Ascorbic acid enhances low-density lipoprotein receptor expression by suppressing proprotein convertase subtilisin/kexin 9 expression

treatment overnight, activity of firefly and Renilla luciferases in cellular lysate using the Dual-Luciferase Reporter Assay E , F , HepG2 cells transfected with control siRNA (si-NC, 25 nM) or siRNA against SREBP2 (si-SREBP2, 25 nM) for 48 h ， followed by treatment with AA (20  ) or vehicle in fresh medium for 12 h. Protein levels of LDLR, SREBP2 precursor (p) and mature form (m) were determined by Western blot ( E) . The normalized intensity of mature SREBP2 [(m) SREBP2)] vs .  -actin was calculated. Expression of LDLR and SREBP2 mRNA was examined by qRT-PCR (F) . I , HepG2 cells were pre-treated with actinomycin D (Act D, 10 nM) for 2 h and then treated with AA (20  M) or vehicle for the indicated times. Expression of LDLR mRNA was


2
The increased low-density lipoprotein cholesterol (LDL-cholesterol or LDL) levels in plasma is a pivotal risk factor for the initiation of cardiovascular disease (CVD), the leading cause of death in all countries (1). LDL receptor (LDLR) plays a critical role in anti-atherosclerosis by regulating LDL catabolism. The majority of excess LDL can be cleared by hepatic LDLR (2). LDLR binds LDL to mediate LDL endocytosis and degradation in lysosomes, and then recycles itself to the cell surface (3). Therefore, mutations of LDLR, either heterozygous or homozygous, result in familial hypercholesterolemia (2). It is effective to lower plasma LDL levels by enhancing hepatic LDLR expression. At the transcriptional level, LDLR expression is predominantly activated by sterol regulatory element-binding protein 2 (SREBP2), which binds to the sterol regulatory element (SRE) in the promoter of LDLR to trigger its transcription (4). In addition to SREBP2, proprotein convertase subtilisin/kexin 9 (PCSK9) regulates LDLR expression at the post-translational level (5). PCSK9 can bind LDLR to lead LDLR degradation intracellularly, resulting in suppressed LDL clearance from the circulation (6). Accordingly, reduction of PCSK9 decreases plasma LDL levels efficiently. PCSK9 is mainly synthesized and secreted from the liver, and the circulating PCSK9 has become an important target for treatment of cardiovascular diseases.
Ascorbic acid (also known as vitamin C) is a water-soluble antioxidant that is naturally present in some kinds of food. Due to its capability to scavenge free radicals, the beneficial effects of ascorbic acid on lipid profiles have been demonstrated (11). It differs from mice, humans and guinea pigs are prone to develop ascorbic acid deficiency as the consequence of mutation in L-gulono--lactone oxidase (Gulo), the enzyme is required for the last step of ascorbic acid biosynthesis (12). Therefore, maintaining adequate levels of ascorbic acid supplement is essential to achieve normal body functions and optimal health (13). Most studies have found that chronic ascorbic acid deficiency in guinea pigs leads to increased cholesterol synthesis and attenuated cholesterol conversion into bile acids (14,15). In addition, in Gulo and ApoE double deficient (Gulo -/-Apoe -/-) mice, the total cholesterol (CHO) levels is decreased by increased ascorbic acid intake (16). These findings suggest that ascorbic acid supplement can influence LDL metabolism. However, the precise actions and the involved mechanisms by ascorbic acid remain unknown. In this study, we investigated if ascorbic acid can affect LDL levels and the action is completed by regulating PCSK9 and LDLR expression.

Ascorbic acid inhibits PCSK9 expression in human hepatic cell lines
PCSK9 is mainly present in hepatocytes as a ~72 kDa precursor and secreted from cells after auto-cleaved as a mature form with a molecular weight of ~62 kDa (9,17). We firstly assessed the effects of ascorbic acid on PCSK9 expression in human hepatic cell lines, HepG2 and Huh7 cells. As shown in Fig. 1A, ascorbic acid down-regulated both PCSK9 precursor and mature form in HepG2 cell in concentration-and time-dependent manners. To determine if inhibition of PCSK9 protein expression by ascorbic acid is associated with reduced PCSK9 mRNA expression, total cellular RNA was extracted after treatment and used to determine PCSK9 mRNA levels by qRT-PCR. Consistently, ascorbic acid reduced PCSK9 mRNA expression in the similar patterns to protein (Fig. 1B), indicating reduction of PCSK9 expression by ascorbic acid can be at a transcriptional level. Similar to HepG2 cells, ascorbic acid also reduced PCSK9 protein and mRNA expression in Huh7 cells, another human hepatic cell line (Fig. 1C, D).

