LncRNA Airn maintains LSEC differentiation to alleviate liver fibrosis via the KLF2-eNOS-sGC pathway

Study population analysis was performed as described in the previous study [17]. Briefly, 28 human fibrotic livers and 6 human healthy livers from patients with hepatic hemangioma were obtained from surgical resections without preoperative treatment at Tianjin Third Central Hospital (Tianjin, China). Hepatic fibrosis was scored (stages F0–F4) according to the METAVIR fibrosis staging system by three hepatopathologists blinded to the study protocol and stratified as normal liver (F0), mild fibrosis (F1–F2), and advanced fibrosis (F3–F4). In addition, we collected 47 serum samples from patients diagnosed as cirrhosis at Tianjin Third Central Hospital (Tianjin, China), and 30 matched blood donor volunteers recruited from the same hospital with no medical history. All subjects were of the same ethnicity. Clinical and pathological characteristics including age, gender, ALT, AST, ALB, GTT, and etiologies were recorded and summarized in Additional file 1: Table S1. The study has been approved by the local Ethical Committee of Tianjin Third Central Hospital (Tianjin, China). Written informed consent was obtained from each patient according to the policies of the committee. The study methodologies were conformed to the standards set by the Declaration of Helsinki.

All the animal protocols were in accordance with the Guidelines for Animal Experiments of Tianjin Medical University Animal Care and Use Committee. Airn knockout (Airn-KO) C57BL/6N mice were generated by CRISPR/Cas9 system (Cyagen, Suzhou China). In brief, genomic DNA was extracted from the Airn-KO mice and was PCR amplified using the primers (F: 5′-AGACACATTTAGTTGGTGGTTGGTCG-3′, R: 5′-TCTTCCACACCCAGGTGGCTTT-3′, R: 5′-AGGAAGTAGGCTCATGGGAGGAG-3′). The product of 800 bp was used for the amplicon and the sequence was confirmed by Sanger sequencing. All wild type (WT) and Airn-KO mice were generated from Airn heterozygous mice, they were kept in a standard 12-h light–dark cycle under the specific pathogen-free conditions with free access to water and food. All liver fibrosis models were performed in male mice unless indicated. For CCl₄-induced liver fibrosis model, twenty WT mice and twenty Airn-KO mice were randomly divided into four groups: WT mice intraperitoneally injected oil (WT, n = 10), WT mice intraperitoneally injected CCl₄ (WT + CCl₄, n = 10), Airn-KO mice intraperitoneally injected oil (Airn-KO, n = 10), and Airn-KO mice intraperitoneally injected CCl₄ (Airn-KO + CCl₄, n = 10). They were administered 5% CCl₄ (v/v) (Sigma-Aldrich, St. Louis, MO, USA) dissolved in oil (0.3 ml/kg body weight) thrice per week for 6 weeks via intraperitoneal injection. For BDL-induced liver fibrosis model, forty WT mice and forty Airn-KO mice were randomly divided into four groups. WT mice were treated with sham operation (WT, n = 20), WT mice were treated with BDL (WT + BDL, n = 20), Airn-KO mice were treated with sham operation (Airn-KO, n = 20) and Airn-KO mice were treated with BDL (Airn-KO + BDL, n = 20). Eighteen days later, all mice were sacrificed under anesthesia with 3% sodium pentobarbital (45 mg/kg, ip). For over-expressed Airn model, adeno-associated virus (AAV8) vectors were used to over-express Airn in mice, and AAV8-GFP was used as a control. Thus, forty Balb/c mice were divided into four groups randomly: Mice were treated with oil in combination with injection of AAV8-GFP (AAV8-GFP, n = 10), CCl₄ in combination with injection of AAV8-GFP (AAV8-GFP + CCl₄, n = 10), oil in combination with injection of AAV8-Airn (AAV8-Airn, n = 10), and CCl₄ in combination with injection of AAV8-Airn (AAV8-Airn + CCl₄, n = 10). Mice in AAV8-GFP + CCl₄ group and AAV8-Airn + CCl₄ group were injected with AAV8-GFP and AAV8-Airn respectively via the tail vein 2 weeks after the first injection of CCl₄, and administered 5% CCl₄ (v/v) dissolved in oil (0.2 ml/kg body weight) thrice per week via intraperitoneal injection. AAV8-GFP and AAV8-Airn group animals were injected with an equivalent volume of oil. After treatment with CCl₄ for 8 weeks, all mice were sacrificed under anesthesia with 3% sodium pentobarbital (45 mg/kg, ip).

