Human PCSK9 promotes hepatic lipogenesis and atherosclerosis development via apoE- and LDLR-mediated mechanisms

Abstract

Aims

Proprotein convertase subtilisin/kexin type 9 (PCSK9) promotes the degradation of hepatic low-density lipoprotein (LDL) receptors (LDLR), thereby, decreasing hepatocyte LDL-cholesterol (LDL-C) uptake. However, it is unknown whether PCSK9 has effects on atherogenesis that are independent of lipid changes. The present study investigated the effect of human (h) PCSK9 on plasma lipids, hepatic lipogenesis, and atherosclerotic lesion size and composition in transgenic mice expressing hPCSK9 (hPCSK9tg) on wild-type (WT), LDLR^−/−, or apoE^−/− background.

Methods and results

hPCSK9 expression significantly increased plasma cholesterol (+91%), triglycerides (+18%), and apoB (+57%) levels only in WT mice. The increase in plasma lipids was a consequence of both decreased hepatic LDLR and increased hepatic lipid production, mediated transcriptionally and post-transcriptionally by PCSK9 and dependent on both LDLR and apoE. Despite the lack of changes in plasma lipids in mice expressing hPCSK9 and lacking LDLR (the main target for PCSK9) or apoE (a canonical ligand for the LDLR), hPCSK9 expression increased aortic lesion size in the absence of apoE (268 655 ± 97 972 µm² in hPCSK9tg/apoE^−/− vs. 189 423 ± 65 700 µm² in apoE^−/−) but not in the absence of LDLR. Additionally, hPCSK9 accumulated in the atheroma and increased lesion Ly6C^hi monocytes (by 21%) in apoE^−/− mice, but not in LDLR^−/− mice.

Conclusions

PCSK9 increases hepatic lipid and lipoprotein production via apoE- and LDLR-dependent mechanisms. However, hPCSK9 also accumulate in the artery wall and directly affects atherosclerosis lesion size and composition independently of such plasma lipid and lipoprotein changes. These effects of hPCSK9 are dependent on LDLR but are independent of apoE.

1. Introduction

Proprotein convertase subtilisin/kexin type 9 (PCSK9) increases plasma low-density lipoprotein (LDL) cholesterol (LDL-C) by acting as a ligand for hepatic LDL receptors (LDLR), targeting them to the endo-lysosomal pathway and prohibiting their return to the cell surface.^1–3 The role of PCSK9 in LDL-C metabolism has been extensively documented and reviewed in both humans and animal models.^4,5 In humans, gain-of-function (GOF) mutations in PCSK9, such as the substitution of Asp₃₇₄ with Tyr (PCSK9-D374Y), can lead to extremely severe phenotypes with cholesterol levels of >500 mg/dL (∼13 mmol/L).^6,7 On the other hand, loss-of-function (LOF) mutations are associated with very low LDL-C levels and lifetime risk of cardiovascular disease reduced by between 50 and 86% among carriers compared with non-carriers.^8,9 Therefore, PCSK9 is being pursued as a therapeutic target to increase LDLR expression and reduce plasma LDL-C levels. Neutralizing antibodies that inhibit PCSK9 binding to LDLR have been recently approved for use in Europe and the USA (Alirocumab by Regeneron/Sanofi and Evolocumab by Amgen) to treat some forms of hypercholesterolaemia. Post hoc analyses of two recent clinical trials with these drugs have suggested a cardiovascular benefit from the inhibition of PCSK9.^10,11 However, this does not yet amount to full scientific evidence yet and Phase III clinical trials to evaluate cardiovascular benefits are in progress.^12–14 In mice, deletion of PCSK9 causes increased hepatic LDLR expression with a drop in LDL-C levels by 50%.¹⁵ Conversely, transgenic (tg) expression or adeno-associated gene transfer of wild-type (WT) PCSK9 or the PCSK9 GOF mutation D374Y significantly increases plasma cholesterol.^16–19 We have previously developed²⁰ and characterized²¹ a line of transgenic mice expressing human (h) PCSK9 within the normal physiological plasma range in humans (∼30–3000 ng/mL^22–25), resulting in reduced hepatic LDLR levels, increased total cholesterol and LDL-C levels, and overproduction of triglycerides.^26,27

