Curcumol attenuates liver sinusoidal endothelial cell angiogenesis via regulating Glis‐PROX1‐HIF‐1α in liver fibrosis

Abstract Objective Hepatic sinusoidal angiogenesis owing to dysfunctional liver sinusoidal endothelial cells (LSECs) accompanied by an abnormal angioarchitecture is a symbol related to liver fibrogenesis, which indicates a potential target for therapeutic interventions. However, there are few researches connecting angiogenesis with liver fibrosis, and the deeper mechanism remains to be explored. Materials and Methods Cell angiogenesis and angiogenic protein were examined in primary LSECs of rats, and multifarious cellular and molecular assays revealed the efficiency of curcumol intervention in fibrotic mice. Results We found that curcumol inhibited angiogenic properties through regulating their upstream mediator hypoxia‐inducible factor‐1α (HIF‐1α). The transcription activation of HIF‐1α was regulated by hedgehog signalling on the one hand, and the protein stabilization of HIF‐1α was under the control of Prospero‐related homeobox 1 (PROX1) on the other. A deubiquitinase called USP19 could be recruited by PROX1 and involved in ubiquitin‐dependent degradation of HIF‐1α. Furthermore, our researches revealed that hedgehog signalling participated in the activation of PROX1 transcription probably in vitro. Besides, curcumol was found to ameliorate liver fibrosis and sinusoid angiogenesis via hedgehog pathway in carbon tetrachloride (CCl4) induced liver fibrotic mice. The protein expression of key regulatory factors, PROX1 and HIF‐1α, was consistent with the Smo, the marker protein of Hh signalling pathway. Conclusions In this article, we evidenced that curcumol controlling LSEC‐mediated angiogenesis could be a promising therapeutic approach for liver fibrosis.


| INTRODUC TI ON
Angiogenesis is a hypoxia-induced and growth factor-dependent process in which endothelial cells are budded, accompanied by migration, expansion and lumen formation from the original vascular structure. Liver fibrosis is a common consequence of injury and repair reaction of various chronic liver diseases (CLDs), and is related to angiogenesis closely. 1 There are a large number of pseudolobular fibrous nodules in hepatic tissue during liver fibrosis, which lead to hepatic sinusoidal blood flow disorder and oxygen delivery reduction. These changes make liver intrinsic cells, such as hepatocytes and activated hepatic stellate cells (HSCs), secret angiogenic factors in order to regulate endothelial angiogenesis. Of note, neovascularization destroys hepatic architecture and promotes sinusoidal remodelling, aggravating liver fibrosis. As a result, pathological angiogenesis and fibrogenesis develop in parallel during progression of CLD. 2 Therefore, the inhibition of pathological angiogenesis has also become a significant method to alleviate liver fibrosis.
The LSECs account for the vast majority of liver non-parenchymal cells and are directly involved in hepatic angiogenesis. 3 The natural LSECs lack an organized basement membrane and have numerous fenestrae grouped into sieve plates, which benefit to mass exchange of hepatic sinusoid and resist exogenous invasion. During the fibrogenic progression of CLD, different factors have made LSEC lose their distinctive morphology through a process named capillarization.
Neovascularization exacerbates liver hypoxia and resistance to blood flow, which increases transcription of many hypoxia-sensitive pro-angiogenesis genes such as vascular endothelial growth factor (VEGF), 4 platelet-derived growth factor (PDGF) and angiopoietin. 5 All these processes are regulated by the hypoxia-inducible factor-1α (HIF-1α). 6 Basic and clinical research also showed that inhibition of LSEC angiogenesis and HSC paracrine effects could alleviate liver fibrosis. 5,7 The hedgehog (Hh) pathway is a conserved morphogenic signalling pathway that modulates the fate of LSECs, including capillarization and angiogenesis. 8,9 There is growing evidence showed that liver injury can activate Hh pathway tremendously. 10 The interaction between increasing sonic hedgehog protein (SHH) and membrane receptor Patched can replace the combination of Patched between intracellular co-receptor-like molecule Smoothened (Smo). The latter eventually results in the initiation of Glis-dependent canonical hedgehog signalling. 11 In this progress, Gli-family transcription factors (Gli1, Gli2 and Gli3) that translocate to nucleus from cytoplasm, which regulate the Hh target gene transcription, affect cell viability, proliferation and differentiation. Previous studies disclosed that Hh can activate pro-angiogenic gene via regulating HIF-1α to promote HSC angiogenesis. 12 The potential mechanisms of Hh pathway regulating the phenotype of LSEC, however, remain uncovered. Here, we have mainly investigated how the Hh pathway regulates LSEC angiogenesis.
Curcumol is extracted from the roots of the herb called Rhizoma Curcumae, and it possesses a variety of pharmacological activities including anti-inflammatory and anti-tumour effect. Previous researches had revealed that curcumol can inhibit the proliferation of nasopharyngeal carcinoma CNE-2 cells by regulating insulin-like growth factor 1 receptor (IGF-1R) and its downstream PI3K/Akt/GSK-3beta pathway. 13 Besides, curcumol can suppress breast cancer cell metastasis by inhibiting the expression of MMP-9. 14 These studies suggested that curcumol is possible to participate in cell proliferation and migration correlated with angiogenesis closely. Moreover, a recent study demonstrated that curcumol is capable of alleviating liver fibrosis via triggering HSC necroptosis. 15 Given the anti-fibrotic effect of curcumol, we have investigated whether curcumol can inhibit hepatic fibrosis via regulating LSEC angiogenesis and elucidated the underlying mechanisms.