The immunohistochemistry was performed essentially as described previously [18]. The slides were treated with primary antibody α-SMA (1:200, rabbit polyclonal, Abcam, ab5694), COL1α1 (1:200, rabbit polyclonal, Abcam, ab34710), CD31 (1:25, mouse monoclonal, ab9498), LAMININ (1:200, rabbit polyclonal, Abcam, ab11575), and PCNA (1:800, rabbit monoclonal, Cell Signaling Technology, #13110), overnight at 4 °C. The slides were incubated with secondary antibody (1:500) (HRP-conjugated anti-rabbit IgG), and the reaction products were visualized using diaminobenzidine (DAB) and monitored by microscopy.

Briefly, livers were fixed with glutaraldehyde, postfixed with OsO4, dehydrated with graded alcohols, dried with hexamethyldisilazane, sputter-coated with gold, and examined using a Gemini 300 scanning electron microscope (Zeiss, Germany). Porosity (percentage of LSEC surface occupied by fenestrae) was measured in scanning electron microscopy (SEM) micrographs of cells.

Primary mouse HSC and HC were isolated by pronase/collagenase perfusion digestion followed by density gradient centrifugation, as previously described [17]. In brief, primary LSEC were isolated from the 8-week-old Balb/c mice by in situ perfusion with 30 ml SC1 solution and 30 ml 0.05% Collagenase IV solution sequentially. The cell suspension was centrifuged at 50g for 4 min and then the supernatant was centrifuged at 500g for 8 min at 4 °C. Pelleted cells were resuspended in 10 ml of 18% Nycodenz solution (Sigma-Aldrich, St. Louis, MO, USA); 6 ml of 12% Nycodenz solution, 6 ml of 8% Nycodenz, and 4 ml of DMEM were orderly loaded on the top of the cell suspension. The added gradient centrifugal liquid was centrifuged at 1450g and 4 °C for 22 min without brake. LSECs were recovered from the interface between the 8 and 12% Nycodenz solutions, mixed with an equal volume of DMEM and centrifuged at 600g for 6 min at 4 °C. Cells were resuspended and incubated with anti-CD146 magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany). Finally, LSEC were cultured in collagen IV-coated plates with LSEC medium, and cell viability was determined by the trypan blue exclusion method.

ChIP assays were performed essentially as described previously [18, 19]. Anti-EZH2 (rabbit polyclonal, Abcam, ab186006) and IgG were used to immunoprecipitated chromatin fragments. Finally, qRT-PCR assays were performed to analyze the precipitated chromatin DNA. The immunoprecipitated DNA was quantitated by qRT-PCR. The ChIP primer sequences were as follows: Klf2-pro 1(-272--120) (Forward) 5′-GCGCGCTAACTATGCTGTTG-3′ and Klf2- pro 1 (Reverse) 5′-CGGTATATAAGCCTGGCGGT-3′, Klf2-pro2 (-916--804) (Forward) 5′- GCTCCTTGGATGAGGC-TT-GT-3′ and Klf2-pro2 (Reverse) 5′-AGCATTAGGTTCAAGGCCCC-3′, Klf2- pro3 (-1299--1155) (Forward) 5′-TGTTTGCCTCCGGGGTTAAG-3′ and Klf2- pro3 (Reverse) 5′-GGGGGATGGGCACATCAAAT-3′, Gapdh intron (Forward) 5′-ATCCTGTAGGCCAGGTGATG-3′ and (Reverse) 5′-AGGCTCAAGGGCTTTTAAGG-3′. The data of ChIP was shown as a percentage relative to input DNA.

