Overexpression of Apolipoprotein F Reduces HDL Cholesterol Levels In Vivo
Objective— Apolipoprotein F (ApoF) is a protein component of several lipoprotein classes including HDL. It is also known as lipid transfer inhibitor protein (LTIP) based on its ability to inhibit lipid transfer between lipoproteins ex vivo. We sought to investigate the role of ApoF in HDL metabolism.
Methods and Results— Adeno-associated viruses (AAV) based on serotype 8, were used to overexpress either murine or human ApoF in mice. Overexpression of murine ApoF significantly reduced total cholesterol levels by 28% (P<0.001), HDL by 27% (P<0.001), and phospholipid levels by 19% (P<0.001). Overexpression of human ApoF had similar effects. Human ApoF was nearly exclusively HDL-associated in mice. In agreement with this finding, greater than 90% of the ApoF in human plasma was found on HDL3, with only a small amount on LDL. Overexpression of mouse ApoF accelerated the plasma clearance of [3H]-cholesteryl ether labeled HDL. Plasma from mice overexpressing ApoF showed improved macrophage cholesterol efflux on a per HDL-C basis.
Conclusions— ApoF overexpression reduces HDL cholesterol levels in mice by increasing clearance of HDL-CE. ApoF may be an important determinant of HDL metabolism and reverse cholesterol transport.
- apolipoprotein F
- lipid transfer inhibitor protein
- adeno-associated virus
- HDL cholesterol
- reverse cholesterol transport
Apolipoprotein F (ApoF) is a 29-kDa sialoglycoprotein found in the HDL and LDL fractions of human plasma.1–4 It is also known as lipid transfer inhibitor protein (LTIP) for its ability to inhibit lipid transfer between lipoproteins ex vivo.5 ApoF (LTIP) is an endogenous inhibitor of CETP-mediated cholesteryl ester (CE) and triglyceride (TG) transfer.6,7 In vitro, ApoF can selectively inhibit transfer events involving LDL,8 and thereby indirectly increases the movement of CE from HDL to VLDL.9,10 Accordingly, ApoF has been postulated to play an important role in HDL metabolism and reverse cholesterol transport (RCT).11,12
ApoF originates in the liver as a 326-aa precursor, which consists of a small signal peptide followed by a large proprotein of 290 amino acids. The proprotein is cleaved to release the 162 amino acid C terminus which makes up the mature secreted form of the protein.3 ApoF is an unusual apolipoprotein in that it is heavily glycosylated with both O- and N-linked sugar groups. These glycosylations render the protein very acidic with an isoelectric point of 4.5, and result in a molecular mass about 40% greater than predicted.
Recent shotgun proteomics studies have identified ApoF as a component of the human HDL particle.13 This renewed our interest in ApoF as a candidate gene involved in HDL metabolism. Little is known about the effects or function of ApoF in vivo, and no overexpression model has been reported. To test the hypothesis that ApoF modulates HDL metabolism, we generated a recombinant adeno-associated virus serotype 8 (AAV8) vector to somatically overexpress ApoF in mouse liver. In this report, we show that stable overexpression of ApoF (1) reduces HDL cholesterol levels, (2) accelerates plasma clearance of HDL-CE, and (3) results in HDL particles that have increased ability to accept cholesterol from the macrophage.
An expression clone encoding mouse ApoF (Genbank BC010815) was obtained from the American Type Culture Collection. A human ApoF expression clone (Genbank NM_001638) was purchased from Origene. The ApoF cDNAs were subcloned into the pcDNA3.1(+)/V5-His TOPO vector (Invitrogen) to generate C-terminal tagged versions. Site-directed mutagenesis was preformed using the Quikchange II XL kit (Stratagene). Primer sequences used for cloning are listed in Supplemental Table I. All clones were verified by sequencing at the Department of Medical Genetics DNA Sequencing Facility (University of Pennsylvania, School of Medicine).
