Patient selection with widely varied HDL-C levels (63 control patients and 76 patients with CAD confirmed by coronary angiography) without lipid-lowering therapy, anthropometric data, and laboratory tests was described previously [20]. Only male patients were included in control and CAD groups. The crude HDL preparations (pHDL) were prepared by the precipitation of apoB-containing lipoproteins with PEG 7000–9000 [6]. The concentrations of choline-containing phospholipids and total cholesterol in HDL preparations were measured by enzyme methods using Sentinel CH. SpA (lot 90398; ref: 17320, Milano, Italy) and HUMAN GmbH (lot 23005, ref 10028, Wiesbaden, Germany) kits, respectively, and apoA-I was measured by immunonephelometry with AU 480 (Beckman Coulter Inc, Brea, CA, USA).

ApoA-I content in pre$\beta$- and $\alpha$-fractions of pHDL was measured by agarose gel electrophoresis followed by immunodetection of apoA-I [6]. PeakFit version 4.12 (SeaSolve Software Inc. Framingham, MA, USA) and ImageJ version 1.54h (https://wsr.imagej.net/ij/index.html) software were used to determine band intensities with baseline subtraction at the conditions of linear response of each band intensity to the amount of loaded sample.

Urea-induced pHDL denaturation after incubation for 6 h at 25 °C was followed by the accumulation of lipid-free pre$\beta$-apoA-I detected in the immunoreplica of agarose gel. ApoA-I dissociation was analyzed with the fractions of pre$\beta$-signal detected after the treatment with 0 M, 4.25 M, and 7.5 M urea [6]. The degree of apolipoprotein dissociation at 4.25 M urea as a mid-transition zone that is accomplished at 7.5 M urea was characterized by dissociation parameter D (Eqn. 1):

(1)$$\mathrm{D}=\left(\mathrm{pre}{\beta }_{4.25}^{\prime }-\mathrm{pre}{\beta }_0^{\prime }\right)/\left(\mathrm{pre}{\beta }_{7.5}^{\prime }-\mathrm{pre}{\beta }_0^{\prime }\right)$$

where $pre{\beta }_{4.25}^{{}^{\prime }}$, $pre{\beta }_0^{{}^{\prime }}$, and $pre{\beta }_{7.5}^{{}^{\prime }}$ are pre$\beta$ fractions with 4.25 M, 0 M, and 7.5 M urea, respectively. The dependence of urea-induced apoA-I release on phospholipid level in pHDL preparations was characterized by the partition coefficient K between aqueous and lipid phases adjusted for water molarity (Eqn. 2):

(2)$$K=(F/B)[\mathrm{pHDL}-\mathrm{PL}]$$

where F and B are the numbers of molecules of lipid-free apoA-I in water and bound to pHDL phospholipids, respectively [21]. The F/B ratio was calculated from the D parameter measured at 4.25 M urea (Eqn. 3):

(3)$$F/B=D/(1-D)$$

Cholesterol efflux was measured with the murine macrophage cell line RAW 264.7, which is a standard cell culture for the measurement of ABCA1-mediated cholesterol efflux [15, 22]. The cell line was obtained from the European Collection of Cell Cultures and validated by speciation by DNA-bar sequencing of the COX subunit 1 gene and tested negative for mycoplasma. Efflux from cells to pHDL as a cholesterol acceptor was measured with a fluorescent probe, BODIPY-cholesterol. The basal and cAMP-stimulated effluxes were measured, and ABCA1-mediated efflux was calculated as the difference between stimulated and basal effluxes and expressed as a percent of effluxed cholesterol [6].

To separate the contributions of both pre$\beta$-HDL and mature $\alpha$-HDL in ABCA1-mediated efflux, a simple kinetic theorem for a two-substrate reaction, when substrates react with a single enzyme without the formation of any ternary complex [23], was applied for a total value of ABCA1-mediated efflux with pHDL as a cholesterol acceptor. The total initial rate of reaction v_t⁢o⁢t when pre$\beta$-HDL and $\alpha$-HDL act as competitive inhibitors of each other for the binding to the ABCA1 transporter is described by Eqn. 4:

