Targeted deletion of endothelial lipase increases HDL particles with anti-inflammatory properties both in vitro and in vivo.

Previous studies have shown that targeted deletion of endothelial lipase (EL) markedly increases the plasma high density lipoprotein cholesterol (HDL-C) level in mice. However, little is known about the functional quality of HDL particles after EL inhibition. Therefore, the present study assessed the functional quality of HDL isolated from EL−/− and wild-type (WT) mice. Anti-inflammatory functions of HDL from EL−/− and WT mice were evaluated by in vitro assays. The HDL functions such as PON-1 or PAF-AH activities, inhibition of cytokine-induced vascular cell adhesion molecule-1 expression, inhibition of LDL oxidation, and the ability of cholesterol efflux were similar in HDL isolated from WT and EL−/− mice. In contrast, the lipopolysaccharide-neutralizing capacity of HDL was significantly higher in EL−/− mice than that in WT mice. To evaluate the anti-inflammatory actions of HDL in vivo, lipopolysaccharide-induced systemic inflammation was generated in these mice. EL−/− mice showed higher survival rate and lower expression of inflammatory markers than WT mice. Intravenous administration of HDL isolated from EL−/− mice significantly improved the mortality after lipopolysaccharide injection in WT mice. In conclusion, targeted disruption of EL increased HDL particles with preserved anti-inflammatory and anti-atherosclerotic functions. Thus, EL inhibition would be a useful strategy to raise ‘good’ cholesterol in the plasma.

method ( 12 ) followed by the calculation of cholesterol effl ux as previously reported ( 13 ).

Plasma cytokine and corticosterone determination EL
Ϫ / Ϫ and WT mice were given an intraperitoneal injection of LPS (40 mg/kg) or vehicle. Two hours later, the plasma was isolated and immediately frozen at Ϫ 80°C. Plasma levels of cytokine and corticosterone levels were determined by using Q-plex ® mouse cytokine array (Biolegend, Santa Fe, CA) and a Coat-A-Count corticosterone kit (Diagnostic Products, Los Angeles, CA), respectively.
Peritoneal macrophages were collected from WT and EL Ϫ / Ϫ mice 4 days after an intraperitoneal injection of 1 ml of 4% thioglycollate medium. The mouse peritoneal macrophages or human THP-1 cells were stimulated with LPS, and mouse or human TNF-␣ production in the culture medium was measured using a commercially available kit from R&D Systems (Minneapolis, MN) or Bio source (Camarillo, CA), respectively.

Immunoblot and histological analyses
Western blotting was performed using anti-VCAM-1 (Santa Cruz Laboratory, Santa Cruz, CA), inducible nitric oxide synthase (iNOS) (BD Transduction Laboratories, Bedford, MA), apolipoprotein (apo)A-1 (Abcam, Cambridge, UK), or anti-␤ -actin antibody (Sigma-Aldrich) as described previously ( 14 ). The relative densities of the acquired bands were determined by using the ImageJ software version 1.38 from the National Institutes of Health (available on the World Wide Web).
For histology, mice were euthanized with an overdose of pentobarbital, and the lungs were removed followed by the infl ation at a pressure of 20 cm water with 4% paraformaldehyde/PBS. Lungs were fi xed overnight, embedded in paraffi n, sectioned, and stained with hematoxylin and eosin.

Real-time PCR
Quantitative real-time PCR was performed as previously reported ( 14 ). PCR primers for mouse TNF-␣ and GAPDH were purchased from Takara-Bio Perfect Real Time Support System (Takara, Shiga, Japan).

Lung wet-to-dry weight ratio and myeloperoxidase activity
A lung wet-to-dry weight (W/D) ratio was used as a parameter of lung water accumulation after the LPS injection as previously described ( 11 ). Briefl y, lung wet weight was determined immediately after removal of the lung. Lung dry weight was determined after the lung had been completely dried in an oven at 50°C for 24 h. Lung myeloperoxidase (MPO) activity was analyzed as previously reported ( 11 ).