Recombinant AAVs were engineered to overexpress Beta Galactosidase (Lac Z), mouse ApoF (mApoF), or human ApoF (hApoF). AAV8 was generated as previously described14 using a chimeric packaging construct in which the Rep gene from AAV2 has been fused with the Cap gene from AAV8.
HEK293 cells were grown in Dulbecco modification of Eagle media (DMEM) containing 10% fetal bovine serum and 1% antibiotic/antimycotic. Cells were seeded into 12-well plates and transiently transfected with 1 μg of plasmid DNA with Lipofectamine (Invitrogen). Twenty-four hours later, the media were changed to serum free DMEM. The cells were incubated overnight and harvested the following day.
Immunoblotting was performed as described previously.15 Antibodies and dilutions are described in the supplemental methods (available online at http://atvb.ahajournals.org).
Male wild-type C57BL/6 mice were obtained from Jackson Labs. Mice were fed a standard chow diet (LabDiet number 5053) ad libitum. AAVs were delivered by intraperitoneal injection. In all cases, fasting plasma was obtained by retro-orbital bleeding while the mice were under isofluorane anesthesia. Animal experiments were approved by the University of Pennsylvania Institutional Animal Care and Use Committee.
Lipid and Lipoprotein Analyses
Total cholesterol, HDL cholesterol, phospholipids, triglycerides, and alanine aminotransferase (ALT) levels were measured with the use of diagnostic reagents from Sigma. Pooled plasma from each group (160 μL), was separated by fast protein liquid (FPLC) gel filtration. Total cholesterol in each fraction was measured using the Total Cholesterol E kit from Wako.
HDL Kinetic Study
Human HDL3 was isolated from pooled human plasma and labeled with [3H]-cholesteryl hexadecyl ether (cholesteryl-1,2-3H, Perkin Elmer) as described previously.16 The fractional catabolic rate (FCR) was determined from the area under the plasma decay curves fitted to a bicompartmental model using the WinSAAM software assuming a plasma volume equivalent to 3.5% of body weight.
Reverse Cholesterol Transport Assay
In vivo RCT assay was performed as previously described using acetylated LDL loaded J774 macrophages labeled with [3H]-Cholesterol.17
Cholesterol Efflux Assays in Bone Marrow–Derived Macrophages
Efflux assays were performed as previously described.18 Briefly, cells were loaded with 2 μCi / ml of [3H]-Cholesterol and 25 μg/ml of acetylated LDL for 24 hours. The cells were preincubated for 18 hours with an LXR agonist (GW3965, 10 μmol/L) to upregulate ABCA1 and ABCG1 expression. Where indicated, cells were treated with 20 μmol/L probucol for 2 hours before starting. Efflux to pooled plasma (final 2.5%) was performed for 2 hours at 37°C. Free cholesterol (FC) efflux was calculated as the percentage of the counts in the media divided by the sum of the counts in the media and the FC counts in the cells. ABCA1-dependent efflux was defined as the total FC efflux minus the FC efflux for cells pretreated with probucol (specific ABCA1 inhibitor). Efflux was normalized to HDL-C levels by dividing by the plasma HDL cholesterol concentration (mg/dl).
All values are shown as the mean±SD. A 2-tailed Student t test was used to test for statistical significance. Detailed methods are available as supplemental materials.
Proteolytic Processing and N-Linked glycosylation of ApoF
To investigate the effects of ApoF on lipoprotein metabolism, we first examined the protein in vitro. To study the synthesis and processing of ApoF, we made C-terminally tagged versions of mouse and human ApoF (Figure 1A) and transiently transfected them into HEK293 cells. Human ApoF produced 2 sets of bands in the media: a doublet around 33 kDa (normally 29 kDa without the tag), as well as a higher band at 55 kDa (Figure 1, lane 2). In contrast, mouse ApoF produced a single band in the media around 35 kDa (Figure 1, lane 9). To delineate the differences between the proprotein and mature ApoF, we mutated arginine 164 at the predicted proprotein cleavage site to alanine (R165 in mApoF). This substitution completely inhibited processing to the mature form (Figure 1, lanes 7 and 10), revealing that the upper band in the media represents the ApoF proprotein.