(4)$$v_{\text{tot}}=\frac{\frac{V_1}{K_1}\left[S_1\right]+\frac{V_2}{K_2}\left[S_2\right]}{1+\frac{\left[S_1\right]}{K_1}+\frac{\left[S_2\right]}{K_2}}$$

where V_i, K_i, and [S_i] are the maximal velocities, Michaelis constants and substrate concentrations for pre$\beta$-HDL and $\alpha$-HDL as first and second substrates, respectively, with the relation V₁ $>$ V₂. Both [S₁] and [S₂] are apoA-I concentrations in the pre$\beta$-band and $\alpha$-band. The treatment of Eqn. 4 in the double reciprocal plots for a first substrate at a constant [S₂]/[S₁] ratio R results in Eqn. 5:

(5)$$\frac{1}{v_{tot}}=\frac{\frac{1}{\left[S_1\right]}+\frac{1}{K_1}+\frac{R}{K_2}}{\frac{V_1}{K_1}+\frac{R{V}_2}{K_2}}$$

For a series of lines in double reciprocal plots for a first substrate with various R values, the common intersection point does not depend on R, and abscissa and ordinate values are determined by Eqns. 6,7, respectively:

(6)$$\frac{1}{\left[S_1\right]}=\frac{V_1-V_2}{K_1{V}_2}$$

(7)$$\frac{1}{V}=\frac{1}{V_2}$$

A common maximal velocity V_m at the constant R is determined by Eqn. 8:

(8)$$V_m=\frac{V_1{K}_2+R{V}_2{K}_1}{K_2+R{K}_1}$$

In turn, the K₂ value is derived from Eqn. 8 (Eqn. 9):

(9)$$K_2=\frac{R{K}_1(V_m-V_2)}{V_1-V_m}$$

V_m varies within low V₂ and high V₁ borders. Generally, V_m non-linearly depends on R. The linear plots according to Eqn. 5 were constructed for fully denatured and intact pHDL sets, and the data were subsequently treated with Eqns. 6,7,8,9 to derive kinetic parameters. The V₁ and K₁ values were measured without S₂ interference with pHDL fully denatured with 7.5 M urea. Before mixing, predenatured HDL was diluted 1:100 to lower urea concentration. The residual urea concentration not exceeding 75 mM did not influence efflux.

The expression of 65 genes involved in HDL metabolism and atherogenesis measured by us earlier by real-time PCR in peripheral blood mononuclear cells [20, 24] was compared to apoA-I dissociation from pHDL in the present study. Housekeeping genes included Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), Lactate dehydrogenase A (LDHA), and Ribosomal protein L3 (RPL3) genes. The relative expression of all target genes was normalized by the geometric mean of three reference genes.

Statistica version 13 (TIBCO Software Inc. (http://tibco.com)) software was used for statistical analysis. The associations between variables were analyzed by the Spearman rank order correlation coefficient. The statistical significance limit was accepted as p 0.05. Nonlinear curve fitting was done with OriginPro 9.0.0 SR1 (OriginLab Corporation, MA, USA). The significance of the difference between different dataset fits was checked with the F-statistic.

The dissociation of apoA-I from pHDL induced by urea treatment and measured by the D parameter significantly decreased with the increase of absolute and apoA-I-normalized cholesterol and phospholipid (PL) levels in pHDL from CAD patients (Table 1). The fit of dissociation data, passed through zero, to the linear model of apoA-I distribution between lipid-free (F) and lipid-bound (B) states (Eqn. 2) results in the values of the partition coefficient between aqueous and lipid phases adjusted for water molarity K = 0.239 $\pm$ 0.012 (R² = 0.874) for control pHDL and K = 0.368 $\pm$ 0.020 (R² = 0.822) for pHDL from CAD patients (Fig. 1). Thus, a 1.54-fold prevalence of apoA-I distribution between water and lipid phases exists in CAD relative to controls (p = 0.000) with the value of the difference in the standard free energy $\mathrm{\Delta }$$\mathrm{\Delta }$G⁰ = $\mathrm{\Delta }{G}_{CAD}^0$ – $\mathrm{\Delta }{G}_{control}^0$ = –RTln(K(CAD)/K(control)) = –1.2 kJ/mol; the difference value is positive for apolipoprotein distribution between lipid and water phases.

Fig. 1.

ApoA-I distribution between water and lipid phases of pHDL predenatured by 4.25 M urea. (A) pHDL from control patients (n = 59). (B) pHDL from CAD patients (n = 73). The 95% prediction bands are given as thin dashed lines. The Y-intercepts passed through zero according to the distribution model. Slopes were 0.239 $\pm$ 0.012 (R² = 0.874) for control and 0.368 $\pm$ 0.020 (R² = 0.822) for CAD. Slope differences were significant with p = 0.000.

