CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is based on, claims a priority benefit from, and incorporates herein by reference, U.S. Provisional Patent Application No. 61/924,472, filed January 7, 2014, and entitled "Apolipoprotein C Biomarkers and Methods for Cholesterol Efflux and Cardiovascular Health."

TECHNICAL FIELD

[0002] This disclosure relates to methods and materials involving biomarkers for cholesterol metabolism and cardiovascular health.

BACKGROUND OF THE INVENTION

[0003] The human apolipoprotein C family consists of 3 members, apoC-l, apoC-ll and apoC- III, which are distributed in chylomicrons, VLDL and HDL. They are portrayed as members of the same family due to similar molecular weights, lipoprotein distribution and coincident purification. T he apoC lipoproteins are structurally polymorphic.

[0004] Apolipoprotein C-ll, apoC-ll, is a minor component of the lipoproteins VLDL, chylomicrons, LDL and HDL. The protein functions as a cofactor of several lipases such as lipoprotein lipase. In this way, the protein participates in the secretion and catabolism of triglyceride-rich lipoproteins. The protein is expressed in the liver as a 79 amino acid monomer. The protein upon entering circulation is subject to posttranslational modifications, whereby the N-terminal 6 amino acids of the apoC-ll are truncated (Figure 1 ). This truncated form of apoC-ll is termed "apoC-ll active" is readily detected through the use of mass spectrometry (Figure 2).

[0005] Apolipoprotein C- III, apoC-IM, is component of all the lipoproteins present in circulation, particularly VLDL. ApoC-IM functions with apoC-ll to regulate the triglyceride metabolism in circulation, in that, while apoC-ll acts as a cofactor for several lipases such as lipoprotein lipase, apoC-IM is an inhibitor of the same lipases. ApoC-IM is expressed in the liver as a 79 amino acid monomer. Before the protein is transported into the extracellular matrix, an O-linked glycosylation is added to a threonine residue at position 74 in the sequence of apoC- III. The O-linked glycosylation has a base structure of Gal-GalNAc. Additionally one or two sailic acid motifs can be attached to the base structure (Figure 3). As apoC-IM is an inhibitor of lipases associated with the removal of triglyceride, the protein has been correlated to higher triglyceride levels and has been shown to increase in dysfunctional HDL in heart patients. These studies have focused on concentration alone. [0006] However, little is known about what relationship, if any, exists between post- translationally modified apoC-ll or apoC-lll and lipoprotein particle size or concentration. In addition, the relationship between these post-translationally modified proteins and cholesterol efflux efficiency of high density lipoprotein (HDL) particles, as a measure of HDL function, has not been assessed.

SUMMARY OF THE INVENTION

[0007] ApoC-ll has traditionally been associated with the removal of triglyceride rich lipoproteins from circulation. While apoC-ll is a minor component in the proteome of HDL, whose primary function is the removal of free cholesterol from the periphery to the liver for excretion, the concentration levels of apoC-ll has been associated with a redistribution of the subclasses of HDL with increasing apoC-ll concentrations corresponding to increases in small HDL particles and associated with decrease cholesterol efflux.

[0008] Thus, embodiments described herein relate to the link between apoC and lipoprotein size/concentration and HDL-mediated cholesterol efflux. In one embodiment, we determined the relative percent abundance of active (truncated) apoC-ll. Several positive correlations were found that suggest that the measure of active apoC-ll can act as a biomarker associated with lipoprotein size and concentration and well as HDL mediated cholesterol efflux and thus coronary artery disease.

[0009] Apo-C-lll is posttranslationally modified by the addition of O-linked glycosylation. Through mass spectrometry four modifications of apoC-lll can be viewed and monitored: 1 ) apoC-lll lacking glycosylation, 2) apoC-lll-Gal-GalNAc - glycosylated apoC-lll, 3) apoC-lll-Gal- GalNAc-Sialic Acid and 4) apoC-lll-Gal GalNAc-2xSialic Acid. In another embodiment, the link between post-translationally modified apoC-lll and lipoprotein size/concentration and HDL mediated cholesterol efflux is disclosed. The relative percent abundances of apoC-lll were determined in the plasma of each individual.

[0010] Other features and advantages of the invention will be apparent from the following detailed description and figures, and from the claims.

BRIEF DESCRIPTION OF THE FIGURES

[0011] Figure 1 depicts the apolipoprotein C-ll sequence (native and active).

[0012] Figure 2 shows an apolipoprotein C-ll spectrum.

[0013] Figure 3 depicts apolipoprotein C-lll glycosylations.

[0014] Figure 4 depicts an example of spectra showing apo C-lll protein variants. [0015] Figure 5 depicts a correlation between apoC protein variants and lipid particle distribution.

[0016] Figure 6 depicts an example of spectra showing apo C-lll protein variants 6-11 .

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] The embodiments herein involve the characterization and monitoring of

posttranslational modifications of apolipoprotein C-ll and C-lll. Some modifications are shown to be significantly correlated with cholesterol efflux and lipoprotein size, and thereby applied to predict atherosclerosis and cardiovascular disease and to monitor the response to therapy.

[0018] Abbreviations

VLDL Very low density lipoprotein

HDL High density lipoprotein

PTM Post-translationally modified

MSIA Mass spectrometry immunoassay

TC Total cholesterol

TG Triglyceride

HDL-C High density lipoprotein cholesterol

LDL-C Low density lipoprotein cholesterol

NMS Nuclear magnetic resonance spectroscopy

MALDI-TOF Matrix-assisted laser desorption/ionization time-of-f light

RPA Relative percentage abundance

LCAT Lecithin cholesterol acyltransferase

[0019] As genetic variants are rare in the apoC gene, the majority of polymorphisms are contributed by post-translational protein modifications. These post-translationally modified (PTM) forms have previously been identified using gel electrophoresis. The high resolving power of mass spectrometry techniques now allows for the detection of a greater range of polymorphisms.

