Original Research

Low-Density Receptor–Dependent and Low-Density Lipoprotein Receptor–Independent Mechanisms of Cyclosporin A–Induced Dyslipidemia

Maaike Kockx, Elias Glaros, Betty Kan, Theodore W. Ng, Jimmy F.P. Berbée, Virginie Deswaerte, Diana Nawara, Carmel Quinn, Kerry-Anne Rye, Wendy Jessup, Patrick C.N. Rensen, Peter J. Meikle, Leonard Kritharides

Objective—Cyclosporin A (CsA) is an immunosuppressant commonly used to prevent organ rejection but is associated with hyperlipidemia and an increased risk of cardiovascular disease. Although studies suggest that CsA-induced hyperlipidemia is mediated by inhibition of low-density lipoprotein receptor (LDLr)–mediated lipoprotein clearance, the data supporting this are inconclusive. We therefore sought to investigate the role of the LDLr in CsA-induced hyperlipidemia by using Ldlr-knockout mice (Ldlr−/−). Approach and Results—Ldlr−/− and wild-type (wt) C57Bl/6 mice were treated with 20 mg/kg per d CsA for 4 weeks. On a chow diet, CsA caused marked dyslipidemia in Ldlr−/− but not in wt mice. Hyperlipidemia was characterized by a prominent increase in plasma very low–density lipoprotein and intermediate-density lipoprotein/LDL with unchanged plasma high-density lipoprotein levels, thus mimicking the dyslipidemic profile observed in humans. Analysis of specific species by liquid chromatography–tandem mass spectrometry suggested a predominant effect of CsA on increased very low–density lipoprotein–IDL/LDL lipoprotein number rather than composition. Mechanistic studies indicated that CsA did not alter hepatic lipoprotein production but did inhibit plasma clearance and hepatic uptake of [14C]cholesteryl oleate and glycerol tri[3H]oleate-double-labeled very low–density lipoprotein–like particles. Further studies showed that CsA inhibited plasma activity and increased levels of C-III and proprotein convertase subtilisin/kexin type 9. Conclusions—We demonstrate that CsA does not cause hyperlipidemia via direct effects on the LDLr. Rather, LDLr deficiency plays an important permissive role for CsA-induced hyperlipidemia, which is associated with abnormal lipoprotein clearance, decreased lipoprotein lipase activity, and increased levels of apolipoprotein C-III and proprotein convertase subtilisin/kexin type 9. Enhancing LDLr and lipoprotein lipase activity and decreasing apolipoprotein C-III and proprotein convertase subtilisin/kexin type 9 levels may therefore provide attractive treatment targets for patients with hyperlipidemia receiving CsA. (Arterioscler Thromb Vasc Biol. 2016;36:00-00. DOI: 10.1161/ATVBAHA.115.307030.) Key Words: apolipoprotein C-III ◼ hyperlipidemia ◼ immunosuppression ◼ lipolysis ◼

yperlipidemia is observed in 40% to 60% of organ trans- hyperlipidemia in patients, such as post-transplantation obe- Hplant recipients and has been linked to the use of immu- sity, multiple drug therapy, and diabetes mellitus, CsA mono- nosuppressant agents such as corticosteroids, sirolimus, and therapy can independently lead to elevated plasma cyclosporine A (CsA).1–3 Transplant hyperlipidemia is char- and levels in humans which are reversible on ces- acterized by high plasma cholesterol and triglyceride levels.2 sation of immunosuppression therapy.4 Specifically, CsA increases very low–density lipoprotein The immunosuppressive effect of CsA is mediated by inhi- (VLDL) and low-density lipoprotein (LDL) concentrations bition of protein phosphatase 2B (calcineurin) and subsequent and shows variable effects on plasma high-density lipo- activation of the transcription factor nuclear factor of transcrip- protein. Although multiple factors potentially contribute to tion. The mechanism(s) by which CsA leads to hyperlipidemia

Received on: December 11, 2015; final version accepted on: April 20, 2016. From the ANZAC Research Institute (M.K., D.N., W.J., L.K.) and Department of Cardiology (L.K.), Concord Hospital, University of Sydney, Sydney, Australia; Centre for Vascular Research (E.G., C.Q.) and Department of Pathology (B.K.), University of New South Wales, Sydney, Australia; Baker IDI Heart and Diabetes Institute, Melbourne, Australia (T.W.N., P.J.M.); Department of Medicine, Division Endocrinology, and Eindhoven Laboratory for Experimental Vascular Medicine, Leiden University Medical Centre, Leiden, The Netherlands (J.F.P.B., P.C.N.R.); Centre for Innate Immunity and Infectious Diseases, Hudson Institute of Medical Research, Clayton, Australia (V.D.); Lipid Research Group, School of Medical Sciences, University of New South Wales Australia, Sydney, Australia (K.-A.R.). The online-only Data Supplement is available with this article at http://atvb.ahajournals.org/lookup/suppl/doi:10.1161/ATVBAHA.115.307030/-/DC1. Correspondence to Leonard Kritharides, Department of Cardiology, Concord Repatriation General Hospital, Concord, New South Wales 2139, Australia. E-mail [email protected] © 2016 American Heart Association, Inc. Arterioscler Thromb Vasc Biol is available at http://atvb.ahajournals.org DOI: 10.1161/ATVBAHA.115.307030

Downloaded from http://atvb.ahajournals.org/1 at University of Sydney on May 11, 2016 2 Arterioscler Thromb Vasc Biol July 2016

Nonstandard Abbreviations and Acronyms Materials and Methods Materials and Methods are available in the online-only Data apo apolipoprotein Supplement. CE cholesteryl ester CO cholesteryl oleate Results CsA cyclosporin A CsA Treatment Hprt Hypoxanthine guanine phosphoribosyl transferase Mice were administered 20 mg/kg per d CsA or vehicle (con- IDL intermediate-density lipoprotein trol) via osmotic minipumps. This CsA dose has previously LDL low-density lipoprotein been shown to reduce transplant rejection in mice while not LDLr LDL receptor inducing liver and kidney toxicity.12,13 Whole-blood CsA lev- LPL lipoprotein lipase els were 1087±124 ng/mL and 711±91 ng/mL after 1 week PCSK9 proprotein convertase subtilisin/kexin type 9 and 4 weeks, respectively, consistent with levels reported by SM sphingomyelin others using a similar delivery method.14 Mouse weights were TO trioleate/triolein not affected (23.3±0.3 versus 23.2±0.3 g, for control and CsA, VLDL very low–density lipoprotein respectively). Histological examination of liver and kidney wt wild-type sections revealed no signs of toxicity after CsA treatment (Figure I in the online-only Data Supplement). −/− are most likely not related to nuclear factor of transcription It is known that Ldlr mice have increased hepatic lipid inhibition1,4 as tacrolimus (FK506), another nuclear factor of content compared with C57Bl/6 wild-type (wt) mice and transcription–inhibiting immunosuppressant, does not induce are more vulnerable to develop nonalcoholic steatohepatitis hyperlipidemia.3 Animal and cell studies have indicated that after high-fat feeding, although they do not develop fatty 15 CsA may affect many aspects of lipid metabolism.1 In 1996, livers on a chow diet. This was supported by our histologi- Rayyes5 and Winegard6 suggested that CsA increased plasma cal studies (Figure I in the online-only Data Supplement). LDL and VLDL levels via decreased low-density lipopro- Hepatic cholesterol, cholesteryl ester (CE), and triglyc- −/− tein receptor (LDLr) activity. The precise role of the LDLr eride contents were higher in Ldlr mice compared with in mediating CsA-mediated hypercholesterolemia, however, wt mice, most marked in triglyceride content (115 versus is unclear. Cell culture studies identified that CsA can inhibit 30 nmol/mg; P<0.05; Table 1). In the wt mice, CsA treat- ment increased hepatic CE content from 11 to 20 nmol/mg β-VLDL uptake via the LDLr in the human hepatoma HepG2 cell line.6 CsA was also found to reduce LDLr-mediated bind- (P<0.01) without significantly increasing other . In Ldlr−/− mice, CsA caused no significant changes in hepatic ing and uptake of LDL in HepG2 cells, and this was associated lipid content. with a reduction in hepatic LDLr mRNA levels.5 However, in vivo studies did not corroborate these findings. Hepatic LDLr protein levels decreased in mice treated with CsA7 but were CsA Treatment Increases Plasma Lipid −/− unaffected in rats8 even though both models demonstrated Levels in Ldlr Mice but not in wt Mice CsA-induced hyperlipidemia. These studies suggest that CsA- CsA treatment did not affect plasma total cholesterol or tri- mediated hyperlipidemia may be related to effects on targets glyceride levels in wt mice (Figure 1A and 1B). However, −/− other than the LDLr, such as activation of hepatic cholesterol in Ldlr mice fed a chow diet, CsA markedly elevated total synthesis or inhibition of lipoprotein lipolysis. In vivo studies plasma cholesterol (263±12 versus 527±31 mg/dL, control in rats demonstrated that CsA reduced LDL production and versus CsA, respectively; P<0.0001) and triglyceride levels catabolism,9 while in renal transplant recipients reduced catab- olism of -like emulsion particles, and a reduced Table 1. Hepatic Tissue Lipid Levels 10,11 fractional catabolic rate of VLDL has been observed. Control CsA Given the current and increasing clinical importance −/− of contemporary therapies such as statins and proprotein Ldlr convertase subtilisin/kexin type 9 (PCSK9) inhibitors in TG 115±30 86±28 promoting LDLr-mediated lipoprotein uptake, we sought Cholesterol 25±2 28±2 to investigate the role of the LDLr in CsA-induced hyper- CE 30±4 46±16 lipidemia by studying Ldlr-knockout mice. We report that CsA induces hyperlipidemia in chow-fed Ldlr−/− mice but wt not in wild-type (wt) C57Bl/6 mice, indicating that LDLr TG 30±10 34±8 deficiency has an indirect role in the development of CsA- Cholesterol 17±1 21±2 induced hyperlipidemia. We further show that CsA inhibits CE 11±1 20±2* lipoprotein clearance in the Ldlr−/− mouse and modulates plasma apolipoprotein C-III (apoC-III) levels, lipoprotein Lipid levels were determined by LC/MSMS. Data are presented nmol per mg protein and is mean±SEM from 7 to 8 mice per group. CE indicates cholesteryl lipase (LPL) activity, and PCSK9 levels, suggesting a com- esters; CsA, cyclosporin A; LC/MSMS, liquid chromatography–tandem mass plex interaction among CsA, LDLr, and lipoprotein metabo- spectrometry; TG, triglyceride; and wt, wild-type. lism in vivo. *P<0.01 for control vs CsA.

