Ambient ultrafine particles alter lipid metabolism and HDL anti-oxidant capacity in LDLR-null mice.

Exposure to ambient particulate matter (PM) is a risk factor for cardiovascular diseases. The redox-active ultrafine particles (UFPs) promote vascular oxidative stress and inflammatory responses. We hypothesized that UFPs modulated lipid metabolism and anti-oxidant capacity of high density lipoprotein (HDL) with an implication in atherosclerotic lesion size. Fat-fed low density lipoprotein receptor-null (LDLR−/−) mice were exposed to filtered air (FA) or UFPs for 10 weeks with or without administering an apolipoprotein A-I mimetic peptide made of D-amino acids, D-4F. LDLR−/− mice exposed to UFPs developed a reduced plasma HDL level (P < 0.01), paraoxonase activity (P < 0.01), and HDL anti-oxidant capacity (P < 0.05); but increased LDL oxidation, free oxidized fatty acids, triglycerides, serum amyloid A (P < 0.05), and tumor necrosis factor α (P < 0.05), accompanied by a 62% increase in the atherosclerotic lesion ratio of the en face aortic staining and a 220% increase in the cross-sectional lesion area of the aortic sinus (P < 0.001). D-4F administration significantly attenuated these changes. UFP exposure promoted pro-atherogenic lipid metabolism and reduced HDL anti-oxidant capacity in fat-fed LDLR−/− mice, associated with a greater atherosclerotic lesion size compared with FA-exposed animals. D-4F attenuated UFP-mediated pro-atherogenic effects, suggesting the role of lipid oxidation underlying UFP-mediated atherosclerosis.

at 0.2 mg/ml in response to death from subcutaneous injection. The averaged water consumption of mice was 4.03 ml per day. There was no signifi cant difference in the weight of mice among groups. Five mice were initiated in the control groups (FA) and 10 in the UFP groups. One mouse died in the FA group, one in the FA+D-4F group, and two in the UFP group during the fi rst week of subcutaneous D-4F injection. Another mouse in the UFP+D-4F group died from a wound infection due to fi ghting at week 8 (supplementary Table II ) .

Lipid profi ling
Total cholesterol was measured by enzymatic assay using a 96well microplate-based absorbance method in accordance with the manufacturer's protocol (Thermo Trace/DMA reagent #TR13303; Fisher #NC9767174, Fisher Diagnostics, Fremont, CA; cholesterol standard 300 mg/dl Fisher #NC9343697, Thermo Scientifi c). HDL cholesterol was determined by employing Lipi-Direct precipitation reagent (PRDI Reference Diagnostics, Bedford, MA). Triglyceride levels were measured by using triglyceride reagent (Fisher, Infi nity, catalog number TR22321) in accordance with the manufacturer's protocols. Nonfasting LDL levels were calculated using the formula: total cholesterol − HDL − triglyceride/5.

Quantifi cation of LDL oxidation and plasma free metabolites of arachidonic and linoleic acids
LDL was obtained by fast protein liquid chromatography (FPLC) using a Superpose 6 10/300GL column as previously described ( 18 ). Briefl y, 100 l of plasma was applied to the column, and the samples were eluted in a mobile phase (0.15 M NaCl, 0.01% NaN3, and 2 mM EDTA, pH 7.5) at a rate of 0.3 ml/min in 30 fractions of 1.0 ml. Fractions containing LDL were extracted. The level of LDL oxidation was assessed by measuring the levels of arachidonic acid and linoleic acid metabolites including HETEs, HODEs, prostaglandin D2 (PGD2), prostaglandin E2 (PGE2), and thromboxane (TXB2) were determined by liquid chromatography, electron spray ionization, and tandem mass spectrometry (LC-ESI-MS/MS) as previously described ( 19,20 ). The levels of free oxidative products of arachidonic and linoleic acids in plasma were also measured by LC-ESI-MS/MS as described ( 19,20 ). In each instance, a deuterium-labeled internal standard was included to correct for extraction effi ciency and to facilitate quantifi cation. Extreme care was taken to prevent lipid oxidation during the analyses by including 20 uM Butylated hydroxytoluene(BHT) and other anti-oxidants.

