, Silicon Valley, CA, USA).

Tissue microarray and samples
The tissue microarray was purchased from Zhuohao Pharmaceutical Technology Co., Ltd (LVC1606, Shanghai, China). The tissue chip contains detailed pathological and survival information of HCC patients. The tissues from 15 HCC patients after tumor resection were obtained with informed consent, and all the experiments were approved by the Fifth Medical Center of Chinese PLA General Hospital and the Ethics Committee at the Stem Cell and Regenerative Medicine laboratory. The pathological and survival information of the patients was listed in Data S1.

mRNA decay analyses
Hepatocellular carcinoma cells were transfected with hnRNPU siRNAs for 72 h and harvested as control group or treated with 5 lM actinomycin D (HY-17559; MedChem Express, Monmouth Junction, NJ, USA) for mRNA decay and harvested at indicated time points. Total RNA extraction and reverse transcription as described previously [19], and the c-MYC mRNA levels were determined by the DDC t method using RPLP0 as internal control for three times. The primers used in this assay were listed in Table S1.

Statistical analysis
The immunohistochemistry (IHC) staining intensity was analyzed using IMAGE-PRO PLUS 6.0 software (Media Cybernetics, Inc., Rockville, MD, USA). Statistical graphs were generated by GRAPHPAD PRISM 8 (GraphPad Software, Inc., San Diego, CA, USA). Correlation between hnRNPU expression and pathological grade was determined by Fisher's exact test. The statistical significance of differences between two groups was determined by t-test. The differences among multiple groups were statistically analyzed by two-way ANOVA test. All data are presented as the Mean AE SD. *P < 0.05 was considered statistically significant.

hnRNPU is upregulated in HCC and related to clinical stage and prognosis of HCC patients
To investigate the clinical significance of hnRNPU in HCC progression, we detected the expression profile of hnRNPU in human liver tissues. We first analyzed the mRNA expression of hnRNPU in patient samples by using HCCDB [20], including HCC tissues (n = 268), paratumor tissues (n = 243), cirrhotic livers (n = 40), and healthy livers (n = 6). Comparing with the healthy livers, the expression level of hnRNPU is upregulated in cirrhotic tissues and gets higher in HCC samples. Moreover, the HCC tissues exhibit significantly elevated hnRNPU expression than the paratumor tissues, although hnRNPU is also slightly upregulated in paratumors than healthy livers (Fig. 1A). Next, we examined the protein levels of hnRNPU by performing IHC staining of an HCC tissue array (n = 58). We found a strong nucleic staining of hnRNPU in tumor tissues compared with a weak or negative staining in the paired nontumor tissues (Fig. 1B). The quantitative analysis of the hnRNPU staining intensity further revealed a significant upregulation of hnRNPU expression in tumors than in adjacent normal tissues (Fig. 1C). Moreover, the higher expression levels of hnRNPU are correlated with the higher TNM stages (Fig. 1D). Patients with high hnRNPU expression in liver cancer tend to have shorter survival time (Fig. 1E), and this is consistent with the result derived from the Human Protein Atlas database (Fig. 1F). The result indicated a potential prognostic value of hnRNPU for HCC. Taken together, we demonstrated that the elevated hnRNPU expression might play important roles in HCC progression.
hnRNPU regulates the proliferation, drug resistance, and chemotherapy-induced apoptosis of HCC cells As deregulated cell proliferation and apoptosis have been regarded as the hallmarks of cancer cells [21], we wondered whether hnRNPU participated in the regulation of these processes. Two stable hnRNPU-knockdown HCC cell lines were established and validated by WB analysis (Fig. 2A). We found that hnRNPU knockdown significantly slowed HCC proliferation (Fig. 2B), and also reduced the HCC resistance to chemotherapeutic agents, Dox, and 5-FU (Fig. 2C). Moreover, comparing to the low expression level of the apoptosis marker, cleaved caspase-3, induced by Dox or 5-FU in HCC cells, hnRNPU knockdown resulted in a further accumulation of cleaved caspase-3 ( Fig. 2D, Fig. S1A), suggesting that hnRNPU knockdown weakened the apoptosis resistance of HCC cells to chemotherapies. The above observations were also repeated in hnRNPU-overexpressed HCC cells (Fig. 2E). We found that hnRNPU overexpression led to increased proliferation (Fig. 2F), drug resistance (Fig. 2G), and chemotherapy-induced apoptosis resistance (Fig. 2H, Fig. S1B) in HCC cells.
