Perturbation of MicroRNA-370/Lin-28 homolog A/nuclear factor kappa B regulatory circuit contributes to the development of hepatocellular carcinoma

Authors


  • Potential conflict of interest: Nothing to report.

  • This work was supported by grant no. 81071842, Distinguished Young Scholars (30825020), Key Program (81230011), and Creative Research Groups (81221061) from the National Natural Science Foundation of China and grants from Shanghai Science and Technology Committee for the key projects (11JC1416200 and 10431903600) and the Rising-Star Program (11QA1408800).

Abstract

MicroRNA 370 (miR-370) is located within the DLK1/DIO3 imprinting region on human chromosome 14, which has been identified as a cancer-associated genomic region. However, the role of miR-370 in malignances remains controversial. Here, we report that miR-370 was repressed in human hepatocellular carcinoma (HCC) tissues and hepatoma cell lines. Using gain-of-function and loss-of-function experiments, we demonstrated that miR-370 inhibited the malignant phenotype of HCC cells in vitro. Overexpression of miR-370 inhibited growth and metastasis of HCC cells in vivo. Moreover, the RNA-binding protein, LIN28A, was identified as a direct functional target of miR-370, which, in turn, blocked the biogenesis of miR-370 by binding to its precursor. LIN28A also mediated the suppressive effects of miR-370 on migration and invasion of HCC cells by post-transcriptionally regulating RelA/p65, which is an important effector of the canonical nuclear factor kappa B (NF-κB) pathway. Interleukin-6 (IL-6), a well-known NF-κB downstream inflammatory molecule, reduced miR-370 but increased LIN28A levels in HCC. Furthermore, miR-370 levels were inversely correlated with LIN28A and IL-6 messenger RNA (mRNA) levels, whereas LIN28A mRNA expression was positively correlated with IL-6 expression in human HCC samples. Interestingly, reduction of miR-370 expression was associated with the development of HCC in rats, as well as with aggressive tumor behavior and short survival in HCC patients. Conclusions: These data demonstrate the involvement of a novel regulatory circuit consisting of miR-370, LIN28A, RelA/p65 and IL-6 in HCC progression. Manipulating this feedback loop may have beneficial effect in HCC treatment. (Hepatology 2013; 58:1977–1991)

Abbreviations
Ab

antibody

Ad-GFP

adenovirus containing green fluorescent protein

Ad-miR-370

adenovirus expressing miR-370

CCA

cholangiocarcinoma

CMV

cytomegalovirus

Cpt1α

carnitine palmitoyl transferase 1 alpha

DEN

diethylinitrosamine

HCC

hepatocellular carcinoma

H&E

hematoxylin and eosin

IHC

immunohistochemical

IL-6

interleukin-6

IMH

nontransformed immortalized human hepatocytes

IP

immunoprecipitation

ISH

in situ hybridization

JSH-23

4-methyl-N1-(3-phenylpropyl)benzene-1,2-diamine

LIN28

Lin-28 homolog A

miR-370

microRNA 370

mRNA

messenger RNA

miRNA

microRNA

MMP-9

matrix metalloproteinase 9

NF-κB

nuclear factor kappa B

NOD/SCID

nonobese diabetic severe compromised immunodeficient

OS

overall survival

pcDNA

plasmid DNA with a cytomegalovirus promoter

PCR

polymerase chain reaction

RIP assay

RNA-binding protein immunoprecipitation assay

SC

subcutaneous

SD

standard deviation

siRNA

short interfering RNA

TNF-α

tumor necrosis factor alpha

UTR

untranslated region

WT

wild type.

Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide, especially in Asia.[1] Most HCCs develop on a background of chronic inflammation caused by hepatitis virus, toxins, metabolic impairment, or autoimmune hepatopathy.[2] Inflammatory molecules can provide signals that promote the proliferation and metastasis of HCC cells.[2, 3] The transcription factor, nuclear factor kappa B (NF-κB), is a key modulator of inflammatory response and plays a pivotal role in the regulation of inflammatory signal transduction pathways in the liver.[4] Activation of NF-κB is also widely viewed as a link between inflammation and the pathogenesis of various cancers, including HCC.[4, 5]

MicroRNAs (miRNAs) are a class of small noncoding RNA molecules that regulate post-transcriptional events.[6] Aberrant expression of many miRNAs is implicated in the onset and development of HCC.[7, 8] MicroRNA 370 (miR-370) is located within the DLK1/DIO3 imprinting region on human chromosome 14.[9] It was first cloned from human embryonic stem cells, but had a very low expression level.[10] Several studies have identified the DLK1/DIO3 domain as a cancer-associated genomic region,[11] implicating the involvement of miR-370 in cancer pathogenesis. Nevertheless, the role of miR-370 in malignances remains controversial. Substantial evidence demonstrates that miR-370 serves as a tumor suppressor in malignant cholangiocytes,[12, 13] leukemia cells,[14] and oral squamous carcinoma cells.[15] In contrast, several studies have reported that overexpression of miR-370 contributes to the progression of gastric carcinoma, prostate cancer, and acute myeloid leukemia.[16-18] In addition, miR-370 has been shown to affect lipid metabolism in the liver by directly targeting carnitine palmitoyl transferase 1 alpha (Cpt1α) and up-regulating liver-enriched miRNA miR-122,[19] indicating that miR-370 may be important for hepatic function.