The cDNA of full-length Airn was sequentially amplified by PCR and ligated into the lentiviral shuttle pCCL.PPT.hPGK.IRES.eGFP/pre [18] to generate the over-expression plasmid (LV-Airn and the empty plasmid as the LV-Control). The cloning primer sequences were as follows: LV-Airn BamHI Forward 5′- CGCGGATCCAATAATCTCCACCCCCTG-3′, LV-Airn BamHI Reverse 5′- CGCGGATCCTTAAGACCCTGTTGAAATTT-3′. These plasmids were used to produce lentivirus in HEK-293T cells with the packaging plasmids. Infectious lentiviruses were harvested at 36 and 60 h after transfection and filtered through 0.45 μm PVDF filters for in vitro experiments. AAV8 vectors for capsid screening were produced by transfecting AAV-293 cells using polyethylenimine (PEI) with an AAV vector plasmid (pAAV.CMV.PI.EGFP.WPRE.Bg-h, addgene, #105530) and helper plasmid including pAAV2/8 (addgene, #112864) and pAdDeltaF6 (addgene, #112867). For in vivo experiments, AAV vectors were produced in large scale and purified through iodixanol gradient density centrifugation, and full AAV particles were collected from the 40–60% interface of iodixanol phase for in vivo experiments. Male mice at 6–8-week-old were injected by tail vein with 1×10¹² pfu/mouse genome copies of AAV8-GFP or AAV8-Airn. The cloning primer sequences were as follows: AAV8-Airn HindIII Forward 5′- CCCAAGCTTAATAATCTCCACCCCCTG-3′, AAV8-Airn BamHI Reverse 5′- CGCGGATCCTTAAGACCCTGTTGAAATTT-3′.

The non-tumorigenic mouse hepatocyte cell line AML12 was cultured in DMEM with 10% fetal bovine serum (FBS, Gibco, Gaithersburg, MD, USA), 1 × insulin-transferrin-sodium selenite media supplement (ITS, Sigma-Aldrich), 40 ng/ml dexamethasone, 100 U/ml penicillin, and 100 μg/ml streptomycin. The cell line HEK293T or AAV293 were maintained in DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin, both cells were cultured in humidified air at 37 °C and 5% CO₂.

Primary LSECs (1×10⁶/well) were seeded in collagen-coated 6-well plates and transfected with siAirns and siRNA-Control for 48 h; RNA and protein of cells were harvested and isolated. siAirns and siRNA-Control were gained from GenePharma, and the sequences are as follows: siAirn-1 (mouse), sense 5′-CCAGUUACCACGCAGACAUTT-3′ and antisense 5′-AUGUCUGCGUGGUAACUGGTT-3′, siAirn-2 (mouse), sense 5′- CCGUCACCAUGUGUCCUUUTT-3′ and antisense 5′- AAAGGACACAUGGUGACGGTT-3′, siAirn-3 (mouse), GCAGCUCUCAUCUGUGUUATT-3′ sense 5′- and antisense 5′- UAACACAGAUGAGAGCUGCTT-3′, NC (mouse), sense 5′-UUCUCCGAACGUGUCACGUTT-3′, and anti-sense 5′-ACGUGACACGUUCGGAGAATT-3′.

Cytoplasmic and nuclear RNA and protein isolation were performed with PARIS™ Kit (Invitrogen, Grand Island, NY, USA), following the manufacturer’s instruction and were performed essentially, as described previously [18].

Briefly, primary LSECs infected with two separated siAirn were lysed with Trizol reagent. Total RNA was qualified and quantified using a Nano Drop and Agilent 2100 bioanalyzer (Thermo Fisher Scientific, MA, USA). The library was amplified with phi29 to make DNA nanoball (DNB) which had more than 300 copies of one molecule, DNBs were loaded into the patterned nanoarray, and single-end 50 bases reads were generated on BGIseq500 platform (BGI-Shenzhen, China). The threshold we used to screen up- or downregulated mRNAs was fold change >1.4 and p<0.05. The transcriptome sequencing data have been deposited in NCBI Gene Expression Omnibus (GEO) under the following accession number: GSE174175.

