Apolipoprotein A-I Assumes a “Looped Belt” Conformation on Reconstituted High Density Lipoprotein*

Apolipoprotein A-I (apoA-I) plays a central role in the reverse cholesterol transport pathway; however, the structural basis for its antiatherogenic effects remains poorly understood. Here we employ EPR spectroscopy and fluorescence resonance energy transfer to elucidate the conformation and relative alignment of apoA-I monomers on discoidal (9.4 nm) reconstituted high density lipoprotein (rHDL). EPR spectroscopy provided evidence for an extended helical secondary structure. Position 139 since it was the only residue examined to display a dynamic motional character consistent with a flexible loop structure. The EPR spectra of nitroxide probes at positions 133 and 146 exhibit spin coupling, indicating that these positions are proximal to an apoA-I paired counterpart on the perimeter of rHDL. fluorescence resonance energy transfer studies employing engineered apoA-I variants possessing a single tryptophan (energy donor) and/or a single cysteine (whose thiol moiety was covalently labeled with an extrinsic energy acceptor) provided evidence that paired apoA-I molecules around the perimeter of rHDL align in an extended antiparallel conformation. Taken together with the observation that the EPR spectra of nitroxide probes positioned at intervening sequence positions (134-145) do not exhibit spin coupling, this has led us to propose a “looped belt” model, wherein residues 133-146 comprise a flexible loop segment that confers to apoA-I an intrinsic ability to adapt its structure to accommodate changing particle lipid content. Specifically, in the looped belt model, with the exception of amino acids 134-145, apoA-I aligns with its counterpart in a helix 5-helix 5 registry, centered at position 139.

A majority of high density lipoprotein (HDL) 2 functionality is derived from the ability of apolipoprotein A-I (apoA-I) to sequester phospholipid and cholesterol and interact with plasma enzymes and cellular receptors. During reverse cholesterol transport, HDL interacts with lecithin:cholesteryl acyltransferase (LCAT) and cellular receptors, including ATPbinding cassette transporter protein A-1 (ABCA1) and the scavenger receptor class B type I in an ordered fashion that is reflected by HDL particle lipid composition. The coordinated interaction of specific HDL subpopulations with enzymes and receptors is essential for the directed trafficking of cholesterol in reverse cholesterol transport. This process is recognized as a primary means by which HDL exerts its antiatherogenic character and participates in lipid metabolism. As a result, the plasma concentration of apoA-I is one of the best indicators of susceptibility to cardiovascular disease (1).
When combined with phospholipid vesicles, apoA-I organizes the lipid into discoidal complexes (2). The x-ray crystal structure determination of an N-terminal truncated apoA-I (5) has led to a model of apoA-I on reconstituted HDL (rHDL) wherein apoA-I adopts an extended ␣ helical "belt" conformation that circumscribes the perimeter of the rHDL disc (6). Whereas considerable evidence supports the "belt" model, the structural organization of apoA-I associated with rHDL remains uncertain, and alternative models of apoA-I persist.
The most commonly presented model of apoA-I on rHDL is the "picket fence" conformation, wherein a series of 22 amino acid amphipathic ␣ helices, connected by hairpin loops, align parallel to the fatty acyl chains of the phospholipids in the bilayer disc (3,4) (Fig. 1). This conformational representation of apoA-I has fallen out of favor due to the mounting evidence that apoA-I adopts an extended ␣ helical conformation on the rHDL.
Recently, Maiorano and colleagues (7) proposed that apoA-I can adopt one of two possible extended ␣ helical conformations, including the extended belt model described above or a model wherein the apoA-I molecule loops back on itself, forming a "hairpin" structure ( Fig. 1). The hairpin model was initially proposed by Tricerri et al. (8), based on fluorescence resonance energy transfer (FRET) data, where N-and C-terminal amino acids (residues 9 and 232, respectively) were found to be proximal. Li et al. (9) proposed that apoA-I adopts a belt conformation in which one apoA-I molecule aligns in variable helix registries relative to its paired counterpart on rHDL. These data are consistent with the results of Oda et al. (10), who showed that the C-terminal one-third of apoA-I adopts an extended ␣ helix conformation on rHDL.
