Role of 3β-Hydroxysteroid-Δ24 Reductase in Mediating Antiinflammatory Effects of High-Density Lipoproteins in Endothelial Cells
Objective— The purpose of this study was to investigate the ability of high-density lipoproteins (HDLs) to upregulate genes with the potential to protect against inflammation in endothelial cells.
Methods and Results— Human coronary artery endothelial cells (HCAECs) were exposed to reconstituted HDLs (rHDLs) for 16 hours before being activated with tumor necrosis factor-α (TNF-α) for 5 hours. rHDLs decreased vascular cell adhesion molecule-1 (VCAM-1) promoter activity by 75% (P<0.05), via the nuclear factor-kappa B (NF-κB) binding site. rHDLs suppressed the canonical NF-κB pathway and decreased many NF-κB target genes. Suppression of NF-κB and VCAM-1 expression by rHDLs or native HDLs was dependent on an increase in 3β-hydroxysteroid-Δ24 reductase (DHCR24) levels (P<0.05). The effect of HDLs on DHCR24 is dependent on SR-BI but not ABCAI or ABCGI. Silencing DHCR24 expression increased NF-κB (1.2-fold, P<0.05), VCAM-1 (30-fold, P<0.05), and NF-κB p50 (4-fold, P<0.05) and p65 subunits (150-fold, P<0.05). TNF-α activation of siDHCR24-treated cells increased expression of VCAM-1 (550-fold, P<0.001) and NF-κB (9-fold, P<0.001) that could no longer be suppressed by rHDLs.
Conclusions— Results suggest that antiinflammatory effects of rHDLs are mediated partly through an upregulation of DHCR24. These findings raise the possibility of considering DHCR24 as a target for therapeutic modulation.
High-density lipoproteins (HDLs) have multiple cardio-protective properties. The best known relates to their ability to promote efflux of cholesterol from macrophages.1 They also inhibit endothelial inflammation.2–5
HDLs inhibit expression of vascular cell adhesion molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1), and E-selectin both in vitro4–6 and in vivo3,7,8 in a process that may be secondary to a suppression of NF-κB activity.4,5,9 We now provide evidence that an HDL-mediated inhibition of VCAM-1 in activated endothelial cells is via blockade of signaling through the classical NF-κB pathway.
Antiinflammatory effects of HDLs are apparent with both native HDLs and discoidal reconstituted HDLs (rHDLs) that contain only apolipoprotein A-I (apoA-I) and phosphatidylcholine.6,10 Furthermore, the ability of rHDLs to inhibit VCAM-1 expression in TNFα-activated human umbilical vein endothelial cells (HUVECs) persists even if the rHDLs are removed from the culture medium several hours before the cells are activated by TNFα.11
A similar phenomenon also occurs in vivo. Insertion of a nonocclusive collar around a carotid artery of normal-fed rabbits induces expression of VCAM-1 and ICAM-1 in the endothelium and the infiltration of neutrophils into the artery wall. Intravenous infusions of a small amount of rHDLs suppress both the infiltration of neutrophils and the expression of endothelial VCAM-1 and ICAM-1.7 This antiinflammatory effect of rHDLs persists even if rHDLs are given 24 hours before applying the inflammatory insult, at which time virtually all of the injected rHDLs have been removed from the circulation.8
On the basis of these findings, we speculated that exposure of endothelial cells to rHDLs induces expression of a protein with the potential to protect against a subsequent inflammatory insult. We now show that incubation of human coronary artery endothelial cells (HCAECs) with either rHDLs or native HDLs induces expression of the antioxidant protein, 3β-hydroxysteroid-Δ24 reductase (also known as 24-dehydrocholesterol reductase - DHCR24) in an SR-BI dependent manner. Furthermore, we show that silencing DHCR24 expression increases NF-κB and VCAM-1 levels in both nonactivated and TNFα-activated HCAECs with an associated loss of the antiinflammatory effects of rHDLs.
