Procollagen C-endopeptidase Enhancer Protein 2 (PCPE2) Reduces Atherosclerosis in Mice by Enhancing Scavenger Receptor Class B1 (SR-BI)-mediated High-density Lipoprotein (HDL)-Cholesteryl Ester Uptake*

Background: Extracellular matrix protein PCPE2 is linked to alterations in HDL size and concentration. Results: PCPE2 protects against diet-induced atherosclerosis by promoting HDL catabolism, reverse cholesterol transport, and SR-BI-mediated uptake of HDL-cholesteryl ester. Conclusion: PCPE2 mediates HDL function by reducing lipid and immune cell accumulation in the artery. Significance: These findings establish a role for the extracellular matrix glycoprotein PCPE2 in SR-BI-mediated HDL function and the prevention of atherosclerosis. Studies in human populations have shown a significant correlation between procollagen C-endopeptidase enhancer protein 2 (PCPE2) single nucleotide polymorphisms and plasma HDL cholesterol concentrations. PCPE2, a 52-kDa glycoprotein located in the extracellular matrix, enhances the cleavage of C-terminal procollagen by bone morphogenetic protein 1 (BMP1). Our studies here focused on investigating the basis for the elevated concentration of enlarged plasma HDL in PCPE2-deficient mice to determine whether they protected against diet-induced atherosclerosis. PCPE2-deficient mice were crossed with LDL receptor-deficient mice to obtain LDLr−/−, PCPE2−/− mice, which had elevated HDL levels compared with LDLr−/− mice with similar LDL concentrations. We found that LDLr−/−, PCPE2−/− mice had significantly more neutral lipid and CD68+ infiltration in the aortic root than LDLr−/− mice. Surprisingly, in light of their elevated HDL levels, the extent of aortic lipid deposition in LDLr−/−, PCPE2−/− mice was similar to that reported for LDLr−/−, apoA-I−/− mice, which lack any apoA-I/HDL. Furthermore, LDLr−/−, PCPE2−/− mice had reduced HDL apoA-I fractional clearance and macrophage to fecal reverse cholesterol transport rates compared with LDLr−/− mice, despite a 2-fold increase in liver SR-BI expression. PCPE2 was shown to enhance SR-BI function by increasing the rate of HDL-associated cholesteryl ester uptake, possibly by optimizing SR-BI localization and/or conformation. We conclude that PCPE2 is atheroprotective and an important component of the reverse cholesterol transport HDL system.

In humans there is an inverse correlation between the concentration of plasma HDL and the relative risk of developing cardiovascular disease (CVD) 5 (1)(2)(3)(4). To explain this association, many studies have suggested that HDL plays a central role in the pathway called reverse cholesterol transport (RCT), a process that returns cholesterol to the liver for excretion (5,6). In the first step, ATP-binding cassette transporter A1 (ABCA1) lipidates apoA-I with phospholipid and cholesterol to form nascent HDL (nHDL) (5,(7)(8)(9). These nHDL are rapidly converted to mature HDL by plasma lecithin-cholesterol acyltransferase, an enzyme that esterifies free cholesterol yielding mature spherical HDL particles with a cholesteryl ester (CE)-rich core. The RCT pathway is complete when the scavenger receptor class B1 (SR-BI) in the liver removes CE from mature HDL, leaving lipid-poor apoA-I to be recycled or filtered by the kidney (10). Because the liver is the primary site of HDL metabo-lism, the steady-state concentration of plasma HDL depends largely on the rate of nHDL formation balanced by the catabolism of HDL.
For some time, epidemiology studies used plasma HDL concentration to help calculate the relative risk of developing CVD. That approach has been criticized because plasma HDL concentration does not adequately predict an individual's CVD risk (11,12). Recent studies have shown that an individual's cholesterol efflux capacity is a more accurate measure of assessing CVD risk (13)(14)(15). Efflux capacity measures the potential of an individual's apoB-depleted plasma to mobilize cellular cholesterol, suggesting that the inherent properties of one's plasma HDL are responsible for driving this process.
Cholesterol efflux is an essential physiological process that produces nHDL particles. It is now recognized that the composition and structure of nHDL, before modification in plasma, differ significantly from plasma or mature HDL. These nHDL particles contain three apoA-I monomers and are highly enriched in free cholesterol and sphingomyelin, having a composition similar to lipid rafts (16).
nHDL formation is not a spontaneous process occurring when apoA-I and phospholipid come in contact. Rather, nHDL formation requires carefully orchestrated conformational changes in the apoprotein structure that permit shielded, hydrophobic residues within multiple monomers of apoA-I to become accessible and bind phospholipid (16). This process yields ϳ11-nm diameter nHDL particles carrying over 100 molecules of free cholesterol/particle (17), an efficient process to efflux cholesterol from the cell.
Several mechanisms have been proposed to explain how nHDL is assembled via ABCA1 on the membrane surface (7,18,19). One hypothesis has binding of lipid-poor apoA-I to ABCA1 as the first step in the process. However, other studies suggest that low affinity, high capacity membrane binding sites are involved before apoA-I binds to ABCA1 (20 -23). Interaction between apoA-I and membrane-bound accessory protein(s) could potentially promote conformational changes within lipid-poor apoA-I and in so doing enhance the rate of nHDL particle formation (12,24,25).
Recent reports suggest that procollagen C-endopeptidase enhancer protein 2 (PCPE2) participates in promoting cholesterol efflux. A connection between PCPE2 and HDL was inferred when significant correlations were observed between PCPE2 single nucleotide polymorphisms and plasma HDL cholesterol concentrations (26). PCPE2 is a 52-kDa glycoprotein encoded by the PCOLCE2 gene (27,28). It shares 43% amino acid sequence identity with PCPE1 (29). Both PCPE1 and PCPE2 are located in the extracellular matrix, where they facilitate bone morphogenetic protein 1 (BMP1) cleavage of C-terminal procollagen propeptides. PCPE2 and PCPE1 have different tissue distributions and heparin-binding affinities, suggesting a functional divergence (29). PCPE2 is heavily expressed in heart tissue in contrast to PCPE1. Both PCPE1 and PCPE2 have two CUB (Complement C1r/C1s, Uegf, Bmp1) domains separated by a short linker region (30 -32), with each domain consisting of about 110 residues containing a ␤-sandwich fold that mediates a variety of protein-protein interactions. The CUB domains also contain a homologous Ca 2ϩ bind-ing site that mediates ionic interactions between protein partners, similar to that described in the low-density lipoprotein receptor family (33). The tandem CUB domains are followed by a C-terminal netrin-like (NTR) domain (27,29). CUB domains are commonly found in extracellular membrane proteins that mediate protein-protein interactions (33)(34)(35)(36), whereas NTR domains bind to cell surface glycosaminoglycans (37,38), presumably anchoring PCPE2 to the extracellular matrix.
