Effects of Nonlipolytic Ligand Function of Endothelial Lipase on High Density Lipoprotein Metabolism in Vivo*

Endothelial lipase (EL) influences high density lipoprotein (HDL) metabolism in vivo and mediates bridging and uptake of HDL particles independent of its lipolytic activity in vitro. To determine whether EL has a nonlipolytic ligand function in HDL metabolism in vivo, 1 × 1011 particles of a recombinant adenovirus encoding human EL (AdEL), catalytically inactive human EL (AdELS149A), or control (Adnull) were injected into wild-type, apoA-I transgenic, and hepatic lipase knockout mice. ELS149A protein was expressed at higher levels than wild-type EL. EL and ELS149A protein were both substantially increased in the postheparin plasma compared with preheparin, indicating that both the wild-type and mutant EL were bound to cell-surface heparan sulfate proteoglycans. Overexpression of wild-type EL was associated with a significantly increased postheparin-plasma phospholipase activity and dramatically decreased levels of total cholesterol, HDL cholesterol, phospholipids, and apoA-I. Injection of AdELS149A did not result in increased phospholipase activity confirming that ELS149A was catalytically inactive. Expression of ELS149A did not decrease lipid or apoA-I levels in wild-type and apoA-I transgenic mice yet led to an intermediate reduction of total cholesterol, HDL cholesterol, and phospholipids in hepatic lipase-deficient mice compared with control and EL-expressing mice. Our study demonstrates for the first time that EL has both a lipolytic and nonlipolytic function in HDL metabolism in vivo. Lipolytic activity of EL, however, seems to be most important for its effects on systemic HDL metabolism.

LPL is directly associated with HDL cholesterol levels in humans (1), but the effects of LPL on HDL metabolism are thought to be largely indirect through its effects on triglyceride-rich lipoproteins. HL was shown to directly modulate HDL metabolism by converting larger HDL to smaller HDL particles through hydrolysis of both triglycerides and phospholipids (2). Independent of their lipolytic activity, LPL and HL can act in cellular lipoprotein metabolism as ligands that mediate the binding and uptake of lipoproteins via proteoglycans and/or receptor pathways. Overexpression of catalytically inactive LPL in transgenic mice resulted in increased triglyceride-rich lipoprotein particle uptake and reduced triglyceride levels (3,4). Overexpression of catalytically inactive HL significantly lowered apoB-containing lipoproteins in apoE and LDL receptor knockout mice (5,6). Reduction of HDL levels by catalytically inactive HL was only apparent in mice that were HLdeficient (5)(6)(7).
Endothelial lipase (EL) is a new member of the lipase gene family (8,9). Overexpression as well as loss-of-function studies suggest that EL, like LPL and HL, plays an important role in lipoprotein metabolism. Hepatic overexpression of EL using an adenoviral vector resulted in markedly reduced HDL cholesterol levels in mice (8). Transgenic overexpression of EL under the control of the endogenous promoter resulted in modestly reduced HDL cholesterol levels (10). Conversely, inhibition of mouse EL activity in wild-type, apoA-I, and HL knockout mice using a specific antibody resulted in significantly increased HDL cholesterol and phospholipid levels (11). In the EL knockout mouse model, total cholesterol, HDL cholesterol, and phospholipids were significantly increased (10,12).
Recent in vitro data demonstrated that EL mediates cellular binding and uptake of HDL particles (13,14) and the selective uptake of HDL-associated cholesteryl esters independent of its lipolytic properties (13). We therefore hypothesized that expression of catalytically inactive EL in vivo would result in decreased HDL cholesterol and apoA-I levels. We mutated the putative active site serine 149 of EL to alanine (EL S149A ). This mutation produces a normally expressed full-length but catalytically inactive enzyme. We then used a recombinant adenovirus to express catalytically inactive EL S149A in wild-type, apoA-I transgenic, and HL-deficient mice. Unexpectedly, in contrast to wild-type EL, overexpression of catalytically inactive EL S149A did not result in reduced HDL and apoA-I levels in wild-type and apoA-I transgenic mice. In HL-deficient mice, however, overexpression of catalytically inactive EL moderately reduced total cholesterol, HDL cholesterol, and phospholipid levels, consistent with the concept that EL, like LPL and HL, can modulate lipoprotein metabolism in vivo independent of its lipolytic function. Catalytic activity of EL, however, appears to be the main mechanism by which EL lowers HDL levels.