| Cell culture
Primary LSECs were isolated from male SD rats as described. 8 Isolated LSECs were cultured in complete medium made from Dulbecco's modified Eagle's medium (DMEM) (Procell) supplemented with foetal bovine serum (FBS) (Gibco) (9:1) and 1% double antibiotics (Streptomycin and aspirin) and placed in 37°C and 5% CO 2 incubator. LSECs entering logarithmic growth period were used in experiments. Cell morphology was assessed via microscope with a Leica Qwin System (Leica).

| Cell viability assay
LSECs were planted in 96-well plates (about 10 4 cells per well) and treated with various reagents when they would be 70%-90% confluent for 24 hour. MTT (Biosharp) was dissolved in PBS at a concentration of 5 mg/mL. The configured MTT was dissolved into the total solution and added to the 96-well plates (20 μL per well). Then, 96well plates were put in incubator for 6 hours. Dimethyl sulphoxide (DMSO) was added to the 96-well plates (150 μL per well) to dissolve the bottom crystal. The spectrophotometric absorbance was measured by a spectrophotometer (Molecular Devices) at 490 nm. We set up six duplicate wells for each group in this experiment.

| Western blot analysis
The LSEC or fresh liver from mice was dissolved with lysis buffer consisting RIPA, PMSF and protease inhibitor (100:1:1). Protein concentration was measured by BCA kit (Beyotime Biotech). We

Conclusions:
In this article, we evidenced that curcumol controlling LSEC-mediated angiogenesis could be a promising therapeutic approach for liver fibrosis. evaluated the level of target protein expression according to the depth of bands until the β-actin had the same expression. The strips were quantified with Image Lab (NIH). Nuclear proteins and cytoplasm proteins were separated using a Nuclear and Cytoplasmic Extraction Kit (Wanleibio).

| Quantitative reverse transcription (PCR)
TRIzol reagent (Sigma) was used to extract total RNA from treated LSECs, and then, the RNA was subject to reverse transcription to cDNA using the kits provided by Yeasen Biotech Co., Ltd after purification. Then, the real-time PCR was performed using the SYBR Green Master Mix. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the invariant control. The primers of all genes are listed in Table 1.

| Tube formation assay
In tube formation assay, a total volume of 150 μL (10 μg/mL) Corning Matrigel Basement Membrane Matrix covered in 24-well plates was pre-cooled for 24 hours, then put in incubator at 37°C for 30 minutes to solidify. LSECs were seeded into wells (nearly 5 × 10 5 per well) after treated with different reagents. Plates were incubated at 37°C and then observed every hour under microscope, and tube length was measured via ImageJ software (NIH).