Although PCSK9 is a key determinant in cholesterol homeostasis, its role in triglyceride metabolism is not completely understood. Plasma PCSK9 levels correlate directly with both plasma triglyceride levels and triglyceride-rich-lipoprotein (TRL) markers in normal,^28,29 but not in obese subjects.³⁰ In addition, in human subjects carrying the PCSK9 S127R GOF mutant, levels of apoB100 were increased due to higher levels of VLDL, IDL, and LDL.³¹ It was also shown that hepatic apoB lipoprotein secretion and degradation is negatively regulated by the LDLR and requires the interaction with ApoE or ApoB,³² thus implying that degradation of LDLR might promote apoB lipoprotein production and secretion. Several mouse studies further suggested a role for PCSK9 in hepatic triglyceride production: (i) adenoviral expression of PCSK9 in fasting mice induced hypertriglyceridaemia due to massive accumulation of TRL by the liver³³; and (ii) increased hepatic TRL output was observed in mice with transgenic expression of PCSK9-D374Y.¹⁷ However, the cellular mechanisms leading to increased hepatic triglyceride production are not clear. In contrast, the role of PCSK9 in intestinal apoB-TRL production, more relevant postprandially, has been studied more extensively by us and others, with data showing that the absence of PCSK9 reduces apoB48 secretion and protects mice from postprandial hypertriglyceridaemia.^15,34 We, and others, have further shown that PCSK9 increases the expression and secretion of intestinal apoB through transcriptional and post-transcriptional mechanisms involving both the sterol-regulatory element-binding protein (SREBP) and non-SREBP pathways.^26,34,35

Atherosclerosis is a complex disease in which lipid accumulation and increased inflammation promote lesion growth and affect its composition. Evidence suggests that an inflammatory subset of monocytes (Ly6C^hi) is predominantly recruited into the forming lesion.^36,37 A correlation between PCSK9 plasma levels and atherosclerotic lesion size has been recently found in Indian-Asian patients with chronic coronary artery disease not on lipid-lowering drugs.³⁸ Recent trials support a role for anti-PCSK9 therapies in reducing clinical cardiovascular events.^10,11 Studies conducted in mice show increased atherosclerosis lesion size in murine (m) PCSK9 overexpressing and PCSK9-D374Y expressing mice due, at least in part, to an increase in plasma lipid levels.^17,18 In LDLR^−/− mice, a widely used mouse model of atherosclerosis, the absence of overexpression of mPCSK9 has been shown to have little to no effect on plasma cholesterol levels and atherogenesis.^16,39 In contrast, the effects of PCSK9 in apoE^−/− mice are less clear. While the absence of PCSK9 has no effect on plasma lipids or atherogenesis in apoE^−/− mice,^16,39 the overexpression of mPCSK9 was associated with increased atherosclerotic lesion size without affecting plasma lipid levels.^16,18 We have recently reported that marrow-derived PCSK9 contributes to lesion inflammation by increasing the infiltration of Ly6C^hi monocytes into the artery wall of apoE^−/− mice,⁴⁰ and our data support other findings showing a direct link between PCSK9 and macrophage inflammatory response.^41,42

Here, we investigated for the first time the mechanisms of action of physiologically relevant hPCSK9 levels on hepatic lipid and lipoprotein production and on atherogenesis in vitro and in vivo. Our results support a scenario where the systemic expression of hPCSK9 increases hepatic lipogenesis through transcriptional and post-transcriptional effects that are both LDLR- and apoE-dependent. Our data further show that hPCSK9 significantly affects atherosclerotic lesion size and composition in apoE^−/− mice, but not in LDLR^−/− mice. Thus, hPCSK9 can also affect lesion development through mechanisms that are not related to systemic lipid changes but only involve PCSK9–LDLR interaction in the plaque. These results further support the value for the potential cardiovascular benefits of anti-PCSK9 therapies.

2. Methods

2.1. Reagents

RIPA buffer and protease inhibitor cocktail were purchased from SIGMA (St Louis, MO, USA). Bis-Tris precast gels for electrophoresis were purchased from Life Technologies (Grand Island, NY, USA) and nitrocellulose membranes from GE Healthcare Bio-Sciences (Pittsburgh, PA, USA). Rabbit polyclonal antibodies to fatty acid synthase (FAS) and diglyceride acyltransferase 1 (DGAT1) were purchased from Abcam Inc. (Cambridge, MA, USA). Mouse monoclonal antibody to apoB was purchased from Santa Cruz Biotechnologies Inc. (Santa Cruz, CA, USA). Mouse monoclonal antibody to GAPDH was purchased from Novus Biological (Littleton, CO, USA). The antibody directed toward microsomal triglyceride transfer protein (MTP) was a kind gift from Dr Larry Swift. Rabbit polyclonal antibody directed to β-actin, goat polyclonal antibody to rabbit HRP, luminol, p-coumaric acid, and hydrogen peroxide were purchased from SIGMA (St Louis, MO, USA). Cholesterol and triglyceride kits were purchased from Raichem (San Diego, CA, USA). The apoB ELISA kit for mouse apoB was purchased from antibodies-online (Atlanta, GA, USA). RNeasy kit for RNA extraction was purchased from QIAGEN, and the iScript cDNA synthesis kit was from BIORAD (Hercules, CA, USA). TaqMan Gene Expression Assays and Taqman Universal PCR Mastermix were purchased from Life technologies (Grand Island, NY, USA). Biotin rat anti-mouse Ly6C was purchased from BD Biosciences (San Jose, CA, USA). Streptavidin-AlexaFluor 488 was purchased from Life Technologies (Grand Island, NY, USA).