Ascorbic acid enhances LDLR expression and LDL uptake
Inhibition of PCSK9 expression can result in increased LDLR protein levels (18). In order to detect whether ascorbic acidreduced PCSK9 can increase LDLR protein expression, we evaluated LDLR protein levels by Western blot, and found LDLR expression in HepG2 cells was increased by ascorbic acid in a concentration-dependent manner ( Fig. 2A). In addition, ascorbic acid increased LDLR mRNA expression (Fig. 2B), indicating it may enhance LDLR levels at both transcriptional and translational levels. Meanwhile, we found a significant increase of Dil-LDL uptake by HepG2 cell in response to ascorbic acid treatment (Fig. 2C), suggesting the enhanced LDLR activity by ascorbic acid. In parallel, ascorbic acid induced LDLR protein and mRNA expression, as well as uptake of Dil-LDL by Huh7 cells (Fig. 2D-F).
Furthermore, we treated primary hepatocytes isolated from C57BL/6J mice with ascorbic acid. Similarly, ascorbic acid increased LDLR while decreased PCSK9 expression in murine primary hepatocytes (Fig. 2G), which indicates that regulation of PCSK9/LDLR expression by ascorbic acid is not species-dependent.

Ascorbic acid attenuates PCSK9 expression by activating FoxO3a
To identify the molecular mechanism by which ascorbic acid inhibits PCSK9 expression in hepatocytes, we detected effects of ascorbic acid on transcription factors responsible for PCSK9 expression, such as PPAR and FoxO3a. Although it has been reported that ascorbic acid deficiency up-regulates PPAR in the liver of senescence marker protein 30 deficient mice (19), we observed that both PPAR protein and mRNA levels were not altered by ascorbic acid (Fig. 3A, B). In contrast, ascorbic acid clearly increased FoxO3a protein and mRNA expression (Fig. 3A, B). We further examined the transcriptional activity of PPAR and FoxO3a. Consistently, mRNA levels of CD36 and fatty acid-binding protein 4 (FABP4), the downstream genes of PPAR (20), were not affected by ascorbic acid either (Fig. 3C). In contrast, mRNA levels of superoxide dismutase 1 (SOD1) and catalase (CAT), the downstream genes of FoxO3a (21), were obviously enhanced by ascorbic acid (Fig. 3D). Furthermore, increased FoxO3a nuclear translocation was determined in cells treated with ascorbic acid (Fig. 3E, F).
To investigate if ascorbic acid can decrease FoxO3a phosphorylation to enhance its nuclear translocation since the phosphorylated FoxO3a retains in the cytoplasm (22), we determined total and phosphorylated FoxO3a levels, and found ascorbic acid increased total FoxO3a but decreased phosphorylated FoxO3a levels ( Fig. 3G).
To further determine the role of FoxO3a in ascorbic acid-inhibited PCSK9 expression, we used FoxO3a siRNA to inhibit FoxO3a protein and mRNA expression (Fig. 4A, B). As expected, inhibition of FoxO3a expression not only increased PCSK9 expression, but also abolished the effect of ascorbic acid on PCSK9 expression (Fig. 4A,  B). Moreover, FoxO3a siRNA attenuated ascorbic acid-induced LDLR expression (Fig.  4A), which should be due to the failure of PCSK9 inhibition in the presence of FoxO3a siRNA.
It has been reported that PCSK9 promoter contains a FoxO3a binding motif, the insulin response element (IRE), with a conserved sequence of TGTTTA (from -381 to -376) (9). To clarify the role of FoxO3a in regulation of PCSK9 transcription by ascorbic acid, we constructed a normal human PCSK9 promoter and the promoter with IRE mutation (-381 CGTCTT -376, the underlined letters are mutated nucleotides). We found that ascorbic acid reduced PCSK9 promoter activity in a concentrationdependent manner (Fig. 4C). However, mutation in IRE attenuated the inhibitory effect of ascorbic acid on PCSK9 transcription. In addition, FoxO3a siRNA abolished ascorbic acid-inhibited PCSK9 promoter activity (Fig. 4D). These results collectively suggest that FoxO3a plays a central role in suppressing PCSK9 transcription by ascorbic acid.

