Introduction

Liver fibrosis occurs as a wound-healing response to liver injury induced by various types of chronic liver diseases and is characterized by an excessive accumulation of extracellular matrix (ECM) components, predominantly collagens [1, 2]. The excessive deposition of ECM proteins replaces functional tissue and disrupts the normal liver architecture. During fibrosis progression, hepatic stellate cells (HSCs) become activated and transdifferentiate into myofibroblastic cells that express activation marker, such as α-smooth muscle actin (α-SMA), which is a key event in liver fibrogenesis [3]. The secretion of excessive ECM proteins and profibrogenic mediators such as transforming growth factor-β1 (TGF-β1) and connective tissue growth factor by activated HSCs results in an imbalance between ECM protein generation and their degradation in liver fibrosis [4]. Therefore, the suppression of activated HSCs is considered as a potential target for liver fibrosis.

Curcumin (i.e. diferuloylmethane, C₂₁H₂₀O₆, molecular weight of 368.4 g) is a polyphenol isolated from yellow curry pigment of turmeric and has demonstrated anti-inflammatory and antioxidant activities [5, 6]. Recently, it was reported that the activated HSCs are inhibited by curcumin [7, 8]. However, the mechanisms responsible for the antifibrotic effects induced by curcumin still remain unclear. Phosphatase and tensin homologue deleted on chromosome 10 (PTEN), which is a tumour suppressor in various cancers, is a negative regulator of liver fibrosis. Both phosphatidylinositol 3-kinase (PI3K)/Akt and extracellular signal-regulated kinase (ERK) pathways are targeted by PTEN through its lipid phosphatase and protein tyrosine phosphatase activity, respectively [9, 10]. The activated PI3K/Akt and ERK pathways as a result of the loss of function of PTEN lead to a reduction of cellular apoptosis and an increase in mitogen signalling [11, 12]. Recently, Bian et al. [13] demonstrated that the DNA methyltransferase1 (DNMT1)-mediated PTEN hypermethylation contributed to the activation of HSCs and liver fibrogenesis in rats. These results suggest that the gene for PTEN plays a vital role in the development of liver fibrosis. However, the role of PTEN in the effects of curcumin has not been studied.

The role of PTEN is important in the development of liver fibrosis and the expression of PTEN is associated with the methylation level of its promoter. DNA methylation, which is a type of epigenetic modifications in mammals, plays an important role with respect to affecting normal transcriptional regulation. Genome-wide DNA hypomethylation can induce chromosomal instability and spurious gene expression. Mammalian DNA is dominantly methylated in the C-5-position of complimentary CpG bp DNMTs, including DNMT1, DNMT3a and DNMT3b [14]. DNMT1 plays a role in the maintenance of specific DNA methylation patterns during DNA replication, whereas DNMT3a and DNMT3b are both essential for de novo methylation. Interestingly, the expression of DNMTs could be regulated by microRNAs (miRNAs), as confirmed by luciferase activity assays. This shows that miRNA-mediated epigenetic regulation is involved in disease-related.

miRNAs are endogenous, small, noncoding RNAs, 21–23 nucleotides in length, that negatively regulate gene expression by imperfect base pairing with the 3′ UTR region of their target messenger RNAs for the promotion of mRNA degradation or translational repression [15, 16]. Recently, emerging studies have indicated that miRNAs are central players in antifibrotic and profibrotic gene regulation during liver fibrosis [17]. For example, the miR-29 family consists of miR-29a, miR-29b and miR-29c, which differ only in two or three bases and can repress profibrogenic gene expression during liver fibrosis [18, 19]. Furthermore, DNMTs are predicted as miR-29 family (miR-29a, miR-29b and miR-29c) target genes by targetscan in human cell lines. Fabbri et al. [20] showed that DNMT3a and DNMT3b could be directly targeted by miR-29 in lung cancer cell lines. In addition, it was found that miR-29b could induce global DNA hypomethylation in acute myeloid leukaemia cells by targeting DNMT3a and DNMT3b directly and DNMT1 indirectly [21]. However, knowledge of the function and role of miR-29 in rat liver fibrosis is still not clear, and more efforts are needed to obtain a better understanding in this unexplored field.

