Primate‐specific microRNA‐637 inhibits tumorigenesis in hepatocellular carcinoma by disrupting signal transducer and activator of transcription 3 signaling

MiR‐637 (microRNA‐637) is a primate‐specific miRNA belonging to the small noncoding RNA family, which represses gene regulation at the post‐transcriptional expression level. Although it was discovered approximately 5 years ago, its biomedical significance and regulatory mechanism remain obscure. Our preliminary data showed that miR‐637 was significantly suppressed in four HCC cell lines and, also, in most of the hepatocellular carcinoma (HCC) specimens, thereby suggesting that miR‐637 would be a tumor suppressor in HCC. Simultaneously, the enforced overexpression of miR‐637 dramatically inhibited cell growth and induced the apoptosis of HCC cells. The transcription factor, signal transducer and activator of transcription 3 (Stat3), is constitutively activated in multiple tumors, and aberrant Stat3 activation is linked to the promotion of growth and desensitization of apoptosis. Our study showed that Stat3 tyrosine 705 phosphorylation and several Stat3‐regulated antiapoptotic genes were down‐regulated in miR‐637 mimics‐transfected and Lv‐miR637‐infected HCC cells. In addition, miR‐637 overexpression negatively regulated Stat3 phosphorylation by suppressing autocrine leukemia inhibitory factor (LIF) expression and exogenous LIF‐triggered Stat3 activation and rescued cell growth in these cells. A nude mice model also demonstrated the above‐described results, which were obtained from the cell model. Furthermore, we found that LIF was highly expressed in a large proportion of HCC specimens, and its expression was inversely associated with miR‐637 expression. Conclusion: Our data indicate that miR‐637 acted as a tumor suppressor in HCC, and the suppressive effect was mediated, at least in part, by the disruption of Stat3 activation. (HEPATOLOGY 2011)

H epatocellular carcinoma (HCC) is a primary malignancy of the liver 1 and is the fifth most common solid tumor worldwide and the third leading cause of cancer-induced death. 2 Despite much being known about the major etiological factors, such as hepatitis B virus (HBV) and hepatitis C virus infection, the genetic and biochemical understanding of its pathogenesis remains fragmented. Currently, the therapeutic options are limited mainly to liver trans-plantation and surgical resection, and a large majority of nonresectable HCC patients thus represent a significant need for more effective treatment.
MicroRNAs (miRNAs) belong to the family of small noncoding RNAs, which repress gene regulation at the post-transcriptional expression level. 3 With developing technology and more detailed research, it is well known that miRNAs have been involved in various biological processes, including carcinogenesis, 4
From the 1  aberrantly expressed miRNAs play essential roles in HCC progression and directly regulate cell proliferation and survival. 5 For example, let-7, 6 miR-101, 7 miR-122, 8 miR-29, 9 and miR-195 10 were down-regulated in HCC samples and acted as suppressors in tumorigenesis. However, other miRNAs, such as miR-17-5p, 11 miR-18a, 12 and miR-221/222, 13 were observed to function as oncogenes in HCC. miR-637 (microRNA-637) was first identified from colorectal tumor tissue in 2006 14 and is a primate-specific miRNA, and Lin et al. reported that it was expressed in four primate species' genomes. 15 The internal expression of miR-637 is found in many tissues, including liver tissue. However, the precise role of miR-637 is still not fully understood.
In the present study, we investigated the potential association between miR-637 and HCC. We examined the expression level of miR-637 in HCC specimens and cell lines and tested its effect on cell growth and apoptosis. In addition, we also investigated the potential role of miR-637 on HCC tumorigenesis in a murine model. Finally, we explored the underlying mechanism of miR-637 functions in HCC. Therefore, our study will provide a better therapeutic candidate for HCC.
Clinical Specimens. Fifty-two paired primary HCC specimens and matched adjacent nontumor liver tissues were collected from tumor resection (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009) in the Prince of Wales Hospital, the Chinese University of Hong Kong (CUHK; Hong Kong, China). Among them, 41 paired specimens were from male and 9 paired specimens were from female patients. At collection, 39 patients were defined as stage II and 13 patients were classified between stage II and stage III (TNM classification). Informed consent was obtained from each patient, and this study was approved by the Ethics Committee of the Prince of Wales Hospital at CUHK.