Studies in PCPE2 Ϫ -deficient mice provide the first direct experimental evidence linking PCPE2 to HDL metabolism (43). PCPE2 knock-out mice had elevated concentrations of enlarged plasma HDL (44), despite displaying defective ABCA1mediated cholesterol efflux capacity. Thus, reduced cholesterol efflux is paradoxically associated with elevated plasma HDL cholesterol in PCPE2-deficient mice. A recent genome-wide siRNA screen also suggests that PCPE2 affects the efficiency of HDL apoA-I secretion (45).
Because elevated plasma HDL levels are associated with reduced atherosclerosis, we sought to investigate whether elevated concentrations of enlarged HDL in PCPE2-deficient mice were atheroprotective or atherogenic. We have suggested mechanisms to explain the role of PCPE2 and provided evidence demonstrating the participation of PCPE2 in HDL cholesterol catabolism and reverse cholesterol transport.

Experimental Procedures
Animals and Diets-PCPE2-deficient mice in the C57BL/6 background were obtained from the Max Planck Institute (43) and then crossed with LDLr Ϫ/Ϫ mice to generate LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice. Both LDLr Ϫ/Ϫ and LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mice have been fully crossed onto the C57BL/6 background as described previously (46,47). Starting at 6 weeks of age, mice were fed for 12 weeks either a standard chow diet or an atherogenic diet containing 0.1% cholesterol and 10% calories from palm oil (47). Mice were fasted for 3 h prior to being anesthetized with ketamine/xylazine (200 mg/kg ketamine and 10 mg/kg xylazine) followed by euthanasia and blood collection by cardiac puncture. The Animal Care and Use Committee of the Wake Forest University School of Medicine approved all procedures used in the current study. All mice were housed in a temperature-controlled room and maintained in a 12-h light/12-h dark cycle at the Wake Forest School of Medicine vivarium.
Plasma Lipoprotein Isolation and Characterization-Blood was collected in EDTA containing tubes then centrifuged at 15°C for 10 min at 7000 rpm. The plasma fraction was collected and stored at Ϫ80°C. Total plasma cholesterol (Wako Diagnostics Cholesterol E) was determined using an enzymatic assay kit. For total lipoprotein cholesterol distribution, 15 g of total plasma cholesterol was applied to a Superose 6 10/300GL column (GE Healthcare) with online mixing of enzymatic reagent (Cholesterol (Liquid) Reagents Set, Pointe Scientific Inc.) as described previously (48). Isolation of the total lipoprotein fraction for compositional analysis was carried out by first obtaining the d Ͻ 1.225 g/ml fraction after density gradient ultracentrifugation using KBr as described previously (46). The d Ͻ 1.225 g/ml fractions were separated into lipoprotein classes using size-exclusion fast protein liquid chromatography (FPLC) (46). Aliquots corresponding to VLDL, LDL, and HDL classes were pooled and stored at Ϫ80°C for composition analysis or used immediately for particle size determination. The HDL particle size was determined by a comparison with standards of known Stokes diameter following separation using Novex (Life Technologies) 4 -20% non-denaturing gels and then stained with Coomassie Blue G-250 dye as described previously (49). Mouse plasma apoA-I concentration was measured using an enzyme-linked immunosorbent assay as described previously (49).
Quantification of Pro-apolipoprotein A-I by Mass Spectrometry-The abundance of pro-apoA-I and apoA-I was assessed from purified HDL and whole plasma from PCPE2 Ϫ/Ϫ and C57BL/6 mice using liquid chromatography-tandem mass spectrometry. Whole plasma aliquots were separated on 12% SDS-PAGE as described previously (16, 50 -52). The gel was stained with SimplyBlue (Life Technologies), and the protein band corresponding to 28,000 Da was excised from the gel, minced, and repeatedly dehydrated with acetonitrile. The gel pieces were rehydrated with a cold, freshly prepared solution containing 20 ng/l trypsin in 10 mM ammonium bicarbonate, pH 7.8, 0.1% (w/v) RapiGest SF TM , and 1 mM CaCl 2 . The final trypsin to apoA-I mass ratio was ϳ1:20. After incubation on ice for 10 min, the digests were incubated for 18 h at 37°C. The digestion solution was removed, and gel pieces were covered with 200 l of acetonitrile/formic acid/water (v/v/v, 50:5:45). After sitting for 10 min the solvent was transferred to a fresh tube. The extraction was repeated and the combined aliquots acidified to an HCl:apoA-I ratio of 1:10 (v/v) using 500 mM HCl. After the acidified solution was incubated for 35 min at 37°C, the sample was centrifuged for 10 min at 13,000 rpm. The supernatant was transferred to a fresh tube before mass spectrometry. Survey scans were performed on each peptide mixture using a Waters Q-TOF API-US mass spectrometer equipped with a Waters CapLC and Advion Nanomate source. Acquisition was controlled using Mass-Lynx TM 4.0 software. Peptides were loaded onto a PLRP-S trapping column, 0.5-mm diameter ϫ 2.0-mm length, containing 3-m diameter particles with a pore diameter of 100 Å. Peptides were loaded onto the trapping column in water/acetonitrile/formic acid (97:3: 0.2) at 500 nl/min. After switching of the trapping column, inline separation was accomplished on a 0.3 ϫ 150-mm PCRP-S column packed with 3-m diameter particles with 100 Å pores at 470 nl/min. The following gradient elution was used: solvent A (25 mM formic acid in 97% water and 3% acetonitrile) and solvent B solvent B (25 mM formic acid in 3% water 97% acetonitrile). The gradient profile was: 2% solvent B for 3 min with a linear increase to 40% B at 90 min and then to 80% B in 5 min. At 95 min, the composition was ramped to 2% B over 5 min and equilibrated with 2% B for 30 min. Positive ion survey scans were recorded in the continuum mode with a scan window of 300 to 1500 m/z for 2 s. The source temperature was 80°C and cone voltage was 45 V. Experimental m/z was corrected using mouse apoA-I tryptic fragment 7, m/z ϭ 533.7462. Ions within Ϯ0.051 m/z of the theoretical ions were sequenced. Product ion MS/MS spectra were acquired in the continuum mode from 50 to 1500 m/z using a data-directed charge state-selective collision energy and an accumulation time of 2 s. Sequence analysis of the MS/MS spectra was performed with a fragment ion tolerance of Ϯ0.05 m/z.
