Search

High-density lipoproteins (HDL) remove cholesterol from peripheral tissues and thereby help to prevent atherosclerosis. Nascent HDL are discoidal complexes composed of a phospholipid bilayer surrounded by protein alpha-helices that are thought to form extensive stabilizing interhelical salt bridges. Earlier we showed that HDL stability, which is necessary for HDL functions, is modulated by kinetic barriers. Here we test the role of electrostatic interactions in the kinetic stability by analyzing the effects of salt, pH, and point mutations on model discoidal HDL reconstituted from human apolipoprotein C-1 (apoC-1) and dimyristoyl phosphatidylcholine (DMPC). Circular dichroism, Trp fluorescence, and light scattering data show that molar concentrations of NaCl or Na(2)SO(4) increase the apparent melting temperature of apoC-1:DMPC complexes by up to 20 degrees C and decelerate protein unfolding. Arrhenius analysis shows that 1 M NaCl stabilizes the disks by deltaDeltaG* approximately equal 3.5 kcal/mol at 37 degrees C and increases the activation energy of their denaturation and fusion by deltaE(a) approximately equal deltaDeltaH* approximately equal 13 kcal/mol, indicating that the salt-induced stabilization is enthalpy-driven. Denaturation studies in various solvent conditions (pH 5.7-8.2, 0-40% sucrose, 0-2 M trimethylamine N-oxide) suggest that the salt-induced disk stabilization results from ionic screening of unfavorable short-range Coulombic interactions. Thus, the dominant electrostatic interactions in apoC-1:DMPC disks are destabilizing. Comparison of the salt effects on the protein:lipid complexes of various composition reveals an inverse correlation between the lipoprotein stability and the salt-induced stabilization and suggests that short-range electrostatic interactions significantly contribute to lipoprotein stability: the better-optimized these interactions are, the more stable the complex is.

To probe the role of protein conformation in the formation and kinetic stability of discoidal lipoproteins, thermal unfolding and refolding studies were carried out using model lipoproteins reconstituted from dimyristoylphosphatidylcholine (DMPC) and selected mutants of human apolipoprotein C-1 (apoC-1). Circular dichroism (CD) spectroscopy and electron microscopy show that the Q31P mutant, which has alpha-helical content in solution (33%) and on DMPC disks (67%) similar to that of the wild type (WT), forms disks of smaller diameter, = 13 nm, compared to 17 nm of the WT-DMPC disks. The L34P mutant, which is largely unfolded in solution, forms disks with alpha-helix content and diameter similar to those of the WT. The R23P mutant, which is fully unfolded in solution, forms disks that have similar diameter but reduced alpha-helix content (40%) compared to the WT-DMPC disks (65%). Remarkably, despite large variations in the alpha-helix content or the disk diameter among different mutant-DMPC complexes, the mutations have no significant effect on the unfolding rates or the Arrhenius activation energy of the disk denaturation, E(a) = 25-29 kcal/mol. This suggests that the kinetic stability of the discoidal complexes is dominated by the lipid-lipid rather than the protein-lipid interactions. In contrast to the heat denaturation, the lipoprotein reconstitution upon cooling monitored by CD and light scattering is significantly affected by mutations, with Q31P forming disks in the broadest and R23P in the narrowest temperature range. Our results suggest that the apolipoprotein helical structure in solution facilitates reconstitution of discoidal lipoproteins but has no significant effect on their kinetic stability.

