Introduction

Liver fibrosis is the final outcome of chronic hepatic damage caused by multiple etiologies, such as virus infections, alcohol over-consumption, and autoimmune or metabolic diseases1,2. In cirrhosis—the end stage of liver fibrosis—liver malfunction with severe complications arises due to the distortion of liver architecture caused by aberrant extracellular matrix (ECM) deposition. However, an important characteristic of liver fibrosis that can provide new therapeutic strategies is its reversibility, even at an advanced stage, upon removal of the causative injury3,4.

Hepatic non-parenchymal cells are involved not only in fibrogenesis but also in fibrosis resolution. Hepatic stellate cells (HSCs) and portal fibroblasts (PFs) are mesenchymal cells capable of transforming into myofibroblasts that aberrantly produce ECM proteins under fibrogenic conditions2,5. These cells are removed by apoptosis6 or senescence7 during fibrosis resolution; however, some remain in a quiescent state8. The molecular mechanisms underlying the transition from fibrogenesis to fibrosis resolution are not fully understood; however, shifting the balance between pro-fibrotic pathways, such as TGF-β1 and PDGF signaling9,10, and anti-fibrotic ones, such as peroxisome proliferator-activated receptor gamma (PPARγ) and cAMP pathways11,12,13, may mediate this transition.

Under fibrotic conditions, liver sinusoids undergo vascular remodeling, which includes angiogenesis and capillarization characterized by the loss of fenestrae and the formation of basement membrane14. As differentiated liver sinusoidal endothelial cells (LSECs) function to maintain the quiescent status of HSCs, loss of LSEC differentiation under fibrosis contributes to HSC activation15,16. Forced re-differentiation of LSECs by a soluble guanylate cyclase activator promotes fibrosis regression in rodent models17, suggesting that the normalization of LSECs to their differentiated state is a prerequisite for proper liver fibrosis resolution.

AKAP12 is a scaffold protein known to integrate effector proteins such as protein kinase A, protein kinase C, and protein phosphatase 2B for the β-adrenergic receptor pathway, through which it has a role in regulating cAMP compartmentalization and the re-sensitization of β-adrenergic receptors18,19. As AKAP12 is widely expressed by several cell types, it is reported to have diverse physiological functions, such as inhibition of angiogenesis20, promotion of meningeal reconstruction21,22, and tumor suppression23. In connection to chronic liver diseases, AKAP12 is reported to be downregulated by an epigenetic mechanism in human hepatocellular carcinoma24. However, little is known regarding its expression and function in normal and fibrotic livers.

In this study, we aim to investigate the molecular mechanisms by which non-parenchymal cells—in particular, PFs and LSECs—modulate hepatic fibrosis resolution. By examining the expression profiles of AKAP12 in normal and fibrotic livers, we investigate the role of this protein in hepatic fibrogenesis and fibrosis resolution.

Materials and methods

Animals and fibrosis induction

All mouse experiments were reviewed and approved by the Committee for Care and Use of Laboratory Animals at Seoul National University according to the Guide for Animal Experiments edited by the Korean Academy for Medical Sciences. Wild-type (WT) and AKAP12-deficient (AKAP12 KO) C57BL/6 mice were bred and maintained as described previously25. Hepatic fibrosis was induced by feeding mice a 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC)-supplemented diet; 0.1% DDC (Sigma) was mixed with normal mouse chow and fed to weight-matched 8- to 10-week-old male mice. The mice were kept on the DDC diet for 3 weeks, following which they were returned to a normal diet for the next 2 weeks. At the end of the experimental period, the mice were killed via deep anesthesia and killed by cardiac perfusion.

Human tissue specimens

Tissues from normal and fibrotic human livers were obtained through biopsy specimens from patients who visited the Asan Medical Center, University of Ulsan, College of Medicine, Seoul, Korea, for diagnosis. Informed consent was obtained from all patients. The patients’ clinical characteristics are summarized in supplementary Table 1. Fibrosis grades were determined by gross and microscopic examination of the liver tissues. The study was approved by the Institutional Review Board of the Asan Medical Center, Seoul, Korea.