[0020] The study of these PTM forms provides insights into the link between protein structure and function. The objective of one or more of the following examples was to describe the heterogeneity of apoC in human plasma using mass spectrometric immunoassay (MSIA), and to examine their relationship with lipoprotein particle size and cholesterol efflux capacity. EXAMPLE 1

[0021] Study Participants: 397 subjects of Chinese descent were studied (224 patients diagnosed with type-2 diabetes mellitus [DM] and 173 healthy individuals to act as healthy controls). The study was approved by the IRB of the National University Health System in Singapore. The patient's age, gender, medical history and BMI were recorded upon entry into the study. Plasma samples were collected and HbA1 c, triglycerides, LDL and HDL measured. Furthermore, the plasma and the concentration of chylomicrons, very low density lipoprotein, low density lipoprotein, high density lipoprotein subclasses, based on particle size, were measured using nuclear magnetic spectroscopy.

[0022] Cholesterol efflux capacity was quantified in the cohort described above with the use of a previously validated high thoroughput assay (Khera AV et al, NEJM, 201 1 ). This assay quantifies total efflux mediated by pathways of known relevance in cholesterol efflux from macrophages (i.e., ATP-binding cassette transporter A1 [ABCA1] and G1 [ABCG1 ], scavenger receptor B1 , and aqueous diffusion).

[0023] Briefly, J774 macrophages were plated and radiola belled with 2 Ci of 3H-cholesterol per milliliter. ABCA1 was up-regulated by means of a 6-hour incubation with 0.3 niM 8-(4- chlorophenylthio)-cyclic AMP. Subsequently, efflux mediums containing 2.8% apolipoprotein B- depleted serum were added for 4 hours. All steps were performed in the presence of the acyl- coenzyme Axholesterol acyltransferase inhibitor CP1 13,818 (2 g per milliliter). Liquid scintillation counting was used to quantify the efflux of radioactive cholesterol from the cells. The quantity of radioactive cholesterol incorporated into cellular lipids was calculated by means of isopropanol extraction of control wells not exposed to patient serum. Percent efflux was calculated by the following formula: [(microcuries of 3H-cholesterol in mediums containing 2.8% apolipoprotein B-depleted serum-microcuries of 3H-cholesterol in serum-free

mediums)÷microcuries of 3H-cholesterol in cells extracted before the efflux step]* 100. All assays were performed in duplicate. To correct for interassay variation across plates, a pooled serum control was included on each plate, and values for serum samples from patients were normalized to this pooled value in subsequent analyses.

[0024] Mass spectrometric immunoassay (MSIA): Quantitative MSIA was used to determine the concentration of apoC-ll and its variants. Initially, the affinity pipettes derivatized with antibodies towards apoC-ll and Lysozyme (which served as an internal reference standard - IRS), were mounted on the head of the Multimek 96-channel pipettor. The MSIA extraction method started with an assay buffer rinse (PBS, 0.1 % Tween, 10 aspirations and dispense cycles, 100 μί volumes each). Next, the pipettes were immersed into a microplate containing the analytical samples and 250 aspirations and dispense cycles were performed (100 μί volumes each) allowing for affinity capture of all the proteins from the samples. The pipettes were then rinsed with assay buffer (100 cycles), and twice with water (10 cycles and 20 cycles, 100 μΙ_ aspiration/dispense each).

[0025] For elution of the captured proteins, 6 μΙ_ aliquots of MALDI matrix (15 g/L sinapic acid in aqueous solution containing 33 % (v/v) acetonitrile, and 0.4 % (v/v) trifluoroacetic acid) were aspirated into the affinity pipettes, and after a 10 second delay (to allow for the dissociation of the protein from the capturing antibody), the eluates were dispensed directly onto a 96-well formatted MALDI target. Following drying, linear mass spectra were acquired from each sample spot, using Bruker's Autoflex III smartbeam MALDI-TOF instrument (Bruker, Billerica, MA) operating in positive ion mode, in the mass range from 5 to 30 kDa, with 200 ns delay, 20.00 kV and 18.45 kV ion source voltages and signal suppression up to 500 Da. Approximately 5,000 laser-shot mass spectra were saved for each standard and sample.

[0026] The mass spectra were internally calibrated using protein calibration standard I, and further processed with Flex Analysis 3.0 software (Bruker Daltonics). Peaks representing intact apoC-ll and active apoC-ll (at 8914.4 m/z and 8204.2 m/z, respectively), and the peak from Lysozyme (the I RS), were integrated baseline-to-baseline using Zebra 1 .0 software (Intrinsic Bioprobes Inc), and the peak area values were tabulated in a spreadsheet. To distinguish between noise and low intensity signals, the peak areas were corrected individually with baseline noise bin signals picked from a close proximity to the integrated signals. The corrected apolipoprotein peak areas were then normalized by the IRS peak areas. The standard curve for apoC-l l was generated by plotting the normalized peak areas of each standard protein against their concentrations, and used to determine the concentration of native and active (truncated) apoC-ll. The normalized peak areas for apoC-ll and its PTMs were summed up, the total apolipoprotein C-ll concentration was determined using the corresponding standard curve, and the concentration of the individual PTMs (post-translational modifications) were determined based on their percentage of the total concentrations. The control plasma and control sample mix were used to assess intra- and inter-day variability.

[0027] Data Analysis: All values were given as means ± SD unless otherwise stated. The statistical analyses were carried out using the SPSS software version 17 (SPSS, Chicago, I L). Pearson correlations were found between the relative percent abundance compared to clinical, biochemical measures and cholesterol efflux. Correlations were considered significant for a p<0.050. The concentration and relative percentage abundance of apoC-ll was log- transformed to improve normality. Multivariate analysis using the general linear model was used to adjust for covariates.