Cells were lysed with cell lysis buffer (Cell Signaling Technology) supplemented with protease inhibitor cocktail, 1% PMSF, and 1% phosphatase inhibitor. Protein concentrations were measured by the BCATM Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA, USA) using BSA as standard. Appropriate amount of protein samples (40 μg for liver tissues; 25–50 μg for cells) along with 4× loading buffer and ddH₂O were boiled for 4 min and then subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Following by electrophoresis, the separated proteins were blotted onto polyvinylidene fluoride (PVDF) membranes in transfer buffer with constant current of 300 mA for 3 h at 4 °C. Then the PVDF membranes were sequentially washed with TBST containing 0.2% Tween-20, blocked with 5% nonfat milk in TBST and incubated with the interested primary antibodies diluted in TBST containing 0.2% Tween-20 overnight at 4 °C. Antibodies used for immunoblotting included VEGFR2 (1:1000, rabbit monoclonal, Cell Signaling Technology, #9698), eNOS (1:1000, mouse monoclonal, Abcam, ab76198), VE-Cadherin (1:1000, rabbit polyclonal, Abcam, ab33168), PCNA (1:1000, rabbit monoclonal, Cell Signaling Technology, #13110), α-SMA (1:1000, rabbit polyclonal, Abcam, ab5694), COL1α1 (1:1000, rabbit polyclonal, Abcam, ab34710), MMP2 (1:1000, rabbit monoclonal, Abcam, ab92536), TIMP1 (1:1000, mouse monoclonal, Santa Cruz, sc-21734), KLF2 (1:1000, mouse polyclonal, Santa Cruz, sc-166238). GAPDH were severed as control for total protein amount. Then, all of membranes were incubated with the HRP-conjugated secondary antibody for 1 h at room temperature. Every experiment was repeated at least three times independently.

Immunofluorescence analysis was performed as described previously [20]. Antibodies used for confocal microscopy included VEGFR2 (1:200, rabbit monoclonal, Cell Signaling Technology, #9698), VE-Cadherin (1:200, rabbit polyclonal, Abcam, ab33168), F4/80 (1:50, rabbit monoclonal, Abcam, ab16911), α-SMA (1:200, rabbit polyclonal, Abcam, ab5694), COL1α1 (1:200, rabbit polyclonal, Abcam, ab34710), Ki67 (1:200, rabbit monoclonal, Abcam, ab16667), PCNA (1:800, rabbit monoclonal, Cell Signaling Technology, #13110), and KLF2 (1:50, mouse polyclonal, Santa Cruz, sc-166238). Cells were incubated with FITC-conjugated secondary antibodies (Thermo Fisher Scientific, HRP-conjugated anti-rabbit IgG or anti-mouse IgG, Alexa Fluor 594) in a dark environment at room temperature for 1 h. Next, cells away from light were stained with DAPI (Sigma, USA) for 20 min. Finally, the slides were mounted with an anti-fade mounting medium and were observed with a Zeiss confocal microscope LSM700.

qRT-PCR analysis was performed as described previously [17]. The sequences of primers are listed as follows.

Cell Counting Kit-8 (CCK-8) assay CCK-8 assay was carried out for detection of cell proliferation in AML12 cells. AML12 cells were inoculated in 96-well plates with 2×10³ cells per well. Absorbance at 450 nm was recorded at specified time points using the CCK-8 kit, based on which the viability curve was plotted.

Cells were inoculated in 24-well plates (10⁴ cells/ well) and labeled with 10 μmol/L EdU at 37 °C for 2 h. After 15 min fixation in 4% paraformaldehyde, cells were incubated with phosphate buffered saline (PBS) containing 0.3% Triton-100 for 15 min. After washing with PBS, 125 μL of dying solution was applied per well and incubated in dark for 30 min. The nucleus was stained with 1×Hoechst 33342 for 10 min. EdU-positive cells, Hoechst-labeled nucleus, and merged ones were captured under a microscope.

Data were expressed as mean ± SD. All the statistical analyses were performed with the SPSS 13.0 (IBM, Armonk, NY, USA). Statistical analyses were performed using either Student’s t test (two-group comparison) or one-way analysis of variance (more than two groups) followed by post hoc comparison, and differences with p<0.05 were considered significantly.

