TGF-β-induced hepatocyte lincRNA-p21 contributes to liver fibrosis in mice

Hepatocyte death, as well as the following inflammatory and fibrogenic signaling cascades, is the key trigger of liver fibrosis. Here, we isolated hepatocytes from CCl4-induced fibrotic liver and found that hepatocyte lincRNA-p21 significantly increased during liver fibrosis. The increase of hepatocyte lincRNA-p21 was associated with the loss of miR-30, which can inhibit TGF-β signaling by targeting KLF11. We revealed that lincRNA-p21 modulated miR-30 availability by acting as a competing endogenous RNA (ceRNA). The physiological significance of this interaction is highlighted by the feedback loop, in which lincRNA-p21 works as a downstream effector of the TGF-β signaling to strengthen TGF-β signaling and mediate its role in promoting liver fibrosis by interacting with miR-30. In vivo results showed that knockdown of hepatocyte lincRNA-p21 greatly reduced CCl4-induced liver fibrosis and inflammation, whereas ectopic expression of miR-30 in hepatocyte exhibited the similar results. Mechanistic studies further revealed that inhibition of miR-30 impaired the effects of lincRNA-p21 on liver fibrosis. Additionally, lincRNA-p21 promoted hepatocyte apoptosis in vitro and in vivo, whereas the proliferation rate of hepatocyte was suppressed by lincRNA-p21. The pleiotropic roles of hepatocyte lincRNA-p21 suggest that it may represent an unknown paradigm in liver fibrosis and serve as a potential target for therapy.


Results
LincRNA-p21 is upregulated in the hepatocyte during liver fibrosis. As a first attempt to investigate the deregulation of hepatocyte lncRNAs during liver fibrogenesis, we isolated hepatocytes from 5 CCl 4 -induced fibrotic liver and 5 oil-treated sham liver. Total RNA were prepared and used to conduct transcriptome RNA sequencing (RNA-seq). Basing on an absolute fold change cutoff value of 1 in log2 scale, we identified 111 aberrantly expressed lncRNA transcripts (63 upregulated and 48 downregulated) in the hepatocytes from fibrotic liver compared with those from oil-treated liver (Fig. 1A). Among the highly expressed lncRNAs, we focused on the intergenic lncRNA-p21, which was initially identified as a transcriptional target of p53.
To confirm the RNA-seq result, we determined lincRNA-p21 expression in the isolated hepatocyte by qRT-PCR, showing that hepatocyte lincRNA-p21 significantly increased after CCl 4 -injection (Fig. 1B). We next examined the lincRNA-p21 expression in the fibrotic liver and isolated HSCs and found that hepatic lincRNA-p21 increased after CCl 4 -treatment. However, the increase in the HSCs wasn't significant (Fig. 1B). Northern blot and Semi-qRT-PCR results further confirmed the upregulation of lincRNA-p21 in hepatocyte during liver fibrosis ( Fig. 1C and D).
Using the standard curve method, we measured the absolute copy numbers of lincRNA-p21 and miR-30s in AML12 treated with or without TGF-β1, showing that the abundance of lincRNA-p21 and miR-30 are comparable (Fig. 2C). Thus, lincRNA-p21 may be able to function as a ceRNA for miR-30. To confirm the interaction between lincRNA-p21 and miR-30, we inserted the lincRNA-p21 cDNA downstream of the firefly luciferase reporter gene. Transfection of miR-30 greatly decreased the luciferase activity of the wild type reporter with normal binding sites for miR-30, but not that with the mutant binding sites. In contrast, miR-30 antagomir inhibited endogenous miR-30 and increased the luciferase activity (Fig. 2D). Next, we transfected the luciferase reporter into AML12 cells together with increasing amounts of pCI-lincRNA-p21. The luciferase activity increased in response to pCI-lincRNA-p21 in a dose-dependent manner, suggesting that ectopically expressed lincRNA-p21 sequestered endogenous miR-30 and prevented it from suppressing luciferase expression (Fig. 2E). Meanwhile, pCI-lincRNA-p21Mut, in which the predicted miR-30 binding site was mutated, failed to increase the luciferase activity (Fig. 2E).