Lin28, consisting of Lin28 homolog A (Lin28A) and its homolog, Lin28B, is a functionally conserved RNA-binding protein originally characterized in Caenorhabditis elegans as a major regulator of developmental timing.[20, 21] Emerging evidence suggests that Lin28 plays crucial roles not only in development, but also in pluripotency, metabolism, and carcinogenesis in mammals.[21] Despite its wide expression in the early stage of developing tissues, Lin28 is undetectable in most adult organs.[22] Interestingly, both LIN28A and LIN28B are up-regulated in diverse human malignancies, including ovarian, breast, colon, lung, and liver cancer, as well as in chronic myeloid leukemia and germ cell tumors.[23-26] Higher expression of LIN28A/LIN28B is associated with more-advanced tumor grade and poorer prognosis.[23, 27] Functional studies have also suggested that LIN28A and LIN28B facilitate the carcinogenesis and development of cancers, including HCC.[23, 24, 26, 28-32] Both LIN28A and LIN28B promote the proliferation of HCC cells, whereas LIN28B also enhances the transformation and invasion of HCC.[23, 24, 31, 32] However, the tumor-promoting mechanisms of LIN28 in HCC remain largely unknown.

In this study, we clarified the role of miR-370 in HCC and elucidated the contribution of the miR-370/LIN28A/NF-κB circuit to the progression of HCC. We speculate that manipulation of this feedback loop could be explored as a novel strategy for the treatment of HCC.

Materials and Methods

Human Tissues

Human liver tissue samples (excluding the samples on the tissue microarray) were obtained from patients who underwent surgical resection and were diagnosed by professional pathologists at the Eastern Hepatobiliary Surgery Hospital (Shanghai, China) and Changzheng Hospital (Shanghai, China), with written informed consent. HCC tissues with typical macroscopic features were collected from the central part of tumor nodules, which were also examined with hematoxylin and eosin (H&E) staining to confirm the diagnosis. The paired adjacent nontumoral tissues without histopathologically identified tumor cells were collected from at least 5 cm away from the tumor border. All human experiments were approved by the ethics committee of the Second Military Medical University (Shanghai, China).

Animal Model

To detect the effect of miR-370 on tumorigenicity in vivo, HCC cells infected with adenovirus expressing miR-370 (Ad-miR-370) or control virus adenovirus containing green fluorescent protein (Ad-GFP) were transplanted subcutaneously (SC) into both flanks of Balb/c nude mice. To explore the effect of miR-370 on metastasis, MHCC-LM3 cells stably expressing luciferase and infected with Ad-GFP or Ad-miR-370 were injected by the tail vein into nonobese diabetic severe compromised immunodeficient (NOD/SCID) mice. Mice were monitored using the IVIS200 imaging system (Caliper Life Sciences, Hopkinton, MA) once a week and sacrificed 8 weeks after cell transplantation. Tumor nodules on the lungs were counted and histopathologically analyzed with H&E staining. To investigate the antitumor effect of miR-370 in vivo, Ad-miR-370 or Ad-GFP was injected into the xenografts of an SC-implanted tumor model in Balb/c nude mice. All animal experiments were performed according to protocols approved by the institutional animal care and use committee at the Second Military Medical University.

RNA-Binding Protein Immunoprecipitation Assay

RNA-binding protein immunoprecipitation (RIP) assays were performed as described previously,[29] with minor modifications. Briefly, Lin28A primary antibody (ab46020; Abcam, Cambridge, MA) was used for endogenous LIN28A immunoprecipitation (IP) in PLC/PRF/5 cells; anti-FLAG Ab-conjugated agarose beads (A2220; Sigma-Aldrich, St. Louis, MO) were used for IP of ectopically expressing LIN28A in MHCC-97H cells transfected with pFlag-cytomegalovirus (CMV)-2 empty vector, pFlag-CMV-LIN28A vector, and pFlag-CMV-LIN28A vector with C161A mutation.

In Situ miRNA Hybridization Assay

Tissue microarray slides of 50 paired HCC and adjacent cancer-free samples were obtained from Xinchao Biotechnology (Shanghai, China). In situ miRNA hybridization (ISH) assays were performed according to the manufacturer's instructions. Stained slides were scanned using a ScanScopeXT (Aperio Technologies, Vista, CA) scanner and analyzed with Aperio Spectrum software (Aperio Technologies).

Statistical Analysis

All statistical analyses were performed using SPSS software (version 17.0; SPSS, Inc., Chicago, IL). Statistical tests for data analysis included two-tailed Student t, log-rank, Wilcoxon's matched pairs, Mann-Whitney's U, and chi-square tests. A P value <0.05 was considered statistically significant.

Detailed materials and methods can be found in the Supporting Materials.