For quantification of the ratio of mature and pro-apoA-I, the N-terminal tryptic peptides for mouse pro-apoA-I (WHV-WQQDEPQSQWDK, protonated m/z ϭ 1996.8942) and mature mouse apoA-I (DEPQSQWDK, protonated m/z ϭ 1132.4911) were synthesized (GenScript), and the m/z, purity, and sequence were verified by mass spectrometry. The retention time, mass, and sequence were used to identify the N-terminal sequence from all samples analyzed.
For the analysis of intact pro-or mature mouse apoA-I from purified HDL, mass spectrometric analysis was performed using a Waters Q-TOF equipped with an Advion Triversa Nanomate source as described previously (53). Samples (0.4 M) were prepared in 1:1 (v:v) acetonitrile-water containing 0.2% formic acid and then introduced into the mass spectrometer at 500 nl/min. Acquisition parameters were adjusted to maximize resolution.
Lipid Composition-Total and free cholesterol determination were performed on aliquots from FPLC-separated and pooled HDL that had been spiked with the internal standard, cholesterol-3,4-13 C 2 (Sigma-Aldrich), and subjected to lipid extraction as previously described (16). To measure free cholesterol an aliquot was removed, evaporated under argon, dissolved in hexane and then injected on a Finnigan TSQ Quantum XLS tandem mass spectrometer interfaced to a TRACE gas chromatograph (GC-MS/MS), as described previously (16,47), by select reaction monitoring (SRM) with the following parameters in the positive ion mode: scan time, 0.1 s; collision energy, 10 V; emission current, 25 A; electron energy, Ϫ42 eV; source and transfer line temperature, 260°C; and flow rate, 1 ml/min. Analysis was carried out using a DB-1 column (10.4 m, 250 m inner diameter) with a 0.25-M film thickness. For quantifying total cholesterol, the remaining sample was dried under a stream of nitrogen, redissolved in 1 ml of ethanol, mixed with 100 l of 50% (w/w) aqueous potassium hydroxide, and then saponified for 1 h at 60°C. After extraction, total cholesterol was measured on the Quantum XLS. Cholesteryl ester was calculated as the difference between free cholesterol and total cholesterol.
Plasma triglycerides were analyzed using mass spectrometry from mouse plasma after preliminary separation from other polar and nonpolar lipids using aminopropyl solid-phase extraction columns using the procedure of Kaluzny et al. (54) with minor modifications. The nonpolar fraction was dried under nitrogen and loaded onto the second solid-phase extraction column dissolved in 500 l of hexane. Cholesteryl esters were removed from the column with 3-4-ml washes of 4% CH 2 Cl 2 in hexane. Then triglycerides were eluted with 3-8-ml washes of CH 2 Cl 2 :diethyl ether:hexane (14:1:84). Quantitation was by gas chromatography-mass spectrometry after saponifi-cation to generate methylated fatty acids (55,56). The fatty acids were quantified to obtain the total fatty acid composition. The moles of fatty acids were summed and divided by 3 to obtain the number of moles of triglyceride. C17 triglyceride was used as the internal standard.
RNA Isolation, cDNA Synthesis, and Quantitative Real-Time-PCR-Tissues harvested at necropsy were flash-frozen with liquid nitrogen and kept at Ϫ80°C until needed. Briefly, tissue was quickly thawed, weighed, and then added to a 14-ml screw cap polypropylene tube containing 1 ml of prechilled TRIzol (Life Technologies). The tissue was immediately homogenized using a Polytron PT 1200C homogenizer, and RNA was isolated following the manufacturer's instructions. Purified RNA pellets were thoroughly resuspended in Ultra-Pure water (Sigma-Aldrich) on ice. RNA concentrations were determined at 260 nm using a Beckman DU800 spectrophotometer.
Using 1-2 g of RNA, cDNA was synthesized using the qScript cDNA Supermix (Quant BioSciences) according to the manufacturer's instructions. Quantitative real-time PCR was performed using the Applied Biosystems 7500 Fast Real Time PCR system with Fast Start Universal SYBR Green Master Mix (Roche Applied Science) according to the manufacturer's instructions. Reactions were set up in triplicate using a 1-g/l cDNA template. Following amplification, Ct values were obtained from Applied Biosystems sequence detection software, and the ⌬Ct values were determined for the gene of interest compared with that for mouse GAPDH for each animal run as a control. For each tissue the percent relative abundance of mouse PCPE2 and PCPE1 was derived from the ⌬Ct max Ϫ ⌬Ct/ ⌬Ct max ) ϫ 100. Primers designed using the Universal Probe Library were purchased from Integrated DNA Technologies. The following are the mouse sequences used: SR-BI forward, GCCCATCATCATCTGCCAACT, and reverse, TCCTGGG-AGCCCTTTTTACT; PCPE1 forward, TTACGTGGCAAGT-GAGGGTTT, and reverse, TGTCCAGATGCACTTCTTGT-TTG; and PCPE2 forward, TCACATGTGGCGGCATTCT, and reverse, CAGGAAAACCTTCACTGCCAAT.
Immunoblotting-Immunoblotting was performed subsequent to total tissue protein isolation following extraction using a radioimmune precipitation assay buffer (RIPA, Cell Signaling Technologies) or using detergent-free lipid raft isolation as described. For tissues extracted with radioimmune precipitation assay buffer, aliquots of equal protein content were run on 12% SDS-PAGE. All sample aliquots were diluted with 4ϫ LDS buffer (Novex) to which solid DTT had been added to achieve a 100 mM final concentration. Samples were heated to 70°C for 10 min. Transfer to PVDF membrane (PerkinElmer Life Sciences) was accomplished after treatment with Tris-glycine, pH 8.3, buffer using a semi-dry blot (Bio-Rad) apparatus for 30 min at 10 V. Membranes were blocked with 5% nonfat dry milk in 10 mM Tris-glycine, pH 7.4, buffer and processed. The primary antibodies used (PCPE1 (Novus Biologicals), PCPE2 (Abcam), GAPDH (Ambion), and SR-BI (Novus Biologicals)) were diluted 1:1000 and incubated with the membrane at 4°C overnight. The blots were washed and then incubated with an HRPconjugated anti-mouse IgG secondary antibody (GE Healthcare) at a 1:10,000 dilution for 1 h at room temperature. Blots were washed again and then incubated with SuperSignal West Pico chemiluminescence substrate (Pierce) and visualized with a FujiFilm LAS-3000 camera. Band intensities were compared using MultiGauge software (FujiFilm).