2.2. Mice

C57BL/6 (WT), apoE^−/−, and LDLR^−/− mice were purchased from the Jackson Laboratories (Bar Harbor, ME, USA) and housed at Vanderbilt University Medical Center and Oregon Health and Science University. Transgenic mice expressing hPCSK9 under a CMV promoter (hPCSK9tg) were generated²⁰ and characterized²¹ in our laboratory and backcrossed to apoE^−/− or LDLR^−/− mice to obtain hPCSK9tg/apoE^−/− or hPCSK9tg/LDLR^−/− mice, respectively. Female mice (10–12 weeks old) were placed on a high-fat diet (HFD) providing 42% of calories from fat and containing 0.2% cholesterol (Catalogue # TD.88137, Harlan laboratories, Indianapolis, IN, USA), and then the extent of atherosclerosis was examined. LDLR^−/− and apoE^−/− background mice were fed a HFD for 8 and 6 weeks, respectively, in order to generate early-stage lesion at similar sizes. WT background mice were fed a HFD for 24 weeks. Mice were anaesthetized in 5% isoflurane using Rodent Anesthesia Machine for Small Animals (Kent Scientific, Winston-Salem, NC, USA), and blood was collected through the retro orbital plexus. Mice were euthanized using 100% CO₂ added to the existing air in the mice chamber at a rate of 2.5 L/min. After visual indication of at least 1 min without breathing, we used cervical dislocation as a secondary means to assure death. All animal experiments were carried out in compliance with National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee of Vanderbilt University and Oregon Health and Science University.

2.3. Immunoblotting

Liver tissues were lysed in RIPA buffer containing protease inhibitor, and protein concentration was measured using the Lowry method.⁴³ Samples were loaded onto 4–12% Bis-Tris precast gels for electrophoresis. The proteins were then transferred to nitrocellulose membranes. Membranes were incubated with primary antibodies against FAS, DGAT1, apoB, MTP, GAPDH, or β-actin with the appropriate secondary antibody conjugated to HRP. Signal was detected by use of a mixture of luminol, p-coumaric acid, and hydrogen peroxide in 100 mM Tris (pH 8.5). Intensity of each band was quantified using Image J software and normalized to β-actin.

2.4. Aortic sinus lesion analysis

At the end of the experiment, the mice were sacrificed and their hearts flushed with saline, embedded in optimum cutting temperature (OCT) medium, and snap-frozen in dry ice. The inferior vena cava was cut to allow the perfusate to exit. Frozen sections of 10 µm thickness were taken in the region of the proximal aorta starting from the end of the aortic sinus and for 300 µm distally, according to the technique of Paigen et al.⁴⁴ Sections were stained with Oil-Red-O and counterstained with haematoxylin. Quantitative analysis of lipid-stained lesions was performed on sections starting at the end of the aortic sinus. The lipid-stained lesions were measured by digitizing morphometry and reported as area (µm² per lesion per mouse).

2.5. Plasma lipoprotein and protein analysis

Mice were fasted overnight prior to blood collection. Plasma cholesterol or triglyceride measurements were performed using the Cholesterol or Triglyceride Reagent kit from RAICHEM, respectively. Mouse lipoproteins were separated from plasma by size exclusion chromatography using a Superose 6 column on a fast protein liquid chromatography (FPLC) system commonly used in our laboratory.²¹ ApoB levels in mouse plasma were analysed by ELISA.

2.6. Microsomal triglyceride transfer protein activity assay

Small pieces (50–70 mg) of liver were collected and homogenized in 0.5 mL of the same buffer. HepG2 cells were washed twice with cold PBS and once with 5 mL of buffer (1 mM Tris–HCl, pH 7.6, 1 mM EGTA, and 1 mM MgCl₂) at 4°C. Livers and cells were collected, vortexed, and centrifuged (TLA 120.2 rotor, 90 000 rpm, 4°C, 1 h). Supernatants were tested for MTP activity using a kit (Chylos Inc., NY, USA) as previously described.²⁶ The triglyceride transfer activity of MTP was measured as % lipid transfer normalized to the protein concentration.

2.7. Immunohistochemistry

For lesion staining of hPCSK9, 5 µm cryosections of the proximal aortas were fixed in ice-cold acetone for 20 min and allowed to dry for 5 min at room temperature. Sections were stained as previously described by Giunzioni et al.⁴⁰ In short, immunostaining was performed with an anti-PCSK9 primary antibody (Hal103) and a rat anti-mouse Ly6C antibody conjugated to biotin. Nuclei were stained with Vectashield containing DAPI. All images shown are projections of a series of 0.5 µm optical sections through the tissue.

