A c-Myc-MicroRNA functional feedback loop affects hepatocarcinogenesis

Authors


  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the National Natural Science Foundation of China (81071622 and 81171896), the National Basic Research Program of China (973) project (2012CB967000), and the Ph.D. program foundation of the Ministry of Education of China (20110141110015).

Abstract

c-Myc (Myc) plays an important role in normal liver development and tumorigenesis. We show here that Myc is pathologically activated in and essential for promoting human hepatocellular carcinoma (HCC). Myc induces HCC through a novel, microRNA (miRNA)-mediated feedback loop comprised of miR-148a-5p, miR-363-3p, and ubiquitin-specific protease 28 (USP28). Myc directly binds to conserved regions in the promoters of the two miRNAs and represses their expression. miR-148a-5p directly targets and inhibits Myc, whereas miR-363-3p destabilizes Myc by directly targeting and inhibiting USP28. Inhibition of miR-148a-5p or miR-363-3p induces hepatocellular tumorigenesis by promoting G1 to S phase progression, whereas activation of them has the opposite effects. The Myc-miRNA feedback loop is dysregulated in human HCC. Conclusion: These results define miR-148a-5p and miR-363-3p as negative regulators of Myc, thus revealing their heretofore unappreciated roles in hepatocarcinogenesis. (HEPATOLOGY 2013;57:2378–2389)

Hepatocellular carcinoma (HCC) is among the most common human cancers and the third most frequent cause of cancer death.1 Risk factors for HCC include hepatitis B virus, hepatitis C virus, aflatoxin B1, heavy alcohol consumption, and vinyl chloride exposure.1-4 However the mechanisms underlying hepatocellular carcinogenesis are poorly understood, although metabolic fitness and innate levels of immunity are known to play critical roles at the early stages of HCC in mice.4 Multistage hepatocarcinogenesis is influenced by genetic and epigenetic changes as well as microenvironmental factors. Included among the former are mutation and/or inactivation of tumor suppressor genes such as TP53 and Rb and the activation of oncogenes such as Ras and c-Myc (hereafter Myc).1-7

Myc, a helix-loop-helix leucine zipper (HLH-ZIP) transcription factor, dimerizes with Max, another HLH-ZIP protein, and binds to E-box sequences to activate transcription of target genes or microRNAs (miRNAs).8 Myc also acts as a transcriptional repressor by interacting with and suppressing other transcription factors and by modulating chromatin status.8 Myc is a downstream effector of many signaling pathways, and its expression is tightly regulated by many factors, including miRNAs.8-10 Through a myriad of such downstream targets, Myc plays important roles in cell growth, survival, metabolism, and tumorigenesis.5-12 Myc is frequently amplified and overexpressed in many different human malignancies, including HCC.2, 3, 13, 14 Up-regulation of Myc and the reprogramming of transcription signature are critical steps in HCC progression in mice,15 and Myc is one of the critical genes activated in cancers believed to be caused by infection with HBX virus.16 Transforming growth factor-β1 and E2F1 may contribute to the promotion and progression of liver carcinogenesis in Myc transgenic mice.17, 18 However, precisely how Myc contributes to hepatocarcinogenesis at the molecular level has not been well characterized.

Here we report that Myc is pathologically activated in and essential for several of the phenotypes associated with human HCC. Contributing to hepatocellular tumorigenicity is Myc's repression of two miRNAs, miR-148a-5p and miR-363-3p, that comprise a negative feedback loop involving Myc itself and ubiquitin-specific protease 28 (USP28)19; Myc directly binds the conserved regions in the promoters of miR-148a-5p and miR-363-3p and represses their expression. These miRNAs function as tumor suppressors that promote cell cycle arrest and inhibit tumor growth. We also report that miR-148a-5p directly targets and inhibits Myc, whereas miR-363-3p destabilizes Myc indirectly by directly targeting and inhibiting USP28, which promotes the proteasome-mediated degradation of Myc protein. Finally, we show that this Myc-miRNA feedback loop is dysregulated in human HCC. These results help to clarify the regulatory mechanism by which Myc is overexpressed in this disease.

Abbreviations

DMEM, Dulbecco's modified Eagle's medium; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GFP, green fluorescent protein; HCC, hepatocellular carcinoma; HLH-ZIP, helix-loop-helix leucine zipper; IgG, immunoglobulin G; IP, immunoprecipitation; miRNA, microRNA; mRNA, messenger RNA; RPE, retinal pigmented epithelium; RT-PCR, reverse-transcription polymerase chain reaction; siRNA, small interfering RNA; USP28, ubiquitin-specific peptidase 28; UTR, untranslated region.