| Co-immunoprecipitation (Co-IP) assay
Treated LSECs were extracted at 4°C in RIPA buffer containing protease inhibitors. Cell lysates diluted to 1 mg/mL protein concentration were co-incubated by IP-grade antibodies against HIF-1α (1:200, CST). After gentle shaking at 4°C overnight, protein A/G PLUS-Agarose (Santa Cruz Biotech) was added to the lysate/antibody mixture and incubated with gentle rocking at 4°C for 4 hours.
Then, the immunoprecipitates were collected by centrifugation and washed three times with cell lysis buffer, then boiled for 5 minutes with the same volume of 4× loading buffer. Proteins were resolved by 10% SDS-PAGE and subjected to Western blotting assays.

| Ubiquitination assay
After treated LSECs were incubated for 24 hours, partial groups were treated with MG132 (10 mg/mL) for another 6 hours. Cell lysates diluted to 1 mg/mL protein concentration were pulled down by IP-grade antibodies against HIF-1α at 4°C overnight. Proteins were resolved by 10% SDS-PAGE. We evaluated the ubiquitination levels of treated cells by anti-ubiquitin antibody.

| Animals and procedures
Male ICR mice (20-25 g body weight) were purchased from Qinglongshan Animal Co, Ltd. All animals were raised in Nanjing  15 All animals got 24 hours fasting at the end of the experiment and then executed to take blood and liver.

| Serum biochemistry
Blood samples were allowed to stand for 1 hours after taken from the eyelids of mice, and serum was extracted after centrifugation and liquated. The serum levels of alkaline phosphatase (ALP), aspartate aminotransferase (AST), alanine aminotransferase (ALT), hyaluronic acid (HA), laminin (LN) and procollagen type III (PCIII) were measured by ELISA kits (Nanjing Yifeixue Biotech).

| Histological analysis
Liver tissues were soaked in 5% paraformaldehyde (PFA) and embedded in paraffin. Haematoxylin-eosin staining (HE) was used for pathological assessments according to the organizational structure. Masson staining and Sirius red staining were used for evaluating collagens. The microscope (ZEISS Axio Vert.A1) was used to take photographs for these staining sections at random fields. Liver tissues were fixed in 0.5% glutaraldehyde and taken photographs at surface of hepatic sinusoid with scanning electron microscope.

| Immunofluorescence staining
For cell assays, treated LSECs were fixed with 4% PFA at 37°C for

| Statistical analysis
Data were presented as mean ± SD, and the results were analysed using the GraphPad Prism 7.0 (GraphPad Software). The significance of difference was determined by one-way analysis of variance with post hoc Dunnett's test. All of the experiments were repeated at least three times on separate occasions. P < .05 was considered statistically significant.

| Curcumol inhibits LSEC angiogenesis in vitro
In view of the important role of pathological angiogenesis in the occurrence and development of liver fibrosis, the effect of curcumol on LSEC angiogenesis was further investigated. As the hepatic fibrosis occurs, the activated HSCs secrete a large amount of extracellular matrix and the important angiogenic factor VEGF, thereby affecting the perihepatic sinus microenvironment. 12 More studies revealed that hepatocytes can also express VEGF to change the phenotype of LSEC. 16 Therefore, we used VEGF-A (40 ng/mL) in vitro to mimic the living environment of LSECs under pathological condition. 8 Firstly, cell viability assay provided three rationale dosages of curcumol, at which they did not significantly change cell viability, to be applied to the subsequent experiment in vitro ( Figure 1A and Figure S1A).
Primary LSECs isolated from rats would lose their fenestrae in 2 days of culture, and medium dose of curcumol (20 μmol/L) had no impact on this progress ( Figure 1B). The number and the size of fenestrae of LSECs are based on the oxygen content surroundings. There are various factors controlling LSEC capillarization, and curcumol did not have capacity to reverse this progress in vitro. We found that capillary marker endothelin-1 mRNA level remained unchanged; however, the angiogenic factor mRNA level was inhibited by curcumol ( Figure 1C).
Similar results were obtained in terms of the corresponding proteins ( Figure 1D,E). Furthermore, the tube formation assay showed that curcumol significantly inhibited LSEC-mediated tube formation ( Figure 1F). Altogether, these data demonstrated that the angiogenic properties of LSECs were negatively regulated by curcumol in vitro.