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 Kockx et al Mechanisms of Cyclosporin A–Induced Dyslipidemia 3

Figure 1. Cyclosporin A (CsA) increases plasma cholesterol and triglyceride levels in Ldlr-knockout mice (Ldlr−/−) mice. Wild-type (wt) and Ldlr−/− mice were treated with 20 mg/kg per d CsA or vehicle (Ctrl) for 4 wk. Plasma cholesterol (Chol; A and C) and triglycerides (TG; B and D) and apolipoprotein (apo) levels (E) were determined by enzymatic assays and Western blotting. Data shown are mean±SEM for 8 mice each. Corresponding Chol levels in mmol/L are 2.3±0.2 and 2.3±0.4 for wt Ctrl and CsA and 6.8±1.0 and 13.6±2.4 for Ldlr−/− Ctrl and CsA, respectively. Corresponding TG levels in mmol/L are: 1.6±0.3 and 1.6±0.5 for wt Ctrl and CsA and 3.5±0.9 and 5.7±1.3 for Ldlr−/− Ctrl and CsA, respectively. Representative Western blots for 8 mice per treatment are shown, bar graph represents mean±SEM with vehicle control set as AU=1. **P<0.001; ***P<0.0001.

(156±13 versus 253±19 mg/dL in control versus CsA, respec- 1.9-fold, 4.2-fold, and 2.5-fold for VLDL-triglyceride, tively; P<0.001; Figure 1C and 1D). The increase in plasma VLDL-cholesterol, and IDL/LDL-cholesterol, respectively lipid levels was paralleled by an increase in plasma apoB100, (Figure 2C and 2D). apoB48, and apoE levels, whereas no differences in plasma Insulin tolerance tests demonstrated a similarly rapid apoA-I were observed (Figure 1E). initial decline in blood glucose levels within 15 minutes of Analysis of plasma lipoprotein fractions by fast pro- insulin administration in both control and CsA-treated mice tein liquid chromatography established that wt mice had a (Figure IIA in the online-only Data Supplement). Blood glu- prominent high-density lipoprotein peak and barely detect- cose levels, however, remained lower in the CsA-treated mice able VLDL/intermediate-density lipoprotein (IDL)/LDL at 60 minutes, suggesting increased rather than decreased fractions (Figure 2A and 2B), and this lipoprotein profile insulin sensitivity with CsA, which was supported by quan- did not change after CsA treatment. In contrast, Ldlr−/− tification of the glucose area under the curve (Figure IIB in mice treated with vehicle had prominent VLDL, IDL/ the online-only Data Supplement). These data indicate that LDL-cholesterol, and VLDL-triglyceride levels that were although CsA-treated mice exhibit marked hyperlipidemia, markedly higher after administration of CsA, increasing they are not insulin resistant.

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 4 Arterioscler Thromb Vasc Biol July 2016

Figure 2. Cyclosporin A (CsA) increases plasma very low–density lipoprotein (VLDL) and low-density lipoprotein (LDL) levels in Ldlr- knockout mice (Ldlr−/−) mice. Plasma VLDL, LDL and high-density lipoprotein (HDL) were separated by fast protein liquid chromatography. Cholesterol (A and C) and triglyceride (B and D) in each fraction were determined as described in the Materials and Methods available in the online-only Data Supplement. A and B, Wild-type (wt) mice. C and D, Ldlr−/− mice. Data shown are mean±SEM from 5 to 7 mice each fractionated individually.

Lipidomics of Plasma and Lipoprotein Fractions levels were all significantly elevated after CsA treatment To investigate if the changes in lipid levels were related to (Table 2). In addition, significant increases in plasma levels changes in particle number or qualitative changes in particle of diacylglycerol and phosphatidylcholine were observed. composition, detailed lipidomic analysis of whole plasma Analysis of VLDL, IDL/LDL, and high-density lipoprotein and lipopoprotein fractions was undertaken using liquid revealed that although total levels of VLDL and IDL/LDL chromatography–tandem mass spectrometry. Confirming our were elevated in plasma (Figure 2), CsA treatment did not data from Figure 2, plasma cholesterol, CE, and triglyceride affect the amount of cholesterol, CE, or triglyceride per unit of

Table 2. Effect of CsA on Major Plasma Lipid Species in Ldlr−/− Mice

Plasma, nmol/mL VLDL, nmol/PC IDL/LDL, nmol/PC HDL, nmol/PC Cholesterol Control 1879±188 1422±156 1016±78 532±72 CsA 4297±573* 1188±58 1208±86 536±55 CE Control 5578±284 2912±477 3637±137 2826±321 CsA 11460±1074† 2814±208 3718±173 2789±280 DG Control 27±4 96±10 46±4 28±5 CsA 45±6‡ 48±6 26±3 24±4 TG Control 377±63 409±63 95±14 5±2 CsA 773±130‡ 286±41 80±11 11±5 PC Control 1818±62 … … … CsA 2262±59§ … … … CE indicates cholesteryl esters; CsA, cyclosporin A; DG, diacylglycerol; HDL, high-density lipoprotein; IDL, intermediate-density lipoprotein; LDL, low-density lipoprotein; PC, phosphatidylcholine; TG, triglyceride; and VLDL, very low–density lipoprotein. Data are presented as nmol/mL for plasma and expressed as nmol lipid per nmol PC content for VLDL, IDL/LDL, and HDL fractions and are mean±SEM from 7 to 8 mice per group. *P<0.01; †P<0.001; ‡P<0.05; §P<0.0001.

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 Kockx et al Mechanisms of Cyclosporin A–Induced Dyslipidemia 5 phosphatidylcholine (Table 2) or per unit of apoB (Table II in the newly synthesized particles. Taken together, these results the online-only Data Supplement), indicating that the choles- indicate that CsA-induced hyperlipidemia is not explained by terol, CE, or triglyceride content per lipoprotein particle was an increase in hepatic VLDL production or secretion. largely unaffected. Taken together, these results suggest that CsA increases VLDL-IDL/LDL particle numbers rather than CsA Inhibits Plasma Lipoprotein Lipase Activity affecting particle composition. and Clearance of VLDL-Like Particles Analysis of minor lipid classes detected by liquid chroma- CsA has been reported to decrease plasma LPL activity in tography–tandem mass spectrometry indicated that although humans and in rats,16,17 which may partly explain increased most lipid species within VLDL or IDL/LDL were unchanged plasma VLDL and IDL/LDL levels. We therefore determined after CsA treatment, sphingomyelin (SM) decreased relative LPL activity after CsA treatment in both Ldlr−/− and wt mice. to phosphatidylcholine within VLDL and IDL/LDL fractions CsA did not affect LPL activity in wt mice (Figure 4A); how- (Table 3), whereas LPC and PI increased in VLDL only. These ever, LPL activity was significantly lower in CsA-treated were not significantly different relative to apoB (Table III in Ldlr−/− mice (352±71 versus 185±17 mU/mL for control and the online-only Data Supplement). Further examination of SM CsA, respectively; P<0.05; Figure 4B). To determine whether species identified differences in response to CsA according to decreased LPL activity affected VLDL particle size, VLDL acyl chain length and saturation (Table IV in the online-only was isolated by density ultracentrifugation, and particle size Data Supplement). Specifically, SM 33:1, 34:1, 36:1, 36:2, was analyzed using dynamic light scattering. VLDL particle 42:1, 31:1, 35:1, 37:2, 41:1, and 41:2 decreased significantly size was 73.0±14.5 nm in control and 75.8±11.4 nm in CsA- in VLDL after CsA treatment. In IDL/LDL, 34:1, 36:1, 36:2, treated mice, suggesting that CsA treatment did not affect par- 42:1 31:1, and 35:1 were decreased after CsA treatment. SM ticle size (Figure IV in the online-only Data Supplement). 38:2 increased significantly in IDL/LDL, but not in VLDL. ApoC-III is an important endogenous inhibitor of LPL Taken together, these results suggest that the predominant activity, and previous reports have observed increased apoC- effect of CsA is to increase plasma VLDL and IDL/LDL par- III levels in transplant recipients treated with CsA.18,19 Western ticle number rather than lipoprotein particle composition. blot analysis indicated that CsA increased plasma apoC-III levels in Ldlr−/− mice but did not in wt mice (Figure 4C and CsA Does Not Affect Hepatic VLDL 4D). These results support the possibility that in Ldlr−/− mice, Synthesis or Secretion CsA-induced hyperlipidemia is at least in part related to inhi- As the lipidomic analysis suggested that CsA treatment bition of LPL activity, which may be mediated by increased increases VLDL-IDL/LDL particle number, we investigated apoC-III levels. whether CsA increases hepatic VLDL production and secre- To determine whether CsA attenuated the clearance of tion. CsA significantly increased hepatic mRNA levels of , the effect of CsA on in vivo clearance of tri[3H] several genes involved in lipid metabolism, specifically Sterol oleate (triolein, TO)- and [14C]cholesteryl oleate (CO)-double- regulatory element–binding transcription factor 1 and its tar- labeled VLDL-like particles was investigated. Previous studies get genes fatty acid synthase and Stearoyl-coA desaturase-1 have indicated that these particles show comparable plasma (Figure 3A). 3-Hydroxy-3-methyl-glutaryl-CoA reductase, kinetics and tissue uptake activity as endogenous triglyceride- the rate-limiting enzyme in cholesterol synthesis, and PCSK9 rich lipoproteins.20 Ldlr−/− mice showed delayed clearance and were also increased but did not reach significance (P=0.06). decreased hepatic uptake of particle-derived [14C]CO com- None of these mRNA species were upregulated by CsA in pared with wt mice (Figure 5; Figure V in the online-only Data wt mice (Figure III in the online-only Data Supplement). Supplement), which is in line with previous reports.21 CsA Although CsA increased hepatic mRNA levels of fatty acid did not affect plasma clearance of [3H]TO or [14C]CO in wt synthase and stearoyl-coA desaturase-1 in Ldlr−/− mice, it had mice (Figure 5A and 5B); however, it significantly decreased no effect on their respective hepatic protein levels (Figure 3B). clearance of [3H]TO (P<0.0001) and [14C]CO (P<0.0001) in In addition, protein levels of microsomal triglyceride transfer Ldlr−/− mice (Figure 5C and 5D). In the presence of CsA, the protein, the protein responsible for VLDL assembly, were also plasma half-life of TO was increased from 4.2 to 7.6 minutes unaffected by CsA treatment (Figure 3B). (P<0.001) and of CO from 3.7 to 7.6 minutes (P<0.001). Because an effect of CsA on microsomal triglyceride Tissue accumulation of [3H] and [14C] was also assessed. transfer protein activity could not be excluded, in vivo hepatic CsA significantly reduced tissue uptake of [3H]TO-derived triglyceride secretion rates were determined after injection [3H]oleate in the heart (Figure 5F) but not in other tissues in of Triton WR1339 to block lipolysis. Triglyceride secretion Ldlr−/− mice. Importantly, CsA decreased total accumulation rates were unaffected after CsA (4.2±0.6 mg/min for control of lipoprotein-derived [14C]CO (Figure 5H) into the liver by versus 3.8±0.5 mg/min for control versus CsA, respectively; 32±4% (P<0.05) supporting inhibition of particle clearance by Figure 3C). To obtain rates of secretion of newly synthesized the liver. In summary, these results indicate that CsA-induced apoB, [35S]-methionine was injected before blocking lipolysis. hyperlipidemia in the Ldlr−/− mouse is most likely mediated by The rate of secretion of newly synthesized VLDL-[35S]apoB decreased plasma clearance of lipoproteins. (Figure 3D) did not differ significantly (P=0.21) between con- trol and CsA-treated mice. In addition, the VLDL-triglyceride/ CsA Increases Plasma PCSK9 Levels in Ldlr−/− [35S]apoB ratio (Figure 3E) in these experiments was unaf- PCSK9 has been recently identified as a major contribu- fected by CsA (P=0.26), suggesting unchanged composition in tor to LDL clearance, especially via the LDLr, and is itself