Histological analysis and immunohistochemistry

Immunofluorescence staining was performed on paraffin sections or frozen sections. The paraffin sections were de-paraffinized and stained with Picro-Sirius Red (Abcam, Cambridge, MA, USA) according to the manufacturer’s instructions to visualize the deposition of ECM proteins. To stain the paraffin sections, antigen retrieval was conducted for 30 min at 95 °C in Tris or citrate buffer, depending on antibody compatibility. After blocking with 5% normal donkey serum (Sigma-Aldrich) in PBS solution, the sections were incubated overnight at 4 °C with primary antibodies for AKAP12 (I. Gelman, Roswell Park Cancer Institute), aSMA (Dako), laminin (Sigma), SE-1 (R&D Systems), CD31 (BD Biosciences), Thy-1 (BD Biosciences), collagen I (Southern Biotech), collagen III (Southern Biotech), Desmin (R&D Systems), VEGFR2 (Cell Signaling), F4/80 (Abcam), SE-1 (R&D Systems), and collagen IV (Southern Biotech). After extensive washing in PBS containing 0.1% Tween-20 solution, the sections were treated with Alexa 488- or 546-conjugated secondary antibodies (1:750, Invitrogen, Carlsbad, CA, USA) for 1 h at room temperature followed by counter staining with Hoechst (Sigma). Fluorescent images were taken under a confocal microscope (Carl Zeiss AG, Oberkochen, Germany), and immuno-positive areas were quantified by ImageJ software. Unless specified otherwise, 5–10 images from a single animal were quantified, and the mean value was considered to be a representative value for the animal.

RNA isolation and qPCR

Liver tissues were incubated in Trizol Reagent (Invitrogen) for 5 min, followed by homogenization using TissueLyser II (Qiagen, Hilden, Germany), and total RNA was extracted according to the manufacturer’s instructions. Two micrograms of RNA from each sample was reverse-transcribed with Moloney murine leukemia virus (MMLV) reverse transcriptase (Promega Corp., Madison, WI, USA). Quantitative real-time PCR was then performed using a StepOnePlus Real-Time PCR system (Applied Biosystems, Foster city, CA, USA) with a RealHelix qPCR kit (NanoHelix, Seoul, Korea). The relative mRNA levels were normalized using housekeeping genes, such as GAPDH, TBP, and 18S rRNA. The primer sequences used in this study are provided in Supplementary Table 2.

Cell culture and transfection

Primary human liver sinusoidal endothelial cells (hLSECs, Cell Biologics, Chicago, IL, USA) were maintained in complete human endothelial medium (Cell Biologics). Cells were transfected with control or SMARTpool siRNAs against human AKAP12 (GE Dharmacon) at passages 5 to 7 using oligofectamine transfection reagent (Invitrogen). Primary mouse portal fibroblasts were cultured using the biliary tree expansion method after intrahepatic collagenase perfusion as described previously26. Control or SMARTpool siRNAs against mouse AKAP12 (GE Dharmacon) were transfected at passages 1 to 2 using JetPrime transfection reagent (Polyplus transfection, Strasbourg, France).

mRNA sequencing analysis

Total RNA was isolated from mouse liver tissue using TRIzol® RNA Isolation Reagents (Life technologies, Carlsbad, CA). RNA integrity was confirmed by a bioanalyzer using an Agilent RNA 6000 Pico Kit (Agilent, Santa Clara, CA). The isolated total RNA was processed for preparing a mRNA sequencing library using a TruSeq stranded mRNA sample preparation kit (Illumina, San Diego, CA) according to the manufacturer’s instructions. Briefly, mRNA was isolated from 400 ng of total RNA using RNA purification beads for polyA capture, followed by enzymatic shearing. After first and second strand cDNA synthesis, A-tailing and end repair were performed for ligation of the proprietary primers, which incorporated unique sequencing adaptors with an index for tracking Illumina reads from multiplexed samples run on a single sequencing lane. For each library, an insert size of ~220 bp was confirmed by Bioanalyzer using an Agilent DNA Kit (Agilent, Santa Clara, CA), and quantification of library was measured by real-time PCR using a CFX96 real-time system (BioRad, Hercules, CA). Samples were sequenced with the Illumina NextSeq 500 platform with paired-end, 75-bp reads for mRNA-seq using a NextSeq 500/550 High Output Kit version 2 (for 150 cycles). Raw image data were transformed by base-calling into sequence data and stored in FASTQ format.

