This paper is available on line at ...homepages.uc.edu/~davidswm/Proteome_function.pdf · floating cells. Total cellular efflux was determined by dividing the 3H-cholesterol in the

Mapping Atheroprotective Functions andRelated Proteins/Lipoproteins in SizeFractionated Human Plasma*□S

Debi K. Swertfeger§, Hailong Li§, Sandra Rebholz§¶, Xiaoting Zhu§, Amy S. Shah�,W. Sean Davidson¶, and Long J. Lu‡§**

HDL has been shown to possess a variety of cardio-protective functions, including removal of excess choles-terol from the periphery, and inhibition of lipoproteinoxidation. It has been proposed that various HDL subpar-ticles exist, each with distinct protein and lipid composi-tions, which may be responsible for HDL’s many func-tions. We hypothesized that HDL functions will co-migratewith the operational lipoprotein subspecies when sepa-rated by gel filtration chromatography. Plasma from 10healthy male donors was fractionated and the proteincomposition of the phospholipid containing fractions wasanalyzed by mass spectrometry (MS). Each fraction wasevaluated for its proteomic content as well as its ability topromote cholesterol efflux and protect low density lipo-protein (LDL) from free radical oxidation. For each func-tion, several peaks of activity were identified across theplasma size gradient. Neither cholesterol efflux or LDLantioxidation activity correlated strongly with any singleprotein across the fractions. However, we identified mul-tiple proteins that had strong correlations (r values >0.7,p < 0.01) with individual peaks of activity. These proteinsfell into diverse functional categories, including those tra-ditionally associated with lipid metabolism, as well asalternative complement cascade, innate immunity andclotting cascades and immunoglobulins. Additionally, thephospholipid and cholesterol concentration of the frac-tions correlated strongly with cholesterol efflux (r � 0.95and 0.82 respectively), whereas the total protein contentof the fractions correlated best with antioxidant activity

across all fractions (r � 0.746). Furthermore, two previ-ously postulated subspecies (apoA-I, apoA-II and apoC-1;as well as apoA-I, apoC-I and apoJ) were found to havestrong correlations with both cholesterol efflux and anti-oxidation activity. Up till now, very little has been knownabout how lipoprotein composition mediates functionslike cholesterol efflux and antioxidation. Molecular &Cellular Proteomics 16: 10.1074/mcp.M116.066290, 680–693, 2017.

The risk of cardiovascular disease has been shown to beinversely related to high density lipoprotein cholesterol (HDL-C)1 levels in large human cohorts (1, 2). Although the exactmechanism that underlies this relationship has not been iden-tified, numerous functions that are seen as atheroprotectivehave been attributed to HDL. For example, studies haveshown that injecting 3H-cholesterol-labeled macrophagesinto mice that overexpress ApoA-I, the most abundant proteinon HDL, results in a significant increase of 3H-cholesteroldetected in the HDL and feces (3). This data supports thewidely accepted ‘reverse cholesterol transport’ (RCT) hypoth-esis (4) which invokes HDL as the primary vehicle for move-ment of excess cholesterol out of the periphery, in which cellsgenerally lack the ability to catabolize cholesterol, and back tothe liver for excretion through the feces. Aside from RCT, HDLhas been shown to have other potentially anti-atheroscleroticeffects. It has well documented antioxidative properties andhas been shown to prevent oxidative modification of lowdensity lipoprotein (LDL) thus reducing macrophage foam cellgeneration in the vessel wall (5). It can also inhibit the expres-sion of cell adhesion molecules on endothelial cells to preventinappropriate capture of circulating monocytes (6–9), andreduce the activity of macrophage chemotactic factor 1 whichsignals the infiltration of surface-adhered monocytes into the

From the ‡School of Information Management, Wuhan University,Wuhan 430072, China; §Division of Biomedical Informatics, Cincin-nati Children’s Hospital Research Foundation, 3333 Burnet Avenue,Cincinnati, Ohio 45229–3039; ¶Center for Lipid and ArteriosclerosisScience, Department of Pathology and Laboratory Medicine, Univer-sity of Cincinnati, 2120 East Galbraith Road, Cincinnati, Ohio 45237–0507; �Division of Endocrinology, Cincinnati Children’s HospitalResearch Foundation, 3333 Burnet Avenue, Cincinnati, Ohio 45229–3039

Received December 7, 2016, and in revised form, February 13,2017

Published, MCP Papers in Press, February 21, 2017, DOI 10.1074/mcp.M116.066290

Author contributions: D.K.S., A.S.S., W.D., and L.J.L. designedresearch; D.K.S. and S.R. performed research; D.K.S., H.L., and X.Z.analyzed data; D.K.S., H.L., X.Z., and W.D. wrote the paper.

1 The abbreviations used are: HDL, high density lipoprotein; LDL,low density lipoprotein; VLDL, very low density lipoprotein; apo, apo-lipoprotein; CVD, cardiovascular disease; RCT, reverse cholesteroltransport; PL, phospholipid; CH, cholesterol; UC, ultracentrifugation;GF, gel filtration; PR, propagation rate; PCC, Pearson correlationcoefficient; ATCC, American Type Culture Collection; 8-Br-cAMP,8-Bromoadenosine 3�,5�-cyclic monophosphate sodium salt.

Research© 2017 by The American Society for Biochemistry and Molecular Biology, Inc.This paper is available on line at http://www.mcponline.org

crossmark

680 Molecular & Cellular Proteomics 16.4

http://crossmark.crossref.org/dialog/?doi=10.1074/mcp.M116.066290&domain=pdf&date_stamp=2017-2-21

vessel wall (10, 11). These varied functions may all contributeto HDL’s well documented association with atheroprotection.

HDL’s ability to carry out multiple atheroprotective func-tions may be explained by its compositional heterogeneity.Recent proteomic studies show HDL is highly compositionallyheterogeneous, composed of particles, most all of whichcontain apoA-I, with related physico-chemical properties, butdiffer widely in their compositions and functions (12–17). Ad-ditionally, correlational network analysis identified distinctprotein clusters, which may represent specific HDL subspe-cies (15). Many of the anti-atherosclerotic functions of HDLhave been attributed to the class as a result of in vitro exper-imentation with ultracentrifugally isolated samples. Previouswork has shown that in ultracentrifugally isolated HDL sub-classes, dense HDL3 was the most effective at inhibition ofLDL oxidation, and that several proteins were highly corre-lated with this activity (15, 18). However, it is becoming clearthat ultracentrifugation may selectively isolate certain subspe-cies and alter the proteomic content of others (19, 20). Indeed,studies using gel filtration separation of human plasma havesuggested the presence of many additional proteins that areeither not found in UC-isolated HDL or are found in muchlower abundance (12).

We have hypothesized that HDL’s role in CVD may bemodulated by distinct HDL subspecies. To test this, we stud-ied two of the most widely recognized functional roles of HDL,cholesterol efflux and ability to prevent LDL oxidation. Insteadof using ultracentrifugation, we separated plasma from 10healthy, normolipidemic males by gel filtration and comparedthe fractions in the functional assays at equal volumes. Thisallowed us to understand the contribution of different HDLsize species in relation to other plasma components includingLDL. We also tracked the proteome across all fractions inorder to derive relationships between the protein contents ofthe lipoprotein species and their functions.

MATERIALS AND METHODS

Study Population and Plasma Collection—Ten healthy, normolipi-demic males (fasting cholesterol � 195 mg/dl; triglycerides � 150mg/dl) between the ages 18 and 40 (mean age 26.2 years) wererecruited. Inclusion criteria included non-smokers, body mass � 24.9,no history of taking lipid lowering medications, no diabetes, no historyof heart disease and C-reactive protein less than 1.0 mg/dl. Fastedblood was collected for a lipid panel. A second vial of blood wascollected for proteomic and functional analysis using citrate as ananticoagulant, and spun at �1590 � g for 15 min at room temperatureto isolate plasma. Plasma was stored at 4 °C until subjected to gelfiltration chromatography, within 2 h. Participants provided informedconsent according to an approved protocol as overseen by theinstitutional review board at Cincinnati Children’s Hospital MedicalCenter.

Plasma Fractionation—Plasma (354 �l) was fractionated on threeSuperdex 200 columns in series (10/300 GL; GE Healthcare Life-sciences, Pittsburgh, PA) as previously described (12). Fractions (1.5ml) were collected and stored at 4 °C for up to 1 week for furtheranalysis. Cholesterol and phosphatidylcholine containing phospho-lipid (PL) content of the fractions were analyzed within 24 h using

enzymatic kits (Pointe Scientific, Canton, MI and Wako, Richmond,VA). Protein content was measured using the Markwell modifiedLowry assay (21).