Lentiviral miR-637 Overexpression Plasmid Construction and Lentivirus Production. A 295 base-pair (bp) fragment of pre-miR637 encompassing the stem loop was amplified, then cloned into lentiviral vector pLVTHM. The production and purification of the lentivirus were performed as described before. 16 A lentiviral vector expressing a scramble RNA was used as the control.
Colony Formation Assay. HepG2 and Bel7404 cells were infected with lentiviral pre-miR637 (Lv-miR637) or Lv-Ctrl (lentiviral control) and cultured for 72 hours, then replated in six-well plates at 5 Â 10 2 per well. After 2 weeks, cells were washed twice with phosphate-buffered saline, fixed with methanol/acetic acid (3:1; v:v), and stained with 0.5% crystal violet. The number of colonies was counted under a microscope. 17 RNA Extraction, Reverse Transcription, and Quantitative Real-Time Polymerase Chain Reaction. Total RNA was extracted using Trizol reagent (Invitrogen). For the miR-637 expression assay, total RNA was reversely transcribed using the NCode TM miRNA First-Strand complementary DNA Synthesis kit (Invitrogen). To measure messenger RNA (mRNA) levels of LIF and other target genes, total RNA was reversely transcribed using the ImProm-II TM Reverse Transcription System (Promega, Madison, WI). All quantitative real-time polymerase chain reaction (qRT-PCR) samples were performed by using SYBR Green PCR master mix (Roche Applied Science, Indianapolis, IN) on an ABI 7500 Real Time PCR System. Primers are listed in Supporting Table 1. U6 or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an endogenous control, and fold changes were calculated by means of relative quantification (2 ÀDDCt ), as described by Livak et al. 18 Western Blotting. Cell lysates were separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis (10%) and transferred to polyvinylidene fluoride membranes (Millipore, Billerica, MA). Membranes were blocked with 5% skimmed milk for 1 hour and incubated with rabbit polyclonal anti-pStat3 (tyrosine 705 phosphorylated Stat3; Cell Signaling Technology, Inc., Danvers, MA) and anti-Stat3 (signal transducer and activator of transcription 3; Santa Cruz Biotechnology, Santa Cruz, CA) antibody (with a dilution of 1:1,000) overnight. Then, they were followed by the horseradishperoxidise-labeled corresponding immunoglobulin G (1:5,000) for 1 hour. Finally, enhanced chemiluminescence (Pierce, Rockford, IL) was used to visualize the results, and GAPDH was used as an internal control.
Enzyme-Linked Immunosorbent Assay. Protein concentrations of LIF secreted by cells were determined by using enzyme-linked immunosorbent assay (ELISA) kits (Invitrogen). Briefly, the samples and an anti-LIF/human interleukin for DA cells conjugate were added to the coated wells of the 96-well plates, then incubated for 4 hours at room temperature. After being washed 3-5 times, they were incubated with chromogenic solution for 30 minutes. Absorbency was measured at 450 nm on the Wallace Victor-1420 multilabel counter (Perkin-Elmer). All experiments were repeated in quadruple.
Luciferase Assays. A 786-bp fragment of LIF 3 0 -UTR (untranslated region) (nt 231-1016; wild type [WT]) was amplified by PCR and cloned into the pMIR reporter vector (Promega). To investigate whether this UTR would be directly targeted by miR-637 in HCC cells, the predicted target site was mutated by site-directed mutagenesis (Mu). These reporter vectors were referred as to WT and Mu vectors, respectively. For luciferase assays, the WT or Mu vector and miRNAs were cotransfected into HepG2 cells and each experiment was repeated in triplication. Luciferase activity was measured at 28-30 hours after cotransfection by using the luciferase reporter assay system (Promega). Total protein concentration was determined at 595 nm using a Bradford assay (Bio-Rad, Hercules, CA) on a spectrophotometer (Tecan, Grödig, Austria). Luciferase activity was normalized by total protein content.
Subcutaneous Tumor Inoculation. Female BALB/c nude mice (4-6 weeks old) were purchased from the Laboratory Animal Services Center of CUHK. Usage and treatment of nude mice were approved by the Animal Experimental Ethics Committee of CUHK. A total of 1 Â 10 6 Lv-miR637-infected HepG2 cells were injected subcutaneously (SC) into the dorsal flank of nude mice. Each group contained 5 mice. Tumor size was measured every 3 days. When SC tumors reached the volume of 500 mm 3 , mice were sacrificed and tumors were dissected. Tumor volumes (V) were calculated by the formula, V¼1/2 Â S 2 Â L, where S and L are the shortest and longest diameter of the tumor, respectively.