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 6 Arterioscler Thromb Vasc Biol July 2016

Table 3. Lipidomics of VLDL, LDL, and HDL in Control vs CsA-Treated Ldlr−/− Mice

VLDL, nmol/PC IDL/LDL, nmol/PC HDL, nmol/PC Species in class SM Ctrl 184.8±5.3 170.9±2.2 124.7±3.1 19 CsA 155.8±5.7* 159.7±3.5* 121.4±4.1 Cer Ctrl 14.7±0.9 8.5−0.5 5.6±0.5 6 CsA 7.5±0.5 6.1±0.4 6.2±0.8 dhCer Ctrl 2.3±0.3 1.2±0.1 1.0±0.1 6 CsA 1.0±0.1 0.8±0.1 1.1±0.2 MHC Ctrl 17.6±0.6 16.8±1.0 5.0±0.3 6 CsA 15.1±1.2 15.1±1.3 5.8±0.5 DHC Ctrl 0.5±0.1 0.4±0.1 0.3±0.1 6 CsA 0.3±0.03 0.4±0.1 0.3±0.1 THC Ctrl 2.4±0.4 1.5±0.2 2.0±0.3 6 CsA 0.7±0.1 0.9±0.2 2.6±0.4 GM3 Ctrl 0.3±0.1 0.2±0.1 0.1±0.1 6 CsA 0.2±0.1 0.2±0.1 0.1±0.1 LPC Ctrl 50.9±11.0 50.5±9.8 229.8±43.5 21 CsA 65.6±13.5† 59.3±11.0 242.2±42.7 LPC Ctrl 0.8±0.1 0.6±0.1 2.5−0.4 1 (O) CsA 0.7±0.1 0.6±0.1 2.0−0.4 PC Ctrl … … … 32 CsA … … … PC (O) Ctrl 13.1±0.6 13.1±0.8 14.3±0.5 27 CsA 13.6±0.7 13.5±0.6 15.9±0.9 PC (P) Ctrl 4.1±0.5 4.2±0.2 3.9±0.6 10 CsA 4.1±0.1 4.1±0.2 4.0±0.3 LPE Ctrl 3.2±0.5 3.1±0.3 7.1±1.2 1 CsA 5.6±1.2 4.3±0.6 8.2±1.3 PE Ctrl 5.2±0.8 3.4±0.6 2.2±0.3 26 CsA 6.7±0.9 5.5±1.1 4.5±0.9 PE (O) Ctrl 2.8±0.3 2.8±0.5 1.7±0.2 11 CsA 2.0±0.1 2.0±0.1 2.0−0.2 PE (P) Ctrl 4.1±0.4 3.9±0.4 2.4−0.2 11 CsA 2.6±0.1 2.7±0.1 1.9±0.2 PG Ctrl 0.1±0.1 0.1±0.1 0.1±0.1 3 CsA 0.1±0.1 0.2±0.1 0.2±0.1 LPI Ctrl 1.5±0.5 3.0±0.4 58.3±9.7 1 CsA 4.0±0.8‡ 4.0±0.7 47.7±10.1 PI Ctrl 77.2±1.3 89.0±2.3 140.0±7.9 20 CsA 92.5±2.8‡ 98.0±3.5 157.5±6.9 PS Ctrl 0.2±0.1 0.1±0.1 0.1±0.1 6 CsA 0.1±0.1 0.1±0.1 0.2±0.1 Lipid levels were determined by LC/MSMS. Data are presented nmol per PC content and are mean±SEM from 7 to 8 mice per group. Not corrected PC content is 48 057±7133 (VLDL-control), 167 427±34 105 (VLDL-CsA), 67 992±5537 (LDL-control), 124 396±13 617 (LDL-CsA), 46 682±5787 (HDL-Ctrl), and 46 970±13 256 (HDL-CsA). Cer indicates ceramide; CsA, cyclosporin A; DHC, dihydrohexoceramide; dhCer, dihydroceramide; GM3, GM3 ganglioside; LC/MSMS, liquid chromatography–tandem mass spectrometry; LPC, lysophosphatidylcholine; LPC(O), lysoalkylphosphatidylcholine; LPE, lysophosphotidylethanolamine; LPI, lysophosphatidylinositol; MHC, monohexosylceramide; PC, phosphatidylcholine; PC(O), alkylphoshatidylcholine; PC(P), alkenylphosphatidylcholine; PE, phosphatidyalethanolamine; PE(O), alkylphosphatidylethanolamine; PE(P), alkenylphosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SM, shingomyelin; and THC, trihexosylceramide. *P<0.001; †P<0.01; ‡P<0.05; §P<0.0001.

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 Kockx et al Mechanisms of Cyclosporin A–Induced Dyslipidemia 7

Figure 3. Cyclosporin A (CsA) does not affect hepatic lipid production or hepatic very low–density lipoprotein (VLDL) secretion. Ldlr- knockout mice (Ldlr−/−) mice were treated with 20 mg/kg per d CsA or vehicle (Ctrl) for 4 weeks. Hepatic mRNA (A) and protein (B) levels of genes involved in lipid metabolism and hepatic VLDL (C) and (apoB) secretion rates (D) and VLDL-Trig/apoB ratios (E) were determined as described in Materials and Methods available in the online-only Data Supplement. Protein levels were corrected for α-tubulin levels to account for differences in total protein loaded per lane. Data shown are mean±SEM of at least 7 mice per group. *P<0.05; **P<0.0001. FAS indicates fatty acid synthase; MTP, microsomal triglyceride; and SCD1, stearoyl-CoA desaturase-1. elevated under conditions of Ldlr−/− deficiency.22 Importantly, Discussion it remains possible that PCSK9 has effects on lipoprotein We have demonstrated that CsA-induced dyslipidemia does metabolism additional to its effects on the LDLr expression.23 not require inhibition of the LDLr and does not require the We therefore investigated whether CsA affected PCSK9 levels presence of a high-fat Western diet. Our studies identify a in Ldlr−/− mice. We were unable to detect PCSK9 in plasma complex interaction among CsA, lipoprotein clearance path- from wt mice, with or without CsA exposure (Figure VI in the ways, and the LDLr such that LDLr deficiency seems to be online-only Data Supplement). In contrast, there was abun- permissive for CsA-induced hyperlipidemia. Treatment with dant PCSK9 in Ldlr−/− mice, and this increased after expo- CsA markedly increases plasma cholesterol and triglyc- sure to CsA (Figure 6A), with a strong correlation between eride levels in chow-fed Ldlr−/−, mice and this is associated blood CsA and PCSK9 levels in these mice (Figure 6B). There with a marked increase in plasma VLDL and IDL/LDL lev- was also a strong correlation between LDL-cholesterol and els, decreased lipolysis, hepatic lipoprotein clearance, and PCSK9 levels in Ldlr−/− mice treated with CsA (Figure 6C), increased plasma PCSK9 levels. whereas plasma PCSK9 only weakly correlated with LDL- The first key finding of our study was that CsA induced cholesterol in vehicle-treated mice (Figure 6D), suggesting a marked hyperlipidemia in the Ldlr−/− mice. Thus, although CsA concurrently acts via an LDLr-independent pathway to previous cellular studies suggested that CsA may modulate elevate PCSK9 and LDL-cholesterol in Ldlr−/− mice. Other LDLr activity levels in HepG2 cells,5,6 direct inhibition of receptors involved in lipoprotein clearance, some of which LDLr activity by CsA is clearly not essential for the devel- have been identified as PCSK9 targets,24,25 were investigated. opment of hyperlipidemia. The second important obser- CsA treatment did not affect hepatic levels of LRP1 or syn- vation is that we did not observe dyslipidemia in wt mice decan-1 nor levels of apoER2 or VLDLr in brain or heart, with functional LDLr fed a chow diet and exposed to CsA. respectively (Figure 6E–6H). This most likely indicates that LDLr-mediated lipoprotein

