MiR-21 Simultaneously Regulates ERK1 Signaling in HSC Activation and Hepatocyte EMT in Hepatic Fibrosis

Background MicroRNA-21 (miR-21) plays an important role in the pathogenesis and progression of liver fibrosis. Here, we determined the serum and hepatic content of miR-21 in patients with liver cirrhosis and rats with dimethylnitrosamine-induced hepatic cirrhosis and examined the effects of miR-21 on SPRY2 and HNF4α in modulating ERK1 signaling in hepatic stellate cells (HSCs) and epithelial-mesenchymal transition (EMT) of hepatocytes. Methods Quantitative RT-PCR was used to determine miR-21 and the expression of SPRY2, HNF4α and other genes. Immunoblotting assay was carried out to examine the expression of relevant proteins. Luciferase reporter assay was performed to assess the effects of miR-21 on its predicted target genes SPRY2 and HNF4α. Primary HSCs and hepatocytes were treated with miR-21 mimics/inhibitors or appropriate adenoviral vectors to examine the relation between miR-21 and SPRY2 or HNF4α. Results The serum and hepatic content of miR-21 was significantly higher in cirrhotic patients and rats. SPRY2 and HNF4α mRNA levels were markedly lower in the cirrhotic liver. MiR-21 overexpression was associated with enhanced ERK1 signaling and EMT in liver fibrosis. Luciferase assay revealed suppressed SPRY2 and HNF4α expression by miR-21. Ectopic miR-21 stimulated ERK1 signaling in HSCs and induced hepatocyte EMT by targeting SPRY2 or HNF4α. Downregulating miR-21 suppressed ERK1 signaling, inhibited HSC activation, and blocked EMT in TGFβ1-treated hepatocytes. Conclusions MiR-21 modulates ERK1 signaling and EMT in liver fibrosis by regulating SPRY2 and HNF4α expression. MiR-21 may serve as a potentially biomarker as well as intervention target for hepatic cirrhosis.


Introduction
Hepatic fibrosis is characterized by excessive production and deposition of the extracellular matrix (ECM), leading to the destruction of the normal hepatic parenchyma and disruption of the liver structure [1][2][3][4]. A well-documented event critical to the development of hepatic fibrosis is the activation and proliferation of resident hepatic stellate cells (HSCs) [5][6][7]. Recent evidence also implicates activated fibroblasts in hepatic fibrosis. These activated fibroblasts are transformed from hepatocytes and biliary epithelial cells through the epithelial-mesenchymal transition (EMT) and contribute to liver fibrogenesis [8][9][10].
The extracellular signal-regulated kinase 1 (ERK1) signaling pathway is implicated in both HSC activation and EMT of hepatocytes and biliary epithelial cells. Specifically, activation of the ERK1 signaling pathway promotes HSC activation [11][12][13]. ERK1 is a critical player in this signaling pathway. Our previous study showed that suppression of ERK1 expression could inhibit HSC activation and block EMT of hepatocytes and biliary epithelial cells [14]. Hepatocyte nuclear factor 4a (HNF4a), an important transcriptional factor for hepatocyte differentiation and function, is downregulated in human cirrhotic liver [15]. In a rat model of cirrhosis, HNF4a significantly suppresses EMT of hepatocytes and alleviates dimethylnitrosamine-induced fibrosis [16]. These findings together indicate that both the ERK1 signaling pathway and EMT may play critical roles in hepatic fibrogenesis and represent a promising therapeutic target in liver fibrosis.
MicroRNAs (miRNAs) are a class of endogenous, small (18-24 nucleotides), non-coding single-stranded RNAs that negatively regulate gene expression through binding to the 39-untranslated region (UTR) of target mRNAs [17]. Dysregulation of miRNAs contributes to the development of a variety of diseases, including liver fibrosis [18,19]. MiRNA-21 is highly expressed at the onset of fibrosis in many organs, including the human liver [20][21][22]. Importantly, miR-21 stimulates the proliferation and activation of fibroblasts in different organs with fibrosis, which may involve the PTEN/Akt, NF-kappa B (NF-kB) and ERK1 signaling pathways [20][21][22][23][24][25]. Additional studies further implicate miR-21 in the activation of HSCs [21]. Moreover, our previous study showed that TGFb1 negatively regulated sprouty2 (SPRY2) expression in HSC activation and ectopic HNF4a expression in the hepatocytes of rats with fibrotic livers was associated with blocked EMT and reduced miR-21 expression. In this study, a computational algorithm analysis suggested that SPRY2 and HNF4a contain putative miR-21 binding sites. Based on these findings, we speculated that miR-21 could modulate hepatic fibrogenesis by targeting SPRY2 and HNF4a in HSCs and hepatocytes. In the current study, we examined serum and hepatic content of miR-21 in patients with liver cirrhosis and in rats with dimethylnitrosamine-induced hepatic cirrhosis. Effects of miR-21 on SPRY2 and HNF4a in HSCs and hepatocytes were also examined.