Preparation of Reconstituted High-Density Lipoproteins
ApoA-I was purified from human plasma by sequential ultracentrifugation (Beckman Coulter), delipidation, and anion-exchage chromatography as previously described.12 Discoidal rHDLs containing apoA-I as their sole protein constituent and 1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphatidylcholine (PLPC, Avanti Polar Lipids) as their sole lipid were prepared by the cholate dialysis method.13
Preparation of High-Density Lipoproteins From Rabbit and Human Plasma
Twenty New Zealand White rabbits (Institute of Medical and Veterinary Sciences, Adelaide, South Australia) were randomly assigned to experimental groups. Procedures were approved by the Sydney South West Area Health Services Animal Welfare Committee. Chronic arterial inflammation was induced by 6 weeks of feeding a diet containing 0.2% cholesterol-enriched chow. Animals were given intravenous infusions (1 mL/min via a catherized ear vein) of either saline or rHDL containing 25 mg of apoA-I (8 mg/kg) on days 3 and 5 of the final week of the study. At the end of the 6-week study, blood was collected via cardiac puncture in tubes containing disodium EDTA and placed immediately on ice before animals were euthanized.
Human and rabbits HDLs were isolated by sequential ultracentrifugation (see supplemental Methods, available online at http://atvb.ahajournals.org).
Human coronary artery endothelial cells (HCAECs, Cell Applications) were cultured in HCAEC growth medium (Cell Applications). Unless otherwise stated, HCAECs were incubated for 16 hours with rHDLs, native human HDLs (1 mg/mL) or rabbit HDLs (32 μmols/L final apoA-I concentration), or PBS (vehicle control), then stimulated with or without TNFα (0.1 ng/mL) for 5 hours at 37°C in a 5% CO2 incubator. For the time course experiment, cells were seeded in a 12-well plate at a cell density of 1×105 cells/mL (1 mL per well) and preincubated with rHDLs (apoA-I 16 μmols/L) or PBS for 0, 1, 2, 4, 8, 12, or 16 hours. The media was then removed, cells washed twice with 500 μL PBS and fresh media added. At 0, 4, 8, 24, 72, or 120 hours post rHDL exposure, total RNA was isolated using TRI reagent (Sigma-Aldrich).
Cells were also incubated with 0.5 μmol/L simvastatin (Sigma-Aldrich) or 54 μmol/L cyclodextrin (Sigma-Aldrich) for 16 hours before isolating RNA.
Transient Cell Transfections and Luciferase Measurements
HCAECs (1×105 cells per well of a 12-well plate) were transfected using Effectene (Qiagen) with one of the following plasmids: IκBα-EGFP reporter vector (Clontech Laboratories), NF-κB-luciferase reporter vector (Promega Corporation), or VCAM-1 promoter-luciferase reporter vectors.14 Luciferase activity was assayed using the Dual-Luciferase Reporter System (Promega). IκBα levels were measured by flow cytometry in a Cytomics FC 500 flow cytometer (Beckman Coulter) using CXP software (Beckman Coulter).
For silencing experiments, HCAECs were transfected with DHCR24 siRNA (Santa Cruz Biotechnology), siABCAI, siABCGI, or SR-BI siRNAs (Qiagen). See supplemental Methods.
IκB Kinase Assay
HCAECs (1×105 cells per 12-well plate) were pretreated with rHDLs or PBS for 16 hours, then stimulated with TNFα for 7.5 minutes. IκB kinase (IKK) activity was measured using IKK substrate peptide (Millipore) and Kinase-Glo reagent (50 μL, Promega, supplemental Methods).
Human NF-κB Target Gene Array Analysis
The TransSignal NF-κB Target Gene array was used to measure human NFκB target gene expression (Panomics, see supplemental Method).
Total RNA extracted from HCAECs using TRI reagent (Sigma- Aldrich) was normalized to a concentration of 100 ng/μL using the SYBR Green II assay (Molecular Probes, Invitrogen) and iSCRIPT/ iQ SYBR Green Supermix used to perform RT-PCR in a BioRad iQ5 thermocycler. Relative changes in mRNA levels were determined by the ΔΔCT method,15 using β2-microglobulin and 18S levels as controls. Primer pair sequences are included in the supplemental Table I.
Affymetrix Human Genome Microarray
Total RNA (10 μg) was extracted using TRI reagent (Sigma- Aldrich). RNA integrity was determined using an Experion RNA analyzer (BioRad). Samples were then sent to the Australian Genome Research Facility (AGRF; Melbourne, Australia). Microarray image acquisition and differentiated gene analysis were completed using Partek Genomics Suite software (Partek Incorporated).