| HIF-1α is involved in curcumol inhibition of LSEC angiogenesis
We next sought to investigate the underlying regulation mechanism of LSEC angiogenesis by curcumol, given that various transcription factors were induced in fibrotic liver due to hypoxia environment. 17 Previous studies had presented that VEGF was a major target gene  . Scale bar = 20 μm. F, Tubulogenesis assay was visualized and quantified with ImageJ (n = 3). Scale bar = 500 μm. * P < .05, ** P < .01 and *** P < .005 versus control group; # P < .05, ## P < .01 and ### P < .005 versus model group of HIF-1α, and we found hypoxia could increase the expression of HIF-1α, VEGF-A and VEGFR2 ( Figure S2A). The curcumol decreased the mRNA and protein levels of HIF-1α after VEGF-A induction ( Figure 2A-C). Interestingly, we found that VEGF-A failed to change the HIF-1α level of LSEC living at low density ( Figures S2B and S3A).
It may be that rapid cell growth induced by VEGF-A leads to highdensity LSEC hypoxia. Then, we used a HIF-1α inhibitor named acriflavine (ACF) to confirm the role of HIF-1α in LSEC angiogenesis.
ACF (0.5 μmol/L) reduced the rise of pro-angiogenic properties of LSECs induced by VEGF-A that were equal to curcumol ( Figure 2D).
Furthermore, immunofluorescence staining for angiopoietin 2 had obtained a similar result ( Figure 2E). The tube formation assay also showed that ACF (0.5 μmol/L) could decrease the length of the tubes that resemble the effect of curcumol ( Figure 2F). Overall, these data convincingly revealed that curcumol inhibited pro-angiogenic properties of LSECs by regulating HIF-1α.

| Curcumol inhibits LSEC angiogenesis by inhibiting Hh signalling pathway
As we know, activated HIF-1α transcription and enhanced HIF-1α protein stability could lead to an increase in HIF-1α expression. 18 Figure 3A,B). But we also found that curcumol did not change the level of Patched significantly.
Moreover, curcumol could decrease the abundance of Gli1 in nucleus of LSECs, which impaired the hedgehog signalling transcription-promoting activity ( Figure 4A). Immunofluorescence staining assays also confirmed above ( Figures 3C and 4B). Subsequently, to further insure whether hedgehog signalling directly participates in the impairment of LSEC angiogenesis by curcumol, we used cyclopamine at 3 μmol/L and SAG at 0.3 μmol/L to block or sensitize hedgehog signalling pathway. As shown in Figure 4C, SAG could offset the inhibiting effect of curcumol on LSEC angiogenesis, whereas cyclopamine exerted synergistic action with curcumol. Nuclear transcription factor HIF-1α that functions as upstream of angiogenic genes was regulated by hedgehog signalling.
As expected, the tube formation assay also confirmed the capacity of hedgehog signalling on LSEC angiogenesis ( Figure 4D). Collectively, these findings provided a support that Hh signalling pathway was critically involved in curcumol-mediated LSEC angiogenesis.

| PROX1 regulated by hedgehog signalling pathway maintains accumulation of HIF-1α in LSECs
Numerous studies have identified PROX1 as an important promoter of HCC angiogenesis, 19,20 and the levels of PROX1 were positively correlated with HIF-1α protein stability. 21 Therefore, we speculated that PROX1 could maintain accumulation of HIF-1α in LSECs and promote LSEC angiogenesis. Firstly, transfection efficiency of PROX1 siRNA or PROX1 plasmid was analysed by determining the levels of protein and mRNA, and the results demonstrated that the PROX1 successfully decreased or increased in LSECs, respectively ( Figure 5A).
Transfecting LSECs after treated with VEGF-A, we found that PROX1 siRNA apparently repressed the expression of HIF-1α and pro-angiogenic factors in LSECs, whereas PROX1 plasmid elevated their expression ( Figure 5B).
The PROX1-mediated elevation in HIF-1α level might be due to an  Figure 5C). Since HIF-1α is degraded through an ubiquitin-dependent mechanism, we wondered that PROX1 can inhibit HIF-1α ubiquitin by recruiting a deubiquitinase called USP19. Interestingly, PROX1 was found to co-localize with USP19 and/or HIF-1α in LSECs via confocal laser scanning microscope ( Figure 5D). Moreover, the amount of USP19 co-precipitated with HIF-1α was increased in PROX1-overexpressing LSECs, while it was diminished in PROX1knockdown LSECs ( Figure 5E). These results suggested that PROX1 inhibits HIF-1α ubiquitination via its association with USP19. We further examined the level of HIF-1α ubiquitination. Ubiquitin-conjugated