Here we apply a combination of site-directed spin label electron paramagnetic resonance spectroscopy (SDSL-EPR) and FRET techniques to examine the structure and relative alignment of apoA-I on discoidal rHDL. The results obtained are consistent with a model wherein paired apoA-I molecules reside on rHDL in an extended ␣ helical conformation aligned in an antiparallel arrangement around the edge of the rHDL disc. Unique EPR spectral features reveal that a central region of the apoA-I sequence constitutes a conformationally adaptable "loop" segment. From this finding, we have formulated a looped belt model for apoA-I structure on rHDL.

EXPERIMENTAL PROCEDURES
Production of Recombinant ApoA-I Protein-Human apoA-I was expressed in bacteria using the pNFXex bacterial expression vector as previously described (11). The pNFXex plasmid was transformed into the Escherichia coli strain BL21 (DE-3) pLysS and cultured in NZCYM medium containing 50 g/ml ampicillin. Expressed proteins were purified via Hi-Trap nickel-chelating columns (GE Biosciences, Inc., Piscataway, NJ) as described (12). Individual Cys substitutions were introduced to the apoA-I cDNA by either primer-directed PCR mutagenesis or by the Mega-Primer PCR method (13). The mutations were verified by dideoxy automated fluorescent sequencing.
Site-directed N-(iodoacetyl)-NЈ-(1-sulfo-5-naphthyl)ethylenediamine (AEDANS) Labeling-8 mg of a given Cys-substituted apoA-I in reducing buffer (20 mM NaPO 4 , 0.5 M NaCl, 3 M guanidine, 0.25 mM dithiothreitol, pH 7.4) was applied to a 1-ml Hi-Trap chelating column, preloaded with 0.1 M NiSO 4 . A 10-fold molar excess of AEDANS (Molecular Probes, Inc., Eugene, OR), with respect to apoA-I concentration, was introduced by passage onto the apoA-I-loaded Hi-Trap chelating column. Following AEDANS labeling, selected apoA-I samples were subjected to circular dichroism analysis to confirm that the secondary structure content of the molecule was unaffected by the presence of the AEDANS moiety.
Preparation of rHDL-Reconstituted nascent HDL were prepared by a modified method originally described by Nichols et al. (14,15). A solution of 22 mM sodium deoxycholate (in 0.5 ml) was added to an equal volume of 16.3 mM 1-palmitoyl-2oleoyl-sn-glycerophosphocholine in Tris-buffered saline, pH 8. The mixture was vortexed and incubated at 37°C until clear. 3 mg of a specified apoA-I was added to the solution (1:2 (w/w) apoA-I:2-oleoyl-sn-glycerophosphocholine), followed by incu-bation at 37°C for 1 h. Sodium deoxycholate was removed by extensive dialysis against Tris-buffered saline, pH 7.4. rHDL were recovered by KBr density gradient ultracentrifugation at 50,000 ϫ g for 3 h in a Beckman Optima TLA 100.4 rotor. Fractions containing both protein and lipid were pooled and dialyzed against Tris-buffered saline, pH 7.4. The size of rHDL was determined by native gradient gel electrophoresis (14). SDS-PAGE was performed using precast 4 -20% polyacrylamide gradient gels (Invitrogen) according to Laemmli (16).
EPR Spectroscopy-EPR measurements were carried out in a JEOL X-band spectrometer fitted with a loop-gap resonator (17,18). An aliquot (10 l) of spin-labeled lipid-free apoA-I or apoA-I rHDL at a final concentration of ϳ5 mg/ml in Trisbuffered saline, pH 7.4, was placed in a sealed quartz capillary contained in the resonator. Spectra of samples were obtained by a single 60-s scan over 100 G at a microwave power of 2 milliwatts and a modulation amplitude optimized to the natural line width (1.5-2.5 G) at room temperature (20 -22°C), as described previously (19). The accessibility of spin-labeled sites to polar (20 mM chromium oxalate) and nonpolar (O 2 in equilibrium with air) relaxation agents was determined from their collisional frequency (⌸) with paramagnetic relaxers using power saturation EPR as previously described (20,21). The hydrophobicity contrast parameter was calculated from the equation, ⌽ ϭ ln(⌸ nonpolar /⌸ polar ).