Western Blot Analysis
Protein extracts were subjected to Western blot analysis and probed with a goat polyclonal antihuman DHCR24 antibody (Santa Cruz Biotechnology; 1:1000 dilution) and the mouse monoclonal antihuman β-actin antibody (Chemicon; 1:10 000 dilution). Direct chemiluminescence imaging of the blots was performed using the ChemiDoc XRS (BioRad) imaging system. Quantity One software (BioRad) was used to quantify band densities.
NF-κB Nuclear Translocation Assay
Nuclear extracts were extracted from HCAECs using the NucBuster protein extraction kit (Merck & Co). The NF-κB NoShift transcription factor assay kit was used to measure NF-κB levels (Merck & Co).
Data are expressed as mean±SEM. Significant differences in treatments were determined by 1-way ANOVA with Bonferroni post test analysis used to determine significance. Significance was set at P<0.05.
Role of NF-κB in the Antiinflammatory Properties of rHDLs
To determine the role of NF-κB in the rHDL-mediated suppression of VCAM-1 in activated HCAECs, we studied the VCAM-1 promoter that is controlled mainly by 3 nuclear factors, activator protein-1 (AP-1), GATA, and NF-κB.16 Nuclear levels of AP-1 and NF-κB both increase in response to TNFα17 and both are involved in TNFα stimulation of VCAM-1 expression.18 To test the posssibility that effects on AP-1 and NF-κB may contribute to the ability of rHDLs to inhibit TNFα-induced VCAM-1 expression, a full-length VCAM-1 promoter DNA fragment (pFFLVCAM-1) and also a shorter fragment of the VCAM-1 promoter DNA that retained the NF-κB sites but not the AP-1 binding site (pF2VCAM-1) were generated by PCR and subcloned into the pGL3-luciferase reporter vector (Figure 1A). Transfected HCAECs were exposed to rHDLs for 16 hours before being activated with TNFα. Promoter activity of both FFLVCAM-1 and F2VCAM-1 were significantly inhibited by rHDLs (P<0.005; Figure 1A) indicating that effects on the NF-κB sites fully explain the ability of rHDLs to inhibit VCAM-1 expression in activated endothelial cells.
Effect of rHDLs on Signaling Through the IKK/IκBα/NF-κB Pathway
Prior exposure of HCAECs to rHDLs suppressed the TNFα-induced IKK activity (P<0.05; Figure 1B). To determine the effect on IκBα levels and NF-κB activity, HCAECs were transfected with an IκBα-EGFP reporter vector or a NF-κB-luciferase reporter vector. Exposure of the transfected cells to rHDLs increased IκBα levels (Figure 1C) and suppressed NF-κB–mediated DNA transcription as measured by a decrease in luciferase activity (Figure 1D).
Effect of rHDLs on TNFα-Stimulated NF-κB Target Gene Expression in HCAECs
To investigate effects of rHDLs on the expression of all NF-κB target genes, HCAECs were exposed to rHDLs or PBS for 16 hours before being activated with TNFα for 5 hours. Exposure to rHDLs suppressed a range of NF-κB–regulated genes (as assessed by macroarray) (Figure 2A), including those belonging to the cell adhesion molecule (CAMs) and metalloproteinase families (Figure 2B), the cytokine and growth factor families (Figure 2C), genes involved in cell cycle regulation (Figure 2D) and AGER, DDH1, HMG14, PRG1, and LAMB2 (Figure 2E). Real-time PCR verified the findings for a selection of genes (supplemental Table II).
Effect of rHDLs on DHCR24 Expression in HCAECs
The above experiments suggest that preincubation of HCAECs with rHDLs induced changes in gene expression so as to make them resistant to subsequent inflammatory insult. To investigate this, HCAECs were incubated with rHDLs (apoA-I 16 μmols/L) or PBS for 16 hours before measuring mRNA levels by Affymetrix human genome U133 Plus 2.0 arrays (54 675 probe sets). The relative expression levels of the differentially expressed genes are shown in the online supplement “Gene List” (supplemental Table III) at http://atvb.ahajournals.org. The top differentially expressed genes are shown in Table 1. All of the differentially expressed genes were filtered by Ingenuity Pathway Analysis database (http://www.ingenuity.com). This analysis showed that regulated genes functioned in cholesterol metabolism, inflammation, or oxidative stress (supplemental Figure I), including 25 antioxidant proteins or regulators of antioxidants proteins (supplemental Figure II). One gene, DHCR24, was highlighted in the top 10 gene list (3rd most upregulated gene Table 1) and clustered in the cholesterol synthesis biosynthesis pathway (supplemental Figure I). DHCR24 catalyzes the final step in the cholesterol biosynthesis pathway, desmosterol to cholesterol. DHCR24 is also a potent H2O2 scavenger and has antioxidant19 and antiapoptotic effects in neuronal cells.20 Real-time PCR and Western blot analysis confirmed that rHDLs increased DHCR24 mRNA levels by 8-fold (Figure 3A) and protein levels by ≈2-fold (Figure 3B). The difference in expression level measured for mRNA and protein most probably reflects the increased sensitivity of real-time PCR versus antibody-based chemiluminescence-dependent Western blot analysis. The rHDL-induced increase in DHCR24 mRNA levels persisted even in the presence of TNFα (Figure 3A).