Hemodynamic and echocardiographic studies
Six hours after the LPS (20 mg/kg) injection, heart rate and systolic blood pressure were measured by the tail-cuff method. Transthoracic 2-dimensional echocardiography (SONOS 5500, Philips, Andover, MA) was performed under light anesthesia with Avertin (0.005 ml/g of 2.5% solution, ip) 8 h following the LPS (20 mg/kg) challenge ( 15 ).

Assays for evaluating the HDL binding to LPS
An in vitro assay was used to quantify the LPS-binding capacity of HDL as previously reported ( 16 ). In brief, isolated HDL was and promote infl ammatory processes and is referred to as dysfunctional HDL ( 2 ). These lines of evidence have stimulated tremendous interest in the functional quality and therapeutic potential of HDL.
Endothelial lipase (EL) is a new member of the lipase gene family ( 4,5 ). EL exhibits preferential substrate specifi city for phospholipids on HDL particles and promotes HDL catabolism ( 6 ). In fact, EL-defi cient mice ( EL Ϫ / Ϫ ) showed an elevated plasma HDL-C level, whereas EL overexpression results in a reduced HDL-C level ( 7,8 ). These fi ndings suggest that a selective inhibitor of EL, if available, would be useful to increase plasma HDL-C levels, although little is known concerning the functional qualities of HDL particles after EL inhibition. Therefore, the present study focused on the HDL quality after EL disruption both in in vitro and in vivo infl ammatory processes.

Cell culture
Human umbilical vein endothelial cells (HUVECs), THP-1, and J774 cells were purchased from the American Type Culture Collection (Manassas, VA). For the stimulation assay, subconfl uent HUVEC was treated with fresh complete medium containing 10 ng/ml tumor necrosis factor (TNF)-␣ (Biosource, Camarillo, CA) for 24 h in the presence of mouse plasma (50 µL) or HDL (250 µg of protein).

Murine model of LPS-induced endotoxin shock
All animal studies were performed in accordance with the Institutional Guidelines of Kobe University and the Institute for Laboratory Animal Research (ILAR) Guide for the Care and Use of Laboratory Animals. Eight-to twelve-week-old male EL Ϫ / Ϫ mice ( 7 ), which were bred with the C57Bl/6 background more than 25 times, and age-matched male wild-type (WT) C57BL/6 mice (Japan Charles River, Osaka, Japan) were used in this study. The body weight of EL Ϫ / Ϫ and WT mice at 8 weeks of age was 24.5 ± 2.4 and 27.9 ± 0.9 g (NS, not signifi cant), respectively, and that at 12 weeks of age was 31.0 ± 1.6 and 33.2 ± 3.0 g (NS), respectively. Plasma was obtained from WT mice (Plasma-WT) and ) by cardiac puncture, and HDL fraction was obtained from the pooled plasma of WT mice (HDL-WT) and EL Ϫ / Ϫ mice (HDL-EL Ϫ / Ϫ ) by ultracentrifugation as described elsewhere ( 9 ). A mouse model of LPS-induced endotoxin shock was generated as described previously ( 10,11 )