hnRNPU regulates the self-renewal and tumorigenic potential of HCC cells Considering that self-renewal of HCC cells is a key process in tumor initiation and progression [22], we further investigated the regulatory role of hnRNPU in the self-renewal of HCC cells. We found that hnRNPU knockdown significantly decreased the colony-forming ability of HCC cells compared with the scrambles (Fig. 3A). CD133, a surface marker of liver cancer stem cells, was also downregulated after hnRNPU knockdown in HCC cells (Fig. 3B), while hnRNPU overexpression increased CD133 expression rate of HCC cells (Fig. S2). Through the 3D tumorsphere culture, a method to enrich and maintain stemlike cells in vitro, we further confirmed that hnRNPU played an important role in supporting the tumorsphere forming potential of HCC cells (Fig. 3C). These results indicated that hnRNPU regulated the self-renewal of HCC cells in vitro. Besides, in the limiting dilution assay, the gold standard for defining self-renewal ability of cells in vivo, we transplanted 10 6 /10 5 / 10 4 PLC-scramble cells and the same number of PLC-hnRNPU-knockdown cells into left and right flanks of NOD/SCID mice, respectively. After 4 weeks, we calculated the tumor incidence (Fig. 3D) and the frequency of tumor initiating cells was analyzed by the extreme limiting dilution analysis (ELDA) software [23] (Fig. 3E). The tumor initiating cell frequency in the scramble group and knockdown group was 1/318 005 and 1/1 089 636, respectively, suggesting that hnRNPU knockdown significantly reduced the tumorigenic potential of HCC cells. These results indicated that hnRNPU regulated the self-renewal of HCC cells, both in vitro and in vivo.
c-Myc regulates hnRNPU at transcriptional level, and hnRNPU stabilizes the mRNA of c-MYC in HCC cells Then, we wondered the reason of hnRNPU upregulation in the HCC progression. By analyzing the genetic alterations of hnRNPU in TCGA-LIHC samples, we noticed that only in a few cases the high hnRNPU mRNA expression might be attributed to the genome amplification, while in most of the cases showed no correlations (Fig. S3). Thus, we started to seek evidence from the transcriptional regulation of hnRNPU. Intriguingly, several c-Myc-binding sites, the enhancer box sequences (E-boxes) [24], were identifies in the promoter region of hnRNPU. Moreover, c-Myc has been widely uncovered to be related to the spliceosome and abnormal splicing events in cancers [25]. Splicing factors such as HNRNPA1, HNRNPA2, PTB [26], SRSF1 [27], and PRMT5 [28] have been reported to be the direct transcriptional targets of c-Myc. This led us to speculate whether hnRNPU would be a potential and novel transcriptional target of c-Myc in HCC. Thus, we performed the dual-luciferase reporter assay to test the transcriptional regulation of c-Myc on hnRNPU. The luciferase activity initiated by hnRNPU promoter (À2197 bp to À1664 bp, with three E-box motifs; Fig. 4A) was upregulated following pCDH-c-Myc transduction, while further downregulated by c-Myc inhibitors, 10058-F4 [29], and 10074-G5 [30] (These inhibitors block c-Myc-Max dimerization and inhibit c-Myc target gene expression). The hnRNPA1 (a known c-Myc target gene) promoter was used as positive control. This result indicated that c-Myc might play a positive role on regulating hnRNPU transcription (Fig. 4B). Further, the ChIP assay validated the binding of c-Myc in the promoter region of hnRNPU (Fig. 4C). In addition, c-Myc inhibitors downregulated the hnRNPU expression at both mRNA level (Fig. 4D) and protein level (Fig. 4E), while c-Myc overexpression promoted the expression of hnRNPU (Fig. 4F). All the above results indicated that c-Myc was a positive transcriptional regulator of hnRNPU in HCC cells.