Hepatocyte miR-30 inhibits liver fibrosis. We previously found that hepatic miR-30s decreased in the fibrotic liver and HSC-specific upregulation of miR-30 prevented liver fibrosis 31 . Here, we examined miR-30s expressions in the hepatocyte after CCl 4 treatment and found that they decreased about 1~4-fold (Supplementary Figure S2A). Thus, we hypothesized that hepatocyte lincRNA-p21 and miR-30 are inversely associated and involved in liver fibrosis. To test this, we constructed adenovirus AdH-miR-30 and AdH-NC that can specifically express miR-30b or control in hepatocyte in vivo under the control of albumin promoter. AdH-miR-30 could significant increased miR-30b expression in AML12, but not in the cultured HSC cell line HSC-T6 (Supplementary Figure S2B). Two days before the first injection of CCl 4 , AdH-miR-30 or AdH-NC was injected into mice via tail vein. Ectopic expression of miR-30 greatly inhibited CCl 4 -induced liver fibrosis as observed by histological examination (Fig. 3A), and significantly decreased collagen deposition and hepatic hydroxyproline level (Fig. 3B). Notably, TGF-β1, Col1a1 and tissue inhibitor of metalloproteinase-1 (TIMP-1) were also greatly reduced in the AdH-miR-30-injected mice. Administration of AdH-miR-30 led to miR-30b increase in the liver tissue (Fig. 3C). Moreover, miR-30b expression increased in the hepatocytes of AdH-miR-30-injected mice, but not in the HSCs (Fig. 3D).
Liver inflammation, triggered by injured hepatocytes, is one of the most characteristic features of chronic liver injury and associated with the development of fibrosis. CCl 4 administration induced significant inflammatory cell infiltration in liver. Notably, increased infiltration of macrophages was limited in AdH-miR-30 group mice (Fig. 3E). Consistent with the histology results, hepatic expression of inflammatory genes, including interleukin-6 (IL-6), chemokine ligand 2 (CCL2) and IL-1β, were suppressed in AdH-miR-30 group (Fig. 3F).
Knockdown of lincRNA-p21 in hepatocyte prevents liver fibrosis. We next investigated the role of hepatocyte lincRNAp21 in liver fibrosis. Mice were injected in the tail vein before the first injection of CCl 4 with the adenovirus AdH-shlincp21 or AdH-NC, which can express lincRNA-p21 shRNA or control under  Figure S3A). Compared to the AdH-shNC-infected mice, AdH-shlincp21-infected mice showed a marked reduction of hepatic fibrosis as observed by histological examination (Fig. 4A), and significantly decreased collagen deposition and hepatic hydroxyproline level (Fig. 4B). LincRNA-p21 silencing also significantly reduced the expression of α-SMA, Col1a1, TGF-β1, TIMP-1 and is associated with miR-30. AML12 cells were transfected with biotinylated miR-30b (Biotin-miR-30b) or its biotinylated mimic control (Biotin-NC) for 24 h. Cells were then harvested for biotin-based pull-down. After washing and enrichment of beads/RNA complex, RNA was eluted from the streptavidin beads. Left, lincRNA-p21 enrichment was determined by qRT-PCR and normalized to control without any transfection. Right, Eluted RNA was amplified by RT-PCR with oligonucleotides spanning region of lincRNA-p21 and run on 1% agarose gel. Total inputs (Input-Biotin-NC and Input-Biotin-miR-30b) are indicated as total RNA isolated from Biotin-NC or Biotin-miR-30b-transfected AML12 cells. The results are shown as fold change compared with control group. Data are the mean ± SEM of three independent experiments. *P < 0.05, **P < 0.01. NS, no significant change.   (Fig. 4C). Additionally, administration of AdH-shlincp21 reduced the infiltration of hepatic macrophages (Supplementary Figure S3B) and the upregulation of hepatic IL-6, IL-1 and CCL2 in CCl 4 -treated mice (Fig. 4D).