Ethical statements
Written informed consent was obtained from all study participants. Acquisition and use of human tissue specimens or sera were carried out in accordance with established institutional and national ethical guidelines regarding the use of human tissues for research. All experimental procedures were performed in accordance with the Regulations for the Experimental Use of Animals by the State Council of the People's Republic of China. The animals were sacrificed under sodium pentobarbital anesthesia, with efforts to minimize animal suffering in accordance to the ARRIVE guidelines (http://www.nc3rs.org.uk/page. asp?id=1357).

MiRNAs and adenoviral vectors
MiR-21 mimic, miR-21 inhibitor (anti-miR-21), the control miRNA and small interfering RNAs (siRNAs) against SPRY2, ERK1 and HNF4a were synthesized by GenePharma (Shanghai, China). The sequences are listed in Table S1. Replication-deficient E1 and E3 adenoviral vectors, AdERK1, AdSPRY2, AdHNF4a and the control vector-AdGFP that express ERK1, SPRY2, HNF4a, and green fluorescent protein (GFP), respectively, as well as adenoviral vectors AdshERK1 (expressing siRNA targeting ERK1 mRNA), AdshHNF4a (expressing siRNA against HNF4a) and AdshNC (containing scrambled siRNA) were prepared as previously described [14,16]. The cDNA fragments of ERK1, SPRY2 and HNF4a were generated through RT-PCR from rat HSCs or hepatocytes as detailed elsewhere in the text. The fragments of siRNA against ERK1 and HNF4a (GenePharma) were synthesized, and all of the fragments were inserted into the pAd-Track-Shuttle vector carrying GFP gene respectively, to yield pAd-Track vectors (pAd-Track-ERK1, pAd-Track-SPRY2, pAd-Track-HNF4a, pAd-Track-GFP, pAd-Track-shERK1, and pAd-Track-shHNF4a). Homologous recombination was performed using 1 mg PmeI-linearized pAd-Track vectors with 0.1 mg supercoiled circular viral backbone vector pAd-Easy-1 in Escherichia coli BJ5183 cells. After packaging in 293 cells, recombinant replication-deficient adenoviruses were generated. These adenoviruses could express both of the exogenous genes and GFP, allowing direct observation of green fluorescence with microscopy to evaluate the efficiency of infection. None of the adenoviral vectors contained the 39-UTR region of the related genes.

Serum and tissue acquisition
Human sera were obtained from 20 healthy individuals and 20 patients with liver cirrhosis (Child-Pugh classification level: B or C). Venous blood samples were collected and stood at room temperature for about 30 min before centrifugation at 8206g for 10 min at 4uC. The supernatant was transferred into new tubes and centrifuged at 16,0006g for 10 min at 4uC to completely remove any cell debris. The resulting serum was stored in new Eppendorf tubes at 280uC. Normal or cirrhotic human liver tissues (n = 7 each) were also obtained. The clinical features of the study subjects are shown in Table 1. Additionally, serum and liver tissue specimens were prepared from male Sprague-Dawley (SD) rats (weighing 200-250 g, n = 10) with liver cirrhosis induced by intraperitoneal injection with 1% dimethylnitrosamine (10 mg/kg; Sigma, St. Louis, MO) for 3 consecutive days per week up to 4 weeks [16].