Cholesterol effl ux study
J774 cells were incubated in DMEM containing 0.5% FBS with an Liver X receptor agonist (T0, 3 mol/L; Sigma-Aldrich) and acetylated low density lipoprotein (50 g protein/ml) for 24 h. Then, cholesterol effl ux was performed for 8 h at 37°C in DMEM/0.2% BSA with the 250 g/ml of HDL-WT or HDL-EL Ϫ / Ϫ . At the end of the 8 h incubation, the lipid fractions were extracted from the effl ux medium or cells by the Bligh and Dyer Next, we assessed the effect of HDL on Cu 2+ -induced oxidation of LDL and found no difference in the anti-oxidative effects between HDL-EL Ϫ / Ϫ and HDL-WT ( Fig. 1B ); the lag time in HDL-EL Ϫ / Ϫ and HDL-WT was 44.8 ± 5.6 versus 46.7 ± 4.5 s; the T max was 107.1 ± 6.9 versus 115.7 ± 2.2 s, respectively. Furthermore, we checked the plasma activities of HDL-associated anti-oxidative enzymes including PON-1 and PAF-AH. Both paraoxonase and arylesterase activities were signifi cantly higher in the Plasma-EL Ϫ / Ϫ than those in Plasma-WT (top panels in Fig. 1C, D ), which appeared to be in proportion to the HDL-C levels in these mice. However, those activities in HDL were similar if the same amounts of HDL protein were evaluated (bottom panels in Fig. 1C, D ), which may refl ect the comparative antioxidative properties of HDL particles against Cu 2+ -induced LDL oxidation ( 21 ). Similarly, the PAF-AH activities were signifi cantly higher in Plasma-EL Ϫ / Ϫ than those in Plasma-WT, whereas those activities in HDL were similar if the same protein amounts of HDL-EL Ϫ / Ϫ or HDL-WT were evaluated ( Fig. 1E ). In addition, we assessed the ability of HDL-WT and HDL-EL Ϫ / Ϫ in cholesterol effl ux but observed no difference (19.1 ± 4.6, 24.1 ± 4.3%, P = NS).

LPS-neutralizing capacity of plasma and HDL are increased in EL ؊ / ؊ mice
It is well known that HDL can form a complex with LPS ( 22 ), which results in neutralization of its activity and attenuation of the subsequent release of infl ammatory cytokines including TNF-␣ , interleukin (IL)-1, and IL-6 ( 23 ). Therefore, we compared the LPS-neutralizing capacity utilizing LAL assays ( 24 ). This assay revealed that the plasma LPS-neutralizing capacity was higher in EL Ϫ / Ϫ mice than that in WT mice ( Fig. 2A , top panel). To determine the relative effects of the plasma HDL-C level on the LPS-neutralizing capacity compared with the alteration in HDL quality, we evaluated the LPS-neutralizing capacity in a same amount of protein, cholesterol, or phospholipids, of HDL-WT and HDL-EL Ϫ / Ϫ . Interestingly, HDL-EL Ϫ / Ϫ showed an increase in the LPS-neutralizing capacity when compared with immobilized for 2 h at 37°C on 96-well plates. Nonspecifi c binding was blocked with 1% BSA for 1 h at 37°C. LPS was biotinylated with EZ-link TM Biotin-LC-Hydrazide (Pierce, Rockford, IL) and incubated for 1 h at 37°C. After washing the plates, the bound LPS was detected by addition of streptavidin-peroxidase and tetramethylbenzidine (TMB).

2+ -induced LDL oxidation
Human LDL (density, 1.019-1.063 g:ml) was isolated from the plasma of healthy volunteers by sequential ultracentrifugation as described previously ( 18 ). LDL (100 g/ml) in PBS was incubated with freshly prepared CuSO4 (5 mol/L) at 37°C in the presence or absence of the isolated HDL (50 g/ml). Diene formation was measured as the increase in absorbance at 234 nm every 5 min, monitored by a spectrophotometer (UV-1600, Shimadzu). The lag time and maximum time ( T max ) were determined as described previously ( 18 ).

Analysis of HDL particles and fatty acid composition
Concentrations of phospholipids, cholesterol, and triglycerides in the isolated HDL fraction were measured using by the commercially available kits (Wako, Osaka, Japan). The apolipoprotein in the HDL fraction was analyzed by SDS-PAGE ( 7 ). Fatty acid composition of isolated HDL was performed using gas chromatography as reported elsewhere ( 19 ). The HDL particle size was analyzed by HPLC (LipoSEARCH ® ) from Skylight Biotech, Inc. (Akita, Japan).

Statistics
The results are expressed as the mean ± SEM. The signifi cance of the differences among the experimental groups was determined by a one-way ANOVA followed by the Bonferroni test for multiple comparisons. The level of statistical signifi cance was set at p < 0.05.