[0028] Results: The predominant apoC-ll form is circulation is the native form with a mean RPA of 92.2 ± 2.6%. The Pearson correlations for the relative percent abundance (RPA) of active (truncated) apoC-ll are listed in Table 1 to 3. Of note, there is a statistically significantly positive correlation between the RPA of active (truncated) apoC-ll and cholesterol efflux (Table 1 ). The RPA of active (truncated) apoC-ll was significantly correlated with HDL-C concentration and HDL particle size (Tables 2 and 3). Similarly, the RPA of active (truncated ApoC-ll) was significantly correlated with LDL-C concentration and LDL particle size (Tables 2 and 3). These relationships were observed in both subjects with and without DM. The RPA of active apoC-ll was inversely correlated with VLDL particle size, but this was observed only in subjects in DM (Table 4).

[0029] The majority of research in probing the functionality of HDL has focused on the protein apolipoprotein A-l . The reason for this is that apoA-l, in addition to being the most abundant protein in the HDL, is a co-factor of LCAT, which is the first protein in the reverse cholesterol transport pathway. In contrast, apoC-ll has mostly been associated with VLDL and chylomicrons as a cofactor for lipoprotein lipase. In circulation, the protein is activated via truncation of the N-terminal amino acids. Based on the results found in this cohort, we found that the RPA of active (truncated) apoC-ll can act as a marker of HDL efflux capacity and changes to the distribution of HDL and LDL size without purification of the lipoproteins. In addition, RPA of active apoC-ll may act as a marker for VLDL particle size specifically in individuals with DM.

[0030] While this modification is only a small percentage of the total apoC-ll content, a decrease in the relative percent abundance of this modification of apoC-ll is correlated with decrease HDL efflux capacity. While the assay used in this experiment utilized mass spectrometric immunoassay, ability to detect and/or quantify the truncation of apolipoprotein C- II can be carried out using multiple mass spectrometric platforms and possibly through an ELISA colorimetric assay. The monitoring of truncated apoC-ll levels can be applied to predict atherosclerosis and cardiovascular disease and to monitor the response to lipid-lowering therapies.

EXAMPLE 2

[0031 ] Study Participants: 397 subjects of Chinese descent completed the study (224 patients diagnosed with type-2 diabetes mellitus [DM] and 173 healthy individuals to act as healthy controls. The study was approved by the IRB of the National University Health System in Singapore. The patient's age, gender, medical history and BMI were recorded upon entry into the study. Plasma samples were collected and HbA1 c, triglycerides, LDL and HDL measured. Furthermore, the plasma and the concentration of chylomicrons, very low density lipoprotein, low density lipoprotein, high density lipoprotein subclasses, based on particle size, were measured using nuclear magnetic spectroscopy. [0032] Cholesterol efflux capacity was quantified in the cohort described above with the use of a previously validated high thoroughput assay (Khera AV et al, NEJM, 201 1 ). This assay quantifies total efflux mediated by pathways of known relevance in cholesterol efflux from macrophages (i.e., ATP-binding cassette transporter A1 [ABCA1 ] and G1 [ABCG1 ], scavenger receptor B1 , and aqueous diffusion). Briefly, J774 macrophages were plated and radiolabeled with 2 μθί of 3H-cholesterol per milliliter. ABCA1 was up-regulated by means of a 6-hour incubation with 0.3 mM 8-(4-chlorophenylthio)-cyclic AMP. Subsequently, efflux mediums containing 2.8% apolipoprotein B-depleted serum were added for 4 hours. All steps were performed in the presence of the acyl-coenzyme A holesterol acyltransferase inhibitor CP1 13.818 (2 g per milliliter).

[0033] Liquid scintillation counting was used to quantify the efflux of radioactive cholesterol from the cells. The quantity of radioactive cholesterol incorporated into cellular lipids was calculated by means of isopropanol extraction of control wells not exposed to patient serum. Percent efflux was calculated by the following formula: [(microcuries of 3H-cholesterol in mediums containing 2.8% apolipoprotein B-depleted serum-microcuries of 3H-cholesterol in serum-free mediums)÷microcuries of 3H-cholesterol in cells extracted before the efflux step]x 100. All assays were performed in duplicate. To correct for interassay variation across plates, a pooled serum control was included on each plate, and values for serum samples from patients were normalized to this pooled value in subsequent analyses.

[0034] Mass spectrometric immunoassay (MSIA): Quantitative MSIA was used to determine the concentration of apoC-lll and its variants. I nitially, the affinity pipettes derivatized with corresponding antibodies towards apoC-lll and Lysozyme (the I RS), were mounted on the head of the Multimek 96-channel pipettor. The MSIA extraction method started with an assay buffer rinse (PBS, 0.1 % Tween, 10 aspirations and dispense cycles, 100 μΙ_ volumes each). Next, the pipettes were immersed into a microplate containing the analytical samples and 250 aspirations and dispense cycles were performed (100 μΙ_ volumes each) allowing for affinity capture of all the proteins from the samples. The pipettes were then rinsed with assay buffer (100 cycles), and twice with water (10 cycles and 20 cycles, 100 μΙ_ aspiration/dispense each).

[0035] For elution of the captured proteins, 6 μΙ_ aliquots of MALDI matrix (15 g/L sinapic acid in aqueous solution containing 33 % (v/v) acetonitrile, and 0.4 % (v/v) trifluoroacetic acid) were aspired into the affinity pipettes, and after a 10 second delay (to allow for the dissociation of the protein from the capturing antibody), the eluates were dispensed directly onto a 96-well formatted MALDI target. Following drying, linear mass spectra were acquired from each sample spot, using Bruker's Autoflex III smartbeam MALDI-TOF instrument (Bruker, Billerica, MA) operating in positive ion mode, in the mass range from 5 to 30 kDa, with 200 ns delay, 20.00 kV and 18.45 kV ion source voltages and signal suppression up to 500 Da. Approximately 5,000 laser-shot mass spectra were saved for each standard and sample.

