PU.1-deficient mice are resistant to thioacetamide-induced hepatic fibrosis: PU.1 finely regulates Sirt1 expression via transcriptional promotion of miR-34a and miR-29c in hepatic stellate cells

Hepatic stellate cells (HSCs) are pericytes located in the liver perisinusoidal space, which is the major cell type involved in the healing and scar formation in response to liver damage. Stellate cells are in a quiescent state in normal conditions. When the liver is damaged, stellate cells can change into an activated state, which is characterized by proliferation, contractility, and chemotaxis [1,2]. Genes regulating proliferation and migration have attracted close attention. NAD-dependent deacetylase sirtuin-1 (SIRT1 in human and Sirt1 in mouse) has been shown to deacetylate a wide range of substrates and involved in the regulation of many physiological functions, including energy metabolism, gene transcription, and cell aging regulation [3]. Moreover, SIRT1 functions as an apoptotic inducer and inhibits cell proliferation and activation in multiple types of cells, such as several cancers and adipocytes [4–6]. Recently, a couple of studies showed that Sirt1 was sharply reduced in aged or fibrotic liver and had a negative effect on the proliferation, migration, and activation of HSCs [7,8]. Sirt1-KO mice displayed lower energy expenditure, dysregulation of fatty acid homeostasis, a higher level of inflammation, and more serious fibrosis than wild-type (WT) mice [9]. The deletion of SIRT1 enhanced liver injury and promoted liver fibrosis. Therefore, it is meaningful to clarify which transcription factors or non-coding RNAs (ncRNAs) are involved in the regulation of SIRT1 expression in this process.

PU box binding protein (PU.1) is a vital transcription factor which is famous for its fundamental role in hematopoiesis and immune system development [10,11]. PU.1 promotes transcription of hundreds of genes including many protein-coding genes and a few of non-protein-coding genes, which offers PU.1 a variety of biological functions not restricted in hematopoietic lineages [12,13]. For example, a much recent research showed that PU.1 promotes the transcription of miR-191, an miRNA targetting the adipogenic initial transcription factor CCAAT enhancer binding protein β (C/EBPβ), to inhibit adipogenic differentiation in 3T3-L1 preadipocytes [14]. However, the role of PU.1 in the activation of HSCs and liver fibrosis is still indistinct.

In the present study, we explored the regulatory role of PU.1 in the expression of Sirt1 in mouse HSCs. Interestingly, we found that PU.1 deficiency did not change Sirt1 mRNA but increased Sirt1 protein expression. Then, we comfirmed that PU.1 negatively regulated Sirt1 protein expression through transcriptionally promoting the Sirt1-targetting miRNAs, miR-34a and miR-29c. Moreover, we investigated the effect of PU.1 on activation of HSCs in vitro and progression of TAA-induced liver fibrosis in vivo.

Newborn male C57BL/6J mice, including PU.1 single allele deficient (PU.1^+/−) and WT mice were purchased from Shanghai Laboratory Animal Center of Chinese Academy of Sciences. All animal procedures described herein were approved by the Ethical Committee of Yantaishan Hospital in accordance with the Chinese Experimental Animal Welfare and Ethical Norms promulgated in 2016. The mice were killed immediately after the experimentation in a Mobile Anesthetic Workstation (Harvard Apparatus, Harvard Bioscience Inc., Cambridge, MA).

Mouse primary HSCs were isolated from C57BL/6J PU.1^+/− and WT mice using a pronase/collagenase digestion method as previously described [15]. The cells were identified by morphology observation and immunohistochemistry assay to confirm their features and purity (Supplementary Figure S1). The cells were then grown on untreated Petri dishes (BD Falcon, Franklin Lakes, NJ), being maintained in a low-glucose DMEM/F12 medium supplemented with 10% FBS, 4 mM l-glutamine and P/S (100 U/100 μg per ml). Cells were cultured in a 37°C humidified cell culture chamber with 5% CO₂.

Overexpression vectors pcDNA-PU.1, pcDNA-Sirt1, and pcDNA-3.1 empty vector were kindly presented by Laboratory of Animal Fat Deposition and Muscle Development, Northwest A&F University (Yangling, China). Sirt1 siRNA and oligo mimics/inhibitors of miRNAs were designed and synthesized by Ribobio Co., Ltd. (Guangzhou, China). For transfection, cells were seeded into six-well plates at a density of 10⁵/ml. On preadipocytes reaching 70–80% confluence, serum-free DMEM/F12 medium was used to starve the cells for 2 h. Then the serum-free medium was replaced by fresh DMEM/F12 medium supplemented with 10% FBS, 2 μg of each plasmid, or 60 pmol of each oligo were transfected with X-tremeGENE HP DNA Transfection Reagent (Roche) according to the manufacturer’s instructions. After incubation for 18 h, the medium was changed into fresh (DMEM/F12 medium supplemented with 10% FBS). After another 48 h, the cells were harvested for the following assays.