In the present study, the expression and methylation status of PTEN were analyzed both in vitro and in vivo after curcumin treatment. It was found that increased PTEN was associated with the DNA methylation level of PTEN by curcumin. It is necessary to understand the role of DNMTs and miR-29, as well as to confirm whether DNMTs and/or miR-29 are responsible for the effects induced by curcumin. Therefore, the present study aimed to investigate the mechanism underlying the increase in PTEN as a result of curcumin.

Results

Curcumin induces PTEN up-regulation and contributes to the inactivation of Akt and ERK signalling pathways

To explore the role of curcumin in PTEN expression, as well as Akt and ERK signalling pathways in liver fibrosis, the levels of PTEN, p-ERK and p-Akt were detected in vitro and in vivo after curcumin treatment. Compared with the control, mRNA and protein levels of PTEN were elevated in TGF-β1-treated HSC-T6 cells, primary HSCs and carbon tetrachloride (CCl₄)-treated rats after curcumin treatment (Fig. 1). Of note, immunoblot analysis showed that the expression of p-ERK and p-Akt was significantly reduced in HSC-T6 cells and primary HSCs after curcumin treatment, whereas the expression of total Akt and ERK proteins remained unchanged (Fig. 1B). Similarly, protein levels of p-ERK and p-Akt were also decreased in liver tissues from CCl₄-treated rats after curcumin treatment compared to those from CCl₄-treated rats (Fig. 1B). These results indicate that curcumin increased PTEN expression and inhibited profibrotic signalling pathways, such as the Akt and ERK pathways.

PTEN up-regulation induced by curcumin plays an important role in the suppression of the activated HSCs

Activated HSCs are suppressed by curcumin. To confirm whether PTEN played a role in the antifibrotic effect induced by curcumin, cell growth and the apoptosis were detected in curcumin-treated HSCs with PTEN short interfering RNA (siRNA). As shown in by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) assay, the growth rate was reduced to 61.3% and 74.5%, respectively, in curcumin-treated HSC-T6 cells and primary HSCs compared to untreated cells (Fig. 2A). Caspase3 activity was increased to 2.47-fold and 1.95-fold, respectively, in curcumin-treated HSC-T6 cells and primary HSCs compared to untreated cells (Fig. 2B). However, these effects were almost blocked down by the silencing of PTEN expression. We also detected some molecules that had already been reported to be up-regulated in activated HSCs and involved in liver fibrosis, including α-1 (I) collagen (Col1A1), α-2 (I) collagen (Col1A2), α-1 (III) collagen (Col3A1) and α-SMA. The mRNA levels of Col1A1, Col1A2, Col3A1 and α-SMA were obviously decreased in curcumin-treated HSC-T6 cells and primary HSCs, whereas they were significantly increased in curcumin-treated HSC-T6 cells and primary HSCs with PTEN siRNA (Fig. 2C). Immunoblot analysis further confirmed that the suppression of type I collagen, type III collagen and α-SMA by curcumin was reversed by the loss of PTEN (Fig. 2D). These results suggest that curcumin reduced ECM protein and α-SMA via PTEN expression.

PTEN down-regulation as a result of its promoter hypermethylation is restored by curcumin

To confirm whether the methylation status of PTEN promoter was responsible for reduced PTEN expression, the methylation of PTEN promoter was detected by methylation-specific PCR (MSP). In vitro, MSP analysis showed that curcumin treatment reduced aberrant hypermethylation of PTEN in activated HSC-T6 cells and primary HSCs (Fig. 3A, B). In vivo, the hypermethylation of PTEN promoter in the liver tissues from CCl₄-treated rats was also reversed by curcumin (Fig. 3C). Combined with the previous study, we found that the loss of PTEN methylation played an important role in the restoration of PTEN expression by curcumin.