Administration of AdH-shlincp21 reduced hepatic lincRNA-p21 (Fig. 4E). We then isolated hepatocyte and HSC from the fibrotic liver. LincRNA-p21 greatly decreased in the hepatocytes from AdH-shlincp21 group, but not in the isolated HSCs (Fig. 4E). Moreover, the miR-30s in the isolated hepatocytes from AdH-shlincp21 group mice significantly increased, suggesting that AdH-shlincp21 might prevent liver fibrosis by increasing miR-30 in hepatocyte (Fig. 4F).
Hepatocyte lincRNA-p21 regulates liver fibrosis through interacting with miR-30. We reasoned that, if hepatocyte lincRNA-p21 regulates liver fibrosis by interacting with miR-30, inhibition of miR-30 would show inhibitory effects on the protective function of AdH-shlincp21 in liver fibrosis. Inhibition of miRNAs with modified antisense oligo-ribonucleotides has been proven to be efficient in silencing endogenous miRNAs in multi tissues 32,33 . To test our hypothesis, anti-miR-30, a phosphorothioate-modified antisense oligonucleotides specific for miR-30, and scrambled control (SCR), were intravenously injected into CCl 4 -treated mice weekly during the liver fibrosis development. AdH-shlincp21 was injected in the tail vein two days before the first treatment of CCl 4. In SCR-injected group, AdH-shlincp21 significantly inhibited liver fibrosis. However, in anti-miR-30 group, AdH-shlincp21 failed to exert the inhibitory effects ( Fig. 5A and B). The expression of hepatic profibrogenic markers (α-SMA, Col1a1, TGF-β1, CTGF and TIMP-1) also significantly increased in anti-miR-30 group (Fig. 5C). In addition, hepatic expression of IL-6, IL-1 and CCL2 significantly increased in anti-miR-30b group mice (Fig. 5D).
AdH-shlincp21 prevented the increase of hepatocyte lincRNA-p21 and led to a significant increase of miR-30b in the isolated hepatocytes from fibrotic liver (Fig. 5E). However, the injection of miR-30 antisense oligonucleotides decreased miR-30b in the hepatocyte (Fig. 5F). Collectively, our results suggest that hepatocyte lincRNA-p21 contributes to liver fibrosis by interacting with miR-30.
LincRNA-p21 is induced by TGF-β in the hepatocyte. Given the prominent role of TGF-β/Smad signaling in hepatocyte death and hepatic fibrogenesis, we investigated the regulation of hepatocyte lincRNA-p21 in response to TGF-β1. We found that lincRNA-p21 was induced by TGF-β1 in a dose-and time-dependent manner ( Fig. 6A and B). The inhibitor of type I TGF-β receptor kinase SB431542 completely blocked the increase of TGF-β-induced lincRNA-p21 (Supplementary Figure S4A). Since lincRNA-p21 was initially identified as a transcriptional target of p53, we examined the expression of p53 in TGF-β1-treated AML12 cells. We found that TGF-β1 didn't increase p53 production, and thus excluded the possibility that TGF-β1 increased lincRNA-p21 expression via p53 (Supplementary Figure S4B).
We searched for potential Smad-binding sites in the upstream region of lincRNA-p21 using rVista 2.0 (http://rvista.dcode.org/), and found 3 conserved binding sites at −1520 bp, −944 bp and −530 bp upstream of lincRNA-p21 in mouse genome. The predicted murine lincRNA-p21 promoter includes the 1400 bp upstream and 200 bp downstream of the first exon 34 . To identify the cis-element responsible for TGF-β induction, we constructed a set of pGL3-Basic luciferase reporters carrying various genomic sequences spanning from −1,825 to + 115 (relative to transcription start site), and subjected them to luciferase assays. The construct encompassing −1825 to +115 bp had a higher promoter activity in response to TGF-β1 (Fig. 6C). We then cloned the −1825 bp to −1426 bp fragment into pGL3-Promoter luciferase reporter, showing that the this fragment contained the TGF-β-response element (Fig. 6D). Finally, we mutated the predicted Smad-binding sites (from −1520 to −1512 bp) and found the mutation abolished the TGF-β-induced luciferase activities (Fig. 6D). Chromatin immunoprecipitation (ChIP) also demonstrated that the antibody against Smad3 could immunoprecipitate the predicted DNA fragments from TGF-β1-treated hepatocyte (Fig. 6E).