[0036] The mass spectra were internally calibrated using protein calibration standard I, and further processed with Flex Analysis 3.0 software (Bruker Daltonics). Peaks representing intact apoC-lll, apoC-lll with Gal-GalNac (glycosylated), ApoC-lll with Gal-GalNac-Sialic Acid and apoC-lll with Gal-GalNac-2xSialic Acid (at 8764.7 m/z, 91 30.0 m/z, 9421 .2 m/z and 9712.5 m/z, respectively) variants, and the Lysozyme peak (serving as the IRS), were integrated baseline- to-baseline using Zebra 1 .0 software (Intrinsic Bioprobes Inc), and the peak area values were tabulated in a spreadsheet.

[0037] To distinguish between noise and low intensity signals, the peak areas were corrected individually with baseline noise bin signals picked from a close proximity to the integrated signals. The corrected apolipoprotein peak areas were then normalized by the I RS peak areas. Peaks representing intact apoC-lll, apoC-lll with Gal-GalNac, ApoC-lll with Gal-GalNac-Sialic Acid and apoC-lll with Gal-GalNac-2xSialic Acid (8764.7 m/z, 91 30.0 m/z, 9421 .2 m/z and 9712.5 m/z respectively) variants were integrated and tabulated in a spreadsheet for determination of relative percent abundances of the three glycosylated motifs and

unglycosylated apolipoprotein C- III.

[0038] The normalized peak areas for apoC-l ll and its PTMs were summed up, the total apolipoprotein C-lll concentration was determined using the corresponding standard curve, and the concentration of the individual PTMs were determined based on their percentage of the total concentrations. The control plasma and control sample mix were used to assess intra- and inter-day variability.

[0039] Data Analysis: All values were given as means ± SD unless otherwise stated. The statistical analyses were carried out using the SPSS software version 1 7 (SPSS, Chicago, I L). Pearson correlations were found between the relative percent abundance compared to clinical, biochemical measures and cholesterol efflux. Correlations were considered significant for a p<0.050. The concentration and relative percentage abundance of apo C-l ll was log- transformed to improve normality. Multivariate analysis using the general linear model was used to adjust for covariates.

[0040] Results: The majority of circulating apoC-lll is post-translationally modified with the native (unmodified) apoC-lll comprising 6.0 ± 3.1 % of the all circulating apoC-lll . The most common modified forms detected were: 1 ) apoC-lll with Gal-GalNac (glycosylated apoC-lll), 2) apoC-lll with Gal-GalNac-Sialic Acid (mono-sialic acid apoC-lll) and 3) apoC-lll with Gal- GalNac-2xSialic Acid (di-sialic acid apoC-lll). Of these, the mono-sialylated form is the predominant modified form with a mean RPA of 38.0 ± 4.0%.

[0041 ] The Pearson correlations for the relative percent abundance (RPA) of the native and 3 modified forms of apoC-lll are listed in Table 5 to 8. Of note, the RPA and concentration of mono-sialic acid apoC-lll was inversely correlated with cholesterol efflux per HDL particle. Correspondingly, the RPA of di-sialic acid apoC-lll was positively correlated with cholesterol efflux per HDL particle (Table 5). Analyzing in subgroups showed that this relationship was specifically in subjects with DM.

[0042] The RPA and concentration of native apoC-lll was positively correlated with HDL-C levels, and the RPA and concentration of di-sialic acid apoC-lll was inversely correlated with triglyceride and LDL-C levels (Table 6). I n terms of correlations with lipid particle sizes, the RPA and concentration of mono-sialic acid apoC-lll was positively correlated with VLDL particle size in keeping with its function as an inhibitor of lipoprotein lipase. Correspondingly, the RPA of di- sialic acid apo-CIII was inversely correlated with VLDL particle size (Table 7). In addition, the RPA of native apoC-lll was significantly correlated with HDL and LDL particle sizes (Table 7).

[0043] Based on the results found in this cohort, we found that the RPA and concentration of mono-sialic acid apoC-lll can act as a marker of HDL efflux capacity in individuals with DM. In addition, the RPA and concentration of native apoC-lll can act as a marker of HDL and LDL size distribution. The measurements of apoC-l ll can be performed without purification of the lipoproteins. While the assay used in these experiment utilized mass spectrometric immunoassays, ability to detect and/or quantify mono- and disialic acid apolipoprotein C-lll can carried out using multiple mass spectrometric platforms and probably through an ELISA colori metric assay.

[0044] The majority of research in probing the functionality of HDL has focused on the protein apolipoprotein A-l . The reason for this is that apoA-l, in addition to being the most abundant protein in the HDL, is a co-factor of LCAT, which is the first protein in the reverse cholesterol transport pathway. In contrast, apoC-lll has mostly been associated with VLDL and chylomicrons in that the protein is an inhibitor for lipase lipoprotein, which is associated with triglyceride levels in circulation. While apoC-lll concentration has been associated with cardiovascular health, concentration alone cannot fully describe apoC-lll effects on efflux. This study has shown that the posttranslational modifications of apoC-lll can have a dynamic effect on HDL function and lipoprotein particle sizes. The biomarkers discussed in this submission provide insight into dysfunctional HDL, particularly in patients with DM. [0045] The measure of HDL-C in circulation has been used as a surrogate marker for cardiovascular risk. However, a significant portion of the individuals experiencing a cardiovascular incident is found to have HDL-C within the "normal" ranges. It is also significant to note that the recent clinical trials of HDL-cholesterol increasing agents have found no benefit from the increase in HDL in at-risk patients and in some cases the therapeutics have been found to increase the incidents of cardiovascular incidents. Both of these finds outline the need to find new biomarkers for HDL functionality rather than HDL-cholesterol concentration. This disclosure fills that need in that the RPA and concentration of mono- and disialic acid apoC-lll can act as a biomarker for cardiovascular health, especially in individuals with DM.

Table 1. Pearson's correlation between apoC-ll and cholesterol efflux.