Total RNA was extracted with TRIzol agent (Invitrogen) and then was used to synthesize cDNA using MMLV reverse transcriptase (Takara, Dalian, China) and oligo (dT) 18 primer following the manufacturer’s protocol. The quantitive real-time PCR (qPCR) were carried out in a final volume of 25 μl, using SYBR Premix Ex Taq (TaKaRa), 0.4 mM of each primer, and 250 ng of cDNA template. Each individual sample was run in triplicate wells. The reactions were initially denatured at 95°C for 1 min followed by 35 cycles at 94°C for 15 s, 56°C for 30 s, and 72°C for 20 s. The density of SYBR Green I and the threshold cycle (C_t) value were analyzed by iQ™5 Optical System Software 2.1 (Bio–Rad, Hercules, CA). The changes in transcript abundance were calculated using the 2^−ΔΔC_t method. The primers applied in the qPCR reactions are listed in Table 1.

Fifty micrograms of protein of each sample were separated by SDS/PAGE (12% gel) and electrotransferred on to PVDF membrane (Millipore, Boston, MA) for immunoblot analysis. The following primary antibodies were used: anti-PU.1 (Cell Signaling, 1:300), anti-Sirt1 (Abcam, 1:300), type I collagen (Abcam, 1:500), anti-α-smooth muscle actin (α-SMA) (Abcam, 1:500), and anti-β-actin (Santa Cruz Biotechnology, 1:800) which was regarded as the internal control. After incubation with the appropriate HRP-conjugated secondary antibody (Abcam), protein levels were detected using a ChemiDoc XRS imaging system and analysed by software Quantity One (Bio–Rad).

The relationships between the transcription factor PU.1 and its potential target DNA sequences (promoter sequences of pri-miR-34a and pri-miR-29c) were predicted in ChIPBase, an online database decoding transcription factor binding maps and transcriptional regulation: http://rna.sysu.edu.cn/chipbase/regulation_browse.php?organism=mouse&assembly=mm10&protein=Spi1&sample_id=MOUMM00068&type=miRNA&upstream=30kb&downstream=10kb&motif_status=Y&target_symbol.

Cells were cross-linked by adding formaldehyde to a final concentration of 1% at room temperature for 10 min. After washing four times with 20 ml PBS in 50-ml conical tubes, cells were scraped and swelled in hypotonic swelling buffer (25 mM HEPES (pH 7.8), 1.5 mM MgCl₂, 10 mM KCl, 0.1% NP-40, protease inhibitor cocktail from Sigma) and incubated on ice for 10 min. Following centrifugation at 2000 rpm for 5 min, the nuclei were lysed in SDS lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris) and sonicated with Branson 150 sonicator. Antibody against PU.1 was used for ChIP, and qPCR was carried out with specific primers to amplify the PU.1-binding region of the miR-34a promoter (forward: 5′-CAG CCT GGA GGA GGA TCG A-3′; reverse, 5′-TCC CAA AGC CCC CAA TCT-3′) and miR-29c promoter (forward: 5′-GCA AAT CGC GTA GTA TCA G-3′; reverse, 5′-TGA CAG AGA CTC CAG ACT-3′).

Following transfection for 48 h, proliferation of the cells was detected with Cell Counting Kit-8 solution (Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer’s instructions.

Cell migration was detected with the Transwell migration assay. Following transfection for 48 h, 3.5 × 10⁴ cells were grown in the top chamber with a non-coated membrane (24-well insert; 8 μm; Corning) with serum-free medium. Medium containing 10% serum was used as a chemoattractant in the lower chamber. The cells were incubated for 24 h. Then a cotton swab was used to remove the non-migrated cells in the upper chamber, and the filters were individually stained with 2% Crystal Violet. The migrated cells adhering to the underside of the filter were examined and counted under a light microscope (Olympus IX70; Olympus Corporation, Osaka, Japan).

Animal model of liver fibrosis was established by intraperitoneal injections of thioacetamide (TAA; Sigma–Aldrich, Munich, Germany). TAA (200 mg/l) was administered in WT and PU.1^+/− mice for 6, 10, or 16 weeks, respectively.