Materials and Methods

E-box Motif Searching, Homology Analysis, and Computational Prediction of miRNA Targets.

The preselection of promoter-associated E-boxes in CpG islands was performed using bioinformatics methods. Briefly, E-boxes were identified within 3 kb upstream and 1 kb downstream of predicated transcription start sites of all annotated human miRNA genes. The conservation of these E-boxes between human and mouse was analyzed using alignment software from the National Center for Biotechnology Information website.

Computational prediction of miRNA targets was performed using the online databases miRDB (www.mirdb.org), miRanda (www.miranda-im.org), miRwalk (www.ma.uni-heidelberg.de/apps/zmf/mirwalk/), and RNAhybrid (http:// bibiserv.techfak.uni-bielefeld.de/rnahybrid/). miRBase (www.mirbase.org/) was used to analyze miRNA information and CpG Island Searcher Program (http://cpgislands.usc.edu/) was used to analyze CpG island regions. The GenBank accession numbers of Myc and USP28 messenger RNA (mRNA) are NM_001706 and NM_020886, respectively.

Plasmids.

Human miRNA expression vectors were performed according to the protocols recommended by the manufacturer (BLOCK-iT Pol II miR RNAi Expression Vector Kits, Invitrogen). Human pre-miRNAs, including approximately 350 bp containing stem-loop structures, were polymerase chain reaction (PCR)-amplified from genomic DNA and cloned into BamHI and XhoI or BglII and SalI sites of pcDNA 6.2-GW/EmGFP-miR vector (Catalog no. K4936-00, Invitrogen). Human Myc or USP28 3′ untranslated region (UTR) fragments surrounding miR-148a-5p or miR-363-3p responsive elements were cloned into SalI and BamHI sites of pEGFP-C1 vector (Clontech Laboratories, Inc.) immediately downstream of green fluorescent protein (GFP) with stop codon TAA. pBabe-MycER plasmid was purchased from Addgene (Cambridge, MA) (Addgene plasmid 19128). All plasmid sequences were verified by direct sequencing. The sequences of all primers are provided in Supporting Table 1.

Table 1. The E-boxes of 21 miRNA Loci Are Conserved Between Human and Mouse
miRNAConserved E-BOXReported to be Regulated by MycFunctionTargetReference
  1. E-boxes from the proximal promoters of 21 of 94 human miRNA loci with promoter E-boxes are conserved among their murine orthologs. miRNAs from 10 of these 21 loci have been reported to be Myc-regulated. miRNAs from the other 11 loci have not been reported to be regulated by Myc. CACGTG(−890) means that an E-box, CACGTG, is located at 890 bp upstream from the transcription starting site.

  2. Abbreviation: TS, tumor suppressor.

hsa-let-7iCACGTG(−890)+TSMyc34
hsa-mir-129CACGTG(−2631)TSSox435
hsa-mir-148aCACGTG(−1328)TSDnmt3b36
hsa-mir-17CACATG(2171)/CATGTG(−1477)+OncogeneE2F132
hsa-mir-18aCATGTG(−1615)+OncogeneK-ras32
hsa-mir-193aCACGTG(−715)TSC-kit37
hsa-mir-193bCACGTG(−971)TSMcl-138
hsa-mir-196a-1CACGTG(−1066)TSHoxc839
hsa-mir-19aCATGTG(−1763)+OncogeneE2F132
hsa-mir-200cCATGTG(−2096)TSBMI140
hsa-mir-20bCACGTG(−2041)+OncogeneHIF-1α32
hsa-mir-23bCACATG(−1776/−1607/−1492)/CATGTG(−1490)+MetabolismGlutaminase41
hsa-mir-24-1CACATG(−2420/−2305)/CATGTG(−2303)OncogeneHNF4α42
hsa-mir-27bCACATG(−1844/−1729)/CATGTG(−1727)TSPPAR43
has-mir-29CACATG(−1912)+TSCDK644
hsa-mir-320aCACGTG(−267)TSNRP-145
hsa-mir-363CACGTG(−2466)   
hsa-mir-378CACGTG(−2006/−1835)+OncogeneTOB246
hsa-mir-615CACATG(−380)   
hsa-mir-92a-2CACGTG(−2306)+OncogeneDKK332
hsa-mir-9-3CACGTG(−32)+MetastasisCDH147

Cell Culture and Transfections.