The expression of DNMT3b was reduced by curcumin in vitro and in vivo

DNMTs, including DNMT1, DNMT3a and DNMT3b, are responsible for the regulation of the global patterns of DNA methylation. DNMT1 plays a role in the maintenance of specific DNA methylation patterns during DNA replication, whereas DNMT3a and DNMT3b are essential for de novo methylation. To investigate whether DNMTs were involved in the loss of PTEN methylation by curcumin, quantitative real-time PCR and immunoblot analysis were performed to detect the mRNA and protein levels of DNMTs in activated HSCs and CCl₄-treated rats, respectively. Curcumin treatment obviously reduced the mRNA and protein levels of DNMT3b in primary HSCs and TGF-β1-treated HSCs, whereas the mRNA and protein levels of DNMT1 and DNMT3a were not changed by curcumin (Fig. 4). Similar results were also observed in vivo. Only the mRNA and protein levels of DNMT3b were suppressed by curcumin in the liver tissues from CCl₄-treated rats (Fig. 4). These results suggest that DNMT3b played an important role in the loss of PTEN methylation by curcumin.

PTEN expression was enhanced by the knockdown of DNMT3b and its promoter methylation was associated with DNMT3b activity

To further confirm the role of DNMT3b in curcumin-mediated PTEN expression, DNMT3b siRNA was transfected into curcumin-treated HSCs, and the expression of PTEN, ECM proteins and α-SMA was detected. It was found that the DNA methylation level of PTEN was down-regulated in curcumin-treated HSCs with DNMT3b siRNA (Fig. 5A, B). Consistent with these results, the mRNA level of PTEN was significantly increased in curcumin-treated HSCs with DNMT3b siRNA (Fig. 5C). DNMT3b knockdown also decreased HSC activation in curcumin-treated HSCs. The mRNA and protein levels of Col1A1, Col1A2 and α-SMA were obviously inhibited in curcumin-treated HSCs with DNMT3b siRNA (Figs 2D and 5E). These data showed that DNMT3b was involved in the loss of PTEN methylation, which was down-regulated by curcumin. We also showed that DNMT3b knockdown was sufficient to up-regulate PTEN in HSCs without curcumin treatment (Fig. 5D). Because the PTEN promoter methylation was not regulated by DNMT1 in cells treated with curcumin, it is necessary to confirm whether PTEN promoter methylation is associated with DNMT3b activity. This hypothesis was confirmed by chromatin immunoprecipitation (ChIP) assays. The band densities of PTEN were obviously decreased in chromatin extracted from HSCs transfected with DNMT3b siRNA, rather than that from the control (Fig. 6B, C). It was also found that the enrichment of PTEN promoter region (−1876 to −1661 nucleotides) (Fig. 6A) was significantly reduced in cells transfected with DNMT3b siRNA compared to the control (Fig. 6D, E). These data demonstrate that the PTEN promoter region is associated with the activity of DNMT3b.

MiR-29b was involved in curcumin-induced PTEN expression

Recently, it was suggested that miR-29 family could induce global DNA hypomethlylation by targeting DNMTs, including DNMT1, DNMT3a and DNMT3b [21]. For this reason, we selected miR-29 as a target for the next experiment. To confirm whether the members of miR-29 family were regulated by curcumin, the expression levels of miR-29a, miR-29b and miR-29c were detected by real-time PCR. Compared with the control, only the expression level of miR-29b was elevated in TGF-β1-treated HSC-T6 cells, primary HSCs and CCl₄-treated rats after curcumin treatment (Fig. 7A). Next, we investigated the effects of miR-29b overexpression on DNMT3b and PTEN mRNA expression. Compared with the curcumin group, the mRNA expression level of DNMT3b was markedly suppressed to 47% and 32%, respectively, in HSC-T6 cells and primary HSCs transfected with miR-29b mimics (Fig. 7B). By contrast, the mRNA level of PTEN was obviously increased by ~ 20–39% after the transfection of miR-29b mimics compared to the curcumin group (Fig. 7B). Also, we showed that miR-29b overexpression was sufficient to up-regulate PTEN and down-regulate DNMT3b in HSCs without curcumin treatment (Fig. 5D). These results suggest that miR-29b might be involved in the loss of PTEN methylation by curcumin.