In our prior study, we found that Airn was significantly upregulated in liver fibrosis according to the microarray data. (The microarray data discussed in the previous article have been deposited in NCBI Gene Expression Omnibus and are accessible through GEO Series accession number GSE89147) [18]. As both mouse and human Airn have various isoforms, we first detected these Airn isoforms in liver tissues or hepatocytes. Of all the isoforms, NR_002853 (mouse Airn) and NR_047514 (human AIRN) showed high expression in liver tissues and remarkably upregulated in fibrotic liver, compared with other isoforms (Additional file 1: Fig. S1A, B). To explore the role of Airn in liver fibrosis, the expression of Airn was initially verified in mouse liver fibrosis model. The results showed that the expression of Airn was dramatically increased in mouse CCl₄- and BDL-induced fibrotic liver tissues compared with normal liver tissues (Fig. 1A, B). Moreover, total RNAs were isolated from the serums of 47 patients with liver fibrosis and found that the level of AIRN was increased in these patients compared with 30 healthy volunteers (Fig. 1C). ROC curve analysis revealed the potential diagnostic performance of AIRN in discriminating patients with liver fibrosis from healthy subjects (AUC = 0.796; p<0.05) (Fig. 1D). In addition, the expression of human AIRN was also detected in 6 healthy liver tissues, 16 mild fibrotic liver tissues (F1–F2) and 12 advanced fibrotic liver tissues (F3-F4) [17]. As shown in Fig. 1E, the expression of AIRN was significantly upregulated in mild fibrotic livers but not in advanced fibrotic livers.

To assess the expression of Airn in different cell types of healthy and fibrotic livers, primary liver parenchymal cells and non-parenchymal cells, including HSC, LSEC, and KC, were isolated from livers of healthy and fibrotic mice respectively. The purity of primary liver non-parenchymal cells was examined by confocal microscopy detection of VE-Cadherin and VEGFR2 (LSEC marker) [21, 22], α-SMA (HSC marker), and F4/80 (KC marker) (Additional file 1: Fig. S1C). Meanwhile, qRT-PCR analysis was preformed to further detect the expression of Lyve1, VE-cadherin, Vegfr2, F4/80, Cd11b, and α-SMA (Additional file 1: Fig. S1D). Based on these observations, LSEC preparations appeared about 94% purity. Our data showed that the highest expression level of Airn was found in HSC and LSEC followed by HCs and KCs (Additional file 1: Fig. S2A). Moreover, the expression of Airn was upregulated in primary LSECs from fibrotic livers of mice treated with CCl₄ for 2, 4, and 8 weeks (Fig. 1F), correlating with the reduction of expression of Vegfr2 and eNos, yet correlating also with the enhancement of the expression of endothelin-1(Et-1) and Laminin (Fig. 1F). Moreover, Airn was increased in injured primary HC and activated HSC, correlating with the enhancement of the expression of Ki67, Col1α1, and α-SMA (Fig. 1G and Additional file 1: Fig. S2B). However, Airn expression had no significant difference in primary KCs of mice treated with CCl₄ for various time periods (Additional file 1: Fig. S2C). Overall, the data showed an obvious enhancement of Airn in primary LSEC, HC, and HSCs at 2 weeks after CCl₄ treatment, indicating Airn is involved in the initiation of fibrosis. The increased Airn in these cells maintaining a high level until significant fibrosis could be observed at 8 weeks of CCl₄ treatment, together with the finding that AIRN was increased in human mild fibrotic livers, suggested that Airn plays a role in the progression of liver fibrosis.

In order to elucidate the function of Airn during liver fibrogenesis in vivo, we generated Airn knockout (Airn-KO) mice via CRISPR/Cas9 system (Additional file 1: Fig. S3A-D) and subsequently constructed the CCl₄-induced liver fibrosis model. Airn-KO mice were normal in appearance and mating, and a detailed histological examination of major internal organs did not reveal any morphological abnormalities (Additional file 1: Fig. S4). The CCl₄ group mice developed serious liver fibrosis and knockout of Airn further aggravated the CCl₄-induced liver fibrosis as demonstrated by macroscopic examination, hematoxylin and eosin (H&E), sirius red staining, serum ALT, AST level, and liver hydroxyproline content (Fig. 2A and Additional file 1: Table S2). Moreover, knockout of Airn notably facilitated the upregulation of CCl₄-induced α-SMA and COL1α1, whereas it significantly suppressed compensatory upregulation of CCl₄-induced PCNA by IHC (Fig. 2A). In addition, western blot and qRT-PCR analysis showed that deficiency of Airn significantly promoted the upregulation of CCl₄-induced α-SMA, COL1α1, MMP2, and TIMP1 (Fig. 2B, C), suggesting Airn deficiency aggravated CCl₄-induced liver fibrosis. On the other hand, we investigated whether Airn regulated LSEC differentiation or capillarization in vivo. Scanning electron microscopy (SEM) results showed that the number of fenestrae were noticeably decreased in CCl₄-induced mice and were further decreased in CCl₄-induced Airn-KO mice (Fig. 2D). Furthermore, the expression of the genes related to LSEC capillarization including CD31 and LAMININ was observably enhanced in CCl₄-induced Airn-KO mice when compared with CCl₄-induced WT mice (Fig. 2C, D), suggesting that Airn deletion worsened CCl₄-induced LSEC capillarization.