ApoF is glycosylated with both O-linked and N-linked sugar groups,12 increasing its size from the predicted 17.5 kDa to the observed molecular mass of 29 kDa.3 Human ApoF contains 3 predicted N-linked glycosylation sites not found in the mouse protein (Figure 1A; supplemental Figure I). Two of these sites are in the proprotein (N118 and N139), and one lies in the mature peptide (N267). Site directed mutagenesis was used to mutate these asparagine residues to alanine. The proprotein in the media was glycosylated at both N118 and N139 (Figure 1B, lanes 3 to 5), as mutation of either or both of these residues reduced the molecular mass of the upper band. The mature form of ApoF appears to be variably glycosylated at N267, as mutation of this residue abolished the upper band of the doublet in the media (Figure 1B, lane 6). Similar results were seen in the cell lysates (Figure 1B, bottom panel). ApoF in the cell lysates was found to be both PNGase F and EndoH-sensitive, consistent with N-glycosylated proteins that have not yet been processed by Mannosidase II in the medial Golgi. ApoF in the media was PNGase F-sensitive but Endo H-resistant, indicating that the secreted forms of ApoF are maturely glycosylated (supplemental Figure II).
Generation of a Polyclonal Antibody to Human ApoF and Detection of ApoF in Human HDL
An adeno-associated virus (AAV) encoding human ApoF was used to generate a rabbit polyclonal antibody. The antibody did not recognize mouse ApoF in plasma (supplemental Figure IIIA, lane 1- control) even when overexpressed (supplemental Figure IIIA, lane 3- mApoF). Two sets of bands were visible in plasma from mice overexpressing human ApoF, including a doublet at 29 kDa and an additional band around 45 kDa (supplemental Figure IIIA, lane 4- hApoF). The upper band is most likely uncleaved ApoF proprotein (Figure 1B). We used this antibody to examine ApoF in human lipoproteins. Surprisingly, greater than 90% of the ApoF was found on HDL3, with only a small amount on LDL (supplemental Figure IIIB). ApoF was not detected on VLDL or HDL2. The ApoF proprotein could not be detected in human plasma (supplemental Figure IV).
AAV8-Mediated Overexpression of ApoF In Vivo
AAVs based on serotype 8 were constructed to overexpress human or mouse ApoF under the control of the liver-specific thyroxine binding globulin promoter.19 Wild-type male C57BL/6 mice were injected with 1×1012 genome copies (GC) of AAV encoding either Lac Z or hApoF and followed for 4 weeks. Human ApoF overexpression reduced HDL-C by 15% (P<0.05). Phospholipids paralleled the decrease in HDL with a 19% reduction (P<0.05). Triglyceride levels were not affected. In addition, plasma alanine aminotransferase (ALT) levels were not substantially elevated (mean <150 IU/L), confirming that the lipid effects are not attributable to hepatotoxicity (data not shown). Pooled plasma from mice infected with AAV-Lac Z or AAV-hApoF was separated by FPLC. Human ApoF overexpression reduced cholesterol in all fractions. The HDL-C peak was noticeably decreased, but not shifted in either direction, indicating that particle size was not affected (Figure 2). ApoF was found to be strictly HDL-associated in these mice. Interestingly, human ApoF appeared to preferentially associate with the larger mouse HDL fractions. The same strategy was used to overexpress mouse ApoF in mice. Mouse ApoF overexpression gave effects similar to those observed for human ApoF. Total cholesterol levels were reduced by 28% (P<0.001) at 4 weeks. Accordingly, HDL-C decreased by 27% (P<0.001), and phospholipids fell by 19% (P<0.001) relative to control animals (Figure 3). As with human ApoF, plasma triglycerides and ALT levels were not affected. ApoF was overexpressed 25-fold relative to the endogenous ApoF message in the liver as determined by quantitative RT-PCR (data not shown). Mouse ApoF reduced cholesterol across all lipoprotein fractions (supplemental Figure V).