HIF-1α was pulled down by anti-HIF-1α and subjected to Western
blotting with anti-ubiquitin. Upon PROX1 overexpression, the quantity of the ubiquitin-conjugated HIF-1α was markedly diminished, and an opposite experimental result was found in PROX1-knockdown LSECs ( Figure 5F). In addition, we found that the agonist of hedgehog signalling increased the mRNA and protein levels of PROX1 significantly; the antagonist, however, had a feeble effect ( Figure 5G,H). We concluded that the primary LSECs express little activity of hedgehog signalling so that they react to its antagonist inactively. Taken together, these results indicated that transactivation of PROX1 by canonical hedgehog signalling could inhibit HIF-1α ubiquitination.

| Curcumol inhibits LSEC angiogenesis via regulating PROX1
We next determined the functional contribution of PROX1 to inhibition of LSEC angiogenesis by curcumol. LSECs were treated by similar method as above, and the activation of PROX1 was determined via different technical methods. As shown in Figure 6A

| Curcumol inhibits LSEC angiogenesis and attenuates liver fibrosis in mice
We finally examined the effect of curcumol in vivo using male mice with CCl 4 -induced liver fibrosis. Besides, we raised the Smo HSCs. Phenotypic stability of LSECs is vital to maintain quiescent HSCs and protect hepatocytes. 16,28 Hepatic sinusoidal angiogenesis owing to dysfunctional LSECs accompanied by an abnormal angioarchitecture is a symbol related to liver fibrogenesis, which indicates a potential target for therapeutic interventions. 3 For instance, sorafenib, 29 a tyrosine-kinase inhibitor, has been proved to block sinusoid remodelling and facilitate liver fibrosis resolution. Recent studies also explained that different active components in natural medicinal plants could alleviate liver fibrosis by inhibiting pathological angiogenesis. [30][31][32][33] We previously reported that curcumol mitigated liver pathological changes via promoting activated HSC necroptosis. 15 However, the effects of curcumol on LSEC angiogenesis in liver fibrosis had not been investigated, and the underlying mechanism remained to be uncovered. In this article, we confirmed firstly that curcumol ameliorated angiogenesis of LSECs in a murine liver fibrosis model.
As we know, a large amount of nutrients and oxygen are consumed in the process of angiogenesis. 34 To adapt to hypoxia, cells reduce oxygen consumption and maintain homeostasis that are mediated by HIF-1α. 17 We investigated a potential mechanism of curcu- in HSCs. 31 Here, we found that curcumol can inhibit the markers of Hh pathway dose-dependently in LSECs, and agonist or antagonist of this signalling pathway was used to confirm the effect of the canonical Hh pathway on HIF-1α protein expression.
Curcumol functioned as an inhibitor of HIF-1α through declining the nuclear level of Gli1. Besides, non-canonical Hh pathway appeared to contribute to cholangiocarcinoma progression, which implied us that Hh has a great influence on the microenvironment of liver. 35 HIF-1α is kept under tight regulation, and in normal physiological condition, it is the most short-lived proteins known. The ubiquitin-proteasome system practically kept steady-state levels F I G U R E 7 Curcumol attenuates liver fibrosis in mice. A, After a week of adaptive feeding, mice were randomly assigned to six groups and injected i.p. with CCl 4 except for control group for 8 wk to induce liver fibrosis. The transfection adenoviruses were given twice in the 3rd and 5th weeks, respectively. Besides, 30 mg/kg curcumol was injected i.p. in other day during the last 4 wk. B, Liver sections were stained with Sirius red reagents, Masson reagents and H&E for collagen and histological examinations (n = 3). Scale bar = 100 μm. C, Determination of serum levels of liver injury markers ALP, AST and ALT (n = 6). D, Determination of serum levels of hydroxyproline content (n = 6). E, Determination of serum levels of fibrotic markers HA, LN, PC-Ⅲ and Ⅳ-C (n = 6). * P < .05, ** P < .01 and *** P < .005 versus control group; # P < .05, ## P < .01 and ### P < .005 versus model group low. Until recently, studies of post-translational modifications of HIF-1α were expanded to hydroxylation, 36 phosphorylation, 37 methylation 38 and acetylation. 