In Vitro Transfection-HEK293 cells were transfected with EL, EL S149A , or a control encoding plasmid (green fluorescent protein) using LipofectAMINE (Invitrogen) according to the manufacturer's protocol. 20 h after transfection, the transfection mixture was replaced with serum-free medium containing heparin (10 units/ml). Heparin (10 units/ml) was added again 47.5 h after transfection, and media were harvested after 48 h.
Recombinant Adenovirus-The recombinant adenoviruses AdEL and Adnull were generated as described previously (8). To generate AdEL S149A , EL S149A cDNA was subcloned into HindIII-XbaI restriction sites of pAdCMVlink (pAdCMVEL S149A ) (15). pAdCMVEL S149A was linearized with NheI and cotransfected into HEK293 cells along with adenoviral DNA digested with ClaI. Viruses were propagated in HEK293 cells, and purified by cesium chloride ultracentrifugation. The purified virus was stored in 10% glycerol/phosphate-buffered saline at Ϫ80°C.
Animal Study-Wild-type C57BL/6 mice were obtained from Taconic Farms; human apoA-I transgenic (C57BL/6-TgN(APOA1)1Rub) and HL-deficient mice (B6.129P2-Lipc tm1Unc ) were from The Jackson Laboratory. All mice were fed a chow diet. Mice (n ϭ 4/group) were injected intravenously via the tail vein with 1 ϫ 10 11 particles of AdEL, AdEL S149A , or Adnull. Blood was drawn from the retroorbital plexus after 4 h of daytime fasting before and several time points after virus injection. Post-heparin plasma was obtained at day 3 after virus injection 5 min after intravenous injection of 100 units/kg heparin.
Lipid and Lipoprotein Analysis-Plasma total cholesterol, HDL cholesterol, and phospholipid levels were measured enzymatically on a Cobas Fara II (Roche Diagnostics) using Sigma reagents (Sigma). Plasma apoA-I levels were quantified using an immunoturbidometric assay (Sigma) on the Cobas Fara.
Gel Filtration Chromatography of Total Plasma-Pooled plasma samples (120 l) were subjected to fast protein liquid chromatography (FPLC) gel filtration (Amersham Biosciences) using two Superose 6 columns in series as described (8). Fractions (0.5 ml) were collected, and cholesterol concentrations were determined using an enzymatic assay (Wako Pure Chemical Industries).
Western Blot Analysis of EL Expression-Mouse plasma samples after heparin-Sepharose treatment or conditioned media from HEK293 cells were subjected to 10% SDS-PAGE and transferred to a nitrocellulose membrane (Hybond ECL, Amersham Biosciences), and human EL was detected with rabbit anti-EL peptide sera (1:3000) and goat anti-rabbit peroxidase-conjugated antisera (1:5000) as the secondary antibody (8).
Lipase Assays-Triglyceride lipase and phospholipase activity were measured as described previously (16). Triglyceride lipase activity was measured according to a modification of the method of Nilsson-Ehle and Schotz (17). The assay tubes contained, in a total volume of 0.3 ml, 0.05 M Tris-HCl, pH 8.0, 0.75% bovine serum albumin, 3.4 mM triolein, ϳ250 M phosphatidylcholine, and culture medium or mouse plasma. Samples were incubated for 1 h at 37°C. Reactions were stopped and products were extracted by the method of Belfrage and Vaughan (18).  (18) except that 100 g of lysopalmitoylphosphatidylcholine per ml was included as carrier in the organic extraction mix. Liberation of more than 2.5 nmol of FFA/h and more than 1.5 nmol of FFA/h compared with background for the triglyceride lipase and the phospholipase assay, respectively, are consistently statistically significantly different from background (p Ͻ 0.05).
Statistical Analysis-Lipase assay results are presented as means Ϯ S.E. Graphs are presented as means Ϯ S.D.