SREBP2 signaling also correlates to activation of LDLR expression by ascorbic acid
Although inhibition of PCSK9 resulted in increased LDLR protein levels, we also observed induction of LDLR mRNA by ascorbic acid at high concentrations (Fig. 2B,  F). Thus, in addition to the regulation by PCSK9, ascorbic acid may activate LDLR expression by other pathways. SREBP2 is the transcription factor activating LDLR expression at the transcriptional level. We initially determined the effect of ascorbic acid on SREBP2 expression and maturation, and observed that ascorbic acid enhanced SREBP2 expression and maturation (Fig. 4E).
To clarify the role of SREBP2 in regulating LDLR transcription by ascorbic acid, we used SREBP2 siRNA to inhibit SREBP2 expression. As expected, inhibition of SREBP2 expression abolished the enhancement of LDLR protein and mRNA expression by ascorbic acid (Fig. 4E, F). Correspondingly, we found ascorbic acid enhanced LDLR promoter activity at a high concentration (Fig. 4G, left half). However, the activation was abolished when SRE in LDLR promoter was mutated (Fig. 4G, right half). Moreover, SREBP2 siRNA attenuated ascorbic acid-enhanced LDLR promoter activity (Fig. 4H).
In addition, we determined the effect of ascorbic acid on stability of LDLR mRNA. Treatment of HepG2 cells with actinomycin D inhibited RNA transcription, therefore, reduced LDLR mRNA levels were determined with time of actinomycin D treatment due to the mRNA degradation. In contrast, compared to actinomycin D alone, co-treatment with ascorbic acid increased LDLR mRNA levels ( Fig. 4I), suggesting ascorbic acid reduces LDLR mRNA degradation or increases its stability. Therefore, the data above suggest that in addition to activation by reduced PCSK9, ascorbic acid also activates LDLR expression by enhancing its stability and SREBP2 transcriptional activity.

Ascorbic acid stimulates LDLR expression and restrains PCSK9 secretion from mouse liver
To determine the physiological relevance of ascorbic acid on PCSK9/LDLR expression, C57BL/6J mice fed either normal chow or a high-fat diet (HFD, containing 21% fat and 0.5% cholesterol) were i.p. injected ascorbic acid solution or vehicle (saline) daily for one week. Although wild type mice do not depend on exogenous ascorbic acid supplement, as shown in Fig. 5A and F, injection with ascorbic acid for one week still moderately increased ascorbic acid levels in the liver. Compared with control mice, ascorbic acid treatment did not cause differences to the animals, such as the ratio of liver weight to bodyweight and lipid profiles (Fig. 5B, G).
Similar to results in in vitro studies, PCSK9 levels in the liver of mice fed normal chow were decreased by ascorbic acid (Fig.  5C, E). PCSK9 is mainly synthesized in the liver and is rapidly secreted into plasma after its maturation through a self-engaged autocatalytic cleavage (17). Thus, we evaluated the secretion of PCSK9 using an ELISA kit, and found ascorbic acid also significantly reduced circulating PCSK9 levels compared with control mice (27±3 vs. 41±3 ng/ml, Fig. 5D). Consistently, ascorbic acid inhibited PCSK9 expression and secretion in HFD-fed mice (Fig. 5H, I).
Consequently, ascorbic acid treatment significantly enhanced LDLR expression in the liver (Fig. 5C, E, H, J). In addition to LDRL, associated with changes of FoxO3a and SREBP2 expression by ascorbic acid, expression of the target genes of FoxO3a