*values were log-transformed to improve normality

Table 2. Pearson's correlation between apoC-ll and lipid levels.

*values were log-transformed to improve normality

Table 3. Pearson correlation coefficients between apoC-ll and lipid particle sizes

*values were log-transformed to improve normality

Table 4. Comparison between Pearson correlation coefficients between RPA of active apoC-ll and lipid parameters in subjects with and without DM.

*values were log-transformed to improve normality Table 5. Pearson's correlation between apoC-lll and cholesterol efflux.

*values were log-transformed to improve normality

Table 6. Pearson's correlation between apoC-lll and lipid levels.

Table 7. Pearson correlation coefficients between apoC-lll and lipid particle sizes.

*values were log-transformed to improve normality

Example 3

[0046] Materials and Methods: We recruited 401 subjects of Chinese descent from the community and outpatient clinic of a tertiary hospital, National University Hospital (Singapore). Demographic factors, medical and medication history were collected using a interviewer- administered questionnaire. Height, weight and blood pressure were measured. Venous blood was drawn after a 10-hour overnight fast and analyzed for HbA1 c, total cholesterol (TC), triglycerides (TG), and HDL-C. These biochemical analyses were carried out at the National University Hospital Referral Laboratory, which is accredited by the College of American Pathologists. TC, TG and HDL-C levels were measured using an automated autoanalyzer (ADVIA 2400, Bayer Diagnostics). LDL-C levels were calculated using the Friedewald formula. The study was approved by the Institutional Review Board of National University Hospital and all subjects gave informed consent.

[0047] Mass-Spectrometric Immunoassays (MSIA) Reagents: Polyclonal goat anti-human antibodies to apoC-l (Cat.no. 31A-G1 b), apoC-ll (Cat.no. 32A-G2b), apoC-lll (Cat.no. 33A- G2b), and ultra-pure human apolipoprotein C-l (Cat.no. 31 P-UP201 ) were obtained from Academy Bio-medical Co. (Houston, TX, USA). Native human apolipoprotein CM (MD-26- 0012P), native human apolipoprotein C-lll (Cat.no. MD-26-0013P), and mouse anti-hen egg lysozyme (Cat.no. 128-10094) that was used as an internal reference standard were obtained from RayBiotech (Norcross, GA, USA). Protein calibration standard I (Cat.no. 206355) was purchased from Bruker (Billerica, MA). Phosphate buffered saline (PBS) buffer (Cat.no. 28372), MES buffered saline (Cat.no. 28390) and 1 ,1 ' Carbonyldiimidazole (97%) (CDI, Cat.no. 530-62- 1 ) were obtained from Thermo Scientific (Waltham, MA, USA). Tween20 (Cat.no. P7949), trifluoracetic acid (TFA, Cat.no. 299537-100G), sinapic acid (Cat.no. 85429-5G), sodium chloride (Cat.no. S7653), HEPES (Cat.no. H3375), ethanolamine (ETA; Cat.no. 398136), albumin from bovine serum (BSA; Cat.no. A7906) and lysozyme from chicken egg white (Cat.no. L7651 ) were obtained from Sigma Aldrich (St. Louis, MO, USA). Acetone (Cat.no. 0000017150) was obtained from Avantor Performance Materials (Center Valley, PA, USA). The acetonitrile solution (ACN; Cat.no. A955-4), hydrochloric acid (HCI; Cat.no. A144-212) and N- methylpyrrolidinone (NMP; Cat.no. BP1 172-4) were obtained from Fisher Scientific (Thermo Fisher Scientific, Waltham, MA, USA). Affinity pipettes fitted with porous microcolumns were obtained from Thermo Scientific (Tempe, AZ, Cat.no. 991 CUS01 ).

[0048] Affinity pipettes derivatization: The affinity pipettes activation was executed using a Multimek 96 channel pipettor (Beckman Coulter, Brea, CA). The antibody solution used for derivatization contained anti-apoC-l, anti-apoC-ll, anti-apoC-lll and anti-lys antibodies, prepared in MES buffer (0.32 g anti-apoC-l, 2.25 g anti-apoC-ll, 2.5 g anti-apoC-lll, and 0.40 g anti-Lys, per tip). For the antibody coupling, a total of 750 cycles of aspirations and dispenses of 50 μΙ_ volumes were performed, followed by rinses with ETA (50 cycles, 100 μΙ_ aspiration/dispense volumes) and HBS-N buffer (50 cycles, 100 μΙ_). The antibody-derivatized pipettes were stored at +4°C until used.

[0049] Preparation of standards and analytical samples: A standard mix containing three purified human apolipoproteins (apoC-l:apoC-ll:apoC-lll 1 :1 :1 ), was prepared to a final concentration of 10 mg/L (total apolipoproteins). Six-point curves were generated by serially diluting the standard apolipoproteins mix in PBS buffer containing 3 g/L BSA, to 5 mg/L, 2.5 mg/L, 1 .25 mg/L, 0.625 mg/L, 0.313 mg/L and 0.156 mg/L. Lysozyme (Lys) in concentration of 0.1 mg/mL was used as an internal reference standard (IRS) for quantitation. This IRS concentration was sufficient to saturate the antibody and create a constant signal throughout the analyses. A control sample mix containing 1 mg/mL from apoC-l, apoC-ll, and apo-C-lll was analyzed in triplicates with each run, along with a control plasma sample whose apoC concentrations were previously determined.

[0050] The human plasma samples were diluted 40-fold in PBS with 0.1 % Tween prior to the analysis. The analytical samples were prepared by combining 30 μί of the diluted plasma (or standards) with 30 μί of a 0.1 mg/mL solution of lysozyme and 60 μί PBS with 0.1 % Tween.