Thermal unfolding of discoidal complexes of apolipoprotein (apo) C-1 with dimyristoyl phosphatidylcholine (DMPC) reveals a novel mechanism of lipoprotein stabilization that is based on kinetics rather than thermodynamics. Far-UV CD melting curves recorded at several heating/cooling rates from 0.047 to 1.34 K/min show hysteresis and scan rate dependence characteristic of slow nonequilibrium transitions. At slow heating rates, the apoC-1 unfolding in the complexes starts just above 25 degrees C and has an apparent melting temperature T(m) approximately 48 +/- 1.5 degrees C, close to T(m) = 51 +/- 1.5 degrees C of free protein. Thus, DMPC binding may not substantially increase the low apparent thermodynamic stability of apoC-1, DeltaG(25 degrees C) < 2 kcal/mol. The scan rate dependence of T(m) and Arrhenius analysis of the kinetic data suggest an activation enthalpy E(a) = 25 +/- 5 kcal/mol that provides the major contribution to the free energy barrier for the protein unfolding on the disk, DeltaG > or = 17 kcal/mol. Consequently, apoC-1/DMPC disks are kinetically but not thermodynamically stable. To explore the origins of this kinetic stability, we utilized dynode voltage measured in CD experiments that shows temperature-dependent contribution from UV light scattering of apoC-1/DMPC complexes (d approximately 20 nm). Correlation of CD and dynode voltage melting curves recorded at 222 nm indicates close coupling between protein unfolding and an increase in the complex size and/or lamellar structure, suggesting that the enthalpic barrier arises from transient disruption of lipid packing interactions upon disk-to-vesicle fusion. We hypothesize that a kinetic mechanism may provide a general strategy for lipoprotein stabilization that facilitates complex stability and compositional variability in the absence of high packing specificity.

Solution structure of lipid-free apolipoprotein C-1 (apoC-1, 6.6 kD) was analyzed by circular dichroism (CD) of 15 mutants containing single Pro or Ala substitutions in predicted alpha-helical regions. While the majority of Pro substitutions induce complete (L11P, L18P, R23P, I29P, M38P, W41P, T45P) or partial (G15P, L34P) helical unfolding, similar substitutions at other sites (A7P, Q31P, V49P, L53P) do not cause large changes in the secondary structure or stability. The results suggest that lipid-free apoC-1 is comprised of two dynamic helices that are stabilized by interhelical interactions and are connected by a short linker containing residues 30-33. We propose that the minimal folding unit in the lipid-free state of this and other exchangeable apolipoproteins comprises the helix-turn-helix motif formed of four 11-mer sequence repeats. Comparison of the helical content in lipid-free and lipid-bound apoC-1 suggests that lipid binding shifts the conformational equilibrium toward preexisting highly helical conformation. Remarkably, near-UV CD spectra of wild type and mutant apoC-1 are not significantly altered upon thermal or chemical unfolding and thus result from residual aromatic clustering that is retained in the unfolded state. Correlation of far- and near-UV CD of the mutant peptides suggests that the hydrophobic cluster containing W41 is essential for the helical stability and may form a helix nucleation site in apoC-1.

The high-resolution conformation of human apoC-I in complexes with sodium dodecyl sulfate (SDS) is presented. As estimated from CD data, apoC-I adopts 54% helical secondary structure when bound to SDS, which is similar to the helical content previously found with phospholipids. The NMR-derived conformation of apoC-I is composed of two amphipathic helices, residues 7-29 and 38-52, separated by a flexible linker. The N-terminal helix contains a mobile hinge involving residues 12-15. The hydrophobic side chains cluster on the nonpolar face of both helices, thus forming two discrete lipid-binding sites in the N-terminal helix and one in the C-terminal helix. As suggested by amide proton resonance line widths and deuterium exchange rates, the N-terminal helix is more flexible and may bind less tightly to the detergent than the C-terminal helix. The different mobility of both helices appears to be related to side-chain composition, rather than length of the amphipathic helix, and may play a role in the function of apoC-I as an activator of lecithin:cholesterol acyltransferase (LCAT). A model is suggested in which the C-terminal helix serves as a lipid anchor while the N-terminal helix may hinge off the lipid surface to make specific contacts with LCAT.