HDL ApoA-I Turnover-Diet-fed mouse HDL was purified using a combination of ultracentrifugation and FPLC. The purified HDL apoA-I was then labeled with 125 I using the Mc-Farlane method (57) as described previously (58,59). Sufficient ICl was used to have 0.5 mol of radiolabeled iodine for every mole of HDL apoA-I and prevent damage to the protein or labeling lipid. Labeled HDL apoA-I was purified from free iodine using a desalting column (Bio-Rad), and the eluate was dialyzed exhaustively against 0.15 M NaCl, pH 7.4. On the day of the study, anesthetized mice were retro-orbitally injected with 125 I-radiolabeled HDL (10 ϫ 10 6 cpm) after isoflurane administration. Approximately 30 l of blood was collected from the contralateral retro-orbital sinus at 0.08, 0.17, 0.5, 1, 3, 6, 22, 28, and 48 h. Radioactivity was measured at each time point using a Beckman Gamma 4000 counter. The fractional catabolic rates and half-lives of the 125 I-labeled HDL were determined from plasma decay curves assuming a one-pool model (58). Blood volumes were estimated to be 5.85 ml/100 g of body weight.
Histology and Immunofluorescence Microscopy-At the time of sacrifice, a 21-gauge butterfly needle was inserted into the left ventricle; the atrium was clipped, and the heart and aorta were perfused with saline for 5 min and then removed and cleaned of fat and adventitia. Mouse heart and aorta were embedded in optimal cutting temperature (OCT) medium and then sectioned for staining and immunohistochemistry. For quantification of total and percent lesion area, OCT-embedded aortas were cut into sequential 6-m sections using a Leica cryostat at Ϫ50°C. Tissue sections were stained overnight in 0.5% Oil Red O dissolved in propylene glycol and then counterstained with hemotoxylin. Masson's trichrome stain was used to assess the connective tissue content of the aortic root. Sections were digitized using a Nikon microscope and Image-Pro Plus 6.2 software and then quantified using NIS-Elements software (Nikon Instruments Inc.). Eight to 10 sections at intervals of 30 m were used for morphological analyses and were averaged to obtain a value for each animal. The results are expressed as both the percentage of lesion area and absolute lesion area.
For immunofluorescence microscopy, purified rat antimouse CD68 (AbD FA-11-Serotec) at a 1:100 dilution was used as the primary antibody. Slides were incubated for 1 h in PBS buffer containing 2% fetal calf plasma, washed with PBS, and then incubated with a 1:300 dilution of Cy3-conjugated goat anti-rat IgG antibody (Rockland) for 30 min at room temperature. After the slides were washed briefly with PBS, they were stained with DAPI to visualize the nuclei and then mounted with Fluoro-Gel (Electron Microscopy Sciences). Immunofluorescence was visualized using a Nikon Eclipse TE2000-S microscope, and the total and percentage of CD68 staining was quantified using NIS-Elements software.
Picrosirius red staining of mouse tissues was performed on formalin-fixed tissue embedded in paraffin and then cut serially in 6-m sections. Collagen fibers were stained using picrosirius red (Polysciences Inc.) and viewed with polarized light as described (60).
In Vivo Reverse Cholesterol Transport Studies-Diet-fed mice received an intraperitoneal injection of 0.6 ϫ 10 6 dpm of [ 3 H]cholesterol-labeled J774 cells as described previously (61). Blood was collected at 6, 24, and 48 h, and the plasma was counted for radioactive content (62). Feces collected from 0 to 48 h and subjected to lipid extraction were then counted for radioactive content of neutral and acidic sterols as described previously (63).
Tissue Culture and Transfection-CHO cells were grown in DMEM/F-12 medium (1:1) supplemented with 10% (v/v) FBS, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. Transient transfections were performed using Lipofectamine 2000 according to the manufacturer's protocol (Life Technologies) with 2.5 g of DNA and 12 l of Lipofectamine 2000/well of a confluent 6-well plate. The human PCPE2-pCMV6 plasmid used in the transfection was obtained from OriGene (Rockville, MD) and contains a Myc-DDK tag at the N terminus, which adds 10 amino acids.

Selective Uptake of HDL Labeled with [ 3 H]Cholesteryl
Oleyl Ether-HDL isolated as described above, using a combination of ultracentrifugation and FPLC, was labeled with cholesteryl [1a,2a-3 H]oleoyl ether ( 3 H-COE) (American Radiolabeled Chemicals Inc.) using recombinant cholesterol ester transfer protein (CETP, Roar Biomedical Inc.) as described (64). An aliquot containing 10 g of labeled HDL (based on protein)/ml of plasma-free DMEM medium plus 0.5% BSA was incubated with confluent CHO cells in a 6-well plate at 37°C for 1.5 h. After incubation the medium was removed, and the cells were washed extensively to remove surface-bound labeled HDL. The cell monolayer was lysed with 1 ml of 0.1 N NaOH, and aliquots of the lysed cells were counted for their 3 H radioactivity content and normalized to total cell protein as determined by the Lowry assay (65).
Surface Staining of CHO Cells-Transfected CHO cells were harvested from 6-well plates and stained for SR-BI and PCPE2 using fluorescence-activated cell sorting (FACS) as described previously (66). Briefly, 3 ϫ 10 5 cells were stained with polyclonal rabbit anti-SR-BI-APC conjugate (Alexa Fluor 647) (Bioss Antibodies) and mouse anti-MycDDK-FITC conjugate (Alexa Fluor 488) (AbD Serotec) for 30 min at 4°C. Samples were washed and resuspended in 2% paraformaldehyde in PBS before acquisition of samples using a system (BD Biosciences).
Statistical Analysis-GraphPad Prism version 5 was used for statistical analysis, and the data are reported as mean Ϯ S.D. Differences between groups were evaluated by independent t tests and by analysis of variance with Tukey's post hoc test.

Results
The Absence of PCPE2 Affects Lipoprotein Distribution and Concentration-PCPE2-deficient mice in the C57BL/6 background (43) were characterized with respect to HDL size and concentration for comparison with the previously published phenotype (44). Fig. 1A shows the cholesterol distribution following FPLC separation of the major lipoprotein classes from the d Ͻ 1.225 g/ml density fraction. We see that PCPE2-deficient mice possess a greater concentration of enlarged HDL particles (Fig. 1A, dashed line) compared with control C57BL/6 mice that express PCPE2 (solid line). The increase in the HDL cholesterol concentration in PCPE2-deficient mice was accompanied by an elevation in plasma apoA-I concentrations as shown in Table 1. These changes in HDL levels were independent of statistically significant changes in LDL or VLDL cholesterol levels between genotypes. A trend to higher plasma triglycerides was observed for chow-fed PCPE2-deficient mice, as exemplified by the increased VLDL peak in Fig. 1A, but the trend was not statistically significant. Non-denaturing gradient gel separation confirmed that PCPE2 Ϫ/Ϫ mice had larger HDL particle sizes and diameters compared with HDL from control mice, as shown in Fig. 1B and consistent with the previous study (44). The cholesterol to apoA-I composition was determined in HDL particles purified by a combination of density ultracentrifugation and FPLC. Fig. 1C shows that HDL particles from PCPE2 Ϫ/Ϫ mice were enriched in total cholesterol when normalized to apoA-I content, likely accounting for the increase in particle size. However, HDL from both genotypes had similar ratios of esterified to total cholesterol (Fig. 1C, EC/TC), suggesting that plasma lecithin-cholesterol acyltransferase activity was similar for the two genotypes and that conversion of free cholesterol to esterified cholesterol was not involved in particle enlargement.