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 8 Arterioscler Thromb Vasc Biol July 2016

Figure 4. Cyclosporin A (CsA) decreases plasma lipoprotein lipase (LPL) activity in Ldlr-knockout mice (Ldlr−/−) mice. Wild-type (wt) and Ldlr−/− mice were treated with 20 mg/kg per d CsA or vehicle (Ctrl) for 4 weeks. Heparin-releasable lipase activity in plasma (A and B) and plasma apolipoprotein C-III (apoC-III) levels (C and D) were determined using a commercially available kit and Western blotting as described in the Materials and Methods available in the online-only Data Supplement. Equal volumes of plasma (1l) were loaded. Data shown are mean±SEM (n=7). Representative Western blots for n=4 mice are shown. *P<0.05. clearance plays a protective role against CsA-mediated Analysis of individual lipid species in VLDL and IDL/ hyperlipidemia. LDL fractions indicated that although absolute levels of lipid Many insights into the role of lipoprotein receptors have species in plasma were increased, lipid composition per par- arisen from animal models completely deficient in these ticle (per phosphatidylcholine or apoB) did not change for the receptors, even though complete deficiency is rare in humans vast majority of lipid species. These data suggest that the pre- (eg, apoE−/−, LDLr −/−, and SRB1−/−). On this basis, our results dominant effect of CsA was to increase plasma lipoprotein in LDLr−/− mice establish a proof of principle that the LDLr particle number without substantially affecting particle com- has an important role in determining whether CsA causes position. The 1 lipid species that did change in both VLDL hyperlipidemia. In addition, our findings also imply that and IDL/LDL was SM, which was decreased by 16% and 7% lower LDLr-mediated clearance activity, as observed in het- for VLDL and IDL/LDL, respectively (Tables 2 and 3; Tables erozygous familial hypercholesterolemia, may promote CsA- II and III in the online-only Data Supplement). Decreased SM mediated hyperlipidemia and that enhanced LDLr-mediated content of LDL can be associated with increased LDL aggre- clearance after exposure to statins and PCSK9 inhibitors, may gation and binding to proteoglycans, but this depends on the mitigate against the effects of CsA. action of shingomyelinase and the concomitant accumulation CsA markedly increased plasma VLDL and IDL/LDL of ceramides.28 Because the ceramide content of lipoproteins levels, whereas it did not affect plasma high-density lipo- in this study was not affected by CsA, it is unclear whether the protein levels, in agreement with observations of human small decrease in SM content would have any physiological CsA-induced hyperlipidemia.2 Because elevated VLDL is consequences. typical of insulin resistance, we investigated whether the The increase in VLDL and IDL/LDL caused by CsA hyperlipidemia in CsA-treated LDLr−/− mice was associated seems to be mediated by effects on clearance rather than with insulin resistance. We found that CsA-treated mice hepatic production or synthesis. CsA did not significantly were not insulin resistant but instead seemed to demon- increase the levels of proteins important in lipoprotein synthe- strate improved insulin sensitivity. Improved insulin sen- sis. In addition, VLDL secretion was directly measured and sitivity in response to CsA has been reported recently in found to be unchanged by CsA. In contrast, CsA inhibited obese high-fat–fed C57Bl/6 mice.26 However, caution must plasma clearance and hepatic uptake of VLDL-like particles be exercised in applying these findings to clinical studies containing both [3H]TO and [14C]CO in Ldlr−/− mice, indi- as immunosuppressive drugs such as CsA increase the risk cating that CsA-induced hyperlipidemia in a LDLr-deficient of developing diabetes mellitus, most likely via effects on environment is primarily a function of decreased hepatic insulin secretion.27 clearance. We identified at least 3 candidate mechanisms by

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 Kockx et al Mechanisms of Cyclosporin A–Induced Dyslipidemia 9

Figure 5. Cyclosporin A (CsA) inhibits plasma clearance and hepatic uptake of very low–density lipoprotein (VLDL)–like particles in Ldlr- knockout mice (Ldlr−/−)mice. Wild-type (Wt) and Ldlr−/− mice were treated with 20 mg/kg per d CsA or vehicle (Ctrl) for 4 weeks. Mice were injected with VLDL-like particles, double-labeled with glycerol [3H]TO and [14C]CO. [3H]- (A and B) and [14C]- (C and D) clearance from plasma as well as [3H] (E and F) and [14C] (G and H) uptake by organs and tissues were determined at 15 minutes after injection for wt and 30 minutes for Ldlr-/- mice. Data are expressed as percentage of the injected dose and for organs per gram tissue. GonWAT indi- cates gonadal white adipose tissue; and scWAT, subcutaneous white adipose tissue. Values are means±SEM (n=5–7 per group). *P<0.05; ***P<0.0001. which this occurs. Plasma LPL activity was inhibited, and this with lower triglyceride levels and that antisense therapy to was associated with a concomitant increase in its inhibitory apoC-III decreases plasma triglyceride levels in humans con- cofactor apoC-III. Effects on lipases and apoC-III after CsA firming mechanistic significance.30 ApoC-III is an inhibitor treatment have been observed in humans as well as in ani- of LPL and also inhibits lipoprotein clearance by blocking mal models17,18 but not in all previous studies.29 Recent studies the engagement of apoE and apoB to their receptors.31 We have shown that genetic deficiency of apoC-III is associated observed that CsA did not increase apoC-III levels in wt mice,

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 10 Arterioscler Thromb Vasc Biol July 2016

Figure 6. Cyclosporin A (CsA) increases plasma proprotein convertase subtilisin/kexin type 9 (PCSK9) levels in Ldlr-knockout mice (Ldlr−/−) mice. Mice were treated with 20 mg/kg per d CsA or vehicle (Ctrl) for 4 weeks. Plasma PSCK9 levels (A) were determined by Western blotting. Equal volumes (1l) of plasma were separated. Correlations between plasma PCSK9 and blood CsA levels (B), PCSK9 and plasma low-density lipoprotein (LDL)-cholesterol levels in CsA-treated (C) and vehicle-treated Ctrl mice (D) were determined using linear regression analysis. Hepatic LRP1 (E), hepatic syndecan-1 (F), brain apoER2 (G), and heart VLDLr (H) protein levels were deter- mined by Western blotting. Protein levels were corrected for α-tubulin levels and Ctrl values were set at 1. Bar graph shown is mean±SEM (n=7–9). A representative Western blot for 4 to 5 mice is shown. *P<0.05. suggesting that the CsA-induced increase in apoC-III levels LPL activity was inhibited by CsA is functionally important requires impaired LDLr activity and presumably impaired in this animal model because VLDL-triglyceride content and lipoprotein clearance. It is unclear whether the extent to which particle size were unchanged. In humans heterozygous for

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 Kockx et al Mechanisms of Cyclosporin A–Induced Dyslipidemia 11