The similar values for free energy change have been obtained for the exchange of apoA-I variants between discoidal HDL-bound and lipid-free pools (1–3 kJ/mol) [25] and the lipid transfer for nanodisks consisting of apoA-I and DMPC (1.7 kJ/mol) [26]. ApoA-I in HDL from CAD patients possesses higher free energy and, thus, exhibits a higher dissociation. Of note, the increased PL:apoA-I and Chol:apoA-I ratios for pHDL from CAD patients relative to controls were measured throughout the whole range of HDL-C concentration (data not shown). To reveal the associations between apoA-I dissociation and lipoprotein metabolism, the expression of 65 genes in peripheral blood mononuclear cells, sensitive to HDL metabolism and atherogenesis [20, 24], was analyzed. The significant correlations between D and gene expression are included in Table 1. For control patients, the positive correlations of the D parameter with the ABCA1 and TLR8 expression levels were revealed. For patients with CAD, the positive (for PRKACB, HMGCR, PLTP, CD36, SLPI, and OLR1) and inverse (for CUBN, ALB, and IL1R1) associations of gene expression in peripheral blood with the degree of apoA-I dissociation were revealed.

First, the efflux measurements were done with intact pHDL as cholesterol acceptors. Both for control and CAD patients, there were significant inverse correlations between apoA-I dissociation and ABCA1-mediated cholesterol efflux (Table 1).

Second, due to the involvement of lipid-free and lipid-bound apoA-I in cholesterol transport by the ABCA1 transporter, the decrease of efflux efficiency in CAD may be associated with the less efficient efflux to each or both acceptors. To separate individual contributions, the functionality of lipid-free apoA-I was studied (Fig. 2). To induce apoA-I dissociation from the HDL lipid phase to the water phase, pHDL from control and CAD patients were denatured for 6 h at 25 °C by 4.25 M urea at the middle of the denaturation transition. The ABCA1-mediated efflux for 2 h to pHDL, at fixed or serial dilutions, was then measured, and the data were fitted to Michaelis-Menten kinetics. The kinetic parameters are given in Table 2. The concentration of lipid-free pre$\beta$-apoA-I dissociated from $\alpha$-HDL was calculated as a product of dissociation parameter D and apoA-I level in the $\alpha$-band estimated by agarose gel electrophoresis and apolipoprotein immunodetection (D$\mathrm{\times }$$\alpha$-apoA-I). The dependencies were well fitted to the equation for one substrate reaction and did not differ by the F-test for control and CAD. The V_m, K_m, and V_m/K_m kinetic parameters did not differ also for pHDL from control and CAD patients by Student’s t-test. Of note, the kinetic parameters, checked by the F-statistic, did not change significantly with pHDL predenatured with 7.5 M urea (Table 2) at the completion of the denaturation transition of pHDL from control or CAD patients, which assumes no contribution of HDL remaining at the transition midpoint to the cholesterol efflux. Thus, the lipid-free apoA-I seems to be equally efficient as an acceptor of cholesterol effluxed by ABCA1 in CAD relative to control samples. We used this approach to separate the contributions of lipid-free and HDL-bound apoA-I as first and second substrates, respectively, to the kinetics of ABCA1-mediated efflux based on the simple kinetic theorem of two substrates competing for one enzyme (Eqn. 4). The initial rates of efflux to intact and fully denatured pHDL from CAD patients with low HDL-C were treated as a function of pre$\beta$-apoA-I concentration [S₁] in double reciprocal plots (Fig. 3). The major prerequisite for a plot linearity is a constant [S₂]/[S₁] ratio R (Eqn. 5). The efflux was measured separately with two intact pHDLs with similar R values and five denatured pHDLs, and the data were concatenated to increase the statistical power. The measurements with denatured pHDL were necessary to derive V₁ and K₁ values without S₂ interference. The V₂ and K₂ values for intact pHDL with an R value 13.0 were derived with Eqns. 6,7,8,9. The complete set of mean values of kinetic parameters included V₁ = 17.5 %/2 h, K₁ = 0.62 µg/mL, V₂ = 5.3 %/2 h, and K₂ = 2.13 µg/mL. Thus, efflux efficiency, expressed as the V_m/K_m ratio, for lipid-free apoA-I was 11.4-fold higher than that for HDL-bound apoA-I in HDL from CAD patients with low HDL-C levels.

Fig. 2.

The kinetics of ABCA1-mediated efflux to pHDL predenatured by 4.25 M urea. (A) pHDL from control patients (n = 29). (B) pHDL from CAD patients (n = 31). The 95% prediction bands are given as thin dashed lines.