HDL isolated from EL ؊ / ؊ mice shows athero-protective properties in vitro
First, we compared plasma lipid profi le between WT and EL Ϫ / Ϫ mice. As shown in Table 1 , we confi rmed that EL Ϫ / Ϫ mice are 59% higher in HDL-C levels than WT mice. HDL-phospholipids were 78% higher in EL Ϫ / Ϫ mice than in WT mice. In addition, HDL particles in EL Ϫ / Ϫ mice were signifi cantly larger than those in WT mice. Thus, HDL-EL Ϫ / Ϫ was found to be rich in cholesterol and phospholipids with increased particle sizes, which is consistent with previous studies ( 7,8 ). Because HDL has been shown to inhibit expression of VCAM-1 induced by pro-infl ammatory cytokines ( 20 ), we checked the effect of HDL-WT or HDL-EL Ϫ / Ϫ on the cytokine-induced VCAM-1 expression in HUVEC. The VCAM-1 expression induced by TNF-␣ was markedly reversed by coincubation with HDL-WT and HDL-EL Ϫ / Ϫ ( Fig. 1A ). There was no difference in the inhibitory effect between HDL-WT and HDL-EL Ϫ / Ϫ . The HDL fraction was obtained by ultracentrifugation using pooled plasma, and levels of cholesterol, triglyceride, and phospholipid were determined by standard biochemical assays. The HDL particle size was measured by HPLC. a p < 0.05. b p < 0.01 versus WT mice (n = 8).
Plasma-EL Ϫ / Ϫ does not merely refl ect the increase in the plasma HDL-C or apoA-1 levels in EL Ϫ / Ϫ mice. There was no difference in the LPS-neutralizing capacity between HDL-EL Ϫ / Ϫ and HDL-WT ( Fig. 2A , third and fourth panels) when the equivalent cholesterol or phospholipids were evaluated.
Notably, apoA-1 content in HDL fraction was similar between HDL-EL Ϫ / Ϫ and HDL-WT ( Fig. 2B ) when the equivalent protein was evaluated by SDS-PAGE. The fi nding indicates that the increased LPS-neutralizing capacity of C-E: Activities of HDL-associated enzymes, PON-1 (paraoxanase, C; arylesterase, D) and platelet-activating factor acetylhydrolase (PAF-AH) (E), were determined in the plasma or HDL of WT or EL Ϫ / Ϫ mice. Bars represent mean ± SEM. * p < 0.05 (n = 10 for whole plasma, and n = at least 3 for HDL from pooled plasma).
or HDL -EL Ϫ / Ϫ resulted in a signifi cant decrease in the TNF-␣ production when compared with that observed with Plasma-WT or HDL-WT, respectively ( Fig. 2E ). Thus, HDL-EL Ϫ / Ϫ with high phospholipid content showed high binding affi nity with LPS and attenuated the bioactivity of LPS.

Analysis of fatty acid composition of isolated HDL fraction
To extend the understanding of the characteristics of HDL-WT and HDL-EL Ϫ / Ϫ , we assessed fatty acid composition using GC-MS. This high-throughput analysis revealed that EL defi ciency substantially alters the composition of fatty acids in HDL particles ( Fig. 3A ). Total fatty acid content was signifi cantly higher in HDL-EL Ϫ / Ϫ than HDL-WT ( Fig. 3B ), which was likely to refl ect the increased phospholipids content in HDL-EL Ϫ / Ϫ . Fatty acids are shown to modulate infl ammatory processes; saturated fatty acids (SFAs) directly stimulate TLR2or TLR4-receptor signaling ( 25 ), and n-3 PUFAs exert a range of anti-infl ammatory actions whereas n-6 PUFAs are a source of prothrombotic and pro-infl ammatory eicosanoids.
Previous studies have demonstrated that the phospholipid bilayer of the HDL surface can bind to the lipid-A, an anchor protein of LPS, to neutralize its activity ( 3 ). To clarify the mechanisms by which HDL-EL Ϫ / Ϫ has increased LPS-neutralizing capacity, therefore, we compared the lipid content in isolated HDL. HDL-EL Ϫ / Ϫ was rich in phospholipids and cholesterol (Table 1), whereas the apoA-1 content in HDL fraction was similar in HDL-EL Ϫ / Ϫ and HDL-WT ( Fig. 2B ). As a result, we found that the phospholipid-or cholesterol-content in the HDL particle, which was calibrated with apoA-1 content, was markedly higher in HDL-EL Ϫ / Ϫ than that in HDL-WT ( Fig. 2C ). Moreover, we directly assessed the interaction of HDL-EL Ϫ / Ϫ or HDL-WT with LPS. LPS binding capacity of HDL was signifi cantly higher in HDL-EL Ϫ / Ϫ than in HDL-WT ( Fig. 2D ). These fi ndings support the notion that the LPSbinding and neutralization are associated with the increase in the phospholipid-or cholesterol-content in the HDL particle. Furthermore, we checked the subsequent cytokine production by comparing LPS-induced TNF-␣ production in THP-1 cells. Preincubation of the LPS with Plasma-EL Ϫ / Ϫ