Statistical Analysis. Data are expressed as mean 6 standard deviation (SD). Statistical analysis was performed by using the independent t-test. A P value of less than 0.05 was considered statistically significant.

miR-637 Expression Was Decreased in Human
HCC Cell Lines and Most Clinical Specimens. First, the expression level of miR-637 in a panel of human HCC cell lines (HepG2, Hep3B, Bel7404, and Huh-7) was analyzed, and we found that the expression of miR-637 was decreased in four common HCC cell lines, when compared to the immortalized nontumorigenic cell line, MIHA ( Fig. 1A; P < 0.01). We also examined miR-637 expression in a liver biopsy from a chronic HBV patient. Our results showed that the expression level of miR-637 in the biopsy was approximately 2-fold to that of MIHA cells. To further demonstrate whether the expression level of miR-637 would be related to cell proliferation, MIHA cells were starved for 72 hours by deduction of FBS from 10% to 0.5% in culture medium. We quantified miR-637 levels at different time points before and after resupplemented FBS in culture medium. Compared with nonstarved cells, the expression level of miR-637 was significantly increased by over 3-fold (P < 0.01). However, miR-637 levels were obviously decreased after cell proliferation was stimulated by the supplementing of FBS in culture media (Fig. 1B). Twelve hours after resupplementing with FBS, interestingly, the miR-637 level was significantly reduced by approximately 50%, compared with the starved cells. At 24 hours after resupplementing with FBS, the miR-637 level completely returned to the same level as the healthy MIHA cells. These results indicated that the expression of miR-637 was inversely correlated with the proliferation status of hepatocytes (Fig. 1B).
To demonstrate whether miR-637 was also downregulated in HCC specimens, we examined the expression of miR-637 in 52 paired HCC specimens and nontumor liver tissues. The expression level of miR-637 was more significantly reduced in HCC specimens than in the adjacent nontumor tissues (Fig. 1C) MiR-637 Overexpression Induced Growth Inhibition and Apoptosis in HCC cells. To determine whether miR-637 could be used as a potential anticancer therapeutic candidate, HepG2 and Bel7404 cells were transiently transfected with miRNAs and cell growth was measured. Compared with the NC control, the miR-637 level was increased nearly 1,000 times in HepG2 and Bel7404 cells after transfection with miR-637 mimics (Supporting Fig. 1A). miR-637 suppressed cell growth in HepG2 cells by 35% (P < 0.01) and in Bel7404 cells by 24% (P < 0.01), whereas anti-miR637 enhanced cell growth in HepG2 cells by 25% (P < 0.01) and in Bel7404 cells by 18% (P < 0.01) at 72 hours ( Fig. 2A).
We next used a lentiviral vector to stably restore the expression of miR-637 in HepG2 and Bel7404 cells, then examined its effect on cell growth. Mature miR-637 was strongly expressed in Lv-miR637-infected HepG2 and Bel7404 cells (Supporting Fig. 1B), and the growth inhibition induced by Lv-miR637 infection was similar to that induced by miR-637 mimics transfection in HepG2 and Bel7404 cells (Fig. 2D). In addition, colony formation assay showed that Lv-miR637-infected HepG2 and Bel7404 cells displayed much fewer, and smaller, colonies, compared with Lv-Ctrl-infected cells (Fig. 2E,F; P < 0.01).