Lipoprotein cholesterol distribution was then evaluated after feeding an atherogenic diet (46, 47, 67) for 12 weeks (results shown in Fig. 1E). As expected, both LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ and LDLr Ϫ/Ϫ mice had substantial increases in VLDL and LDL cholesterol levels. Table 2 shows a comparison of these values with their chow-fed counterparts. Interestingly, although diet-fed LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice did not show a significant difference in LDL cholesterol concentrations between genotypes, there was a statistically significant increase (ϳ90%) in VLDL and HDL cholesterol concentration. These results suggest that feeding an atherogenic diet to hypercholesterolemic mice that lack PCPE2 appears to also affect VLDL in addition to HDL metabolism.
PCPE2 Protects against Diet-induced Atherosclerosis-Groups of male LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ , LDLr Ϫ/Ϫ , and LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mice were fed an atherogenic diet. After 12 weeks of diet feeding their hearts and aortas were embedded into OCT, sectioned, and stained with Oil Red O. The extent of atherosclerotic lipid accumulation was assessed by staining aortic root sections with the neutral lipid stain followed by quantification, as shown in Fig. 2A.

TABLE 1 Plasma cholesterol and triglycerides
Fasting plasma cholesterol and triglyceride concentrations were obtained from equal numbers of male and female mice fed a chow diet for 12 weeks.

PCPE2 Modulates SR-BI Function and Atherosclerosis
tionally defined HDL particles in their plasma. Taken together, these results show that higher concentrations of enlarged HDL in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice do not offer significant protection against the progression of atherosclerosis. Furthermore, aortic lipid deposition in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice could not be distinguished from that seen in mice having no apoA-I-associated HDL, suggesting that the HDL particles in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice are dysfunctional in nature. Staining for the presence of CD68ϩ cells in the aortic root is commonly used as a marker for macrophage infiltration. Representative stained aortic root sections are shown in Fig. 2D: LDLr Ϫ/Ϫ , LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ , and LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mice displayed from left to right, respectively. CD68ϩ staining is expressed as the percent of CD68ϩ area in Fig. 2E and as total CD68ϩ area in m 2 in Fig. 2F. As observed with neutral lipid staining, LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ and LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mouse aortas had about a ϳ45% increase in CD68ϩ staining cells compared with aortas from LDLr Ϫ/Ϫ mice.
Reduced Collagen Content in PCPE2-deficient Mice-Given the reported enhancer properties of PCPE2 on BMP1 catalytic function (68, 69), we measured the connective tissue content in the aortic root using Masson's trichrome. Fig. 3A shows representative aortic roots sections from 12-week diet-fed LDLr Ϫ/Ϫ and LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice. Adjacent to each section is the quantification of the connective tissue content, as shown by dark blue staining (Fig. 3A) for each genotype. Whether the data were calculated as the percent of connective tissue area or as total connective tissue area (m 2 ), the LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ aorta had ϳ28% less connective tissue than the LDLr Ϫ/Ϫ aorta using Masson's trichrome staining. Because collagen types I  and III are prevalent in mouse heart and aorta, it follows that the regulation of procollagen processing must play an important role in normal tissue maintenance during disease progression (70). The reduction in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ aortic root connective tissue was confirmed by the picrosirius red staining of paraffin-embedded aortic root sections show in Fig. 3B, a-f. In Fig. 3B, a, c, and e, different views are offered of a representative slide taken from LDLr Ϫ/Ϫ mouse aortic root, whereas the views in b, d, and f are from LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mouse aortic root. Fig. 3B, a and b, shows stained aortic root at ϫ5 magnification using light microscopy, and c and d panels show the same slide but under polarizing filters. Fig. 3B, e and f, shows the same slides as above but at a ϫ20 magnification under polarizing filters, as indicated by the boxed areas in c and d. The birefringence of collagen fibers under polarized light shows a shift in color between the various components, red, orange, yellow, and green, and represents the order of decreasing thickness of the fibers. In this case, the LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mouse aortic root appears to have significantly less birefringence of all thicknesses of collagen fibers when compared with the LDLr Ϫ/Ϫ aortic root, again confirming the results of the Masson's trichrome staining. Because PCPE1 is highly similar in structure and function to PCPE2 with respect to procollagen processing, we examined whether the expression of PCPE1 might be elevated in the PCPE2-deficient state. Fig. 3C compares PCPE1 and PCPE2 mRNA expression in the artery and liver from the two genotypes of mice. No elevation of PCPE1 was seen in either tissue in light of the absence of PCPE2, although these results do show significant expression of both PCPE1 and PCPE2 in the liver of LDLr Ϫ/Ϫ mice.
Based on the mRNA expression data we then examined the picrosirius staining of collagen fibers in the hepatic triad of livers from LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ and LDLr Ϫ/Ϫ mice (shown in Fig. 3D). Here, under polarizing light the picrosirius-stained hepatic portal triad from LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice was compared with that from LDLr Ϫ/Ϫ mice and showed that both the thickness and amount of collagen are significantly reduced in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice.
PCPE2 Was Not Essential for in Vivo Pro-apoA-I Processing-PCPE2 had been reported to enhance the BMP1-mediated cleavage of propeptides from procollagen (27,29,71) during assembly in the extracellular matrix. These studies suggest that cleavage of apoA-I's propeptide is the rate-limiting step to form nHDL (39,71,72) and the primary defect in PCPE2-deficient mice. Investigators have hypothesized that if PCPE2 and BMP1 work together to remove the pro-segment of pro-apoA-I, then HDL concentrations in mice lacking PCPE2 expression might be affected. Francone et al. (44), having measured HDL apoA-I from PCPE2-deficient mice by isoelectric focusing and mass spectrometry, report an elevation in the amount of plasma pro-apoA-I over mature apoA-I.