Liver tissue was fixed in formaldehyde and embedded by paraffin. The embedded samples were sliced into 6-μm sections and then stained with HT15-1KT Masson’s trichrome kit (Sigma) according to the manufacturer’s instructions. Morphometric analysis of hepatic fibrosis was performed using semiquantitative fibrosis scores based on the Ishak Scoring System (Grading of Hepatic Necrotic Inflammation).

Fifty grams of liver tissue was homogenated in 1 l DMEM/F12 medium. ROS and MDA contents were respectively detected with Total ROS Detection Kit for Fluorescence Microscopy/Flow Cytometry and Mouse MDA ELISA Detection Kit (BestBio, Shanghai, China) according to the manufacturer’s instructions.

Sixty microliters of 10% chloral hydrate was intraperitoneally injected into mice for anesthesia. Approximately 5 min later, iodophor was used to disinfect the superficial coat of right eye, and then the mice were fixed and punctured for orbital vein blood. Serum was separated for analyzing the activity of ALT and AST in the Yantaishan Hospital utilizing a Roche P module analyzer.

Total collagen was determined by hydroxyproline quantitation. Mouse liver tissue was hydrolyzed with 6 N HCl at 110°C overnight. The hydrolysate was filtered through 45-μm filters, and the filtrate was dissolved in 50% isopropanol. Hydroxyproline contents were detected with a General Hydroxyproline (Hyp) ELISA Kit (Sigma). Absorbance of each sample was measured at 450 nm using a microplate reader (Packard BioScience, Meriden, CT, U.S.A.). Hydroxyproline levels were expressed as mg hydroxyproline per gram liver tissue.

All statistical analyses were performed using SPSS 19.0 statistical software (SPSS, Inc.). Data were presented as means ± S.E.M. Comparisons were made by one-way ANOVA. Significance was set at P<0.05.

Primary HSCs were isolated from WT and PU.1^+/− mice. The expression of PU.1 and Sirt1 in the HSCs were detected with qPCR and Western blotting. The results showed that the levels of PU.1 mRNA and protein were reduced by approximately 50% in the HSCs from PU.1^+/− mice (Figure 1A,C). PU.1 depletion did not influence the expression of Sirt1 mRNA (Figure 1B) but caused a modest increase in that of Sirt1 protein (by approximately 40%, P<0.05, Figure 1C). These data suggested that PU.1 might regulate the expression of Sirt1 gene at a stage after transcription.

MiR-34a and miR-29c are two validated miRNAs that could target Sirt1. HSCs were respectively isolated from WT and PU.1^+/− mice, and the pcDNA-PU.1 expression vector was transfected into PU.1^+/− HSCs. The levels of miR-34a and miR-29c were detected in the WT HSCs, PU.1^+/− HSCs and pcDNA-PU.1-transfected PU.1^+/− HSCs. Compared with the WT HSCs, PU.1^+/− HSCs displayed significantly lower levels of miR-34a and miR-29c, while PU.1 overexpression inversed the decrease in their levels (Figure 2A). Besides, PU.1^+/− HSCs displayed a higher level of Sirt1 protein, and PU.1 sharply suppressed Sirt1 protein expression (Figure 2B).

Then, PU.1^+/− HSCs were transfected with the pcDNA-PU.1, pcDNA-Sirt1, miR-34a mimic or miR-29c, and the levels of Sirt1, miR-34a, and miR-29c were detected. The results showed that PU.1 overexpression significantly increased the levels of miR-34a and miR-29c (Figure 2C), and decreased the Sirt1 protein level in PU.1^+/− HSCs (Figure 2D). Overexpression of either miR-34a or miR-29c did not influence the level of PU.1, but sharply reduced the Sirt1 protein expression (Figure 2D).

For PU.1, overexpression had a markedly positive effect on the expression of miR-34a and miR-29c, we suspected that PU.1 might be a transcription factor of miR-34a and miR-29c. ChIPBase, an online database decoding transcription factor binding maps and transcripitional regulation, revealed that miR-34a and miR-29c are of the several hundred miRNAs potentially regulated by the transcripition factor PU.1. ChIP-qPCR was used to verify the binding of PU.1 with the promoters of miR-34a and miR-29c. The results showed that, compared with IgG control, we detected abundant levels of target transcripts in the protein–DNA complexes bound by PU.1 antibody (Figure 3A,B), indicating that PU.1 was a transcription factor directly regulating miR-34a and miR-29c expression.

