Lipid-bound structure of an apolipoprotein E-derived peptide.

Apolipoprotein (apo) E regulates plasma lipid homeostasis through its ability to interact with the low density lipoprotein (LDL) receptor family. Whereas apoE is not a ligand for receptor binding in buffer alone, interaction with lipid confers receptor recognition properties. To investigate the nature of proposed lipid binding-induced conformational changes in apoE, we employed multidimensional heteronuclear NMR spectroscopy to determine the structure of an LDL receptor-active, 58-residue peptide comprising residues 126-183 of apoE in association with the micelle-forming lipid dodecylphosphocholine (DPC). In the presence of 34 mm DPC the peptide forms a continuous amphipathic helix from Glu131 to Arg178. NMR relaxation studies of DPC-bound apoE-(126-183), in contrast to apoE-(126-183) in the presence of TFE, are consistent with an isotropically tumbling peptide in solution giving a global correlation time of approximately 12.5 ns. These data indicate that the helical peptide is curved and constrained by a lipid micelle consisting of approximately 48 DPC molecules. Although the peptide behaves as if it were tumbling isotropically, spectral density analysis reveals that residues 150-183 have more motional freedom than residues 134-149. These molecular and dynamic features are discussed further to provide insight into the structural basis for the interaction between apoE and the ligand binding repeats of the LDL receptor.

The fundamental importance of apolipoprotein (apo) 1 E in plasma lipoprotein metabolism is illustrated by transgenic and gene disruption experiments in mice. Transgenic animals overexpressing apoE manifest decreased plasma cholesterol levels and are protected against diet-induced atherosclerosis (1) while apoE-null mice display dramatically elevated plasma lipoprotein and cholesterol levels and are highly susceptible to dietinduced atherosclerosis (2,3). When complexed to lipoproteins, apoE mediates whole particle uptake and removal from the circulation via members of the LDL receptor family.
Biophysical studies reveal that apoE is comprised of two structural domains, a 22-kDa N-terminal (NT) domain and a 10-kDa C-terminal domain (4,5). The N-and C-terminal domains are connected by a flexible, unstructured, region encompassing amino acids 191-216 that is susceptible to proteolytic cleavage. Studies conducted with isolated domains reveal that the NT domain contains amino acids responsible for binding to the LDL receptor (6). Several lines of evidence have led to a consensus that localizes the receptor-binding site to residues 136 -150 (7). This region of the protein is rich in basic amino acids, and their proposed role in receptor interactions is consistent with studies demonstrating loss of receptor binding following chemical modification of lysine and arginine residues (8,9). In the absence of lipid, the isolated NT domain is not recognized by the LDL receptor. On the other hand, complexation with phospholipids results in a particle that binds efficiently to the LDL receptor (6). These data indicate that a lipid binding-induced conformational adaptation of apoE, which can be mimicked by the isolated NT domain, is an essential feature of apoE function as a ligand for receptor-mediated endocytosis of plasma lipoproteins.
X-ray crystallography of lipid-free apoE3-NT has yielded high resolution structures (10,11). This domain exists as an elongated globular four-helix bundle. Each ␣-helix segment is amphipathic, orienting its hydrophobic face toward the center of the bundle. The structure of lipid-free apoE3-NT provides a useful starting point for development of models of how the helix bundle alters its structure upon interaction with lipid surfaces to adopt a receptor-active conformation. Weisgraber et al. (12) studied the surface properties of apoE3-NT at the air/water interface on a monolayer balance. These investigators concluded that the protein spreads on the surface to occupy a volume greater than can be accounted for by the globular helix bundle conformation. More recently, NMR spectroscopy studies have provided evidence for a major conformational change in the NT domain upon interaction with lipid (13). In 1994, Weisgraber (7) proposed an "open conformation" model in which the loop segment that connects helix 2 and helix 3 in the bundle functions as a hinge, about which the protein opens to expose a continuous hydrophobic surface. Raussens et al. (14) investigated the structural organization of apoE3-NT in dimyristoylphosphatidylcholine (DMPC) complexes by infrared spectroscopy. These investigators presented a model wherein apoE-NT adopts an open conformation, circumscribing the perimeter of the disc bilayer with its helical axes aligned perpendicular to the fatty acyl chains of DMPC, to adopt a receptor active conformation (6,15). Support for this model has been * This work was supported by Grant HL64159 from the NHLBI, National Institutes of Health, the Protein Engineering Network Centers of Excellence and the Canadian Institutes of Health Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  obtained from studies employing fluorescence resonance energy transfer to evaluate distance relationships between specific sites in the protein as a function of lipid binding (16,17). Lu et al. (18) provided evidence for the conformational opening model by demonstrating that apoE-NT dependent transformation of DMPC bilayer vesicles into disc complexes is abolished when helical segments in the bundle conformation are tethered by disulfide bond engineering.