Negative correlation between ascorbic acid and PCSK9 or LDL levels in Gulo -/mouse and human serum
Naturally, C57BL/6J mice are able to synthesize ascorbic acid. They are not a hyperlipidemic model even though they are fed HFD. Therefore, technically it is not easy to manipulate ascorbic acid levels in vivo to influence LDL levels in mice naturally. To further clarify the effect of ascorbic acid on cholesterol metabolism in vivo, we investigated the effect of ascorbic acid deficiency on regulation of LDL metabolism in Gulo -/mice, the animals are fully dependent on exogenous ascorbic acid supplement.
After removal of ascorbic acid from the drinking water for 3 weeks, compared with wild type mice, Gulo deficiency reduced ascorbic acid more than 60% in the liver (Fig.  6A). Associated with reduced ascorbic acid, serum PCSK9 levels and hepatic PCSK9 expression were increased ( Fig. 6B-D), which resulted in reduced hepatic LDLR expression in Gulo -/mice (Fig. 6C, D). Correspondingly, serum LDL levels were significantly increased, while HDL levels were decreased, thereby resulting in decreased ratio of HDL to LDL (Fig. 6E).
In contrast to ascorbic acid deficiency, injection of HFD-fed Gulo -/mice ascorbic acid solution for two weeks substantially increased ascorbic acid levels in the liver ( Fig 6F). Associated with increased hepatic ascorbic acid levels, serum PCSK9 levels and hepatic PCSK9 expression were reduced ( Fig. 6G-I). Meanwhile, hepatic LDLR expression was activated (Fig 6H, I).
Consequently, serum LDL levels were reduced, while HDL levels and the ratio of HDL to LDL were increased by exogenous ascorbic acid supplement in Gulo -/mice ( Fig  6J). In addition, we observed reduced SREBP2 and FoxO3a activity (decreased HMGCR, HMGCS, SOD1 and CAT, Fig. 6D) by ascorbic acid deficiency, and activated both (activated HMGCR, HMGCS, SOD1 and CAT, Fig. 6I) by ascorbic acid supplement in Gulo -/mice fed HFD. Taken together, these results suggest the importance of hepatic ascorbic acid in maintaining cholesterol homeostasis.
The correlation between PCSK9 and circulating cholesterol levels has been well established (11,23,24). Our results above suggest that ascorbic acid deficiency increased PCSK9 expression, which can make substantial contribution to increased LDL levels in Gulo -/mice. Similar to guinea pigs or Gulo -/mice, our human beings are not able to produce ascorbic acid either, and our life solely depends on the exogenous ascorbic acid supplement. Therefore, we speculated that ascorbic acid levels in humans could be one of determinants for circulating PCSK9, LDL and total cholesterol (CHO) levels. To determine it, we collected serum samples form 24 volunteers and determined levels of CHO, LDL, PCSK9 and ascorbic acid. Consistent with the previous study, serum CHO and LDL levels were positively correlated to PCSK9 levels (data not shown). Interestingly, we observed that PCSK9 levels were negatively correlated to ascorbic acid levels in human serum (Fig. 6K). Furthermore, serum ascorbic acid levels were negatively correlated to serum CHO and LDL levels (Fig. 6L, M), suggesting the role of ascorbic acid in regulating PCSK9 expression and cholesterol metabolism in humans.