Fluorescence Spectroscopy-Fluorescence measurements were conducted by excitation of samples containing rHDL at 295 nm, and emission spectra were monitored between 300 and 575 nm on a PerkinElmer LS50B Luminescence Spectrometer (PerkinElmer Life Sciences). Scans were performed on 0.5 mg/ml protein with a 3-nm slit width and scan rate of 30 nm/min. The distance between specified donor (Trp) and acceptor (AEDANS) fluorophores was calculated as described by Selvin (22).

SDSL-EPR Analysis of Putative Helix Junctions in
ApoA-I rHDL-A distinguishing feature of the picket fence model versus the belt and hairpin models is the presence of short loops that connect a series of 22-amino acid ␣ helical segments (Fig.  1A), several of which are punctuated by proline residues (3) (see Fig. 3A). SDSL-EPR experiments were conducted to distinguish whether the conformation adopted by apoA-I on rHDL corresponds to a contiguous elongated ␣ helix or a series of antiparallel ␣ helices connected by reverse turns. The extrinsic nitroxide spin label probe, MTSSL, was introduced at specified locations in apoA-I by creating a series of single Cys substitution apoA-I variants (summarized in Table 1) and covalent attachment via the thiol side chain. Following MTSSL labeling and reconstitution into rHDL, the particles were characterized by native pore-limiting gradient PAGE (Fig. 2). The data indicate that rHDL generated with different MTSSL-labeled Cys substitution variants were indistinguishable from rHDL prepared using wild type apoA-I in terms of average particle size (ϳ9.4-nm diameter) and heterogeneity.
EPR spectra of rHDL bearing MTSSL-labeled apoA-I, wherein the nitroxide probe was introduced at residue positions spanning amino acids 83-91 (corresponding to the helix 2-helix 3 junction (6)), are shown in Fig. 3B. The targeted sites resemble side chains attached to a backbone of fixed secondary structure (10,23). With the exception of position 87 (Fig. 3B), the spectra suggest that the side chains have significant rotational freedom, consistent with surfaces sites along an ␣ helix.
Analysis of side chain dynamics and chemical accessibility of this (residues 83-91) and other putative helix junctions ( Table  1) was performed to confirm the continuous ␣ helical interpretation of EPR spectral line shapes (24). Because rHDL-associated apoA-I is probably oriented asymmetrically with respect to the rHDL lipid surface, variations in residue chemical accessibility will reflect the periodicity of the secondary structure. For lipid or membrane-associated proteins, the relative accessibility of the nitroxide spin label to polar and nonpolar relaxation agents provides the contrast parameter (⌽), which identifies the hydrophobicity of the side chain environment (10,24). When spin labels attached to sites within the 83-91 sequence of rHDL-bound apoA-I are evaluated for polarity, the amplitude of ⌽ is modulated at periodicity consistent with that of an ␣-helix (Fig. 3C). The more hydrophobic positions (⌽ Ͼ Ϫ0.5) cluster to a common face of the ␣ helix (Fig. 3D), whereas the more hydrophilic residues cluster to the opposing face of the helix. Similar EPR spectra and solvent accessibility patterns were observed at other putative interhelix regions examined, indicative of contiguous ␣ helix secondary structure (data not shown). These results are incongruent with a model wherein rHDLassociated apoA-I is composed of a series of 22-residue length antiparallel ␣ helices (i.e. the picket fence model). On the other hand, nitroxide probes located at residue positions from 139 to 146 (corresponding to the helix 5-helix 6 junction) did not yield EPR spectra consistent with a contiguous ␣ helix (see below).