Trimming, alignment, and expression analysis

Three paired-end 75 bp reads from three replicates each of wild-type mouse sample and AKAP12 knock out mouse sample were trimmed for both PCR and sequencing adapters with Cutadapt. Trimmed reads were aligned to the mm10 mouse reference genome using STAR27. Quantification of gene expression was performed with Cufflinks to calculate FPKM values28.

Differential expression analysis

Significant differential expression was analyzed using edgeR29 at the gene level. A heatmap for differentially expressed genes was generated using R language. Functional classification was performed using DAVID30.

Gene set enrichment analysis

Gene set enrichment analysis (GSEA) was used to determine the extent to which expression profiles were enriched for a priori defined sets of genes from biologically coherent pathways31. GSEA was performed using version 2.2 of GSEA run on all gene sets in version 5.2 of the Molecular Signatures Database to correct for multiple hypotheses testing; the FDR threshold was set at ≤0.25.

Statistical analysis

All data are expressed as the mean ± SEM. One-way ANOVA, followed by Tukey’s tests or two-tailed Student’s t-tests, was used for statistical analyses. Differences between groups were considered significant for P  0.05. All statistical analyses were performed using GraphPad Prism V.5.00 (GraphPad Software, San Diego, California, USA).

Data availability

The RNA-seq data are publicly available through the National Center for Biotechnology Information’s Gene Expression Omnibus under accession number GSE104030.

Results

Expression patterns of AKAP12 in normal and fibrotic mouse livers

Immunofluorescence staining of normal mouse liver tissues showed that AKAP12 was expressed along the endothelium and in cells surrounding the portal tract but not in the central vein (Fig. 1a). Double immunofluorescence staining identified AKAP12-expressing cells in the portal tract as PFs as the AKAP12 signal merged with the PF markers thy-1 (Fig. 1b, upper row), desmin, and PDGFRα but not with EpCAM or CD31 (Supplementary Figure 1a). Interestingly, AKAP12-positive cells in the liver parenchyma were LSECs, as revealed by colocalization of AKAP12 with SE-1 signals (Fig. 1b, bottom row) and sinusoidal basement membrane marker collagen type IV (Supplementary Figure 1b). Desmin-positive HSCs were in close proximity to AKAP12-positive cells, but they rarely overlapped (Supplementary Figure 1b). The livers of AKAP12 KO mice exhibited disorganized sieve plates with a ~25% reduction in the number of fenestrae per unit area (Supplementary Figure 2a). However, under normal circumstances, we found no indication of liver disease or histological/functional changes due to the AKAP12 gene (data not shown). To investigate the role of AKAP12 in hepatic fibrosis, we used a biliary fibrosis model induced by DDC-feeding. Aberrant ECM accumulation was induced after 3 weeks of DDC-feeding, by which time AKAP12 expression drastically decreased in both PFs and LSECs (Supplementary Figure 2b and c). We then investigated the course of fibrogenesis and fibrosis resolution by feeding mice a DDC diet for 3 weeks, followed by feeding them a normal diet for 2 weeks (Fig. 1c). ECM accumulation, which was observed from the first week of DDC-feeding, gradually increased until the third week. Fibrosis resolution began as soon as the mice began feeding on a normal diet; regression in ECM deposition was clearly evident by the second week of fibrosis resolution (Fig. 1d). AKAP12 expression in the portal tract and sinusoids decreased sharply from the first week of DDC-feeding but was restored during fibrosis resolution (with the removal of DDC-induced injury), even reaching up to 80% of its normal level after 2 weeks of fibrosis resolution (Fig. 1e, f). These data indicate that AKAP12 expression in PFs and LSECs in the liver is dynamically altered during fibrogenesis and fibrosis resolution.