Cells, transfection and infection
Rat HSC cell line HSC-T6 [26] (kindly provided by Dr. SL. Friedman), which is considered activated mature fibroblast with typical features of myofibroblasts, and human embryonic kidney cell line HEK-293 (ATCC, Manassas, VA) were cultured in Dulbecco's Modified Eagle's Medium (DMEM) (Gibco) with 10% fetal calf serum (FBS) (Gibco). Primary HSCs and hepatocytes were isolated from male SD rats as previously described [16]. Briefly, rats were starved for 18 h and anesthetized with an intraperitoneal injection of 10% chloral hydrate (35 mg/kg). The portal vein was identified and used to perfuse the liver with Krebs Ringer buffer containing EDTA and then with collagenase type IV solution. When the liver showed swelling, lobe bubbling, and the appearance of cellular debris, the liver tissues were bluntly dissected in ice-cold Krebs Ringer Buffer to obtain cell suspension. For HSCs isolation, the cell suspension was repeatedly centrifuged at 506g (2 min each round) until no visible pellet was observed, followed by filtering through a 250 mm mesh filter to remove tissue fragments. Then, the cell suspension was centrifuged at 2006g for 10 min to yield a pellet of non-parenchymal cells containing the HSCs. At last, HSCs were isolated from the non-parenchymal cell suspension by 8.2% Nycodenz (Nycomed, Oslo, Norway) density gradient centrifugation at 14006g and extensive washing. The cells were seeded into dishes (35-mm) at 1610 6 cells/dish and cultured in DMEM with 10% FBS. For hepatocytes isolation, the cell suspension was filtered through a 250 mm mesh filter, and then through a 60 mm mesh filter to obtain single cell suspension. The suspension was divided into several 50 ml conical tubes, washed and centrifuged at 1506g for 3 min at 4uC for three times. The hepatocytes were seeded into dishes (35-mm) at 1610 6 cells/dish and cultured in hepatocyte culture medium as previously described [16]. Trypan blue exclusion showed that cell viability was greater than 90%. For detection of the expression of miRNA and its target genes in activated HSCs and EMT hepatocytes, the cells were cultured for 48 h, followed by stimulating with TGFb1 (3 ng/ml; 48 h for hepatocytes, 7 days for HSCs). The cell culture media with TGFb1 was changed every 3 days. In order to determine the effects of upregulating miR-21 on HSC activation and hepatocyte EMT, cells were cultured in medium without TGFb1 and transfected with miR-21 mimics (100 pmol), or the control miRNA (100 pmol) for 48 h using Lipofectamine TM 2000 (Invitrogen, Carlsbad, CA) as instructed by the manufacturer. To examine the effect of downregulating of miR-21 on HSCs and hepatocytes, the cells were treated with TGFb1 (3 ng/ml) for 48 h for hepatocytes and 5 days for HSCs. Then, the cells were transfected with miR-21 inhibitors (100 pmol) or its control miRNA (100 pmol) for 48 h. Additionally, cells were infected with appropriate adenoviral vectors at a multiplicity of infection (MOI) of 50 as detailed elsewhere in the text.

Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR)
Total cellular RNA was extracted from cells, liver tissues or serum using the mirVana RNA isolation kit (Invitrogen). To examine the level of miR-21 in tissues, RNA (2 mg) was polyadenylated with ATP using a poly (A) polymerase (New England BioLabs, Ipswich, MA) at 37uC for 2 h, reversetranscribed with Super Script III (Invitrogen) and a poly (T) adapter, and subjected to SYBR Green-based real-time PCR analysis (Takara, Dalian, China). To examine the level of miR-21 in serum, we used RT and qPCR kits specifically for accurate miRNA analysis (Applied Biosystems). A TaqMan microRNA Reverse Transcription Kit (Applied Biosystems, USA) was used to perform RT reactions. Then, the RT reactions were incubated for 30 min at 16uC, 30 min at 42uC, 5 min at 85uC, followed by maintaining at 4uC. For real-time PCR, 1.33 ml diluted RT products were mixed with 10 ml of 26 Taqman PCR master mixture (No AmpErase UNG), 1 ml TaqMan MicroRNA Assay and Nuclease-free water in a final volume of 20 ml. All reactions were run on the ABI 7300 (Applied Biosystems, USA) with the following conditions: 95uC for 30 s followed by 40 cycles at 95uC for 10 s, and 60uC for 30 s. For detection of mRNA transcript levels, total RNA was reverse-transcribed using Super Script III with random primers and subjected to SYBR Green-based realtime PCR analysis. MiR-21 was normalized against miR-238 in the serum and against Rnu6-2 in liver tissues. The mRNA was normalized against b-actin. All primer sequences are listed in Table S2.