To investigate this effect further, we first measured the ratio of intact pro-apoA-I to mature apoA-I in plasma HDL from PCPE2-deficient mice by mass spectrometric analyses. Fig. 4, A and B, shows the masses of directly infused delipidated apoA-I from the plasma of C57BL/6 (A) and PCPE2 Ϫ/Ϫ mice (B), respectively, after transformation of the electrospray spectrum. These results demonstrate that in both samples the most abun-dant mass was consistent with mature mouse apoA-I (27,950 Ϯ 1.4 Da (73)), with only a small amount of mouse pro-apoA-I (28,815 Ϯ 1.4 Da), which includes the mass of the pro-sequence WHVWQQ.
To quantify the amount of pro-apoA-I in plasma, the unique T1 peptide for the mature apoA-I (DEPQSQWDK) tryptic peptide and T1 for pro-apoA-I (WHVWQQDEPQSQWDK), which includes the additional six pro-amino acids, were synthesized and used as standards. The sequence of each synthetic peptide was verified by MS/MS sequencing, and then LC/MS was used to establish the retention times and ionizability of the two different T1 peptides. The ion-current ratio of equal molar amounts of the two synthetic T1 peptides, T1 apoA-I /T1 pro-apoA-I , was found to be 1.07.
Aliquots of mouse plasma were separated on 12% SDS-PAGE. The gel band corresponding to apoA-I at ϳ28,000 Da was excised and subjected to in-gel digestion using trypsin (51). Following digestion, peptides were exhaustively extracted from the gel slice, separated by reverse-phase HPLC, and identified by monitoring the ϩ2 charge state of each peptide (566.75 m/z for mature apoA-I and 998.95 m/z for pro-apoA-I). Fig. 4 shows the summed selective ion electrospray mass chromatograms for apoA-I from C57BL/6 (C) and PCPE2 Ϫ/Ϫ mouse plasma (D). Note that the distribution of both pro-and mature apoA-I was virtually identical for both genotypes. The critical factor for this analysis is that the molar ion intensities of each peptide were nearly identical; therefore, the measured ion intensities could be used to quantify the relative amount of each peptide. This interpretation of our data completely changes the interpretation based on a previous study (44) that reported an elevation in the amount of pro-apoA-I. Thus, our studies suggest that the role of PCPE2 in HDL enlargement is not due to the conversion of pro-apoA-I to mature apoA-I. Overall, we found that both genotypes had very little (ϳ4%) pro-apoA-I, and these results suggest that although PCPE2 may enhance pro-apoA-I processing in vitro, in vivo the absence of PCPE2 does not appear to be rate-limiting.
PCPE2 Affects HDL ApoA-I Catabolism Rates-To investigate the possibility that increased concentrations of enlarged HDL in plasma were related to the reduced catabolism of plasma HDL in PCPE2-deficient mice, HDL was purified from the plasma of diet-fed LDLr Ϫ/Ϫ and LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice using a combination of ultracentrifugation and FPLC and then radiolabeled with 125 I. Purified 125 I-labeld HDL particles were injected into diet-fed recipient mice, and the disappearance of radioactivity from plasma was followed over time, as shown in Fig. 5, A and B. 125 I-labeled HDL from LDLr Ϫ/Ϫ mice injected into LDLr Ϫ/Ϫ mice had a fractional catabolic rate (FCR) of 0.106 Ϯ 0.010 pools/h as compared with a FCR of 0.076 Ϯ 0.011 when the same HDL was injected into LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice (mean Ϯ S.D. of n ϭ 3 mice). A similar trend was observed when 125 I-labeled HDL from LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice was injected into LDLr Ϫ/Ϫ versus LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice, resulting in an FCR of 0.097 Ϯ 0.030 versus 0.062 Ϯ 0.008, respectively. From these data it appears the rate of turnover for HDL isolated from LDLr Ϫ/Ϫ or LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice was a function of the recipient genotype and not a result of HDL particle origin. In conclusion, these results suggest that PCPE2 exerts its effects on HDL metabolism by assisting in particle remodeling and/or removal. There was a trend suggesting that the larger LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ -derived HDL particles were removed more slowly in both genotypes, but this trend did not reach statistical significance.
PCPE2 and Reverse Cholesterol Transport-As PCPE2 is expressed in the extracellular matrix of mouse liver (29), we next determined how its absence might alter macrophage to liver reverse cholesterol transport (61). To this end, atherogenic diet-fed LDLr Ϫ/Ϫ and LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice received an intraperitoneal injection of [ 3 H]cholesterol-loaded J774 foam cells (61). The amount of [ 3 H]cholesterol appearing in the plasma and feces was measured for 0 -48 h, and the results are presented as the percent of injected dose (shown in Fig. 5, C and  D, respectively). Six hours after the [ 3 H]cholesterol-loaded J774 cells were injected, there was a significant difference in the appearance of [ 3 H]cholesterol in plasma between the two genotypes (Fig. 5C). LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mouse plasma showed a 2-fold higher concentration of [ 3 H]cholesterol, which remained relatively constant over the course of the study. In contrast, LDLr Ϫ/Ϫ mice showed a comparatively slower increase in plasma radioactivity, which reached a maximum by 24 h and then decreased by 48 h. The excretion of neutral [ 3 H]cholesterol in LDLr Ϫ/Ϫ mouse feces was statistically FIGURE 4. Mass spectrometric analysis and the ratio of mature to pro-apoA-I in HDL from PCPE2 ؊/؊ mouse plasma. A and B, molecular masses of HDL apoA-I after maximum entropy transformation of the electrospray spectrum of FPLC purified HDL apoA-I from chow-fed C57BL/6 mice (A) and PCPE2 Ϫ/Ϫ mice (B). The most abundant mass in each spectrum was identical within the Ϯ1.4 Da S.D. to the mass of mature mouse apoA-I at 27,950 Da. The minor spectrum peak is consistent with the mass of mouse pro-apoA-I at 28,815 Da. C and D, the relative intensity of apoA-I tryptic peptides from LC-MS/MS analysis. ApoA-I from whole mouse plasma was separated by 12% SDS-PAGE followed by in-gel trypsin digest and then analyzed for the N-terminal tryptic pro-apoA-I peptide and mature apoA-I peptide: C57BL/6 mouse plasma (C) and PCPE2 Ϫ/Ϫ mouse plasma (D) as described under "Experimental Procedures." Peptides were separated by reverse-phase HPLC and analyzed for tryptic peptide of the N-terminal T1 for mature peptide (m/z 1132.48) and the T1 propeptide (m/z 1995.88), shown in selective ion chromatograms. The mature murine T1 tryptic peptide has the amino acid sequence DEPQSQWDK, and the unprocessed pro-form has the sequence WHVWQQDEPQSQWDK.

PCPE2 Modulates SR-BI Function and Atherosclerosis
greater than that in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mouse feces (Fig. 5D). Although a trend for greater 3 H acidic sterol excretion was seen for LDLr Ϫ/Ϫ compared with LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice, it was not statistically significant. Together, the plasma and fecal analyses indicate that in the absence of PCPE2 less macrophage cholesterol was removed by the RCT pathway to the liver.