Fig. 1: AKAP12 is mainly expressed in PFs and LSECs of normal liver, and its expression is dynamically altered during fibrosis progression and resolution.
figure 1

a Paraffin sections from normal WT livers were subjected to immunofluorescence staining for AKAP12 (green). Note the specific immunoreactivity around the portal vein and liver sinusoids, which was not detected in AKAP12 KO livers (inlet). PV portal vein, CV central vein. b Double immunostaining of liver sections from WT mice for AKAP12 and other cell-type markers. AKAP12(+) cells in the portal tract are shown merged with thy-1(+) PFs, and AKAP12(+) cells in the liver parenchyma are shown merged with SE-1(+) LSECs. c–e Mice fed a normal diet (ND) or DDC-containing diet for 1–3 weeks (DDC 1–3W) or DDC diet for 3 weeks followed by a normal diet for 1–2 weeks (RES 1, 2W) (c). Liver sections from each experimental group stained with Sirius Red (d), AKAP12 and αSMA (e), and AKAP12 or laminin (f). Graphs show morphometric quantification of staining-positive areas relative to control liver sections. Data are expressed as the mean ± SEM. n = 7–9 per group. Scale bars represent 100 µm in a, b, e, and f, and 500 µm in d

Full size image

Expression patterns of AKAP12 in normal and fibrotic human livers

Immunohistochemical analyses were used to detect AKAP12 in normal and fibrotic human liver specimens to understand its correlation with human liver conditions. In normal human livers, AKAP12 immunoreactivity was prominent along liver sinusoids and the areas surrounding the portal tract (Fig. 2a), analogous to its expression profile in normal mouse livers. We then analyzed human liver samples with fibrosis caused by three different etiologies, namely, hepatitis B virus (HBV), hepatitis C virus (HCV), and NBNC (non-HBV/non-HCV). Fibrosis grading into stage 1 (portal fibrosis), stage 2 (periportal fibrosis), stage 3 (septal fibrosis), and stage 4 (cirrhosis) was done by pathologists. Liver tissues from a total of 28 patients—4 samples with normal liver histology and 2 samples for each stage of hepatic fibrosis caused by each etiology mentioned previously—were subjected to immunostaining for AKAP12. Intriguingly, AKAP12 immunoreactivity was greatly reduced, even at the early stages of fibrosis (stage 1), irrespective of etiology (Fig. 2b). The average AKAP12-positive area in fibrotic liver samples relative to normal liver samples was the lowest in HBV-induced fibrosis, but the differences among the various etiologies were not statistically significant (Fig. 2c). Besides this, although the mean AKAP12-positive area was the smallest in cirrhotic livers (stage 4), the differences among the various stages were not statistically significant (Fig. 2d). These results suggest that the AKAP12 gene is an early responder to fibrotic changes caused by a wide range of hepatic injuries and that it is involved in the pathogenesis of human liver fibrosis.

Fig. 2: AKAP12 is downregulated in fibrotic human livers.
figure 2

a Paraffin sections from normal liver biopsies were stained for AKAP12 (green) and αSMA (red). AKAP12 immunoreactivity was prominent in the portal tract (middle) and sinusoids (right). The inset within the last image to the right shows anti-rabbit IgG staining. Scale bars represent 100 µm. b Representative images of AKAP12 (green) and αSMA (red) staining on human liver specimens with fibrosis from stage 1 to 4 caused by hepatitis B virus (HBV, top row), hepatitis C virus (HCV, middle row), or non-HBV/non-HCV (NBNC, bottom row). Scale bar represents 100 µm. c, d Graphs represent relative AKAP12-positive area analyzed by etiologies (c) or fibrosis stages (d) (n = 4 in normal, n = 8 in each etiology group, n = 6 in each stage group). Data are expressed as the mean ± SEM. **P < 0.01, ***P < 0.001

Full size image

Effect of AKAP12 deficiency on DDC-induced liver fibrosis and fibrosis resolution