Western blotting assays
Proteins were extracted with RIPA lysis buffer (Bioteke Co, Beijing, China) containing a cocktail of protease inhibitors and phosphatase inhibitors, and separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane (HAHY00010, Millipore, Billerica, MA). The membrane was probed with specific primary antibodies (Table S3) overnight at 4uC, followed by incubation with appropriate donkey-anti-mouse or donkey-anti-rabbit secondary antibodies (GIBCO BRL, Rockville, CA). Protein bands were visualized using the Odyssey infrared imaging system (LI-COR Biotechnology, NE). GADPH was used as a loading control.

Cell migration assays
The migration of HSC-T6 cells was examined using a transwell assay [27]. Briefly, 1610 4 cells transfected with miR-21 inhibitors or the control miRNA were seeded in transwell inserts with 8.0mm-pore transwell filters (BD, NJ), and the wells under the inserts were coated with fibronectin (50 mg/ml) and 500 ml DMEM containing 10% FBS was added as chemoattractants. After 24 and 48 h, the migration of HSCs was quantified by counting the number of cells stained by 0.5% crystal violet.

Luciferase reporter assay
Reporter plasmids psiCHECK-2-SPRY2 and psiCHECK-2-HNF4a were constructed as described previously [28]. The 280bp fragment of the SPRY2 39-UTR containing the miR-21 target sequence was amplified from total DNA of primary rat HSCs by PCR and cloned into the psiCHECK-2 dual luciferase reporter plasmid (Promega, Madison, WI) at the 39-end of the coding sequence of Renilla reniformis luciferase to produce psiCHECK-2-SPRY2. Similarly, the luciferase reporter plasmid psiCHECK-2-HNF4a containing a 280-bp fragment of the HNF4a 39-UTR with the miR-21 target sequence was prepared. To determine sequence specificity, we also constructed the plasmids psiCHECK-2-mt-SPRY2 or psiCHECK-2-mt-HNF4a in which the conserved targeting sequence of miR-21 was mutated (from AUAAGCU to CUCGAGC). MiR-21 mimics and reporter plasmids were cotransfected into HEK-293 cells using Lipofectamine TM 2000. Renilla constructs were transfected as an internal control. After 48 h incubation, firefly luciferase and Renilla luciferase activities were measured using a luciferase assay system (Promega). All luciferase activity readings were normalized relative to the activity of the Renilla luciferase control. All experiments were performed in triplicate.

Statistical analysis
Analysis of variance (ANOVA) and Student's t-test were used for comparison of normally distributed data between the groups and paired data. Data not normally distributed were compared using the Mann-Whitney test. All results were reported as mean 6 SD of at least three independent experiments and a P value ,0.05 was considered statistically significant.

Results
Liver fibrosis is associated with broad changes in the expression of miR-21 and its targeted genes, ERK1 signaling and EMT-associated genes We first examined the expression of miR-21 in the serum of patients with hepatic cirrhosis or rats with dimethylnitrosamineinduced liver cirrhosis, which was confirmed by histopathologic examination (Fig. S1A and S1B). The serum content of miR-21 was 2.3 fold higher in patients with liver cirrhosis than in healthy control subjects (Fig. 1A). Significantly higher serum miR-21 content was also seen in the cirrhotic rats (Fig. 1B). Consistently, we also observed a marked increase in the miR-21 content in the cirrhotic liver tissues of humans and rats ( Fig. 1C and 1D). Our quantitative RT-PCR further revealed that the mRNA transcript levels of ERK1 and its downstream target ribosomal protein S6 kinase 2 (RSK2) were significantly higher in the cirrhotic liver tissues ( Fig. 1C and 1D). The mRNA transcript levels of EMTassociated gene vimentin were approximately 2.6 to 3.1 fold higher in the cirrhotic liver of both humans and rats ( Fig. 1C and 1D) while those of E-cadherin were markedly lower compared to normal livers ( Fig. 1C and 1D). Screening for targets of miR-21 using Target Scan and PicTar yielded SPRY2 and HNF4a as the putative targets. Our RT-PCR assays further showed that the mRNA transcript levels of both SPRY2 and HNF4a were markedly lower in the cirrhotic liver of both humans and rats ( Fig. 1C and 1D). Additionally, TGFb1 markedly increased miR-21 levels in primary HSCs and hepatocytes of rats in a time-dependent manner ( Fig. 1E and 1F). Similar increase was observed in the transcript levels of ERK1 and RSK2 and fibrogenesis-associated genes a-smooth muscle actin (a-SMA) and tissue inhibitor of metalloproteinase 1 (TIMP1) in primary HSCs (Fig. 1E). The mRNA transcript levels of SPRY2 in primary HSCs (Fig. 1E) and HNF4a in primary hepatocytes (Fig. 1F) were noticeably decreased. Meanwhile, the mRNA transcript levels of vimentin were markedly higher while those of E-cadherin were markedly lower in the primary hepatocytes. Furthermore, TGFb1 caused a marked reduction in certain liver specific genes, including albumin (ALB) and cytochrome P450 (CYP) 1a2 (CYP1a2) in primary hepatocytes (Fig. 1F). Western blotting assays additionally showed that SPRY2 and HNF4a expression was markedly suppressed in TGFb1-treated HSCs or hepatocytes ( Fig. S2A and S2C) and cirrhotic rat liver tissues ( Fig. S2B and S2D).