Notably, we have previously reported that TGF-β1 reduced miR-30 in hepatocyte 35 . To ascertain the underlying mechanism responsible for miR-30 decrease in response to TGFβ, we determined the expression of pri-miR-30s in TGFβ-treated AML12 cells, showing that TGFβ didn't obviously suppress the transcription of pri-miR-30s (Supplementary Figure S4C). Thus, TGFβ-induced lincRNA-p21 might be responsible for the decrease of miR-30.
LincRNA-p21 activates TGF-β/Smad signaling in the hepatocyte. In our previous study, we found that miR-30 blunted TGF-β/Smad signaling in HSCs by targeting KLF11, which suppressed the transcription of inhibitory Smad7 in TGF-β/Smad pathway 31 . Here, we further revealed that the inhibition of KLF11 by miR-30 resulted in the upregulation of Smad7 in hepatocytes (Fig. 7A). In the isolated hepatocytes from fibrotic liver injected with AdH-miR-30, ectopic expression of miR-30b led to decrease of KLF11 and increase of Smad7 in hepatocyte in vivo (Fig. 7A). Western blot showed that miR-30b greatly inhibited the phosphorylation of Smad2 and Smad3 in TGF-β1-treated AML12 cells (Fig. 7B). Collectively, these results provide convincing evidence that miR-30 can suppress TGF-β/Smad signaling by targeting KLF11 in hepatocyte.
We next sought to investigate whether lincRNA-p21 is involved in TGFβ/Smad signaling through regulating KLF11. Luciferase reporter containing the wild or mutant KLF11 3′UTR was transfected into AML12 cells together with pCI-lincRNA-p21. Ectopic expression of lincRNA-p21 induced the increase of luciferase activity in the reporter containing the KLF11 3′UTR (Supplementary Figure S5A). However, the increase of luciferase activity was suppressed by miR-30b (Supplementary Figure S5B). Reciprocally, knockdown of lincRNA-p21 reduced the luciferase activity. The suppression of luciferase activity by lincRNA-p21 siRNA was reversed by miR-30 antagomir (Supplementary Figure S5C).
Given that lincRNA-p21 is highly inducible by TGFβ, we explored the involvement of lincRNA-p21 in TGFβ-induced hepatocyte apoptosis. Knockdown of lincRNA-p21 significantly inhibited the TGFβ-induced apoptosis and led to the decrease of proapoptotic genes and the increase of anti-apoptotic Bcl-XL (Supplementary Figure S6A and B). In the TGFβ-induced apoptotic AML12 cells, the cleavage of caspase 3 was inhibited by lincRNA-p21 siRNA (Supplementary Figure S6C).
Finally, we investigated the in vivo role of lincRNA-p21 in hepatocyte apoptosis. Hepatocyte apoptosis remarkably increased in CCl 4 -induced fibrotic liver as assessed by counting the number of apoptotic terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling (TUNEL)-positive hepatocytes. However, knockdown of lincRNA-p21 blunted the increase of hepatocyte apoptosis (Fig. 8H). We also determined the serum alanine aminotransferase (ALT) levels, showing that knockdown of lincRNA-p21 significantly decreased serum ALT in the CCl 4 -treated mice (Fig. 8I).
LincRNA-p21 contributes to regulating hepatocyte growth in fibrotic liver. Increased hepatocyte death and growth inhibition enhance the secretion of pro-inflammatory and profibrogenic cytokines. We found that enforced expression of lincRNA-p21 in AML12 cells resulted in a significant decrease in viability (Supplementary Figure S7A). Conversely, knockdown of lincRNA-p21 increased AML12 cell viability (Supplementary Figure S7B). In TGFβ-treated AML12 cells, lincRNA-p21 siRNA prevented the TGF-β-induced  Figure S7D). Additionally, proliferating hepatocytes significantly increased in livers from AdH-shlincp21 group compared with those of AdH-shNC group as evaluated by immunohistochemical staining of Ki67 (Supplementary Figure S7E).