RESULTS
The lipolytic properties of wild-type EL and EL S149A were first analyzed in vitro. HEK293 cells were transfected with EL, EL S149A , or control. EL and EL S149A protein were expressed at very similar levels ( Fig. 1), but EL S149A had no detectable triglyceride or phospholipase activity (Table I), confirming that EL S149A was catalytically inactive.
To determine the effect of EL S149A on HDL metabolism in vivo, we injected wild-type mice with 1 ϫ 10 11 particles of an adenovirus encoding EL (AdEL), EL S149A (AdEL S149A ), or no transgene (Adnull). Protein expression was determined by Western blot analysis in the pre-and postheparin plasma at day 3 after virus injection (Fig. 2). EL and EL S149A protein were both substantially increased in the postheparin plasma compared with preheparin, indicating that both the wild-type and mutant EL were bound to cell-surface heparan sulfate proteoglycans (HSPGs). Overexpression of EL resulted in significantly increased phospholipase activity in postheparin plasma. In contrast, injection of AdEL S149A did not result in altered triglyceride or phospholipase activity (Table II). The changes in plasma total cholesterol, HDL cholesterol, and phospholipids during the course of the study are depicted in Fig. 3, A-C. As   shown previously (8), overexpression of wild-type EL resulted in a marked reduction of all plasma lipids lasting over the course of the study. In contrast, overexpression of EL S149A did not result in decreased total cholesterol or phospholipid levels compared with base line and showed slight reduction of HDL cholesterol only at day 5 after virus injection. Plasma lipoprotein profiles determined at day 10 after virus injection by FPLC analysis showed no reduction of HDL cholesterol in AdEL S149Ainjected mice compared with control (Fig. 3D). In order to determine the effects of this human catalytically inactive EL S149A on human-type HDL particles, we injected human apoA-I transgenic mice with 1 ϫ 10 11 particles AdEL, AdEL S149A , or Adnull (Fig. 4). Table III summarizes the postheparin triglyceride lipase and phospholipase activities at day 3 after virus injection. As expected, EL-expressing mice showed significantly increased phospholipase activity in post-heparin plasma, whereas expression of EL S149A did not result in increased phospholipase activity compared with control. Overexpression of wild-type EL resulted in significantly decreased total cholesterol, HDL cholesterol, phospholipid and apoA-I levels (Fig. 5, A-D). Overexpression of catalytically inactive EL, however, did not result in reduction of any of these parameters over the course of the study.
Dugi et al. (7) reported that expression of catalytically inactive HL resulted in reduced HDL levels only in mice that were HL-deficient, suggesting that the presence of endogenous mouse HL may account for the lack of effect observed in other mouse models (5,6). We therefore tested the hypothesis that endogenous mouse HL may also mask the effects of catalytically inactive EL, as hepatic EL and HL may have overlapping functions in HDL metabolism. HL knockout mice were injected with 1 ϫ 10 11 particles of AdEL, AdEL S149A , or Adnull (Fig. 6). As seen in wild-type and apoA-I transgenic mice, EL-expressing mice showed increased phospholipase activity, whereas expression of catalytically inactive EL did not result in increased lipase activity compared with control (Table IV). Overexpression of wild-type EL resulted in a dramatic and sustained reduction of total cholesterol, HDL cholesterol, and phospholipid levels. Interestingly, expression of catalytically inactive EL in HL-deficient mice led to an intermediate reduction of total cholesterol, HDL cholesterol, and phospholipid levels compared with that seen in wild-type EL expressing mice and control (Fig. 7, A-C), lasting over the course of the study. Plasma lipoprotein profiles determined at day 5 after virus injection confirmed the reduction of HDL cholesterol in AdEL S149A -expressing mice compared with control (Fig. 7D).

DISCUSSION
Our study demonstrates the following: 1) mutation of the putative active site serine 149 of wild-type EL to alanine generates a catalytically inactive yet normally expressed fulllength EL S149A , confirming serine 149 to be part of the catalytic triad of EL (Ser 149 -Asp 173 -His 254 ); 2) EL can be released into the plasma by heparin, suggesting that EL, like LPL and HL, is bound to cell-surface HSPGs; and 3) EL has both a lipolytic and a nonlipolytic function in the metabolism of HDL in vivo, yet lipolytic activity of EL appears to be the main mechanism by which EL modulates systemic HDL levels.