MiR-21 targets the 39-UTR of SPRY2 and HNF4a
As our computational algorithm analysis showed that SPRY2 and HNF4a were candidate target genes of miR-21, we investigated whether miR-21 directly targeted SPRY2 and HNF4a expression by co-transfecting SPRY2 or HNF4a 39-UTR reporter plasmids containing putative miR-21 binding sites and miR-21 mimics into HEK-293 cells. Luciferase assays showed that miR-21 mimics caused a 42.6% and 41.4% reduction in luciferase activities of cells transfected with SPRY2 39-UTR ( Fig. 2A) and HNF4a 39-UTR reporter plasmids (Fig. 2B), respectively. Co-transfection of miR-21 mimics and SPRY2 or HNF4a 39-UTR reporter plasmids containing mutant miR-21 binding sites, on the other hand, failed to suppress luciferase activities ( Fig. 2A and 2B). These results indicated that miR-21 directly targeted SPRY2 and HNF4a mRNA to modulate their expression.

MiR-21 regulates ERK1 signaling in HSCs by targeting SPRY2
We further examined the effects of miR-21 and SPRY2 on each other and on ERK1 signaling in HSCs. We found that miR-21 mimics caused a marked reduction in the mRNA transcript and protein levels of SPRY2 in primary quiescent HSCs (Fig. 3A-3C). Meanwhile, we observed that miR-21 mimics caused a noticeable increase in the mRNA transcript and protein levels of ERK1 and RSK2. In addition, AdSPRY2 infection of primary HSCs treated with TGFb1 for 48 h suppressed the expression of miR-21 accompanied by a marked decline in the mRNA transcript levels of ERK1 and RSK2 (Fig. 3D), whereas miR-21 level slightly increased after primary HSCs were transfected with siRNA against SPRY2 (Fig. S3). These results suggest the presence of a modulatory loop between miR-21 and SPRY2. We then transfected primary quiescent HSCs with miR-21 mimics followed by AdSPRY2 infection 48 h later. We found that SPRY2 significantly attenuated miR-21-induced increase in the mRNA transcript levels of ERK1 and RSK2 (Fig. 3E), suggesting that SPRY2 may function downstream of miR-21. Moreover, AdERK1 is synergetic with miR-21 mimics for the activation of ERK1 signaling in primary HSCs, and AdshERK1 could enhance the inhibitory effect of miR-21 inhibitor on the ERK1 signaling pathway in activated HSCs (Fig.S4A and S4B).

MiR-21 promotes EMT of primary hepatocytes by targeting HNF4a
Given our current finding that miR-21 directly targeted HNF4a and our previous finding that HNF4a significantly suppressed the EMT of hepatocytes [16], we further examined the effects of miR-21 and HNF4a on each other and on the EMT of primary hepatocytes. As shown in Fig. 4A, 4B and 4C, miR-21 mimics caused a marked reduction in the mRNA and protein levels of HNF4a in primary isolated hepatocytes along with a noticeable increase in the mRNA and protein levels of vimentin and a significant decrease in the mRNA transcript levels of E-cadherin. Similar results were observed in TGFb1-treated hepatocytes (Fig.  S5). AdHNF4a infection of primary hepatocytes that had been treated by TGFb1 for 48 h decreased miR-21 levels, along with a significant reduction in the mRNA transcript levels of vimentin and a noticeable increase in the mRNA transcript levels of Ecadherin (Fig. 4D). We then transfected primary hepatocytes with miR-21 mimics followed by AdHNF4a infection 48 h later. We found that HNF4a significantly suppressed miR-21 and the mRNA transcript levels of vimentin while it markedly increased the mRNA transcript levels of E-cadherin (Fig. 4E), indicating that HNF4a may function downstream of miR-21. Furthermore, we transfected primary hepatocytes with miR-21 mimics followed by AdshHNF4a infection 48 h later. We found that AdshHNF4a slightly enhanced miR-21 level in primary hepatocytes (Fig. 4F). AdshHNF4a and miR-21 alone or in combination reduced the mRNA transcript levels of E-cadherin, but increased the mRNAtranscript levels of vimentin. These results suggest the presence of a modulatory loop between miR-21 and HNF4a.