We also demonstrated the ability of native HDLs isolated from both human and rabbit plasma (Figure 3C) to increase DHCR24 expression. In contrast, human LDLs had no effect on DHCR24 expression (Figure 3C).
rHDLs increased DHCR24 expression in a dose-dependent manner (Figure 3D), with effects apparent at apoA-I levels as low as 8 μmol/L.
In previous studies, we reported that rHDLs inhibited VCAM-1 expression in activated endothelial cells even if the rHDLs had been removed before activation with TNFα.11 This was also true for DHCR24 expression. HCAECs required at least an 8-hour exposure to rHDLs to induce DHCR24 expression (Figure 3E). However, once induced, the effect persisted for at least 8 hours even after the rHDLs had been removed with DHCR24 expression returning to the uninduced state by 24 hours (Figure 3F).
Effects of DHCR24 on VCAM-1 Expression and NF-κB Activity
To determine whether DHCR24 is involved in the ability of rHDLs to suppress inflammatory gene expression in HCAECs, cells were transfected with siRNA targeted against DHCR24. siRNA decreased DHCR24 mRNA levels by about 70% (data not shown) which was associated with a 30-fold increase in VCAM-1 expression (P<0.05, supplemental Figure IIIA). When DHCR24-silenced HCAECs were treated with TNFα, VCAM-1 expression was increased 550-fold compared with levels in nonactivated cells. In contrast to the control cells, rHDLs did not significantly suppress VCAM-1 expression in TNFα-activated DHCR24-silenced cells. DHCR24-silenced HCAECs also showed increased expression of the NF-κB subunits, p50 (4-fold, P<0.05, supplemental Figure IIIB) and p65 (50-fold, P<0.05, supplemental Figure IIIC) and a 25% (P<0.05) increase in NF-κB activity, relative to controls (data not shown).
To determine whether these effects of HDLs on DHCR24 were related to cell cholesterol synthesis or efflux, we treated HCAECs with cyclodextrin to promote cholesterol efflux or simvastatin to block cholesterol synthesis. Cyclodextrin had no effect on DHCR24 or VCAM-1 expression (Figure 4A). Predictably, simvastatin decreased DHCR24 but this was not associated with an increase in VCAM-1 expression, suggesting that inhibition of cholesterol synthesis was not responsible for the elevated VCAM-1 expression in silenced-DHCR24 cells. In keeping with the cyclodextrin result, we found that lipid-free AI had no effect on DHCR24 or VCAM-1 expression (Figure 4B). Furthermore, silencing ABCAI also had no effect on DHCR24 expression or on the ability of rHDLs to increase DHCR24 expression. Similarly, silencing ABCGI had no effect on DHCR24 expression. Silencing SR-BI, however, did reduce DHCR24 levels both in the presence and absence of rHDLs, suggesting that SR-BI is involved in the ability of rHDLs to increase DHCR24 expression (Figure 4C). Together, this final series of experiments suggested that cholesterol synthesis/efflux is not involved in the regulation of DHCR24 by HDLs or in the effects of DHCR24 in HCAECs.
This study demonstrates that rHDL-mediated suppression of NF-κB activity in activated endothelial cells4,5,11 is achieved via the classical IKK/IκBα/NF-κB signaling pathway, with rHDLs suppressing IKK, increasing IκBα levels, thereby suppressing NF-κB nuclear translocation. Furthermore, suppression of NF-κB activity by rHDLs explains the ability of rHDLs to inhibit VCAM-1 expression and monocyte adhesion to activated endothelial cells. The ability of rHDLs to inhibit many NF-κB target genes may also explain both the known antiinflammatory and antiapoptotic properties of rHDLs.