Results

miR-370 Inhibits Malignant Phenotype of HCC Cells In Vitro

We examined miR-370 expression in 20 paired primary HCC and surrounding nontumorous liver specimens. miR-370 was down-regulated (HCC/nontumorous specimens <0.5) in 16 cases (80%; Fig. 1A). We also detected miR-370 expression in nontransformed immortalized human hepatocytes (IMH) and hepatoma cell lines. miR-370 levels were significantly decreased, relative to the IMH control, in all of the 11 tested hepatoma cell lines (Supporting Fig. 1A). Enforced expression of miR-370 by transfecting cells with plasmid DNA with a cytomegalovirus promoter (pcDNA)-miR-370 reduced colony formation of HCC cells (Fig. 1B and Supporting Fig. 1B). Overexpression of miR-370 significantly decreased proliferation of HCC cells (Fig. 1C and Supporting Fig. 1C,D). In contrast, inhibition of miR-370 enhanced cell proliferation (Fig. 1D and Supporting Fig. 1E). Flow cytometry assay showed that overexpression of miR-370 promoted apoptosis of HCC cells, whereas inhibition of miR-370 attenuated serum starvation-induced apoptosis of HCC cells (Supporting Fig. 2). We also examined the effects of miR-370 on migration and invasion of the highly invasive MHCC-LM3 and YY-8103 cells. miR-370 overexpression markedly reduced, whereas miR-370 inhibition increased, migration and invasion of these cells (Fig. 1E,F and Supporting Fig. 3A-D). Interestingly, miR-370 inhibition also markedly enhanced migration and invasion of IMH cells (Supporting Fig. 3E,F).

Figure 1.

miR-370 suppresses malignant phenotype of HCC cells in vitro. (A) Real-time PCR quantification of miR-370 expression in 20 paired HCC and adjacent nontumorous tissues (NT). miR-370 expression was normalized against an endogenous control (U6). (B) Representative images (top) and quantification (bottom) of colony formation assay in MHCC-97H cells transfected with pcDNA3.0 (vector) or pcDNA-miR-370 (miR-370). (C) Proliferation of MHCC-97H cells transfected with miR-370 mimic (miR-370) or negative control (NC) was determined as described in the Supporting Materials. (D) miR-370 inhibitor enhanced the proliferation of MHCC-97H cells. (E) Migration (left) and invasion (right) assays of MHCC-LM3 cells transfected with miR-370 mimic and NC for 72 hours. (F) Migration (left) and invasion (right) assays of MHCC-LM3 cells treated with miR-370 inhibitor and NC inhibitor for 48 hour. ★★P < 0.01; ★★★P < 0.001 by two-tailed Student t test. Experiments were performed in triplicate, and data are shown as mean ± SD.

miR-370 Suppresses Growth and Metastasis of HCC Cells In Vivo

To further investigate the effect of miR-370 on tumorigenesis of HCC cells in vivo, MHCC-97H or YY-8103 cells infected with Ad-miR-370 or control adenovirus Ad-GFP were SC transplanted into the flanks of Balb/c nude mice. Xenografts were detected in 37.5% (3 of 8) of mice as early as day 14 and in all subjects by day 33 after inoculation in mice receiving MHCC-97H cells infected with Ad-GFP (Fig. 2A). No xenografts were observed until day 33 in mice receiving MHCC-97H cells infected with Ad-miR-370, and only small nodules were identified in 50% (4 of 8) of mice by day 38 (Fig. 2A). Xenografts were significantly smaller in the Ad-miR-370 group, compared to the control group, at every time point (Supporting Fig. 4A). Consistently, xenograft weight was significantly reduced in the Ad-miR-370 group (Fig. 2B). Real-time polymerase chain reaction (PCR) analysis showed a significant increase in miR-370 levels in the Ad-miR-370 group, relative to the Ad-GFP control (Supporting Fig. 4B). Similar results were obtained with YY-8103 cells (Supporting Fig. 4C,D). We further investigated the effect of miR-370 on HCC metastasis in vivo in NOD/SCID mice injected with luciferase-labeled MHCC-LM3 cells infected with Ad-miR-370 or Ad-GFP. Luciferase signals were detected in lungs in all mice in the Ad-GFP group by ex vivo imaging, but in only 2 of 5 mice in the Ad-miR-370 group 8 weeks after cell transplantation (Fig. 2C and Supporting Fig. 4E). Number of tumor foci on lungs was also significantly reduced in the Ad-miR-370 group (Fig. 2D). Histologic analysis confirmed reduced tumor foci, which were composed of paratypic HCC cells in the Ad-miR-370 group (Fig. 2D). We then explored the antitumor effect of miR-370 on an established HCC cell transplanted SC tumor model in Balb/c nude mice. Intratumoral injection of Ad-miR-370 significantly reduced the growth and weight of MHCC-97H xenografts (Fig. 2E and Supporting Fig. 4F). Real-time PCR confirmed the increased expression of miR-370 in Ad-miR-370-treated tumor nodules (Supporting Fig. 4G). Histological analysis revealed that the tumor nodules were composed of HCC cells arranged in a trabecular pattern, as proved by H&E staining (Fig. 2F). Additionally, Ad-miR-370-treated tumor nodules displayed decreased Ki-67 expression (Fig. 2F) and contained more apoptotic cells (Supporting Fig. 5).

Figure 2.