Downregulation of miR-21 targets ERK1 signaling and SPRY2 in HSC activation and blocks EMT of TGFb1treated hepatocytes by targeting HNF4a
To determine whether downregulating miR-21 impacted on ERK1 signaling and HSC activation, we transfected TGFb1treated primary HSCs and HSC-T6 cells with miR-21 inhibitors and found that miR-21 inhibitors markedly reduced the levels of miR-21 along with an apparent decline in the mRNA transcript levels of ERK1 and its downstream target RSK2 (Fig. 5A and 5D). Furthermore, miR-21 inhibitors significantly upregulated the mRNA transcript levels of SPRY2. These findings were further confirmed by immunoblotting assays (Fig. 5C and 5F). Expectedly, miR-21 inhibitors markedly reduced the expression of profibrotic genes in activated HSCs, including TGFb1, a-SMA, Col I and III, and TIMP1, and augmented the expression of MMP-9 and MMP-13 (Fig. 5B, E and 5F). The migration ability of HSC-T6 cells was also significantly suppressed by miR-21 downregulation (Fig. 5G and 5H). The results showed that downregulation of miR-21 could inhibit ERK1 signaling and HSC activation by targeting SPRY2. Moreover, miR-21 inhibitors also caused a significant increase in the mRNA and protein levels of HNF4a, along with enhanced mRNA transcript levels of Ecadherin and certain liver specific genes, and attenuated the expression of a cluster of mesenchymal markers and profibrotic genes in TGFb1-treated hepatocytes (Fig. 5I-5K). The data revealed that decreased miR-21 expression could block the TGFb1-induced EMT process in hepatocytes by directly promoting HNF4a expression.