Mass Spectrometry Sample Preparation and Analysis—Plasmafractions (300 �l) were prepared for mass spectrometry (MS) aspreviously described (22). Briefly, samples were delipidated usingchloroform: methanol, and then treated with dithiothreitol and iodo-acetamide (Sigma Aldrich, St. Louis, MO) to reduce and carbamidom-ethylate the proteins prior to trypsinization. Trypsinized proteins weresubjected to MS using a nanoLC-MS/MS (AB Sciex5600 �TripleTOF)mass spectrometer as previously described (22) with the exceptionthat the trap column used was a Chrom XP C18-CL NanoLC TrapColumn (350 �m � 0.5 mm with a 3 �m packed bed) from Eksigent(Dublin, CA). Peakview version 2.1 was used to convert the raw datafiles into the peaklist (.mgf) files. The resulting mass spectra wereanalyzed with Mascot (version 2.2.2, www.matrixscience.com) andX! Tandem (version 2001.01.01.1) search engines against theUniProtKB/Swiss-Prot Protein Knowledgebase (2011, containing540,958 entries). Search criteria included: human taxonomy, andcarbamidomethylation as a variable modification; peptide tolerancewas set at �20 ppm, MS/MS tolerance set to � 0.6 Da, and up to 3missed trypsin cleavages allowed. Peptide and protein identificationfrom the MS/MS were validated using Scaffold software (version3.3.1) and only peptides with �95% identification probability andproteins with �99% identification probability were included in theanalysis. Additionally, at least 2 peptides from each protein wererequired to be considered in the analysis. False discovery rates wereless than 0.6% for peptide identification (calculated as the percentageof the sum of exclusive spectrum counts of decoy proteins divided bythe sum of exclusive spectrum counts of target proteins) and lessthan 0.1% for protein identification (calculated as number of decoyproteins divided by the number of target proteins). Using this method,we did not make any quantitative comparisons between differentproteins in each fraction. However, because equal volumes of eachfraction were used for analysis, the resulting spectral counts shouldbe proportional to the relative abundance of a given protein acrossfractions. Previous studies have confirmed that the spectral countsacross fractions correspond to their relative abundance acrossfractions based on immunological analyses (13, 15). When noted, theprotein list was selected for proteins that had previously been identifiedas being either LDL- or HDL- associated, as found on the LDL and HDLwatch list (http://homepages.uc.edu/�davidswm/HDLproteome.htmlhttp://homepages.uc.edu/�davidswm/LDLproteome.html). The com-plete list of proteins detected by MS can be found in Supplemental Figs.(supplemental Fig. S1). Originally, 206 proteins were identified, of which78 proteins were considered LDL- and HDL- associated proteins. Fifty-seven of the proteins excluded belonged to the immunoglobulin familyand 11 proteins were keratins. The mass spectrometry proteomics datahave been deposited to the ProteomeXchange Consortium via thePRIDE (23) partner repository with the dataset identifier PXD005520 and10.6019/PXD005520.

Cholesterol Efflux Assay—RAW 264.7 macrophage cells (ATCC,Manassas, VA) (4 � 105 cells) were grown in 48-well plates. Cells wereexchange labeled for 16 h with 0.5 �Ci 3H-cholesterol (Perkin Elmer,Waltham, MA) per well, in the presence of 0.3 mM 8-Bromoadenosine3�,5�-cyclic monophosphate sodium salt (8-Br-cAMP) (Sigma Al-drich). The next day, unincorporated 3H-cholesterol was washed off,and cells were treated for 6 h with or without cholesterol acceptors orfractions, in the presence of 8-Br-cAMP. Cholesterol acceptors usedfor controls included lipid-free apoA-I (10 �g/ml), or ultracentrifugallyisolated HDL (10 �g/ml protein). To evaluate the efflux capacity ofplasma fractions, 20 �l of each fraction was added to the cell me-dium. 3H-cholesterol detected in the medium was measured usingliquid scintillation counting after filtering the medium to remove any

Atheroprotective Functions in Size Fractionated Plasma

Molecular & Cellular Proteomics 16.4 681

http://www.matrixscience.comhttp://homepages.uc.edu/∼davidswm/HDLproteome.htmlhttp://homepages.uc.edu/∼davidswm/LDLproteome.htmlhttp://www.mcponline.org/cgi/content/full/M116.066290/DC1

floating cells. Total cellular efflux was determined by dividing the3H-cholesterol in the medium, by the total 3H-cholesterol contentextracted from a separate set of control wells that were measured atthe end of the overnight labeling period. Background efflux wassubtracted from each fraction. To compare between experiments,efflux was normalized to a standard UC-HDL pool that was includedin each experiment. All fractions were evaluated in triplicate. Higherefflux indicates better HDL function. PeakFit software (Systat Soft-ware, San Jose, CA) was used to identify component peaks from theresulting cholesterol efflux curves using the residuals best fit method.The resulting component peaks correlated well with the original dataand had an r2 value of 0.98.

LDL Oxidation Assay—LDL oxidation was measured using thepropagation rate (PR) of oxidation in a 96-well UV plate. LDL (20 �gcholesterol) was added to each well. To measure the antioxida-tive capacity of the plasma fractions, 50 �l of each fraction waspre-incubated with the LDL at 37 °C for 10 min. Oxidation productswere measured by the production of conjugated dienes, initiated bythe addition of 2,2,- Azobis (2-methylpropionamidine) dihydrochloride(AAPH, Acros Organics, Geel, Belgium) (5 mM final concentration).Accumulation of conjugated dienes was measured at A234 for 8 h at37 °C. The maximal slope of each curve represents the propagationrate (PR). Data are expressed as percent inhibition of LDL oxidation,as represented by PR, relative to LDL oxidation in the absence of anyadditions. Higher inhibition of LDL oxidation indicates “better” HDLfunction. Fractions from each volunteer were analyzed in duplicate.PeakFit software was used to identify individual component peaksfrom the resulting oxidation curves using the residuals best fitmethod. The resulting component peaks correlated well with theoriginal data and had an r2 value of 0.98.

Correlations of Plasma Fraction Proteomic Data with FunctionalData—Pearson correlation coefficients (PCC), as mathematicallyshown below, were calculated to assess the relationship betweenprotein total spectral counts and functional profiles across fractions ofindividual peaks. Assume that Xij

k is the spectral count of protein Px inthe plasma fraction j of peak k from subject i. Given one functionalassay F, let Yij

k denote functional activity measurement in fraction j ofpeak k from the same subject i. Then, for each protein and functionpair (Px, F), the PCC of peak k was calculated by:

Rk �Px,F ��i�j�Xijk � X� �Yijk � Y�

��i�j�Xijk � X� 2��i�j�Yijk � Y� 2where X� and Y� are the mean of spectral count Xij

k and functionalactivity Yij

k. Thus, PCC, as represented by Rk, simultaneously reflectsboth intra- and inter-subject correlation of plasma fractions proteomicdata with functional data. The p value was calculated using the t testto test the null hypothesis that there is not a linear correlation betweenthe abundance of proteins and functional activities against the alter-native that there is a nonzero linear correlation. In this study, wedefined Rk as the local correlation of individual peaks.

Correlations of HDL Subspecies with Functional Data—Linear cor-relation of HDL subspecies abundance and functional activity wasalso assessed using PCC. Putative HDL subspecies that are sup-ported by multiple pieces of evidence were selected from our priorwork (14). These putative subspecies were identified using a networkbased approach searching for common migration patterns of proteinsseparated using three orthogonal separation techniques. Networkmining was performed using a maximal clique algorithm so that a totalof 30 HDL subspecies were identified.

Existence of these subspecies was tested within individual plasmafractions by examining the non-zero MS spectral counts of all proteinmembers. If all protein members of an HDL subspecies co-existedwithin that fraction, we calculated the mean of the normalized spectral

counts of all members except apoA-I to estimate the elution patternsof the HDL subspecies. Using the normalized data effectively re-moved the bias of the more abundant proteins on the overall distri-bution pattern of the putative subspecies, and instead shifted theemphasis to the correlation of the complex with activity. ApoA-I wasleft out of the mean calculation because it is likely a member ofmost HDL subspecies, and the relative contribution of apoA-I to anygiven subspecies cannot be determined. Finally, PCC was obtainedto represent the relationship of HDL subspecies with functionalactivities.

Experimental Design and Statistical Rationale—Proteomic contentwas analyzed on 18 fractions from 10 volunteers (total 180 samples).Triplicate measurements of cholesterol efflux capacity were obtainedfor each plasma fraction from all volunteers. Duplicate measurementsof antioxidative capacity were made on 18 fractions from all 10volunteers. All data are expressed as mean � standard deviationunless otherwise specified. When necessary, Bonferroni multiple cor-rections were applied to account for multiple comparisons.

RESULTS

Study Populations—Table I defines the characteristics ofthe study population. The mean age of the subjects was26.2 y. In addition to having normal values for total choles-terol, HDL-C, LDL-C, triglycerides, and glucose, their C-reac-tive protein levels were below 1.0 mg/dL.

Lipoprotein Distribution of Fractionated Plasma—Plasmasamples from each individual were fractionated using threetandem Superdex 200 columns, as described in Materials andMethods. Previous work (12, 13) has shown that the lipopro-teins, as identified by the combination of PL, cholesterol andprotein, elute from the columns in fractions 13–30, whichcorrespond to a size range of roughly 30 to 927 kDa. Fig. 1shows the distribution profiles of cholesterol, PL and proteinacross this range. Two distinct cholesterol and PL peaks wereobserved. Proteins typically found in VLDL/LDL were foundin the first peak (fractions 15–20) (supplemental Fig. S1).Proteins typically associated with HDL were seen in thesecond peak (fractions 20–28) (supplemental Fig. S1). Be-cause these separations depend on molecular size and notdensity, it is not accurate to refer to these using the densitybased designations of VLDL, LDL or HDL. However, to beconsistent with convention, we refer to these ranges offractions as the VLDL/LDL and HDL size ranges. Albuminand other smaller free plasma proteins eluted in fractions28–30 (12).