Using an alternate approach Raussens et al. (19) studied a fragment derived from human apoE. A 58-residue peptide encompassing the receptor binding region of apoE was generated by CNBr cleavage of recombinant apoE3-(1-183), purified and characterized. Far UV CD spectroscopy of the peptide showed that it is unstructured in aqueous solution. Importantly, however, apoE3-(126 -183) efficiently transforms anionic phospholipid vesicles into LDL receptor competent, peptide-lipid complexes. Analysis of these complexes by electron microscopy revealed disc-shaped particles with an average diameter of 13 Ϯ 3 nm. Subsequently, the structure of this receptor-active apoE peptide was determined by NMR experiments conducted in the presence of the lipid mimetic cosolvent trifluoroethanol (TFE) (20). In 50% TFE, apoE-(126 -183) forms a continuous amphipathic ␣-helix over residues Thr 130 -Glu 179 . To extend these findings and investigate the structural organization of apoE-(126 -183) in the presence of lipid, we have determined the structure of apoE-(126 -183) in complex with the micelleforming single acyl chain phospholipid dodecylphosphocholine (DPC). Electrostatic and geometric features of the apoE-(126 -183) DPC-bound structure suggest that apoE binds to the LDL receptor by interacting with more than one of its ligand binding repeats. The results are discussed in terms of structural determinants responsible for apoE conformational adaptability and binding to the LDL receptor family.
NMR Spectroscopy-NMR experiments were performed on ϳ2 mM 15 N-labeled apoE-(126 -183) in the presence of 34 mM DPC-d 38 , in 500 l of 10% H 2 O/90% D 2 O, pH 6.0, containing 0.01% (w/v) NaN 3 and 2 mM 2,2-dimethyl-2-silapentane-5-sulfonate as an internal chemical shift reference. NMR experiments were carried out at 25°C on Varian INOVA 500 MHz and Unity 600 MHz NMR spectrometers. Data were processed using NMRPIPE (21) and analyzed using NMRVIEW (22). Complete 1 H and 15 N spectral assignments of apoE-(126 -183) were obtained using gradient-enhanced three-dimensional 15 N-edited TOCSY and NOESY ( mix , 75 ms) experiments to identify spin systems and inter-residue connectivities as described by Wü thrich (23). Confirmation of side-chain assignment was obtained through the use of threedimensional HNHB and two-dimensional natural abundance 13 C-HSQC spectra. 15 N T 1 , T 2 , and heteronuclear NOE relaxation data were recorded at 25°C on both 500 and 600 MHz spectrometers using the enhanced sensitivity gradient pulse sequences developed by Farrow et al. (24). The T 1 relaxation decay was sampled at 11 time points on each spectrometer: 11 .0 ms on the 600 MHz spectrometer. The exponential decay curves for T 1 and T 2 peak intensities were fit using the in-house program Xcrvfit (inhouse written program; www.pence. ualberta.ca/software). 1 H-15 N NOE values were obtained from the ratio of the peak intensity from proton-saturated and unsaturated spectra. Reduced spectral density mapping was carried out as described in Farrow et al. (25).
Structure Calculation-An ensemble of 147 apoE-(126 -183) structures was computed from the distance and dihedral angle restraints available (Table I) starting with an extended chain using a simulated annealing protocol (26,27) in X-PLOR version 3.851 (28). Inter-proton distance restraints were derived from three-dimensional 15 N-edited NOESY experiments recorded with a mix of 75 and 120 ms. Distances were calibrated according to Slupsky and Sykes (29). backbone dihedral angles were calculated based on measured 3 J HN-H␣ coupling constants in an HNHA experiment (30) and the Karplus equation (31). dihedral angle restraints were obtained from the ratio of the d N␣ (i,i)/ d ␣N (iϪ1,i) in the three-dimensional 15 N-edited NOESY spectrum (32). The statistical values for the 50 lowest energy structures are presented in Table I. Families of structures were extracted from the ensemble of structures using the program NMRCLUST (33). For clustering, the backbone heavy atoms of residues 134 -149 of the 50 structures were superimposed, and clustering was based on residues 133-177.