Fig. 3.

The dependency of the total initial rates of ABCA1-mediated cholesterol efflux v_tot on S₁ concentration in double reciprocal plots for intact and denatured pHDL from the CAD group with low plasma HDL-C levels. The data were fitted to a linear function 1/v_t⁢o⁢t = intercept + slope $\mathrm{\times }$ (1/[S₁]). The efflux was measured with five individual pHDL pre-denatured with 7.5 M urea (filled symbols) and treated with concatenated data: the intercept was 0.0573 $\pm$ 0.003 (p = 0.000), the slope was 0.035 $\pm$ 0.002 (p = 0.000), and R² = 0.94. For two intact pHDL with a mean value of R = 13.0 (open symbols), the intercept was 0.127 $\pm$ 0.009 (p = 0.000), the slope was 0.017 $\pm$ 0.003 (p = 0.002), and R² = 0.83. The kinetic parameters with S₂ were calculated from the ordinate and abscissa of the intersection point.

We used PEG 7000–9000 to prepare crude HDL for subsequent cholesterol efflux measurements. ApoB depletion by PEG is the most frequent approach in such studies [15, 22, 27] due to quick and efficient plasma/serum fractionation with preservation of HDL integrity. To avoid possible contamination by apoB-containing aggregated lipoproteins, we discarded the thin surface layer after the centrifugation step, if present. Importantly, the procedure, if performed at room temperature, neither involves apoA-I dissociation [28] nor an overall efflux change [29]. These properties are crucial in the preservation of individual efflux steps with lipid-free apoA-I and $\alpha$-HDL-bound apoA-I. Also, HDL isolation by ultracentrifugation leads to apoA-I loss and alters HDL size distribution [22], while chromatography leads to sample dilution and the additional step of fraction concentration; both procedures are time-consuming. The knowledge of HDL heterogeneity by size or density could detail the origin of differences in apoA-I dissociation induced by urea; however, it preserves the major conclusion on the similar cholesterol-accepting properties of lipid-free apoA-I from CAD and control patients. We used murine macrophages RAW 264.7 as a standard cell line to test the cholesterol-accepting capacity of individual pHDL when ABCA1 properties were inwardly preserved. These measurements focus on the interaction between the cholesterol acceptor and the transporter and not on the effect of the intervention on ABCA1 expression [15]. We used the intact and c-AMP-treated cells to specifically induce ABCA1 expression in murine macrophages and included the control sample on the different plates to perform inter-plate standardization of ABCA1-mediated cholesterol efflux. BODIPY-cholesterol is also a more sensitive probe for ABCA1-mediated efflux than radiolabeled cholesterol [30].

There are three major findings in the present study revealed by the comparison of compositional and functional properties of HDL from CAD and control patients.

First, for HDL from CAD patients, apoA-I partitioned more into the water phase during urea-induced HDL denaturation. ApoA-I dissociation is determined by the equilibrium [7, 31] and/or kinetic [32] nature of HDL stability. The first condition implies apoA-I distribution between lipid and water phases by mass law action, while the second condition includes apolipoprotein shedding from the HDL surface induced by metabolic transformation. Metabolic remodeling that includes the transitions between different lipoprotein structures is accompanied by the apolipoprotein A-I dissociation from the HDL surface and particle fusion; lipolytic and transfer activities are the most important. CETP- and HL-induced dissociation of lipid-free apoA-I from HDL with the subsequent formation of discoidal HDL occurs in the presence of VLDL or LDL [33, 34]. This apoA-I dissociation also occurs at HDL thermal or chemical denaturation [6, 21, 32, 35]. The chaotropic agent-induced apoA-I dissociation is irreversible [21], and HDL denaturation induces apoA-I unfolding and partitioning from HDL to water (Fig. 1). Partitioning also occurred in the absence of denaturants [7] that may result in the appearance of lipid-free apoA-I as a principal ligand for the ABCA1 transporter. The spontaneous exchange of apoA-I between lipid-free and lipid-bound forms [8, 36] differs from dissociation, as no net transfer occurs in the exchange. Irrespective of the nature of dissociation, this process results in the existence of up to 10% of total plasma apoA-I in lipid-free form.