Up-regulation of miR-370 inhibits malignant behavior of HCC cells in vivo. (A) HCC-free survival of mice transplanted with MHCC-97H cells infected with Ad-miR-370 or Ad-GFP were analyzed by Kaplan-Meier's method and compared using the log-rank test (n = 8 in each group). (B) Representative images (top) and weight (bottom) of xenografts derived from MHCC-97H cells infected with Ad-miR-370 and Ad-GFP, respectively. Horizontal line at the bottom indicates median value. P = 0.0078 by Wilcoxon's matched pairs test. (C) Images (top) and statistical analysis (bottom) of luciferase signals in lungs of NOD/SCID mice transplanted with MHCC-LM3 cells infected with Ad-GFP and Ad-miR-370, respectively, at 8 weeks after inoculation (n = 5 in each group). Horizontal line at the bottom indicates median value. P = 0.0317 by Mann-Whitney's U test. (D) Representative H&E-stained sections of lung tissues (top) and number of tumor foci (bottom) in lungs collected from Ad-GFP- and Ad-miR-370-infected groups. Arrow indicates tumor foci in the lung. Horizontal line at the bottom indicates median value. P = 0.0157 by Mann-Whitney's U test. Scale bar = 100 μm. (E) Representative images (top) and weight (bottom) of xenografts established by SC transplantation with MHCC-97H cells and then injected with Ad-miR-370 or Ad-GFP (n = 11 in each group). Horizontal line at the bottom indicates median value. P = 0.0029 by Wilcoxon's matched pairs test. (F) H&E staining performed on paraffin-embedded sections obtained from xenografts injected with Ad-GFP or Ad-miR-370 showed a typical trabecular HCC pattern in both the Ad-GFP and Ad-miR-370 groups. IHC staining showed a reduction of Ki-67 expression in the Ad-miR-370 group, compared to the Ad-GFP group. Scale bar = 100 μm.

LIN28A Is a Functional Mediator of miR-370

To investigate the underlying molecular mechanisms by which miR-370 exerts its antitumor effect, we screened for putative targets of miR-370 by performing an in silico complementarity search using TargetScan (www.targetscan.org/) and PicTar (http://picta.mdc-berlin.de/). This approach identified LIN28A, an evolutionarily conserved molecule across many species, as a potential downstream target of miR-370 (Fig. 3A and Supporting Fig. 6A). LIN28A messenger RNA (mRNA) and protein levels were decreased in HCC cells by ectopic expression of miR-370 (Fig. 3B) and increased by miR-370 inhibitor (Fig. 3C). Immunohistochemical (IHC) analysis also revealed decreased LIN28A in Ad-miR370-treated MHCC-97H xenografts (Supporting Fig. 6B). Reporter assay revealed that overexpression of miR-370 decreased the luciferase activity of the wild-type (WT) LIN28A 3′ untranslated region (UTR) by 59.4% (P < 0.0001; Fig. 3D). Deletion or point mutation of the target sequence on the LIN28A 3′ UTR diminished the effect of miR-370 on LIN28A, indicating that LIN28A is a direct downstream target of miR-370 (Fig. 3D and Supporting Fig. 6C,D).

Figure 3.

Identification of LIN28A as a direct effector of miR-370. (A) Predicted consequential pairing of target region in the 3′ UTR of LIN28A (top) and miR-370 (bottom) using the prediction program, TargetScan. (B) Real-time PCR quantification (top) and western blotting analysis (bottom) of LIN28A expression in HCC cells transfected with miR-370 mimic or negative control (NC) for 24 hours. Expression of LIN28A mRNA was normalized against β-actin. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (C) Transfection of miR-370 inhibitor or control inhibitor for 24 hours promoted the expression of LIN28A at both mRNA (top) and protein levels (bottom) in HCC cells. (D) Effect of miR-370 on the LIN28A 3′ UTR. A luciferase reporter plasmid carrying either the LIN28A 3′ UTR (WT) or LIN28A 3′ UTR with miR-370 target-sequence deletion (mutant) was cotransfected into HEK293T cells with miR-370 mimic or NC. (E and F) Lentivirus expressing LIN28A without the 3′ UTR reversed effects of miR-370 on migration (E) and invasion (F) of MHCC-LM3 cells. P < 0.05; ★★P < 0.01; ★★★P < 0.001 by two-tailed Student t test. Experiments were performed in triplicate, and data are shown as mean ± SD.

Enforced expression of LIN28A promoted proliferation of MHCC-97H cells, whereas knockdown of LIN28A inhibited their proliferation (Supporting Fig. 7A,B). In addition, overexpression of LIN28A significantly augmented, whereas down-regulation of LIN28A suppressed, migration and invasion of HCC cells (Supporting Fig. 7C,D). Importantly, the suppressive effects of miR-370 on migration and invasion of HCC cells were substantially reduced by infection with a lentiviral expression vector of LIN28A without the 3′ UTR (Fig. 3E,F and Supporting Fig. 7E). Overall, these findings demonstrate that down-regulation of LIN28A contributes to the functional role of miR-370 in HCC cells.