Discussion
As a water-soluble antioxidant, ascorbic acid has multiple biological functions. It is essential for collagen synthesis and enzyme activity maintenance (13,25). Besides scurvy, many other pathogenesis of ascorbic acid deficiency have been observed. Lack of ascorbic acid production in mice exacerbates diabetic kidney injury, suggesting that compensation for ascorbic acid loss is an effective treatment for diabetic kidney injury (26). The ascorbic acid deficiency also enhances tumor necrosis factor -induced insulin resistance (27).
In vascular biology, ascorbic acid deficiency can reduce collagen content of atherosclerotic plaques, resulting in formation of the vulnerable plaques (16). Physiological concentrations of ascorbic acid inhibits LDL oxidation in vitro (28). In addition, low levels of ascorbic acid cause dyslipidemia in guinea pigs, especially when they are fed a high-cholesterol or fat diet (14). Many studies suggest that advisable intake of ascorbic acid has a positive effect on lipid profiles. Ascorbic acid deficiency is a prevalent complication in hemodialysis patients due to its loss during dialysis sessions (29). Thus, supplementing ascorbic acid to hemodialysis patients for three months can significantly increase serum ascorbic acid levels, associated with increased HDL and reduced CHO, LDL and triglyceride levels (29). Supplementing more than 500 mg/day of ascorbic acid for a minimum 28 days can result in a significant decrease in serum LDL or triglyceride levels in patients with hypercholesterolemia (30). Ascorbic acid supplement also decreases triglyceride and LDL levels in type 2 diabetic patient (31). In addition, C57BL/6J mice received a HFD supplement with ascorbic acid (1% w/w) for 15 weeks had reduced serum LDL levels by 50%, suggesting that ascorbic acid may suppress HFD-induced visceral obesity and nonalcoholic fatty liver disease (NAFLD) (32). In HFD-fed Gulo -/-Apoe -/mice, supplement of ascorbic acid substantially increased serum ascorbic acid levels, which is associated with obviously reduced plasma CHO levels (16). In our study, we found that ascorbic acid withdrawal resulted in substantial reduction of ascorbic acid in Gulo -/mice (62.2%, Fig.  6A). Associated with decreased ascorbic acid levels, increased LDL and decreased HDL levels were determined in Gulo -/mice compared with C57BL/6J mice (Fig. 6E). However, exogenous ascorbic acid supplement failed to reduce LDL levels in C57BL/6J mice (Fig. 5B, G), although LDLR expression was activated in the liver (Fig. 5C, H). This may be due to that C57BL/6J mice are naturally able to produce ascorbic acid, which results in that the animals have a constant high ascorbic acid level, while the short-term exogenous supplement of ascorbic acid to C57BL/6J mice is hard to manipulate ascorbic acid levels substantially (27.8%, Fig. 5A). In addition, C57BL/6J mice is not a hypercholesterolemia model since the increased CHO levels by HFD feeding is mainly contributed by increased HDL but not LDL levels (Fig. 5B vs. 5G). In contrast, supplement of ascorbic acid to its deficient mice or patients can clearly ameliorate lipid profiles (16, [29][30][31][32]. The amelioration of lipid profiles by ascorbic acid supplement should be attributed to inhibition of PCSK9 expression and activation of LDLR expression in the liver (Fig 6F-J).
In spite of the clear beneficial effects of ascorbic acid on lipid metabolism, the underlying mechanisms have not been fully elucidated. PCSK9 plays a critical role in cholesterol homeostasis. Lowering PCSK9 levels results in increase of intracellular LDLR protein abundance and contributes to metabolism/reduction of LDL (33). Amelioration of lipid metabolism by ascorbic acid implies that the action may be completed, at least in part, by inhibiting PCSK9 expression. Indeed, in this study we found that PCSK9 levels were negatively correlated to ascorbic acid levels, which was associated with the positive correlation to CHO and LDL levels in both human and Gulo -/mouse serum (Fig. 6). At the molecular level, we determined ascorbic acid decreased PCSK9 expression and increased LDLR expression in HepG2 or Huh7 cells in vitro (Fig. 1, 2) and in the liver (Fig. 5, 6). SRE is present in PCSK9 promoter and activation of SREBP2 increases PCSK9 gene expression (34). Although ascorbic acid activated SREBP2, it significantly attenuated PCSK9 protein and mRNA levels (Fig. 1), indicating that inhibition of PCSK9 expression by ascorbic acid is independent of SREBP2 pathway. Recently, FoxO3a has been identified as an important regulator of lipid metabolism and liver functions (35,36). FoxO3a suppresses PCSK9 transcription by blocking the binding of HNF-1 with the promoter (9). In this study, we determined that ascorbic acid activated FoxO3a expression/transcriptional activity while had little effect on either PPAR expression and its transcriptional activity (Fig. 3), suggesting FoxO3a is the major player in ascorbic acid-inhibited PCSK9 (Fig. 4A-D).
The post-translational regulation of LDLR expression is primarily controlled by PCSK9. Interestingly, in addition to protein levels, we also found increased LDLR mRNA level by ascorbic acid (Fig. 2B, F), suggesting additional mechanism involved in ascorbic acid-activated LDLR expression. Indeed, we observed that ascorbic acid enhanced SREBP2 transcriptional activity. It also increased LDLR mRNA stability ( Fig.  4E-I). Therefore, ascorbic acid activates LDLR expression by regulating multiple pathways.
In conclusion, we found that ascorbic acid reduced PCSK9 expression/secretion and enhanced LDLR expression as well as the underlying mechanisms (Fig. 7). Importantly, we validated that human serum PCSK9 and LDL levels were negatively correlated to ascorbic acid levels. Our study provides the molecular mechanisms by which supplement of ascorbic acid can be helpful in improving lipid profiles in species who are not able to produce ascorbic acid endogenously.

