Heparan sulfate proteoglycans present PCSK9 to the LDL receptor

Abstract

Coronary artery disease is the main cause of death worldwide and accelerated by increased plasma levels of cholesterol-rich low-density lipoprotein particles (LDL). Circulating PCSK9 contributes to coronary artery disease by inducing lysosomal degradation of the LDL receptor (LDLR) in the liver and thereby reducing LDL clearance. Here, we show that liver heparan sulfate proteoglycans are PCSK9 receptors and essential for PCSK9-induced LDLR degradation. The heparan sulfate-binding site is located in the PCSK9 prodomain and formed by surface-exposed basic residues interacting with trisulfated heparan sulfate disaccharide repeats. Accordingly, heparan sulfate mimetics and monoclonal antibodies directed against the heparan sulfate-binding site are potent PCSK9 inhibitors. We propose that heparan sulfate proteoglycans lining the hepatocyte surface capture PCSK9 and facilitates subsequent PCSK9:LDLR complex formation. Our findings provide new insights into LDL biology and show that targeting PCSK9 using heparan sulfate mimetics is a potential therapeutic strategy in coronary artery disease.PCSK9 interacts with LDL receptor, causing its degradation, and consequently reduces the clearance of LDL. Here, Gustafsen et al. show that PCSK9 interacts with heparan sulfate proteoglycans and this binding favors LDLR degradation. Pharmacological inhibition of this binding can be exploited as therapeutic intervention to lower LDL levels.

Fig. 1

PCSK9 cell surface binding and activity depend on HSPG. a A heparin pentasaccharide (SANORG, sticks) (PDB ID 1E03) docked at the electrostatic surface (red negative; blue positive) of the HSPG-binding site in PCSK9 (PDB ID 3H42) with positively charged amino acids (R: arginine, H: histidine) indicated. b Superposition of SANORG with PCSK9 (ribbon). c Non-permeabilized HepG2 cells-expressing PCSK9 (green) after treatment with or without heparinase I. Nuclei were stained with Hoechst (blue). The experiment was repeated three times with similar results. d PCSK9 binding to heparin was analyzed by affinity chromatography. e PCSK9 mutant with all six amino acids highlighted in a, b substituted for alanines (mut 5) showed complete loss of heparin binding. Substitution of R93 alone for cysteine (R93C) also resulted in complete loss of heparin binding. f, g HepG2 cells incubated 18 h with WT PCSK9 and PCSK9 mut 5 (10 nM of each, n = 4). h PLA analysis of non-permeabilized HepG2 cells showed that co-localization of LDLR and PCSK9 was markedly reduced upon incubation of cells with heparin (500 µg/ml). The experiment was repeated three times with similar results. i, j LDLR levels in HepG2 cells following incubation with heparin (500 µg/ml, 18 h, n = 4). k Heparin treatment resulted in accumulation of PCSK9 in the culture supernatant (n = 3). l Effect of heparin on LDLR and PCSK9 mRNA levels. m Heparin (50 and 500 µg/ml) had no effect on the interaction between PCSK9 and LDLR in a cell-free assay. A PCSK9 antibody (αPCSK9, 5 nM) directed against the LDLR-binding site is included as positive control (n = 3). Bar graphs show mean values (n as indicated) with s.e.m. error bars. Statistical significance was evaluated using a two-tailed Student’s t-test. Scale bars are 10 µm. Supplementary Fig. 10 shows uncropped gel images

Fig. 2

Heparin mimetics are potent PCSK9 inhibitors. Structure of the heparin mimetics dextran sulfate (a), pentosan sulfate (b), suramin (c), and the phosphorothioate oligonucleotide S-dC-36 (d) (R = SO₃ ⁻ or H). Binding curves and the dose-dependent effects on LDLR levels in HepG2 cells (n = 3–5, bar graphs show mean values with s.e.m. error bars) are shown below each mimetic structure. Statistical significance was evaluated using a two-tailed Student’s t-test. Summary of binding constants and representative LDLR Western blots are shown in Supplementary Fig. 5d, e

Fig. 3

Structure of PCSK9 in complex with a dextran sulfate disaccharide. a Structure of dextran sulfate disaccharide on the electrostatic surface (blue: positive; red: negative) of the pro-domain of PCSK9 (PDB ID: 5OCA) with amino acids indicated (H: histidine and R: arginine). b Side view of the binding site containing conformation 1 of dextran sulfate with interacting residues indicated (Q: glutamine, T: threonine). c Simulated annealing Fc-Fo composite omit map (positive density show in green and contoured at 3σ) of dextran sulfate in two alternative conformations (confirmations 1 (cyan) and 2 (magenta)). Interacting residues of PCSK9 are indicated and show together with a simulated annealing 2Fc-Fo composite omit map (blue and contoured at 1σ)