TABLE ICharacteristics of study population

n 10

Age (years) 26.2 � 6.3BMI (kg/m2) 22.9 � 1.5TC (mg/dL) 160.2 � 18.4TG (mg/dL) 81.4 � 24.9LDL-C (mg/dL) 92.8 � 17.7HDL-C (mg/dL) 51.1 � 10.5Glucose (mg/dL) 91.5 � 8.0Systolic blood pressure (mm Hg) 120.8 � 10.5Diastolic blood pressure (mm Hg) 72 � 7.5



http://www.mcponline.org/cgi/content/full/M116.066290/DC1http://www.mcponline.org/cgi/content/full/M116.066290/DC1

Distribution of Proteins Detected by Mass SpectrometryAcross Triple Superdex Fractions—Individual plasma frac-tions were delipidated and prepared for LC-MS/MS analysis.Proteins were identified in Scaffold using Mascot and X! Tan-dem, as described in Materials and Methods. The abundance(as represented by total spectrum counts) of individual pro-teins across the fractions were summed across all subjectsand normalized. For normalization of each protein, we sub-tracted the mean from the summed spectral counts of indi-vidual fractions, and then divided the individual values by itsstandard deviation. The resulting values represent the relativedistribution of an individual protein across fractions. The re-sulting 206 proteins were then further selected using the LDLor HDL watch list (Materials and Methods) to identify proteinsthat were known to associate with LDL or HDL, resulting in alist of 78 proteins. We then applied a hierarchical clusteringapproach in GENE-E software (24) to group the proteinsbased on their distribution patterns across fractions. The heatmap shown in Fig. 2 illustrates the distribution of each proteinand the clustering result. The first cluster is comprised ofapoB-containing lipoproteins, and were found in the larger PLcontaining fractions (fraction 15–17), also called large VLDL/LDL fractions. Subunits of fibrinogen that are commonly ob-served in LDL were distributed across fraction 17–20 (cluster2), which are defined as small VLDL/LDL fractions. ApoA-I andapoA-II were found in all PL containing fractions, as previouslyreported (12). However, there was significantly more apoA-Iand apoA-II in the second PL peak (fraction 20–28, cluster 4)compared with the first. ApoA-I and apoA-II have been rec-ognized as a scaffold for other HDL-associated proteins (13,25–28). The overlap in the clusters 3–5 indicate that someproteins in cluster 4 (e.g. apoA-I or apoA-II) may carry otherproteins in each of cluster 3, 4 or 5, comprising separatesubspecies. Thus, we defined fractions 20–24 as large HDLsize fractions and fractions 25–28 as small HDL size fractions.The proteins in cluster 6 were mostly distributed in fractions

28–30, which we called the minimally lipidated/free proteinsrange. Each of these newly defined size-based fractions maycontain one or more PL-associated subspecies. BecauseapoA-I and apoA-II were detected in every PL containingfraction (13–30), we analyzed the HDL functional activities inall fractions.

Cholesterol Efflux Capacity of Plasma Fractions—The abilityof each PL containing fraction (fractions 13–30) to promotecholesterol efflux was measured for all ten participants. Themouse macrophage cell line, RAW 264.7 cells, were exchangelabeled with 3H-cholesterol in the presence of 8-Br-cAMP toupregulate ABCA1. Fig. 3A shows the cholesterol efflux ca-pacity (mean and standard deviation) of each fraction. PeakFitsoftware (29) was used to mathematically model individualpeaks of efflux from the entire range of phospholipid contain-ing fractions. This helped us to better identify the differentcomponents contributing to total cholesterol efflux capacity.Using the residuals best fit method, three peaks of activitywere identified, as shown in Fig. 3B. Peak 1 corresponds tothe LDL range, peak 2 to the HDL range, and peak 3 to theminimally lipidated/free protein range of fractions. The two PL-rich peaks clearly demonstrate a stronger cholesterol effluxcapacity than the third peak in the minimally lipidated/free pro-teins range, again when compared on an equal volume basis.

Correlation of Phospholipid with Cholesterol Efflux ofPlasma Fractions—Previous reports have suggested a stronglink between PL content and cholesterol efflux. In order todetermine whether or not PL was a strong predictor of thecholesterol efflux activity of the plasma fractions, we looked atthe correlation of PL, total cholesterol and protein content ofthe fractions and their ability to stimulate cholesterol efflux.Fig. 4A shows a very strong linear correlation between theaveraged PL content of each fraction from all 10 volunteersand the corresponding averaged efflux capacity of each frac-tion (r 0.953, p � 0.001). Previous studies have shown thatthe PL content of HDL is directly related to its ability to effluxcholesterol, likely because the PL provides a large sink forcholesterol solubilization (30–35). Total cholesterol alsoshowed a high linear correlation (r 0.82, p � 0.001) withefflux capacity, although not as strong as the PL correlation(Fig. 4B). Interestingly, it appears that as the cholesterol con-tent of the fractions increases, there appears to be a levelingoff of the cholesterol efflux activity. Indeed, the data can alsofit a 2-site saturation with nonspecific binding curve modelwith a higher correlation (r 0.89, p � 0.001) (supplementalFig. S2). However, cholesterol efflux is a complex, multicom-ponent process. It is thus, difficult to interpret the significanceof this type of correlation. In contrast, the protein content inindividual fractions does not exhibit any discernable correla-tion with efflux activity (Fig. 4C).

Correlation of MS Identified Proteins to Cholesterol Efflux—Next, we sought to determine if any individual proteins, incontrast to total protein content of each fraction, are corre-lated with cholesterol efflux. Because we have shown that

FIG. 1. Characterization of fractionated human plasma. Humanplasma (354 �l) was separated on 3 Superdex 200 columns in series.Fractions (1.5 ml) were collected and analyzed for PL (open circles),cholesterol (closed circles) and protein content (closed diamonds).Data represent the average data from 10 volunteers, except for theprotein data, which show data averaged from 3 volunteers.



http://www.mcponline.org/cgi/content/full/M116.066290/DC1http://www.mcponline.org/cgi/content/full/M116.066290/DC1

total phospholipid is the strongest predictor of efflux activity,we limited our search to proteins that have already beenshown to be associated with LDL or HDL by mass spectrom-eter studies. Proteins in each fraction, identified by MS, werecorrelated with the individual cholesterol efflux peaks usingPCC. Table II shows the top ranked proteins based on theircorrelation with cholesterol efflux activity of each peak. Inpeak 1 (LDL/VLDL range), the top efflux correlated protein is

apoB, the major component of LDL. In peak 2, apoA-I andapoA-II are strongly correlated with efflux activity, as well asimmunoglobulins and serum amyloid P. They are distributedacross both large and small HDL fractions. This is consistentwith data showing apoA-I and apoA-II’s ability to promotecholesterol efflux (36–40). In peak 3, we identified severalminimally lipidated/free proteins, including anti-thrombin 3,�-1-antitrypsin, transthyretin and albumin. Fig. 5 shows the

FIG. 2. Proteomic heat map with hierarchical clustering of MS identified proteins across the PL containing plasma fractions.Individual proteins were identified by MS and abundance of proteins across fractions was then normalized as described in Materials andMethods. The list has been filtered to contain only those proteins found on the human HDL or LDL watch list.



distribution of the proteins most highly correlated with eachpeak’s cholesterol efflux activity.

Antioxidative Capacity of Plasma Fractions—We also as-sessed the plasma fractions from our 10 participants for theirantioxidative capacity. The capacity of each fraction to inhibitLDL lipid peroxidation was measured using AAPH to initiateLDL oxidation (Materials and Methods). Inhibition of oxidationis calculated as the percent decrease in PR relative to LDLonly. When antioxidant activity is plotted across fractions, Fig.6A shows that there are 3 peaks of activity where fractionsinhibit LDL oxidation. Individual peaks of antioxidative activitywere deconvoluted using PeakFit software. Fig. 6B shows theidentified component curves, as well as the sum of the indi-vidual curves, as identified using the residuals best fit method.Peak 1 (fractions 16–21) is found in the LDL size range frac-tions, peak 2 (fractions 22–27) is found in the HDL range offractions, whereas peak 3 (fractions 26–20) represents frac-tions containing minimally lipidated or free protein. Althoughthe PeakFit software identified three component peaks of

oxidation, similar to the efflux curves, we noted that thesepeaks are not identical to the component efflux peaks.

Protein Content Correlates with Antioxidant Capacity—Sim-ilar to cholesterol efflux activity, we sought to determine if thetotal PL, cholesterol or protein content of the plasma fractionsare correlated to the corresponding antioxidative capability.Fig. 7 shows that, unlike cholesterol efflux which correlatedwith PL and cholesterol content, antioxidative activity ofplasma was most closely correlated with total plasma proteinlevel (Fig. 7C; r 0.746, p � 0.0005). This is consistent withprevious work determined that total plasma proteins ac-counted for a major portion of the total plasma antioxidantcapacity (41). Interestingly, the data in Fig. 7C can also fit amodel of saturation with higher statistical accuracy (r 0.87,p � 0.0001) (supplemental Fig. S3), however, the interpreta-tion of this is hindered by the complexity of the system.Because the antioxidant activity of the fractions are likely dueto multiple components in each fraction, we are unable todetermine the meaning of this curve fit. On the other hand,

FIG. 3. Cholesterol efflux across all triple superdex fractions. A, Total cholesterol efflux capacity of each fraction, loaded at equal volume,is shown. Efflux measurements were performed on each fraction in technical triplicates. Data are background subtracted (fraction 13 was setas background) and normalized to a standard UC-HDL sample that was analyzed on each plate to compare across experiments. Data shownrepresent the average and standard deviation of 10 volunteers. B, PeakFit software v.4.12 was used to identify individual component peaksderived from the cholesterol efflux data using the residuals method. Individual peaks (peak 1, closed circles; peak 2, open circles; peak 3,closed triangles) identified using the residuals methods, as well as the sum of the peaks (open triangles) are shown.