Human liver tumor-derived cell lines used in this study—including HepG2, Bel-7402, FHCC98, and Huh-7—were cultured under standard cell culture conditions in Dulbecco's modified Eagle's medium (DMEM) (GIBCO, Life Technologies) supplemented with 10% fetal bovine serum (GIBCO), 1% L-glutamine, 1% penicillin-streptomycin, and 1% nonessential amino acids in a 5% CO2-humidified chamber. Primary human foreskin fibroblasts (American Type Culture Collection Manassas, VA) were grown as described.20 Human retinal pigmented epithelium (RPE) cells immortalized with hTERT were cultured in DMEM:F12 medium, whereas HEK293T cells were cultured as described for HCC. Normal human hepatocytes HL-7702 were grown in RPMI1640 (GIBCO) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin, maintained at 37°C and 5% CO2.

Small interfering RNA (siRNA), miRNA mimics, and miRNA inhibitors were purchased from RiboBio Co., Ltd. (Guangzhou, China). siRNA sequences 1 and 2 against USP28 were referenced from Zhang et al.21 (sequence 1, CUGCAUUCACCUUAUCAUU; sequence 2, UUGGUUUAGUGCUGUUAUU), siRNA sequence against Myc (AGACCUUCAUCAAA AACAUUU) was referenced from Napoli et al.,22 and siRNA against GFP was designed from RiboBio Co., Ltd. (Guangzhou, China). The inhibitor of Myc/Max dimerization 10058-F4 (sc-213577) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). For transfection, plasmids (1-5 μg), siRNAs (100 nM final concentration), miRNA mimics (50 nM final concentration), or miRNA inhibitors (100 nM final concentration) were transfected into appropriate cells using Lipofectamine 2000 (Invitrogen), 48 hours after transfection cell pellets were collected and subjected to RNA isolation and immunoblot analysis.

RNA Extraction.

RNA was extracted using TRIzol (Invitrogen) from the indicated cell lines according to the manufacturer's protocol. DNA contamination was removed with RNAse-free DNase I.

Stem-Loop Reverse-Transcription PCR (RT-PCR) and Real-Time RT-PCR. Stem-loop RT for mature miRNAs was performed according to the manufacturer's protocol.

All reagents for stem-loop RT were obtained from Promega, Inc. (Madison, WI) and RiboBio. PCR products were analyzed on 3% agarose gels. Small RNA U6 was used as an internal control. Real-time PCR was performed with SYBRGreen (Bio-Rad). Primers and other reagents of mature miRNA assays were purchased from RiboBio. Primers of real-time PCR for other genes are listed in Supporting Table 1.

Chromatin Immunoprecipitation Assay and Real-Time PCR.

Chromatin immunoprecipitation (IP) assays were performed according to Yi et al.23 Briefly, intracellular protein-DNA complexes were cross-linked in situ by the addition of 1% of formaldehyde. Total lysates were then sonicated and subjected to chromatin-conjugated IP using specific antibodies. After reversal of cross-links, precipitated DNA was purified and analyzed by real-time PCR with specific primers (Supporting Table 1.).

Cell Cycle and Soft Agar Colony Assay.

Cell cycle analyses were performed on propidium iodide–stained nuclei using a MoFlo XDP-Flow Cytometer (Beckman Coulter, Inc). Data were analyzed by single-histogram statistics.20, 24

For colony assays, 1 × 103 cells of the indicated type were plated in triplicate in soft agar (0.35% low melting point agarose on top of 0.7% bottom agarose) in six-well plates and fed intermittently with DMEM. Colonies were enumerate after 2 weeks by staining with methylene blue after methanol fixation.25

Tumorigenicity Assays.

Four-week-old male BALB/c nude mice were purchased from Shanghai SLAC Laboratory Animal Co. and maintained in microisolator cages. Tumorigenicity assays and tumor volume measurements were performed as previously described.20 Briefly a total of 1 × 107 indicated cells were suspended in 100 μL serum-free DMEM and injected subcutaneously in the flanks of animals. Tumor growth was monitored every three days for a total period of 30 to 40 days. Tumor volumes were calculated by the equation V (mm3) = a × b × c/2, where a is the length, b is the width, and c is the height.

Patient Samples. Primary HCC samples were obtained from patients undergoing tumor resection.