To investigate the role of AKAP12 in fibrosis progression, we examined the degree of fibrosis in WT and AKAP12 KO mice in response to a DDC challenge; afterwards, we also studied fibrosis resolution after the removal of the DDC challenge. Comparisons of ECM deposition levels after 3 weeks of DDC-feeding showed no significant differences between WT and AKAP12 KO livers (Fig. 3a). However, AKAP12 KO mice displayed reduced fibrosis resolution after being shifted to a normal DDC-free diet for 2 weeks, as revealed by Sirius Red (Fig. 3a), collagen I (Fig. 3b), and collagen III staining (Fig. 3c). Reductions in fibrosis-related ECM gene expression during fibrosis resolution were more apparent than histological regression (Fig. 3d), suggesting that most of the ECM-producing cells were either removed or returned to their quiescent state. In contrast, mRNA levels of these ECM components were approximately two- to threefold higher in AKAP12 KO mice than in WT mice during fibrosis resolution (Fig. 3d). Despite impaired fibrosis resolution in AKAP12 KO mice, the serum levels of liver enzymes returned to normal 2 weeks after the mice were shifted to a normal diet (Supplementary Figure 3).

Fig. 3: AKAP12 KO mice display reduced fibrosis resolution accompanied by impaired reversion of activated myofibroblasts and vascular remodeling.
figure 3

a–c Liver sections from WT or AKAP12 KO mice fed a normal diet (ND), DDC-containing diet for 3 weeks (DDC), and DDC-containing diet for 3 weeks followed by a normal diet for 2 weeks (RES), which were stained with Sirius Red (a) and IHC against collagen I (b), and collagen III (c). Scale bars indicate 500 µm. Graphs are morphometric quantifications of areas stained by Sirius Red, collagen I, and collagen III. Data are expressed as the mean ± SEM. n = 4 mice for ND, n = 10 mice for DDC, n = 11–12 mice for RES (a). n = 2 mice for ND, n = 5 mice for DDC, n = 5–6 mice for RES (b, c). d Relative mRNA levels of Col1a1, Fibulin1, and Elastin in liver tissues were quantified by qPCR with specific primers. n = 2 mice for ND, n = 5 mice for DDC, n = 5 mice for RES. Data are expressed as the mean ± SEM. e, f Liver sections were subjected to IHC to detect laminin and desmin (e) or VEGFR2 and F4/80 (f). Scale bars indicate 100 µm. Graphs show relative areas positive for each type of staining. Data are expressed as the mean ± SEM. n = 2 mice for ND, n = 5 mice for DDC, n = 5–6 mice for RES. *P < 0.05; **P < 0.01; ***P < 0.001; NS not significant

Full size image

On the basis of the expression patterns of AKAP12 and the lower levels of fibrosis resolution observed in AKAP12 KO mice, we hypothesized that the restoration of AKAP12 expression during fibrosis resolution led to the deactivation of myofibroblasts and normalization of remodeled sinusoids. Although the population of desmin-positive myofibroblasts grew in the portal area during fibrogenesis and then decreased during fibrosis resolution in WT mice, a significant number of desmin-positive myofibroblasts remained in the livers of AKAP12 KO mice during fibrosis resolution (Fig. 3e). Vascular remodeling, which includes angiogenesis and capillarization of sinusoids, is characteristic of hepatic fibrosis irrespective of etiology14,32. DDC-induced fibrosis also led to aberrant angiogenesis within the fibrotic area, as revealed by VEGFR2 staining (Fig. 3f). Furthermore, DDC-induced liver injury led to the formation of a sinusoidal basement membrane as revealed by laminin staining (Fig. 3e). Vascular remodeling was reversed as fibrosis resolved in WT mice. In AKAP12 KO mouse livers, however, VEGFR2-positive and laminin-positive staining was higher compared to that in WT mouse livers during fibrosis resolution (Fig. 3e, f), implying that the reversion of remodeled sinusoids to differentiated ones in AKAP12 KO mice is lower than that in WT mice. Although F4/80-positive myeloid cell populations increased during fibrosis and then reduced as fibrosis resolved, we did not observe significant differences in the number of these cells between WT and AKAP12 KO mice (Fig. 3f). Taken together, these results suggest that the reduced fibrosis resolution observed in AKAP12 KO mice is associated with impaired normalization of myofibroblasts and liver sinusoids.