Besides, it was reported that hnRNPU stabilizes the c-MYC mRNA through associating with IGF2BP1 in an RNA-dependent fashion [31], and we wondered whether hnRNPU regulates c-MYC mRNA stability in HCC cells. We knocked down the hnRNPU expression and measured the levels of c-MYC mRNA at indicated time point after actinomycin D treatment, which act as a transcriptional inhibitor. The results showed that c-MYC mRNA was decreased faster (4 h after actinomycin D treatment) in hnRNPU-knockdown Hep3B cells than hnRNPU-normal Hep3B cells, and the decreasing levels were more significant in hnRNPU-knockdown HCC cells than control cells after 8-h actinomycin D treatment (Fig. 4G). Correspondingly, the protein levels of c-Myc were also reduced in hnRNPUknockdown HCC cells but increased in hnRNPU-overexpressed HCC cells (Fig. 4H, Fig. S1). It was suggested that hnRNPU could also stabilize c-MYC mRNA and regulate its protein expression.

hnRNPU positively correlates with c-Myc in HCC samples
Moreover, in clinical HCC patient samples, a positive correlation between c-Myc and hnRNPU was identified at mRNA level in the TCGA-LIHC cohort (Fig. 5A). This observation still holds true at the protein level in 15 primary HCC samples (Fig. 5B), as the representative colocalized IHC staining pictures show (Fig. 5C, Tumor No. 1 and Tumor No. 2). Interestingly, a micro-metastatic foci with both c-Myc and hnRNPU overexpression were identified in the negative staining background of the paratumor tissue of Patient No. 2 (Fig. 5C, Paratumor No. 2). This finding further confirmed the close correlation of the two proteins in HCC.

hnRNPU overexpression partially abolishes the effects of c-Myc inhibitors in HCC cells
We considered that c-Myc is a strong oncogene which regulates numbers of downstream pathways to maintain cancer cell survival and self-renewal [32,33], and the mutual regulation between c-Myc and hnRNPU. We wondered what role hnRNPU plays on the c-Myc-mediated biological roles in HCC cells. Therefore, we used c-Myc inhibitors to block c-Mycmediated biological functions, and detected the effect of hnRNPU overexpression on c-Myc function inhibition. By introducing exogenous hnRNPU in HCC cells, we found that hnRNPU overexpression partially abolished the growth inhibitory effects of 10058-F4 and 10074-G5 (Fig. 6A). The colony-forming ability in hnRNPU-overexpressed cells also showed higher resistance to c-Myc inhibitors (Fig. 6B). Also, the number and volume of tumorspheres were significantly reduced by 10058-F4 or 10074-G5, while hnRNPU overexpression partially reversed this selfrenewal inhibition caused by the two c-Myc inhibitors (Fig. 6C). Not only that, in HCC cells treated with 10058-F4 or 10074-G5, cleaved caspase-3 expression was induced in 48 h, yet hnRNPU overexpression suppressed the induction of cleaved caspase-3 (Fig. 6D, Fig. S4), suggesting that hnRNPU overexpression strongly rescued HCC cells from apoptosis induced by c-Myc inhibitors.
All the results suggested that hnRNPU overexpression in HCC cells could partially rescue the proliferation and self-renewal inhibitory effects as well as apoptosis induction of c-Myc inhibitors, which suggested an important role of hnRNPU in c-Myc mediating biological roles in HCC cells.

c-Myc overexpression partially reverses the effects of hnRNPU knockdown in HCC cells
Since hnRNPU takes part in c-Myc-mediated biological roles in HCC cells, will c-Myc also influences the biological roles of hnRNPU in HCC cells? Therefore, we also explored the effect of c-Myc overexpression on the biological function of hnRNPU-knockdown HCC cells. HCC cells were transfected with c-Myc overexpression vector (pCDH-RFP-c-Myc), hnRNPU-knockdown vector (pScioR-GFP-sh-hnRNPU), and the corresponding control vector (pCDH-RFP-NC/pS-cioR-GFP-scramble). The transfected cells were enriched by flow cytometry (Fig. S5) and verified by WB (Fig. 7A). We found that c-Myc overexpression could also partially reverse the inhibitory effects on proliferation (Fig. 7B), colony forming (Fig. 7C), and tumorsphere forming (Fig. 7D) caused by hnRNPU knockdown. The results suggested that c-Myc was partly involved in the hnRNPU-mediated biological roles in HCC cells.