Effects of ApoF Overexpression on HDL Turnover
We hypothesized that ApoF might decrease HDL cholesterol levels by increasing cholesteryl ester removal from the plasma. A kinetic study was performed to examine the plasma clearance of human HDL3-labeled cholesteryl ether. Mice were injected with AAV-LacZ or AAV-mApoF three weeks before the kinetic study. We chose to overexpress mouse ApoF in this and subsequent experiments, because it is plausible that human ApoF may not as efficiently interact with mouse SR-BI or other receptors in the liver. At 2 weeks, total cholesterol in the plasma was reduced by 19% (P<0.05), and HDL-C decreased by 20% (P<0.05) (supplemental Figure VIA). Plasma clearance of HDL-CE was significantly faster in the mice overexpressing mouse ApoF. This was reflected in a 24% increase in the fractional catabolic rate- 0.220±0.028 HDL3 pools/hr for mApoF versus 0.178±0.012 HDL3 pools/hr for LacZ (P<0.05; supplemental Figure VIB).
ApoF Overexpression and Reverse Cholesterol Transport
Because ApoF overexpression accelerated the plasma clearance of HDL-CE, we wondered whether ApoF would promote RCT. To address this possibility, we performed an in vivo RCT assay. Mice overexpressing Lac Z or mApoF were injected with [3H]-cholesterol labeled J774 macrophages and followed for 48 hours. ApoF overexpression reduced HDL-C levels in these mice by 21% relative to control animals (P<0.001), consistent with previous observations (data not shown). Surprisingly, despite a markedly reduced HDL peak in the FPLC profile, the 3H counts in HDL were almost identical to control animals at 48 hours after macrophage injection (Figure 4A). This suggests that despite lower HDL cholesterol levels, ApoF-containing HDL is still able to accept nearly normal levels of macrophage-derived cholesterol. Plasma counts of [3H]-cholesterol were not significantly different throughout the experiment (Figure 4B). This might be expected as plasma counts are a result of both macrophage cholesterol efflux, and clearance of HDL cholesterol by the liver. Interestingly, liver counts were significantly increased with ApoF overexpression (34%, P<0.05; Figure 4C), consistent with the HDL kinetic study. Bile and Fecal 3H counts were not significantly different between groups (Figure 4D and 4E), and notably were not decreased with ApoF despite the reduction in HDL cholesterol.
ApoF and Macrophage Cholesterol Efflux
To characterize the functional activity of HDL from control and ApoF overexpressing mice, we examined the capacity of plasma to promote cholesterol efflux from bone marrow–derived macrophages. Despite a roughly 20% decrease in HDL cholesterol and phospholipids, plasma from mice overexpressing mouse ApoF demonstrated only 5% lower capacity to promote efflux from cells (P<0.05; Figure 5A). When free cholesterol efflux was normalized to HDL cholesterol concentration, plasma from ApoF animals showed 19% higher efflux (P<0.001). This higher per unit efflux capacity was attributable primarily to changes in ATP-binding cassette transporter A1 (ABCA1)-independent efflux, as determined by pretreatment with probucol (22% increase, P<0.001; Figure 5B). ABCA1-dependent efflux trended toward an increase per HDL-C, but this did not reach statistical significance. Additional results from experiments addressing the role of ApoF in cholesterol efflux as well as CETP inhibition are included in supplemental information.
We used AAVs based on serotype 8 to overexpress ApoF in mouse liver. Human and mouse ApoF overexpression both reduced HDL cholesterol levels, providing the first concrete in vivo evidence that ApoF is involved in HDL metabolism. Overexpression of mouse ApoF improved the clearance of HDL cholesterol ester from the plasma, and resulted in HDL particles that are more efficient acceptors of macrophage-derived cholesterol. In addition, we have gained important insights into the biology of ApoF as it relates to HDL cholesterol and RCT.