Experimental procedures Data availability statement
All the data related to this work have been contained within the manuscript.

Materials
Ascorbic acid was obtained from

Cell culture
HepG2 or Huh7 cells, the human hepatic cell lines, were cultured in high glucose DMEM medium containing 10% fetal bovine serum (FBS), 50 g/ml penicillin and streptomycin, and 2 mM glutamine. Human embryonic kidney 293T cell line was cultured in RPMI 1640 medium containing 10% FBS, 50 g/ml penicillin/ streptomycin, and 2 mM glutamine. Mouse primary hepatocytes were isolated from C57BL/6J mice by a collagenase perfusion method as described (37). All the cells were cultured in a humidified atmosphere incubator at 37°C and 5% CO2.

Western blot and immunofluorescent staining
The cells were dispensed in a 6-well culture plate at density of 5×10 5 cells per well. After reaching ~80% confluence, cells received treatment in medium containing 2% FBS. Expression of LDLR, PCSK9, PPAR SREBP2, phosphorylated FoxO3a and FoxO3a protein in total cellular extract was determined by Western blot as described (38). To prepare nuclear and cytoplasmic proteins, cells were harvested using the Nuclear and Cytoplasmic Protein Extraction Kit (Keygen Biotech, Nanjing, China) according to the manufacturer's protocol. FoxO3a expression in cells was also determined by immunofluorescent staining as described (39). The samples were observed and photographed with a fluorescence microscope after staining. After capture, the mean fluorescence intensity (MFI) of all immunofluorescent images was calculated as described (40) using morphometry software (IP Laboratory, Scanalytics, Rockville, MD) by a person who was blinded to the treatment. For Western blot, bands from five repeated experiments were scanned and the density of band was quantified with statistical analysis.

Real-time quantitative reverse transcription PCR (qRT-PCR)
Total RNA was isolated from cells or a piece of liver using the Total RNApure reagent (Zomanbio, Beijing, China) according to the manufacturer's instructions. Total RNA (1 g) was converted into cDNA using the reverse transcription kit (Vazyme, Nanjing, China). The real time PCR was then performed using a SYBR green PCR master mix with primers listed in Table 1. mRNA expression of genes was normalized by glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA in the corresponding samples.

LDL uptake assay
To analyze LDLR activity in HepG2 or Huh7 cells, cells (1×10 5 cells/well) were cultured in glass bottom cell culture dishes (Corning, NY, USA) overnight. Cells were then treated with vehicle or ascorbic acid (20 M) for 12 h in 2% serum medium. Cells were switched into serum-free medium containing Dil-LDL (20 g/ml) and incubated for 4 h at 37°C. After washed twice with PBS and covered by Vectashield mounting medium for 30 min, cells were stained with DAPI solution for nucleus. Finally, cells were observed and photographed with a fluorescence microscope.

Luciferase reporter assay
Human PCSK9 and LDLR promoters were constructed as described (41). The human PCSK9 promoter with mutation of FoxO3a binding site or IRE (named as phPCSK9-IREmut) and the human LDLR promoter with SRE mutation (named as phLDLR-SREmut) were constructed using the TransStartFastPfu DNA Polymerase Kit (TransGen Biotech, Beijing, China) with primers including the corresponding IRE or SRE mutation (Table 2), respectively. The PCR product was digested and then ligated into pGL4 luciferase reporter vector. The promoter activity was determined using the Dual-Luciferase ® Reporter Assay System (Promega, Madison, WI, USA) (41).