39 In hepatocellular carcinoma, the homeobox protein PROX1 is overexpressed in hepatocytes and is required for cell migration. The results of assays suggested that the amino-terminal two-thirds (amino acids 1-570) of PROX1 was responsible for the interaction with HIF-1α, upregulating HIF-1α expression to induce EMT response in HCC cells. 21 Another research had showed that PROX1 could prevent p65 ubiquitination by recruiting USP7 to inhibit HCC angiogenesis. 19 Therefore, we F I G U R E 8 Curcumol inhibits liver sinusoidal endothelial cell angiogenesis through regulating Hh signalling pathway in mice. A, Analysis of sinusoidal fenestrae (arrowhead) in liver tissues (n = 3). Scale bar = 1 μm. B, Immunohistochemical analyses of endothelial markers CD31, CD34 and vWF in liver tissues (n = 3). Scale bar = 100 μm. C, Western blot analysis of endothelial markers and angiogenic factors (n = 3). D, Immunofluorescence analysis of endothelial markers CD31 accompanied by HIF-1α, VEGF-A, Smo and PROX1 in liver tissues (n = 3). Scale bar = 20 μm. * P < .05, ** P < .01 and *** P < .005 versus control group; # P < .05, ## P < .01 and ### P < .005 versus model group focused on the ubiquitination of HIF-1α that a reversible process is dependent on deubiquitylating enzymes (DUBs It is noteworthy that many controversial issues still remained unsolved among this research. We confirmed the anti-angiogenic effect of curcumol through Hh signalling pathway; however, the direct target of drug was still unknown. Glis family transcription factors could affect various genes in nuclear, but the specific regulation mechanism had not been revealed. The effect of curcumol on the LSEC capillarization in vitro and in vivo is inconsistent. The maintenance of fenestrae structure depends on various factors.
The loss of fenestrae in the cultured LSECs was rescued by silencing DLL4 in vitro, 41 and BMP9 is a key paracrine secreted from HSCs to control LSEC fenestration. 42 Primary LSECs almost pursued to lose their fenestrae out of normal environment regardless of the slight effect of curcumol in vitro. The process of hepatic angiogenesis is divided into many stages, and the effects of drug therapy on angiogenesis at different stages are also different. 7 LSEC renewal differs in physiological and in pathological conditions. 43 It is impressively difficult to distinguish the characteristics of angiogenesis between liver regeneration and liver cirrhosis.
According to our experiment in vivo, curcumol was given to mice in the early stage of liver fibrosis and showed great resistance to hepatic fibrosis and angiogenesis. However, we still think that curcumol could have multiple targets to achieve its pharmaceutical characteristics. Our previous research had investigated whether curcumol is able to alleviate liver fibrosis via triggering HSC necroptosis. 15 Activated HSCs can secrete angiogenic factors to regulate LSECs, and it also participates in liver angiogenesis directly. 12 Inhibition of activated HSCs by curcumol would attenuate LSEC-mediated angiogenesis. LSECs account for the majority of liver non-parenchymal cells and are the most important part of neovascularization, so we draw a conclusion that curcumol could alleviate liver fibrosis via inhibiting LSEC-mediated angiogenesis.
A future perspective could be represented by the investigation of curcumol for the modulation of cellular processes involved in LSEC-mediated angiogenesis in the liver.
In summary, we demonstrated that canonical hedgehog signalling regulated LSEC angiogenesis through transactivation of PROX1 and HIF-1α. Natural medicine curcumol could alleviate sinusoid pathological angiogenesis by impairing hedgehog signalling ( Figure 9).
These findings provided a novel molecular mechanism for curcumol to be a new anti-fibrotic agent for liver fibrosis.

F I G U R E 9
Diagram illustrates the mechanisms of curcumol regulating liver sinusoidal endothelial cell-mediated angiogenesis in liver fibrosis

ACK N OWLED G EM ENTS
Our research was supported by the National Natural Science

CO N FLI C T O F I NTE R E S T S
All authors confirm that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this article.