LIF Was a Direct Target of miR-637. The abovedescribed significant effects of miR-637 on HCC proliferation and apoptosis prompted us to dissect the potential mechanism of miR-637 in tumorigenesis. As is widely known, miRNAs exert their function through suppressing the expression of their target gene(s). To identify the targets of miR-637, we performed bioinformatics analysis using the miRanda and Findtar online programs. Although hundreds of different targets were predicted by each program, the common targets predicted by miRanda and Findtar associated with cell proliferation and apoptosis were only LIF and Stat3. To further verify these predicted targets, Targetscan also was performed, and results showed that LIF and Stat3 were the targets of miR-637. In our biological experiments, Stat3 was demonstrated to be one pseudotarget of miR-637 through luciferase reporter and western blotting (data not shown) in hepatoma cells. Among these candidate targets, the 3 0 UTR of LIF was predicted to have two binding sequences, which matched the miR-637 seed (Fig. 3A). Therefore, the target sequence (nt 232-1017) of LIF 3 0 UTR (WT) or the site-mutated sequence (Mu) (Fig. 3A) was cloned into the luciferase reporter vector, pMIR. Then, the WT or Mu vector was cotransfected with NC, miR-637, or anti-miR637 into HepG2 cells. For the LIF WT vector, a significant repression of luciferase activity was induced by cotransfection with miR-637, in comparison with the NC cotransfection (P < 0.01), whereas luciferase activity displayed no remarkable difference between cotransfection with NC and with anti-miR637 (Fig. 3B). Moreover, luciferase activity of the LIF Mu vector was unaffected by a simultaneous transfection either with miR-637, NC, or with anti-miR637 (Fig. 3B).
By using qRT-PCR and ELISA analyses, we further found that LIF mRNA and protein expression levels were significantly reduced by miR-637 transfection, whereas anti-miR637 obviously enhanced its expression (Fig. 3C,D). Taken together, all these results strongly suggested that LIF was a direct target of miR-637 in HepG2 cells.
Stat3 Activation Was Disrupted by Enforced Expression of miR-637 Through Suppressing LIF Expression. Interleukin-6 (IL-6) and related cytokines are known to activate Stat3, and LIF is a member of the cytokines, which rapidly induces Stat3 phosphorylation. We have demonstrated that LIF was a direct target of miR-637 and its expression was decreased by miR-637. We, therefore, wondered whether miR-637 would mediate Stat3 phosphorylation. Lv-miR637 and miR-637 mimics were used in the study. We found that the activated Stat3 was significantly reduced in the Lv-miR637-infected HepG2, compared to the Lv-Ctrl-infected, cells, thus suggesting that miR-637 blocked the tyrosine (Y705) phosphorylation of Stat3 (Fig. 4A). A similar result was observed in HepG2 cells, which had been transfected with miR-637 mimics (Fig. 4B). Moreover, we investigated whether anti-miR637 would promote the expression of phosphorylated Stat3 in HepG2 cells (Fig. 4C). In addition, the decreased pStat3 level in miR-637-transfected HepG2 cells was associated with the decreased expression of several Stat3-regulated cell-survival target genes, including Cryab, Mcl1, and Bcl-xl, 19 whereas anti-miR637 promoted their expressions at mRNA level (Fig. 4D).
To elucidate whether the growth-suppressive and apoptotic effects of miR-637 were mediated by the repression of LIF in HepG2 cells, we performed gainand loss-of-function studies. We first silenced LIF by using its specific siRNA to test whether the reduced expression could mimic the suppressive effect of miR-637. LIF expression was reduced by specific small interfering RNA of LIF (LIF SiRNA)1 (LIF-Si1) and SiRNA2 (LIF-Si2) at the mRNA and protein levels (Supporting Fig. 3A,B). HepG2 cells were transfected with LIF-Si1, and we then examined cell growth and apoptosis. We found that LIF knockdown led to obvious cell-growth inhibition, and that it induced more apoptotic cells (33%) than those that had been induced by miR-637 mimics (28.9%; Fig. 5A,D,E). Subsequently, we evaluated whether LIF knockdown could suppress the phosphorylation of Stat3. HepG2 cells were transfected with LIF-Si1 or LIF-Si2 for 48 hours, then, cells were harvested for western blotting. Both LIF-Si1 and LIF-Si2 decreased Stat3 phosphorylation in HepG2 (Fig. 5B), and they also suppressed the expression of several antiapoptotic Stat3-regulated genes at the mRNA level (Fig. 5C).
On the other hand, exogenous LIF was introduced for the gain-of-function studies. We treated the infected cells with exogenous LIF (20 ng/mL), and found that exogenous LIF counteracted the cell-growth suppression of miR-637 (Fig. 6A) and that it activated Stat3 within 30 minutes (Fig. 6B). A quantitative analysis of apoptotic cells showed that LIF introduction suppressed the apoptosis induced by miR-637 in HepG2 cells (Fig. 5D,E). Also, the expression levels of Stat3-regulated genes in Lv-miR637-infected HepG2 cells were rescued by exogenous LIF (Fig. 6C). Persistent Stat3 activation frequently happens in various tumor cells, 20,21 and miR-637 functions as an anticancer candidate by suppressing activated Stat3 expression in HCC cells.