Measurement of HDL oxidation index
The HDL oxidation index (HOI) was measured as an indicator of the anti-oxidant capacity of HDL that prevents LDL oxidation. The fl uorescent intensity of dichlorofl uorescein-diacetate (DCFH-DA) during LDL air oxidation in the presence or absence of HDL was measured as described previously ( 13,21 ). The HOI was calculated as the ratio of dichlorofl uorescein (DCF) fl uorescence in the presence of HDL over the value in the absence of HDL.

Analyses of atherosclerotic lesion size
Lesion quantifi cation of atherosclerosis was performed in accordance with procedures described ( 22,23 ). For en face lesion quantifi cation of entire aortas, the aortas were dissected after perfusion-fi xation, opened longitudinally from the heart to the iliac artery bifurcation, and pinned on a black wax pan for Sudan IV staining. For aortic sinus lesion quantifi cation, the upper portion of the heart and proximal aorta was obtained, embedded in OCT compound, and stored at Ϫ 70°C. Serial 5 m thick cryosections of the aortic root were collected. These metabolism and reducing HDL anti-oxidant capacity. Low density lipoprotein receptor-null (LDLR Ϫ / Ϫ ) mice fed on a high-fat diet (HFD) were exposed to fi ltered air (FA) versus UFPs collected near downtown Los Angeles in the presence or absence of D-4F administration. We quantifi ed HDL anti-oxidant capacity, PON activity, serum amyloid A (SAA) and tumor necrosis factor ␣ (TNF-␣ ) levels, plasma lipid profi les, LDL oxidation, and oxidative products of arachidonic and linoleic acids, namely, hydroxyeicosatetraenoic acids (HETEs) and hydroxyoctadecadienoic acids (HODEs), in relation to the atherosclerotic lesion size. D-4F was administered to elucidate the role of plasma lipoproteins and HDL functionality underlying UFP-induced atherosclerosis.

UFP collections and sample preparation
The collection of the UFPs was conducted at the University of Southern California (USC) campus located in the urban regions of Los Angeles, in close proximity to a network of major freeways. These aerosols represent a mixture of pollution sources, including fresh ambient PM from areas impacted by heavy-duty diesel trucks and light-duty gasoline vehicles, as well as PM generated by photochemical oxidation of primary organic vapors. The USC site is representative of the numerous urban regions in the USA where high concentrations of PM are freshly emitted from vehicular traffi c on nearby freeways. UFP samples were collected by a high-volume particle sampler as previously described ( 15 ). The collected PM samples were extracted from the fi lter substrates, and were reaerosolized for the exposure experiment as previously described ( 16 ). The size distribution of UFPs was comparable to the representative aerosols collected in downtown Los Angeles at the USC site (supplementary Fig. I) as previously described ( 17 ). Chemical analysis was also performed on the UFPs (for details refer to Supplementary Methods). The chemical compositions of UFPs and the concentration (in terms of ratios in total PM) of the main chemical constituents of UFPs including inorganic ions; organic, elemental, and total carbon content; and selected redox-active metal species is presented in supplementary Table I