To exclude the possibility that Airn alters the metabolism or toxicity of CCl₄ rather than by altering cell responses, the results was confirmed in a BDL-induced mouse liver fibrosis model. As shown in Additional file 1: Fig. S5A-C and Additional file 1: Table S3, the mice of BDL group developed serious liver fibrosis, and knockout of Airn further aggravated BDL-induced liver fibrosis as demonstrated by macroscopic examination, H&E staining, sirius red staining, IHC, serum ALT, AST level, liver hydroxyproline content, western blot, and qRT-PCR. As expected, knockout of Airn notably facilitated the upregulation of BDL-induced expression of LSEC capillarization marker genes CD31 and LAMININ assessed qRT-PCR and IHC (Additional file 1: Fig. S5C, D). Taken together, our data clearly revealed that Airn deficiency aggravated CCl₄- and BDL-induced liver fibrosis and LSEC capillarization in vivo.

To test whether over-expression of Airn would alleviate liver fibrosis in vivo, AAV8-Airn or AAV8-GFP was intravenously injected into the CCl₄-treated or oil-treated mice via the tail vein 2 weeks after the first injection of CCl₄. After 8 weeks of CCl₄ treatment, qRT-PCR analysis confirmed that Airn was over-expressed in the liver fibrosis model (Fig. 3A). Over-expression of Airn greatly alleviated CCl₄-induced liver fibrosis as demonstrated by macroscopic examination, H&E staining, sirius red staining, serum ALT, AST level, and liver hydroxyproline content (Fig. 3B and Additional file 1: Table S4). Moreover, Airn over-expression markedly suppressed upregulation of CCl₄-induced α-SMA and COL1α1, whereas obviously promoted compensatory upregulation of PCNA by IHC (Fig. 3B). In addition, qRT-PCR and western blot analysis showed that over-expression of Airn suppressed upregulation of CCl₄-induced α-SMA, COL1α1, MMP2, and TIMP1 (Fig. 3C, D). On the other hand, SEM analysis indicated that the number of fenestrae was significantly decreased in CCl₄-induced mice, while Airn over-expression rescued the reduction of fenestrae (Fig. 3E). Furthermore, Airn over-expression ameliorated CCl₄-induced LSEC capillarization assessed by IHC and qRT-PCR for CD31 and LAMININ (Fig. 3C, E), demonstrating that Airn over-expression suppressed CCl₄-induced LSEC capillarization. Taken together, these results confirmed that over-expression of Airn alleviated CCl₄-induced liver fibrosis in vivo and was associated with LSEC capillarization.

The in vivo data showed that Airn alleviated CCl₄-induced LSEC capillarization. Therefore, to investigate whether Airn was involved in LSEC capillarization and differentiation in vitro, we first synthesized three specific siRNAs against Airn (siAirn-1, siAirn-2, and siAirn-3). Among them, siAirn-1 and siAirn-2 showed an efficient knockdown effect and were applied in subsequent experiments. RNA-seq analysis showed that knockdown of Airn decreased LSEC differentiation-associated genes including Flt4 (Vegfr3), Lyve-1, and Stab1, but increased LSEC capillarization-associated genes including Gabre, Lama1, and Lama2 [23] (Additional file 1: Fig. S6A). qRT-PCR analysis further verified that knockdown of Airn significantly reduced Vegfr2, eNos, Lyve-1, and VE-cadherin expression, whereas markedly enhanced Laminin and Angpt2 expression (Fig. 4A). Moreover, the protein level of VEGFR2, eNOS, and VE-Cadherin was markedly decreased in Airn-silenced LSECs assessed by western blot and confocal microscopy (Fig. 4B, C), indicating that Airn silencing promoted LSEC capillarization. On the other hand, over-expression of Airn remarkably enhanced the expression of Vegfr2, eNos, Lyve-1, and VE-cadherin, while significantly suppressed the expression of Laminin and Angpt2 in primary LSEC (Fig. 4D). Consistently, the protein level of VEGFR2, eNOS, and VE-Cadherin was notably increased in Airn over-expressing LSEC (Fig. 4E, F). Additionally, the expression of Vegfr2 and eNos in the primary LSECs isolated from AAV8-Airn mice exhibited remarkable enhancement, but the expression of Laminin demonstrated a profound reduction, in comparison with the AAV8-GFP mice (Additional file 1: Fig. S6B). Taken together, our results demonstrated that Airn maintained LSEC differentiation in vitro.