FIG. 4. Correlation between cholesterol efflux and PL, cholesterol or protein levels of plasma fractions. Scatterplots showing thecorrelation of PL (A), cholesterol (B) and protein (C) concentrations to the efflux capacity of individual fractions. Data represent the average PL,cholesterol and protein content of each fraction plotted against the average normalized efflux of that same fraction. Pearson correlationcoefficient for PL is 0.950 and the p � 0.0001; for cholesterol is 0.82, p � 0.0001 and for protein is 0.06 (n.s.).



http://www.mcponline.org/cgi/content/full/M116.066290/DC1

neither PL nor CH demonstrated a clear correlation to anti-oxidative capability (Fig. 7A, 7B).

Correlation of MS Identified Proteins to Inhibition of LDLOxidation—We assessed the correlation of individual proteinswith antioxidant activity using the Pearson’s correlation coef-ficient. As with cholesterol efflux, we selected proteins thatare lipoprotein associated. No specific proteins correlatedwith the entire set of fractions and their ability to inhibit LDLoxidation. However, significant correlations were identifiedbetween individual proteins and individual peaks of activity(Table III). It is interesting to note that in the small VLDL/LDLrange fractions, the most highly correlated proteins were threesubunits of fibrinogen, followed by �-2 macroglubulin. In theHDL range, in addition to some immunoglobulins, the mosthighly correlated proteins included apoA-II, serum amyloid P,inter-�-trypsin inhibitor heavy chain H1, apoA-I and a fewdifferent complement components. In the minimally lipidated/free protein fractions, the antioxidation correlated proteinsinclude �-1 antitrypsin, albumin and serotransferrin. Thesehighly correlated proteins are overlaid with the three anti-oxidation peaks in Fig. 8A (peak 1), 8B (peak 2) and 8C(peak 3).

Correlation of Previously Identified Putative Subspecieswith HDL Functions—We further wanted to determine if therewas any correlation of cholesterol efflux or antioxidant activitywith lipoprotein subspecies that we had previously postulatedusing a systematic computational approach (14). As de-scribed in the Methods section, PCC coefficient was calcu-lated to determine how well each putative subspecies corre-lated with either cholesterol efflux or antioxidant activity.Using the list of putative subspecies previously identified, weidentified 2 subspecies that are highly correlated with choles-terol efflux activity and 3 subspecies that associate signifi-cantly with antioxidant activity. As shown in Fig. 9A, choles-

terol efflux activity in peak 2 was highly correlated withcomplex 1: apoA-I, apoA-II, and apoC-I, with a PCC of 0.961as well as complex 2: apoA-I, apoC-I, and apoJ, with a PCCof 0.897. Fig. 9B shows the subspecies highly associated withantioxidant activity. Two subspecies were identified that cor-related highly with peak 2. ApoA-I, apoA-II and apoC-I com-prised one correlated complex, whereas the other wasapoA-I, apoC-I and apoJ. Both subspecies that correlatedwith peak 2 of antioxidant activity had a PCC value of greaterthan 0.8, indicating a strong correlation. Additionally, the fi-brinogen complex (FibA, FibB, and FibG) is highly correlatedwith antioxidant activity in peak 1, with a PCC of 0.98 indi-cating a nearly perfect correlation with antioxidant activity.The correlation of the fibrinogen complex with antioxidantfunction is in agreement with Olinescu et al. and Abudu et al.(42, 43), who have also shown that fibrinogen has antioxidantactivity.

DISCUSSION

HDL is a heterogeneous population of particles with distinctsize, charge and compositional characteristics. Given thewide variety of functions associated with HDL proteins, itstands to reason that different subpopulations may exhibitdifferent functionalities. Here, we investigated cholesterol ef-flux and protection of LDL from oxidation, arguably two ofHDL’s most attributed functions. Rather than study the HDLparticles, purified by potentially perturbing ultracentrifugation(19), we investigated these functions in the context of size-fractionated human plasma. This allowed the analysis of rel-atively native lipoproteins and the correlation of these func-tional activities with the proteomic composition of thefractions. Based on our proteomic analysis of size basedfractionation of human plasma, we have indeed shown that avariety of HDL subspecies exist, as defined by their hetero-

TABLE IICorrelation of cholesterol efflux activity with MS spectral counts

Efflux Peak 1 Efflux Peak 2 Efflux Peak 3

r p r p r p

apolipoprotein B 0.944 4.29E-09 apolipoprotein A-II 0.993 2.36E-16 antithrombin-III 0.967 6.88E-11IGHM 0.912 1.31E-07 Ig kappa chain C region 0.978 3.02E-12 �-1-antitrypsin 0.965 1E-10C4b-binding protein � chain 0.749 0.0003 Ig gamma-1 chain C region 0.974 1.02E-11 transthyretin 0.958 4.1E-10fibrinogen � chain 0.720 0.0008 apolipoprotein A-I 0.951 1.46E-09 albumin 0.935 1.35E-08fibrinogen � chain 0.710 0.0010 serum amyloid P 0.944 4.16E-09 �-1-acid glycoprotein 1 0.849 8.39E-06fibrinogen � chain 0.709 0.0010 complement C3 0.894 5.83E-07 serotransferrin 0.836 1.55E-05apolipoprotein(a) 0.564 0.0147 inter � trypsin inhibitor 4 0.838 1.41E-05 �-1-acid glycoprotein 2 0.830 2.04E-05

apolipoprotein C-I 0.832 1.87E-05 apolipoprotein H 0.826 2.37E-05apolipoprotein J 0.764 0.0002 angiotensinogen 0.765 0.0002paraoxonase 1 0.760 0.0003 vitamin D binding protein 0.754 0.0003apolipoprotein D 0.658 0.0030 apolipoprotein A-IV 0.697 0.0013PtdIns-glycan-specific

phospholipase D0.646 0.0038 kallistatin 0.690 0.0015

ceruloplasmin 0.614 0.0067 retinol binding protein 4 0.684 0.0017lecithin:cholesterol

acyltransferase0.564 0.0149 gelsolin 0.655 0.0032

complement C2 0.560 0.0157 zinc-�-2-glycoprotein 0.650 0.0035lumican 0.530 0.0238 �-2-HS-glycoprotein 0.622 0.0059plasma kallikrein 0.505 0.0327 plasminogen 0.617 0.0064

hemopexin 0.563 0.0149



geneous protein compositions. Furthermore, we demon-strated that the two HDL atheroprotective functions investi-gated were differentially influenced by the various plasmacomponents. For example, cholesterol efflux appears to bemost highly associated with phospholipid content of the frac-tion, whereas inhibition of LDL oxidation was more closelyassociated with total protein content. Each atheroprotectivefunction was correlated with different sets of proteins. Finally,we identified 2 separate putative subspecies that were highlycorrelated with both cholesterol efflux and inhibition of LDLoxidation (apoA-I, apoA-II, apoC-I; and apoA-I, apoC-I andapoJ), suggesting that together these proteins may have arole in HDL function.

Our analysis of the protein distribution of size separatedplasma further delineates the heterogeneity of lipoproteins. InFig. 2, we showed hierarchical clustering of proteins based ontheir distribution patterns. Similar co-migration patterns oftwo proteins indicate they may reside on the same particles.Thus, we defined five fraction ranges each containing proteinsthat migrated with similar patterns: large VLDL/LDL, smallVLDL/LDL, large HDL, small HDL and minimally lipidated/freeprotein fractions. These hierarchical grouping of fractions im-ply at least five types of subspecies. In fact, it is likely that theactual number of PL-associated subspecies is far more thanfive, because the number of proteins residing on the sameparticles is likely small because of their biophysical limitations.In any given PL-containing fraction, we detected between 13and 54 known HDL- or LDL-associated proteins. However, wedo not expect that all proteins reside on a single particlebecause of size constraints. Therefore, we expect that thereare multiple PL-containing subspecies in each fraction. Asverification of our method, we found one known HDL subspe-cies, trypanosome lytic factor (TLF) (27, 44, 45), which con-tains three main proteins, apoL-I, haptoglobin-related protein(HPR), and apoA-I, in fractions 19–23 (supplemental Fig. S1),demonstrating that our approach to separation does not dis-rupt the lipoprotein particles. Thus, according to our hierar-chical clustering analysis, ApoL-I and HPR are grouped in thecluster 3 across large HDL fractions, whereas apoA-I, likelyworking as a scaffold for proteins in cluster 3–5, is groupedwith cluster 4, which contains proteins found in both large andsmall HDL fractions (cluster 3 and 5) (Fig. 2). Considering thenumber of proteins detected in each fraction and their possi-ble combinations, dozens of subspecies may exist. In a recentstudy, a series of HDL subspecies candidates were inferredusing a network-based computational method (14). Experi-mental validation is still necessary to further our understand-ing of these plasma subspecies.