Mouse exposure to UFPs in the presence or absence of D-4F
All animal experiments were performed in the Vivarium in the Ray R. Irani Building in compliance with USC Institutional Animal Care and Use Committee protocol. Age-matched LDLR Ϫ / Ϫ male mice under the C57BL/6 background (stock number 002207, Jackson Laboratory) were exposed to UFPs or High-effi ciency particulate air(HEPA)-FA (used as control group) via whole-body animal exposure chambers on a HFD (D12492: 5.24 kcal/g, 34.9 g% fat, 26.2 g% protein, 26.2 g% carbohydrate; Research Diets) starting at the age of 90 days. In reference to the previous exposure duration/dosage of exposure ( 7 ), the mice were exposed to UFPs at mass concentrations of approximately 360 g/m 3 for 5 h per day and 3 days per week for 10 weeks (please see Supplementary Methods for details). During exposure, a scanning mobility particle sizer (SMPS Model 3080, TSI Inc.) was used, in parallel with the animal exposure chambers, to monitor particle sizes and concentrations.
D-4F was initially administered via subcutaneous injection at 0.2 mg/ml/mouse, and saline was injected as the control. After the fi rst week, D-4F was administered via drinking water whereas D-4F administration restored PON activity ( P < 0.01) ( Fig. 2B ). Spearman analysis further revealed a correlation between the reduction in PON activity and an increase in HOI ( = Ϫ 0.72, P < 0.01) ( Fig. 2C ). Thus, these fi ndings suggest that UFPs decreased HDL anti-oxidant capacity associated with reduced PON activity.

Exposure to UFPs increased plasma levels of SAA and TNF-␣
PM promotes infl ammatory responses with relevance to the initiation of atherosclerosis ( 2 ). Akin to the C-reactive protein in humans, SAA is a systemic infl ammatory marker in mice. UFP exposure signifi cantly increased SAA levels in LDLR Ϫ / Ϫ mice ( P < 0.01), whereas D-4F attenuated UFPs increased plasma levels of SAA ( Fig. 4A ). PM exposure has also been reported to increase levels of infl ammatory cytokines including TNF-␣ ( 2, 29 ), and to activate NF-B signaling ( 30,31 ). While UFP exposure signifi cantly increased plasma TNF-␣ levels, D-4F did not signifi cantly alter the levels of TNF-␣ ( Fig. 4B ). Hence, UFP exposure promoted systemic infl ammatory responses.
Spearman correlation analysis revealed that HDL levels inversely correlated with atherosclerotic lesions ( = Ϫ 0.84, sections were stained with Oil Red O. The images of the aorta and lesions were captured by using an Olympus SZX12 trinocular dissection microscope with 1× and 0.5× objectives in a blinded fashion (Olympus U-SPT and Diagnostic Instruments), 0.6× HRP060-CMT camera adapters, Q Imaging Micropublisher 5.0 RTV Digital camera, and Image Pro Plus image analysis software v4.5 (Media Cybernetics). The ratios of the area covered by atherosclerotic lesions to that of the entire aorta (en face staining) were calculated and used as lesion scores for our correlation analysis. The mean value of lipid staining areas of aortic sinus cross-sections was calculated as additional lesion score.

Measurement of PON activity and levels of SAA and TNF-␣
PON activity was determined using paraoxon as substrate by measuring the absorbance at 412 nm in the presence of 4-nitrophenol as described ( 24 ). Briefl y, the activity was measured at 25°C by adding 50 l of serum to 1 ml Tris-HCl buffer (100 mM at pH 8.0) containing 2 mM CaCl 2 and 5 mM of paraoxon. The rate of generation of 4-nitrophenol was measured at 412 nm. Enzymatic activity was computed by a molar extinction coeffi cient of 17,100 M Ϫ 1 ·cm Ϫ 1 . Plasma SAA levels were determined by ELISA with Invitrogen's kit as previously described ( 25 ). Plasma levels of TNF-␣ were measured using ELISA kit form BioLegend following the manufacturer's instructions.

Statistical analyses
Data were expressed as mean ± SD. Multiple comparisons were made by one-way ANOVA, and statistical signifi cance for pairwise comparison was determined by post hoc analyses using the Turkey test. Spearman correlation analysis was performed between lesion size and various measurements. P < 0.05 was considered statistically signifi cant.