Discussion
It is widely accepted that ERK1 signaling plays a vital role in HSC activation in hepatic fibrogenesis [13,14]. Additionally, hepatocyte undergoing EMT contributes to the accumulation of fibrogenic myofibroblasts [10,16]. Blocking EMT may hold the promise of reversing liver fibrogenesis. Our previous study demonstrated that targeted co-suppression of ERK1 signaling in HSCs and EMT of hepatocytes could inhibit transformation of liver parenchymal and mesenchymal cells to activated fibroblasts, and significantly attenuate hepatic fibrosis [14,16]. These results prompted us to speculate that the therapeutic strategy targeting the ERK1 signaling pathway in HSCs and EMT of hepatocytes may render the fibrotic liver more amenable to treatment compared with conventional therapy.
In the current study, we observed markedly increased miR-21 contents in the serum and hepatic tissues of patients with liver cirrhosis and rats with dimethylnitrosamine-induced hepatic Figure 1. Liver fibrosis is associated with broad changes in the expression of miR-21 and its targeted genes, ERK1 signaling and EMT-associated genes. Serum miR-21 contents in cirrhotic patients (n = 20) and normal subjects (n = 20) (A) and in rats with dimethylnitrosamineinduced liver cirrhosis (B). MiR-21 expression was normalized against miR-238 in serum. **P,0.01 vs. normal controls. Quantitative RT-PCR examination of the expression of miR-21 and its targeted genes SPRY2 and HNF4a, ERK1 and its downstream target RSK2, and epithelial-mesenchymal transition (EMT)-associated genes E-cadherin and vimentin in the liver tissues of cirrhotic patients (n = 7) (C) and rats (n = 10) (D). *P,0.05 and **P, 0.01 vs. normal subjects or rats. Primary HSCs were treated with TGFb1 for 7 days (E) and primary hepatocytes for 48 h (F) as detailed in methods. Quantitative RT-PCR examination of the expression of miR-21 and its targeted genes SPRY2 and HNF4a, ERK1 and RSK2, E-cadherin, vimentin and MMP-9, liver fibrogenesis-associated genes a-SMA and TIMP1, and liver specific genes ALB and CYP1a2. MiR-21 expression was normalized against Rnu6-2 and mRNA expression was normalized against b-actin. *P,0.05 and **P,0.01 vs. non-stimulated HSCs or hepatocytes. Each value represents the mean with the SD (error bars) for triplicate samples. doi:10.1371/journal.pone.0108005.g001  cirrhosis. The findings are consistent with the report of upregulated miR-21 in fibrotic liver biopsies of HCV patients [22]. Recently, certain dysregulated miRNAs have been considered as non-invasive candidate diagnostic biomarkers for liver diseases [38,39]. It may be worthwhile to investigate whether plasma miR-21 could serve as a diagnostic marker or prognostic predictor in cirrhotic patients. ERK1 signaling and EMT are known to have significant impact on fibrogenesis [14,16,40]. Spry1-4 is potent inhibitor of the MAPK pathway and SPRY2 is abundant in the liver [41][42][43]. Previous studies have shown that miR-21 contributes to colon cancer and gliomas along with downregulation of SPRY2 to stimulate ERK signaling [44,45]. Additionally, HNF4a, the liverenriched transcription factor, could attenuate hepatic fibrosis by improving liver function and alleviating EMT in hepatic fibrosis [15,16]. Interestingly, our computational prediction suggests that miR-21 may directly target the 39-UTR of SPRY2 and HNF4a.
Our luciferase reporter assays further demonstrated that miR-21 downregulated SPRY2 and HNF4a expression. We also found that overexpression of SPRY2 and HNF4a led to significantly subdued levels of miR-21 in HSCs and hepatocytes, suggesting the presence of autoregulatory feedback loops between miR-21 and SPRY2 or HNF4a.
To further confirm that dysregulated miR-21 simultaneously promotes fibrotic transformation of HSCs and hepatocytes, we treated primary HSCs or hepatocytes with miR-21 mimics. Expectedly, our data suggested that miR-21 could contribute to the accumulation of fibroblasts not only by stimulating HSC activation via inhibiting SPRY2 expression to increase ERK1 signaling, but also by triggering EMT of hepatocytes via downregulating HNF4a. In view of previous studies and our present work, miR-21 is integrally involved in multiple pathways and profibrotic network in fibroblast transformation of quiescent HSCs and hepatocytes, including TGFb1/smad, NF-kB, ERK1 signaling and EMT, suggesting that miR-21 may serve as a 'super' regulatory miRNA in liver fibrosis. To our knowledge, this is the first report of the roles of miR-21 in simultaneous myofibrobalst transformation of both liver parenchymal and mesenchymal cells in liver fibrosis.
Interestedly, our results showed miR-21 and SPRY2/HNF4a mutually modulated the expression of each other, suggesting that there was a regulatory feedback loop between miR-21 and its target genes. The precise mechanism, however, remains unclear. It has been documented that some target genes could bind to the regulatory miRNA to influence the level of the miRNA. For example, our previous study has found that LIN28A, the target gene of miR-370, blocked the biogenesis of miR-370 by binding to its precursor [46]. We speculate that SPRY2 and HNF4a might exert the feedback regulatory effect with the similar mechanism.