ApoA-I and apoA-II (HDL’s main structural proteins) weredetected in all phospholipid containing fractions, even in theVLDL/LDL size range. Thus, all phospholipid containing frac-tions were examined for both cholesterol efflux and antioxi-dation activity. Correlational analysis examining the proteomiccontent of each fraction with the HDL functions revealed that

FIG. 5. Overlay of proteins that correlate most highly with indi-vidual peaks of cholesterol efflux activity. Distribution of MS spec-tral counts across fractions are shown overlaid with the cholesterolefflux activity for the proteins most highly correlated with efflux of theindividual peaks for A, the first peak, [closed circles, apoB; opencircles, IGHM; closed triangles, C4BPA; open triangles fibrinogen �chain; gray circle, cholesterol efflux] B, the second peak [closedcircle, apoA-II; open circle, IGKC; closed triangle, IGHG1; open trian-gle, ApoA-I; and gray circle, cholesterol efflux] and C, the third peak[closed circle, Antithrombin 3; open circle, �-1-antitrypsin; closedtriangle, transthyretin; open triangle, albumin and gray triangle, cho-lesterol efflux.



http://www.mcponline.org/cgi/content/full/M116.066290/DC1

similar, but different subsets of proteins were linked to eachfunctional activity. For example, cholesterol efflux activity wasdetected in multiple peaks across the fractions, indicating that

cholesterol efflux is related to more than one subspecies, andthose multiple PL species may coordinate with each other topromote cholesterol efflux. In an attempt to identify subspe-

FIG. 6. Inhibition of LDL oxidation bygel filtration fractionated plasma. A,Plasma fractions (50 �l) were incubatedwith LDL and AAPH to initiate lipid peroxi-dation. Percent inhibition of LDL oxidationwas calculated using the propagation rate(PR) in the presence of each plasma frac-tion relative to PR of LDL only. B, Individ-ual components of the total oxidationcurve were identified using PeakFit. Threepeaks were identified that likely contributeto the entire range of antioxidation activity(peak 1, closed circles; peak 2, open cir-cles; peak 3, closed triangles, and thesum of all peaks, open triangle). The r2

value for the peak fit was 0.98.

FIG. 7. Correlation of antioxidative activity with total PL, cholesterol or protein content of fractions. Scatterplots showing a relativelystrong correlation with total protein content of fractions (r 0.746, p � 0.0005) (C), but the lack of correlation of PL (A) or CH content (B) offractions with antioxidative capacity of plasma fractions, (PCC of r �0.271 (PL) and �0.0243 (CH), and p n.s.).

TABLE IIICorrelation of antioxidant activity with MS spectral counts

Oxidation peak 1 Oxidation peak 2 Oxidation peak 3

r p r p r p

fibrinogen � chain 0.986 8.32E-14 Ig gamma-1 chain C region 0.975 7.71E-12 �-1-antitrypsin 0.934 1.43E-08fibrinogen � chain 0.976 5.26E-12 Ig kappa chain C region 0.973 1.37E-11 albumin 0.934 1.47E-08fibrinogen � chain 0.972 1.91E-11 apolipoprotein A-II 0.958 3.93E-10 serotransferrin 0.918 7.98E-08�2-macroglobulin 0.905 2.57E-07 serum amyloid P 0.949 2.01E-09 angiotensinogen 0.890 7.4E-07fibronectin 0.726 0.0006 inter � trypsin inhibitor 4 0.916 9.41E-08 antithrombin-III 0.889 7.98E-07apolipoprotein B 0.605 0.0078 complement C3 0.876 1.91E-06 transthyretin 0.883 1.21E-06IGHM 0.557 0.0164 apolipoprotein A-I 0.832 1.89E-05 apolipoprotein H 0.873 2.23E-06

apolipoprotein C-I 0.804 5.86E-05 gelsolin 0.835 1.62E-05apolipoprotein J 0.701 0.0012 �-2-HS-glycoprotein 0.813 4.07E-05ceruloplasmin 0.604 0.0080 plasminogen 0.791 9.26E-05complement C2 0.560 0.0157 �-1-acid glycoprotein 1 0.766 0.0002paraoxonase 1 0.539 0.0210 �-1-acid glycoprotein 2 0.740 0.0004Ptdlns-glycan-specific

phospholipase D0.501 0.0342 hemopexin 0.734 0.0005

�-1-antichymotrypsin 0.650 0.0035complement C9 0.599 0.0086�-1B-glycoprotein 0.576 0.0123vitamin D binding protein 0.567 0.0140apolipoprotein A-IV 0.564 0.0147�-2-antiplasmin 0.556 0.0167vitronectin 0.548 0.0184



cies that may be related to cholesterol efflux, we analyzed thetop proteins whose distribution across each efflux peak weremost highly correlated with cholesterol efflux activity. In termsof individual proteins, we found the top correlated proteins

consist of diverse functional categories including lipoproteinstraditionally involved with lipid metabolism, immunoglobulins,proteins associated with the alternative complement cascadeand innate immunity as well as many proteins associated withthe clotting cascade. The lipoproteins most commonly asso-ciated with cholesterol efflux were identified in peak 2 (HDLrange) and include the most abundant and central structuralapolipoproteins in HDL, apoA-I and apoA-II. Other lipopro-teins with high correlations in peak 2 include apoC-1 andapoJ (clusterin). Each of these apolipoproteins have previ-ously been shown to have the ability to efflux cholesterol fromcells (46–49). Besides these apolipoproteins well known fortheir efflux capacity, multiple immunoglobulins and variouscomponents of the alternative complement pathway, includ-ing complement C3, C5, C6 and C7 were found to be highlycorrelated with efflux activity in peak 2, along with proteins ofthe innate immune response, serum amyloid P and inter-�-trypsin inhibitor H4. In the LDL size range fractions, it appearsthat an apoB containing particle is most highly correlated withefflux activity. These studies utilized radio-labeled free cho-lesterol to specifically measure cholesterol efflux in one direc-tion, i.e. from the cell to the lipoproteins. It has been shownthat the net mass transfer of cholesterol between LDL andcells results in a net influx of cholesterol mass to the cell (50).Nevertheless, our work clearly shows that cellular cholesterolcan end up in LDL populations, suggesting that it participatesin cellular cholesterol homeostasis in more ways than simplecholesterol loading. Other proteins found to be highly corre-lated with efflux activity in peaks 1 and 3 include multiplechains of fibrinogen in peak 1 and anti-thrombin 3, �-1 anti-trypsin and albumin in peak 3. Many of these proteins havewell established functions in regulating the clotting cascade(51). However, it is unclear whether they contribute directly tocholesterol efflux, or if they are cargo proteins on lipoproteinparticles that have efflux capacity because of the presence ofapoA-I, because it is well established that minimally lipidatedapoA-I is the primary mediator of ABCA1 dependent choles-terol efflux (52, 53). Interestingly, we also found albumin to behighly correlated with efflux activity in peak 3. Consistent withour findings, multiple studies have previously shown that al-bumin has the capacity to accept cholesterol from cells (54,55). Each of these highly correlated proteins were also de-tected in our previous studies (12, 22), which analyzed onlyphospholipid bound proteins, suggesting that our highly cor-related proteins are lipid bound.

Inhibition of LDL oxidation is another potentially importantatheroprotective function of HDL. Similar to cholesterol effluxactivity, antioxidation function is clearly associated with mul-tiple subspecies in multiple peaks of activity across PL con-taining fractions. Indeed, there have been many studies show-ing a variety of antioxidants present in plasma. These includeboth small molecules, such as ascorbate, urate and vitamin E,as well as proteins, such as paraoxonase, transferrin, andalbumin (41, 56–64) Although our study does not address the

FIG. 8. Overlay of top identified proteins that correlate withinhibition of LDL oxidation. Distribution of MS spectral countsacross fractions are shown overlaid with the antioxidant activity of theindividual peaks for A, the first peak, [closed circle, fibrinogen � chain;open circle, fibrinogen � chain; closed triangle, fibrinogen � chain;open triangle, �-2-macroglobulin; gray circle, anti-oxidation activity]B, the second peak [closed circle, IgHG1; open circle, IgKC; closedtriangle, apoA-II; open triangle, serum amyloid P; gray circle, antioxi-dation activity] and C, the third peak of activity [closed circle, �-1-antitrypsin; open circle, albumin; closed triangle, serotransferrin; opentriangle, angiotensinogen; gray circle, anti-oxidation activity].



small molecules association with our antioxidation activity, wedid, however, find some interesting associations with specificproteins. Proteomic analysis showed that the proteins mosthighly correlated with antioxidative activity in peak 1 are fi-brinogen �, � and � chains, the three subunits of fibrinogenthat come together to form the fibrinogen complex. Severalprevious studies have detected fibrinogen associated withlipoprotein particles (12, 15, 16, 65, 66), but it is unclearwhether the association is specific, or is simply because of theabundance of the fibrinogen complex non-specifically stickingto the lipoprotein particles. The fibrinogen complex has amass of �342 kDa, and its peak elution fraction is fraction 18(Fig. 2). According to our calibration standards, proteins andprotein complexes with a MW of 340–417 kDa will elute infraction 18. Thus, using our sizing information, we are unableto determine whether this fibrinogen complex is PL-associ-ated or not. Nevertheless, fibrinogen has previously beenshown to have antioxidant activity (42, 43), which lends sup-port to our findings that fibrinogen is highly correlated withantioxidant activity in peak 1. Furthermore, correlation analy-sis demonstrated that the fibrinogen complex (FibA, FibB andFibG) was also highly correlated to antioxidant activity in peak1 (Fig. 9B, r 0.98). In peak 2, apoA-I and apoA-II are bothhighly correlated with antioxidant function. This effect may bebecause of these proteins specifically, as these proteins haveboth been shown to have antioxidative properties (67–71).Additionally, apoA-I and apoA-II may be the structural pro-teins of particularly effective subspecies of HDL containingother antioxidant proteins. For instance, ceruloplasmin andparaoxonase, both proteins with documented antioxidantproperties (60, 72, 73), were also found to be highly correlatedwith antioxidation activity in peak 2. Other proteins in peak 2that were highly correlated with antioxidation include comple-ment components, as well as other inflammatory and immuneresponse proteins (immunoglobulins, serum amyloid P, in-ter-�-trypsin inhibitor), indicating the potential overlap andinteraction of immune response and antioxidation pro-

cesses. Antioxidative activity in peak 3 is highly correlatedwith albumin and serotransferrin, both proteins known tohave antioxidant properties (62, 74, 75). Other interestingproteins found in peak 3 that had significant correlationswith antioxidant activity include both hemopexin and apoA-IV, both of which have previously demonstrated antioxidantactivity (76, 77).