[0051] MSIA: The affinity pipettes were mounted on the head of the Multimek 96-channel pipettor. The MSIA extraction method started with an assay buffer rinse (PBS, 0.1 % Tween, 10 aspirations and dispense cycles, 100 μί volumes each). Next, the pipettes were immersed into a microplate containing the analytical samples and 250 aspirations and dispense cycles were performed (100 μΙ_ volumes each) allowing for affinity capture of all the proteins of interest from the samples. The pipettes were then rinsed with assay buffer (100 cycles), and twice with water (10 cycles and 20 cycles, 100 μΙ_ aspiration/dispense each).

[0052] For elution of the captured proteins, 6 μΙ_ aliquots of MALDI matrix (15 g/L sinapic acid in aqueous solution containing 33 % (v/v) acetonitrile, and 0.4 % (v/v) trifluoroacetic acid) were aspired into the affinity pipettes, and after a 10 second delay (to allow for the dissociation of the protein from the capturing antibody), the eluates were dispensed directly onto a 96-well formatted MALDI target. Following drying, linear mass spectra were acquired from each sample spot, using Bruker's Autoflex III smartbeam MALDI-TOF instrument (Bruker, Billerica, MA) operating in positive ion mode, in the mass range from 5 to 30 kDa, with 200 ns delay, 20.00 kV and 18.45 kV ion source voltages and signal suppression up to 500 Da. Approximately 5,000 laser-shot mass spectra were saved for each standard and sample.

[0053] The mass spectra were internally calibrated using protein calibration standard I, and further processed with Flex Analysis 3.0 software (Bruker Daltonics). All peaks representing apolipoproteins and their post-translational modifications, along with the IRS, were integrated baseline-to-baseline using Zebra 1 .0 software (Intrinsic Bioprobes I nc), and the peak area values were tabulated in a spreadsheet. To distinguish between noise and low intensity signals, the peak areas were corrected individually with baseline noise bin signals picked from a close proximity to the integrated signals. The corrected apolipoprotein peak areas were then normalized by the IRS peak areas. Three standard curves, for apoC-l, apoC-ll and apoC-lll, were generated by plotting the normalized peak areas of each standard protein against their concentrations, and used to determine the concentration of each apoC protein and the corresponding protein post-translational modifications.

[0054] The normalized peak areas for each apolipoprotein and its protein variants were summed up, the total apoC-l, C-ll , and C-lll concentration was determined using the corresponding standard curve, and the concentration of the individual protein variants were determined based on their percentage of the total concentrations. The control plasma and control sample mix were used to assess intra- and inter-day variability.

[0055] Cholesterol efflux assay: Cholesterol efflux capacity was quantified using a modified version of a previously validated high thoroughput assay. This assay quantifies total efflux mediated by pathways of known relevance in cholesterol efflux from macrophages (i.e., ATP- binding cassette transporter A1 [ABCA1 ] and G1 [ABCG1 ], scavenger receptor B1 , and aqueous diffusion). Briefly, J774 macrophages were plated and radiolabeled with 1 Ci of 3H- cholesterol per milliliter. ABCA1 was up-regulated by means of a 4-hour incubation with 0.3 mM 8-(4-chlorophenylthio)-cyclic AMP.

[0056] Subsequently, efflux to medium containing the equivalent of 1 .0 % apolipoprotein B- depleted serum was followed for 3 hours. Liquid scintillation counting was used to quantify the efflux of radioactive cholesterol from the cells. The quantity of radioactive cholesterol in the medium as well as in cellular lipids (after isopropanol extraction) was determined. Percent efflux was calculated by the following formula: [counts per minute (cpm) of 3H-cholesterol in medium containing 1 .0% apolipoprotein B-depleted serum - cpm of 3H-cholesterol in serum- free medium]÷ [cpm of 3H-cholesterol in medium + cpm 3H-cholesterol in cells ] ^χ 100. All assays were performed in duplicate. To correct for interassay variation across plates, a pooled serum control from 10 healthy volunteers was included on each plate, and values for serum samples from patients were normalized to this pooled value in subsequent analyses.

Cholesterol efflux per particle = normalized efflux / HDL particle number as determined by NMR.

[0057] Nuclear magnetic spectroscopy (NMS): The concentration of chylomicrons, VLDL, LDL, HDL subclasses, based on particle size, were measured using NMS on plasma samples (Liposcience, Raleigh, NC). The concentration of each subclass was derived from the measured amplitudes of the distinct lipid methyl group NMR signals they emit. Weighted average particle sizes in nanometers (nm) were calculated from the subclass levels, and diameters assigned to each subclass. The fraction of large LDL particle was derived by dividing the number of large LDL particles by the total number of LDL particles. This calculation was repeated for the other LDL, HDL and VLDL subclasses.

[0058] Statistical analysis: All values were given as means ± SD unless otherwise stated. Analyses were carried out using SPSS software version 17 (SPSS, Chicago, IL). For samples where the protein variants were not detectable, the relative percent abundance (RPA) values were replaced with half the minimum value of the concentration of the variant of interest. To characterize the associations between RPA and clinical measures, i.e., lipid particle size distribution and cholesterol efflux, we performed a multiple linear regression on RPA and clinical measures adjusting for the following confounders: diabetes mellitus (DM), gender, age, alcohol status, smoking status, body-mass-index (BMI), creatinine and lipid medication status.

[0059] We reported the standardized coefficients and their respective 95% confidence intervals. A likelihood ratio test was carried out to determine the significance of the standardized coefficients. To account for multiple testing, a Bonferrioni correction was used for each variant with 28 outcomes (i.e. we multiple the p-value by 28 and take the minimum between this multiplied value and 1 ). The concentration of apoC-l, C-ll and C-lll were log- transformed to improve normality. To characterize the association visually across various apoC variants and clinical measures, heatmaps were plotted with the hierarchical clustering approach to cluster those with similarly patterns.

[0060] Results: A total of 397 subjects completed the study, of which the data from 337 subjects were analyzed. Sixty subjects were excluded from the analysis due to incomplete demographic, medical history or laboratory data. The baseline characteristics of the population is shown is Table 8. The mean age of the population was 54.8 years (± 1 1 .5), 51 % were female and 57% had a history of diabetes mellitus (DM).