LIN28A Blocks Maturation of miR-370

LIN28 has been shown to function as an oncoprotein by forming a double-negative feedback loop with let-7 in breast cancer.[29] Identification of LIN28A as a target of miR-370 in HCC cells raises the possibility that LIN28A may block the biogenesis of miR-370. Indeed, our results showed that overexpression of LIN28A significantly decreased miR-370 level, whereas substitution of a single amino acid (C161A) required for the RNA-binding affinity of LIN28A[33] efficiently reversed the effect of LIN28A on miR-370 (Fig. 4A). As a positive control, let-7 level was also reduced upon ectopic expression of LIN28A, but not by C161A mutation (Fig. 4A). However, as a negative control, miR-21 level[34] was not influenced by LIN28A (Fig. 4A). In contrast, knockdown of LIN28A by small interfering RNA (siRNA) substantially raised levels of miR-370 and let-7, but not miR-21 (Fig. 4B). RIP assay revealed that both miR-370 and let-7 precursors, but not miR-21 precursor, were highly enriched in LIN28A immunoprecipitates from PLC/PRF/5 cells (Fig. 4C), suggesting direct binding between endogenous LIN28A and pre-miR-370 in HCC cells. To confirm the specificity of binding, MHCC-97H cells were transfected with Flag-LIN28A or empty vector. Subsequent RIP assay displayed significant enrichment of pre-miR-370 and pre-let-7 in Flag-LIN28A immunoprecipitates, which were abolished by C161A mutation (Fig. 4D).

Figure 4.

LIN28A blocks biogenesis of miR-370. (A) Real-time PCR quantification (top) of LIN28A, mature miR-370, let-7, and miR-21 expression in MHCC-97H cells transfected with control vector (Flag), Flag-LIN28A (LIN28A), and Flag-LIN28A with C161A mutation (C161A). Expression of miRNA was normalized against an endogenous control (U6). Expression of LIN28A was normalized against β-actin. Western blotting analysis (bottom) indicated LIN28A expression in MHCC-97H cells transfected with the indicated vectors. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (B) Real-time PCR quantification (top) of LIN28A, mature miR-370, let-7, and miR-21 expression in MHCC-97H cells transfected with siRNA against LIN28A (si-LIN28A) or its negative control NC (NC). Western blotting analysis (bottom) indicated LIN28A expression in MHCC-97H cells transfected with NC or si-LIN28A. (C and D) Enrichment of precursors of miR-370 (pre-miR-370), let-7 (pre-let-7), and miR-21 (pre-miR-21) in endogenous LIN28A immunoprecipitates from PLC/PRF/5 cells (C) and in Flag-LIN28A immunoprecipitates from MHCC-97H cells transfected with indicated plasmids (D) were assessed by RNA-binding protein IP assay coupled with real-time PCR. Western blotting analysis indicated LIN28A expression in PLC/PRF/5 (C) and MHCC-97H cells (D). P < 0.05; ★★P < 0.01; ★★★P < 0.001 by two-tailed Student t test. Experiments were performed in triplicate, and data are shown as mean ± SD.

LIN28A Activates NF-κB Pathway by Posttranscriptional Regulation of RelA/p65

NF-κB has been reported to transcriptionally activate the expression of LIN28B, but not LIN28A, in breast cancer.[28, 34] However, the effect of LIN28 on NF-κB has not been reported. Our experiments showed that LIN28A overexpression enhanced, whereas LIN28A knockdown suppressed, the activity of a NF-κB luciferase reporter[35] in HCC cells (Fig. 5A and Supporting Fig. 8A). LIN28A inhibition resulted in down-regulation of NF-κB target genes, including interleukin-6 (IL-6), tumor necrosis factor alpha (TNF-α), and matrix metalloproteinase 9 (MMP-9; Fig. 5B), indicating that LIN28A is implicated in activation of the NF-κB pathway in HCC.

Figure 5.

LIN28A activates the NF-κB pathway by promoting RelA/p65 translation. (A) Down-regulation of LIN28A by siRNA (si-LIN28A) inhibited NF-κB luciferase reporter activity in HCC cells. (B) Real-time PCR analysis of expression of the NF-κB regulatory genes, IL-6, TNF-α and MMP-9, in MHCC-97H cells transfected with si-LIN28A or negative control (NC). Gene expression was normalized against β-actin. (C) Western blotting analysis of RelA/p65 in HCC cells infected with lentivirus expressing LIN28A or GFP for 5 days (top) and in HCC cells transfected with si-LIN28A or NC for 48 hours (bottom). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (D and E) Enrichment of RelA/p65 in endogenous LIN28A immunoprecipitates from PLC/PRF/5 cells (D) and in Flag-LIN28A immunoprecipitates from MHCC-97H cells transfected with the indicated plasmids (E) were assessed by RIP assay coupled with real-time PCR. (F) Luciferase activity of psiCHECK2 carrying the RelA/p65 3′ UTR in HCC cells infected with lentivirus expressing LIN28A or GFP for 5 days (left) and in HCC cells transfected with si-LIN28A or NC for 48 hours (right). P < 0.05; ★★P < 0.01; ★★★P < 0.001 by two-tailed Student t test. Experiments were performed in triplicate, and data are shown as mean ± SD.