Although RCT consists of multiple steps, it is largely dependent on the expression of hepatic SR-BI (74). It is also known that attenuation of SR-B1 in mice results in a substantial increase in HDL plasma levels (10). Based on these facts, we next examined the expression of SR-BI in the liver of LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice by both RT-PCR and Western blot analysis (shown in Fig. 5, E and F). Here an immunoblot probing for mouse SR-B1 using liver extracts from two different mice per genotype shows that SR-B1 levels in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice are approximately twice that in LDLr Ϫ/Ϫ mouse livers, whereas GAPDH levels are constant. These findings were further supported by RT-PCR analysis showing a greater abundance of SR-BI mRNA in liver from LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice compared with controls (Fig. 5F), suggesting that both SR-B1 mRNA, as well as protein, is up-regulated in the LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mouse liver. This outcome was unexpected because our earlier results showed a reduction in RCT and HDL catabolism, consistent with a reduction in SR-BI. These results led us to focus next on measuring the effect of PCPE2 expression on SR-BI function.

PCPE2 Enhances SR-BI Cholesteryl Ester Uptake Function-
To determine how PCPE2 expression might affect SR-BI function, we first considered testing primary hepatocytes from LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ and control mice. However, the presence of PCPE2 in the extracellular matrix was not compatible with the use of collagenase during hepatocyte isolation. In addition, reports cautioning against the use of primary hepatocytes when studying SR-BI regulation (75) led us to perform a gain of function study in CHO cells (76). To test the effect of PCPE2 on SR-BI-mediated HDL-cholesteryl ester uptake, we first determined the surface expression of PCPE2 in CHO cells that were either mock-transfected or transfected with a plasmid expressing PCPE2 for 24 h. Fig. 6A shows a histogram of PCPE2 surface expression 24 h after transfection measured by flow cytometry. The dotted line depicts the expression of PCPE2 in control cells, and the shaded solid line depicts the shift in PCPE2 surface expression in PCPE2 transiently transfected CHO cells (Fig. 6A). The 2-fold increase (47926 Ϯ 1025 versus 21336 Ϯ 525, mean Ϯ S.D., n ϭ 3 independent experiments) indicates that PCPE2 transfection significantly increased its expression on the cell surface, consistent with the known function of PCPE2 in the extracellular matrix.
Next, to test the effect of PCPE2 on SR-BI function, CHO cells were mock-transfected or transfected with a construct encoding PCPE2 and then assayed for their ability to take up of n ϭ 5-6 male mice for each genotype. The asterisk indicates statistical significance at p Ͻ 0.001. ns, indicates that the difference was not significant. E, an immunoblot analysis of liver SR-BI levels in atherogenic diet-fed mice. F, relative mRNA abundance of liver SR-BI in atherogenic diet-fed mice. Data represent the results from n ϭ 6 -10 male mice/genotype. The liver mRNA abundance of SR-BI in LDLr Ϫ/Ϫ mouse liver was set at 100%.
HDL-cholesteryl ester. The day after mock or transient transfection with PCPE2, confluent cell monolayers were washed and then incubated with 10 g of labeled 3 H-COE HDL for 1.5 h in triplicate. Following incubation, the monolayer was washed extensively and the cell pellet subjected to radioactivity and protein quantification. Fig. 6B shows the results of these studies and suggests that the overexpression of PCPE2 enhanced the uptake of 3 H-COE HDL. This increase in SR-BI function appeared to be independent of HDL origin, because HDL particles derived from either LDLr Ϫ/Ϫ or LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice showed a similar extent of 3 H-COE uptake.
We next examined the protein extracts from cells that were either mock-transfected or transfected with PCPE2 using immunoblot analysis, as shown in Fig. 6C. In this study, the total cellular SR-BI protein was unchanged by PCPE2 overexpression. PCPE2 protein levels were increased with overexpression, as identified by the 1.2-kDa increase in the endogenous molecular mass due to the MycDDK tag on the exogenously expressed PCPE2.
To further explore the mechanism behind PCPE2 enhancement of SR-BI function, mock-and PCPE2-transfected CHO cells were stained and analyzed for SR-BI surface expression using flow cytometry. Fig. 6D shows a histogram of surface SR-BI expression in control cells, indicated by the dotted line, and in cells overexpressing PCPE2, denoted by a solid line. Interestingly, the control cells contain two populations of cells, one characterized by a low surface frequency of SR-BI/cell and a second population with a significantly higher frequency of SR-BI/cell. Upon overexpression of PCPE2, a shift to one population with a significantly higher frequency of SR-BI/cell is observed (50992 Ϯ 1070 versus 36244 Ϯ 905, mean Ϯ S.D., n ϭ 3 independent experiments). These results suggest that PCPE2 may act to enhance SR-BI function by promoting or maintaining SR-BI expression on the cell surface.

Discussion
The current studies show for the first time that PCPE2, an extracellular matrix-associated protein, confers an atheroprotective function to HDL in vivo. The implication is that in the absence of PCPE2, HDL particles are dysfunctional and cannot protect against the progression of atherosclerosis. In addition, these studies show that the apparent HDL dysfunction results

PCPE2 Modulates SR-BI Function and Atherosclerosis
from reduced HDL-CE catabolism via SR-BI and is not actually due to changes within the particle or its components itself (Fig.  5, A-D). Surprisingly, the extent of atherosclerosis in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice was indistinguishable from that in LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mice, which lack all apoA-I-containing HDL particles (47). Although both LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ and LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mice had similar extents of atherosclerosis, there was an uncoupling of its progression as it relates to total plasma cholesterol levels (77). LDLr Ϫ/Ϫ , apoA-I Ϫ/Ϫ mice had no HDL and half the circulating LDL cholesterol levels as diet-fed LDLr Ϫ/Ϫ mice (46). In contrast, LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice had higher HDL cholesterol concentrations and similar LDL cholesterol concentrations relative to LDLr Ϫ/Ϫ mice but increased atherosclerosis. These findings suggest that the term "dysfunctional HDL," e.g. HDL that is not atheroprotective, does not technically apply in this situation. Rather, HDL-related dysfunctionality was caused by obstruction of the RCT at one or more of its individual steps.