2.8. Flow cytometry analyses of circulating CD11b and Ly-6C^hi monocytes

Peripheral blood was collected from mice fed a HFD diet for 8 weeks. To distinguish monocytes from other blood and spleen cells, FITC fluorochrome-tagged rat anti-mouse CD90.2, B220, GR1 (BD-Pharmingen, San Jose, CA), and NK cells (Caltag, Burlingame, CA, USA) were used. Total monocytes were detected using rat anti-mouse CD11b-PE (BD-Pharmingen, San Jose, CA, USA), and Ly6C^hi monocytes were quantitated using rat anti-mouse Ly6-C labelled with biotin (BD Pharmingen) and streptavidin-linked AlexaFluor 647 (Life Technologies, Carlsbad, CA, USA)

2.9. Quantitative real-time PCR (qRT-PCR)

Total RNA was isolated using RNeasy mini kit and processed for reverse transcription. Relative quantification of mRNA expression was performed with QuantStudio qPCR System using TaqMan Gene Expression Assays. Expression levels were calculated using the ΔΔCT method and normalized to 18S ribosomal RNA as internal control. In vivo gene expression data are presented as fold increase over control. Gene expression in transgenic mice (hPCSK9tg, hPCSK9tg/apoE^−/−, or hPCSK9tg/LDLR^−/− mice) is shown as bar graph, whereas the relative control mice (WT, apoE^−/−, or LDLR^−/−, respectively) are represented by the dashed line, which equals 1.

2.10. Statistical analyses

All the statistical analyses were carried out using GraphPad Prism 6 software. The Mann–Whitney test was used to compare cholesterol, triglyceride, and apoB levels in plasma, as well as atherosclerotic lesion size and composition data between two groups. Quantitative real-time PCR (qRT-PCR) data are presented as fold changes from the appropriate control (hypothetical value of 1) and analysed using a one-sample t-test. In all n ≥ 6, data were tested for normality. Results are presented as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.0001.

3. Results

3.1. The effect of PCSK9 on plasma lipids and apoB is mediated by LDLR and apoE

We generated a transgenic line of mice expressing hPCSK9²⁰ and characterized plasma PCSK9 levels and lipoprotein association in mice on either WT, apoE^−/−, or LDLR^−/− background.²¹ After overnight fasting, hPCSK9 significantly increased plasma cholesterol by 96%, triglycerides by 18% (Figure 1A), and apoB by 57% (Figure 1D) in WT mice, driven by increased levels of VLDL and LDL (see Supplementary material online, Figure S1A). In contrast, the absence of either LDLR or apoE negated the effect of hPCSK9 on fasting plasma lipid (Figure 1B and C, respectively) and apoB (Figure 1E and F, respectively) levels. Fasting plasma lipoprotein profiles from transgenic and control LDLR^−/− and apoE^−/− mice are shown in Supplementary material online, Figure S1B and C, respectively. Of note, PCSK9 expression did increase cholesterol and triglyceride levels in LDLR^−/− mice after feeding (Figure 2A and B). Supplementary material online, Figure S1D shows that plasma hPCSK9 levels in our transgenic mice on HFD are on the high end of the physiological range in humans^22,45 and are approximately 25% higher than those of the same mice on chow diet.²¹ Due to the reciprocal regulation between PCSK9 and LDLR,^21,46,47 the absence of LDLR significantly increased mPCSK9 levels 9.3-fold (not shown) as well as hPCSK9 levels by 1.9-fold (see Supplementary material online, Figure S1D). Hepatic LDLR levels were reduced to a similar level in hPCSK9 transgenic mice independently of the presence or absence of apoE (see Supplementary material online, Figure S1E).

Figure 1.

Plasma lipid and PCSK9 levels in human PCSK9 transgenic mice. Total cholesterol and triglyceride levels (mg/dL) in (A) WT (n = 8) and hPCSK9tg (n = 9) mice; (B) LDLR^−/− (n = 15) vs. hPCSK9tg/LDLR^−/− mice (n = 15); and (C) ApoE^−/− (n = 14) vs. hPCSK9tg/apoE^−/− mice (n = 14). Corresponding plasma apoB levels (μg/mL) in (D) WT (n = 8) vs. hPCSK9tg mice (n = 8); (E) LDLR^−/− (n = 8) vs. hPCSK9tg/LDLR^−/− mice (n = 8); and (F) ApoE^−/− (n = 4) vs. hPCSK9tg/apoE^−/− mice (n = 4). **P < 0.01, ***P < 0.01.

Figure 2.

Effect of fasting and feeding on plasma lipids in human PCSK9 transgenic mice. (A) Changes in cholesterol levels in hPCSK9 transgenic mice vs. comparator (WT, LDLR^−/−, or apoE^−/−) in fasting and fed conditions (n = 6–8). (B) Changes in triglyceride levels in hPCSK9 transgenic mice vs. comparator (WT, LDLR^−/−, or apoE^−/−) in fasting and fed conditions (n = 6–8). *P < 0.05, ***P < 0.01.