Taken together, we suggested the following regulatory signaling: miR-637 suppressed LIF expression, then it blocked Stat3 phosphorylation, which sustained target genes, such as Cryab, Mcl1 and Bcl-xl activity, which, in turn, inhibited cell growth and induced apoptosis (Fig. 4E).
MiR-637 Suppressed Tumor Growth of the HepG2 Cells in Nude Mice. HepG2 cells were infected with Lv-miR637, then were SC injected into the dorsal flank of nude mice. Strikingly, the average tumor volume and weight of the Lv-miR637-infected group were markedly reduced by more than 60% (Fig.  7A,B). At 18 days after inoculation, those mice injected with Lv-Ctrl-infected cells carried larger burdens (Fig. 7C,D). For a better understanding of the molecular mechanism of miR-637 on tumorigenesis in vivo, we examined the expression of Ki-67, pStat3 (Y705), and Bclxl in tumor tissues by using immunohistochemical analysis. Expression levels of Ki-67, pStat3, and Bcl-xl were significantly lower in tumor tissues of mice treated with Lv-miR637, in comparison with mice treated with Lv-Ctrl (Fig. 7E) (brown spots by counterstaining with H&E). These results provided strong evidences to support that miR-637 suppressed tumorigenesis of hepatoma cells in vivo by regulating Stat3 activity.
Up-regulation of LIF Expression Was Inversely Associated With miR-637 Expression in HCC Tissue Specimens. We further examined the expression levels of LIF in 52 cases of HCC patients. The LIF expression level was up-regulated in 69%, whereas the miR-637 expression level was down-regulated in 66% (n ¼ 52) of HCC tumors (T), in comparison with the matched adjacent nontumor liver tissues (N) (Fig. 8A).
Then, we assessed whether miR-637 would be implicated in LIF expression. A significant inverse association between the expression of miR-637, and that of LIF, was observed in the same 52 paired HCC specimens (r ¼ À0.6547; Fig. 8B).

Discussion
MiR-637 is a primate-specific miRNA with a higher expression level in livers, according to the miRbase database. However, the role of miR-637 was inadequately reported in any aspect, even in the liver. In the present study, we first identified miR-637 as a putative tumor suppressor in HCC cells and tissues and provided a basis for the potential application of miR-637 in cancer therapy.
Despite the fact that miR-637 was discovered several years ago, its functions remain obscure. Our results first displayed the suppressed expression of miR-637 in four HCC cell lines and tumor tissues and were reversely correlated with cell proliferation (Fig. 1). To reveal the role of miR-637 in HCC cells, we examined the effect of miR-637 on cell growth in HepG2 and Bel7404 cells. Our results showed that miR-637 could inhibit cell growth and induce apoptosis (Fig. 2), and suppress oncogenesis in a nude mice model with HCC xenograft (Fig. 7), thus demonstrating that it might be a tumor suppressor in HCC.
Regarding the transcription factor, Stat3, it has been reported that aberrant Stat3 activation was linked to the promotion of growth and desensitization of apoptosis in a variety of human tumors. 21,22 Also, Stat3 is constitutively activated in multiple tumors, such as glioblastoma, colorectal carcinoma, and HCC, and acts as an oncogene. 23,24 Activated Stat3, in which the specific tyrosine residues are phosphorylated (Y705) by IL-6 and associated cytokines, is implicated in the proliferation of, and metastasis in, HCC. 25 In addition, Stat3 phosphorylation is highly positive in HCC biopsies. 26 Moreover, SOCS-1, a negative regulator of Stat3, when silenced by methylation, resulted in the constitutive activation of Stat3 in the development of primary human HCC. 27 Collectively, HCC harbors constitutively active Stat3, and the inhibition of active Stat3 expression might be a potential candidate for HCC. Therefore, blocking Stat3 signaling by SiRNA, antisense oligonucleotides, or specific inhibitors has been demonstrated to suppress tumor growth, promote apoptosis, and improve survival. 22 Our data uncovered that miR-637 overexpression decreased activated Stat3 levels, whereas it really did not affect the expression of Stat3 (Fig. 4), suggesting the disruption of Stat3 phosphorylation. In addition, the suppressed expression of the antiapoptotic target genes of Stat3, including Cryab, Mcl1, and Bcl-xl, in miR-637-transfected HepG2 cells promoted sensitivity to apoptosis.