There are several mechanistic scenarios for the role of PCPE2. The first is that PCPE2 confers atheroprotection to apoA-I by enhancing BMP1-mediated catalytic cleavage, converting pro-apoA-I to mature apoA-I, and stimulating ABCA1mediated cholesterol flux (26,44). However, our studies show conclusively that pro-apoA-I processing was not altered in the case of PCPE2 deficiency (Fig. 4). The absence of PCPE2 did not affect the circulating levels of pro-apoA-I in vivo, which remained around 3-6% of total apoA-I for both PCPE2-replete and PCPE2 knock-out mice. Although PCPE2 enhanced pro-apoA-I processing in vitro and was shown to have a high affinity for apoA-I using both surface plasmon resonance and selective co-immuno-precipitation (26), its absence in vivo did affect the ratio of pro-to mature apoA-I. This was not unexpected, because PCPE2 is not essential for catalytic cleavage of pro-apoA-I by BMP1 (39,42). Theoretically, PCPE2 may limit BMP1-mediated processing of pro-apoA-I under high substrate conditions, as reported for the processing of procollagen in the heart following chronic pressure overload (70).
Insight into the molecular mechanism of the protective function of PCPE2 comes from our HDL clearance studies (Fig. 5, A  and B), where the fractional clearance of HDL was delayed in PCPE2-deficient mice, regardless of the source of the HDL. A slower catabolic turnover resulted in increased HDL concentration, which ultimately caused HDL particle enlargement (Fig. 1). In mice, increased HDL size and concentration is a hallmark of SR-B1 deficiency (78). Hepatic specific SR-B1-deficient mice were reported to have higher plasma HDL levels than the controls, with greater aortic lesion formation. Because hepatic SR-B1 mediates selective uptake of HDL-cholesteryl esters and their removal from circulation for excretion into bile (74), it was concluded that in mice SR-BI is a positive regulator of macrophage RCT. As seen in genetically modified mice, humans with genetic variants in SR-BI show a reduced capacity to efflux CE, which leads to greater HDL plasma concentrations (79 -82) and a greater risk for CVD. Our study showed a macrophage to fecal RCT rate significantly lower in LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice and similar to that reported for SR-BI knockout mice. Interestingly, this lower rate occurred despite a 2-fold higher level of SR-BI protein in the livers of LDLr Ϫ/Ϫ , PCPE2 Ϫ/Ϫ mice (Fig. 5, C-F).
The impact of PCPE2 on liver SR-BI functionality was demonstrated by slower plasma HDL-CE turnover, reduced macrophage cholesterol to fecal deliver, and increased plasma HDL size and concentration. Despite elevations in plasma HDL, the aortic lipid and immune cell infiltration was greater in the absence of PCPE2, highlighting the importance of aortic SR-BI in preventing the progression of atherosclerosis. These results prompted us to consider how PCPE2 might influence hepatic SR-BI function.
Transient transfection of CHO cells with PCPE2 gave a 2-fold increase in PCPE2 surface expression (Fig. 6, A and C) and an almost 2-fold increase in the uptake of 3 H-COE from labeled HDL (Fig. 6B). Overexpression of PCPE2 in CHO cells did not change total SR-BI protein levels but did increase the amount of SR-B1 protein measured on the surface of transfected cells (Fig. 6D). In contrast, liver SR-BI expression in PCPE2 knock-out mice was increased 2-fold. The absence of an increase in SR-B1 protein following transfection may be ascribed to the transient nature of the in vitro experiment.
Our studies also provide evidence for a profound reduction in connective tissue content in both the aortic root and the hepatic triad in mice lacking PCPE2 (Fig. 3, A-D). PCPE2 enhances procollagen processing and is abundantly expressed in the heart, aorta, adipose tissue, and trabecular meshwork (27,29,38). The expression of PCPE1, which is related to PCPE2, was not increased as a result of PCPE2 deletion (Fig. 3, C and D) and did not substitute for PCPE2 in arterial tissue. The consequences of these changes in connective tissue content are not known at this time but may be related to the positioning and or stabilization of SR-B1 on the cell surface.
The mechanistic steps involved in SR-BI-mediated aortic cholesterol homeostasis have been studied for years, but questions still remain. How SR-B1 mediates selective CE removal from HDL at the cell surface, leaving lipid-poor apoA-I after transferring CE into the cell, remains unresolved. Nevertheless our results strongly suggest that SR-BI-mediated 3 H-COE removal from HDL is facilitated by surface-bound PCPE2. The mechanism of the influence of PCPE2 on SR-BI uptake of HDL-CE may involve promoting the oligomerization of SR-BI with itself, with PCPE2, or with other proteins. Previous studies have demonstrated that SR-BI forms dimers and tetramers on the cell surface necessary for its function (83)(84)(85). Oligomerization has been postulated to form a hydrophobic channel that facilitates the selective delivery of CE from HDL to the cell. Although putative SR-BI oligomerization domains are thought to localize to the transmembrane segment, it has been suggested that the extracellular domains of SR-BI may also contribute to oligomerization (86). The shift in SR-BI surface stain intensity (Fig. 6D) suggests a significant change in antibody binding properties and, therefore, a change in the conformation of amino acid region 338 -440.
Alternative explanations for the action of PCPE2 on SR-BI include the postulation that PCPE2 binds and sequesters SR-B1 at the cell surface. Previous studies have reported that the intracellular adaptor protein PDZK1 binds the C terminus of SR-BI and is required for receptor function (87,88). In an analogous way, PCPE2 may provide scaffolding or tethering at the cell surface, promoting association of SR-B1 with HDL-CE. Another possibility would be that PCPE2 is involved in SR-BI endocytosis (89 -92). Yet another plausible mechanism is based on the previously shown ability of PCPE2 to bind apoA-I (26); here PCPE2 would interact with apoA-I bound to HDL, thereby altering the protein conformation and in turn enhancing CE removal. As the current report provides a solid foundation for our understanding of PCPE2 function, additional studies will be needed to examine the structures of and potential interactions between PCPE2 and SR-BI, allowing insights into the mechanisms through which PCPE2 controls SR-BI-mediated HDL-CE uptake.
In conclusion, our studies show for the first time that an extracellular matrix protein, PCPE2, is essential for the integrity of the HDL-mediated CE transport system. These studies also show that simply measuring plasma HDL concentration or its associated components in order to assess one's risk of CVD can be misleading. Rather, a combination of HDL production, catabolism, and concentration is essential for a precise analysis of HDL function. The novel finding is that PCPE2 contributes to SR-BI function by enhancing the rate of HDL-CE uptake, thereby stimulating the RCT pathway. Because the molecular events that drive the selective uptake of CE from HDL particles via SR-BI have yet to be completely delineated, the questions generated by this work will lead to new studies designed to define those mechanisms. Overall, these studies will ultimately lead to a precise understanding of how SR-BI mediates selective CE uptake and provide new insights into how extracellular matrix-associated proteins, such as PCPE2, protect against the development of atherosclerosis.