Fig. 4

PCSK9 binding to synthetic glycans. a Dissection of the interaction between PCSK9 and extracellular matrix glycans (numbers refer to Supplementary Table 2) using a synthetic glycan microarray. Natural heparin is included as a positive control. b The heparin structures immobilized in the glycan microarray are shown. Structures interacting with PCSK9 contain repeats of [4)-α-GlcN-6,N-disulfate(1 → 4)-α-IdoA-2-sulfate-(1→]

Fig. 5

Generation of inhibitory mAbs directed against the PCSK9 HSPG-binding site. a Rats were immunized with cDNA encoding a rat PCKS9 chimera encompassing the human HSPG-binding site. The overall sequence identity between rat and human PCSK9 is 77%. b IP of radioactive-labeled PCSK9 (35S-PCSK9) using serum from three immunized rats. Preimmune serum is used as control. See full gel in Supplementary Fig. 7a. c LDLR levels in HepG2 cells following incubation with IgG from immunized animals compared to preimmune IgG (1 µg/ml). d mAbs were tested by IP. Fifteen mAbs specifically precipitated radioactive-labeled PCSK9. e LDLR levels in HepG2 cells incubated with the individual mAbs (n = 3–5). Twelve individual mAbs showed a significant and around 2-fold increase in LDLR, and were tested for binding to PCSK9 mut 5 by IP. f A polyclonal antibody was used as positive control (R&D systems, AF3888). Three mAbs, including 5E11, failed to precipitate PCSK9 mut 5. g Uptake of fluorescently labeled LDL (DiI-LDL) in HepG2 cells incubated with 100 nM PCSK9 alone (control) or in combination with 5E11 or evolocumab as indicated (n = 3). Bar graphs show mean with s.e.m. error bars. Statistical significance was evaluated using a two-tailed Student’s t-test. Supplementary Fig. 10 shows uncropped gel images

Fig. 6

Enzymatic removal of liver heparan sulfate releases PCSK9 and ablates its activity. a, b Infusion of heparinase I prior to the injection of PCSK9 (10 µg) completely inhibits PCSK9-induced degradation of LDLR. Western blot of representative samples is shown in a and quantification of LDLR in b (control n = 7, PCSK9 n = 6, heparinase n = 5, heparinase/PCSK9 n = 5). Heparinase treatment unmasks the antigenicity of the major liver HSPG syndecan-1 (middle panel). Beta-actin is used as loading control (lower panel). c Heparinase I treatment leads to an increase in plasma PCSK9 as measured by ELISA 15 min after injection (control n = 6, heparinase n = 6). d Western blot (representative samples) of liver syndecan-1 is used as control of heparinase injection. Beta-actin is shown as loading control. e–h Transgenic mice with constitutive expression of human heparanase (Hpa-tg) (n = 7) have increased plasma PCSK9 (e), increased liver LDLR (f, g), and reduced plasma cholesterol (h) compared to control WT mice (n = 6). Bar graphs show mean with s.e.m. error bars. Statistical significance was evaluated using a two-tailed Student’s t-test. Supplementary Fig. 11 shows uncropped gel images

Fig. 7

Inhibition of PCSK9:HSPG interaction protects LDLR in vivo. a LDLR is protected in mice co-injected with PCSK9 (10 µg) and heparin (500 µg) or suramin (300 µg) compared to mice injected with PCSK9 alone (control n = 15, PCSK9 n = 10, PCSK9/heparin n = 7, PCSK9/suramin n = 3). b, c A single injection of suramin (50 mg/kg) increased liver LDLR levels (b) and reduced total plasma cholesterol (c) 18 h later in WT mice (control n = 5, suramin n = 8) but not in PCSK9 KO mice (control n = 5, suramin n = 7). d, e Co-injection of PCSK9 (10 µg) with mAb 5E11 (50 µg) protects the LDLR from degradation (control n = 4, PCSK9 n = 7, PCSK9/mAb 5E11 n = 6). Representative blot is shown. f, g PCSK9 with mutated HSPG-binding site (mut 5) (10 µg) is ineffective in inducing LDLR degradation (n = 3 of each). h PCSK9 but not PCSK9 mut 5 significantly increases total cholesterol in mice fed Western-type diet 6 h post injection (control n = 25, PCSK9 n = 12, PCSK9 mut 5 n = 9). i Injected PCSK9 mut 5 remained in circulation for a prolonged time compared to PCSK9 WT in line with reduced capture by HSPG and clearance by LDLR. Bar graphs show mean with s.e.m. error bars. Statistical significance was evaluated using a two-tailed Student’s t-test. Supplementary Fig. 11 shows uncropped gel images

Fig. 8

Model depicting the role of HSPG in PCSK9 activity in the hepatocyte. The model shows a hepatocyte and how LDLR in the absence of PCSK9 mediates endocytosis of bound LDL cholesterol particles followed by lysosomal degradation of LDL, while LDLR recycles to the cell surface. HSPG mediates the capture of PCSK9 and its subsequent presentation to LDLR thereby directing LDLR itself to lysosomes for degradation. In the model, we have depicted HSPG as syndecan but based on the present data we cannot exclude that it may also be a glypican