In our previous study (14), network analysis identified 30potential subspecies that were supported by multiple lines ofevidence. Further analysis of the proteomic compositions ofour plasma fractions yielded support for the existenceof some previously identified potential HDL subspecies. Forinstance, apoA-I, apoC-I and apoJ (clusterin) were foundtogether in peak 2, and were individually found to be sig-nificantly correlated with both cholesterol efflux and antioxi-dation activity (p � 0.05 after Bonferroni multiple correction).We noted that these three proteins appear to comprise anHDL subspecies as indicated in our previous study (14). Theexistence of this HDL subspecies is supported by multiplelines of evidence. First, multiple literature reports have docu-mented an association between pairs of these proteins (41,78–80). Second, recent data from our group has shown thatwhen apoA-I is knocked out in a mouse model, apoC-I wasalso significantly decreased (28). Finally, using co-migrationalanalysis (14), we were able to detect a shift in migrationof apoC-I as well as apoJ in a subject who was deficient inapoA-I, implying that in the absence of apoA-I, apoC-I andapoJ migrate with smaller particles than they would in thepresence of apoA-I. These data, taken together, support thenotion that apoA-I, apoC-I and apoJ may constitute a specificsubspecies of HDL. Similarly, a second previously identifiedputative subspecies was found to be highly correlated withcholesterol efflux activity, as well as antioxidation activity inpeak 2: apoA-I, apoA-II and apoC-I. Indeed, the apolipopro-teins in each of these complexes have demonstrated effluxpotential; it will be interesting to determine whether a lipopro-tein complex or complexes, containing these efflux-capable

FIG. 9. Correlation of previouslyidentified putative subspecies withcholesterol efflux and antioxidantactivities in individual peaks. Pearsoncorrelation coefficients were calculatedfor the identified subspecies as de-scribed in Methods. A, PCC of each sub-species compared with cholesterol ef-flux activity. Black bars, apoA-I, apoA-II,apoC-I; gray bars, apoA-I, apoC-I, apoJ.B, PCC of each subspecies comparedwith antioxidation activity. Black bars,fibrinogen � chain, fibrinogen � chain,and fibrinogen � chain; light gray bars,apoA-I, apoA-II, and apoC-I; dark graybars, apoA-I, apoC-I and apoJ.



proteins, will act synergistically to enhance cholesterol effluxcompared with each protein on its own. Because each ofthese subspecies is highly correlated with both cholesterolefflux and antioxidant activity, and because many of theseexchangeable lipoproteins are capable of performing similarfunctions, it is difficult to ascertain the importance of onesubspecies compared with another in carrying out each spe-cific function. The presence of additional proteins on thesesubspecies may account for differences in functional activi-ties. Regardless, this data supports our hypothesis that dis-tinct HDL subspecies may be responsible for various biolog-ical functions related to CVD.

Although our results suggest multiple proteins/subspeciesare associated with cholesterol efflux and antioxidation func-tions, we caution that both efflux and antioxidation associatedproteins were derived from numerical correlation analysis. It isdifficult to know if the correlated proteins are causative orsimply associations. However, correlations between the pro-teins and the two functions are very strong and these candi-dates are actively being studied through interventional exper-imental approaches in our laboratory. This work is among thefirst to go beyond individual proteins and link putative HDLsubspecies to specific functions. Further work will be neededto confirm the existence of specific subspecies and theirdirect role in a specific HDL function. However, our datasuggest that certain plasma proteins may serve as betterbiomarkers than HDL-C for CVD risk assessment, especiallywhen it comes to precision medicine; specific functionalevaluation (e.g. antioxidation and cholesterol efflux activi-ties) may provide richer information than general diseaserisk assessment.

* This work was supported by National Institutes of Health HeartLung and Blood Institute, R01HL67093 and R01HL104136 to W.S.D.,R01HL111829 to L.J.L., and National Natural Science Foundation ofChina No. 31601083. Mass spectrometry data were collected in theUC Proteomics Laboratory on the 5600 � TripleTOF system funded inpart through an NIH shared instrumentation grant (S10 RR027015-01;KD Greis-PI). The content is solely the responsibility of the authorsand does not necessarily represent the official views of the NationalInstitutes of Health.

□S This article contains supplemental material.** To whom correspondence should be addressed: Division of

Biomedical Informatics, Cincinnati Children’s Hospital ResearchFoundation, 3333 Burnet Avenue, MLC 7024, Cincinnati, OH45229-3039. Tel.: 513-636-8720; Fax: 513-636-205; E-mail: [email protected].

REFERENCES

1. Mahmood, S. S., Levy, D., Vasan, R. S., and Wang, T. J. (2014) TheFramingham Heart Study and the epidemiology of cardiovascular dis-ease: a historical perspective. Lancet 383, 999–1008

2. Gordon, T., Castelli, W. P., Hjortland, M. C., Kannel, W. B., and Dawber,T. R. (1977) High density lipoprotein as a protective factor against cor-onary heart disease. The Framingham Study. Am. J. Med. 62, 707–714

3. Zhang, Y., Zanotti, I., Reilly, M. P., Glick, J. M., Rothblat, G. H., and Rader,D. J. (2003) Overexpression of apolipoprotein A-I promotes reversetransport of cholesterol from macrophages to feces in vivo. Circulation108, 661–663

4. Fielding, C. J., and Fielding, P. E. (1995) Molecular physiology of reversecholesterol transport. J. Lipid Res. 36, 211–228

5. Soran, H., Schofield, J. D., and Durrington, P. N. (2015) Antioxidant prop-erties of HDL. Front. Pharmacol. 6, 222

6. Barter, P. J., and Rye, K. A. (1996) High density lipoproteins and coronaryheart disease. Atherosclerosis 121, 1–12

7. Cockerill, G. W., Huehns, T. Y., Weerasinghe, A., Stocker, C., Lerch, P. G.,Miller, N. E., and Haskard, D. O. (2001) Elevation of plasma high-densitylipoprotein concentration reduces interleukin-1-induced expression ofE-selectin in an in vivo model of acute inflammation. Circulation 103,108–112

8. Cockerill, G. W., Rye, K. A., Gamble, J. R., Vadas, M. A., and Barter, P. J.(1995) High-density lipoproteins inhibit cytokine-induced expression ofendothelial cell adhesion molecules. Arterioscler. Thromb. Vasc. Biol. 15,1987–1994

9. Ashby, D. T., Rye, K. A., Clay, M. A., Vadas, M. A., Gamble, J. R., andBarter, P. J. (1998) Factors influencing the ability of HDL to inhibitexpression of vascular cell adhesion molecule-1 in endothelial cells.Arterioscler. Thromb. Vasc. Biol. 18, 1450–1455

10. Gharavi, N. M., Gargalovic, P. S., Chang, I., Araujo, J. A., Clark, M. J.,Szeto, W. L., Watson, A. D., Lusis, A. J., and Berliner, J. A. (2007)High-density lipoprotein modulates oxidized phospholipid signaling inhuman endothelial cells from proinflammatory to anti-inflammatory. Ar-terioscler. Thromb. Vasc. Biol. 27, 1346–1353

11. Navab, M., Imes, S. S., Hama, S. Y., Hough, G. P., Ross, L. A., Bork, R. W.,Valente, A. J., Berliner, J. A., Drinkwater, D. C., Laks, H. (1991) Monocytetransmigration induced by modification of low density lipoprotein incocultures of human aortic wall cells is due to induction of monocytechemotactic protein 1 synthesis and is abolished by high density lipo-protein. J. Clin. Invest. 88, 2039–2046

12. Gordon, S. M., Deng, J., Lu, L. J., and Davidson, W. S. (2010) Proteomiccharacterization of human plasma high density lipoprotein fractionatedby gel filtration chromatography. J Proteome Res 9, 5239–5249

13. Gordon, S. M., Deng, J., Tomann, A. B., Shah, A. S., Lu, L. J., andDavidson, W. S. (2013) Multi-dimensional co-separation analysis revealsprotein-protein interactions defining plasma lipoprotein subspecies. MolCell. Proteomics 12, 3123–3134

14. Li, H., Gordon, S. M., Zhu, X., Deng, J., Swertfeger, D. K., Davidson, W. S.,and Lu, L. J. (2015) Network-Based Analysis on Orthogonal Separationof Human Plasma Uncovers Distinct High Density Lipoprotein Com-plexes. J. Proteome Res. 14, 3082–3094

15. Davidson, W. S., Silva, R. A., Chantepie, S., Lagor, W. R., Chapman, M. J.,and Kontush, A. (2009) Proteomic analysis of defined HDL subpopula-tions reveals particle-specific protein clusters: relevance to antioxidativefunction. Arterioscler. Thromb. Vasc. Biol. 29, 870–876

