Interfacial exclusion pressure determines the ability of apolipoprotein A-IV truncation mutants to activate cholesterol ester transfer protein.

We used a panel of recombinant human apolipoprotein (apo) A-IV truncation mutants, in which pairs of 22-mer alpha-helices were sequentially deleted along the primary sequence, to examine the impact of protein structure and interfacial activity on the ability of apoA-IV to activate cholesterol ester transfer protein. Circular dichroism and fluorescence spectroscopy revealed that the secondary structure, conformation, and molecular stability of recombinant human apoA-IV were identical to the native protein. However, deletion of any of the alpha-helical domains in apoA-IV disrupted its tertiary structure and impaired its molecular stability. Surprisingly, determination of the water/phospholipid interfacial exclusion pressure of the apoA-IV truncation mutants revealed that, for most, deletion of amphipathic alpha-helical domains increased their affinity for phospholipid monolayers. All of the truncation mutants activated the transfer of fluorescent-labeled cholesterol esters between high and low density lipoproteins at a rate higher than native apoA-IV. There was a strong positive correlation (r = 0.790, p = 0.002) between the rate constant for cholesterol ester transfer and interfacial exclusion pressure. We conclude that molecular interfacial exclusion pressure, rather than specific helical domains, determines the degree to which apoA-IV, and likely other apolipoproteins, facilitate cholesterol ester transfer protein-mediated lipid exchange.

dissociates from their surface (5) and thereafter circulates primarily as a lipid-free protein (6). A broad spectrum of physiologic functions has been proposed for apoA-IV in human lipid metabolism (7,8), including specific roles in intestinal lipid absorption (8), intravascular lipoprotein metabolism (9 -12), cellular cholesterol efflux (13,14) by interaction with the ABCA1 transporter (15,16), and regulation of the activity of two key proteins involved in the process of reverse cholesterol transport:lecithin-cholesterol acyltransferase (17,18) and cholesterol ester transfer protein (19,20).
Cholesterol ester transfer protein (CETP) is a 74,000-dalton plasma protein that facilitates the exchange of non-polar lipids among circulating lipoproteins (21). CETP-mediated transfer of cholesterol esters from HDL to very low density lipoprotein is central to the process of reverse cholesterol transport (22). It is well established that lipoprotein lipid composition modulates CETP activity (23), but the role of apolipoproteins is less clear. In studies that used lipid emulsions as model lipoproteins, apolipoproteins A-I, A-II, A-IV, and E stimulated CETP-catalyzed lipid exchange equally well (20,24,25). However, studies that assayed CETP activity using native lipoproteins found that apoA-IV may have distinct concentration-dependant effects on CETP-catalyzed lipid exchange between HDL and LDL (26) and among HDL subfractions (19,27).
Although like all apolipoproteins, apoA-IV has a high content of amphipathic ␣-helical structure (28,29), the amphipathic helices in apoA-IV are very hydrophilic (29), are predominantly of the Y-class (30) and are incapable of deeply penetrating lipid monolayers (31,32). With increasing surface pressure, these helices are sequentially excluded from the interface (33,34). Consequently, the interaction of apoA-IV with lipoproteins is very labile and is sensitive to interfacial pressure (35). We have proposed that these properties enable apoA-IV to act as a barostat which maintains lipoprotein surface pressure within a critical range required for optimal activity of lipolytic enzymes and transfer proteins (33,36,37). In this regard, apoA-IV possesses dynamic interfacial properties that are optimal for stabilizing surface tension and lipid packing at expanding lipid/ aqueous interfaces (34).
Using a panel of recombinant human apoA-IV truncation mutants, Emmanuel et al. (38) found that a specific ␣-helical domain located between residues 117 and 160 determines the catalytic efficiency of apoA-IV in the lecithin:cholesterol acyltransferase reaction. Given the impact of distinctive ␣-helical domains in determining the interfacial properties of apoA-IV, and given the importance of interfacial phenomena in the reg-ulation of the CETP reaction (36), we have used this same panel of truncation mutants to examine the impact of ␣-helical structure, protein conformation, and interfacial activity on the ability of apoA-IV to modulate CETP-catalyzed cholesterol ester transfer between HDL and LDL.