Informed consent was obtained at the Union Hospital in Wuhan and at the Eastern Hepatobiliary Surgery Hospital in Shanghai, China. The diagnosis of HCC was confirmed in each case by histological reviews. None of the patients received chemotherapy prior to hepatectomy.

Co-IP and Immunoblotting.

Co-IP was performed as described.25 Briefly, cells were lysed, disrupted by sonication, and cleared by centrifugation. The supernatants were blocked with normal rabbit immunoglobulin G (IgG) and immunoprecipitated with a 1:1,000 dilution of an anti-USP28 polyclonal antibody (A300-898A; Bethyl Laboratories, Inc.). Control IP with normal rabbit IgG were performed in parallel. Immune complexes were precipitated, washed, and resuspended in sodium dodecyl sulfate—polyacrylamide gel electrophoresis (SDS-PAGE) lysis buffer, boiled, and resolved by SDS-PAGE in 10% polyacrylamide gels. Proteins were then transferred to polyvinylidene difluoride membranes which were then probed with an anti-Myc monoclonal antibody (1:500) (9E10, Santa Cruz Biotechnology) as described.23 The blot was then incubated with a 1:5,000 dilution of HRP-conjugated goat anti-mouse IgG (Santa Cruz Biotechnology), washed, and subjected to chemiluminescence detection as described.25

Results

Myc Inhibition Diminishes the Malignant Properties of HCC Cells.

Because Myc is overexpressed in many human cancers, we first evaluated Myc protein levels in a panel of human normal cell lines with wild-type p53, including primary human foreskin fibroblasts, human RPE cells immortalized with hTERT, human normal hepatocytes (HL-7702 cells), and several human liver tumor cell lines, including HepG2, Bel-7402, FHCC98 (all with wild-type p53), and Huh-7 with mutant p53. As shown in Supporting Fig. 1, Myc was detected in all cases except RPE cells, with the highest levels occurring in HepG2 and BEL-7402 cells.

Figure 1.

Expression of the miRNAs is directly regulated by Myc. (A) Heatmap representation of Myc regulating miRNA levels assessed by real-time stem-loop RT-PCR in HepG2 and BEL-7402 cell lines with inhibition of Myc and HL7702 with induction of Myc. Clustering was performed with MultiExperiment Viewer software. (B) Schematic representation of miR-148a-5p and miR-363-3p genomic loci. E-boxes in the CpG islands are indicated. (C) Real-time PCR chromatin IP analyses of Myc binding to different E-boxes. (D) Confirmation of chromatin IP assay for Myc binding to different regions by conventional PCR and gel analyses. Left: HepG2 cells. Right: BEL-7402 cells. Results are expressed as the mean ± SD from three independent experiments.

The above results confirmed that Myc is commonly overexpressed in human HCC. We therefore examined the role and function of Myc in conferring various HCC malignant phenotypes. Myc-specific siRNA was used to ablate endogenous Myc expression in the HCC-derived cell lines HepG2 and BEL-7402. Compared with cells transfected with control siRNAs, HepG2-si-Myc and BEL-7402-si-Myc cells showed significant reductions in Myc mRNA and protein levels (Supporting Fig. 2A). A significant reduction in soft agar colony formation (Supporting Fig. 2B,C) and an increase in G0/G1 phase arrest (Supporting Fig. 2D) were also seen in HepG2-si-Myc and BEL-7402-si-Myc cells. Similar results were observed when HepG2 and BL-7402 cells were treated with 10058-F4, a small molecule inhibitor of Myc-max dimerization26 (Supporting Fig. 2E,F).

Finally, we asked whether the deregulation of Myc could affect the behavior of the normal human hepatocyte line HL-7702. We therefore generated HL-7702 cells stably transfected with a MycER expression vector and showed that Myc induction with 4-HT induced the transformation of these cells (Supporting Fig. 3A,B). Overall, these data indicate that Myc is essential for maintaining the malignant phenotypes of HCCs and that the enforced expression of Myc can induce some of the phenotypes associated with HCC.

Promoter E-boxes in CpG Islands From 21 Loci are Conserved Between Human and Mouse Among 1,049 miRNA Loci.