In vivo study
The animal study was approved by the Animal Ethics Committee of Hefei University of Technology and performed in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publications No. 8023, revised 1978) and the ARRIVE guidelines. C57BL/6J mice (males, ~8-week-old) were purchased from the Animal Center of Nanjing University. Gulo -/mice (C57BL/6J background) were obtained from the Zhejiang University (Hangzhou, China) (26). During the breeding and maintenance, Gulo -/mice were supplemented with tap water containing 1 g/l ascorbic acid. The water was changed twice a week.
To determine the effect of ascorbic acid supplement on LDLR/PCSK9 expression and lipid profiles in wild type mice, C57BL/6J male mice (~8-week old) were fed either normal chow or HFD containing 21% fat and 0.5% cholesterol, and randomly divided into two groups (5 mice/group). Mice were then i.p. injected ascorbic acid (100 mg/day/kg bodyweight, AA group) or the same volume of saline (Control group) for 1 week. The dose of ascorbic acid was chosen based on previous studies (42,43). To determine the influence of ascorbic acid deficiency on PCSK9/LDLR expression and lipid profiles in vivo, male C57BL/6J (WT) or Gulo -/mice (~8-week old, 5 mice/group) were maintained for 3 weeks without ascorbic acid supplement in the drinking water. To determine the effects of ascorbic acid supplement on PCSK9/LDLR expression and lipid profiles in Gulo -/mice, male Gulo -/mice (~8-week old) were fed HFD and randomly divided into two groups (5 mice/group). Mice were then i.p. injected ascorbic acid (100 mg/day/kg bodyweight, AA group) or the same volume of saline (Control group) for 2 weeks.
All the mice were free to access to food and drinking water during the treatment. At the end of experiment, all the mice were anesthetized and euthanized in a CO2 chamber, followed by collection of liver and blood samples individually. The blood was kept for ~2 h at room temperature. Serum samples were isolated after centrifugation for 20 min at 2,000 g. Serum triglyceride, CHO, HDL and LDL levels were analyzed using an automatic biochemical analyzer (Model #3100, Hitachi High-Technologies Corporation, Tokyo, Japan). Serum PCSK9 levels were determined by the ELISA kit. Liver samples were used to determine protein expression and ascorbic acid levels by Western blot and ascorbic acid assay kit, respectively.

Human serum samples
With a protocol approved by the Clinical Ethics Committee of the First Affiliated Hospital of Anhui Medical University (Hefei, Anhui, China) and adhered strictly to the Declaration of Helsinki Principle 2008, we recruited 24 volunteers at an age range of 28-58 years old. All volunteers provided written informed consent and study protocol. Serum samples were prepared for determination of CHO, LDL, PCSK9 and ascorbic acid levels by automatic biochemical analyzer, ELISA and ascorbic acid assay kits, respectively.

Statistical analysis
All experiments were repeated at least 5 times, and the representative results are presented. All values are expressed as means ± SEM. All the data were initially subjected to a normal distribution analysis with SPSS software (1-sample K-S of non-parametric test). The data in normal distribution was then analyzed by ANOVA with post-hoc test or t-test. The significant difference was considered if p < 0.05 (n≥5).      F). C, E, HepG2 or Huh7 cells were treated with 20 M AA for 12 h. Cells were then switched to serum-free medium containing Dil-LDL (20 g/ml) and incubated for 4 h at 37°C, followed by photograph with a fluorescence microscope. The mean fluorescent intensity (MFI) in images was quantitatively analyzed (original magnification, ×400). G, primary hepatocytes isolated from C57BL/6J mice were treated with AA at the indicated concentrations for 12 h or with 20 M AA for the indicated times. Expression of PCSK9 or LDLR protein was determined by Western blot. *, p < 0.05; **, p < 0.01 vs. control (n = 5).