The most striking finding of this study is the ability of both native and reconstituted HDLs to increase endothelial cell expression of DHCR24, an effect that persisted for at least 8 hours after the HDLs had been removed from the culture media. Together, these findings provide an insight into the previously unexplained observation that the response of endothelial cells to an inflammatory insult is markedly reduced by prior exposure to rHDLs, even if the rHDLs are removed several hours before initiation of the inflammatory insult.11 It may also explain why rabbits injected intravenously with rHDLs 24 hours before inserting a nonocclusive silastic collar around the carotid artery remain protected against the resulting acute vascular inflammation, even though the injected rHDLs have long since been cleared from the plasma.9
DHCR24 catalyzes the final step in cholesterol biosynthesis, the conversion from desmosterol to cholesterol.21 However, in addition, DHCR24 is antiapoptotic and has been shown to regulate cell growth and senescence via an interaction with p53.20,22 The antiapoptotic effects may relate to the ability of DHCR24 to scavenge hydrogen peroxide.19 Because NF-κB is redox sensitive, it is possible that these antioxidant properties suppress both endothelial cell NF-κB activity and the downstream expression of VCAM-1. This possibility is supported by finding that NF-κB activity and VCAM-1 expression were both increased in endothelial cells in which the gene for DHCR24 had been silenced (supplemental Figure III). Because silencing DHCR24 may have reduced cell cholesterol synthesis and because incubation with rHDLs would have increased cholesterol efflux, it was necessary to determine whether any of the effects observed in these studies were secondary to changes in cellular cholesterol. This possibility was excluded by finding that neither simvastatin nor cyclodextrin had any effect on either DHCR24 or VCAM-1 (supplemental Figure III). Rather, it was apparent that rHDLs increased the expression of a number of genes functionally associated with the oxidative stress response, including a number of antioxidant proteins, independent of cholesterol metabolism. Interestingly, the most downregulated gene by HDLs in HCAECs was thioredoxin-interacting protein (TXNIP). TXNIP is an oxidative stress mediator that inhibits thioredoxin (a potent antioxidant), thus the downregulation of TXNIP is in keeping with an augmented antioxidant potential in HCAECs treated with HDLs.
Native as well as reconstituted HDLs increased DHCR24 expression. The observation that lipid-free AI had no effect raises the possibility that phospholipids (rather than apoA-I) may have been responsible for the effect, consistent with previous reports that phospholipid vesicles are as effective as rHDLs in suppressing TNFα-activation of VCAM-1 expression in human endothelial cells.23
These studies provide a plausible mechanism to explain previous observations that rHDLs inhibit cytokine-induced cell adhesion molecule expression,4,6,24 metalloproteinase expression,25 apoptosis,26 and thrombus formation27 in cell and animal models of inflammation. Specifically, we have shown that rHDLs suppress the cytokine-induced expression of many of the NF-κB gene targets, possibly explaining an ability of HDLs to retard plaque progression or promote plaque stability. If a reduced expression of DHCR24 were to contribute to development of atherosclerosis, it is possible that an HDL-mediated increase in its expression may limit plaque growth or stabilize vulnerable atherosclerotic plaques and reduce their potential to rupture and cause clincal events.
In conclusion, rHDLs have been shown to inhibit VCAM-1 expression in activated endothelial cells via suppression of NF-κB activity. We have also shown that the suppression of NF-κB activity is achieved via blockade of the classical NF-κB signaling pathway rather than through the downregulation of individual subunit expression of NF-κB per se. This effect of HDL then leads to decreased expression of many NF-κB target genes. And finally, we have shown that rHDLs stimulate endothelial cell expression of DHCR24 in a process that may explain part of the antiinflammatory properties of HDLs. This involvement of DHCR24 raises the possibility that it should be considered as a future target for therapeutic modulation.
Sources of Funding
This work was supported by the Bruce and Joy Reid Foundation and the National Health and Medical Research Council (NH&MRC) of Australia.
K.C.Y.M. and X.H.L. contributed equally to this study.
Received October 2, 2008; revision accepted March 11, 2009.