Discussion
lncRNAs play crucial roles in specific cell types, tissues and developmental conditions by various mechanisms. Correspondingly, most lncRNAs exhibit tissue or cell-specific expression 37 . Here, using RNA sequence, we identified differentially expressed lincRNA-p21 in hepatocyte during liver fibrosis. Similarly, lincRNA-p21 increased in the livers of mice treated with the carcinogen furan, and thus implying an extensive role of lincRNA-p21 in the response to liver injury 38 . Of note, lincRNA-p21 also increased and contributed to pulmonary fibrosis in acute respiratory distress syndrome 39 . However, the deregulation of lincRNA-p21 in the progression of fibrosis still remains elusive. A recent report showed that lincRNA-p21 decreased in human cirrhotic liver and a higher dose CCl 4 -induced fibrotic liver 40 . Given the fact that lncRNAs are poorly conserved among species, more investigations are needed to confirm the regulation of lincRNA-p21 under different pathological conditions. Previous reports showed that the transcription of lincRNA-p21 is regulated by various signaling pathways under different conditions 18,21 . Here, using both in vitro and in vivo studies, we observed a significant increase of hepatocyte lincRNA-p21 after TGF-β1 treatment, and identified the promoter region responsible for TGF-β/ Smad signaling. Since TGF-β1 increased during liver fibrogenesis triggered by various causes, it's tempting to speculate that TGF-β1-mediated upregulation of lincRNA-p21 may work as a specific downstream effector of TGF-β signaling in hepatocyte, and therefore represent a hitherto unknown paradigm in liver fibrosis. miRNAs with profibrogenic or antifibrogenic functions have been implicated in diverse animal models and human patients. However, few studies identified their roles in hepatocyte. Here, our results demonstrate that hepatocyte miR-30 greatly inhibits fibrotic TGF-β/Smad signaling by targeting KLF11 and consequently prevents liver fibrosis. KLF11 isn't a member of Smad family. However, the induction of KLF11 strengthens TGF-β/Smad signaling cascade through suppressing the Smad7 expression. In addition, previous studies showed that KLF11 overexpression mimics the TGF-β-induced effects in different types of cells 41,42 . Taken together, inhibition of KLF11 could be an effective way to suppress the excessive TGF-β/Smad signaling during liver fobrosis.
The increase of lincRNA-p21 in hepatocyte was associated with the loss of miR-30 during liver fibrosis. In both in vitro and in vivo systems, we observed that downregulation of lincRNA-p21 was sufficient to suppressed TGF-β signaling and liver fibrosis. These phenomena depend on the interaction between lincRNA-p21 and miR-30. The presence of competitive miR-30 antagomir abolished the inhibitory effects of lincRNA-p21 knockdown on TGF-β signaling and liver fibrogenesis, indicating that lincRNA-p21 functions as a ceRNA. Basing on these results, we propose that TGF-β-induced lincRNA-p21 in turn strengthens TGF-β signaling by interacting with miR-30, thus forming a positive feedback loop to ensure lincRNA-p21 expression and mediate the role of TGF-β in promoting liver fibrosis.
To date, the mechanism of miR-30 deregulation in various states is mostly unknown. Here, we provide the first evidence that TGF-β-induced lincRNA-p21 inhibited miR-30 by directly binding to them. Moreover, the transcribing of pri-miR-30 wasn't affected by TGF-β, and thus strongly suggesting the underlying mechanism responsible for miR-30 decrease in response to TGF-β. However, at this stage, we can't exclude the possibility that the decrease of miR-30 may be triggered by other mechanisms in liver fibrosis. To clarify this issue, further studies will be needed.