LIN28A is a post-transcriptional modulator of mRNAs,[36] and we therefore sought to determine its effect on the translation of RelA/p65, which plays an important role in canonical NF-κB pathway transduction.[4] Real-time PCR failed to detect any significant effect of LIN28A on RelA/p65 mRNA levels (data not shown). However, RelA/p65 protein levels were significantly increased by LIN28A overexpression and decreased by LIN28A knockdown (Fig. 5C). Direct binding of RelA/p65 mRNA and LIN28A, which was abolished by C161A mutation, was detected by RIP assay (Fig. 5D,E). Furthermore, LIN28A overexpression increased, whereas LIN28A repression decreased, the activity of the luciferase reporter gene carrying the RelA/p65 3′ UTR (Fig. 5F). Interestingly, the effects of LIN28A on HCC cell migration and invasion were reversed by inhibition of RelA/p65 NF-κB transcriptional activity with oridonin and 4-methyl-N1-(3-phenylpropyl)benzene-1,2-diamine (JSH-23; Supporting Fig. 8B-E). Overall, these findings suggest that direct binding of LIN28A to RelA/p65 mRNA promotes the translation of RelA/p65, which contributes, at least in part, to the functional role of LIN28A in HCC.

Positive Feedback Loop Involving miR-370, LIN28A, RelA/p65, and IL-6 Aggravates Malignant Phenotype of HCC Cells

In view of the effect of LIN28A on the NF-κB pathway, we speculated that miR-370 may exert its inhibitory effect on HCC by suppression of the NF-κB pathway. As expected, miR-370 overexpression decreased, whereas miR-370 inhibition increased, RelA/p65 protein expression and activity of the NF-κB luciferase reporter in HCC cells (Fig. 6A,B), but RelA/p65 mRNA was unaffected (data not shown). RelA/p65 protein levels were also repressed in MHCC-97H xenografts treated with Ad-miR-370 (Supporting Fig. 9A). Consistently, ectopic expression of miR-370 led to down-regulation of NF-κB target genes (i.e., IL-6, TNF-α, and MMP-9), whereas inhibition of miR-370 exerted the opposite effect (Fig. 6C). Reduced expression of these NF-κB target genes was also observed in MHCC-97H xenografts treated with Ad-miR-370 (Supporting Fig. 9B). Interestingly, the effect of miR-370 on RelA/p65 protein level, activity of the NF-κB luciferase reporter, and NF-κB downstream genes in HCC cells could be abrogated by nontargetable LIN28A (Fig. 6D and Supporting Fig. 9C-E). More important, effects of miR-370 on HCC cell migration and invasion were abrogated by oridonin or JSH-23 (Supporting Fig. 10), supporting the hypothesis that the NF-κB pathway is involved in the suppressive effects of miR-370 on HCC.

Figure 6.

Regulatory circuit involving miR-370, LIN28A, RelA/p65, and IL-6 aggravates the malignant phenotype in HCC cells. (A) Western blotting analysis of RelA/p65 protein in HCC cells transfected with miR-370 mimic or miR-370 inhibitor for 48 hours. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (B) Activity of NF-κB luciferase reporter in HCC cells transfected with miR-370 mimic and negative control (NC; left) or miR-370 inhibitor and NC inhibitor (right) for 48 hours. (C) Real-time PCR analysis of expression of the NF-κB regulatory genes, IL-6, TNF-α, and MMP-9, in MHCC-97H cells transfected with miR-370 mimic and NC (left) or miR-370 inhibitor and NC inhibitor for 48 hours (right). Gene expression was normalized against β-actin. (D) Nontargetable LIN28A reversed the effect of miR-370 on RelA/p65 protein levels. HCC cells were infected with lentivirus expressing LIN28A without the 3′ UTR or GFP for 5 days and then transfected with miR-370 mimic and NC for 48 hours. Cells were collected for western blotting analysis of RelA/p65 protein levels. (E) Real-time PCR analysis of miR-370 levels in MHCC-97H cells stimulated with recombinant IL-6 (10 ng/mL). (F) Schematic representation of miR-370/LIN28A/RelA/p65/IL-6 regulatory circuit. P < 0.05; ★★P < 0.01; ★★★P < 0.001 by two-tailed Student t test. Experiments were performed in triplicate, and data are shown as mean ± SD.

IL-6 is a downstream target gene of NF-κB that plays a crucial role in hepatocarcinogenesis.[4, 5] Previous studies reported that IL-6 inhibited miR-370 through modulation of DNA methylation in human cholangiocarcinoma (CCA).[12, 13] In this study, we confirmed that treatment of HCC cells with IL-6 significantly decreased miR-370 levels, followed by an increase in LIN28A protein (Fig. 6E and Supporting Fig. 11A,B). Furthermore, the methylation inhibitor, 5-aza-2′-deoxycytidine markedly increased expression of primary miR-370 (Supporting Fig. 11C), suggesting that miR-370 is down-regulated in HCC in an epigenetic manner. These results indicate that a positive feedback loop, consisting of miR-370, LIN28A, RelA/p65, and IL-6, is involved in the progression of HCC (Fig. 6F).