The Central Region of ApoA-I on rHDL Forms a Loop Segment-Based on the observation that MTSSL-labeled residues located in the helix 5-helix 6 junction (residues 139 -146) are not consistent with ␣ helix, a scanning SDSL-EPR survey was performed that included all residue positions between 130 and 150. This was accomplished in a manner similar to that described above for the putative interhelix segments listed in Table 1. Among the 20 MTSSL-labeled apoA-I variant-bearing rHDL examined by EPR spectroscopy, position 139 displayed unique spectral characteristics, wherein side chain flexibility and solvent exposure implies that residue 139 occupies a distinct structural element (Fig. 4A), such as a hinge or turn. Furthermore, nitroxide moieties located at positions 133 and 146 in apoA-I exhibited broadened spectral features characteristic of magnetically coupled spins within 15 Å of one another (Fig.  4B). Spin coupling between nitroxide moieties labeled at these positions in apoA-I following incorporation into rHDL is apparent when compared with the corresponding EPR spectra of these MTSSL-labeled apoA-I in a lipid-free state. Note the

TABLE 1 Positions examined by EPR
The residue numbers and amino acids (single-letter code) within apoA-I putative helix junctions that were Cys-substituted and MTSSL spin-labeled are indicated in boldface type. These MTSSL-bearing apoA-I variants were formed into rHDL and examined by EPR.

Helix junction
Amino acids examined degree of peak amplitude attenuation and spectral broadening upon lipid association. Given that the respective apoA-I molecules examined possess a single MTSSL moiety and the observed spin-spin interactions indicate an interspin distance of ϳ15 Å, it can be concluded that apoA-I sequence positions 133 and 146 reside in a location that is proximal to the corresponding sequence position in its paired apoA-I counterpart on rHDL. In contrast to the behavior of the MTSSL probe at positions 133 and 146, spin coupling was not detected when positions 132, 145, and 147 were labeled (Fig. 4C). Partial spin cou-  pling occurred when the probe was introduced at position 134, reflecting a greater interspin distance of ϳ17-19 Å. Based on these data, it is conceivable that positions 133 and 146 define the borders of a loop segment in the conformation adopted by apoA-I in these rHDL. When these data are considered in light of models depicting the alignment and orientation of apoA-I on rHDL (Fig. 4D), we noted specific features that distinguish the belt and hairpin models. In the hairpin model, a turn near the center of the molecule results in predominantly intramolecular interactions for apoA-I monomers aligned around the edge of rHDL disc. By contrast, in the belt model, this central turn is absent, and hence, intermolecular interactions predominate between paired apoA-I around the edge of the rHDL (Fig. 4D). To differentiate between these mutually exclusive models, FRET experiments were designed to investigate the relative molecular alignment of apoA-I regions and the intraversus intermolecular interactions between paired apoA-I in rHDL.

Energy Transfer between Engineered Donor and Acceptor
Fluorophores-In FRET experiments described below, Trp residues in apoA-I served as energy donor (emission maximum 330 -350 nm; excitation 295 nm), whereas the extrinsic fluorescence probe, AEDANS, covalently bound to engineered Cys residues in apoA-I, served as energy acceptor (emission maximum, ϳ467 nm; excitation, 336 nm). To limit the number of potential fluorescence donors in apoA-I, a series of apoA-I variants bearing only one Trp (W@50, W@72, W@90, and W@108 apoA-I, respectively) or lacking Trp altogether (W@Ø apoA-I), were generated. To examine intramolecular FRET, six unique apoA-I variants were created bearing a single Trp and a single Cys substitution ( Table 2). In the first experiment, the position of the energy donor (Trp) was fixed at residue 72 in apoA-I, and an AEDANS was introduced at position 210 (W@72:A@210; Fig. 5B, top). The resulting apoA-I rHDL elicited significant FRET, as judged by the fluorescence emission