Secreted ApoF is known to arise from cleavage of a much larger precursor.3 We created ApoF variants that could not be cleaved to generate the mature peptide. Surprisingly, the proprotein was still secreted into the media. The presence of maturely glycosylated proprotein in the media suggests that ApoF is processed to the mature form at the cell surface or after secretion. Human ApoF contains the sequence RVGRS at the cleavage site, whereas the mouse sequence is RAKRS. This RAKR sequence in mouse is a consensus furin recognition sequence of the format R-X-R/K-R. It will be interesting to determine whether ApoF is a bona fide target of furin or if another proprotein convertase is responsible for its processing. Interestingly, the ApoF proprotein could be detected in plasma and on isolated HDL when overexpressed in mice (supplemental Figures III and IV). Although we did not detect the ApoF proprotein in human plasma, our data suggests that it can be secreted. Cleavage of the secreted proprotein may represent a regulatory mechanism for converting a potentially inactive version of ApoF to its active form.
ApoF is known to be heavily glycosylated with both O- and N-linked sugar moieties. Glycosylation is a requirement for full LTIP activity.12 Based on sequence, it has long been thought that asparagine 267 in the mature peptide is N-glycosylated.3 We have shown that the human ApoF protein is in fact variably N-glycosylated at N267, giving rise to the doublet seen at 29 kDa in plasma. In addition, we have shown that the ApoF proprotein is glycosylated at N118 and N139. These glycosylations are Endo H sensitive in cell lysates, but resistant in the media (supplemental Figure II). The presence of maturely glycosylated Endo H-resistant proprotein in the media supports the idea that ApoF is processed after secretion. It is curious that the human protein has evolved these 3 N-linked glycosylation sites, including 1 in the mature peptide. Despite 60% total amino acid identity, mouse and human ApoF have a great deal of sequence divergence near the C-terminus (supplemental Figure I). It is tempting to speculate that these differences may have evolved for the purpose of fine tuning CETP activity in humans.
In our experiments, human ApoF was strictly HDL-associated in wild-type mice. A small amount could be found on LDL in mouse strains lacking the LDL receptor (supplemental Figures VII and VIII). In agreement with this, ApoF in human plasma was found predominantly in the HDL3 fraction, with a very small amount on LDL (supplemental Figure IIIB). The presence ApoF on human HDL is consistent with very early reports.1–3,7 ApoF was originally purified and cloned using human HDL as the starting material.3 This conflicts with reports that ApoF is predominantly LDL-associated.5,8 A recent study has reconciled this discrepancy, confirming that 75% of the ApoF protein is present on HDL, and only about 25% on LDL.4 LTIP activity was found only in the LDL fractions in this report, whereas the HDL associated ApoF existed as part of a 470-kDa inactive complex with a density in the HDL3 range. These findings are in general agreement with our observations. It is noteworthy that we found ApoF on HDL3 and not HDL2 (supplemental Figure IIIB). This is surprising, because human ApoF preferred to associate with the larger HDL particles in the mouse profile (Figure 2B).4 As has been suggested,4 it is possible that the protein content of ApoF-containing particles places their density in the HDL3 range despite a larger size.
Overexpression of mouse and human ApoF both significantly reduced HDL cholesterol levels. This suggests that despite their differences, mouse and human ApoF share some overlapping functional properties. ApoF overexpression had no effect on SR-BI and ABCA1 mRNA or protein levels (supplemental Figure IX and X), ruling out obvious effects at the level of the liver that would impact HDL production or selective uptake. Because neither mouse nor human ApoF overexpression significantly altered the size of the HDL particles (by FPLC), it is unlikely that the effects are due to altered LCAT activity (Figure 2 and supplemental Figure V). Furthermore, the ratio of cholesteryl ester to free cholesterol did not change with ApoF overexpression (data not shown), nor was the lipid composition of the HDL from these mice affected (supplemental Table II).