Down-Regulation of miR-370 Is Associated With Development of HCC in Rats and With Aggressive Tumor Behavior and Short Survival in HCC Patients

We further validated the roles of miR-370 and LIN28A in the development of HCC by using real-time PCR to examine miR-370 and LIN28A mRNA levels in liver tissues from diethylinitrosamine (DEN)-treated rats, 86 paired primary HCC and adjacent nontumorous liver tissues from primary HCC patients with complete clinical data (excluding the 20 pairs of samples referred to above), and 24 healthy human liver tissue samples. miR-370 expression was significantly down-regulated in cirrhotic liver tissues from rats at week 11 after DEN administration, compared to normal rat livers, and was further reduced in HCC tissues, relative to adjacent fibrotic tissues (Fig. 7A). In contrast, LIN28A mRNA was gradually up-regulated during the development of DEN-induced HCC (Fig. 7B). Consistently, miR-370 expression was substantially repressed in the surrounding nontumorous livers from HCC patients, compared to healthy human livers (median, 0.859 and 0.003, respectively; P < 0.001; Mann-Whitney's U test), and was further reduced in HCC tissues (Fig. 7C). In contrast, LIN28A mRNA levels were increased 19-fold in nontumorous livers, compared to healthy livers. LIN28A mRNA was only slightly augmented in HCC tissues, relative to surrounding nontumorous tissues (median, 3.59 × 10−4 and 6.08 × 10−4, respectively; Fig. 7D). Correlation studies displayed that miR-370 levels were inversely correlated with LIN28A and IL6 mRNA levels in human HCC specimens (Fig. 7E), and LIN28A expression was positively correlated with IL-6 expression (Fig. 7E). Specifically, low expression of miR-370 was more likely in human HCC specimens with high levels of LIN28A and IL-6 mRNAs, whereas high expression of LIN28A was more likely in human tissues with high IL-6 levels. More interesting, clinicopathologic analysis demonstrated that down-regulation of miR-370 in human HCCs was significantly correlated with aggressive pathologic characteristics, including larger tumor size (P = 0.028), advanced tumor stage (P = 0.004), presence of venous invasion (P = 0.003), tumor microsatellite formation (P = 0.009), and presence of capsular invasion (P = 0.003; Table 1). To further validate the relation between miR-370 levels and survival of HCC patients, a tissue microarray, which contained 50 paired HCC samples and adjacent cancer-free samples obtained from HCC patients with a median follow-up of 33.5 months (range, 2.0-72.0; standard deviation [SD]: 24.4), was used for in situ hybridization of miR-370. Kaplan-Meier's analysis revealed that lower miR-370 levels were correlated with shorter overall survival (OS) in HCC patients (Fig. 7F).

Figure 7.

miR-370-mediated circuit is perturbed in HCC. (A and B) Real-time PCR analysis of miR-370 (A) and LIN28A mRNA (B) in liver tissues from rats treated with DEN. Samples were collected from normal rat liver (0W), fibrotic liver (11W), liver tumor tissue (19W-T), and adjacent nontumoral tissue (19W-N) at the indicated times (n = 5 in each group). miR-370 levels were normalized against U6. LIN28A expression was normalized against β-actin. Horizontal line indicates median value. P < 0.05; ★★P < 0.01 by Mann-Whitney's U test. (C and D) Real-time PCR analysis of miR-370 (C) and LIN28A mRNA (D) in 24 healthy human liver tissues (NL), 86 paired primary HCCs (HCC), and adjacent nontumorous livers (NT). P < 0.05; ★★★P < 0.001 by Mann-Whitney's U test. (E) Percentages of specimens with low or high miR-370 expression, relative to levels of LIN28A (top) and IL6 (middle) mRNAs. Percentage of specimens with low or high LIN28A expression, relative to IL-6 mRNA level (bottom). The median value of all 86 HCC samples was chosen as the cut-off point for separating HCCs with high expression of miR-370, LIN28A, and IL-6 from HCCs with low levels of these molecules. ★★P < 0.01; ★★★P < 0.001 by chi-square test. (F) Kaplan-Meier's analysis of OS in 50 patients with HCC. The median value of miR-370 level of all 50 HCC samples was chosen as the cut-off point. P = 0.015 by the log-rank test.

Table 1. miR-370 Expression Correlates With Aggressive Phenotype
miR-370
VariablesAll Cases (n = 86)Low Expression (n = 43)High Expression (n = 43)P Valuea
  1. The median value of all 86 HCC samples was chosen as the cut-off point.

  2. a

    Chi-square test.

  3. b

    Mean age.

  4. Abbreviations: HBsAg, hepatitis B surface antigen; AFP, alpha-fetoprotein.

Age, years
≤51.8b3918210.516
>51.8472522 
Sex
Male7238340.234
Female1459 
HBsAg
Absent10551.000
Present763838 
Liver cirrhosis
Yes5127240.510
No351619 
AFP (ng/mL)
≤202811170.167
>20583226 
Differentiation
I-II9270.156
III-IV774136 
Tumor size, cm
≤5174130.028
>5693930 
Tumor multiplicity
Single6931380.058
Multiple17125 
Tumor staging
TNM I-II5320330.004
TNM III-IV332310 
Venous invasion
Absent4214280.003
Present442915 
Tumor microsatellite
Absent5019310.009
Present362412 
Capsular invasion
Present4429150.003
Absent421428 

Discussion

Previous studies have demonstrated reduced miR-370 levels in gastrointestinal stromal tumors,[37] bladder cancer,[38] neuroblastoma cells,[39] and oral squamous cell carcinoma.[15] However, the role of miR-370 in hepatocarcinogenesis remains elusive. The current study revealed that miR-370 expression was gradually reduced during the development of HCC in DEN-treated rats. This decrease in miR-370 was observed in all tested hepatoma cells and in most HCC tumor samples. Moreover, we also demonstrated the suppressive effects of miR-370 on the malignant phenotype of HCC cells both in vitro and in vivo by gain-of-function and loss-of-function experiments. Low expression of miR-370 in HCCs was associated with an aggressive disease phenotype, including advanced tumor stage, larger tumor size, and the presence of venous invasion, microsatellite tumors, and capsular invasion. Importantly, HCC patients with lower levels of miR-370 had shorter OS. All these data suggest that miR-370 may play a crucial role in the carcinogenesis and progression of HCC and may represent a novel therapeutic target and prognostic marker for HCC.