3.2. PCSK9 mediates hepatic lipogenesis through SREBP and non-SREBP pathways

We have previously reported that PCSK9 increases apoB-TRL intestinal production through both LDLR-dependent and -independent mechanisms.²⁶ Figure 3A shows that hPCSK9 modulates hepatic expression of key genes involved in the SREBP1 pathway, inducing a significant 1.5-fold increase in SRE-binding factor 1 (SREBF1), 1.7-fold increase in acetyl-CoA carboxylase, and a 2.6-fold increase in FAS compared with WT mice. These increases were seen only in mice challenged with HFD. In the absence of either LDLR or apoE, hPCSK9 expression did not increase any of the hepatic SREBP1 genes, irrespective of diet (Figure 3B or C, respectively).

Figure 3.

Sterol-regulated genes in human PCSK9 transgenic mice. mRNA levels of SREBP1-related genes in livers of (A) WT vs. hPCSK9tg; (B) LDLR^−/− vs. hPCSK9tg/LDLR^−/− mice; and (C) apoE^−/− vs. hPCSK9tg/apoE^−/− mice. Corresponding fatty acid synthase (FAS) protein levels in livers of WT vs. hPCSK9tg mice fed either chow diet (D) or HFD (E). Black bars indicate mRNA levels in livers of animals fed a chow diet; grey bars refer to livers of mice fed a HFD. Results are expressed as fold changes compared with the relative control (WT, LDLR^−/−, or apoE^−/−). **P < 0.01, ***P < 0.001. n = 6 or 7 in each group for all genes.

In WT mice, hPCSK9 expression also increased hepatic FAS protein levels, the rate-limiting enzyme for de novo lipogenesis. The increase was modest but significant under chow diet (1.9-fold, P < 0.01) and magnified by HFD feeding (6.1-fold, P < 0.01) (Figure 3D and E, respectively). No effects were found in FAS protein levels in hPCSK9 expressing mice on LDLR^−/− or apoE^−/− mice, irrespective of diet (see Supplementary material online, Figure S2).

To further investigate the effect of PCSK9 on the lipogenic pathway, we studied other key genes and proteins that are not regulated by the SREBP pathway. PCSK9 expression significantly increased apoB, MTP, DGAT1, and apoE expression levels in HFD-fed mice on WT background by 1.8-, 2.3-, 2.3-, 2-fold, respectively. No changes were found in chow-fed mice on WT background (Figure 4A).

Figure 4.

Genes under different regulation in human PCSK9 transgenic mice. mRNA levels of non-SREBP lipogenic genes in livers of (A) WT and hPCSK9tg mice; (B) LDLR^−/− and hPCSK9tg/LDLR^−/− mice; and (C) apoE^−/− and hPCSK9tg/apoE^−/− mice. Corresponding microsomal triglycerides transfer protein (MTP) activity (% of lipid transfer) in livers of (D) HFD-fed WT and hPCSK9tg mice; (E) LDLR^−/− and hPCSK9tg/LDLR^−/− mice; and (F) apoE^−/− and hPCSK9tg/apoE^−/− mice. Black bars indicate mRNA levels in livers of animals fed a chow diet; grey bars refer to livers of mice fed a HFD. Results are expressed as fold changes compared with the relative control (WT, LDLR^−/−, apoE^−/−). *P < 0.05. n = 6 or 7 in each group for all genes.

In the absence of either LDLR or apoE, hPCSK9 expression did not change any of the measured non-SREBP genes (Figure 4B or C, respectively). In line with the effect on the mRNA expression levels, the activity of MTP was increased in livers of hPCSK9tg mice on WT background fed HFD (Figure 4D), whereas no differences were seen in mice lacking either LDLR or apoE (Figure 4E or F, respectively).

Similar results, showing increased SREBP-dependent and -independent gene expression, were obtained in HepG2 cells transduced with WT-PCSK9 or the PCSK9-D374Y GOF mutation (see Supplementary material online, Figure S3).

3.3. PCSK9 increases aortic sinus lesion size and inflammation in the absence of apoE

It was previously reported that the overexpression or knock-down of PCSK9 affects atherosclerotic lesion cholesterol content and size.¹⁶ The overexpression of hPCSK9 produced small, fatty streak-like lesions in mice on WT background, with aortic lesion size of 35 793 ± 15 674 µm², whereas virtually no lipid accumulation occurred in aortas of control mice (Figure 5A). As expected, due to the lack of significant plasma lipid changes, LDLR^−/− mice showed intermediate-size proximal atherosclerotic lesions that were not significantly affected by hPCSK9 expression (192 912 ± 69 837 µm² in hPCSK9tg/LDLR^−/− mice and 160 570 ± 62 425 µm² in LDLR^−/− mice) (Figure 5B). In contrast, despite no changes in plasma lipids, apoE^−/− mice expressing hPCSK9 had 42% larger proximal aortic lesions compared with control apoE^−/− mice (268 655 ± 97 972 µm² vs. 189 423 ± 65 700 µm², respectively) (Figure 5C). Similar results were also obtained from the analysis of the en face lesion area of the whole aorta (see Supplementary material online, Figure S4).