ApoF overexpression did not substantially change ApoAI or Apo E levels in wild-type mice as assessed by western blotting (supplemental Figure XI). ApoF was overexpressed in human ApoAI transgenic mice to more precisely measure ApoAI levels. Human ApoF overexpression lowered HDL levels by 25% relative to mice receiving the control virus (Lac Z). This drop in HDL cholesterol was accompanied by a 19% drop in ApoAI levels relative to baseline measurements (supplemental Figure XIIA). The reduction in ApoAI was observed in both the HDL and nonlipoprotein fractions in these mice (supplemental Figure XIIB). It is conceivable that the tight association of ApoF with the HDL particle displaces ApoAI, and thereby promotes catabolism of the particle. This seems less likely as the HDL particle size was not altered (by FPLC), and free ApoAI levels (in the nonlipoprotein fraction) were also reduced to a similar extent in these mice (supplemental Figure XIII).
Overexpression of ApoF accelerated the plasma clearance of HDL cholesteryl ether, as reflected by a 24% increase in the fractional catabolic rate. This finding is in agreement with the increased liver counts observed in the RCT study. Because SR-BI and LDL receptor protein levels were not affected (supplemental Figure X), the increased liver uptake is probably attributable to the presence of ApoF on HDL. ApoF may improve the HDL particle’s interaction with SR-BI, or possibly even serve as a weak ligand for another liver-expressed receptor. ApoF could modify the particle in such a way that uptake of CE is improved. Alternatively, ApoF is known to form particles in the HDL density range either along with ApoA-I, ApoA-I, and Apo-II, or with apolipoprotein F as the sole apolipoprotein.1 Perhaps these ApoF-only particles can also play a unique role in shuttling CE to the liver.
Plasma from mice overexpressing ApoF was a better acceptor for macrophag-derived cholesterol on a per HDL-C basis. This increased efflux capacity was attributable primarily to changes in ABCA1 independent efflux, as would be expected from the presence of ApoF on the larger HDL fractions. In loaded J774 macrophages, HDL isolated from mice overexpressing human ApoF was a better acceptor in cholesterol efflux assays (supplemental Figures XIV and XV). The nonlipoprotein fraction, containing free human ApoAI as well as the free human ApoF protein, showed significantly reduced efflux relative to that from control mice. Taken together, these results indicate that ApoF overexpression produces HDL particles that are more efficient acceptors of cholesterol from the macrophage.
ApoF overexpression consistently lowered HDL cholesterol levels in mice. Surprisingly, this did not adversely affect the rate of RCT, as mice overexpressing ApoF had similar counts of macrophage-derived cholesterol in the feces. This paradox can be explained by 2 factors: (1) an increased per unit capacity of HDL to accept cholesterol in cases of ApoF overexpression, and (2) accelerated clearance of HDL-CE from the plasma. Increased flux of macrophage-derived cholesterol through the plasma compartment was able to maintain RCT at levels seen in control mice, despite lower steady state levels of HDL.
We have demonstrated that ApoF overexpression reduces HDL cholesterol levels in mice—the first in vivo data implicating ApoF in HDL metabolism. Furthermore, our results indicate that ApoF has favorable effects on the reverse cholesterol transport pathway which are independent of its predicted effects on CETP function. ApoF overexpression increased clearance of CE from the plasma, and resulted in HDL particles that were better acceptors for macrophage-derived cholesterol. Based on the data, we believe that high levels of ApoF would be atheroprotective. It will be important to determine whether ApoF levels or genetic variants have a relationship to coronary risk in humans.
The rabbit polyclonal antibody used to detect the LDL receptor was a gracious gift of Dr Gene C. Ness.
Sources of Funding
This work was supported by NIH grants 5-P01-HL-059407-09 and 5R01-HL 055323-11.
Original received June 17, 2008; final version accepted October 3, 2008.