Distribution parameter D was inversely associated with the absolute and apoA-I-normalized levels of choline-containing phospholipids and cholesterol in pHDL, opposite to the lack of any association for pHDL from control patients. The increased denaturation of HDL from patients with acute coronary syndrome has been described [35]. Importantly, HDL particles from CAD patients are more enriched with PL and cholesterol normalized by apoA-I level throughout the whole range of HDL-C. The enrichment of HDL with cholesterol, with the concomitant increase of competition between apoA-I and cholesterol for the binding to phospholipid molecules adjacent to apoA-I, is suggested to be involved in the increased apolipoprotein dissociation from HDL in CAD. Cholesterol molecules have been shown to be partially excluded from boundary lipids in reconstituted HDL [37]. The free cholesterol molecules may increasingly accumulate in the vicinity of apoA-I molecules in cholesterol-overloaded pHDL from CAD patients with the additional perturbation of PL structural order and dynamics. What could be a driving force for additional cholesterol accumulation in HDL from CAD patients compared to control patients? Earlier, we measured the upregulated expression of CETP, LPL, and PLTP genes and the downregulated expression of the LCAT gene among 65 genes selected to follow atherogenesis-prone changes in CAD [20, 24]. The differential expression of these genes is suggested to be involved in lipid enrichment of HDL in CAD. Indeed, the enrichment of HDL with phospholipid and cholesterol molecules in CAD could occur due to a decreased phospholipid hydrolysis by LCAT, which possesses phospholipase activity with a concomitant lowering of cholesteryl ester generation. PLTP is capable of transforming small into large HDL particles with a concomitant release of apoA-I [5]; the effect of upregulated expression of the pLTP gene lies in accordance with the increase of apoA-I-normalized cholesterol and phospholipid levels in CAD. The entropically favorable processes of both apoA-I exchange between discoidal HDL-bound and lipid-free pools [25] and the lipid transfer for nanodisks consisting of apoA-I and DMPC [26] with the increased activation entropy and the decrease in the standard entropy have been suggested. The exact contribution of entropy and enthalpy to apoA-I increased dissociation from mature spherical HDL in CAD remains to be determined. The different exposure of the hydrophobic amino acid residues of apoA-I to the aqueous phase in the activated state at apolipoprotein denaturation [21] may be involved. Interestingly, cholesterol in discoidal HDL increased the number of apoA-I tryptophan residues accessible to the aqueous phase but decreased their mean degree of hydration [38].

Second, we revealed significant associations between the apoA-I dissociation parameter and expression of several genes (Table 1). These associations may reveal proatherogenic or atheroprotective effects, and any causal links remain to be separately studied. For HDL from control patients, the D parameter was positively correlated with the ABCA1 and TLR8 expressions. The decrease of lipid-free apoA-I with the increase of ABCA1 activity is suggested to result in the increase of dissociation of lipid-bound apolipoprotein. Of note, the increased apoA-I dissociation seems to possess a proatherogenic effect in control patients due to the positive associations between the D parameter and TLR8 gene expression, which is involved in inflammation [39].

For HDL from CAD patients, the D parameter was associated with the expression of a number of genes involved in lipoprotein metabolism (PRKACB, HMGCR, PLTP, OLR1, CUBN, and ALB) and systemic inflammation (CD36, SLPI, and IL1R1). Interestingly, the expression of apoptosis-related genes IL1R1 and PRKACB with the inverse associations with the D parameter has also been regulated inversely in acute myocardial infarction [40]. The observed inverse associations of the D parameter with CUBN and ALB gene expression may be associated with the increased catabolism of lipid-free apoA-I and the accompanying decrease of the pre$\beta$ fraction in CAD. Indeed, CUBN, as an endocytic receptor, is involved in endocytosis of apoA-I and albumin and maintenance of their blood levels [41]. Mechanistically, PLTP activity promotes both phospholipid transfer from VLDL and LDL to HDL and pre$\beta$₁-HDL formation by loss of apoA-I from larger HDL particles [42, 43], thus explaining the positive association between PLTP gene expression and the D parameter in CAD. However, this association seems not to be significant for control patients due to the efficient transformation of pre$\beta$₁-HDL to larger HDL by LCAT activity in the pre$\beta$₁-HDL cycle [43], which is suggested to be impaired in CAD. For controls, normal PLTP activity with generation of pre$\beta$₁-HDL is balanced by normal LCAT activity with consumption of pre$\beta$₁-HDL. For CAD patients, the increased PLTP activity does not match the decreased LCAT activity. It may be speculated further that PLTP-generated pre$\beta$₁-HDL, not participating efficiently in their metabolism by low LCAT activity in CAD, is increasingly cleared from the circulation because of low CUBN gene expression. Of note, there is a nonconclusive opinion that PLTP activity is a risk factor for human CAD with an unclear effect on cholesterol efflux [44]. Positive association of HMGCR expression with the D parameter only for CAD patients originates from the inverse correlations of HMGCR expression with plasma HDL-C, pHDL-C:apoA-I, and pHDL-PL:A-I (data not shown), which in turn are inversely correlated with the D parameter. Thus, HMGCR-induced increase of cholesterol synthesis in CAD is associated with the increase of apoA-I dissociation from HDL.