LPL deficiency, VLDL particle size was not affected,32 imply- Rather, the absence of the LDLr plays an important permis- ing that partial LPL deficiency does not necessarily affect sive role for CsA-induced hyperlipidemia, which is related to VLDL particle size. abnormal lipoprotein clearance, LPL activity, and increased We also made the novel observation of increased levels apoC-III and PCSK9. Enhancing LDLr and LPL activity and of plasma PCSK9 in Ldlr−/− mice treated with CsA, with a decreasing apoC-III and PCSK9 levels may provide attractive striking correlation between blood CsA and PCSK9 levels. treatment targets for transplant patients receiving CsA. Elevated levels of PCSK9 in Ldlr−/− mice have been reported and are consistent with high circulating levels of PCSK9 Acknowledgments in humans with LDLr deficiency.22,33 That CsA increases We thank the South Eastern Area Laboratory Service of the Prince PCSK9 levels in the absence of functional LDLr is intriguing. of Wales Hospital for performing blood CsA measurements, the Other potential PCSK9 targets such as the VLDLr,24 LRP125, Key Centre for Polymers and Colloids, and the Molecular Biology 24 Facility both at the University of Sydney for assistance in particle apoER2 , or syndecan-1 levels were not affected by CsA, size determination and Dr Blake Cochran from the University of New implying the presence of a yet unrecognized clearance path- South Wales for reviewing insulin resistance data. ways for PCSK9 which are modulated by CsA. The absence of any correlation between PCSK9 and LDL-C in vehicle-treated Sources of Funding mice emphasizes the mechanistic link among CsA, eleva- This work was supported by the National Heart Foundation (grant tion of PCSK9, and elevation of LDL levels. Interestingly, in aid G10S5192) and the National Health and Medical Research PCSK9 was predicted to be a direct binding partner for CsA Council of Australia (program grant 482820). P.C.N. Rensen is an in a human structural proteome-wide characterization of CsA Established Investigator of the Dutch Heart Foundation (2009T038). targets.34 Recently, induction of a gain-of-function mutant of PCSK9 increased VLDL-triglyceride and IDL/LDL levels in Disclosures Ldlr−/− mice35 and supports the possibility of elevated PCSK9 None. levels contributing to CsA-induced hyperlipidemia in our Ldlr−/− mice. References Blood CsA levels achieved in our study were 711±91 ng/ 1. Kockx M, Jessup W, Kritharides L. Cyclosporin A and atherosclerosis– cellular pathways in atherogenesis. Pharmacol Ther. 2010;128:106–118. mL after 4 weeks of treatment which are slightly higher than doi: 10.1016/j.pharmthera.2010.06.001. levels achieved in humans (100–400 ng/mL)36 but are very 2. Kobashigawa JA, Kasiske BL. Hyperlipidemia in solid organ transplanta- consistent with other animal studies, which report CsA lev- tion. Transplantation. 1997;63:331–338. els of ≈1300 ng/mL at administered dose of 50 mg/kg per d37 3. Miller LW. Cardiovascular toxicities of immunosuppressive agents. Am J Transplant. 2002;2:807–818. 14,38 and between 360 and 1066 ng/mL CsA at 20 mg/kg per d. 4. Kockx M, Kritharides L. Cyclosporin A-induced hyperlipidemia. 2012. In: Published literature indicates that blood CsA levels can cor- Gerhard K, ed. Lipoproteins- Role in Health and Diseases. 2012:337–354. relate with plasma lipid levels13,39 and may be relevant to some 5. Rayyes OA, Wallmark A, Florén CH. Cyclosporine inhibits catabolism of low-density lipoproteins in HepG2 cells by about 25%. Hepatology. of our findings. 1996;24:613–619. doi: 10.1002/hep.510240325. Although previous animal studies of CsA-induced 6. Winegar DA, Salisbury JA, Sundseth SS, Hawke RL. Effects of cyclospo- hyperlipidemia have provided contradictory mechanistic rin on cholesterol 27-hydroxylation and LDL receptor activity in HepG2 cells. J Lipid Res. 1996;37:179–191. conclusions, several previous studies share the observation 7. Wu J, Zhu YH, Patel SB. Cyclosporin-induced dyslipoproteinemia is asso- that the effects of CsA are more prominent under condi- ciated with selective activation of SREBP-2. Am J Physiol. 1999;277(6 pt tions which either interfere with normal lipid homeostasis 1):E1087–E1094. or confer baseline hyperlipidemia. Matsumoto et al13 found 8. Vaziri ND, Liang K, Azad H. Effect of cyclosporine on HMG-CoA reduc- tase, cholesterol 7alpha-hydroxylase, LDL receptor, HDL receptor, VLDL that CsA increased plasma lipid levels in hyperlipidemic receptor, and lipoprotein lipase expressions. J Pharmacol Exp Ther. apoE-knockout mice and Emeson et al40 and Ragheb et al41 2000;294:778–783. reported a strong dyslipidemia in C57Bl/6 mice and New 9. López-Miranda J, Vilella E, Pérez-Jiménez F, Espino A, Jiménez- Perepérez JA, Masana L, Turner PR. Low-density lipoprotein metabolism Zealand White rabbits treated with CsA only when they were in rats treated with cyclosporine. Metabolism. 1993;42:678–683. fed a high-fat diet. Taken together with our own findings, 10. De Lima JJ, Latrilha Mda C, Toffoletto O, Ianhez LE, Krieger EM, it is likely that CsA perturbs multiple pathways relevant Maranhão RC. Plasma kinetics of chylomicron-like emulsion in renal to lipoprotein metabolism and that the residual capacity of transplant patients receiving cyclosporin-based immunosuppression. Clin Cardiol. 1998;21:411–413. lipoprotein clearance pathways determines the final altera- 11. Hoogeveen RC, Ballantyne CM, Pownall HJ, Opekun AR, Hachey DL, tions in plasma lipoproteins after exposure to CsA. We Jaffe JS, Oppermann S, Kahan BD, Morrisett JD. Effect of sirolimus on interpret the LDLr as providing important capacity for lipo- the metabolism of apoB100- containing lipoproteins in renal transplant patients. Transplantation. 2001;72:1244–1250. protein clearance which compensates for the effects of CsA 12. Ishida O, Ochi M, Miyamoto Y, Kuta Y, Akiyama M. Suppression by on other pathways. This has important clinical implications cyclosporine of cellular and humoral reactivity after peripheral nerve because it implies that simultaneously targeting both permis- allografts in mice. Transplantation. 1989;48:824–829. 13. Matsumoto Y, Hof A, Baumlin Y, Hof RP. Differential effect of cyclospo- sive pathways such as the LDLr (eg, with statins and PCSK9 rine A and SDZ RAD on neointima formation of carotid allografts in apo- inhibitors) and direct pathways (eg, LPL, PCSK9, apoC-III, lipoprotein E-deficient mice. Transplantation. 2003;76:1166–1170. doi: with direct biological therapies) may mitigate CsA-induced 10.1097/01.TP.0000090393.75600.32. hyperlipidemia in patients. 14. Hippert C, Dubois G, Morin C, Disson O, Ibanes S, Jacquet C, Schwendener R, Antignac C, Kremer EJ, Kalatzis V. Gene transfer may be In conclusion, we have demonstrated that CsA does not preventive but not curative for a lysosomal transport disorder. Mol Ther. cause hyperlipidemia via direct effects on the LDL receptor. 2008;16:1372–1381. doi: 10.1038/mt.2008.126.

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 12 Arterioscler Thromb Vasc Biol July 2016

15. Bieghs V, Van Gorp PJ, Wouters K, Hendrikx T, Gijbels MJ, van Bilsen M, only aggregation, of low density lipoprotein (LDL) particles. Two distinct Bakker J, Binder CJ, Lütjohann D, Staels B, Hofker MH, Shiri-Sverdlov mechanisms leading to increased binding strength of LDL to human aortic R. LDL receptor knock-out mice are a physiological model particularly proteoglycans. J Biol Chem. 1998;273:29127–29134. vulnerable to study the onset of inflammation in non-alcoholic fatty liver 29. Ichimaru N, Takahara S, Kokado Y, Wang JD, Hatori M, Kameoka H, disease. PLoS One. 2012;7:e30668. doi: 10.1371/journal.pone.0030668. Inoue T, Okuyama A. Changes in lipid metabolism and effect of simvas- 16. Espino A, Lopez-Miranda J, Blanco-Cerrada J, Zambrana JL, Aumente tatin in renal transplant recipients induced by cyclosporine or tacrolimus. MA, Paniagua JA, Blanco-Molina A, Jimenez-Pereperez JA, Rodriguez Atherosclerosis. 2001;158:417–423. M, Pérez-Jiméz F. The effect of cyclosporine and methylprednisolone on 30. Gaudet D, Alexander VJ, Baker BF, Brisson D, Tremblay K, Singleton plasma lipoprotein levels in rats. J Lab Clin Med. 1995;125:222–227. W, Geary RS, Hughes SG, Viney NJ, Graham MJ, Crooke RM, Witztum 17. Derfler K, Hayde M, Heinz G, Hirschl MM, Steger G, Hauser AC, Balcke JL, Brunzell JD, Kastelein JJ. Antisense inhibition of apolipoprotein C-III P, Widhalm K. Decreased postheparin lipolytic activity in renal transplant in patients with hypertriglyceridemia. N Engl J Med. 2015;373:438–447. recipients with cyclosporin A. Kidney Int. 1991;40:720–727. doi: 10.1056/NEJMoa1400283. 18. Tur MD, Garrigue V, Vela C, Dupuy AM, Descomps B, Cristol JP, Mourad 31. Sacks FM. The crucial roles of E and C-III in apoB lipo- G. Apolipoprotein CIII is upregulated by anticalcineurins and rapamycin: protein metabolism in normolipidemia and hypertriglyceridemia. Curr implications in transplantation-induced dyslipidemia. Transplant Proc. Opin Lipidol. 2015;26:56–63. doi: 10.1097/MOL.0000000000000146. 2000;32:2783–2784. 32. Blackett P, Tryggestad J, Krishnan S, Li S, Xu W, Alaupovic P, Quiroga C, 19. Kimak E, Solski J, Baranowicz-Gaszczyk I, Ksiazek A. A long-term study Copeland K. Lipoprotein abnormalities in compound heterozygous lipo- of dyslipidemia and dyslipoproteinemia in stable post-renal transplant protein lipase deficiency after treatment with a low-fat diet and orlistat. J patients. Ren Fail. 2006;28:483–486. doi: 10.1080/08860220600778878. Clin Lipidol. 2013;7:132–139. doi: 10.1016/j.jacl.2012.11.006. 20. Berbée JF, Boon MR, Khedoe PP, et al. Brown fat activation reduces 33. Raal F, Panz V, Immelman A, Pilcher G. Elevated PCSK9 levels in hypercholesterolaemia and protects from atherosclerosis development. untreated patients with heterozygous or homozygous familial hypercho- Nat Commun. 2015;6:6356. doi: 10.1038/ncomms7356. lesterolemia and the response to high-dose statin therapy. J Am Heart 21. Rensen PC, Herijgers N, Netscher MH, Meskers SC, van Eck M, van Assoc. 2013;2:e000028. doi: 10.1161/JAHA.112.000028. Berkel TJ. Particle size determines the specificity of apolipoprotein 34. Hu G, Wang K, Groenendyk J, Barakat K, Mizianty MJ, Ruan J, Michalak E-containing triglyceride-rich emulsions for the LDL receptor versus M, Kurgan L. Human structural proteome-wide characterization of cyclo- hepatic remnant receptor in vivo. J Lipid Res. 1997;38:1070–1084. sporine A targets. Bioinformatics. 2014;30:3561–3566. doi: 10.1093/ 22. Tavori H, Rashid S, Fazio S. On the function and homeostasis of PCSK9: bioinformatics/btu581. reciprocal interaction with LDLR and additional lipid effects. Atherosclerosis. 35. Roche-Molina M, Sanz-Rosa D, Cruz FM, García-Prieto J, López S, Abia 2015;238:264–270. doi: 10.1016/j.atherosclerosis.2014.12.017. R, Muriana FJ, Fuster V, Ibáñez B, Bernal JA. Induction of sustained 23. Rashid S, Tavori H, Brown PE, Linton MF, He J, Giunzioni I, Fazio hypercholesterolemia by single adeno-associated virus-mediated gene S. Proprotein convertase subtilisin kexin type 9 promotes intesti- transfer of mutant hPCSK9. Arterioscler Thromb Vasc Biol. 2015;35:50– nal overproduction of triglyceride-rich apolipoprotein B lipoproteins 59. doi: 10.1161/ATVBAHA.114.303617. through both low-density lipoprotein receptor-dependent and -inde- 36. Wong SH. Therapeutic drug monitoring for immunosuppressants. Clin pendent mechanisms. Circulation. 2014;130:431–441. doi: 10.1161/ Chim Acta. 2001;313:241–253. CIRCULATIONAHA.113.006720. 37. Vessie EL, Hirsch GM, Lee TD. Aortic allograft vasculopathy is medi- 24. Poirier S, Mayer G, Benjannet S, Bergeron E, Marcinkiewicz J, Nassoury ated by CD8(+) T cells in cyclosporin A immunosuppressed mice. Transpl N, Mayer H, Nimpf J, Prat A, Seidah NG. The proprotein convertase Immunol. 2005;15:35–44. doi: 10.1016/j.trim.2005.02.006. PCSK9 induces the degradation of low density lipoprotein receptor 38. Wranning CA, El-Akouri RR, Groth K, Mölne J, Parra AK, Brännström (LDLR) and its closest family members VLDLR and ApoER2. J Biol M. Rejection of the transplanted uterus is suppressed by cyclosporine A Chem. 2008;283:2363–2372. doi: 10.1074/jbc.M708098200. in a semi-allogeneic mouse model. Hum Reprod. 2007;22:372–379. doi: 25. Canuel M, Sun X, Asselin MC, Paramithiotis E, Prat A, Seidah NG. 10.1093/humrep/del410. Proprotein convertase subtilisin/kexin type 9 (PCSK9) can mediate deg- 39. Kuster GM, Drexel H, Bleisch JA, Rentsch K, Pei P, Binswanger U, radation of the low density lipoprotein receptor-related protein 1 (LRP-1). Amann FW. Relation of cyclosporine blood levels to adverse effects on PLoS One. 2013;8:e64145. doi: 10.1371/journal.pone.0064145. lipoproteins. Transplantation. 1994;57:1479–1483. 26. Yan L, Song K, Gao M, Qu S, Liu D. Modulation of cell-mediated immu- 40. Emeson EE, Shen ML. Accelerated atherosclerosis in hyperlip- nity to suppress high fat diet-induced obesity and insulin resistance. idemic C57BL/6 mice treated with cyclosporin A. Am J Pathol. Pharm Res. 2016;33:395–403. doi: 10.1007/s11095-015-1797-9. 1993;142:1906–1915. 27. Penfornis A, Kury-Paulin S. Immunosuppressive drug-induced diabetes. 41. Ragheb A, Attia A, Elbarbry F, Prasad K, Shoker A. Attenuated combined Diabetes Metab. 2006;32(5 pt 2):539–546. action of cyclosporine a and hyperlipidemia on atherogenesis in rabbits by 28. Oörni K, Hakala JK, Annila A, Ala-Korpela M, Kovanen PT. thymoquinone. Evid Based Complement Alternat Med. 2011;2011:620319. Sphingomyelinase induces aggregation and fusion, but phospholipase A2 doi: 10.1093/ecam/nep225.