It was suggested from all the data above that PU.1 suppressed the expression of Sirt1 through promoting the expression of miR-34a and miR-29c. To confirm that, the pcDNA-Sirt1 expression vector, pcDNA-PU.1 expression vector or inhibitors of miR-34a/miR-29c were individually transfected or co-transfected into WT HSCs. Our data on Western blotting indicated that the suppression by PU.1 overexpression could be inversed by either Sirt1 overexpression or inhibition of miR-34a and miR-29c (Figure 3C). Moreover, the increase in levels of miR-34a and miR-29c could be antagonized by either Sirt1 overexpression or inhibition of miR-34a and miR-29c (Figure 3D,E).

It has been proven that Sirt1 inhibited the proliferation, migration, and activation of HSCs, so PU.1 probably played a positive role in those processes for its suppression on Sirt1 protein. The data on CCK-8 and Transwell migration assays indicated that PU.1 overexpression increased the cell proliferation and migration (Figure 4A,B). Western blotting analysis showed that PU.1 overexpression increased the levels of α-SMA and type I collagen (Figure 4C), which were regarded as markers for HSC activation. As expected, the increase in cell proliferation, migration, and activation was inversed by either Sirt1 overexpression or inhibition of miR-34a/-29c (Figure 4A–C). Moreover, PU.1 overexpression increased the ROS content and expression of proinflammatory genes IL-1β and TGF-β1, and the increase could be inversed by either Sirt1 overexpression or inhibition of miR-34a/-29c (Figure 5A,B). These results indicated that PU.1 positively regulate the proliferation, migration, activation, oxidative stress and inflammatory response in HSCs via suppression of Sirt1 protein expression.

Finally, animal model of liver fibrosis was established by intraperitoneal injection of TAA in WT and PU.1^+/− mice, respectively. At week 16 of TAA adiministration, Masson’s trichrome staining showed that WT mice displayed serious fibrotic features in liver, while PU.1^+/− mice displayed a previously lower fibrotic degree than those of WT mice (Figure 6A). The fibrosis score in the liver of PU.1^+/− mice was much lower than that of WT mice at week 10 and 16 (Figure 6B). The content of total collagen and expression of ColI in the liver of PU.1^+/− mice were significantly lower than that of WT mice at week 10 and 16 (Figure 6C,D). Consistently, the serum levels of ALT and AST, important indexes of liver dysfunction, were also significantly lower in PU.1^+/− mice at weeks 6, 10, and 16 (Table 2). In addition, PU.1^+/− mice also displayed lower levels of inflammatory genes (Figure 7A,B) and oxidative stress (Figure 7C,D). These data above demonstrated that PU.1 deficiency made mice resistant to TAA-induced liver fibrosis and dysfunction.

PU.1 transcription factor was characterized in 1992, which is confirmed as the product of the PU.1 oncogene located in 19q13.3-19q13.4 region of the human genome [16]. For a long time, PU.1 has been restricted as a master controller in the development of hematopoietic lineages. However, it was suggested PU.1 was not a hematopoietic lineage specific transcription factor but also influenced the cell fate and function of non-hematopoietic lineages, such as osteogenic and adipogenic [17,18]. Moreover, much recently, PU. 1 was shown to be lower expressed and identified as a tumor suppuressor in hepatocellular carcinoma, suggesting that PU.1 might play a role in the cell fate or function of the hepatic lineages [12]. In this study, we found that PU.1 had a positive impact on the activation of HSCs and liver fibrosis by negatively regulating the expression of Sirt1 protein. PU.1^+/− mice displayed resistance to TAA-induced liver fibrosis, inflammatory response, and oxidative stress and liver dysfuntion in vivo. To our knowledge, this is the first report that reveals the exact role of PU.1 in the regulation of Sirt1 expression and the activation of HSCs.

As a vital transcription factor, PU.1 promotes transcription of multiple genes, including protein coding and non-protein coding genes including miRNAs [13,19]. Here, we found that, both in WT and PU.1^+/− HSCs, PU.1 overexpression caused a dramatic increase in expression miR-34a and miR-29c, two Sirt1 targetting miRNAs. Moreover, we demonstrated that PU.1 promoted miR-34a and miR-29c expression as a transcription factor by bioinformatics and ChIP-qPCR analyses. Many studies also revealed that PU.1 can transcriptionally promote expression of some miRNAs and functions as a negative regulator of the genes targetted by these miRNAs. For instance, PU.1 directly bound with the promoter of miR-191 host gene, promote miR-191 transcription, and post-transcriptionally suppressed C/EBPβ expression during adipogenesis [14]. Another example is that PU.1 can promote the transcription of miR-615-5p and suppress IGF2-mediated hepatoma cell invasion [12].