Discussion
hnRNPU, a 120 kDa multifunctional protein, regulates the pre-mRNA splicing process through direct binding with target genes, or protein/protein interaction-mediated splicing decisions [34,35]. In addition, the actin-hnRNPU complex is a critical regulator in the initial phase of transcription activation in eukaryotic cells [36]. In HCC cells, hnRNPU-related biological process and downstream targets have also been investigated, including that hnRNPU interacts with TIPUH1, a novel oncogenic protein overexpressed in HCC [37]. The long noncoding RNA H19 is also bound to hnRNPU, inhibiting RNA polymerase IImediated transcription by disrupting the hnRNPUactin complex [38]. In our study, we found that the splicing regulator hnRNPU was overexpressed in HCC tissues and was related to the poor prognosis of HCC patients. The elevated hnRNPU participated in the proliferation, chemotherapy-induced apoptosis, and self-renewal regulation of HCC cells. In order to better understand the functional mechanism of hnRNPU in HCC, we explored its possible upstream regulatory factors and downstream effect factors. Similar to splicing factors HNRNPA1 and HNRNPA2 [26], our results showed that hnRNPU promoter contained several c-Myc-binding sites (E-boxes) and was regulated by c-Myc at the transcriptional level in HCC cells. In addition, hnRNPU exerts a global control of AS for a series of mRNA by regulating U2 snRNP maturation [39], and some splicing variants have an effect on the HCC progression. Such as, the binding of hnRNPU to the pre-Rac1 promotes the inclusion of exon 3b and yielding an oncogenic splicing variant, Rac1b [40]. Interestingly, we were intrigued by the observation that hnRNPU stabilizes the c-MYC mRNA through associating with IGF2BP1 in an RNA-dependent fashion [31], and our results also verified that hnRNPU participated in stabilizing c-MYC mRNA in HCC cells. This is inspiring that c-Myc regulates the transcription of hnRNPU, hnRNPU in turn stabilizes the c-MYC mRNA. It might indicate a regulatory loop between hnRNPU and c-Myc in HCC cells. Knockdown or overexpress one of them, the other one would have similar changes (Figs S1 and S4 and Fig. 4F).
The regulatory loop of hnRNPU and c-Myc is very important for the biological properties of HCC cells. HnRNPU and c-Myc not only regulates each other's expression levels, but also participates in the biological roles of HCC mediated by each other. In Fig. 6, exogenous overexpression of hnRNPU partially reversed the effects of c-Myc inhibitors in HCC cells. It was suggested that hnRNPU is an important effect target of c-Myc and participates in part of the c-Mycmediated biological effects (proliferation, self-renewal and so on). Similarly, in Fig. 7, exogenous overexpression of c-Myc partially reversed the effects of hnRNPU knockdown in HCC cells. This indicated that c-Myc is also partially involved in hnRNPU-mediated biological effects. However, inhibition of either of c-Myc or hnRNPU would lead to the termination of their regulatory loop, regardless of the other is hyperfunction or overexpression, the proliferation and self-renewal ability of HCC cells would be difficult to restore. When we blocked c-Myc function by inhibitors and overexpressed hnRNPU alone in HCC cells (Fig. 6) or overexpressed c-Myc alone in hnRNPUknocked down HCC cells (Fig. 7), the proliferation and self-renewal ability of treated HCC cells were significantly lower than that of NC HCC cells. This indicated that the hnRNPU-c-Myc regulatory loop exerts synergistic effects on the proliferation and self-renewal ability of HCC cells (Fig. 8A).