16. Vaisar, T., Pennathur, S., Green, P. S., Gharib, S. A., Hoofnagle, A. N.,Cheung, M. C., Byun, J., Vuletic, S., Kassim, S., Singh, P., Chea, H.,Knopp, R. H., Brunzell, J., Geary, R., Chait, A., Zhao, X. Q., Elkon,K., Marcovina, S., Ridker, P., Oram, J. F., and Heinecke, J. W. (2007)Shotgun proteomics implicates protease inhibition and complement ac-tivation in the antiinflammatory properties of HDL. J. Clin. Invest. 117,746–756

17. Asztalos, B. F., and Schaefer, E. J. (2003) High-density lipoprotein sub-populations in pathologic conditions. Am. J. Cardiol. 91, 12e–17e

18. Camont, L., Lhomme, M., Rached, F., Le Goff, W., Negre-Salvayre, A.,Salvayre, R., Calzada, C., Lagarde, M., Chapman, M. J., and Kontush,A. (2013) Small, dense high-density lipoprotein-3 particles are en-riched in negatively charged phospholipids: relevance to cellular cho-lesterol efflux, antioxidative, antithrombotic, anti-inflammatory, andantiapoptotic functionalities. Arterioscler. Thromb. Vasc. Biol. 33,2715–2723

19. Munroe, W. H., Phillips, M. L., and Schumaker, V. N. (2015) Excessivecentrifugal fields damage high density lipoprotein. J. Lipid Res. 56,1172–1181

20. van’t Hooft, F., and Havel, R. J. (1982) Metabolism of apolipoprotein E inplasma high density lipoproteins from normal and cholesterol-fed rats.J. Biol. Chem. 257, 10996–11001

21. Markwell, M. A., Haas, S. M., Bieber, L. L., and Tolbert, N. E. (1978) Amodification of the Lowry procedure to simplify protein determination inmembrane and lipoprotein samples. Anal. Biochem. 87, 206–210



http://www.mcponline.org/cgi/content/full/M116.066290/DC1mailto:[email protected]:[email protected]

22. Heink, A., Davidson, W. S., Swertfeger, D. K., Lu, L. J., and Shah, A. S.(2015) A comparison of methods to enhance protein detection of lipo-proteins by mass spectrometry. J. Proteome Res. 14, 2943–2950

23. Vizcaino, J. A., Csordas, A., del-Toro, N., Dianes, J. A., Griss, J., Lavidas,I., Mayer, G., Perez-Riverol, Y., Reisinger, F., Ternent, T., Xu, Q. W.,Wang, R., and Hermjakob, H. (2016) 2016 update of the PRIDE databaseand its related tools. Nucleic Acids Res. 44, D447–D456

24. Gould, J. (2013) GENE-E. http://www.broadinstitute.org/cancer/software/GENE-E/.

25. Gordon, S. M., Li, H., Zhu, X., Shah, A. S., Lu, L. J., and Davidson, W. S.(2015) A Comparison of the Mouse and Human Lipoproteome: Suitabilityof the Mouse Model for Studies of Human Lipoproteins. J. ProteomeRes. 14, 2686–2695

26. Phillips, M. C. (2013) New insights into the determination of HDL structureby apolipoproteins: Thematic review series: high density lipoproteinstructure, function, and metabolism. J. Lipid Res. 54, 2034–2048

27. Shiflett, A. M., Bishop, J. R., Pahwa, A., and Hajduk, S. L. (2005) Humanhigh density lipoproteins are platforms for the assembly of multi-com-ponent innate immune complexes. J. Biol. Chem. 280, 32578–32585

28. Gordon, S. M., Li, H., Zhu, X., Tso, P., Reardon, C. A., Shah, A. S., Lu, L. J.,and Davidson, W. S. (2016) Impact of genetic deletion of platform apo-lipoproteins on the size distribution of the murine lipoproteome. J. Pro-teomics 146, 184–194

29. Rundel, R. (1991) Peakfit: Non-linear Curve-fitting Software-technicalGuide. Jandel Scientific, San Rafael, CA

30. Davidson, W. S., Lund-Katz, S., Johnson, W. J., Anantharamaiah, G. M.,Palgunachari, M. N., Segrest, J. P., Rothblat, G. H., and Phillips, M. C.(1994) The influence of apolipoprotein structure on the efflux of cellularfree cholesterol to high density lipoprotein. J. Biol. Chem. 269,22975–22982

31. Agnani, G., and Marcel, Y. L. (1993) Cholesterol efflux from fibroblasts todiscoidal lipoproteins with apolipoprotein A-I (LpA-I) increases with par-ticle size but cholesterol transfer from LpA-I to lipoproteins decreaseswith size. Biochemistry 32, 2643–2649

32. Fournier, N., Paul, J. L., Atger, V., Cogny, A., Soni, T., de la Llera-Moya, M.,Rothblat, G., and Moatti, N. (1997) HDL phospholipid content and com-position as a major factor determining cholesterol efflux capacity fromFu5AH cells to human serum. Arterioscler. Thromb. Vasc. Biol. 17,2685–2691

33. Davidson, W. S., Gillotte, K. L., Lund-Katz, S., Johnson, W. J., Rothblat,G. H., and Phillips, M. C. (1995) The effect of high density lipoproteinphospholipid acyl chain composition on the efflux of cellular free cho-lesterol. J. Biol. Chem. 270, 5882–5890

34. Fournier, N., Atger, V., Cogny, A., Vedie, B., Giral, P., Simon, A., Moatti, N.,and Paul, J. L. (2001) Analysis of the relationship between triglyceridemiaand HDL-phospholipid concentrations: consequences on the efflux ca-pacity of serum in the Fu5AH system. Atherosclerosis 157, 315–323

35. Yancey, P. G., Kawashiri, M. A., Moore, R., Glick, J. M., Williams, D. L.,Connelly, M. A., Rader, D. J., and Rothblat, G. H. (2004) In vivo modu-lation of HDL phospholipid has opposing effects on SR-BI- and ABCA1-mediated cholesterol efflux. J. Lipid Res. 45, 337–346

36. Johnson, W. J., Mahlberg, F. H., Rothblat, G. H., and Phillips, M. C. (1991)Cholesterol transport between cells and high-density lipoproteins.Biochim. Biophys. Acta 1085, 273–298

37. Yancey, P. G., Bielicki, J. K., Johnson, W. J., Lund-Katz, S., Palgunachari,M. N., Anantharamaiah, G. M., Segrest, J. P., Phillips, M. C., and Roth-blat, G. H. (1995) Efflux of cellular cholesterol and phospholipid tolipid-free apolipoproteins and class A amphipathic peptides. Biochem-istry 34, 7955–7965

38. Mendez, A. J., Anantharamaiah, G. M., Segrest, J. P., and Oram, J. F.(1994) Synthetic amphipathic helical peptides that mimic apolipoproteinA-I in clearing cellular cholesterol. J. Clin. Invest. 94, 1698–1705

39. Gillotte, K. L., Zaiou, M., Lund-Katz, S., Anantharamaiah, G. M., Holvoet, P.,Dhoest, A., Palgunachari, M. N., Segrest, J. P., Weisgraber, K. H.,Rothblat, G. H., and Phillips, M. C. (1999) Apolipoprotein-mediatedplasma membrane microsolubilization. Role of lipid affinity and mem-brane penetration in the efflux of cellular cholesterol and phospholipid.J. Biol. Chem. 274, 2021–2028

40. Hara, H., and Yokoyama, S. (1991) Interaction of free apolipoproteins withmacrophages. Formation of high density lipoprotein-like lipoproteins andreduction of cellular cholesterol. J. Biol. Chem. 266, 3080–3086

41. Wayner, D. D., Burton, G. W., Ingold, K. U., Barclay, L. R., and Locke, S. J.(1987) The relative contributions of vitamin E, urate, ascorbate andproteins to the total peroxyl radical-trapping antioxidant activity of hu-man blood plasma. Biochim. Biophys. Acta 924, 408–419

42. Olinescu, R. M., and Kummerow, F. A. (2001) Fibrinogen is an efficientantioxidant. J. Nutritional Biochem. 12, 162–169

43. Abudu, N., Miller, J. J., and Levinson, S. S. (2006) Fibrinogen is a co-antioxidant that supplements the vitamin E analog trolox in a modelsystem. Free Radic. Res. 40, 321–331

44. Rifkin, M. R. (1978) Identification of the trypanocidal factor in normalhuman serum: high density lipoprotein. Proc. Natl. Acad. Sci. U.S.A.75, 3450–3454

45. Raper, J., Fung, R., Ghiso, J., Nussenzweig, V., and Tomlinson, S. (1999)Characterization of a novel trypanosome lytic factor from human serum.Infect. Immun. 67, 1910–1916

46. Smith, L. E., Segrest, J. P., and Davidson, W. S. (2013) Helical domains thatmediate lipid solubilization and ABCA1-specific cholesterol efflux in apo-lipoproteins C-I and A-II. J. Lipid Res. 54, 1939–1948

47. Gelissen, I. C., Hochgrebe, T., Wilson, M. R., Easterbrook-Smith, S. B.,Jessup, W., Dean, R. T., and Brown, A. J. (1998) Apolipoprotein J(clusterin) induces cholesterol export from macrophage-foam cells: apotential anti-atherogenic function? Biochem. J. 331, 231–237

48. Remaley, A. T., Schumacher, U. K., Stonik, J. A., Farsi, B. D., Nazih, H., andBrewer, H. B., Jr. (1997) Decreased reverse cholesterol transport fromTangier disease fibroblasts. Acceptor specificity and effect of brefeldinon lipid efflux. Arterioscler. Thromb. Vasc. Biol. 17, 1813–1821