Thermal and chemical unfolding of lipid-free apolipoprotein C-1 (apoC-1), a 6-kDa protein component of very low density and high-density lipoproteins, was analyzed by far-UV CD. In neutral 1 mM Na2HPO4 solutions containing 6-7 micrograms/mL protein, the apoC-1 monomer is approximately 30% alpha-helical at 0-22 degrees C and unfolds reversibly from about 22-80 degrees C with Tm = 51 +/- 3 degrees C and van't Hoff enthalpy delta Hv(Tm) = 19 +/- 3 kcal/mol. The apparent free energy of the monomer stabilization determined from the chemical unfolding at 0 degree C, delta G(0 degree C) = 2.8 +/- 0.8 kcal/mol, decreases by about 1 kcal/mol upon heating to 25 degrees C. A small apparent heat capacity increment suggests the absence of a substantial hydrophobic core for the apoC-1 molecule. At pH 7, increasing apoC-1 concentration above 10 micrograms/mL leads to self-association and formation of additional alpha-helices that unfold upon both heating and cooling from room temperature. The CD data indicate that the high-temperature transition reflects a complete monomer unfolding and the low-temperature transition reflects oligomer dissociation into stable monomers. This suggests the importance of hydrophobic interactions for apoC-1 self-association. Close proximity between the high- and low-temperature transitions and the absence of a plateau in the chemical unfolding curves recorded from oligomeric apoC-1 indicate marginal oligomer stability and suggest that in vivo apoC-1 transfer is mediated via the complexes with other apolipoproteins and/or lipids.

Infrared absorption spectra are reported for six apolipoprotein fragments in SDS/D2O. Five of the peptides correspond to proposed lipid-binding domains of human apolipoproteins [apoC-I(7-24), apoC-I(35-53), apoA-II(18-30)+, apoA-I(166-185), apoE(267-289)], and the sixth is the de novo lipid associating peptide LAP-20. The amide I infrared absorption patterns are generally consistent with predominantly helical structures (as determined previously by NMR spectroscopy and distance geometry calculations) and further suggest that apoA-I(166-185) and apoE(267-289) are bound to SDS relatively weakly in comparison to the other four peptides. The latter conclusion is also supported by the temperature dependence of the infrared spectra, as increasing temperature promotes a distinct increase in random coil structure only for apoA-I(166-185) and apoE(267-289). In addition to features readily ascribed to helices, the infrared spectra of all the peptides show absorptions in the spectral region 1630-1635 cm-1 that is usually associated with beta-structure, a motif that is clearly absent from the NMR-derived structures. Parallel difficulties also arose in the analyses of the circular dichroism spectra. We suggest that both the low-frequency infrared absorptions and the ambiguities in interpreting the CD spectra may be due to unusual structures at the peptide C-termini, involving C=O groups that form hydrogen bonds simultaneously either with two solvent molecules or with donors from the backbone (NH) and the solvent (OH). Analogous absorptions may be a general feature of solvent-exposed helices, which suggests a need for caution in assigning amide I bands below 1640 cm-1.

Peptides corresponding to the proposed lipid-binding domains of human apolipoprotein C-I, residues 7-24 (ALDKLKEFGNTLEDKARE) and 35-53 (SAKMREWFSETFQKVKEKL), were studied by CD and two-dimensional 1H NMR spectroscopy. Sodium dodecyl sulfate (SDS) was used to model the lipoprotein environment. Analysis of the CD data shows that both peptides lack well-defined structure in aqueous solution but adopt helical, ordered structures upon the addition of SDS. The helical nature of the peptides in the presence of SDS was confirmed by H alpha secondary shifts. A total of 199 (apoC-I(7-24)) and 266 (apoC-I(35-53)) distance restraints were used in distance geometry and simulated annealing calculations to generate average structures for both peptides in aqueous solutions containing SDS. The backbone (N, C alpha, C = O) RMSD from the average structure of an ensemble of 20 structures was 0.73 +/- 0.22 and 0.48 +/- 0.14 A for apoC-I(7-24) and apoC-I(35-53), respectively. In the presence of SDS, the distance geometry and simulated annealing calculations show that both peptides adopt well-defined amphipathic helices with distinct hydrophobic and hydrophilic faces. The calculated structures are discussed relative to predicted structures. Comparing our CD and NMR results for the apoC-I fragments in SDS with CD results of others obtained in the presence of dimyristoylphosphatidylcholine indicates that SDS may be a better model of the lipoprotein environment.