intensity at 470 nm upon excitation of the sample at 295 nm (Fig. 6A). By contrast, when the AEDANS moiety was located at position 133 or 146 in apoA-I, minimal fluorescence emission at 470 nm was observed, indicating that the donor and acceptor fluorophores in the label pairs (W@72: A@133 and W@72:A@146) reside at a distance beyond significant FRET detection (Fig. 6A). Based on this alignment, additional dual labeled apoA-I bearing a single Trp and an AEDANS moiety were engineered, wherein the Trp and AEDANS placement were predicted to display FRET, either in the antiparallel belt or hairpin configuration (Fig. 5A). As was the case for the single Trp:single AEDANS W@72:A@210 apoA-I variant, rHDL prepared using W@50:A@230, W@90:A@190, or W@108:A@170 apoA-I displayed evidence of FRET, as seen by the intensity of the AEDANS emission peak upon excitation at 295 nm (Fig. 6A).  a Single AEDANS-bearing apoA-I were used as controls for the detection of non-FRET, AEDANS excitation. b Single Trp-bearing apoA-I were used to determine parameters of Trp fluorescence at that location (i.e. quantum yield and total fluorescence). c Dual labeled apoA-I (Trp and AEDANS) were used to establish the relative alignment of apoA-I regions and R 0 and R value measurements. The W@72:A@133 and W@72:A@146 samples were used as controls, evaluating the degree of FRET observable in nonaligned positions. d Mixtures of singly labeled apoA-I in the rHDL formulations were used in experiments designed to distinguish between intra-and intermolecular energy transfer. The W@72 ϩ W@Ø :A@133 and W@72 ϩ W@Ø :A@146 were used as controls, evaluating the degree of intermolecular FRET observable in nonaligned positions. e W@Ø apoA-I combined with dual labeled apoA-I in a 5:1 ratio in rHDL formulations were examined to detect only intramolecular FRET.
FRET Distinguishing Intra-from Intermolecular Alignment-Because predicted sites of interaction are similar for a hairpin conformation and an antiparallel belt alignment, these data cannot distinguish between these two models. This was achieved, however, by producing rHDL using a mixed population of apoA-I (a 1:5 ratio of a given single AEDANS-labeled Trp:Cys apoA-I to W@Ø apoA-I). Under these conditions, ϳ98% of rHDL discs generated will bear either no fluorescent apoA-I molecules or one fluorescent apoA-I molecule. In this manner, only intramolecular FRET is possible. Fluorescence spectra of these samples failed to show FRET when excited at 295 nm (Fig. 6B), indicating that the energy transfer observed in Fig. 6A arises from intermolecular interactions at these paired locales. The small amount of FRET observed for rHDL prepared with W@Ø apoA-I and AEDANS-labeled W@50:L230C (W@50:A@230) apoA-I suggests that sequence positions 50 and 230 on the same apoA-I molecule are proximal to one another.
Confirmation of the intermolecular nature of energy transfer observed in this system was obtained by preparing rHDL from a mixture of engineered apoA-I variants that contain a single Trp (with no Cys) and corresponding apoA-I variants that contain an AEDANS-labeled Cys with no Trp. In this experimental design, only intermolecular FRET is possible. rHDL were prepared using the following variant apoA-I combinations: W@72 ϩ W@Ø:A@210 (Fig. 6C), W@90 ϩ W@Ø:A190, W@108 ϩ W@Ø:A@170, W@72 ϩ W@Ø:A@133, and W@72 ϩ W@Ø: A@146 using a 1:5 ratio of single Trp "donor" apoA-I to AEDANS-labeled single Cys "acceptor" apoA-I. As observed with dual labeled apoA-I (Fig. 6A), no significant FRET is observed with rHDL composed of W@72 ϩ W@Ø:A@133 or W@72 ϩ W@Ø:A@146 (Fig. 6C). Each of the remaining apoA-I pairs displayed similar degrees of FRET as was observed with the corresponding donor:acceptor sites using the design of Fig. 6A. Since FRET in rHDL prepared with W@72:A@210, W@90:A@190, and W@108:A@170 apoA-I arise only from intermolecular energy transfer, these samples were used to calculate the distance between the different donor and acceptor fluorophores. The values obtained, which range from 28.8 Å for W@108:A@170 apoA-I to 22.7 Å for W@50:A@230 apoA-I, are summarized in Table 3.