EXPERIMENTAL PROCEDURES
Lipids-Egg phosphatidylcholine and cholesterol oleate (Sigma) were dissolved in high performance liquid chromatography grade chloroform (Aldrich, Milwaukee, WI) and stored under nitrogen at Ϫ20°C. Phospholipid concentration was determined by phosphorous analysis (39). The self-quenching fluorescent cholesterol ester analog NBDCE was obtained from Molecular Probes (Eugene, OR). Lipids were Ͼ99% pure by TLC on silica gel.
Lipoproteins, Apolipoproteins, and Cholesterol Ester Transfer Protein-LDL and HDL were isolated from human plasma by sequential flotation at 1.019 -1.063 and 1.125-1.25 g/ml, respectively, and dialyzed against 10 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.025% sodium azide. Human apoA-I was isolated from HDL (40). Human apoA-IV was isolated from donors homozygous for the A-IV-1 and A-IV-2 alleles (41). Recombinant human apoA-IV truncation mutants were created by sitedirected mutagenesis of a DNA construct containing the human apoA-IV coding sequence, followed by expression in Escherichia coli (38). Recombinant human CETP was produced in a Chinese hamster ovary cell line expressing a cloned human CETP cDNA; the specific activity of recombinant CETP, as assayed by the transfer of 14 C-cholesterol ester from HDL to LDL, was identical to plasma CETP (42). Protein concentration was determined with bicinchoninic acid (43). All preparations were homogeneous by SDS-PAGE. Mean molecular hydrophobicities and helical hydrophobic moments of recombinant apoA-IV truncation mutants were calculated using a consensus hydrophobicity scale (44).
Circular Dichroism Spectroscopy-Circular dichroism studies were performed on a Jasco J-720 Spectropolarimeter. Spectra of apolipoproteins at 3 M in 50 mM Tris, pH 7.4, 1 mM EDTA, 0.02% sodium azide were recorded at 25°C from 190 to 260 nm using a 1-mm thermostated cell, 1-mm spectral bandwidth, and 2-s time constant. Buffer blanks were digitally subtracted. Percent ␣-helicity was calculated as previously described (28). Thermal denaturation studies were performed by monitoring ellipticity at 222 nm as a function of temperature from 20 to 65°C. The enthalpy of denaturation, ⌬H D , and the thermal denaturation midpoint, T m , were determined from the slope of plots of ⌬G versus 1/T (45).
Fluorescence Spectroscopy-Fluorescence studies were performed on an SLM 8000C spectrofluorometer. Spectra of apolipoproteins at 3 M in 50 mM Tris, pH 7.4, 1 mM EDTA, 0.02% sodium azide were recorded at 25°C from 290 to 370 nm using a 1-cm cell with excitation at 280 nm, 1-s integration, and 4-nm slits on excitation and emission monochromators. Spectra were excitation corrected by reference to a rhodamine quantum counter, and corrected for scatter and Raman emission by digital subtraction of buffer blanks. Chemical denaturation studies were performed by addition of buffered 6 M guanidinium hydrochloride. Quenching studies were performed by addition of buffered 6 M KI; Stern-Volmer quenching constants, K q , were obtained from plots of I o /I versus [KI]; fractional tryptophan exposure was calculated as K q (apolipoprotein)/K q (N-acetyltryptophanamide) (46).