Highlights Cyclosporin A (CsA) is an immunosuppressant commonly used to prevent organ rejection but is associated with hyperlipidemia and an in- creased risk of cardiovascular disease. Although studies suggest that CsA-induced hyperlipidemia is mediated by inhibition of low-density lipoprotein receptor, the data supporting this are inconclusive. We investigated the mechanisms mediating CsA-induced dyslipidemia and show that • The low-density lipoprotein receptor is permissive for CsA-induced dyslipidemia • Is mediated by inhibited lipoprotein clearance • Is associated with inhibited plasma lipoprotein lipase activity and increased apolipoprotein C-III and proprotein convertase subtilisin/kexin type 9 levels • Our studies suggest that the low-density lipoprotein receptor, lipoprotein lipase, apolipoprotein C-III, and proprotein convertase subtilisin/kexin type 9 may provide attractive treatment targets to prevent heart disease in patients receiving CsA.

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016 Low-Density Lipoprotein Receptor−Dependent and Low-Density Lipoprotein Receptor− Independent Mechanisms of Cyclosporin A−Induced Dyslipidemia Maaike Kockx, Elias Glaros, Betty Kan, Theodore W. Ng, Jimmy F.P. Berbée, Virginie Deswaerte, Diana Nawara, Carmel Quinn, Kerry-Anne Rye, Wendy Jessup, Patrick C.N. Rensen, Peter J. Meikle and Leonard Kritharides

Arterioscler Thromb Vasc Biol. published online May 5, 2016; Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 2016 American Heart Association, Inc. All rights reserved. Print ISSN: 1079-5642. Online ISSN: 1524-4636

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://atvb.ahajournals.org/content/early/2016/05/05/ATVBAHA.115.307030

Data Supplement (unedited) at: http://atvb.ahajournals.org/content/suppl/2016/05/04/ATVBAHA.115.307030.DC1.html http://atvb.ahajournals.org/content/suppl/2016/05/04/ATVBAHA.115.307030.DC2.html

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Arteriosclerosis, Thrombosis, and Vascular Biology can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at: http://www.lww.com/reprints

Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular Biology is online at: http://atvb.ahajournals.org//subscriptions/

Downloaded from http://atvb.ahajournals.org/ at University of Sydney on May 11, 2016

Supplement Table I: primers for Real Time RT-PCR

Gene Foward Reverse

Hmgcoar AGCTTGCCCGAATTGTATGTG TCTGTTGTAACCATGTGACTTC

Cyp7α GGGATTGCTGTGGTAGTGAGC GGTATGGAATCAACCCGTTGTC

Cyp27a1 GTGGTCTTATTGGGTACTTGC GGTATGGAATCAACCCGTTGTC

Aco GCCCAACTGTGACTTCCATT GGCATGTAACCCGTAGCACT

Agpat1 CACCCAGGATGTGAGAGTCTG CTGACAACGTCCAGGCGAGG

Sqle CGCGCAGCGGTTACTCTGGT AGTCACGGCCGGGAACCTGT

Mtp TGCGGGTCAACAGAGAGGCG CCTTGACCTTCCCCCGGACCA

Gpat4 GGCAGAGGAGCTGGAGTC TGTTGTGGTACGTAATGATGG

Dgat2 AGTGGCAATGCTATCATCATCGT AAGGAATAAGTGGGAACCAGATCA

Fas GCTGCGGAAACTTCAGGAAAT AGAGACGTGTCACTCCTGGACTT

Hmgcs GCCGTGAACTGGGTCGAA GCATATATAGCAATGTCTCCTGCAA

Scd1 CCGGAGACCCCTTAGATCGA TAGCCTGTAAAAGATTTCTGCAACC

Lipin1 GGTCCCCCAGCCCCAGTCCTT GCAGCCTGTGGCAATTCA

Lipin2 AGTTGACCCCATCACCGTAG CCCAAAGCATCAGACTTGGT

Srebp1 AGCAGCCCCTAGAACAAACAC CAGCAGTGAGTCTGCCTTGAT

Srebp2 TGAAGGACTTAGTCATGGGGAC CGCAGCTTGTGATTGACCT

ApoCIII AGGTACGTAGGTGCCATGC CGTCTTGGAGGCTTGTTCCA

Pcsk9 CAGAGGTCATCACAGTCGGG GGGGCAAAGAGATCCACACA

Hrpt TGACACTGGTAAAACAATGC AACACTTCGAGAGGTCCTTT

Hmgcoar, 3-hydroxy-3-methylglutaryl coenzyme A reductase; Cyp7α, cholesterol 7alpha-hydroxylase; Cyp27a1, sterol 27-hydroxylase; Aco, Acyl coA carboxylase; Agpat1, 1-acylglycerol-3-phosphate O-acyltransferase; Sqle, Squalene monooxygenase; Mtp, Microsomal Triglyceride Transfer Protein; Gpat, Glycerol-3- phosphate acyltransferase; Dgat, Diacylglycerol acyltransferase; Fas, Fatty acid synthase; Hmgcs, HMGcoA synthase; Scd1, Stearoyl-coA desaturase-1; Srebp, Sterol regulatory element binding protein; Apo, apolipoprotein; Pcsk9, Proprotein Convertase Subtilisin/kexin type 9 ;Hrpt, Hypoxanthine guanine phosphoribosyl transferase.

Supplement Table II: Effect of CsA on major plasma lipid species in Ldlr-/- mice

Plasma VLDL IDL/LDL (nmol/mL) (nmol/apoB) (nmol/apoB) Chol Ctrl 1879 ± 188 350 ± 98 67 ± 13 CsA 4297 ± 573* 275 ± 54 120 ± 28

CE Ctrl 5578 ± 284 669 ± 168 237 ± 42 CsA 11460 ± 1074** 630 ± 113 364 ± 77

DG Ctrl 27 ± 4 26 ± 9 3 ± 1 CsA 45 ± 6# 11 ± 3 3 ± 1

TG Ctrl 377 ± 63 111 ± 37 7 ± 2 CsA 773 ± 130# 69 ± 18 8 ± 3

Ctrl 1818 ± 62 262 ± 73 66 ± 12 PC CsA 2262 ± 59*** 233 ± 48 100 ± 23

Chol, cholesterol; CE, cholesteryl esters; DG, diacylglycerol; TG, triglyceride; PC, phosphatidyl choline. Lipid levels were determined by LC/MSMS. Data are presented as nmol/mL for plasma and expressed nmol lipid per apoB content for VLDL and IDL/LDL fractions. ApoB content of VLDL and IDL/LDL particles was determined by Western blotting. Data is mean ± sem from 7- 8 mice per group. # p<0.05, *p<0.01, **p<0.001, ***p<0.0001 v ctrl.