Figure 5.

Aortic lesion size in human PCSK9 transgenic mice. Quantitation (upper panel) and representative sections (lower panel) of Oil-Red-O stained proximal aortas of (A) WT (n = 8) vs. hPCSK9tg (n = 9) mice; (B) LDLR^−/− (n = 15) vs. hPCSK9tg/LDLR^−/− (n = 15) mice; and (C) apoE^−/− (n = 14) vs. hPCSK9tg/apoE^−/− (n = 14) mice. Results are expressed in µm² per section per mouse. *P < 0.05, ***P < 0.001. Scaled bar, 200 µm.

LDLR is the dominant mediator of plasma PCSK9 clearance and hepatic internalization.^21,46 Immunostaining of the aortic sinus showed that hPCSK9 is found in lesions of apoE^−/− mice (Figure 6A) but not in those of LDLR^−/− mice (Figure 6B). Finally, we studied whether accumulation of hPCSK9 in the lesion affects local inflammation. Figure 7A and B shows that hPCSK9 increases the infiltration of inflammatory Ly6C^hi monocytes by 21% (P < 0.05) in mice lacking apoE. In contrast, lesions of hPCSK9 transgenic mice on LDLR^−/− background showed similar levels of Ly6C^hi monocytes compared with control LDLR^−/− mice (Figure 7C andD, respectively). The small lesions in transgenic mice on WT background did not allow for detection of aortic hPCSK9 or Ly6C^hi monocytes. Despite increased Ly6C^hi monocytes infiltration to the atherosclerotic lesion, no differences were found in blood CD11b⁺ and Ly6C^hi levels in transgenic mice expression of hPCSK9 either on WT, LDLR^−/−, or apoE^−/− background (see Supplementary material online, Figure S5).

Figure 6.

Human PCSK9 in the artery wall of transgenic mice. Immunostaining to detect human PCSK9 in atherosclerotic lesions of proximal aortas in hPCSK9tg mice on apoE^−/− (A) or LDLR^−/− background (B). FITC-hPCSK9 staining is green. DAPI staining is blue. Scaled bar, 100 µm.

Figure 7.

Inflammatory monocytes in the artery wall of transgenic mice. Quantification and representative staining of Ly6C^hi-positive cells (vs. all cells) in the lesions of apoE^−/− (n = 6) vs. hPCSK9tg/apoE^−/− (n = 6) mice (A and B, respectively), and LDLR^−/− (n = 8) and hPCSK9tg/LDLR^−/− (n = 6) mice (C and D, respectively). Ly6C staining is green. DAPI staining is blue. *P < 0.05. Scaled bar, 20 µm.

4. Discussion

PCSK9 targets the LDLR for lysosomal degradation, leading to increased LDL-C levels in humans and mice.^2,3 We, and others, have shown that hPCSK9 can also affect the production of intestinal apoB-TRL through both transcriptional and post-transcriptional mechanisms.^26,34,35 Although a role for PCSK9 in hepatic lipoprotein production has been described both in humans^28,29,31 and in mice,^17,33 the underlying cellular mechanisms of action of the effect of PCSK9 on hepatic lipogenesis are not understood. In addition, it remains to be established whether PCSK9 exerts direct effects on atherogenesis in the absence of systemic lipid changes. Our results support a scenario where hPCSK9 modulates plasma cholesterol and triglyceride levels in an LDLR- and apoE-dependent manner that involves hepatic activation of synthesis of sterol- and non-sterol-regulated proteins. We further show that hPCSK9 expression affects atherosclerotic lesion size and composition even in the absence of systemic lipid changes but only if the LDLR is present.

Here, we first demonstrated that in WT mice challenged with a HFD, expression of hPCSK9 generates a pro-atherogenic lipid profile. Whereas elevated cholesterol and apoB levels are the obvious consequence of LDLR degradation and reduced LDL-C uptake, the increases in plasma triglyceride (Figure 1) and TRL levels (see Supplementary material online, Figure S1) are less intuitive and less understood. It has previously been shown that PCSK9 GOF mutations can increase hepatic apoB-TRL production in both humans^28,29,31 and mice.^17,31,33 Thus, we investigated here the cellular mechanisms accompanying the effects of WT hPCSK9 on hepatic lipogenesis. We show that hPCSK9 expression increases hepatic synthesis of both SREBP-dependent and other non-SREBP lipid-modulating genes (Figures 3and4). Notably, the PCSK9-mediated effects on lipogenic genes in the liver occur only when mice are fed the HFD, which suggests that dietary fat and cholesterol are necessary to trigger lipogenic responses upon which PCSK9 has modulatory effects. PCSK9 also exerts post-transcriptional effects on FAS, as protein levels were increased by almost two-fold whereas the mRNA levels were unchanged. This effect was enhanced by HFD feeding, which induced a 2.6-fold increase in FAS mRNA levels, and 6-fold increase in hepatic FAS protein levels. HFD also caused a significant up-regulation of apoE, a protein involved in both hepatic lipoprotein secretion and uptake.^48,49 We further show that in addition to the effect of PCSK9 in the liver (See Supplementary material online, Figure S1) and small intestine²⁶ under fasting conditions, PCSK9 also reduces LDLR and increases apoE accumulation in the liver and small intestine after feeding (see Supplementary material online, Figure S6). Overall, hPCSK9 expression in WT mice reduced LDLR, increased plasma VLDL and LDL levels, induced hepatic lipogenesis, and caused the development of small aortic sinus lesions (Figure 5). The multiple cellular pathways by which PCSK9 acts to modulate plasma lipids warrant a modification of the current paradigm based on PCSK9 solely affecting hepatic LDLR levels and LDL-C uptake.