The enhanced effect of miR-21 on EMT and ERK1 singling prompted us to propose that inhibiting miR-21could ameliorate biological characteristics of activated HSCs as well as block EMT of TGFb1-treated hepatocytes. As expected, inhibiting miR-21 upregulated SPRY2 and restrained ERK1 signaling in activated HSCs, and subsequently suppressed the migration of activated HSCs. Moreover, we demonstrated that decreased miR-21  . MiR-21 inhibitors target ERK1 signaling and SPRY2 in HSC activation and block EMT of TGFb1-treated hepatocytes by targeting HNF4a. Primary HSCs were treated with TGFb1 for 5 d followed by transfection with miR-21 inhibitors. MiR-21 and the mRNA transcript levels of SRPY2, ERK1 and RSK2 (A), a-SMA, TIMP1, MMP-2 and MMP-13 (B) were examined by quantitative RT-PCR. Immunoblotting assays were done to detect the expression of SPRY2, phospho-ERK1, phospho-ERK2 and RSK2 (C). HSC-T6 cells were transfected with miR-21 inhibitors. MiR-21 and the mRNA transcript levels of SRPY2, ERK1 and RSK2, MMP-9, MMP-13, fibrogenesis-associated genes TGFb1, a-SMA, Col I, Col III and TIMP1 were examined by quantitative RT-PCR (D, E). Immunoblotting assays were done to detect the expression of SPRY2, phospho-ERK1, phospho-ERK2 and RSK2, Col I and TIMP1 (F). HSC-T6 cells were transfected with miR-21 inhibitors and transwell cell migration assays were performed as described in methods (G, H). Hepatocytes were treated with TGFb1 for 48 h followed by transfection with miR-21 inhibitors. MiR-21 and the mRNA transcript levels of HNF4a, Ecadherin and vimentin, MMP-2, MMP-9 and a-SMA, liver specific genes CYP1a2 and AFP were examined by quantitative RT-PCR (I, J). Immunoblotting assays were done to detect the expression of HNF4a, E-cadherin and vimentin, MMP2, a-SMA and TIMP1 (K). GADPH was used as a loading control. MiR-21 Simultaneously Regulates ERK1 Signaling and EMT PLOS ONE | www.plosone.org expression in hepatocytes undergoing EMT could increase levels of HNF4a and epithelial markers, restore expression of certain liver specific genes in hepatocytes. The above data further showed that downregulated miR-21 expression exerted the inhibitory effects on ERK1 signaling in HSCs and EMT of hepatocytes simultaneously during hepatic fibrosis.
Taken together, our results revealed significant elevation of circulating, hepatic and cellular miR-21 expression, which is associated with ERK1 signaling and EMT in liver fibrosis. MiR-21 directly interacts with the 39-UTR of SPRY2 and HNF4a, leading to enhanced ERK1 signaling in HSCs and hepatocyte EMT. Downregulating miR-21 suppressed ERK1 signaling, inhibited HSC activation, and blocked EMT in TGFb1-treated hepatocytes. Our data strongly indicated the presence of broad changes in miR-21 expression and its targeted genes, ERK1 signaling and EMT-associated genes in hepatic fibrosis and that miR-21 may be a central player in hepatic fibrogenesis. MiR-21 could serve as a potentially clinically useful biomarker and represent a promising molecular target for hepatic fibrosis. Figure S1 Histopathologic examination of the liver tissue of cirrhotic patients and rats. Histopathologic examination of the liver tissue from the patient with hepatic cirrhosis (A). H&E staining of the liver tissue from rats with dimethylnitrosamine-induced liver cirrhosis. The presence of cirrhosis is further shown by Van Gieson (VG), Masson and Sirius Red staining (B). (TIF) Figure S2 Protein levels of SPRY2 and HNF4a in TGFb1 treated cells and cirrhotic rat liver tissues. The levels of SPRY2 and HNF4a were examined by Western blotting assays in primary HSCs treated with TGFb1 for 7 days (A) and primary hepatocytes treated with TGFb1 for 48 h (C). Representative images of levels of SPRY2 and HNF4a in cirrhotic rat liver tissue are shown (B and D). The proteins were extracted from normal (n = 3) and cirrhotic liver tissues (n = 3) randomly. The figures only showed the representative images. GAPDH was used as a loading control. (TIF) Figure S3 Effect of siRNA against SPRY2 on miR-21 level. Primary rat HSCs were transfected with siRNA against SPRY2. Cells were collected 48 h after siRNA delivery. Quantitative RT-PCR was carried out to detect miR-21 expression. Gene expression folds were normalized against the control. Each value represents the mean with the SD for triplicate samples. (*P,0.05). (TIF) Figure S4 Associative action of ERK1 and miR-21. Primary rat HSCs were treated with AdERK1 and miR-21 mimic (A) or primary HSCs were treated with TGFb1 for 48 h followed by AdshERK1 and miR-21 inhibitor transfection for 48 h (B). Quantitative RT-PCR was carried out to detect the levels of miR-21, ERK1, RSK2 and a-SMA. Gene expression folds were normalized against the control. Each value represents the mean with the SD for triplicate samples. (*P,0.05, **P,0.01). (TIF) Figure S5 Combined effect of miR-21 mimics and TGFb1 on EMT in primary rat hepatocytes. Primary rat hepatocytes were treated with miR-21 mimics and TGFb1 for 48 h. MiR-21 level was examined by quantitative RT-PCR (A). Expression of E-cadherin and vimentin was examined by Western blotting assays (B). GADPH was used as a loading control. Blots representative of at least three independent experiments are shown. (TIF)