EL is a member of the triglyceride lipase gene family comprising pancreatic lipase, LPL and HL. Recent results from overexpression and loss-of-function studies suggest that EL, like other members of this gene family, plays an important role in lipoprotein metabolism (8, 10 -12). LPL and HL are bound to HSPGs and act there to either hydrolyze lipoproteins or mediate uptake of lipoproteins independent of their lipolytic activity. The clusters of positively charged residues in LPL and HL that have been implicated in heparin binding are highly conserved in EL (8,9). In this study, we demonstrated for the first time that EL can be released into the plasma by heparin in vivo, suggesting that EL, like LPL and HL, is anchored to the luminal endothelial surface via HSPGs. Release of EL from HSPGs is accompanied by significantly increased phospholipase activity levels.
Recent in vitro data suggested that EL can promote HDL binding (13,14), holoparticle uptake (13), and selective uptake of cholesteryl esters (CE) independent of its lipolytic activity (13). To determine whether lipolytic activity of EL is required for its effects on HDL metabolism in vivo, we used adenoviral vectors to overexpress wild-type and catalytically inactive EL S149A in several different mouse models. EL S149A was well expressed and could be released by heparin to the same extent as wild-type EL, suggesting that neither expression nor proteoglycan binding of the mutated EL was impaired. Overexpression of wild-type EL resulted in significantly reduced levels of total cholesterol, HDL cholesterol, and phospholipids in all mouse models examined. Unexpectedly, despite even higher protein expression levels compared with wild-type EL, overexpression of EL S149A did not result in reduction of total cholesterol, HDL cholesterol, phospholipid, or apoA-I levels in wildtype and apoA-I transgenic mice. Therefore, lipolytic activity of endothelial lipase appears to be very important for its effects on HDL metabolism. However, the possibility that catalytically inactive EL S149A may serve as a ligand to mediate the binding and uptake of lipoproteins to an extent that does not affect systemic HDL levels in these two mouse models cannot be ruled out. Importantly, overexpression of catalytically inactive EL in HL knockout mice resulted in a significant reduction of   Dugi et al. (7) previously reported that reduction of HDL levels by catalytically inactive HL was only apparent in mice that were HL-deficient. Our data suggest that endogenous mouse HL not only masks the effects of catalytically inactive HL but also accounts for the lack of effect of catalytically inactive EL observed in wild-type and apoA-I transgenic mice. The presence of endogenous active HL may interfere with the effects of hepatic expression of catalytically inactive EL, as EL and HL may have overlapping functions in HDL metabolism. Both EL and HL were shown to mediate binding of HDL to cell-surface HSPGs in vitro (14) and to hydrolyze HDL lipids in vitro (16) and in vivo (5-8, 10, 19, 20). The specific roles of HL versus EL in HDL metabolism are currently not very well understood. We recently proposed that HL may primarily act on triglyceride-enriched HDL particles resulting in reduction in size and increased uptake of cholesteryl esters as well as remnant HDL 2 particles, whereas EL may act primarily on HDL phospholipids resulting in dissociation of apolipoproteins and subsequent catabolism (21). The results of the present study, however, indicate that EL and HL have at least partially overlapping functions in HDL metabolism. Studies in EL-deficient and EL/HL double knockout mice may contribute to a better understanding of the specific roles of HL and EL in HDL metabolism.
In summary, in the present study we tested the hypothesis that EL can alter HDL metabolism independent of its lipolytic function. Overexpression of wild-type EL dramatically lowered total cholesterol, HDL cholesterol, phospholipid, and apoA-I levels in all mice models examined. Expression of catalytically inactive EL S149A did not reduce HDL cholesterol in wild-type and apoA-I transgenic mice but resulted in a moderate reduc-tion of total cholesterol, HDL cholesterol, and phospholipids in mice that were HL-deficient. We therefore conclude that EL can modulate HDL metabolism independent of its lipolytic function, yet the lipolytic activity of EL appears to be the main determinant for its effects on HDL metabolism.