Exposure to UFPs reduced HDL anti-oxidant capacity and PON activity
To investigate the lipid mechanisms underlying UFPmediated atherosclerosis, we determined HDL anti-oxidant capacity by assessing LDL oxidation. Data was expressed as a HOI using a DCF-based cell free assay ( 13,21 ). An elevated HOI represented a decrease in HDL anti-oxidant capacity ( 13,21 ). UFP exposure signifi cantly increased HOI ( P < 0.05), whereas D-4F administration blocked UFP effects on HOI ( P < 0.01) ( Fig. 2A ).
To further assess how UFPs affected HDL anti-oxidant capacity, we quantifi ed plasma activity of PON, a HDLassociated enzyme harboring anti-oxidant properties ( 22,26 ). UFPs signifi cantly reduced plasma PON activity ( P < 0.01), PON activity was also negatively correlated with lesion ratios (supplementary Fig. IV), but D-4F did not significantly rescue this negative correlation. There were no correlations between lesion ratios and SAA levels (supplementary  Ϫ / Ϫ mice. LDLR Ϫ / Ϫ mice were exposed to FA versus UFPs for 10 weeks in the presence or absence of D-4F. Plasma levels of HDL (A) and triglyceride (B) were measured. HDL levels were signifi cantly decreased ( P < 0.01), whereas triglyceride levels were signifi cantly increased ( P < 0.01) in LDLR Ϫ / Ϫ mice exposed to UFPs. Administration of D-4F (gray bars) signifi cantly attenuated UFP-mediated reduction in HDL ( P < 0.01) and increase in triglyceride ( P < 0.05).

Fig. 2. UFPs reduced HDL anti-oxidant capacity and PON activity in LDLR
Ϫ / Ϫ mice. A: HOI was measured by DCF assay as a reverse indicator of HDL anti-oxidant capacity. UFPs signifi cantly reduced HDL anti-oxidant capacity as evidenced by a signifi cant increase in HOI (black bars). D-4F administration abrogated UFP-mediated increases in HOI (gray bars). B: Plasma PON activity was measured after 10 weeks of exposure to UFPs. UFPs signifi cantly reduced plasma PON activity ( P < 0.01), which was restored with D-4F administration ( P < 0.01). C: Spearman correlation analysis indicated that PON activity reversely correlated with HOI ( = Ϫ 0.72, P < 0.01).
HDL levels ( Fig. 1A ) and an increase in triglyceride levels ( Fig. 1B ) in LDLR Ϫ / Ϫ mice, likely due to the different genetic background ( 33 ). In addition, the concentration and duration of UFP exposure as well as diet and age of mice may have infl uenced the lipid profi les. In the previous study, ApoE-null mice on chow diet were exposed to concentrated ambient particles (CAPs), starting at the age of 6 weeks over a 40-day period (5 h per day, 3 days per week, for a total of 75 h). The concentration of UFPs in the exposure chamber was 5.59 (±1.23) × 10 5 particles/cm 3 . In the current study, LDLR Ϫ / Ϫ mice on a HFD were exposed to ambient UFPs at 1.91 × 10 5 /cm 3 starting at the age of 12 weeks for 5 h a day, three days a week, for 10 weeks and a total of 150 h. There were signifi cant inverse correlations between HDL levels and atherosclerotic lesions among all animals, including untreated and D-4F-treated mice (supplementary Fig. III). D-4F treatment led to increased HDL levels and decreased atherosclerotic lesions, supporting the mechanisms by which UFPs promoted atherosclerotic lesion formation, in part, via a reduction in HDL levels in LDLR Ϫ / Ϫ mice.
UFPs also altered PON activity, plasma oxidized fatty acids, and HDL anti-oxidant capacity. Paraoxonase 1 (PON1) is an HDL-associated lactonase harboring anti-oxidant and anti-atherogenic properties ( 34 ). PON1 increases HDL binding to macrophages which, in turn, promotes HDLmediated cholesterol effl ux ( 34 ). The ability of HDL to inhibit LDL oxidation and lipid peroxidation is mediated, in part, through PON ( 35 ). Increasing evidence suggests