Recent studies have shown that certain miRNAs can influence tumor growth and are considered promising targets for diagnosis, prognosis, and treatment of some cancers.27, 28 In order to identify miRNAs directly bound by Myc through the consensus E-box elements CACGTG, CACATG, CATGTG, and CACGCG,29, 30 we performed a large-scale preselection of candidate sites using bioinformatics methods. Among a collection of 1,049 miRNA loci (miRBase-Release 1631), we identified one or more E-boxes in the promoters, first exons, or first introns in 834 cases (Supporting Fig. 4A and Supporting Table 3). We then determined whether these E-boxes are located in CpG islands, because Myc-binding E-boxes are usually located within such contexts. As shown in Supporting Fig. 4A and Supporting Table 3, 94 of the 834 loci resided within CpG islands. Further analyses showed 21 of these loci to be conserved between humans and mice (Supporting Fig. 4A and Supporting Table 3). Interestingly, miRNAs from 10 of these loci have been reported to be regulated by Myc and to mediate a variety of Myc functions32, 33(Table 1 and Supporting Fig. 4B,C). Therefore, miRNAs from the other 11 loci may also be regulated by Myc and have important roles in mediating Myc function.

Myc Regulates miRNA Expression From the Other 11 Loci.

To address whether Myc regulates expression of miRNAs from the remaining 11 loci, we tested the effects of Myc inhibition or Myc induction on the expression of these miRNAs. As shown in Fig. 1A and Supporting Table 2, induction of MycER activation in HL7702 cells resulted in the down-regulation of several of the miRNAs, including miR-148a-5p and miR-363-3p, whereas inhibition of Myc in HepG2 and BEL-7402 cells resulted in up-regulation of other miRNAs, including miR-148a-5p and miR-363-3p. Taken together, Myc differentially regulates these miRNAs in a cell context–dependent manner.

Figure 2.

Effects of miR-148a-5p and miR-363-3p on tumorigenicity and cell cycle. (A) Ectopic expression of miR-148a-5p or miR-363-3p confirmed by stem-loop RT-PCR. (B) Colony formation by the indicated cell lines. (C) Anchorage-independent soft aga growth. Top: photomicrographs of typical soft agar colonies. Bottom: bar graph of average number of colonies per well. (D) Cell cycle analyses. Cells were harvested in log-phase growth. Nuclei were stained with propidium iodide and analyzed via flow cytometry. (E) In vivo tumor growth in nude mice injected with equivalent numbers of the indicated cell lines. Results are representative of three independent experiments or are expressed as the mean ± SD from three replicates. *P < 0.05. **P < 0.01. ***P < 0.001.

Myc Binds Promoter E-boxes in the CpG Islands of MiR-148a and MiR-363 Loci.

Stem-loop quantitative RT-PCR results in all three cell lines indicated miR-363-3p to be the most significantly regulated miRNA in response to Myc activation or inhibition (Supporting Table 2). In addition, prediction of miRNA targets showed the 3′-UTR of Myc to contain one highly conserved miR-148a-5p-binding site from human to dog. Based on these findings, we selected miR-363-3p and miR-148a-5p for additional analyses. To address whether Myc directly binds the promoter regions of miR-148a and miR-363, we conducted a chromatin IP assay in HepG2 and BEL-7402 cells. This revealed that Myc binds both miR-148a and miR-363 promoter regions containing the highly conserved E-box regions in CpG islands (Fig. 1B-D and Supporting Fig. 5). These results provided strong evidence that both miRNAs are directly regulated by Myc. In addition, our unpublished data show that inhibition of both miRNAs by Myc is accompanied by a prominent decrease in active histone marks around the transcription start sites of both miRNAs.

Enforced miR-148a-5p and miR-363-3p Overexpression Inhibit Tumorigenicity and Induce Cell Cycle Arrest in Liver Tumor Cell Lines.

To examine the consequences of relieving the suppression of miR-148a-5p and miR-363-3p in hepatocarcinoma, we tested whether ectopic expression of these miRNAs affected the biology of liver cells. As shown in Fig. 2A-C and Supporting Fig. 6, obvious inhibition cell growth and soft agar colony formation were observed with both miRNAs only in human liver tumor HepG2 and BEL-7402 cells harboring high levels of Myc, whereas no obvious effects were observed in human normal HL7702 hepatocytes with low levels of Myc. Supporting experiments indicated that ectopic expression of miR-148a-5p or miR-363-3p induced a consistent G0/G1 arrest in HepG2 and BEL-7402 cells, but not HL7702 cells (Fig. 2D). Ectopic expression of miR-148a-5p and miR-363-3p also inhibited the migration in HepG2 and BEL-7402 cells (Supporting Fig. 7).

Figure 3.