HSC activation have been commonly recognized as the principal cellular players promoting synthesis and deposition of ECM. The expression of marker genes including collagens I and III, α-SMA, MMPs, and TIMPs are significantly increased in the activated HSC compared with that of quiescent HSCs. Additionally, it has been reported that isolated HSCs remains quiescent when cultured for 3 days in vitro, became partly activated at day 7 and fully activated at day 14 [24]. Since the results in vivo showed that Airn alleviated CCl₄-induced liver fibrosis, the effect of Airn was investigated in HSCs. However, the expression of COL1α1, α-SMA and MMP2 was unchanged in Airn-silenced or -over-expressed primary HSC at day 2 assessed by qRT-PCR, western blot, and confocal microscopy (Fig. 5A–C and Additional file 1: Fig. S7A-B). Similarly, these results were confirmed in primary HSCs at day 12 (Additional file 1: Fig. S7C-F) suggesting that Airn was not directly involved in the regulation of HSC activation. Therefore, we hypothesized that Airn indirectly acted on HSC by maintaining LSEC differentiation. Subsequently, Airn-silenced or -over-expressed primary LSEC was used to co-culture with primary HSC (Fig. 5D). The expression of the fibrotic genes including α-SMA and COL1α1 was significantly upregulated in HSCs co-culturing with Airn-silenced primary LSECs compared with that treated with the control LSECs (Fig. 5E, G), while the expression of these genes was repressed in HSC co-culturing with Airn-over-expressed primary LSEC assessed by western blot and confocal microscopy (Fig. 5F, G). Taken together, these results suggested that Airn inhibited HSC activation indirectly through maintaining LSEC differentiation.

To investigate whether Airn was required for the proliferation of HC in vitro, we knocked down the expression of Airn by siRNA in primary HC and AML12 cells. qRT-PCR, western blot, and confocal microscopy analysis showed that the expression of Ki67 and PCNA was significantly decreased in Airn-silenced primary HCs or AML12 cells (Fig. 6A–C and Additional file 1: Fig. S8A-C). CCK8 assay showed that Airn-silenced notably inhibited cell proliferation (Additional file 1: Fig. S8D). Moreover, over-expression of Airn remarkably increased the expression of Ki67 and PCNA in primary HCs and AML12 cells (Fig. 6D,E and Additional file 1: Fig. S8E-G). CCK8 and EdU assay showed that over-expression of Airn significantly improved cell proliferation (Additional file 1: Fig. S8H, I). Additionally, the expression of the pro-proliferation genes exhibited a profound reduction in the primary HCs isolated from CCl₄-induced Airn-KO mice, in comparison with that of the CCl₄-induced WT mice (Additional file 1: Fig. S9A, B), indicating that Airn promoted HC proliferation directly. While it has been reported that LSECs promoted HC proliferation by paracrine hepatic cytokines including Wnt2a and HGF [7, 21]. Therefore, we investigated whether Airn promoted HC proliferation by LSECs paracrine signal. The results showed that Airn silencing inhibited the mRNA of expression of Hgf and Wnt2a in primary LSEC (Fig. 6F). To further detect whether Airn regulated LSEC paracrine secretion, Airn was silenced by siRNA in primary LSECs and cultured in vitro for 48 h, and the level of HGF was detected in culture supernatants by ELISA. The data showed that Airn silencing downregulated the secretion of HGF in LSEC (Fig. 6G). Next, Airn-silenced primary LSEC were used to co-culture with primary HC (Fig. 6H). Compared with the control LSEC-treated HC, the expression of Ki67 was significantly reduced in HC when co-cultured with Airn-silenced primary LSECs by qRT-PCR and confocal microscopy (Fig. 6I, J). Taken together, Airn not only directly promoted the proliferation of HC, but also promoted the proliferation of HCs through LSEC paracrine secretion of Wnt2a and HGF.