[0061] ApoC-l, C-ll and C-lll heterogeneity: ApoC-l

[0062] The mean apoC-l concentration measured by MSIA was 71 .2 ± 46.9 mg/L. A full- length native apo C-l (57 amino acid), and an A/-terminal truncated protein (des-TP apoC-l) were detected (Figure 4). The full-length native protein was the predominant form, with a mean RPA of 73.9 ± 7.4%. Dipeptidyl-peptidase IV (DPP-IV) inhibitor use was associated with a lower des-TP apoC-l RPA (14.1 % vs 27.3%; p< 0.0050). Four percent of the of the study population was noted to have very low levels of des-TP apoC-l (< 10%) [n=15]. All 15 subjects had a history of DM, and among these, 13 were on DPP-IV inhibitors. Of the remaining 2 subjects, one was on a DPP-IV inhibitor 5 months prior to the study visit and the other was not on a DPP- IV inhibitor.

[0063] ApoC-ll. The mean apoC-ll concentration was 56.2 ± 30.2 mg/L. A full-length native pro-ApoC-ll and a PTM form where the A/-terminal 6 amino acids are truncated (des-TQQPQQ apoC-ll) were detected on MSIA (Figure 4). The predominant form in circulation was native apo C-ll with a mean RPA of 92.2 ± 2.6%.

[0064] ApoC-lll. The mean apoC-lll concentration in the study population was 90.3 ± 53.3 mg/L. Majority of circulating apo C-lll was post-translationally modified (RPA of full-length native apo-CIII was 6.5 ± 3.1 %). Eleven PTM apo C-lll were detected on MSIA (Figure 4). The most common forms were based upon the basic structure of O-linked glycosylation at the threonine residue at position 74 (Hex)(HexNAc), followed by the addition of 1 to 2 sialic acid motifs to this base structure (NauAc).

[0065] The most common forms were apoC-lll ₀ [apo C-lll + (Hex^HexNAc) _! ] , apoC-lll _! [apoC-lll+ (Hex^HexNAc^NauAc) _! ] and apoC-lll ₂ [ApoC3 + (HexMHexNAc^NauAc^j. Of these, apoC-ll , the mono-sialylated form was the predominant PTM form. In addition, C- terminal truncation of apoC-lll _! and apoC-lll ₂ were also detected. The respective RPAs of the protein variants are shown in Table 2. ApoC-lll variants 7 to 12 were rare, and found in only 12 to 24% of subjects in this study.

[0066] Correlations with lipid particle distribution. The direction and strength of correlation between the RPA of each apoC variant and lipid particle concentration are represented in Figure 5 as a heat-map. An increased des-TQQPQQ apoC-ll RPA was associated with a higher fraction of large LDL, large HDL, and small VLDL particles in circulation. Specifically, the RPA of des-TQQPQQ apoC-ll was positively correlated with large LDL, large HDL and small VLDL particle concentrations, and inversely correlated with small LDL particle concentration. The above correlations were not present when absolute des-TQQPQQ apoC-ll concentrations were examined.

[0067] Significant correlations with lipid particle size were also observed for specific sialylated and fucosylated ApoC-ll l variants. Among these, the apoC-lll variant 6 apoC-lll variant 6

[ApoC-lll + (Hex) ₂ (HexNAc) ₂ (Fuc) ₃ ] had the strongest correlations. An increased apoC-lll variant 6 RPA was associated with a higher fraction of large LDL, large HDL and a lower proportion of large VLDL particles in circulation. Specifically, the RPA of apoC-lll variant 6 was positively correlated with large LDL and large HDL particle concentrations, and inversely correlated with small LDL, small HDL and large VLDL particle concentrations. The absolute concentrations of apoC-lll variant 6 were also significantly associated with the above lipid parameters.

[0068] ApoC-lll ₂ was clustered with apoC-lll variant 6 on the heat-map and showed similarities in their correlations with lipid particle sizes. Like apoC-ll l Variant 6, an increased ApoC-lll ₂ RPA was associated with a higher proportion of large LDL, large HDL and a lower proportion of large VLDL particles in circulation. Specifically, the RPA of ApoC-lll ₂ was inversely correlated with small LDL and large VLDL particle concentrations.

[0069] In contrast, a higher RPA of apoC-l l was associated with a lower proportion of large LDL and large HDL particles in circulation. Specifically, the RPA of apoC-l l was positively correlated with small LDL and small HDL particle concentrations, and inversely correlated with large HDL particle concentration.

[0070] Correlation with efflux capacity. The RPA of des-TQQPQQ apoC-ll was positively correlated with cholesterol efflux capacity (p<0.001 ) [Table 10]. This was observed in both DM and non-DM subjects (see appendix Table 3). The standardized coefficient decreased after adjusting for number of HDL particles, but remained statistically significant (Table 1 1 ). [0071] The RPA of apoC-ll (monosialic acid apoC-lll) was inversely correlated with cholesterol efflux capacity per HDL particle, while the RPA of apoC-lll variant 6 [ApoC-lll + (Hex) ₂ (HexNAc) ₂ (Fuc) ₃ ] was positively correlated with cholesterol efflux capacity per HDL particle (Table 1 1 ). This trend was observed in both DM and non-DM subjects but was not statistically significant after correcting for multiple testing. The correlation of these two variants and cholesterol efflux capacity (without adjustment for HDL particle number) was not statistically significant. No significant association was observed between apoC-l or its truncated form and cholesterol efflux capacity.

[0072] Three native (full-length) proteins and 12 PTM apoC protein variants were characterized on MSIA. ApoC-lll exhibited the greatest degree of detectable heterogeneity, with protein variants arising from sialylation, fucosylation and truncation. As these variants are the result of post-translational modifications, examining the relative concentrations of variants within an individual would be more relevant to the associated biological processes, rather than absolute protein concentrations.