DISCUSSION
In the present investigation, a combination of EPR and FRET experiments were performed to systematically examine the veracity of prevailing models of apoA-I conformation on discoidal rHDL (4,25). SDSL-EPR experiments revealed that, across multiple helix junctions and putative reverse turns predicted by the picket fence model, nearly all residues examined were assigned ␣ helix secondary structure. The predominantly contiguous ␣ helix observed throughout the sequence of rHDLassociated apoA-I effectively eliminates the picket fence model, restricting possible conformational arrangements to the belt and hairpin models. Interestingly, earlier FRET experiments provided support for both the belt (26) and the hairpin (27) models. Davidson and Hilliard (28) attempted to resolve this issue by suggesting that both the belt and hairpin conformations may co-reside on the same rHDL disc. Recent studies utilizing a combination of chemical cross-linking and mass spectrometry (29,30) support an extended belt model, similar  Fig. 5B, top) and reconstituted onto rHDL demonstrated significant energy transfer. B, FRET measurements of rHDL discs composed of dual label apoA-I in a 1:5 ratio with W@Ø apoA-I (outlined in Fig. 5B, middle). FRET in an alternative alignment of apoA-I was examined (C ), wherein energy transfer for W@72, W@72 ϩ W@Ø:A@133, W@72 ϩ W@Ø: A@146, and W@72 ϩ W@Ø:A@210 samples was measured. These rHDL samples were composed of single tryptophan-bearing apoA-I in a 1:5 ratio with AEDANS-labeled apoA-I, with the exception of the W@72 sample, which was entirely composed of W@72.

TABLE 3 Calculated FRET values
Values were obtained as described under "Experimental Procedures" using rHDL composed of apoA-I bearing both a single Trp donor and single AEDANS acceptor moiety. Q D is the quantum yield of the donor Trp in the absence of acceptor, determined using the quantum yield of 0.2 for 30 mM Trp in butanol as reference. E is the fractional energy transfer efficiency; J is the overlap integral; R 0 is the distance at 50% efficiency of energy transfer; R is the calculated distance between donor and acceptor moieties. ApoA-I rHDL Conformation JULY 21, 2006 • VOLUME 281 • NUMBER 29

Label
to that described by Segrest et al. (6), wherein apoA-I monomers align in an antiparallel fashion, centered at residue 129. Interestingly, Silva et al. (29) suggest an alternative alignment, wherein apoA-I assumes a "LL 5/2" belt registry that possesses a hingelike feature composed of amino acids 99 -143, similar to that described earlier by Calabresi et al. (31) on the basis of monoclonal antibody binding studies. By contrast, Li et al. (9) suggested that apoA-I monomers adopt a belt conformation yet can align in alternate helix registries with respect to its paired apoA-I counterpart on rHDL.
Herein SDSL-EPR analysis revealed intermolecular spin coupling between paired apoA-I molecules labeled with a nitroxide probe at either position 133 or position 146. Spin coupling in this instance indicates that these sequence positions reside proximal (Ͻ12 Å) to their counterpart on the 9.4 nm rHDL. Given the degree of spin coupling at position 133 or 146 in conjunction with the absence of spin coupling at adjacent or intervening positions, only one interpretation avails itself, that residues 134 -145 form a looplike structural element. Moreover, the degree of spin coupling at positions 133 and 146 indicates that the intervening loop segment is not the result of a transient out-folding of helical residues from the disc edge but is a stable conformational feature. While intriguing, this result is consistent with both hairpin and belt models. FRET studies based on these EPR results confirm the alignment of apoA-I monomers on rHDL and provide evidence that protein-protein contact is intermolecular, substantiating the belt model and eliminating the hairpin model. When considered together with scanning SDSL-EPR data, we conclude that, with the exception of one region, apoA-I adopts ␣ helical secondary structure throughout the length of its sequence.
An interesting concept emerging from the SDSL-EPR studies relates to the putative loop segment in the central region of apoA-I when associated with 9.4-nm rHDL. This conforma-tional feature encompasses residues 133-146 (Fig. 7A).