We next explore the mechanism of Airn in maintaining LSEC differentiation. It has been reported that the LSEC phenotype maintained, at least partly, through eNOS-sGC signaling [21] and we have revealed that Airn promoted the expression of eNOS and regulated the expression of eNOS-sGC downstream target genes (Fig. 4A–E). Therefore, LSEC were treated with sGC agonist YC-1. qRT-PCR and confocal microscopy analysis confirmed that Airn silencing decreased the expression of LYVE-1 and increased the expression of LAMININ, which was abrogated by YC-1 (Fig. 7A, B). Moreover, YC-1 rescued Airn silencing-induced downregulation of Wnt2a, Hgf, and Wnt9b, indicating that the knockdown of Airn in LSEC suppressed HC proliferation via the eNOS-sGC pathway (Fig. 7A). Furthermore, primary LSEC was isolated from WT and Airn-KO mice, subsequently treated with YC-1. The expression of Lyve-1 was downregulated, while Laminin was upregulated in the primary LSEC isolated from Airn-KO mice compared with WT (Additional file 1: Fig. S10A, B). However, augmentation of Laminin induced by knockout of Airn was abolished by YC-1 (Additional file 1: Fig. S10A, B). In addition, the expression of KLF2, which has been reported to have positively regulated the transcription of eNOS [25, 26], was decreased in Airn-silenced primary LSEC (Fig. 7C, D and Additional file 1: Fig. S10C) and increased in Airn-over-expressed primary LSECs assessed by qRT-PCR, western blot, and confocal microscopy (Additional file 1: Fig. S10D-F). Thus, to investigate whether Airn regulated the eNOS-sGC pathway via KLF2, specific siRNA targeting KLF2 was transfected in Airn-over-expressed primary LSEC and the results showed that KLF2 silencing abrogated Airn over-expression-induced upregulation of eNos and downregulation of Laminin assessed by qRT-PCR (Fig. 7E). Taken together, our data demonstrated that Airn maintained LSEC differentiation through the KLF2-eNOS-sGC pathway, thereby inhibiting HSC activation and HC proliferation.

Generally, lncRNAs regulate their target genes by interacting with RNA binding proteins or acting as endogenous competing RNAs for specific microRNAs. To understand the molecular mechanism underlying the effects of Airn on liver fibrosis and LSEC capillarization, we first investigated the cell distribution of Airn using FISH and qRT-PCR. The results showed that Airn was mainly located in the nucleus of primary LSECs and HCs (Additional file 1: Fig. S11A-C). Moreover, our data demonstrated that Airn promoted the expression of KLF2 at both protein and RNA level in primary LSEC. Thus, a bioinformatics tool was used to screen for Airn-interacting proteins. The data showed that the 58 proteins including EZH2, which has been reported to bind to the promoter regions of KLF2 [27, 28], possibly interact with both human and mouse Airn (Additional file 1: Fig. S11D-G). To investigate this interaction, RIP assay was performed and the results showed a significant enrichment of Airn with the EZH2 antibody (Additional file 1: Fig. S12A, B). In addition, to identify the exact region of the Airn responsible for EZH2 interaction, we constructed the full-length Airn, a series of truncated Airn and the indicated antisense probe (Additional file 1: Fig. S12C). The results of RNA pull-down indicated that nucleotides 532 to 869 of Airn could bind EZH2, consistent with the prediction (Additional file 1: Fig. S12D). Moreover, ChIP analysis demonstrated that EZH2 could bind to promoter regions of KLF2 and over-expression of Airn reduced their binding ability (Fig. 7F, G). Taken together, these data demonstrated that Airn interacts with EZH2 and may block the binding site of EZH2 to KLF2, thus releasing the inhibition of KLF2 and LSEC differentiation related genes.