Many studies had reported that c-Myc overexpression is one of the most common drivers of human cancer, and regulates multiple target genes involved in important cell fate control process including proliferation, apoptosis, differentiation, cell-cycle regulation, etc [32,33]. In the initial stage of cancer, c-Myc plays an important role in proliferation and self-renewal of cancer cells [32]. With the development of HCC, c-Myc overexpression leads to the upregulation of hnRNPU expression, which in turn further promotes the c-Myc overexpression. As a result, the proliferation and self-renewal ability of HCC increased with the c-Myc and hnRNPU upregulation. The colocalization and correlation of hnRNPU and c-Myc in HCC tissues as well as their correlation with pathological grade also support this view. Besides, it was reported that many components of the spliceosome (including hnRNPU) might be therapeutic entry points for aggressive c-Myc-driven cancers [41]. In consistent with our results, it showed that c-Myc and hnRNPU regulatory loop might be of great importance for the proliferation and self-renewal of Mean AE SD, *P < 0.05, **P < 0.01 represents difference between NC vs U, and # P < 0.05, ## P < 0.01, represents difference between NCc-Myc inhibitor vs U-c-Myc inhibitor, determined by two-way ANOVA. (B) Colony-forming ability of NC or hnRNPU-overexpressed HCC cells treated with DMSO, 10058-F4 or 10074-G5. Each DMSO group colonies sum area were standardized as 100. Mean AE SD, ***P < 0.001, determined by Student's t-test. (C) Images of the tumorspheres formed by NC or hnRNPU-overexpressed HCC cells treated with DMSO, 10058-F4 or 10074-G5. Mean AE SD, **P < 0.01, ***P < 0.001, by Student's t-test. Bars, 1000 lm. (D) WB analysis of caspase-3 and cleaved caspase-3 expression before and after treatment of 10058-F4 or 10074-G5 in NC or hnRNPU-overexpressed (U) cells.
advanced c-Myc-driven HCC. This positive feedback loop of hnRNPU and c-Myc in HCC might be partially responsible for HCC progression and poor prognosis (Fig. 8B).
c-Myc is the most frequently amplified oncogene in various cancers. To identify the essential downstream effector or interacting protein for c-Myc, researchers screened for genes whose downregulation was lethal for c-Myc-activated cells. They found that c-Myc-hyperactivated cells were more sensitive to perturbations in spliceosome function or core spliceosomal genes [25], than c-Myc-normal cells. As c-Myc has been regarded as an undruggable but attractive therapeutic target in cancer [42], its dependency on splicing machinery makes the spliceosome-targeted therapies feasible in c-Myc-driven cancers [11]. Here, we demonstrated that hnRNPU might be a novel transcriptional target of c-Myc in HCC; more importantly, the positive feedback loop of hnRNPU and c-Myc is functionally essential for HCC proliferation and self-renewal. Thus, breaking the feedback loop or inhibiting hnRNPU expression through RNA interference,  Mean AE SD, **P < 0.01, determined by two-way ANOVA. (C) Colony-forming ability of NC/c-Myc overexpressed HCC cells stable transfected with pSicoR-scramble (scr)/sh-hnRNPU (sh-U) vector. Each NC-scramble group (NC-scr) colonies sum area was standardized as 100. Mean AE SD, **P < 0.01, determined by two-way ANOVA. (D) Tumorspheres formed by NC/c-Myc overexpressed HCC cells stable transfected with pSicoR-scramble (scr)/sh-hnRNPU (sh-U) vector. Mean AE SD, **P < 0.01, by two-way ANOVA. Bars, 1000 lm. neutralization antibodies or disturbing its function by kinase inhibitors, is promising therapeutic approaches for c-Myc-hyperactivated HCC patients with hnRNPU overexpression.
In summary, we reported that hnRNPU might function as a novel transcriptional target of c-Myc and their regulatory loop promoted HCC progression. hnRNPU might be a promising target for the treatment of HCC.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. hnRNPU and c-Myc blots of transfected HCC cells treated before and after chemotherapies. Fig. S2. FCM analysis of CD133 + cell populations rate in NC and hnRNPU overexpressed (hnRNPU) HCC cells.   .   Table S1. The primers used in the study. Table S2. The kits and reagents used in the study. Table S3. The antibodies used in the study. Data S1. Pathological data of HCC tissue chip (LVC1606).