Recently, an x-ray crystal structure has been reported for fulllength apoA-I at 2.4 Å resolution (32). The structure reveals that lipid-free apoA-I is composed of an N-terminal four-helix bundle and two C-terminal helices. These authors proposed a model for the transformation of apoA-I from a lipid-free to a lipid-associated state that is presumed to be an extended ␣ helix. Among the interesting structural features of the fourhelix bundle conformation is the conspicuous presence of a solvent-exposed loop segment at one end of the bundle that connects helix C and helix D. This loop segment corresponds to residues 137-146, precisely the region of lipid-associated apoA-I that we have independently identified as a flexible loop segment. Thus, it is conceivable that, following interaction with lipid surface binding sites, the helix bundle opens with elongation of the molecule into an extended helical conformation. At the same time, it seems plausible that, depending on the lipid environment, the exposed loop segment between residues 137 and 146 may be retained in the lipid-bound state.
The presence of such a loop segment in lipid associated apoA-I would have the effect of bringing residues 143-165 of one apoA-I into proximity with its paired counterpart on a 9.4-nm rHDL (Fig. 7B). Interestingly, this conformation was suggested as a potential arrangement for apoA-I on rHDL (33). Because residues in this region have been implicated in LCAT activation (34 -41), it is conceivable that loop segment conformational adaptability could serve to modulate LCAT activity. For example, as seen in Fig. 7B, it is possible that the size of the loop segment correlates with rHDL diameter. Thus, the loop segment would be larger on a 7.4-nm particle than on a 9.4-nm rHDL and, possibly, nonexistent on larger rHDL containing two apoA-I (10.5-nm particle) (42). From a geometrical standpoint, the existence of a central loop could contribute to the known conformational adaptability of apoA-I on rHDL. Fur- , where the central loop is larger (7.2-nm disc) or nonexistent (10.5-nm disc) (33) relative to a 9.4-nm disc. The relative scale of rHDL discs is indicated (blue circles), and the corresponding number of residues required to circumscribe the rHDL perimeter is shown based on the assumption of 1.5 Å of helical rise per residue. The residues implicated in LCAT activation are indicated in red (residues 143-165).
thermore, centering the loop segment over the nexus of apoA-I alignment allows for conformational adaptation through changes in loop size with minimal disruption of established intermolecular contacts.
The variability of apoA-I structure in this region may also explain the effect of the apoA-I Milano (R173C) mutation on rHDL disc size (43,44), which is restricted in size to 7.8-nm particles and is a poor activator of LCAT (45). The disulfide bridge formed between paired apoA-I at position 173 may lock apoA-I monomers in an alternative alignment (25) wherein the disruption of the protein-protein contacts necessary for adaptive changes in the central loop would create an insurmountable barrier. This hypothesis is supported by the observations of Tian and Jonas (46), wherein introduction of a Cys at position 124 results in a restricted rHDL population composed of 7.8and 8.6-nm diameter particles. On the other hand, introduction of a Cys at position 232 had no such effect on rHDL particle size.
In the looped belt model described herein, a disulfide bond formed at position 124 would lead to a significant realignment of paired apoA-I monomers on rHDL, whereas position 232 is predicted to reside in a proximal alignment and, hence, would impose little constraint on apoA-I monomer registry.
Curtiss et al. (47) have proposed that cholesterol efflux to HDL is facilitated by the release of lipid-free or lipid-poor apoA-I from spherical HDL. The inability of apoA-I Milano to adapt to changes in HDL composition may be a significant contributing factor in the reported antiatherogenic character of this mutant, since it may predispose apoA-I Milano to early release from HDL in the subintimal space. The release of lipidfree apoA-I is an important intermediary in the transfer of cholesterol from cellular membranes to the circulating pool of HDL, since ABCA1-mediated cholesterol transport is specific to lipid-poor apoA-I. A shorter cycling period for apoA-I Milano between lipid-associated and lipid-poor states may enhance the process of cholesterol efflux.
In conclusion, we have arrived at a model for apoA-I on 9.4-nm rHDL discs wherein apoA-I adopts a looped beltlike conformation. By correlating apoA-I structure with the biological properties of various sized HDL, we will gain a more complete understanding of how apoA-I structurally adapts to changes in HDL composition and size and how these changes are manifested in the modulation of HDL function.