[0073] ApoC-l. Native apoC-l is a 57-amino acid protein. In vitro studies and transgenic mice studies have suggested that apoC-l has a role in inhibition of VLDL uptake via the LDL-receptor and LDL-related receptor (LRP). It has also been shown to be an in-vitro activator of LCAT. The N-terminal truncated apoC-l protein variant was first identified by in four human subjects. The truncation involved the removal of 2 neutral N-terminal amino acid, and was postulated to be performed by the dipeptidyl-peptidase-4 (DPP-4) enzyme.

[0074] The N-terminal truncated form forms approximately 30% of total circulating apoC-l. However, we noted very low truncated apoC-l levels in a small subset of DM patients who were on dipeptidylpeptidase-4 (DPP-4) inhibitor therapy, consistent with the hypothesis that apoC-l is a substrate of the DPP-4 enzyme. There are limited studies addressing the effect of DPP-4 inhibitors on lipid levels, and these have suggested a potential benefit in terms of reduction in total cholesterol. In this study, we did not observe any significant correlations between native or truncated apoC-l and lipid particle distribution or cholesterol efflux. This suggests that DPP4- enzyme inhibition is likely to have neutral effect on these parameters.

[0075] ApoC-ll. ApoC-ll exists as 2 forms: a pro-apo C-ll protein, and a mature 79-amino acid protein that is activated by truncation of the A/-terminal hexapeptide in circulation (des- TQQPQQ apoC-ll). It is predominantly distributed in VLDL and chylomicrons. The most well- established function of apoC-ll is as a cofactor of lipoprotein lipase. This has been

demonstrated in both vitro studies and in patients with apoC-ll genetic defects. In this study, individuals with higher proportions of des-TQQPQQ apoC-ll and reduced native apoC-ll were noted to have a shift towards smaller VLDL particles, and larger LDL and HDL particles. The shift towards smaller VLDL particles is in keeping with the known biological function of apo C-ll as an activator of lipoprotein lipase.

[0076] The RPA of des-TQQPQQ apoC-ll was also positively correlated with cholesterol efflux capacity, even though this truncation modification formed only a small percentage of the total apoC-ll in circulation. This correlation with cholesterol efflux capacity was seen despite an associated shift to larger HDL particles, which are known to demonstrate a lower efficiency at cholesterol efflux. The underlying biological reason for this association with cholesterol efflux is uncertain as apoC-ll has not been consistently shown to have an effect on LCAT or CETP activity. This correlation weakened substantially after correction with HDL particle number, suggesting that the effect on cholesterol efflux may be mediated in part by increase in HDL particle numbers, and in part by changes in HDL function.

[0077] ApoC-lll. ApoC-ll l is the most abundant apoC lipoprotein. It also exhibits the greatest degree of heterogeneity. Native apo C-l ll is a 79-amino acid protein distributed in VLDL, chylomicrons and HDL. Three main isoforms of apo C-lll have previously been identified on isoelectric focusing. These are the apoC-lll ₀ a (native form), apoC-lll! (mono-sialyated apoC-lll), and apoC-ll l ₂ (di-sialyated apoC-lll). In addition, minor c-terminal truncated forms due to carboxypeptidase cleavage have also previously been identified on mass spectrometry. We identified seven additional PTM variants: apoC-lll ₀ a (glycosylated apoC-lll) and 6 fucosylated forms.

[0078] Plasma apoC-lll inhibits lipoprotein lipase and apoB-mediated binding of lipoproteins to LDLR, thus increases plasma triglyceride levels. Intracellular apoC-l ll is also known to increase VLDL synthesis and secretion. ApoC-lll levels have been shown to correlate positively with plasma triglyceride levels in human studies. Loss of function mutations in ApoC-lll have recently been shown to be associated with low plasma triglyceride, and importantly, a reduced risk of ischemic cardiovascular disease.

[0079] The RPA of ApoC-lll ₂ (disialylated variant) and apoC-lll variant 6 (fucosylated) was associated with a smaller proportion of large VLDL particles in circulation. In view of the known function of apoC-lll as an inhibitor of lipoprotein lipase, suggesting that disialylation and fucosylation is associated with loss of function of apoC-lll . The RPA of apoC-lll variant 6 was also noted to be positively correlated with cholesterol efflux capacity per HDL particle, but not cholesterol efflux capacity as a whole. This was despite an associated shift towards larger HDL particles that was observed when this variant was more abundant. This suggests that fucosylation of apoC-lll may be associated with qualitative changes at the level of the HDL particle, making it more efficient at cholesterol efflux despite its larger size. This may explain, at least in part, the recently described observation between loss of function mutations in apoC-lll and protection against coronary artery disease.

[0080] This study demonstrates that the apoC lipoproteins undergo significant post- translational modification and exist as a heterogeneous population in circulation. The results also suggest that measuring specific protein variants of apo C-ll and C-lll may serve as useful markers of HDL function. In addition, they highlight nuances to potential markers of HDL function: measurements of apoC-ll truncation may provide a global indicator of HDL function, while measurements of apoC-lll sialylation and fucosylation may provide information on qualitative changes at the level of the HDL particle. These may ultimately serve as useful biomarkers in the monitoring of response to lipid lowering therapy and cardiovascular risk.

Table 8. Baseline characteristics of study population

* Comparison between DM vs Non-DM. If the variable is continuous then a t-test is performed. If it is categorical, a chi-squared test is performed.

** Statin, Niacin, Ezetimibe, Fenofibrate or Omega 3

Table 9. Apo C-lll protein variants (n=337)

Table 10. Correlation between apoC protein variants and cholesterol efflux capacity

* with adjustment for multiple testing (Bonferroni correction) Table 11. Correlation between apoC protein variants and cholesterol efflux capacity per

HDL particle

* with adjustment for multiple testing (Bonferroni correction)

[0081] It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and modifications are within the scope of the following claims.