DISCUSSION
The current study elucidated the lipid mechanisms underlying the effects of UFP exposure to hyperlipidemic LDLR Ϫ / Ϫ mice. UFP exposure led to enhanced atherosclerotic lesions in association with UFP-mediated reduction in HDL anti-oxidant capacity and PON activities, an increase in SAA and TNF-␣ levels, and an increase in oxidation of LDL and lipid metabolites (HETEs and HODEs). Administration of D-4F further provided new insights into the mechanisms whereby UFPs accelerated atherosclerosis via a proatherogenic lipid profi le and systemic infl ammatory effects.
The current data performed in LDLR Ϫ / Ϫ mice recapitulated the previous study by Araujo et al. ( 7 ) in which UFP exposure to ApoE-null mice signifi cantly increased atherosclerotic lesion size. HDL anti-infl ammatory capacity was decreased by UFPs in the previous study as measured by the inhibition of HDL on monocyte binding to endothelial cells. In the current study, UFPs reduced HDL anti-oxidant capacity as measured by its ability to inhibit LDL oxidation. HDL anti-atherogenic properties include the abilities to exert anti-oxidant and anti-infl ammatory actions in the vasculature ( 8,9 ). The data from the current study are in agreement with the report revealing the anti-oxidant capacity of HDL is signifi cantly impaired in patients with acute coronary syndromes as compared with healthy individuals or those with stable coronary artery disease ( 32 ).
Unlike the previous study ( 7 ), the current study revealed that exposure to UFPs led to a signifi cant reduction in  Fig. 3. UFPs signifi cantly increased plasma levels of free oxidized fatty acids, 9-HODE, and 12-HETE. Plasma levels of free oxidized fatty acids (HODEs and HETEs) were analyzed by high-performance liquid chromatography, electrospray ionization, and tandem mass spectrometry (HPLC-ESI-MS/MS) after 10 weeks of UFP exposure. A: UFPs signifi cantly increased 9-HODE levels as compared with FA ( P < 0.01, black bars). Administration of D-4F signifi cantly attenuated these effects ( P < 0.01, gray bars). B: UFPs also signifi cantly increased 12-HETE levels ( P < 0.05, black bars), which were signifi cantly attenuated in the presence of D-4F ( P < 0.001, gray bars).
( Fig. 5 ). One possibility is that D-4F treatment may have not been able to fully normalize HDL atheroprotective properties such as anti-infl ammatory capacity and reverse cholesterol transport which is supported by the fact that D-4F treatment did not completely abrogate UFP effects on SAA levels. Alternatively, UFPs may also enhance atherosclerosis via nonlipid mediated pathways as well.
The limitation of the current study lies in the small numbers of animals per treatment conditions. While UFP exposure engendered statistically signifi cant changes such as lesion score, HDL level, anti-oxidant capacity, PON activity, etc., some of the measurements showed trends of changes consistent with lesion data but fell short by P values. For example, UFPs increased 5-HETE and 13-HODE levels, but the increases were not statistically signifi cant ( P = 0.18 and P = 0.16, respectively). Similarly, D-4F treatment reduced UFP-induced TNF-␣ levels, but the P value was 0.12. Our power analysis ( 40 ) based on our current data suggests an improvement in statistical signifi cance by increasing the number of mice to 10 in all groups in our future investigation if we are to focus on specifi c experimental conditions.
Overall, the strengths of the current study lie in the use of LDLR Ϫ / Ϫ mice and D-4F treatment which provide valuable insights into the mechanisms underlying UFPmediated lipid metabolism and HDL dysfunction in close relation to atherosclerotic lesion formation. Exposure to UFPs in LDLR Ϫ / Ϫ mice recapitulated the previous fi nding that PM accelerated atherosclerosis in ApoE-null mice ( 7 ).
In addition, the current study shows that changes in lipid metabolism and HDL functionality are likely to be pathogenic mediators in UFP-mediated atherogenesis.