However, our findings seem to be in conflict with other studies that have reported a tumor-promoting function for miR-370.[16-18] miRNAs primarily exert their effects by regulating multiple target mRNAs.[6] Known targets of miR-370 include mitogen-activated protein kinase kinase kinase 8,[12] Wingless-type MMTV integration site family, member 10B,[13] forkhead box M1,[14] insulin receptor substrate 1,[15] transforming growth factor beta receptor II,[16] forkhead box protein O1,[17] neurofibromin 1,[18] and Cpt1α.[19] Most of these targets are implicated in cancer pathogenesis, some as oncogenes and others as tumor suppressors. The opposing effects of miR-370 on tumors may thus be attributed to the different functional natures of their target genes in a given cell type or under specific circumstances,[16] making it a context-dependent effector. The RNA-binding protein, LIN28A, and its paralog, LIN28B, are oncoproteins that are involved in many aspects of malignancies.[23, 24, 26, 28-32] The results of the current study suggested that LIN28A was a bona-fide target of miR-370 and promoted the proliferation, migration, and invasion of HCC cells. More important, the nontargetable LIN28A reversed the miR-370-mediated suppression of HCC cell migration and invasion, suggesting that inhibition of the oncoprotein, LIN28A, contributes to the suppressive effects of miR-370 on HCC. The involvement of LIN28A may thus explain, at least in part, the inhibitory roles of miR-370 in HCC.

Lin28 promotes tumor development in at least two independent manners.[36] First, it selectively blocks the biogenesis of a class of miRNAs, such as let-7.[34] Second, it acts as a post-transcriptional regulator by directly binding specific mRNAs.[21] The Lin28/let-7 double-negative feedback loop is one of the best-characterized examples of the modulation between an miRNA and its post-transcriptional regulator.[36] To our knowledge, let-7 is the only miRNA that has been reported to interact reciprocally with Lin28. The current study demonstrated that LIN28A blocked the biogenesis of miR-370 by binding to its precursor. The mutual regulation of LIN28A and miR-370 thus represents another paradigm of the direct interaction between LIN28 and miRNA. The identification of this novel LIN28A/miRNA loop suggests that the double-negative feedback loop between tumor-suppressive miRNA and LIN28A may be a ubiquitous phenomenon in cancer pathogenesis. On the other hand, direct translational modulation of mRNAs is another crucial mechanism by which Lin28 regulates gene expression.[21] Most documented mRNA targets of LIN28A, including insulin-like growth factor-2, Oct4, cyclin A, cyclin B, cyclin-dependent kinase 4, and human epidermal growth factor receptor 2, are important for cell growth, metabolism, and cancer development.[21, 26, 40] Interestingly, we demonstrated that direct binding of LIN28A to RelA/p65 mRNA promoted the translation of RelA/p65. RelA/p65 is the key subunit of the NF-κB family, which functions as an important promoter of liver carcinogenesis.[4] Thus, post-transcriptional modulation of this crucial oncoprotein represents a novel and important mechanism whereby LIN28A may exert its tumor-promoting function, in addition to its effect on miRNAs.

Most cases of HCC arise in cirrhotic livers with persistent inflammation.[1] Deeper understanding of the mechanistic link between inflammation and HCC would help to identify potential therapeutic targets for HCC. Proinflammatory transcription factors, such as NF-κB and signal transducer and activator of transcription 3, and nontranscriptional elements, such as miRNAs, often cooperate in the regulatory networks that link inflammation to cancers.[28, 41, 42] The results of our current study demonstrated that miR-370 suppressed the NF-κB pathway by inhibiting LIN28A, and the biological functions of both miR-370 and LIN28A were reversed by inactivation of the NF-κB pathway. IL-6 is a well-known target of NF-κB and plays a crucial role in inflammation, wound healing, and hepatocarcinogenesis.[4, 5] Consistent with previous studies in CCA that demonstrated reduced miR-370 expression by IL-6 through modulation of DNA methylation,[12, 13] we also observed a decrease in miR-370 levels, followed by an increase in Lin28A protein levels, induced by IL-6 in HCC cells. Therefore, we identified a novel regulatory circuit in HCC consisting of miR-370, LIN28A, RelA/p65, and IL-6. This regulatory loop is perturbed in human HCC tissues, suggesting that the self-reinforcing regulatory feedback circuit is involved in the progression of HCC.

In conclusion, the present study highlights the biological significance of miR-370 in HCC and elucidates a previously unrecognized molecular mechanism underlying the development of HCC. These findings suggest that early intervention to disrupt this loop may have therapeutic potential for HCC.

Ancillary