Fig. 2. LPS-neutralizing capacity of HDL was higher in EL
Ϫ / Ϫ mice. A: Mouse plasma (5 l) or HDL (10 g protein, 50 g cholesterol, or 50 g phospholipids) were incubated with 100 ng/ml of LPS at 37°C for 1 h, and lipopolysaccharide (LPS) neutralization was quantifi ed by the Limulus amebocyte lysate (LAL) assay. B: Apolipoprotein (apo)A-1 protein levels in HDL fractions were similar between HDL-WT and HDL-EL Ϫ / Ϫ when the same concentration of HDL protein (15 g protein) was evaluated by SDS-PAGE. C: Concentration of cholesterol (Cho), phospholipids (PL), and triglycerides (TG), as well as apoA-1 content in isolated HDL were determined, and the ratio of Cho/apoA-1, PL/apoA-1, or TG/ apoA-1 were expressed as a percent value of WT group. D: Isolated HDL was immobilized to 96-well plates and incubated with biotinylated LPS. Bound LPS was detected by peroxidase-conjugated streptavidin and tetramethylbenzidine (TMB). Binding of LPS to HDL is expressed as a value with OD450 nm. E: THP-1 cells were preincubated with plasma (20 l) or HDL (50 g) of WT or EL Ϫ / Ϫ mice and then treated with 1 g/ml of LPS. The TNF-␣ concentration in culture medium was analyzed by ELISA. Bars represent mean ± SEM. * p < 0.05 (n = 5-6).
production in peritoneal macrophages isolated from EL Ϫ / Ϫ and WT mice. The TNF-␣ production by EL Ϫ / Ϫ macrophages was similar to that by WT macrophages in response to LPS (2847.0 ± 119.6 vs. 2663.5 ± 128.3 pg/ml, p = NS). On the other hand, glucocorticoids exert anti-infl ammatory effects by suppressing pro-infl ammatory cytokines and stimulating anti-infl ammatory cytokines. Therefore, we measured plasma corticosterone levels following the LPS injection (40 mg/kg), but found no signifi cant difference be tween WT and EL Ϫ / Ϫ mice (528 ± 47.6 vs. 624 ± 29.7 ng/ml, p = NS).

LPS-induced lung damage was attenuated in EL
؊ / ؊ mice TNF-␣ mRNA expression in the mouse lung was analyzed following the LPS injection. The LPS injection (25 mg/ kg) substantially increased the expression of lung TNF-␣ mRNA levels both in WT and EL Ϫ / Ϫ mice. Notably, the LPS-induced TNF-␣ expression in EL Ϫ / Ϫ mice was lower than that in WT mice ( Fig. 5A ). In addition, protein levels of iNOS and VCAM-1 in the lungs were lower in EL Ϫ / Ϫ mice than WT mice after the LPS injection ( Fig. 5B, C ) Histological examination demonstrated that LPS induced an infi ltration of numerous polymorphonuclear leukocytes and macrophages in the interstitial spaces and marked swelling of the alveolar walls. These infl ammatory Interestingly, HDL-EL Ϫ / Ϫ was signifi cantly lower in SFA than HDL-WT ( Fig. 3C ). Moreover, both the eicosapentaenoic acid (EPA)/arachidonic acid (AA) ratio and the n-3/n-6 fatty acids ratio were signifi cantly higher in HDL-EL Ϫ / Ϫ than in HDL-WT ( Fig. 3D ). These results imply that antiinfl ammatory properties of HDL-EL Ϫ / Ϫ may be partly due to the increased content of the anti-infl ammatory fatty acids.