MiR-29b targeted DNMT3b and induced the loss of PTEN methylation

To explore the molecular mechanism of the loss of PTEN methylation regulated by miR-29b, we predicted the targets of miR-29b using bioinformatics analysis. It was predicted that DNMT1 and DNMT3a were not the targets of miR-29b, whereas miR-29b could interact with the 3′ UTR of rat DNMT3b mRNA using miRDB (http://mirdb.org/miRDB/) (Fig. 8A, B). The sequence of 3′ UTR of DNMT3b mRNA target region was cloned into pMIR-Report™ Luciferase plasmid. The construct was cotransfected into HSCs along with miR-29b precursor or miRNA negative control (miR-NC). β-gal reporter control plasmid was cotransfected to monitor transfection efficiency. MiR-29b precursor significantly reduced luciferase activity driven by the wild-type 3′ UTR of DNMT3b compared to miR-NC in HSCs, whereas it could not inhibit the luciferase activities of mutated type DNMT3b 3′ UTR and empty vector (Fig. 8C). These results suggest that DNMT3b was a target of miR-29b. Next, we investigated the role of miR-29b in the curcumin-mediated expression of DNMT3b and PTEN. It was found that the reduction of curcumin-mediated DNMT3b and the increase in curcumin-induced PTEN were reversed by miR-29b inhibitor in activated HSCs (Fig. 9A, B). Immunoblot analysis also showed that the protein levels of DNMT3b and PTEN had a similar pattern (Fig. 9C). Furthermore, we detected the DNA methylation levels in curcumin-treated HSCs with miR-29b inhibitor. MSP analysis showed that the loss of PTEN methylation induced by curcumin was inhibited by miR-29b inhibitor (Fig. 9D). These results indicate that miR-29b could alter the methylation and expression of PTEN via the target of DNMT3b in curcumin-treated HSCs.

Discussion

The results of the present study show that the methylation of PTEN was reduced by curcumin via the regulation of miR-29b and DNMT3b. Owing to the loss of PTEN methylation in liver fibrosis, the expression level of PTEN was increased, resulting in the inhibition of HSC proliferation and the induction of apoptosis. It has been demonstrated that PI3K/Akt and ERK activation, which play an important role in controlling cancer cell growth and survival, promote HSC activation and liver fibrogenesis [22, 23]. Moreover, the loss of PTEN gene expression causes aberrant activation of the PI3K/Akt and ERK pathways and, consequently, leads to cancer cell proliferation and, ultimately, stimulates tumourigenesis [24]. Consistent with these previous studies, both PI3K/Akt and ERK pathways were activated in the liver tissues of CCl₄-treated rats and in activated HSCs, which were reduced by curcumin. Furthermore, it was found that only DNMT3b, and not DNMT1 or DNMT3a, was involved in the loss of PTEN methylation, which was confirmed as a target of miR-29b and silenced by curcumin-induced miR-29b. These data revealed the involvement of epigenetic modification with respect to suppressing liver fibrosis and as a novel antifibrotic mechanism of curcumin (Fig. 10).

It has been demonstrated that curcumin not only inhibits the proliferation of HSCs, but also stimulates the apoptosis of these cells [8]. In addition, the deposition of collagen was also reduced by curcumin [25]. A recent study showed that the inhibition of cell proliferation and the induction of apoptosis induced by curcumin in activated HSCs were blocked down by the inhibitor of peroxisome proliferator-activated receptor γ, which is a key regulator of the cell cycle and apoptosis [26]. Furthermore, the interruption of nuclear factor-kappaB and ERK signalling by curcumin inhibited the expression level of connective tissue growth factor, which contributes to the production of ECM proteins in activated HSCs [27]. However, the epigenetic modification regulated by curcumin in HSCs has never been explored. In the present study, the antifibrotic effects of curcumin treatment were inhibited by PTEN siRNA, indicating that the knockdown of PTEN contributed to the development of liver fibrosis. This finding was consistent with the previous study by Bian et al. [13], which showed that decreased PTEN expression as a result of its hypermethylation promoter was found during HSC activation. It was also found that PTEN hypermethylation promoter was regulated by DNMT1, which was not consistent with our results, and the expression and role of both DNMT3a and DNMT3b was not further confirmed by Bian et al. [13]. In the present study, only Dnmt3b was changed and PTEN promoter methylation was not regulated by DNMT1 with respect to the effects of curcumin treatment. It was further confirmed that PTEN promoter methylation was associated with DNMT3b activity by ChIP (Fig. 6). Therefore, the alteration of PTEN promoter methylation is a result of continuous DNMT3b activity, which is targeted by curcumin-induced miR-29b. Combined with the previous study, it is demonstrated that the PTEN promoter is associated with DNMTs, including DNMT1 and DNMT3b.