Determination of Interfacial Exclusion Pressure-The interfacial exclusion pressure (⌸ ex ) of recombinant apolipoproteins at the phospho-lipid/water interface was determined using a KSV 5000 Langmuir Film Balance (KSV Instruments, Helsinki, Finland), as previously described (33,36). Egg phosphatidylcholine monolayers were spread at the air/ buffer interface at initial pressures (⌸ i ) of 5-30 mN/m and then apolipoproteins were injected into the subphase to give a final concentration of 50 g/dl, which had been previously determined to yield saturating binding to the interface. The increase in interfacial pressure was then monitored by a Wilhelmy plate until it reached a stable plateau (⌬⌸). ⌸ ex was determined by extrapolation of ⌬⌸ versus ⌸ i . The precision of the interfacial pressure measurements was Ϯ0.1 mN/m. Fluorescent Cholesterol Ester Transfer Assay-Fluorescent cholesterol ester transfer assays were performed using an SLM 8000C spectrofluorometer equipped with a thermostated cell holder. Recombinant HDL (rHDL) containing apoA-I, egg phosphatidylcholine, cholesterol oleate, and the self-quenching fluorescent cholesterol ester analog ND-BCE were prepared by the method of Pittman et al. (47). Reaction mixtures containing NBDCE-labeled rHDL (0.1 mg/ml), LDL (1 mg/ml), apolipoproteins (5 M), and recombinant human CETP (5 g) in a final volume of 150 l were incubated at 37°C in a 3 ϫ 3-mm magnetically stirred microcuvette, and the fluorescence intensity at 535 nm was monitored for 30 min with excitation at 465 nm, a 1-s integration rate, and 4-nm monochromator slits. Transfer rate constants were calculated from the slope of a linear regression of time versus ln(I max Ϫ I x ), where I max is the maximal fluorescence intensity and I x is the fluorescence intensity at time X.

Calculated Hydrophobic Properties of Recombinant ApoA-IV
Truncation Mutants-Recombinant human apoA-IV (r-AIV) and a panel of recombinant apoA-IV truncation mutants in which pairs of 22-mer ␣-helices were sequentially deleted along the primary sequence from the amino to carboxyl terminus were expressed in E. coli (38). Although the calculated mean hydrophobicities, ϽHϾ, and mean helical hydrophobic moments, ϽϾ, of the deleted segments varied considerably, their deletion had a negligible impact on the hydrophobic properties of the remaining molecule (Table I).
Spectroscopic Studies-Circular dichroism spectra of r-AIV revealed a mean residue ellipticity of 16,649 deg cm 2 /dmol at 222 nm, which corresponds to 50% ␣-helical structure (Table  II). Thermal denaturation of r-AIV yielded an enthalpy of denaturation, ⌬H D , of 70.8 kcal/mol with a transition midpoint at 50.3°C. These parameters are similar to those obtained for native human apoA-IV-1 (28,31,45). Likewise, fluorescence spectroscopy of r-AIV revealed several features characteristic of human apoA-IV (28,46). 1) The fluorescence emission of its single tryptophan was blue shifted to 328 nm. 2) Tryptophan fluorescence was resistant to iodide quenching, indicating that the tryptophan resides in a hydrophobic and/or negatively shielded environment. 3) Chemical denaturation induced a multiphasic red-shift in the wavelength of maximum fluorescence emission (data not shown) and an increase in relative quantum yield (I max /I o ) at the denaturation midpoint (Table II). Together these findings suggest that the secondary structure, conformation, and molecular stability of r-AIV are identical to that of native human apoA-IV. ApoA-IV truncation mutants were much more sensitive to thermal denaturation, as evidenced by ⌬H D values that were 19.6 -51.0 kcal/mol lower than either native or the parent recombinant apoA-IV. Interestingly the ␣-helicity of all the truncation mutants was higher, perhaps reflecting recruitment of some local secondary structure with disruption of ordered folding. The tryptophan fluorescence emission in all the deletion mutants was red-shifted compared with native and r-AIV, although iodide accessibility was lower. Moreover, none of the truncation mutants displayed the increase in quantum yield with denaturation that is a signature of native human apoA-IV (28,46). Taken together, these findings demonstrate that deletion of any of the ␣-helical domains in apoA-IV significantly disrupts its tertiary structure and impairs its molecular stability.