Dimayuga P, Zhu J, Chyu KY, Xu XO, Yano J, Shah PK, Nilsson J, Cercek B. Reconstituted HDL containing human apolipoprotein A-I reduces VCAM-1 expression and neointima formation following periadventitial cuff-induced carotid injury in apoE null mice. Biochem Biophys Res Commun. 1999; 264: 465–468.
Cockerill GW, Rye K-A, Gamble JR, Vadas MA, Barter PJ. High-density lipoproteins inhibit cytokine-induced expression of endothelial cell adhesion molecules. Arterioscler Thromb Vasc Biol. 1995; 15: 1987–1994.
Nicholls SJ, Dusting GJ, Cutri B, Bao S, Drummond GR, Rye K-A, Barter PJ. Reconstituted high-density lipoproteins inhibit the acute pro-oxidant and proinflammatory vascular changes induced by a periarterial collar in normocholesterolemic rabbits. Circulation. 2005; 111: 1543–1550.
Xia P, Gamble JR, Rye K-A, Wang L, Hii CS, Cockerill P, Khew-Goodall Y, Bert AG, Barter PJ, Vadas MA. Tumor necrosis factor-alpha induces adhesion molecule expression through the sphingosine kinase pathway. Proc Natl Acad Sci U S A. 1998; 95: 14196–14201.
Xia P, Vadas MA, Rye KA, Barter PJ, Gamble JR. High density lipoproteins (HDL) interrupt the sphingosine kinase signaling pathway. A possible mechanism for protection against atherosclerosis by HDL. J Biol Chem. 1999; 274: 33143–33147.
Matz CE, Jonas A. Micellar complexes of human apolipoprotein A-I with phosphatidylcholines and cholesterol prepared from cholate-lipid dispersions. J Biol Chem. 1982; 257: 4535–4540.
Bustin SA. Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. J Mol Endocrinol. 2000; 25: 169–193.
Minami T, Abid MR, Zhang J, King G, Kodama T, Aird WC. Thrombin stimulation of vascular adhesion molecule-1 in endothelial cells is mediated by protein kinase C (PKC)-δ-NF-κB and PKC-ε-GATA signaling pathways. J Biol Chem. 278: 6976–6984.
Darnay BG, Haridas V, Ni J, Moore PA, Aggarwal BB. Characterization of the intracellular domain of receptor activator of NF-κB (RANK): interaction with tumor necrosis factor receptor-associated factors and activation of NF-κB and c-Jun N-terminal kinase. J Biol Chem. 273: 20551–20555.
Greeve I, Hermans-Borgmeyer I, Brellinger C, Kasper D, Gomez-Isla T, Behl C, Levkau B, Nitsch RM. The human DIMINUTO/DWARF1 homolog seladin-1 confers resistance to Alzheimer’s disease-associated neurogeneration and oxidative stress. J Neurosci. 2000; 20: 7345–7352.
Waterham HR, Koster J, Romeijn GJ, Hennekam RC, Vreken P, Andersson HC, FitzPatrick DR, Kelley RI, Wanders RJ. Mutations in the 3β-hydroxysterol Δ24-reductase gene cause desmosterolosis, an autosomal recessive disorder of cholesterol biosynthesis. Am J Hum Genet. 2001; 69: 685–694.
Baker PW, Rye KA, Gamble JR, Vadas MA, Barter PJ. Ability of reconstituted high density lipoproteins to inhibit cytokine-induced expression of vascular cell adhesion molecule-1 in human umbilical vein endothelial cells. J Lipid Res. 1999; 40: 345–353.
Robbesyn F, Auge N, Vindis C, Cantero AV, Barbaras R, Negre-Salvayre A, Salvayre R. High-density lipoproteins prevent the oxidized low-density lipoprotein-induced epidermal [corrected] growth factor receptor activation and subsequent matrix metalloproteinase-2 upregulation. Arterioscler Thromb Vasc Biol. 2005; 25: 1206–1212.
Nofer JR, Levkau B, Wolinska I, Junker R, Fobker M, von Eckardstein A, Seedorf U, Assmann G. Suppression of endothelial cell apoptosis by high density lipoproteins (HDL) and HDL-associated lysosphingolipids. J Biol Chem. 2001; 276: 34480–34485.
Li D, Weng S, Yang B, Zander DS, Saldeen T, Nichols WW, Khan S, Mehta JL. Inhibition of arterial thrombus formation by ApoA1 Milano. Arterioscler Thromb Vasc Biol. 1999; 19: 378–383.