The main target of PCSK9 is hepatic LDLR, which in turn mediates the clearance of PCSK9 from plasma, regardless of the tissue origin of PCSK9.^21,40,50 It was previously suggested that PCSK9 binding to LDL could affect its activity.^51,52 Although we cannot distinguish the effect of LDL-bound and apoB-free PCSK9, it is noteworthy that the relative amount of LDL-bound PCSK9 in our transgenic model is within the range found in human plasma (25–40% association).²¹ The absence of LDLR in mice or the presence of two receptor-negative LDLR mutations in patients will produce elevated LDL-C levels that are not sensitive to modulations by PCSK9.^16,21,46,53 Similarly, we show that the overexpression of hPCSK9 in mice lacking LDLR does not affect plasma cholesterol, although a mild (non-significant) effect was found on plasma triglyceride and apoB levels (Figure 1). We previously reported that hPCSK9 affects intestinal apoB-TRL production also through an LDLR-independent mechanism.²⁶ Here, we wanted to test whether PCSK9 affects hepatic lipogenesis in an LDLR-independent manner. In the absence of LDLR, PCSK9 does not affect the expression of genes and proteins involved in hepatic lipogenesis (both SREBP- and non-SREBP-mediated), thus suggesting that the previously reported LDLR-independent effects on TRL secretion^26,35 are specific to enterocytes. Our results (Figure 2A and B) further suggest that PCSK9 increases the intestinal, not the hepatic, contribution to plasma cholesterol and triglyceride levels in the absence of LDLR.

Proximal aortic lesion size and en face aortic lesion trended toward increased lesion size in hPCSK9tg/LDLR^−/− mice compared with LDLR^−/− controls (+28%, P = 0.124, and + 17%, P = 0.47, respectively). This trend likely represents a small contribution of intestinal lipoproteins to atherosclerosis. However, the lack of statistical significance does not allow us to make a case for any contribution of intestinal lipoproteins to atherosclerosis in our model.

The removal of LDLR in apoE^−/− mice (LDLR^−/−/apoE^−/− double knockouts) is known to have little effect on plasma lipoproteins relative to apoE deficiency only.⁵⁴ In other words, in the absence of the ligand (apoE), the removal of the receptor (LDLR) does not further aggravate the clearance of the accumulating remnant lipoprotein. Also in our transgenic mice on apoE^−/− background, where ∼70% of hepatic LDLR is removed by the action of hPCSK9, no effects on plasma cholesterol, triglyceride, apoB, or lipoproteins levels were noted (see Supplementary material online, Figure S1 and Figure 1, respectively). Similar results were previously shown in mice either lacking or overexpressing mPCSK9 on apoE^−/− background.^16,39 In the current study, we show that in WT mice the expression of hPCSK9 increases hepatic apoE mRNA levels by approximately two-fold (Figure 4), suggesting the involvement of apoE in PCSK9-mediated hepatic lipogenesis. We also show that the absence of apoE abrogates both transcriptional and post-transcriptional effects of PCSK9 on hepatic lipogenic pathways (see Supplementary material online, Figure 2S, Figures 3and4), suggesting a direct causal role for apoE in PCSK9-mediated hepatic lipogenesis. It was previously shown that the LDLR promotes post-translational degradation of apoB and thereby reduces VLDL particle secretion and that LDLR-dependent apoB degradation occurs after exit from the ER and that it involves both of the LDLR ligands on the VLDL particle, apoB, and apoE.³²