Exclusion Pressure at the Lipid/Water Interface-Examination of the binding of r-AIV to egg phosphatidylcholine monolayers revealed that r-AIV was excluded from the lipid/water interface at pressures above 34.3 mN/m. In comparison, the exclusion pressure was 27.0 mN/m for native human apoA-IV-1, 30.8 mN/m for native apoA-IV-2, and 33.0 mN/m for apoA-I, as previously noted (33). The higher interfacial exclusion pressure of r-AIV relative to the native apolipoproteins could be a consequence of the presence of a hydrophobic MRGS(H) 6 decapetide at its amino terminus, which was added to the recombinant construct to facilitate purification (38). Comparison of the interfacial exclusion pressure of the recombinant apoA-IV truncation mutants with r-AIV yielded the surprising finding that, with the exception of the ⌬h1-2 and ⌬h9 -10 mutants, deletion of amphipathic ␣-helical domains increased the interfacial exclusion pressure (Fig. 1). There was no consistent relationship between either the hydrophobic properties of the excluded helices or the ␣-helicity of the remaining molecule and the interfacial exclusion pressure, although deletion of the exclusively Y class helices (30) near the COOH terminus yielded proteins with the highest surface activity.
Effect on the Rate of CETP-mediated Cholesterol Ester Transfer-Incubation of NDBCE-labeled rHDL with a 10-fold excess of LDL produced little change in fluorescence intensity; however, in the presence of CETP, transfer of the probe from the recombinant HDL to LDL caused a rise in fluorescence emission. The transfer rate constant in the presence of the common human apoA-IV-1 isoprotein was 2.22 ϫ 10 Ϫ3 s Ϫ1 , for the variant human apoA-IV-2 isoprotein it was 2.59 ϫ 10 Ϫ3 s Ϫ1 , and for r-AIV it was 2.92 ϫ 10 Ϫ3 s Ϫ1 . In the presence of recombinant apoA-IV truncation mutants in the reaction mixture, the transfer rate constant varied between 2.45 and 3.29 ϫ 10 Ϫ3 s Ϫ1 . Multiple factor correlation analysis identified no significant relationship between the transfer rate constants and any of the spectroscopic parameters, but a strong correlation was found between the transfer rate constants and interfacial exclusion pressure (r ϭ 0.790, p ϭ 0.002) (Fig. 2). DISCUSSION In solution, apoA-IV adopts a loosely folded "molten globule" conformation, in which the hydrophobic faces of its multiple amphipathic ␣-helices face inwards toward the interior of the molecule (28,46). Our present findings demonstrate that these repeated ␣-helices collectively and cooperatively contribute to the molecular stability of apoA-IV in solution, for deletion of even short helical segments disrupts its tertiary conformation and significantly destabilizes the molecule. This is further suggested by the observation that these same deletions disrupt the otherwise strong self-association of apoA-IV (38). Interestingly, conformational destabilization of apoA-IV was associated with induction of additional ␣-helical structure, probably as a consequence of coil 3 helix transformations in regions of random coil structure, as occurs when it binds to phospholipid vesicles (31), or possibly because of altered self-association. Paradoxically, destabilization of the tertiary structure of the apoA-IV truncation mutants increased their interfacial exclusion pressure, perhaps because apoA-IV must unfold before it can fully adsorb to lipid/water interfaces (31,33), or perhaps because destabilization of apoA-IV in solution increases the free energy gradient for its adsorption to the hydrophobic interface.
The data from this unique panel of apoA-IV truncation mu-  tants do not support a role of specific helical domains in activating CETP. However, the observation of a relationship between interfacial exclusion pressure and cholesterol ester transfer rate sheds additional light on the biophysical role of apolipoproteins in the CETP reaction. CETP catalyzes the movement of non-polar lipids between lipoproteins along chemical concentration gradients (21,23). Apolipoproteins could modulate this process by: 1) controlling the physical interaction of CETP with the surface of acceptor and/or donor particles; 2) modulating the interfacial solubility and accessibility of cholesterol esters; and 3) stabilizing acceptor particles as they acquire additional core lipids.