Transcriptome analysis of WT and AKAP12 KO livers during fibrosis resolution

To investigate the molecular and cellular mechanisms underlying the positive effects of AKAP12 on fibrosis resolution, the transcriptomes of WT and AKAP12 KO livers during fibrosis resolution were analyzed by RNA sequencing (RNA-seq). Three mice were randomly selected from each genotype for RNA-seq analysis. A total of 1033 genes were found to be differentially expressed between the two genotypes with P < 0.05. Among these 1033 differentially expressed genes, 579 genes were upregulated in AKAP12 KO samples. In agreement with the results from immunohistochemical and quantitative polymerase chain reaction (qPCR) analyses, genes belonging to GO terms “extracellular matrix structural constituent,” “collagen binding”, and “extracellular matrix binding” were mostly elevated in AKAP12 KO samples (Fig. 4a, left). In addition, expression levels of genes belonging to the GO terms “positive regulation of angiogenesis” and “basement membrane” were also mostly elevated in AKAP12 KO samples (Fig. 4a, right). Molecular pathways, such as focal adhesion, GTPase-mediated signal transduction, and the adenosine 3′,5′-cyclic monophosphate (cAMP) signaling pathway, which can inhibit fibrosis12, were predicted to be associated with impaired fibrosis resolution in AKAP12 KOs by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway (Fig. 4b) and GO analysis (Fig. 4c, d).

Fig. 4: Transcriptome analysis of WT and AKAP12 KO livers at resolution state.
figure 4

a Heatmap of differentially expressed genes belong to GO terms “extracellular matrix structural constituent,” “collagen binding”, and “extracellular matrix binding” (left) and “positive regulation of angiogenesis” and “basement membrane” (right) showing differences in gene expression levels between WT and AKAP12 KO samples. Each row represents one gene, and each column represents one sample. The scale bar indicates row z-scores of gene expression values, with highly expressed genes depicted in pink and genes with low-level expression depicted in blue. Cluster analysis of the gene expression data was performed to group genes with similar expression patterns across experimental conditions. b Enrichment analysis of the KEGG pathway assignments of genes potentiated in AKAP12 KO livers. c Enrichment analysis of gene ontology biological processes (GOBP) of genes modulated in AKAP12 KO livers. d Gene ontology analysis for molecular function (GOMF) of genes modulated in AKAP12 KO livers

Full size image

Role of AKAP12 in activation of PFs and association with cAMP pathway

In accordance with the RNA sequencing results, qPCR revealed that transcript levels of PDE4a (a member of the PDE4 family of proteins that hydrolyze cAMP) were elevated in AKAP12 KOs during fibrosis resolution (Fig. 5a). It was previously demonstrated that forskolin, a cAMP activator, attenuated CCL4-induced liver fibrosis in a rodent model9. Moreover, rolipram, a PDE4 inhibitor, attenuated bile-duct ligation (BDL)-induced fibrosis in rats33, implying that the cAMP signaling pathway has an inhibitory role in hepatic fibrosis. Consistent with these in vivo studies, treatment of primary cultures of mouse PFs with forskolin significantly reduced the expression of myofibroblast-related genes (Fig. 5b). The inhibitory effect of forskolin on Col1a1, Col15a1, and Elastin gene expression was reversed by co-treatment with a PKA inhibitor, H89, whereas the expression levels of Col3a1 and αSMA were unaffected by PKA inhibition (Fig. 5b); this result means that increased cAMP levels induced by forskolin resulted in the inhibition of myofibroblastic gene expression, which is mediated in part through the activation of PKA.