MiR-637 has been identified as a potent inhibitor of the Janus activated kinase (Jak)/Stat3-signaling pathway. Next, we wanted to know what would regulate the inactivation of Stat3 in miR-637-transfected HepG2 cells. It is well known that IL-6 family cytokines, such as IL-6, LIF, and oncostatin M (OSM), function as autocrine or paracrine growth factors to activate Stat3. 25,28,29 In our bioinformatics assays, neither IL-6 nor OSM was predicted as a target of miR-637. LIF was predicted to be a direct target gene of miR-637 by means of bioinformatics analysis. In addition, our data strongly confirmed that it was a direct target of miR-637 in HCC (Fig. 3). LIF is a multipotent cytokine that binds to the LIFR/gp130 heterodimerized complex that exerts its function through the Jak/Stat3 pathway. 30 Human carcinoma cell lines, including those of liver cancer, produce LIF, 31 and LIF stimulates cell proliferation by the paracrine and autocrine pathways. 32 To further examine the functions of LIF in HCC, we found that LIF silence induced cellgrowth inhibition and the inactivation of Stat3 similar to the phenotypes induced by miR-637 restoration (Fig. 5). Meanwhile, exogenous LIF could rescue the suppressive effect induced by miR-637 and activate Stat3 phosphorylation (Fig. 6), suggesting that LIF is an important factor to support hepatoma cell growth. However, exogenous LIF had only a moderate effect on the proliferation of HepG2 cells and Lv-Ctrlinfected HepG2 cells (Supporting Fig. 4A) and slightly increased Stat3 activation within 30 minutes (Supporting Fig. 4B). This paradox might be caused by the high level of internal LIF in these highly differentiated cancer cells. Therefore, additional exogenous LIF might be somehow redundant and showed less sensitivity.
We have discerned the molecular mechanism involved in the suppression of activated Stat3 by an miR-637 in vitro experiment. Furthermore, an ectopic HCC model was applied to evaluate the effect of miR-637 on HCC in vivo. Our data revealed that Lv-miR637-infected HepG2 cells formed smaller tumors and had lower expression levels of Ki-67 (a cellular proliferation marker), phosphorylated Stat3, and Bclxl, in comparison with the Lv-Ctrl-infected HepG2 cells (Fig. 7). Therefore, miR-637 achieved a beneficial antitumor effect on HCC by suppressing the activation of Stat3. This, therefore, suggests that miR-637 likely acted as an inhibitor of Jak/Stat3 signaling.
In addition, we found that LIF was expressed in human HCC, and that tumors with a low level of miR-637 tend to have high levels of LIF. This, there-fore, indicates that miR-637 might be responsible for LIF expression in human HCC (Fig. 8). A previous study showed that LIF was expressed in human gliomas, and that it subsequently activated the Jak/Stat pathway. 33 This cytokine was also expressed in other tumor types, such as thyroid cancer and rhabdomyosarcomas, 34,35 and that it acts as a mediator of Stat3 activation in some malignant tumors. 28,33,35 We, therefore, suggest that miR-637 would function as a therapeutic target or biomarker in HCC and would thus hold out a significant promise for the development of more successful therapeutic strategies or in improving the diagnoses of HCC patients. Further studies are needed to identify this assertion by using larger HCC samples. In addition, we also need to define whether the molecular mechanism identified in the present study would be functional in other tumor types.
In conclusion, we advance that miR-637 suppresses the constitutive activation of Stat3 by targeting autocrine LIF in human HCC. Furthermore, this suppressive effect is important for the growth of HCC cells in vitro and for the development of liver cancer in vivo. Therefore, miR-637 might be successful against HCC because it inhibits the activation of Stat3 and disrupts Jak/Stat3 signaling partially through suppressing autocrine LIF expression. Fig. 8. LIF was up-regulated in HCC specimens and inversely correlated with miR-637 expression levels. (A) Expressions of miR-637 and LIF in human paired HCC specimens. LIF abundance was normalized to GAPDH and miR-637 expression was normalized to U6. (B) Statistically significant inverse association between miR-637 and LIF expression at mRNA levels in HCC specimens (r ¼ À0.6547).