Third, lipid-free/lipid-poor apoA-I that may originate from the HDL-bound state, both for CAD and control HDL, seems to be equally efficient in ABCA1-mediated efflux. The two dependencies in the double reciprocal plot intersected in the upper right corner (Fig. 3), which is definite evidence of competition of two substrates for a single enzyme [23] with the relation V₁ (pre$\beta$-HDL) $>$ V₂ ($\alpha$-HDL), which is our case. The competition becomes more prominent with the increase of the $\alpha$-apoA-I:pre$\beta$-apoA-I ratio R (13.0 in this case). The above-metioned low CUBN gene expression in CAD may underlie the increased clearance of PLTP-generated pre$\beta$₁-HDL. We describe for the first time the competition between pre$\beta$-HDL and $\alpha$-HDL for ABCA1 when both substrates during kinetic measurements are present at the same fixed ratio R as in plasma. Two efflux kinetics have to be measured separately: the first with fully denatured HDL and the second with intact HDL. The ability of isolated apoA-I [3, 45] and HDL fractions [3, 13, 46] to interact mainly with ABCA1 has been described with conflicting results. However, the absence of both the second substrate and appropriate kinetic treatment largely hindered competition visualization, while only correlations with the level and composition of other lipoproteins were being analyzed. The obtained V_m and K_m values were lower than the analogous mean values (22.3% and 0.64 µg/mL) for ABCA1-mediated efflux to lipid-free apoA-I for 6 h [3]. Of note, only 40–60% of apoA-I molecules were dissociated in water upon pHDL denaturation by 4.25 M urea (compared to 10–20% of pre$\beta$-apoA-I in intact HDL), and V_m would not reach the maximal limiting value for lipid-free apoA-I due to the lower V_m value for remaining lipid-bound apoA-I. However, similar efflux kinetics at two zones of denaturation transition largely exclude this possibility. Of note, the structure of residual HDL with lipid-bound apoA-I, still existing at the transition midpoint, is perturbed by urea, which excludes the possibility of accepting cholesterol by residual HDL.

The decreased ABCA1-mediated efflux efficiency of intact HDL as a mixture of lipid-free and lipid-bound apoA-I in CAD has been described [17, 18]. Despite the attempts to normalize efflux to pre$\beta$₁ levels as an indicator of lipid-poor apolipoprotein, the conventional CEC measurements were performed with HDL as a mixture of lipid-free and lipid-bound apoA-I without the separation of individual efflux reactions with different acceptors [17, 18, 47]. However, the overall reaction velocity for a two-substrate reaction with a single enzyme depends on the concentrations of both substrates depicted by R values (Eqn. 8), and the numerical simulation of this dependency revealed a non-linear increase of the V_m value with the decrease of the R value. The effective value of the maximal reaction velocity V_m, 7.9%/2 h, for intact HDL from CAD patients with low HDL-C levels and with the mean R value of 13.0 lies between the V₁ (17.5 %/2 h) and V₂ (5.3%/2 h) limits. It can be speculated further that the combined assessment of lipid-free and HDL-bound apoAI could improve CAD risk prediction and clarify the existing inconsistencies in HDL functionality in cholesterol efflux, in particular, the significance of large [47] and small [3] HDL. Also, the combined assessment of two substrates could localize more exactly the impairment of cholesterol efflux by oxidative modification of HDL in CAD to lipid and/or protein moieties [48, 49]. In any case, the obtained data strongly suggest the integrity of the structure of lipid-free apoA-I in both control and CAD patients. However, we cannot exclude the possibility of oxidative modification of lipids in HDL. The exact nature of the dependency of V_m on R for control and CAD patients with different HDL-C levels and the contribution of HDL fractions is a matter of ongoing experiments based on the application of the kinetics of lipid-free and HDL-bound apoA-I competing with each other for the ABCA1 transporter. We suggest that this dependency may underlie the diminished cholesterol efflux in CAD.