Myc is a direct target of miR-148a-5p. (A) The Myc 3′-UTR contains one predicted miR-148a-5p binding site conserved from human to Canis familiaris. (B) A GFP reporter construct containing a portion of the Myc 3′ -UTR with its miR-148a-5p binding site was assayed for GFP expression via immunoblotting. (C) Myc and USP28 mRNA expression levels normalized for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) by real-time RT-PCR in HepG2 and BEL-7402 cells. (D) Myc and USP28 protein levels in HepG2 and BEL-7402 cells after transfection with miR-148a-5p mimic. Results are representative of three independent experiments or means ± SD from three replicates. *P < 0.05. **P < 0.01.

To determine whether mir-148a-5p or mir-363-3p could inhibit tumor growth, HepG2 cells ectopically expressing mir-148a-5p or mir-363-3p were injected into the flanks of nude mice. This resulted in a significant decrease of tumor growth compared with the tumors expressing empty vector (Fig. 2E). Taken together, these data suggest that miRNAs induce cell cycle arrest and inhibit tumor growth in HCC cells.

miR-148a-5p Directly Targets Myc While miR-363-3p Destabilizes Myc by Directly Targeting USP28.

To elucidate the molecular mechanism by which both miRNAs induce cell cycle arrest and inhibit tumorigenicity, we performed miRDB, miRanda, miRwalk, and RNAhybrid analyses to identify functional targets of miR-148a-5p and miR-363-3p. These analyses revealed the 3′-UTR of Myc to contain one highly conserved miR-148a-5p binding site from human to Canis familiaris (Fig. 3A), whereas the 3′-UTR of USP28 mRNA, the ubiquitin protease of Myc, contains one highly conserved from human to Equus caballus and the other nonconserved miR-363-3p binding sites (Fig. 4A).

Figure 4.

miR-363-3p directly targets USP28. (A) The USP28 3′-UTR contains two predicted miR-363-3p binding sites. (B) A GFP reporter construct containing a portion of the USP28 3′-UTR with its miR-363-3p binding sites was assayed by immunoblotting. (C) USP28 and Myc mRNA levels were normalized for GAPDH by real-time RT-PCR in HepG2 and BEL-7402 cells following transfection with miR-363-3p mimic. (D) USP28 and Myc protein levels in HepG2 and BEL-7402 cells after transfection with miR-363-3p mimic. Results are representative of three independent experiments or are expressed as the mean ± SD from three replicates. *P < 0.05. **P < 0.01.

Figure 5.

USP28 stabilizes Myc. (A) Myc and USP28 protein levels in HepG2 cells treated or not treated with MG132 after transfection with miR-363-3p mimic. (B) USP28 and Myc mRNA levels normalized for GAPDH by quantitative RT-PCR in HepG2 and BEL7402 after transfection with USP28-specific siRNA. (C) USP28 and Myc protein levels after transfection with USP28-specific siRNA. (D) Focus formation assay by the indicated cells. (E) Myc and USP28 protein levels in HepG2 cells treated or not treated with MG132 after transfected with USP28 specific siRNA. (F) Co-IP of Myc and USP28 from HepG2 cells. HepG2 cell lysates were immunoprecipitated with USP28 antibody or control IgG followed by immunoblotting for Myc. Results are representative of three independent experiments or are expressed as the mean ± SD from three replicates. *P < 0.05. **P < 0.01.

We tested whether Myc or USP28 are direct targets of miR-148a-5p or miR-363-3p, respectively. For this, a GFP reporter assay was employed to detect the potential interaction of each miRNA with the 3′-UTR of targets. The results showed that miR-148a-5p inhibited the GFP expression of a vector containing the predicted miR-148a-5p binding site but not the GFP vector only (Fig. 3B). We also found that miR-363-3p repressed the GFP expression of a vector containing either one or both of the predicted miR-363-3p binding sites, but not the nonconserved binding site (Fig. 4B). These data supported direct inhibition of Myc by miR-148a-5p and USP28 by miR-363-3p.

As expected, ectopic expression miR-148a-5p in HepG2 and BEL-7402 HCC cells resulted in a marked decrease of Myc mRNA and protein and an increase in mir-363-3p. This was further associated with decrease in USP28 mRNA and protein (Fig. 3C,D); ectopic expression miR-363-3p in HepG2 and BEL-7402 cells resulted in a marked decrease of USP28 at both mRNA and protein levels, while in a marked decrease of Myc at protein level but not mRNA level (Fig. 4C,D).