Fig. 5: AKAP12 inhibits activation of PFs in association with the cAMP signaling pathway.
figure 5

a qPCR of liver tissue during fibrosis resolution showing relative mRNA levels of gene members of the PDE4 family (n = 6 mice per genotype). b Primary mouse PFs were treated with either forskolin (10 µM) alone or with a combination of forskolin (10 µM) and H89 (10 µM) for 16 h. Relative mRNA levels of Col1a1, Col3a1, Col15a1, Elastin, and αSMA were measured by qPCR with specific primers. n = 3 for each group. c, d Mouse PFs were transfected with control or AKAP12 siRNAs. Graphs show mRNA expression levels of AKAP12 and myofibroblast-related genes in siAKAP12-treated PFs relative to those in control siRNA-treated PFs (siCtrl) (c). PFs transfected with siAKAP12 were treated with rolipram (20 µM) for 24 h, and the relative levels of ECM gene expression were measured by qPCR (d). n = 3 repeats. Data are expressed as the means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001

Full size image

On the basis of the association of AKAP12 with the cAMP pathway18 and the inhibitory effect of cAMP signaling on myofibroblast activation, we hypothesized that AKAP12 may have an inhibitory role in hepatic fibrosis via modulation of the cAMP pathway. We then examined the role of AKAP12 in regulating the myofibroblastic activation of PFs in a cell-intrinsic manner. Mouse primary PFs were transfected with either control siRNA or AKAP12 siRNA, following which the expression profiles of key myofibroblast marker genes were measured by qPCR. Although the levels of αSMA were comparable, AKAP12 knockdown PFs expressed more Col1a1, Col3a1, Col15a1, and Elastin than controls (Fig. 5c). These effects were, however, abolished by rolipram treatment (Fig. 5d), indicating that AKAP12 knockdown led to the activation of PDE4-mediated cAMP hydrolysis. These results indicate that AKAP12 inhibits the activation of PFs and that this effect is probably associated with its function as a scaffold protein in the cAMP signaling pathway.

Role of AKAP12 in sinusoidal remodeling

Soluble factors are believed to mediate the crosstalk between liver sinusoids and hepatocytes or other non-parenchymal cells34,35. To identify the factor(s) that could mediate the effects of AKAP12 on vascular remodeling, we measured levels of various angiocrine factors. We found that in AKAP12 KO livers, the levels of endothelin-1 (ET-1) are upregulated during fibrosis resolution, whereas the levels of hepatocyte growth factor (HGF), wingless-type MMTV integration site family member 2 (Wnt2), and vascular endothelial growth factor (VEGF) were comparable to those in the livers of WT littermates (Fig. 6a). Interestingly, serum levels of ET-1 were also elevated in AKAP12 KO mice, not only during fibrosis resolution but also under normal conditions (Fig. 6b). To test whether AKAP12 inhibits vascular remodeling by directly regulating LSECs, human primary LSECs (hLSECs) were transfected with siRNAs against AKAP12, following which the expression levels of basement membrane components and the angiogenic capacity of cells were measured. Among the laminins, Laminin α1 and α5 were upregulated by AKAP12 knockdown (Fig. 6c), suggesting that AKAP12 is involved in basement membrane reconstitution. Furthermore, AKAP12 knockdown in hLSECs increased ET-1 gene expression (Fig. 6c) and caused the formation of tube-like structures (Fig. 6d). These results suggest that in LSECs, AKAP12 inhibits basement membrane formation, ET-1 expression, and angiogenesis in a cell-intrinsic manner. To verify the negative relationship between AKAP12 expression and sinusoidal remodeling, we analyzed a publicly available microarray data set (GSE1843) on liver endothelial cells (LECs) isolated from rat livers with or without CCL4-induced cirrhosis36. The data indicated that levels of AKAP12 mRNA were significantly decreased in cirrhotic LECs and negatively correlated with levels of ET-1 or Laminin α5 transcripts (Fig. 6e). Therefore, it is likely that reduced AKAP12 expression levels during fibrogenesis promote sinusoidal remodeling, whereas restoration of AKAP12 expression during fibrosis resolution accelerates the recovery of normal sinusoidal characteristics.