Human apolipoprotein C-I (apo C-I) in solution, in monomeric and oligomeric form, and in micellar complexes with dimyristoylphosphatidylcholine (DMPC), below and above the phase transition temperature of DMPC, was investigated with steady-state and time-resolved fluorescence methods. The environment of the Trp residue of apo C-I, in each physical state, was evaluated from fluorescence spectra and their changes upon KI quenching. Rotational correlation times of Trp residues were obtained from fluorescence anisotropy decay measurements. Static fluorescence anisotropy was determined as a function of temperature for the Trp residues of apo C-I in all physical states and for diphenylhexatriene dissolved in apo C-I X DMPC complexes. It was found that the Trp residues of apo C-I in solution are exposed from 75 to 88% to the aqueous medium, depending on the state of self-association. On the other hand, the Trp residues in apo C-I X DMPC complexes are only 42-45% exposed to KI quenching through an environment distinct from water. Apolipoprotein C-I in all its physical forms had two rotational correlation times associated with Trp motions: a longer one dependent on the size and flexibility of the entire particle and a very short one in the range from 0.2 to 0.4 ns. The later correlation times correspond to local Trp residue motions. These Trp motions were not significantly affected by a transition from the gel to the liquid-crystalline state of the lipid in apo C-I X DMPC complexes, suggesting that there is no coupling between the local motions of lipids and those of Trp side chains of apo C-I.

We have investigated the protease activity, present in human serum, that digests the serum amyloid A (SAA) protein. SAA radiolabeled with 125I was incubated at 37 degrees C with serum and plasma and analyzed for degradation products by alkaline urea-polyacrylamide gel electrophoresis and gel filtration chromatography. Serum initially digested SAA to intermediates of 3000-5000 in molecular weight, and these were further degraded to smaller peptides with prolonged incubation. SAA was not degraded by plasma anticoagulated with ethylenediaminetetraacetic acid (EDTA) or heparin. Recalcification of plasma anticoagulated with EDTA led to the generation of protease activity against SAA whereas EDTA plasma defibrinated with thrombin was inactive. We employed both nonselective and selective protease inhibitors and synthetic substrates for kallikrein and plasmin to further characterize the serum protease. These studies demonstrated that degradation of SAA is not directly attributable to enzymes involved in coagulation, kinin formation, or fibrinolysis, but the unidentified protease may be activated by one of the clotting factors. The specificity of the SAA degradation was demonstrated in experiments with three of the well-characterized apolipoproteins. Apolipoproteins A-I, C-I, and C-III-1, which also associate with the plasma high-density lipoproteins, were not degraded by serum although they were good substrates for purified thrombin and plasmin.

The apoproteins (apo) C-I, C-II, and C-III are low molecular weight amphiphilic proteins that are associated with the lipid surface of the plasma chylomicron, very low density lipoprotein (VLDL), and high-density lipoprotein (HDL) subfractions. Purified apoC-I spontaneously reassociates with VLDL, HDL, and single-bilayer vesicles (SBV) of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine. ApoC-I also transfers reversibly from VLDL to HDL and from VLDL and HDL to SBV. The kinetics of association of the individual apoC proteins with SBV are second order overall and first order with respect to lipid and protein concentrations. At 37 degrees C, the rates of association were 2.5 x 10(10), 4.0 x 10(10) and 3.8 x 10(10) M-1 s-1 for apoC-I, apoC-II, and apoC-III, respectively. Arrhenius plots of association rate vs temperature were linear and yielded activation energies of 11.0 (apoC-I), 9.0 (apoC-II), and 10.6 kcal/mol (apoC-III). The kinetics of vesicle to vesicle apoprotein transfer are biexponential for intermembrane transfer, indicating two concurrent transfer processes. Rate constants at 37 degrees C for the fast component of dissociation were 11.7, 9.5, and 9.9 s-1, while rate constants for the slow component were 1.3, 0.6, and 0.9 s-1 for apoC-I, apoC-II, and apoC-III, respectively. The dissociation constants, Kd, of apoC-I, apoC-II, and apoC-III bound to the surface monolayer of phospholipid-coated latex beads were 0.5, 1.4, and 0.5 microM, respectively. These studies show that the apoC proteins are in dynamic equilibrium among phospholipid surfaces on a time scale that is rapid compared to lipolysis, lipid transfer, and lipoprotein turnover.