49. Remaley, A. T., Stonik, J. A., Demosky, S. J., Neufeld, E. B., Bocharov,A. V., Vishnyakova, T. G., Eggerman, T. L., Patterson, A. P., Duverger,N. J., Santamarina-Fojo, S., and Brewer, H. B. (2001) Apolipoproteinspecificity for lipid efflux by the human ABCAI transporter. Biochem.Biophys. Res. Commun. 280, 818–823

50. Zimetti, F., Weibel, G. K., Duong, M., and Rothblat, G. H. (2006) Measure-ment of cholesterol bidirectional flux between cells and lipoproteins. J.Lipid Res. 47, 605–613

51. Tanaka, K. A., Key, N. S., and Levy, J. H. (2009) Blood coagulation:hemostasis and thrombin regulation. Anesth. Analg. 108, 1433–1446

52. Oram, J. F., Lawn, R. M., Garvin, M. R., and Wade, D. P. (2000) ABCA1 isthe cAMP-inducible apolipoprotein receptor that mediates cholesterolsecretion from macrophages. J. Biol. Chem. 275, 34508–34511

53. Wang, N., Silver, D. L., Costet, P., and Tall, A. R. (2000) Specific bind-ing of ApoA-I, enhanced cholesterol efflux, and altered plasma mem-brane morphology in cells expressing ABC1. J. Biol. Chem. 275,33053–33058

54. Sankaranarayanan, S., de la Llera-Moya, M., Drazul-Schrader, D., Phillips,M. C., Kellner-Weibel, G., and Rothblat, G. H. (2013) Serum albumin actsas a shuttle to enhance cholesterol efflux from cells. J. Lipid Res. 54,671–676

55. Li, X. M., Tang, W. H., Mosior, M. K., Huang, Y., Wu, Y., Matter, W., Gao,V., Schmitt, D., Didonato, J. A., Fisher, E. A., Smith, J. D., and Hazen,S. L. (2013) Paradoxical association of enhanced cholesterol efflux withincreased incident cardiovascular risks. Arterioscler. Thromb. Vasc. Biol.33, 1696–1705

56. Frei, B., England, L., and Ames, B. N. (1989) Ascorbate is an outstandingantioxidant in human blood plasma. Proc. Natl. Acad. Sci. U.S.A. 86,6377–6381

57. Tribble, D. L., van den Berg, J. J., Motchnik, P. A., Ames, B. N., Lewis,D. M., Chait, A., and Krauss, R. M. (1994) Oxidative susceptibility of lowdensity lipoprotein subfractions is related to their ubiquinol-10 and al-pha-tocopherol content. Proc. Natl. Acad. Sci. U.S.A. 91, 1183–1187

58. Karten, B., Beisiegel, U., Gercken, G., and Kontush, A. (1997) Mechanismsof lipid peroxidation in human blood plasma: a kinetic approach. Chem.Phys. Lipids 88, 83–96

59. Huang, J. M., Huang, Z. X., and Zhu, W. (1998) Mechanism of high-densitylipoprotein subfractions inhibiting copper-catalyzed oxidation of low-density lipoprotein. Clin. Biochem. 31, 537–543

60. Watson, A. D., Berliner, J. A., Hama, S. Y., La Du, B. N., Faull, K. F.,Fogelman, A. M., and Navab, M. (1995) Protective effect of high densitylipoprotein associated paraoxonase. Inhibition of the biological activity ofminimally oxidized low density lipoprotein. J. Clin. Invest. 96, 2882–2891

61. Navab, M., Berliner, J. A., Watson, A. D., Hama, S. Y., Territo, M. C., Lusis,A. J., Shih, D. M., Van Lenten, B. J., Frank, J. S., Demer, L. L., Edwards,



http://www.broadinstitute.org/cancer/software/GENE-E/http://www.broadinstitute.org/cancer/software/GENE-E/

P. A., and Fogelman, A. M. (1996) The Yin and Yang of oxidation in thedevelopment of the fatty streak. A review based on the 1994 GeorgeLyman Duff Memorial Lecture. Arterioscler. Thromb. Vasc. Biol. 16,831–842

62. Halliwell, B. (1988) Albumin–an important extracellular antioxidant?Biochem. Pharmacol. 37, 569–571

63. Halliwell, B. (1999) Antioxidant defense mechanisms: from the beginning tothe end (of the beginning). Free Radic. Res. 31, 261–272

64. Sitar, M. E., Aydin, S., and Cakatay, U. (2013) Human serum albumin and itsrelation with oxidative stress. Clin. Lab. 59, 945–952

65. Bancells, C., Canals, F., Benitez, S., Colome, N., Julve, J., Ordonez-Llanos,J., and Sanchez-Quesada, J. L. (2010) Proteomic analysis of electrone-gative low-density lipoprotein. J. Lipid Res. 51, 3508–3515

66. Dashty, M., Motazacker, M. M., Levels, J., de Vries, M., Mahmoudi, M.,Peppelenbosch, M. P., and Rezaee, F. (2014) Proteome of humanplasma very low-density lipoprotein and low-density lipoprotein exhibitsa link with coagulation and lipid metabolism. Thromb. Haemost. 111,518–530

67. Van Linthout, S., Spillmann, F., Riad, A., Trimpert, C., Lievens, J., Meloni,M., Escher, F., Filenberg, E., Demir, O., Li, J., Shakibaei, M., Schimke,I., Staudt, A., Felix, S. B., Schultheiss, H. P., De Geest, B., andTschope, C. (2008) Human apolipoprotein A-I gene transfer reducesthe development of experimental diabetic cardiomyopathy. Circulation117, 1563–1573

68. Kim, C., Lee, J. M., Park, S. W., Kim, K. S., Lee, M. W., Paik, S., Jang, A. S.,Kim do, Uh, J. S., Kim, Y., and Park, C. S. (2016) Attenuation of CigaretteSmoke-Induced Emphysema in Mice by Apolipoprotein A-1 Overexpres-sion. Am. J. Respir. Cell Mol. Biol. 54, 91–102

69. Navab, M., Hama, S. Y., Anantharamaiah, G. M., Hassan, K., Hough, G. P.,Watson, A. D., Reddy, S. T., Sevanian, A., Fonarow, G. C., and Fogel-man, A. M. (2000) Normal high density lipoprotein inhibits three steps inthe formation of mildly oxidized low density lipoprotein: steps 2 and 3. J.Lipid Res. 41, 1495–1508

70. Garner, B., Waldeck, A. R., Witting, P. K., Rye, K.-A., and Stocker, R. (1998)Oxidation of high density lipoproteins: II. Evidence for direct reduction of

lipid hydroperoxides by methionine residues of apolipoproteins AI andAII. J. Biol. Chem. 273, 6088–6095

71. Graham, A., Hassall, D. G., Rafique, S., and Owen, J. S. (1997) Evidence fora paraoxonase-independent inhibition of low-density lipoprotein oxida-tion by high-density lipoprotein. Atherosclerosis 135, 193–204

72. Ehrenwald, E., Chisolm, G. M., and Fox, P. L. (1994) Intact human cerulo-plasmin oxidatively modifies low density lipoprotein. J. Clin. Invest. 93,1493–1501

73. Gutteridge, J. M. (1978) Caeruloplasmin: a plasma protein, enzyme, andantioxidant. Ann. Clin. Biochem. 15, 293–296

74. Sengupta, S., Wehbe, C., Majors, A. K., Ketterer, M. E., DiBello, P. M., andJacobsen, D. W. (2001) Relative roles of albumin and ceruloplasmin inthe formation of homocystine, homocysteine-cysteine-mixed disulfide,and cystine in circulation. J. Biol. Chem. 276, 46896–46904

75. Spranger, T., Finckh, B., Fingerhut, R., Kohlschütter, A., Beisiegel, U., andKontush, A. (1998) How different constituents of human plasma and lowdensity lipoprotein determine plasma oxidizability by copper. Chem.Phys. Lipids 91, 39–52

76. Qin, X., Swertfeger, D. K., Zheng, S., Hui, D. Y., and Tso, P. (1998) Apoli-poprotein AIV: a potent endogenous inhibitor of lipid oxidation. Am. J.Physiol. 274, H1836–1840

77. Tolosano, E., and Altruda, F. (2002) Hemopexin: structure, function, andregulation. DNA Cell Biol. 21, 297–306

78. Zhou, M., Lucas, D. A., Chan, K. C., Issaq, H. J., EFPetricoin 3rd, Liotta,L. A., Veenstra, T. D., and Conrads, T. P. (2004) An investigation into thehuman serum “interactome”. Electrophoresis 25, 1289–1298

79. Ehnholm, C., Bozas, S. E., Tenkanen, H., Kirszbaum, L., Metso, J., Murphy,B., and Walker, I. D. (1991) The apolipoprotein A-I binding protein ofplacenta and the SP-40,40 protein of human blood are different proteinswhich both bind to apolipoprotein A-I. Biochim. Biophys. Acta 1086,255–260

80. James, R. W., Hochstrasser, A. C., Borghini, I., Martin, B., Pometta, D., andHochstrasser, D. (1991) Characterization of a human high density lipo-protein-associated protein, NA1/NA2. Identity with SP-40,40, an inhibitorof complement-mediated cytolysis. Arterioscler. Thromb. 11, 645–652



Documents

This paper is available on line at ...homepages.uc.edu/~davidswm/Proteome_function.pdf · floating cells. Total cellular efflux was determined by dividing the 3H-cholesterol in the