Fig. 6: AKAP12 deficiency leads to ET-1 gene expression, laminin reconstitution, and angiogenesis of LSECs.
figure 6

a qPCR of liver tissue during fibrosis resolution revealed increased expression of ET-1 in AKAP12 KO mice (n = 5 mice per genotype). b Serum levels of ET-1 measured by ELISA from WT or AKAP12 KO mice (n = 3 mice in ND groups, n = 5 mice in DDC groups, n = 6 mice in RES groups). c hLSECs were transfected with control or AKAP12 siRNAs, following which relative mRNA levels were measured by qPCR. Note that AKAP12 knockdown resulted in increased expression of Laminin α1 (lama1), Laminin α5 (lama5), and ET-1 (n = 3 repeats). d Matrigel tube formation assays on hLSECs show an increased number of tube rings in AKAP12 knockdowns (n = 3 repeats). Data are expressed as the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001. e An analysis of GEO data set GSE1843 showing significant reduction in AKAP12 in liver endothelial cells (LECs) isolated from cirrhotic rats (left graph). Pearson correlation of AKAP12 levels and ET-1 (middle graph) or lama5 (right graph) levels shows negative correlation values (n = 3 samples per group)

Full size image

Discussion

As liver fibrosis can be naturally reversed by removing injury-causing agents, understanding the precise mechanisms mediating fibrosis resolution can provide new therapeutic strategies for fibrosis treatment. In this study, we identified AKAP12 as a key molecule that promotes fibrosis resolution. AKAP12 expression in both PFs and LSECs was downregulated during fibrogenesis but was restored during fibrosis resolution. Reduction in AKAP12 expression levels during liver fibrosis was not only observed in a mouse model but was also observed in human liver tissues exhibiting various stages of fibrosis caused by different etiologies. AKAP12 KO mice displayed reduced fibrosis resolution in association with myofibroblast and endothelial characteristics, implying that recovery of AKAP12 expression is required for proper normalization of portal fibroblasts and liver sinusoids during fibrosis resolution.

Dynamic alteration of AKAP12 expression during fibrosis onset and resolution suggests that its expression can be regulated by microenvironmental factors. Our group previously reported that retinoic acid (RA) and TGF-β1 can induce AKAP12 expression to mediate epithelial-to-mesenchymal transition in meningeal reconstruction; however, this effect was abolished by local hypoxia21. We have also shown that during central nervous system (CNS) vascular development, AKAP12 expression is decreased due to hypoxia during active angiogenesis but can be restored through reoxygenation20. Interestingly, TGF-β1 is a potent inducer of hepatic fibrosis, and the liver is a retinol-rich organ that takes up ~70% of dietary retinol37. Moreover, several lines of evidence indicate that tissue hypoxia occurs during hepatic fibrosis and that hypoxia-mediated signaling affects a wide array of cellular responses, such as myofibroblast activation, angiogenesis, and immune responses38,39,40. Therefore, it is possible that these microenvironmental factors modulate AKAP12 expression in hepatic fibrosis.

RNA sequencing analysis strongly suggested that the antifibrogenic role of AKAP12 is associated with the cAMP pathway. Indeed, members of the AKAP family are scaffold proteins that fine-tune cAMP signaling through spatio-temporal regulation18. AKAP12 is mostly localized to the plasma membrane, where it mediates the subcellular compartmentalization of cAMP19. cAMP is a second messenger in the G-protein-coupled receptor (GPCR) signal cascade that regulates a wide range of cellular responses. Reports indicate that in liver fibrosis, either direct activation of cAMP signaling by forskolin12 or inhibition of the cAMP-hydrolyzing enzyme PDE433 attenuated fibrogenesis in animal models; this finding suggests that boosting cAMP signaling could have therapeutic value in treating liver fibrosis. Nevertheless, stimulating cAMP signaling directly has limited therapeutic value due to unacceptable side effects caused by its broad-range effects on numerous cellular processes41. In this regard, using AKAP12 to stimulate cAMP signaling could provide new therapeutic approaches that minimize undesirable side effects.

In conclusion, our results suggest that AKAP12, whose expression is dynamically altered during fibrosis onset and resolution, has a central role in the proper resolution of liver fibrosis.