Although the 3′-UTR of Myc does not contain predicted binding sites for miR-363-3p, ectopic expression miR-363-3p also led to a marked decrease of Myc protein but not mRNA (Fig. 4C,D). The reduction in Myc protein could be prevented, however, if the cells were exposed to MG132, a proteasome inhibitor (Fig. 5A). Because USP28 has been reported to be required for Myc stability in human tumor cells,17 we evaluated USP28 protein levels in a panel of normal human hepatocyte and HCC cell lines. As shown in Supporting Fig. 1A, USP28 was detected in HepG2, BEL-7402, and FHCC98 liver tumor cell lines, whereas it was undetectable in normal cell lines. When HepG2 and BEL-7402 cells were transfected with a USP28 siRNA, we noted a subsequent decrease of Myc at the protein level but not at the mRNA level (Fig. 5B,C) whereas inhibition of USP28 with MG132 resulted in no changes in Myc protein or mRNA levels (Fig. 5E). We also observed that inhibition of USP28 suppressed the malignant phenotype of HepG2 and BEL-7402 cells in a manner that correlated with their reduced Myc levels (Fig. 5D). Co-IP experiments also confirmed that USP28 directly interacts with Myc in HepG2 cells (Fig. 5F). These data indicate that miR-363-3p indirectly destabilizes Myc by directly targeting USP28.

Figure 6.

Suppression of miR-148a-5p or miR-363-3p increases tumorigenicity and induces G0/G1 to S phase progression. (A) miR-148a-5p, miR-363-3p, Myc, and USP28 levels normalized for U6 or GADPH by real-time RT-PCR in HL7702 and FHCC98 cells after transfection with miR-148a-5p– or miR-363-3p–specific inhibitor. (B) Myc and USP28 protein levels in indicated cells after transfection with miR-148a-5p– or miR-363-3p–specific inhibitor. (C) Monolayer colony formation assay of indicated cells. (D) Soft agar colony formation assay of indicated cells. (E) Cell cycle analyses of indicated cells. Cells were harvested in log-phase growth. Nuclei were stained with propidium iodide and analyzed via flow cytometry. (F) In vivo tumor growth in nude mice injected with indicated cells. Each point represents the mean ± SD of three replicates. *P < 0.05. **P < 0.01. ***P < 0.001.

Inhibition of miR-148a-5p or miR-363-3p Induces G0/G1 to S Phase Progression and Increases Tumorigenicity.

Among cell lines used in this study, miR-148a-5p and miR-363-3p express high levels in the normal human hepatocyte cell line HL7702 and human liver tumor cell line FHCC98. To elucidate whether inhibition of miR-148a-5p or miR-363-3p affected cellular properties associated with the malignant phenotype, we inhibited each miRNA using miRNA-specific inhibitors in HL7702 and FHCC98 cells. As shown in Fig. 6A,B, this significantly decreased their miRNA expression and led to increases in their target transcripts and proteins in both cases. It also enhanced several malignant phenotypes (Fig. 6C,D). For example, it promoted G0/G1 to S phase progression in HL7702 and FHCC98 cells (Fig. 6E) and also promoted cell migration (Supporting Fig. 10). Moreover, a significant increase in tumor growth rates were observed when compared with the tumors expressing control miRNA inhibitor in FHCC98 cells (Fig. 6F). These studies show that both gain- and loss-of-function of miR-148a-5p or miR-363-3p affected multiple aspects of the malignant phenotypes in HCCs.

Figure 7.

The Myc-miRNA feedback loop is dysregulated in HCCs. (A) Myc, miR-148a-5p, USP28, and miR-363-3p RNA levels in two normal and four HCC cell lines. (B) RNA levels of Myc, USP28, mir-148a-5p, and mir-363-3p in 27 HCCs and their paired normal tissues. The results shown in (A) and (B) are expressed as the mean ± SD from three independent experiments. (C) Heatmap representation of Myc, miR-148a-5p, USP28, and miR-363-3p RNA levels in 19 normal liver tissues and 27 HCCs from (B). Clustering was performed with MultiExperiment Viewer software. (D) Correlation between different member's RNA expression of the Myc-miRNA feedback loop in 10 HCCs and 8 normal liver tissues from (B). Each point shows an individual hepatocellular tissue sample. R, correlation coefficient. (E) Western blot assays show Myc and USP28 protein levels from the indicated HCC samples. (F) Proposed model of the Myc-miRNA feedback loop in hepatocarcinogenesis.