Supplemental Table III: Lipidomics of VLDL and LDL in Ctrl v CsA-treated Ldlr-/- mice

VLDL IDL/LDL (nmol/apoB) (nmol/apoB)

Sphingomyelin Ctrl 49.3 ± 14.4 11.1 ± 1.9 CsA 36.0 ± 6.9 15.8 ± 3.5 Ceramide Ctrl 3.9 ± 1.2 0.6 ± 0.1 CsA 1.7 ± 0.3 0.6 ± 0.1 Dihydroceramide Ctrl 0.7 ± 0.2 0.08 ± 0.02 CsA 0.2 ± 0.1 0.08 ± 0.02 Monohexosylceramide Ctrl 4.7 ± 1.3 1.1 ± 0.2 CsA 3.7 ± 0.8 1.5 ± 0.4 Dihydrohexoceramide Ctrl 0.1 ± 0.02 0.03 ± 0.01 CsA 0.1 ± 0.01 0.03 ± 0.01 Trihexosylceramide Ctrl 0.5 ± 0.1 0.1 ± 0.03 CsA 0.1 ± 0.02 0.08 ± 0.03 GM3 ganglioside Ctrl 0.1 ± 0.05 0.03 ± 0.01 CsA 0.1 ± 0.02 0.02 ± 0.01 Lysophosphatidylcholine Ctrl 13.8 ± 4.8 3.2 ± 0.9 CsA 16.7 ± 4.6 5.1 ± 1.4 Lysoalkylphosphatyidylcholine Ctrl 0.2 ± 0.1 0.04 ± 0.01 CsA 0.2 ± 0.05 0.05 ± 0.01 Phosphatidylcholine Ctrl 261.7 ± 73.0 65.5 ± 11.5 CsA 233.1 ± 47.6 100.3 ± 22.5 Alkylphosphatidylcholine Ctrl 3.3 ± 0.9 0.9 ± 0.2 CsA 3.2 ± 0.7 1.3 ± 0.3 Alkenylphosphatidylcholine Ctrl 1.0 ± 0.2 0.3 ± 0.1 CsA 1.0 ± 0.2 0.4 ± 0.1 Lysophosphatidylethanolamine Ctrl 0.8 ± 0.2 0.2 ± 0.03 CsA 1.6 ± 0.6 0.4 ± 0.1 Phosphatidylethanolamine Ctrl 1.3 ± 0.5 0.2 ± 0.1 CsA 1.5 ± 0.3 0.6 ± 0.2 Alkylphosphatidylethanolamine Ctrl 0.7 ± 0.2 0.2 ± 0.1 CsA 0.5 ± 0.1 0.2 ± 0.04 Alkenylphosphatidylethanolamine Ctrl 1.0 ± 0.3 0.3 ± 0.1 CsA 0.6 ± 0.1 0.3 ± 0.1 Phosphatidylglycerol Ctrl 0.07 ± 0.04 0.01 ± 0.01 CsA 0.03 ± 0.01 0.02 ± 0.01 Lysophosphatidylinositol Ctrl 0.4 ± 0.1 0.2 ± 0.04 CsA 1.1 ± 0.4 0.4 ± 0.1 Phosphatidylinositol Ctrl 20.3 ± 5.7 5.9 ± 1.1 CsA 21.8 ± 4.6 9.5 ± 1.9 Phosphatidylserine Ctrl 0.04 ± 0.02 0.03 ± 0.01 CsA 0.02 ± 0.08 0.07 ± 0.02

Lipid levels were determined by LC/MSMS. Data are presented relative to apoB content and represent mean ± sem from 7-8 mice per group.

Supplement Table IV: Lipidomics of SM species in VLDL and LDL in Ctrl v CsA-treated Ldlr-/- mice

VLDL IDL/LDL (nmol/PC) (nmol/PC)

SM 32:1 Ctrl 657.5 ± 74.9 575.3 ± 79.6 CsA 479.9 ± 56.5 561.8 ± 53.4 SM 33:1 Ctrl 3474.3 ± 202.6 2276.8 ± 322.0 CsA 2097.6 ± 209.3** 1965.4 ± 136.0 SM 34:1 Ctrl 42719.1 ± 2793.5 36562.6 ± 1603.0 CsA 31134.2 ± 2592.4# 31208.1 ± 1312.6# SM 34:2 Ctrl 3398.4 ± 339.3 3214.5 ± 211.9 CsA 26854 ± 206.6 2791.8 ± 184.7 SM 34:3 Ctrl 7.5 ± 4.3 10.9 ± 4.7 CsA 7.0 ± 2.5 7.6 ± 1.7 SM 36:1 Ctrl 5373.5 ± 560.3 4838.3 ± 241.0 CsA 2914.6 ± 212.9* 3176.8 ± 166.5** SM 36:2 Ctrl 1719.6 ± 173.7 1535.4 ± 150.7 CsA 1014.0 ± 152.9# 953.0 ± 97.1* SM 36:3 Ctrl 137.8 ± 17.2 134.3 ± 16.6 CsA 126.2 ± 11.8 102.4 ± 6.2 SM 38:1 Ctrl 103761.5 ± 3024.1 105263.6 ± 785.3 CsA 98741.1 ± 4049.8 102288.4 ± 1929.7 SM 38:2 Ctrl 1534.2 ± 198.5 720.8 ± 117.7 CsA 1324.9 ± 216.8 1766.1 ± 229.2* SM 42:1 Ctrl 8173.2 ± 554.1 7608.6 ± 497.7 CsA 6344.5 ± 120.9* 5701.6 ± 262.2* SM 31:1 Ctrl 322.5 ± 56.7 217.1 ± 16.6 CsA 104.9 ± 13.5* 127.6 ± 14.7* SM 33:1 Ctrl 42.7 ± 3.5 31.4 ± 4.2 CsA 26.1 ± 2.1* 29.2 ± 4.7 SM 35:1 Ctrl 1242.0 ± 97.0 906.3 ± 65.0 CsA 679.1 ± 45.2** 617.0 ± 55.4* SM 35:2 Ctrl 98.6 ± 16.3 86.5 ± 13.5 CsA 85.2 ± 6.5 78.4 ± 9.5 SM 37:2 Ctrl 88.6 ± 16.6 70.5 ± 12.3 CsA 33.6 ± 4.7* 49.1 ± 10.9 SM 39:1 Ctrl 865.6 ± 171.2 660.7 ± 154.5 CsA 754.2 ± 51.4 650.7 ± 114.8 SM 41:1 Ctrl 6384.6 ± 360.9 5089.0 ± 260.4 CsA 4074.7 ± 166.3*** 4258.8 ± 292.2 SM 41:2 Ctrl 4760.4 ± 294.2 3808.2 ± 316.0 CsA 3151.7 ± 347.3* 3351.4 ± 254.0

Levels of SM species were determined by LC/MSMS. Data is presented nmol per PC content and is mean ± sem from 7-8 mice per group. #p<0.05, *p<0.01, ** p<0.001, ***p<0.0001.

Figure I

A wt Ldlr-/-

Ctrl

CsA

B C wt Ldlr-/- wt Ldlr-/-

Ctrl

CsA

Figure I: CsA does not induce liver or kidney toxicity. Wt and Ldlr -/- mice were treated with 20mg/kg/day CsA or vehicle (Ctrl) for 4 weeks via subcutaneous implanted Alzet Osmotic minipumps. Formalin-perfused livers (A) and kidneys (B) were paraffin embedded and stained with hematoxylin/eosin. Kidney sections were also stained with Periodic acid-Schiff stain (C). Tissue sections shown were viewed using a 20x objective. Scale bar = 200µm. No significant difference was Figure II

A B

15 Ctrl 1000 e ) s CsA n ) o i 800 ** c e L m / s l u 0 l

10 o o 2 g c 600 1 m

u d l x m

o g (

L o

/ 400 l l 5 C B o U A m 200 m ( 0 0 0 15 30 45 60 75 90 105 120 Ctrl CsA Time (min)

Figure II: CsA does not induce insulin resistance in LDLr-/- mice. LDLr-/- mice were treated with 20mg/kg/day CsA or vehicle (Ctrl) for 4 weeks. Non-fasted insulin tolerance tests were performed as described in material and methods. (A) Blood glucose levels in response to i.p. injection of insulin and (B) Glucose area under the curve (AUC). Values are mean ± S.E.M. (n=10 per group). **p<0.01.

Figure III

3 Ctrl CsA

2

1 (fold induction) (fold Hepatic mRNAlevel Hepatic 0

Fas Scd1 Srebp1 Srebp2 Hmgcoar

Figure III: CsA does not affect hepatic mRNA levels of genes involved in lipid synthesis. Wt mice were treated with 20mg/kg/day CsA or vehicle (Ctrl) for 4 weeks. Hepatic mRNA levels were determined by qPCR as described in Materials and Methods.

Figure IV

120 e z i

s 100

e l

c ) 80 i t m r n a

( 60 p

L 40 D L

V 20 0 Ctrl CsA

Figure IV: CsA does not affect VLDL particle size in LDLr-/- mice. LDLr-/- mice were treated with 20mg/kg/day CsA or vehicle (Ctrl) for 4 weeks. VLDL was isolated by density ultracentrifugation and VLDL particle size was determined using dynamic light scattering as described in Material and Methods. Figure V

wt Ldlr-/-

20 20 Ctrl Ctrl CsA CsA 15 15

10 10 H] activity H] H] activity H] 3 3 [ 5 [ 5 * (% of injected dose) 0 (% of injected dose) 0

Liver Heart Liver Heart SpleenKidney ScWATMuscle SpleenKidney ScWATMuscle GonWAT GonWAT SolMuscle SolMuscle

wt Ldlr-/-

100 100 Ctrl Ctrl CsA 80 CsA 80 60 40 60 * C] activity C] C] activity C] 14 14 1

3 [ [ 2 (% of injected dose)

(% of injected dose) 1 0 0

Liver Heart Liver Heart SpleenKidney ScWATMuscle SpleenKidney ScWATMuscle GonWAT GonWAT SolMuscle SolMuscle

Figure V: CsA inhibits hepatic uptake of lipoprotein derived cholesteryl oleate in LDLr-/- mice. Wt and LDLr-/- mice were treated with 20 mg/kg/day CsA or cremaphore vehicle (Ctrl) for 4 weeks. Mice were injected with VLDL-like particles, double-labeled with [3H]TO and [14C]CO. 3H- and 14C- uptake by organs and tissues were determined at 15 min for wt and 30 min for LDLr-/- after injection. Data are expressed as percentage 3H- or 14C-activity of the injected dose per total tissue, with the exception of kidney, white adipose tissue pads and muscle which were determined as uptake in 1 representative tissue. GonWAT, gonadal white adipose tissue, scWAT, subcutaneous white adipose tissue. Values are mean ± S.E.M. (n=5-7 per group). *p<0.05.

Figure VI

Figure VI: Plasma PCSK9 levels are not increased by CsA in wt mice Wt were treated with 20 mg/kg/day CsA or vehicle (Ctrl) for 4 weeks. Plasma PSCK9 levels were determined by Western Blotting. A representative sample from Ctrl Ldlr-/- mice is shown. No increase in PCSK9 was observed after CsA treatment of wt mice.