Hepatocyte apoptosis and cell death is a key trigger of liver fibrogenesis 7,8 . To date, a number of lncRNAs have been shown to modulate apoptosis 43,44 . However, there are conflicting reports on the role of lincRNA-p21 in apoptosis and cell growth 19,20,36 . Here, in vitro and in vivo results confirmed the contribution of lincRNA-p21 to hepatocyte apoptosis during liver fibrosis, showing that inhibition of hepatocyte lincRNA-p21 reduced hepatocyte death and prevented subsequent inflammatory cell recruitment and activation of inflammatory and fibrogenic signal cascade. LincRNA-p21 was initially identified as a transcriptional target of p53. Subsequent investigations suggested that lincRNA-p21 functioned as a downstream repressor in the p53 transcriptional response and promoted p53-mediated expression of p21 20 . Notably, hepatocyte-specific deletion of p53 decreased liver fibrosis through downregulating the profibrogenic mediators 45 . Thus, whether hepatocyte lincRNA-p21 functions as a more general mediator of multiple profibrogenic signaling pathways remains to be defined.
In summary, our present results extend the knowledge of lincRNA-p21 and deepen the understanding of the cross talk between lncRNA and profibrogenic TGF-β signaling cascades in hepatocyte during liver fibrosis. The hypothetical roles of hepatocyte lincRNA-p21 in TGF-β signaling and liver fibrosis are schematically summarized in Supplementary Figure 8. The pleiotropic effects of hepatocyte lincRNA-p21 on liver fibrosis suggest that it could be an effective target for therapy.

Materials and Methods
Antibodies and Reagents. The antibodies used in the study are listed in Supplementary Table 1.
Recombinant human TGF-β1 was from Peprotech (Rocky Hill, NJ). Other reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated.
Cell culture. Cells were incubated in a humidified atmosphere of 5% CO 2 at 37 °C . In this study, recombinant TGF-β1 was added to the medium at a final concentration of 5 ng/ml or as indicated. The culture medium and isolation of hepatocyte and HSC were described in the Supplementary Materials and Methods. Animal treatments. Animal protocols were reviewed and approved by the Animal Care and Use Committee of Nanjing University, and conformed to the Guidelines for the Care and Use of Laboratory Animals published by the National Institutes of Health. Five-week old male C57 mice (20 ± 2 g) were obtained from the Animal Center of Yangzhou University (Yangzhou, China). Animals were maintained under pathogen-limited conditions and had free access to rodent chow and water. Mouse liver fibrosis was induced according to previously described mothed 31 . Briefly, mice were intraperitoneally injected with of 20% CCl 4 solution in sterile mineral oil at a dose of 2.5 ml CCl 4 per kilogram body weight twice per week for three or four weeks. The adenovirus was injected only one time via tail vein at two days before the first CCl 4 injection (1 × 10 9 pfu/mouse).
Histological examination. Histological examinations of liver were performed as described before 31  Northern blot. LincRNA-p21 probe was labelled with digoxigenin (DIG) using a DIG DNA Labelling Kit (Roche). Total RNA extracted from AML12 cells and run on a 1% denatured agarose gel, transferred to positively charged nylon membranes (Millipore) followed by cross-linking through UV irradiation. The membrane was then hybridized with (DIG)-labelled probe overnight. The detection was performed using a DIG luminescent detection kit (Roche) according to the manufacturer's instructions.
Western blot. Western blot was performed as described before 31 . In brief, the cell lysates were separated by SDS-PAGE and then transferred onto PVDF membranes. After incubation with primary antibody, the membranes were washed with PBST and then probed with horseradish peroxidase-conjugated secondary antibody at room temperature. The immunoreactive bands were detected by fluorography using an enhanced chemiluminescence system (Cell Signaling Technology, Beverly, MA). Statistical Analysis. Results in this study are expressed as the means ± standard error of the mean (SEM).
The data were analyzed for normal distribution. Differences between multiple groups were checked using one-way ANOVA with post-hoc Bonferroni correction. Differences between two groups were analyzed by a two-tailed unpaired Student's t test. A value of P < 0.05 was considered statistically significant, and P < 0.01 indicated strongly significant difference.
Other methods used in this study are described in the Supplementary Materials and Methods.