Regarding the first mechanism, the exclusion pressure of a lipid transfer protein must be close to the pressure of donor and acceptor interfaces, so that its binding is reversible; otherwise it could bind too tightly and become "locked" on the interface. Indeed, the calculated pressure of the HDL3 surface is 33 mN/m (33) and the interfacial exclusion pressure of CETP is 31 mN/m (36). Thus, apolipoproteins could control the association of CETP with donor and acceptor lipoproteins by modulating surface pressure. For example, there is little lipid transfer between naked lipid emulsion particles with CETP alone (24,25), for CETP binds tightly to both donor and acceptor particles (48). The presence of apolipoproteins with exclusion pressures higher than CETP enables it to dissociate from donor particles after cholesterol ester loading, and, especially, from expanding acceptor particles after cholesterol ester delivery. If such phenomena predominate, then the efficacy of apolipoproteins in activating or inhibiting CETP will be a function of both their exclusion pressure and their ambient concentration (26).
Regarding the second and third mechanisms, both the microsolubilization of membrane-free cholesterol and phospholipids (49) and the solubility of cholesterol esters in phospholipid monolayers (36) is a function of interfacial pressure. Thus, adsorption of apolipoproteins to acceptor and donor particles could facilitate loading/unloading of cholesterol ester molecules by optimizing their surface accessibility and buffering any changes in the chemical concentration gradient as the reaction proceeds. In this regard, the relationship between the interfacial exclusion pressure of apoA-IV truncation mutants and cholesterol ester transfer rate depicted in Fig. 2 is remarkably similar to the relationship between exclusion pressure and the ability of apolipoproteins to promote phospholipid efflux from free cholesterol-enriched fibroblasts shown in Fig. 6 in the paper by Gillote et al. (49). In addition, by reducing surface tension, adsorption of apolipoproteins to acceptor particles would minimize the free energy cost of expansion, analogous to the role of the insect protein Lp-III in stabilizing hemolymph high density lipophorin particles as they acquire diacylglycerol transported by the lipid transfer particle (50). If such phenomena predominate, then the efficiency with which apolipoproteins activate CETP will be directly proportional to their surface activity (25).
Given that a large fraction of apoA-IV circulates as a lipidfree apolipoprotein (6), and that both plasma apoA-IV levels (5,51) and CETP activity (52) increase in the post-prandial period, it is possible that apoA-IV could modulate CETP activity in vivo. In this regard, the higher exclusion pressure and cholesterol ester transfer rate of the apoA-IV-2 isoprotein raise the question as to whether this variant increases CETP activity in vivo. Several population screening studies have observed lower HDL levels (53,54), higher LDL levels (55,56), and an increased LDL/HDL ratio (54) in individuals carrying an A-IV-2 allele, as would be expected with increased CETP activity (57). However, in the only population survey to directly measure CETP activity, von Eckardstein et al. (58) found lower CETP activity in apoA-IV-1/2 heterozygotes. Possibly, other apoA-IV genetic variants with altered interfacial properties (59) could influence CETP activity.
Finally, these data raise several caveats regarding the use of modified recombinant apolipoproteins in biological studies. First, they illustrate how minor modifications made to facilitate protein purification may have major structural and functional consequences. In this case, addition of a hydrophobic decapeptide to the amino terminus of the r-AIV construct significantly increased its surface binding affinity and its ability to activate CETP. Second, these data illustrate the need for a full biophysical characterization of recombinant apolipoproteins used to probe the function of local domains. Again, the amino-terminal modification of r-AIV appeared to have no affect on its solution structure as assessed spectroscopically, yet it had a major impact on its interfacial activity. Likewise, spectroscopic characterization of the deletion mutants, alone, provided no indication as to how their apparent structural destabilization could increase their catalytic efficacy in the CETP reaction. Failure to fully characterize the biophysical properties of modified recombinant apolipoproteins could also lead to erroneously attributing a specific biologic function to local structure, when in fact, it was determined by global properties, such as interfacial activity. In this regard, however, comparison of our data and that of Emmanuel et al. (38) supports their conclusion that helixes 5 and 6 (residues 117-160) in apoA-IV play a specific catalytic role in lecithin:cholesterol acyltransferase activation.
In summary using a panel of recombinant human apoA-IV truncation mutants, we have found a strong correlation between interfacial exclusion pressure and the ability to activate CETP. We conclude that molecular interfacial exclusion pressure, rather than specific helical domains, determines the degree to which apoA-IV, and most likely other apolipoproteins, facilitate CETP-mediated lipid exchanges.