Figure 3. Induction of FoxO3a expression and nuclear translocation by ascorbic acid.
A, B, HepG2 cells were treated with ascorbic acid (AA) at the indicated concentrations for 12 h. Expression of PPAR and FoxO3a protein (A) and mRNA (B) were determined by Western blot and qRT-PCR, respectively. C, D, Expression of CD36, FABP4, SOD1 and CAT mRNA was examined by qRT-PCR. E, HepG2 cells were treated with 20  AA for 12 h. FoxO3a expression was determined by immunofluorescent staining followed by quantifying MFI in images (original magnification, ×400). F, HepG2 cells were treated with AA at the indicated concentrations for 12 h, followed by extraction of nuclear and cytosolic proteins and determination of FoxO3a expression by Western blot. G, HepG2 cells were treated with AA at the indicated concentrations for 12 h, and expression of FoxO3a and phosphorylated FoxO3a (p-FoxO3a) was determined by Western blot. *, p < 0.05; **, p < 0.01 vs. control (n = 5). Expression of LDLR and SREBP2 mRNA was examined by qRT-PCR (F). I, HepG2 cells were pre-treated with actinomycin D (Act D, 10 nM) for 2 h and then treated with AA (20 M) or vehicle for the indicated times. Expression of LDLR mRNA was determined by qRT-PCR. *, p <0.05; **, p < 0.01 vs. corresponding controls or as indicated (n = 5); #, p < 0.05; ##, p < 0.01 vs. si-NC plus AA (n = 5).

Figure 5. Ascorbic acid activates LDLR expression in the liver and reduces PCSK9 levels in serum of C57BL/6J mice
A-J, ~8-week old male C57BL/6J mice in 4 groups (5 mice/group) were fed normal chow or high-fat diet (HFD: 0.5% cholesterol plus 21% fat), and i.p. injected ascorbic acid (100 mg/day/kg bodyweight, AA group) or saline (Control group) daily for 1 week. At the end of treatment, mouse liver and blood samples were collected and used for following assays. A, F, determination of ascorbic acid content in the liver; B, G, determination of HDL-cholesterol (HDL) and LDL-cholesterol (LDL) levels in serum. C, H, expression of LDLR and PCSK9 protein in the liver was determined by Western blot. D, I, serum PCSK9 levels were determined by ELISA. E, J, expression of PCSK9, LDLR, SREBP2, HMGCR, HMGCS, FoxO3a, SOD1 and CAT mRNA was determined by qRT-PCR. *, p < 0.05; **, p < 0.01 vs. corresponding control (n = 5).

Figure 6. Ascorbic acid ameliorates lipid profiles in Gulo -/mice associated with regulation of PCSK9 and LDLR expressions
A-E, the ~8-week old male C57BL/6J (WT) or Gulo -/-(C57BL/6J background) mice (5 mice/group) were fed normal chow without ascorbic acid supplementation for 3 weeks. F-J, male Gulo -/mice (~8-week old) fed HFD were randomly divided into two groups (5 mice/group), and i.p. injected ascorbic acid (100 mg/day/kg bodyweight, AA group) or the same volume of saline (Control group) for 2 weeks. After treatment, mouse liver and blood samples were collected and used for the following assays. A, F, determination of ascorbic acid content in the liver. B, G, determination of serum PCSK9 levels. C, H, expression of LDLR and PCSK9 protein was measured by Western blot. D, I, expression of PCSK9, LDLR, SREBP2, HMGCR, HMGCS, FoxO3a, SOD1 and CAT mRNA was determined by qRT-PCR. E, J, determination of serum LDL, HDL levels and calculation of the ratio of HDL to LDL. *, p < 0.05; **, p < 0.01 vs. corresponding control (n = 5); K-M, PCSK9, CHO, LDL and ascorbic acid levels in human serum samples (n = 24) were determined. K, serum PCSK9 levels are plotted against ascorbic acid levels in serum; Serum ascorbic acid levels are also plotted against CHO (L) and LDL (M) levels, respectively.

Figure 7. Identifying the role of ascorbic acid in regulating PCSK9-LDLR pathway
Treatment of hepatocytes or mice with ascorbic acid activates FoxO3a, which results in inhibition of PCSK9 transcription and activation of LDLR protein expression. In addition, ascorbic acid is able to activate SREBP2 pathway, which leads to activation of LDLR transcription. These two pathways work together to enhance LDLR activity and ameliorate cholesterol metabolism, particularly in the species relying on exogenous ascorbic acid supplement.