The Myc-miRNA Functional Feedback Loop Is Dysregulated in Human HCC.

In order to investigate whether the Myc-miRNA feedback loop is dysregulated in human HCCs, expression levels of Myc, USP28, mir-148a-5p, and mir-363-3p were quantified in total RNA derived from two normal human cell lines, four human liver tumor cell lines, 27 hepatocarcinomas, and paired normal hepatic tissues. These studies showed increased levels of Myc in 17 of the HCCs, (>1.5 fold change) and increased levels of USP28 in 21 cases. In contrast, mir-148a-5p and mir-363-3p transcripts were reduced in 23 and 11 cases, respectively (Fig. 7A-C). Hence, the Myc-miRNA feedback loop is dysregulated in HCCs. Furthermore, our results generally showed a positive correlation between Myc and USP28 levels, a negative correlation between Myc and mir-148a-5p or mir-363-3p levels, and a negative correlation between USP28 and mir-363-3p levels (Fig. 7A,D). In addition, western blot assays showing Myc or USP28 protein levels were up-regulated in HCC relative to adjacent normal tissues, in human liver tumor cell lines relative to normal human cell lines (Fig. 7E and Supporting Fig. 1A). Taken together, our findings suggest the Myc-miRNA functional feedback loop affecting hepatocarcinogenesis (Fig. 7F).

Discussion

Myc is essential for development and survival5-10, 14, 15, 48 and is a well-known regulator of proliferation, differentiation, and oncogenesis.5-9, 15 Identifying the mechanisms underlying these processes would therefore be expected to be helpful in understanding the molecular pathophysiology of cancer.5-9, 15

Deregulation of Myc acts as an oncogenic driver in many cancers, including HCC.2, 5, 6, 8, 9, 15, 49 For example, the activation of an Myc transcription signature is strongly associated with the malignant conversion of preneoplastic liver lesions,15, 17 and Myc inactivation is sufficient to induce regression of invasive liver cancers.6 Our current data show that inhibition of Myc, via a novel negative feedback mechanism involving mir-148a-5p and mir-363-3p, decreases the malignant phenotypes and induces cell cycle arrest of HCC, thus suggesting that Myc plays a very important role in the tumorigenesis of HCC.

Myc is known to directly regulate miRNAs with oncogenic and tumor suppressor function.5, 7, 10, 32, 41 miRNAs are small, noncoding RNAs that posttranscriptionally regulate gene expression. Recent functional studies have shown that specific miRNAs can act as disease modifiers.27, 28 Myc directly activates the miR-17-92 cluster in human B cells32 and widespread Myc-mediated miRNA repression contributes to lymphomagenesis in mice.33 Cairo et al.7 also reported that Myc directly regulates mir-371-3 and mir-100/let7a-2/mir-125b-1 cluster contributing to the pathogenesis of hepatoblastoma. In the current study, we identified two Myc-repressed miRNAs, miR-148a-5p and miR-363-3p, as contributing to the generation of HCC.

Our data provided evidence that the expression of mir-148a-5p and mir-363-3p inhibits tumorigenicity and induces cell cycle arrest through mechanisms leading to a decrease in Myc. This down-regulation occurs through distinct albeit complementary mechanisms. In the first case, miR-148a-5p directly targets and inhibits Myc, whereas in the second case, miR-363-3p works more indirectly by destabilizing Myc through the targeting of USP28. In this process, we revealed that these miRNAs comprise a negative feedback loop that cooperate to inhibit the translation of Myc and promote the degradation of preexisting Myc protein.

miR-148-5p was first reported by Lujambio et al.36 as an inhibitor of tumor invasion and metastasis of gastric cancer. MiR-363-3p was demonstrated to be down-regulated and to inhibit the growth in T cell lymphomas.50 Tumor-specific hypermethylation or Myc overexpression to suppress miR-148a-5p expression leads to decreased miR-148a-5p levels contributing to tumorigenicity. Future studies should focus on whether miR-148a-5p or miR-363-3p suppresses liver tumorigenesis in liver-specific Myc with full 3′-UTR region transgenic mice. In conclusion, we identified a c-Myc-miRNA feedback loop that regulates hepatocarcinogenesis.

Acknowledgements

We thank HongBing Shu for encouraging suggestions, Edward Prochownik for revising the manuscript, and all members in our laboratory for discussion and technical help.

Ancillary

Advertisement