Recent studies have shown that miRNAs play a critical role in the control of various HSCs functions, such as cell proliferation and apoptosis [19, 28]. A previous study also demonstrated that miR-29b expression was down-regulated during HSC activation in primary culture [19]. ECM proteins such as typeIcollagen could be inhibited by miR-29b via the interaction with 3′ UTRs of Col1A1 and Sp1 [18]. MiR-29b has a profound influence on liver fibrosis, and the loss of miR-29b expression may contribute to the development of liver fibrosis. The miR-29 family could induce global DNA hypomethlylation by targeting DNMTs, including DNMT1, DNMT3a and DNMT3b [21]. Of note, these results were all found in human cell lines. The present study aimed to check whether DNMTs expression is also regulated by miR-29 in the rat. Accordingly, we selected the miR-29 family, and not others, as a research target. Our results showed that miR-29b could reduce DNMT3b expression as a target gene, as confirmed by luciferase activity assays. Combined with our previous studies, the methylation status of PTEN could be regulated by miR-29b via its direct target DNMT3b. There may be other miRNAs regulated by curcumin, although the key role of miR-29b was confirmed in the present study. The effects of reduced DNMT3b protein and increased PTEN protein by curcumin treatment could be blocked down by miR-29 inhibitor. With the loss of miR-29b, the expression of DNMT3b was increased. Furthermore, the decreased expression of miR-29b was found both in vivo and in vitro during liver fibrosis (Fig. 7A). In addition, an effective miRNA repression requires a favourable ratio between the miRNA and its target. The results of the present study showed that the two-fold increased miR-29b by curcumin was blocked down by miR-29b inhibitor, resulting in the restoration of reduced DNMT3b (Figs 7A and 9A, C). These results showed that DNMT3b could be effectively inhibited by curcumin-induced miR-29b.

Elasin, fibrillin and Col3A1, which are components of ECM proteins, were predicted to be miR-29b target genes by targetscan and eventually validated [29]. Transfection of a miR-29b precursor markedly attenuated the expression of Col1A1 and Col1A2 mRNAs in mouse HSCs [19]. However, whether Col1A1, Col1A2 and Col3A1 are the target genes of miR-29 in rat still needs to be confirmed by luciferase activity assays. We also investigated whether PTEN was a target gene for miR-29b in the rat. The results showed that PTEN was not a predicted target gene for miR-29b and PTEN could not be directly regulated by miR-29b. In the present study, it was found that curcumin increased miR-29b expression, which down-regulated the methylation of PTEN and up-regulated its expression in HSCs, resulting in the inhibition of Col1A1, Col1A2, Col3A1 and α-SMA expression. It was also found that miR-29b could inhibit the activation of HSCs via the methylation status of PTEN. However, the mechanism of the direct regulation of miR-29b by curcumin still remains unclear and further studies are warranted.

Collectively, the results obtained in the present study demonstrate that curcumin can up-regulate the expression level of miR-29b, leading to the silencing of DNMT3b and the loss of PTEN methylation, which contributes to the suppression of activated HSCs. The results not only provide a new insight of the role of miRNAs and epigenetic mechanisms in the suppression of liver fibrosis, but also show a new antifibrotic mechanism of curcumin. Furthermore, it is confirmed that miRNAs mediated the epigenetic mechanisms that contribute to the suppression of liver fibrosis.

Materials and methods