Koren E, McConathy WJ, Alaupovic P. Isolation and characterization of simple and complex lipoproteins containing apolipoprotein F from human plasma. Biochemistry. 1982; 21: 5347–5351.
Olofsson SO, McConathy WJ, Alaupovic P. Isolation and partial characterization of a new acidic apolipoprotein (apolipoprotein F) from high density lipoproteins of human plasma. Biochemistry. 1978; 17: 1032–1036.
He Y, Greene DJ, Kinter M, Morton RE. Control of cholesteryl ester transfer protein activity by sequestration of lipid transfer inhibitor protein in an inactive complex. J Lipid Res. 2008.
Wang X, Driscoll DM, Morton RE. Molecular cloning and expression of lipid transfer inhibitor protein reveals its identity with apolipoprotein F. J Biol Chem. 1999; 274: 1814–1820.
Nishide T, Tollefson JH, Albers JJ. Inhibition of lipid transfer by a unique high density lipoprotein subclass containing an inhibitor protein. J Lipid Res. 1989; 30: 149–158.
Morton RE, Greene DJ. Regulation of lipid transfer between lipoproteins by an endogenous plasma protein: selective inhibition among lipoprotein classes. J Lipid Res. 1994; 35: 836–847.
Serdyuk AP, Morton RE. Lipid transfer inhibitor protein activity deficiency in normolipidemic uremic patients on continuous ambulatory peritoneal dialysis. Arterioscler Thromb Vasc Biol. 1997; 17: 1716–1724.
Serdyuk AP, Morton RE. Lipid transfer inhibitor protein defines the participation of lipoproteins in lipid transfer reactions: CETP has no preference for cholesteryl esters in HDL versus LDL. Arterioscler Thromb Vasc Biol. 1999; 19: 718–726.
Morton RE, Nunes V, Izem L, Quintao E. Markedly elevated lipid transfer inhibitor protein in hypercholesterolemic subjects is mitigated by plasma triglyceride levels. Arterioscler Thromb Vasc Biol. 2001; 21: 1642–1649.
Morton RE, Gnizak HM, Greene DJ, Cho KH, Paromov VM. Lipid transfer inhibitor protein (apolipoprotein F) concentration in normolipidemic and hyperlipidemic subjects. J Lipid Res. 2008; 49: 127–135.
Vaisar T, Pennathur S, Green PS, Gharib SA, Hoofnagle AN, Cheung MC, Byun J, Vuletic S, Kassim S, Singh P, Chea H, Knopp RH, Brunzell J, Geary R, Chait A, Zhao XQ, Elkon K, Marcovina S, Ridker P, Oram JF, Heinecke JW. Shotgun proteomics implicates protease inhibition and complement activation in the antiinflammatory properties of HDL. J Clin Invest. 2007; 117: 746–756.
Lagor WR, Heller R, de Groh ED, Ness GC. Functional analysis of the hepatic HMG-CoA reductase promoter by in vivo electroporation. Exp Biol Med (Maywood). 2007; 232: 353–361.
Tietge UJ, Maugeais C, Cain W, Grass D, Glick JM, de Beer FC, Rader DJ. Overexpression of secretory phospholipase A(2) causes rapid catabolism and altered tissue uptake of high density lipoprotein cholesteryl ester and apolipoprotein A-I. J Biol Chem. 2000; 275: 10077–10084.
Tanigawa H, Billheimer JT, Tohyama J, Zhang Y, Rothblat G, Rader DJ. Expression of cholesteryl ester transfer protein in mice promotes macrophage reverse cholesterol transport. Circulation. 2007; 116: 1267–1273.
Kitajima K, Marchadier DH, Miller GC, Gao GP, Wilson JM, Rader DJ. Complete prevention of atherosclerosis in apoE-deficient mice by hepatic human apoE gene transfer with adeno-associated virus serotypes 7 and 8. Arterioscler Thromb Vasc Biol. 2006; 26: 1852–1857.