One of the three proposed lipid-binding regions of the human apolipoprotein C-I (apo-C-I) is an amphipathic helix which extends from residue 33 to residue 53 and includes a single methionine at sequence position 38. The involvement of the sequence around methionine-38 in phospholipid binding has been evaluated with paramagnetic and nuclear reported groups attached to the thiomethyl moiety. This moiety has been spin-labeled with N-(2,2,6,6-tetramethylpiperidinyl-1-oxy)bromoacetamide or 13C enriched with 13CH3I. As determined from its EPR spectrum, the nitroxide at Met-38 of apoC-I had a rotational correlation time (tau C) of 0.22 ns. When dimyristoylphosphatidylcholine (DMPC) was bound to the spin-labeled apoprotein, tau c increased to 0.35 ns, indicating decreased motion for the methionyl side chain. The line width (nu 1/2) and spin--lattice relaxation time (T1) for the thiomethyl resonance of 13C-enriched apoC-I in 10 mM phosphate buffer was 6.0 Hz and 320 ms, respectively. When the protein solution was made 1.6 M in Gdn-HCl, these values changed to 2.6 Hz and 970 ms, respectively. Upon addition of DMPC multilamellar liposomes to [13C]apoC-I in 1.6 M Gdn-HCl, the line width increased to 4.7 Hz and the T1 decreased to 380 ms. These results strongly suggest that methionine-38 of apoC-I resides in a region of the apoprotein which undergoes significant secondary and/or tertiary structural change upon disaggregation/unfolding in Gdn-HCl and upon interaction with phospholipid.

We used antisera to human A and C apolipoproteins to identify homologues of these proteins among the high-density lipoprotein apoproteins of Macaca fascicularis (cynomolgus) monkeys, and NH2-terminal analysis was used to verify the homology. The NH2-terminal sequence of the M. fascicularis apoA-I is identical with that of another Old World species, Erythrocebus patas, and differs from human apoA-I at only 4 of the first 24 residues. M. fascicularis apoA-II contains a serine for cysteine replacement at position 6 and is therefore monomeric like the apoA-II from all species below apes. Human and monkey apoA-II are not otherwise different through their first 25 residues. About 20% of M. fascicularis apoC-I aligns with human apoC-I through residue 22, and 80% lacks an NH2-terminal dipeptide. Otherwise, the monkey apoC-I differs from the human protein at only 2 of 25 positions. Two forms of M. fascicularis apoC-II were identified. ApoC-II1 is highly homologous with human apoC-II, whereas an NH2-terminal hexapeptide is absent from apoC-II2. ApoC-II2 was the predominant species, and apoC-II1 appears to represent a propeptide from which a hexapeptide prosegment is cleaved at a Gln-Asp bond. Both forms of monkey apoC-II are potent activators of lipoprotein lipase. There are two polymorphic forms of M. fascicularis apoC-III, and their electrophoretic mobilities become identical after treatment with neuraminidase. Except for a glycine for serine substitution at position 10, the first 15 NH2-terminal residues of M. fascicularis and human apoC-III are the same.