Materials and Methods

Animals and CsA treatment Ldlr-/-, on a C57Bl/6.J background and wt C57Bl/6.J mice were obtained from JAX Labs in Bar Harbor, USA and the Animal Resources Centre in Perth, Australia, respectively. Mice were fed a chow diet, provided water ad libitum and kept on a 12 h day/night cycle throughout the study. Female mice weighing 18 to 20 g received 20mg/kg/day CsA (Sandimmun® IV) via subcutaneous implanted Alzet osmotic minipumps (model 2004) for 4 weeks. Control mice received minipumps containing vehicle (33% Ethanol and 62% Cremophor®EL; Sigma). All animal work was conducted according to the Animal Care and Ethics Committee guidelines from the University of New South Wales and the Animal Welfare Committee from the Sydney Local Health Network, Sydney, Australia.

Liver and kidney histology Formalin-fixed livers and kidneys were embedded in paraffin and stained with hematoxylin-eosin. Kidney sections were also assessed using Periodic acid-Schiff stain.

Insulin tolerance tests Insulin tolerance tests (ITT) were performed after 4 weeks of CsA treatment. Non- fasted mice received an i.p. injection with 1 U/kg of body weight of human insulin (Sigma) followed by repeated plasma glucose measurements over 2 h using a glucometer (Accuchek Performa, Roche). The area under the curve (AUC) was determined using Graphpad Prism using the trapezoid method1.

Plasma analysis Whole blood CsA levels were determined using liquid chromatography-tandem mass spectrometry (LC-MS). Fasted plasma Chol and TG concentrations were measured using commercial kits (Wako Chemicals). Lipoprotein analysis was performed by fast protein liquid chromatography (FPLC) using a Superose 6 10/300 GL column (GE Healthcare) with elution in 20mM sodium phosphate buffer, pH 7.8 at 0.25mL/min and monitoring at 280nm. 250µl fractions were collected and Chol and TG levels in each fraction were determined using commercial kits (Wako chemicals). Plasma levels of apoAI, apoE, apoB, apoCIII and PCSK9 were determined by Western analysis. Briefly, indicated volumes of plasma (0.5 or 1µl) were separated using 7% NuPAGE Tris-Acetate gels (Thermofisher) for detection of apoB48 and apoB100, while all other apolipoproproteins were detected after separation on 4-12% NUPAGE Bis-Tris gels (Thermofisher). Antibodies used were: mouse anti-apoAI (Meridian Life Sciences), anti-apoE (Meridian Life Sciences), anti-apoB (Abcam), anti-apoC-III (Santa Cruz) and anti-PCSK9 (Abcam). VLDL (d>1.006 g/mL) was isolated by density ultracentrifugation and particle size determined by dynamic light scattering using a Malvern Zetasizer Nano ZS equipped with a 633 nm He-Ne laser at 25°C2.

mRNA and protein analysis Total mRNA was extracted from 50-100 mg pieces of tissue using TRIzol reagent (LifeTechnologies) and a Fastprep tissue homogenizer. RNA was extracted with chloroform and purified using the PureLink RNA Mini Kit (LifeTechnologies) according to the manufacturer’s instructions. 1 µg of RNA was reversed transcribed to cDNA using oligo(dT)primers and Superscript III (LifeTechnologies). Relative realtime PCR was performed using SensiMixPlus SYBR (Bioline) and the primer sequences presented in Supplemental Table S1. Reaction were carried out on a Rotor-Gene 3000 (Qiagen) using the following protocol: 95°C for 10 min, 40 cycles of 95°C for 10 sec and 59°C for 1 min. Hypoxanthine guanine phosphoribosyl transferase (Hprt) was used as a housekeeping gene. Relative mRNA expression between control and CsA treated mice were determined using the ΔΔCt method. Total protein homogenates were prepared in RIPA buffer (50mM Tris, 150mM NaCl, 0.1% SDS, 1% Triton-X100, 0.5% deoxycholate, pH 7.5) from 25-50 mg pieces of liver using a Fastprep tissue homogenizer. Western blots were run using 4-12% NuPAGE Bis-Tris gels (Thermofisher) and mouse anti-MTP (BD Biosciences), FAS (BD Biosciences) SCD1 (Cell signaling technology), receptor-2 (apoER2; Abcam), VLDL-receptor (Novus), Lipoprotein Related Protein-1 (LRP1; Abcam) and syndecan-1 (Abcam).

Hepatic VLDL-TG production rates In vivo VLDL-TG production was determined as described by Schaap et al3. In short, mice were treated with and without CsA for 4 weeks as described above. After a 4 h fast, mice received 150µCi/mouse Express Protein Labeling mix [35S] (Perkin Elmer) through the tail vein. After 30min mice were i.v. injected with 500mg/kg triton WR 1339 to block lipolysis. Blood samples were taken at 0,10, 30 and 60 min after triton injection via tail bleeding. At 120 min mice were euthanized and a blood sample was taken via cardiac puncture. VLDL was isolated from plasma by density ultracentrifugation (Schaap) and [35S]-apoB incorporation was determined by scintillation counting on a Tricarb 2000. Plasma TG levels were determined using commercial kits (Wako chemicals) and TG secretion rates were determined by linear regression.

LPL activity assay Total releasable LPL activity was determined by subtracting activity in preheparin plasma (i.e. mainly hepatic lipase) from lipase activity in postheparin plasma according to Dallinga-Thie et al4 after mice were treated with and without CsA for 4 weeks. LPL activity was determined using a commercial available LPL activity assay (CellBiolabs).

In vivo clearance of VLDL-like particles VLDL-like TG-rich emulsion particles (80nm) containing tracer amounts of glycerol tri[3H]oleate (triolein, TO) and [14C]cholesteryl oleate (CO) were prepared as described previously5. After 4 weeks of CsA treatment [3H]TO/[14C]CO particles (1mg TG containing 1.2x106 µCi [3H]TO and 0.2x106 µCi [14C]CO) were injected via the tail vein (t=0) and blood samples were taken at 2, 5, 10, 15 and 30 min after injection. Plasma [3H] and [14C] levels were determined and decay curves were established using plasma volumes calculated as 0.04706 x body weight (g). At the final time point, mice were euthanized and a blood sample was taken via cardiac puncture followed by perfusion with ice-cold heparin solution (0.1% v/v in saline). Organs and tissues were isolated and analysed for [3H] and [14C] activity content after dissolving in Solvable (Perkin Elmer).

Lipidomic analysis of plasma and lipoprotein fractions Plasma and lipoprotein samples were randomised prior to lipid extraction and analysis. Lipoproteins (100µL) were lyophilized and reconstituted in 10µL phosphate buffered saline prior to lipid extraction. To each sample (10µL), a mixture of internal standards in chloroform:methanol (1:1, 15µL) was added. The internal standards comprised lipids which are either stable isotope labeled or non-physiological and so present in plasma at extremely low concentrations. Total lipid extraction from whole plasma (10µL) or lipoprotein fractions were performed by the addition of a single phase chloroform:methanol (2:1) mixture. Lipid profiling was performed by high performance liquid chromatography (HPLC) electrospray ionisation tandem mass spectrometry using an Agilent 1200 liquid chromatography system combined with an Applied Biosystems API 4000 Q/TRAP mass spectrometer with a turbo ion spray source (350oC) and ABSCIEX Analyst® software (Version 1.5, ABSCIEX, Foster City, California, USA). Precursor ion scans and neutral loss scans, and multiple reaction monitoring were performed as previously described6. Values for each lipid class were calculated as the sum of the individual lipid species. Plasma and lipoprotein lipids were normalized to PC or apoB content.

Data analysis Comparisons between the vehicle-control and CsA-treatment group were assessed by unpaired student’s t test or ANOVA with a Holm-Sidak post hoc test to correct for multiple comparisons. VLDL secretion and clearance one phase decay curves were determined using linear and non-linear regression, respectively. Statistical analyses were performed using Prism 6. Differences were considered significant at two-tailed p < 0.05.

1. McGrath KC, Li XH, Whitworth PT, Kasz R, Tan JT, McLennan SV, Celermajer DS, Barter PJ, Rye K-A, Heather AK. High density lipoproteins improve insulin sensitivity in high-fat diet-fed mice by suppressing hepatic inflammation. J. Lipid Res. 2014;55(3):421–430.

2. Berbée JFP, van der Hoogt CC, Sundararaman D, Havekes LM, Rensen PCN. Severe hypertriglyceridemia in human APOC1 transgenic mice is caused by apoC-I-induced inhibition of LPL. J. Lipid Res. 2005;46(2):297–306.

3. Schaap FG, Rensen PCN, Voshol PJ, Vrins C, van der Vliet HN, Chamuleau RAFM, Havekes LM, Groen AK, van Dijk KW. ApoAV reduces plasma triglycerides by inhibiting very low density lipoprotein-triglyceride (VLDL-TG) production and stimulating lipoprotein lipase-mediated VLDL-TG hydrolysis. J. Biol. Chem. 2004;279(27):27941–27947.

4. Dallinga-Thie GM, Zonneveld-de Boer AJ, van Vark-van der Zee LC, van Haperen R, van Gent T, Jansen H, De Crom R, van Tol A. Appraisal of hepatic lipase and lipoprotein lipase activities in mice. J. Lipid Res. 2007;48(12):2788– 2791.

5. Berbée JFP, Boon MR, Khedoe PPSJ, Bartelt A, Schlein C, Worthmann A, Kooijman S, Hoeke G, Mol IM, John C, Jung C, Vazirpanah N, Brouwers LPJ, Gordts PLSM, Esko JD, Hiemstra PS, Havekes LM, Scheja L, Heeren J, Rensen PCN. Brown fat activation reduces hypercholesterolaemia and protects from atherosclerosis development. Nat Commun 2015;6:6356..

6. Weir JM, Wong G, Barlow CK, Greeve MA, Kowalczyk A, Almasy L, Comuzzie AG, Mahaney MC, Jowett JBM, Shaw J, Curran JE, Blangero J, Meikle PJ. Plasma lipid profiling in a large population-based cohort. J. Lipid Res. 2013;54(10):2898–2908.

A. C57Bl6 wild type B. C57Bl6 wild type with CsA C. LDLr-/- with CsA

C

LDLr LDLr LDLr-/- CsA CsA Non Non - Non - LDLr LDLr LDLr IDL/ IDL/ IDL/ Liver Liver CsA Liver LDL LDL LPL- CsA LDL IDL/ - IDL/IDL/ CIII+ - LDL LDLLDL PCSK9 ++