Whether PCSK9 influences atherosclerosis in apoE^−/− mice has been a subject of previous studies. PCSK9 knock-down or inhibition was shown to have no effect on atherosclerotic lesion size in mice lacking apoE,^16,39 whereas overexpression of PCSK9 in the same mice significantly increased lesion size.^16,18 Since PCSK9 levels in the above studies were not reported, it is difficult to compare these findings and place in a physiological context. Here, we show for the first time that the expression of hPCSK9 at the high end of the physiological range^22–25 increases aortic sinus lesion size by >40% (Figure 5) in the absence of apoE. Also, the accumulation of PCSK9 in atherosclerotic lesions has not been studied to date, likely due to technical limitations with immuno-histological analyses of mPCSK9.⁵⁵ Our results show that hPCSK9 is present in the lesions of hPCSK9 transgenic mice of apoE^−/− background, but not in those of LDLR^−/− background. This strongly suggests that the accumulating PCSK9 in the lesion is trapped by LDLR-mediated uptake in atheroma cells, and not simply from local synthesis. We have recently reported similar findings, showing LDLR-dependent accumulation of PCSK9 in atherosclerotic lesions of mice expressing PCSK9 from bone marrow-derived cells.⁴⁰ It was previously reported that in addition to reducing plasma cholesterol levels, anti-PCSK9 antibodies also inhibit atherosclerosis, improve lesion morphology, and reduce vascular inflammation in mice.⁵⁶ Our data show that the absence of LDLR-mediated trapping of PCSK9 in the artery wall negates the effects of PCSK9 on vascular inflammation (Figure 7) and atherogenesis (Figure 5). Since deletion of PCSK9 does not affect plasma lipids levels in apoE^−/− mice,³⁹ therapeutic inhibition of PCSK9 is also not expected to affect plasma lipids in these mice. However, anti-PCSK9 antibodies are likely to inhibit LDLR-mediated trapping of PCSK9, and thus it seems reasonable to hypothesize that these antibodies will inhibit vascular inflammation and atherosclerosis development in apoE^−/− mice.

PCSK9 can have biological effects in the atheroma by its known action on the LDLR, as we previously showed that removal of macrophage LDLR reduces atherogenesis in the mouse.⁵⁷ However, PCSK9 can also engage with other receptors or proteins that modulate plaque formation.^58,59 We studied whether the accumulation of PCSK9 affects lesion composition in apoE^−/− mice. Recently, several possible associations have been suggested to support an involvement of PCSK9 in the inflammatory process: (i) PCSK9 levels correlate with white blood cell count in patients affected by stable coronary artery disease⁴¹; (ii) the pro-inflammatory cytokine IL-1β disrupts cholesterol-mediated feedback regulation of LDLR expression through mechanisms involving activation of the mammalian target of rapamycin^60,61; (iii) absence of PCSK9 protects against septic shock induced by LPS administration⁶²; and (iv) marrow-derived PCSK9 alters lesion inflammation (without affecting size) by promoting macrophage differentiation.⁴⁰ Our results here show that the accumulation of hPCSK9, as seen in hPCSK9tg/apoE^−/− mice (Figure 6), is accompanied by increased infiltration of Ly6C^hi monocytes into the atherosclerotic lesion (Figure 7). In contrast, when systemic expression of hPCSK9 does not lead to its accumulation in the lesion, as is the case in hPCSK9/LDLR^−/− mice, no changes are seen in the number of atheroma Ly6C^hi monocytes. As Ly6C^hi could not be detected in the aortic wall of WT and hPCSK9tg mice on WT background, it remains to be determined if the lack of LDLR blocked PCSK9-mediated inflammation in the aortic wall or whether the absence of apoE promoted PCSK9-mediated inflammation.

The lack of differences in blood CD11b⁺ and Ly6C^hi monocytes in transgenic mice in all backgrounds (WT, LDLR^−/−, and apoE^−/−) suggests that the effect of PCSK9 on inflammation is not systemic but rather localized to the atherosclerotic lesion. We have recently characterized the direct effect of macrophage-derived PCSK9 on the atherosclerotic lesion and showed that local PCSK9 modulates macrophage differentiation and monocyte infiltration into the artery wall.⁴⁰

It was previously shown that homozygous familial hypercholesterolemia (FH) subjects carrying defective LDLR mutants (<25% of normal LDLR activity) still partially benefit from PCSK9 inhibition with a 20% reduction in LDL-C levels.^24,53 In contrast, homozygous FH subjects carrying negative LDLR mutants (<2% of normal LDLR activity) do not respond at all to PCSK9 inhibition (54, 64). Our data support a scenario where PCSK9 internalization by the LDLR also plays a role in vascular wall inflammation, thus suggesting that in subjects with double-negative LDLR mutants, PCSK9 inhibition is not likely to produce anti-inflammatory effects in the vascular wall.

In conclusion, we show that clinically relevant human PCSK9 levels affect plasma lipid and lipoprotein levels not only through reduced hepatic lipoprotein clearance but also via activation of hepatic lipogenesis, a phenomenon dependent on the presence of both the LDLR and apoE, and influenced by both transcriptional and post-transcriptional events in hepatic lipogenesis. Moreover, we found that PCSK9 promotes atherogenesis both indirectly, by raising plasma lipids, and directly via modulation of the entry of inflammatory monocytes into the artery wall. Our results suggest additional reasons to expect cardiovascular benefits in the ongoing Phase III clinical trials of the recently approved anti-PCSK9 therapies.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Funding

This study was supported by the National Institutes of Health (NHLBI) through grant R01-HL106845 to S.F.