The authors would like to express gratitude for the technical support from Dr. David Davis and graduate students Tyler Beebe, Karen Fang, David Mittelstein, James Hill, Katherine Quigley, and Melody Shen at USC as well as Dr. Fen Yin at University of California, Los Angeles. that PON1 mediates HDL anti-oxidant capacity to prevent LDL oxidation, as measured by a reduction in lipid peroxides ( 22,26 ). In the current study, we observed that UFPs affected HDL anti-oxidant capacity and PON activity, and Spearman analysis corroborated a close correlation between them. Thus, a reduction in PON activity may be intimately related with an UFP-mediated decrease in HDL anti-oxidant capacity.
HDL dysfunction is associated with the pathogenesis of atherosclerosis ( 11,36 ). HDL from diabetic patients had reduced anti-infl ammatory and anti-oxidant properties, accompanied by increased free oxidized fatty acids (HETEs and HODEs) ( 13 ). Increased plasma levels of 9-HODE and 13-HODE impaired the anti-infl ammatory properties of HDL ( 19 ). Here, we demonstrated that UFPs increased the plasma levels of 9-HODE and 12-HETE. While the mechanisms underlying the elevated HETE and HODE levels and reduced PON activity remain to be elucidated, removal of oxidized fatty acid from HDL was reported to restore PON activity ( 37 ). D-4F can reduce HETE and HODE plasma levels via its strong binding affi nity to oxidized fatty acids ( 38 ).
The anti-oxidant activity of HDL has generated a considerable enthusiasm in HDL-based therapy. HDL and its ApoA-I can lose their protective activity to reverse transport cholesterol through changes in protein or lipid composition as well as protein modifi cations ( 7,39 ). D-4F was reported to restore HDL function and to prevent diabetes-induced atherosclerosis ( 13 ). In the current study, D-4F reduced HETEs, HODEs, and the HDL oxidation index, increased PON activity, and attenuated atherosclerotic lesion size. These findings supported the notion that D-4F played an athero-protective role in UFP-exposed LDLR Ϫ / Ϫ mice, and that UFP-mediated reduction in HDL anti-oxidant capacity was implicated in the promotion of atherosclerosis. However, even though D-4F treatment fully inhibited UFP-induced changes in HDL anti-oxidant capacity, exposure to UFPs was still able to enhance atherosclerotic lesion formation among D-4F-treated mice Fig. 4. Exposure to UFPs increased plasma levels of SAA and TNF-␣ in LDLR Ϫ / Ϫ mice. Plasma levels of SAA and TNF-␣ were measured by ELISA after 10 weeks of exposure to UFPs. UFPs signifi cantly elevated SSA levels ( P < 0.05) and TNF-␣ levels ( P < 0.05, black bars). D-4F administration attenuated UFP-mediated elevation in SAA levels ( P < 0.01) but not TNF-␣ levels (gray bars).  5. UFP-mediated atherosclerotic lesion size correlated with a decrease in HDL anti-oxidant capacity. LDLR Ϫ / Ϫ mice were exposed to FA or UFPs in the presence or absence of D-4F. A: En face mouse aortas were stained with Sudan IV to visualize atherosclerotic lesions (representative pictures). B: Atherosclerotic lesion scores were defi ned as the lesion/area ratios, revealing that UFPs (n = 8, P < 0.001) signifi cantly increased atherosclerotic lesion size (black bars). Administration of D-4F attenuated atherosclerotic lesion size (UFPs, n = 9, P < 0.01) (gray bars). C: Correlation between UFP-mediated increase in atherosclerotic lesion size and reduction in HDL anti-oxidant capacity. Spearman analysis revealed that the HOI was positively correlated with lesion size of atherosclerosis in all mice (left panel, = 0.52, P < 0.007) and in the absence of D-4F administration (middle panel, = 0.81, P < 0.002). D-4F treatment led to a loss of correlation between HOI and lesion size (right panel, = Ϫ 0.02, P = 0.96), supporting the notion that D-4F reduced lesion size by restoring HDL anti-oxidant capacity. D: Representative pictures of atherosclerotic lesion in aortic sinus (cross-sections). E: Lesion quantifi cation of aortic sinus as measured by lesion area. UFPs signifi cantly increased lesion area (n = 8, P < 0.0001), which was attenuated by D-4F administration (n = 9, P < 0.001).