The gene expression regulatory mechanism of miRNAs is relatively simple. They negatively regulate their target genes by translational suppression in most cases. Therefore, a function of miRNA depends mainly on its target genes. In the current study, miR-34a and miR-29c both negatively regulated the expression of their target gene Sirt1. Inhibition of miR-34a/-29c could increase Sirt1 expression and inhibited Sirt1-mediated HSC activation and inflammation. They functioned as profibrotic regulators. Very similar to the results of the present study, some researchers also proved that miR-34a promoted activation of HSCs and functioned as a profibrotic regulator [20,21]. For example, Li et al. [20] identified peroxisome proliferator activated receptor γ (PPARγ) as a target gene in HSCs, and inhibition of miR-34a up-regulated PPARγ and down-regulated the expression of α-SMA. However, Kwiecinski et al. [22] proposed that miR-29 family, including miR-29a and miR-29b, could be induced by hepatocyte growth factor or down-regulated by TGF-β1 stimulation, and up-regulation of miR-29 reduced synthesis of collagen-1 and -4. The study did not show the exact role of miR-29c in the activation of HSCs. Moreover, the HSCs in Kwiecinski et al. [22] article were being stimulated by HGF or TGF-β1, but, in our study, HSCs were transfected with plasmids or oligos without any pretreatment. The exact role of miR-29c needs to be further characterized.

Responding to liver damage, stellate cells can change into an activated state, characterized by proliferation, contractility, and chemotaxis. Here, our results showed that Sirt1 overexpression suppressed the cell proliferation, cell migration, expression of activation markers collagen-1 and α-SMA, and production of ROS, IL-1β, and TGF-β1. In addition, down-regulation of Sirt1 by PU.1 overexpression promoted the cell proliferation, cell migration, expression of activation markers collagen-1 and α-SMA, and production of ROS, IL-1β and TGF-β1, which could be antagonized by either Sirt1 overexpression or inhibition of miR-34a/-29c. Moreover, PU.1^+/− mice, which had a higher expression level of Sirt1 protein, displayed obvious resistance to TAA-induced liver fibrosis and dysfunction, manifested by a lower fibrotic score, less collagen content, better liver function, and lower levels of oxidative stress and inflammatory response. Many previous studies also revealed that Sirt1 could suppress liver fibrosis and inflammation. For example, Sirt1 liver KO mice accumulated larger amounts of hepatic lipid and expressed higher levels of inflammatory cytokines than control mice on treatment of ethanol exposure, and serum of Sirt1 liver KO mice had increased levels of ALT and AST [23]. Actually, it has been widely acknowledged that Sirt1 played a preventive role in liver injury and fibrosis.

In conclusion, we found that PU.1 down-regulated Sirt1 expression at the translational level via promotion of miR-34a and miR-29c expression. PU.1 promoted Sirt1-mediated HSC proliferation, migration, and activation in mouse, and PU.1 deficiency made mice resistant to TAA-induced liver fibrosis and dysfunction.

The authors declare that there are no competing interests associated with the manuscript.

The authors declare that there are no sources of funding to be acknowledged.

Q.L. designed the study and prepared the manuscript. Y.Z. and S.Y. performed the experiments. Y.W. and J.W. prepared the figures. W.Y. and Y.L. analyzed the results and revised the manuscript.

ALT

alanine aminotransferase

AST

aspartate aminotransferase

CCK-8

cell counting Kit-8

C/EBPβ

CCAAT enhancer binding protein β

C_t

threshold cycle

DMEM/F12

Dulbecco’s Modified Eagle Media: Nutrient Mixutre F-12

HRP

horseradish peroxidase

HSC

hepatic stellate cell

IGF

insulin-like growth factor

IL-1

interleukin-1

KO

knockout

MDA

malondialdehyde

PPARγ

peroxisome proliferator activated receptor γ

P/S

penicilin and streptomycin

PU

purine

PU.1

PU box binding protein

qPCR

quantitive real-time PCR

ROS

reactive oxygen species

Sirt1

sirtuin-1

TAA

thioacetamide

TGF

transforming growth factor

WT

wild-type

α-SMA

α-smooth muscle actin