EL
؊ / ؊ mice are protected against endotoxin shock To evaluate the role of these anti-infl ammatory actions of HDL in vivo, LPS-induced systemic infl ammation was generated in WT and EL Ϫ / Ϫ mice. As depicted in Fig. 4A , we observed a marked improvement of survival rate in EL Ϫ / Ϫ mice in response to the relatively high dose (80 mg/ kg) of LPS when compared with WT mice. To evaluate the early infl ammatory response in these mice, plasma levels of infl ammatory cytokines were analyzed 2 h after the LPS injection (40 mg/kg). The LPS injection resulted in robust increases in levels of TNF-␣ , IL-1 ␤ , monocyte chemoattractant protein-1, and IFN-␥ both in WT and EL Ϫ / Ϫ mice ( Fig. 4B ). However, these cytokine levels were signifi cantly lower in EL Ϫ / Ϫ mice than in WT mice ( Fig. 4B ). To determine whether the lower levels of infl ammatory cytokines in EL Ϫ / Ϫ mice refl ect the decreased cellular production in response to LPS, we evaluated the LPS-induced TNF-␣

W/D ratio following the LPS injection was lower in EL
Ϫ / Ϫ mice than that in WT mice ( Fig. 5F ).

EL defi ciency attenuates LPS-induced cardiac dysfunction and hypotension
To compare the cardiac infl ammation and function, the TNF-␣ mRNA expression in the heart was analyzed following changes were attenuated in EL Ϫ / Ϫ mice ( Fig. 5D ). We assessed granulocyte infi ltration by MPO activity in the lung as well as lung edema by means of the W/D ratio. These parameters were low and similar in the two mouse groups at baseline ( Fig. 5E, F ). The LPS treatment increased the lung MPO activity, which was signifi cantly lower in EL Ϫ / Ϫ mice than that in WT mice ( Fig. 5E ). Moreover, the lung no difference in blood pressure between two groups at the baseline. However, blood pressure fell from 106.8 ± 5.5 to 90.6 ± 7.6 mmHg ( p < 0.05) by the LPS challenge in WT mice. In contrast, the blood pressure was not affected by the LPS treatment in EL Ϫ / Ϫ mice (from 108.7 ± 2.5 to 108.4 ± 2.5 mmHg, p = NS). The echocardiography revealed that the fractional shortening of the left ventricle after the LPS challenge was signifi cantly lower in WT than in EL Ϫ / Ϫ mice the LPS injection. The LPS injection (25 mg/kg) substantially increased the expression of the heart TNF-␣ mRNA both in WT and EL Ϫ / Ϫ mice, and the expression returned to the basal level 4 h after the LPS challenge ( Fig. 5G ). In EL Ϫ / Ϫ mice, the TNF-␣ induction was attenuated compared with WT mice 1 and 2 h after the LPS administration ( Fig.  5G ). Moreover, we measured blood pressure and performed echocardiographic examinations in these mice and found lipid-A binds to the phospholipids of the HDL surface ( 3 ).
In fact, the present study has shown that phospholipid content in HDL particles was higher in HDL-EL Ϫ / Ϫ than in HDL-WT, which is in line with past reports ( 8,26 ). Therefore, we speculate that the increased phospholipid content may increase the affi nity of the HDL particles to lipid-A and then neutralize bioactivity of LPS. Furthermore, HDL-EL Ϫ / Ϫ was rich in n-3 PUFAs including EPA and poor in SFA compared with HDL-WT, which is likely to make the HDL particles more anti-infl ammatory. The precise mechanism underlying the altered fatty acid composition remains unclear and needs to be clarifi ed by further studies. However, the change in HDL fatty acids suggests that the selectivity of EL phospholipase activity may depend on the fatty acid composition in the substrate HDL phospholipids. On the other hand, previous studies have indicated that apoA-1 itself can neutralize LPS ( 27 ). Because the apoA-1 content relative to the protein content was similar in HDL-WT and HDL-EL , it is considered that the anti-infl ammatory effect of HDL-EL Ϫ / Ϫ is not likely mediated by the change in apoA-1 content. It has been reported that suppression of EL in macrophages directly decreases the secretion of infl ammatory cytokine secretion ( 28,29 ). In the present study, however, there was no difference in TNF-␣ production by isolated macrophages between EL Ϫ / Ϫ and WT mice.
In endotoxin-induced systemic infl ammation, LPS is known to inhibit activities of LPL and HL to increase plasma levels of apoB-containing lipoproteins such as VLDL and its remnants (30)(31)(32). Also, LPS increases EL expression to decrease HDL levels ( 10 ). Thus, EL may modulate the severity of endotoxin shock through plasma HDL levels. Kitchens et al. ( 33 ) have reported that in spite of the decline in HDL levels, HDL remains the dominant LPS acceptor in septic patients, whereas LPS binding shifts to VLDL in some cases. They also reported that the LPS binding and neutralization are largely associated with phospholipid content in lipoprotein subclasses ( 33 ). These fi ndings support that EL disruption may not only increase HDL quantity but also improve HDL quality, and as a result, protect against endotoxin shock.

Administration of HDL-EL ؊ / ؊ improved survival in LPS-induced septic shock
We investigated if the improved survival in EL Ϫ / Ϫ mice was attributable to the increase in anti-infl ammatory HDL particles. WT mice were administered the same amount of HDL-EL Ϫ / Ϫ or HDL-WT from the tail vein followed by the LPS treatment and the survival rate was monitored. Both HDL-WT and HDL-EL Ϫ / Ϫ improved the survival rate in the septic shock model compared with vehicle treatment ( Fig. 6 ). Interestingly, the survival rate was signifi cantly higher in the case of HDL-EL Ϫ / Ϫ than in HDL-WT. These fi ndings indicate that the improved anti-infl ammatory function of HDL-EL Ϫ / Ϫ is relevant in the in vivo endotoxin shock model, and that EL deletion may increase HDL-C with augmented LPS-neutralizing capacity and preserved anti-infl ammatory properties.

DISCUSSION
In the present study, we showed that targeted disruption of EL results in an increase not only in plasma HDL-C levels but also in the HDL quality; HDL-EL Ϫ / Ϫ had a variety of anti-infl ammatory properties, in terms of inhibition of VCAM-1 expression, the HDL-associated anti-oxidative enzymatic activities, and the ability of cholesterol effl ux, whereas the anti-infl ammatory potency was similar between HDL-WT and HDL-EL Ϫ / Ϫ . Moreover, HDL-EL Ϫ / Ϫ exhibited a higher LPS-neutralizing capacity than HDL-WT. These anti-infl ammatory actions of HDL appear to be physiologically relevant in vivo because EL Ϫ / Ϫ mice showed an attenuation in infl ammatory responses after the LPS challenge, resulting in the increased survival rate against endotoxin shock.
Although many of the anti-infl ammatory functions of HDL were similar between EL Ϫ / Ϫ and WT mice, only LPSneutralizing capacity was more potent in HDL-EL Ϫ / Ϫ than in HDL-WT. Previous studies have demonstrated that ) were injected to WT mice through the tail vein (7.5 mg protein/kg), followed by the LPS administration (80 mg/kg, ip). Mice were carefully monitored, and survival rate at various times was recorded (n = 10 in each group, p < 0.05).