RESULTS
Preparation and Characterization of ApoE-(126 -183)-As previously shown, apoE-(126 -183) is insoluble in aqueous solution above pH 4, and below pH 4, the peptide is not structured in water. However, in the presence of either lipids, such as dimyristoylphosphatidylglycerol (DMPG), DPC, or lipidmimicking agents, such as TFE, it adopts a high (70 -80%) helical content (19,20). Previously, it was determined that 13 mM DPC is sufficient to induce full structuring of this peptide (19). For NMR experiments, we prepared the sample by resuspending lyophilized 15 N-labeled apoE-(126 -183) in D 2 O/ H 2 O (9:1, v/v). Owing to residual HCl following lyophilization (19), the pH of the solution was around 3.5. To this solution, in a stepwise manner, deuterated DPC was added from a concentrated stock solution in D 2 O. Upon each addition, one-dimensional 1 H and two-dimensional 1 H-15 N HSQC NMR spectra were recorded. Addition of DPC was stopped when the spectra did not show a change between two consecutive additions (data not shown). The final DPC concentration was 34 mM. Following DPC titration, the pH of the sample was adjusted to 6.0 by stepwise addition of 500 mM NaOD in D 2 O.
The secondary structure of apoE-(126 -183) was determined using NMR spectroscopy based upon NOE connectivities, the H ␣ NMR chemical shift index (CSI) (34,35) and the ratio of d N␣ /d ␣N NOEs (32). A summary of these data is illustrated in Fig. 2. Helical secondary structure was defined as previously described (20). Fig. 2 shows that DPC-bound apoE-(126 -183) is composed of a single ␣-helix spanning the sequence from Glu 131 to Glu 179 , with the first five and last four residues unstructured. Fig. 2 also shows evidence of the simultaneous presence of unambiguous d ␣N(i,iϩ2) and d ␣N(i,iϩ4) NOEs for residues Leu 133 -Val 135 , Ala 152 -Asp 154 , and Gln 156 -Arg 158 . According to Wü thrich (23), d ␣N(i,iϩ4) is characteristic of an ␣-helix whereas d ␣N(i,iϩ2) is characteristic of a 3 10 -helix. The simultaneous presence of both NOEs for the same residue may reflect a degree of internal flexibility for these residues. An amide proton secondary shift plot (Fig. 3) shows a periodicity of 3-4 residues from Thr 130 to Arg 178 , typical of a curved amphipathic ␣-helix (36,37). In general, hydrophobic residues tend to have a downfield deviation for this type of structure. ApoE-(126 -183) deviates somewhat from the ideal situation. For example Arg 134 , Arg 145 , and Gln 156 show unexpected downfield deviations characteristic of hydrophobic residues, while Val 135 , Ala 160 , and Val 161 show upfield deviations. In a helical wheel representation of this region of the peptide (data not shown) Arg 134 , Arg 145 , and Gln 156 are located on the hydrophobic face of the helix, close to the interface between the polar and nonpolar sides of the amphipathic helix. Since Arg and Gln have relatively long side-chain structures, the C␤ and C␥ (as well as C␦ for Arg) acyl groups may be considered as apolar motifs, maintaining hydrophobic interactions with other nonpolar residues in this region while the polar terminus of the side-chain is capable of protruding beyond the hydrophobic face of the helix. Ala 160 is located near the polar-nonpolar interface, close to the polar side. This could explain the small upfield deviation observed for this residue. Val 161 is located on the polar face of the helix and on the predicted convex side of the curvature, explaining the unexpected upfield deviation observed for this hydrophobic amino acid, as the amide proton secondary deviation is related to the hydrogen bond length. Two regions of the peptide, around Ala 152 -Asp 153 and Gln 163 -Gly 169 , show a 3-4 residue periodicity but with a lesser intensity and a non-significant amide shift deviation. These regions might indicate differences in helix curvature around these amino acids since secondary structural data indicate these residues are helical. Indeed, previously it was shown with model peptides that a lack of amide proton secondary shift 3-4-residue periodicity is not due to the absence of a helical conformation but, rather, comes from a less curved structure (37).
The three-dimensional structure of apoE-(126 -183) was calculated from NOE and dihedral restraints as described under "Experimental Procedures." An ensemble of 147 structures was computed, none of which contained distance restraint violations greater than 0.2 Å or dihedral angle violations greater than 2°. The 50 structures with the lowest calculated total energy were selected for further consideration. According to PROCHECK-NMR (38), 99.7% of the non-glycine residues have (, ) angles in the most favored or the additionally allowed regions of the Ramachandran plot for these 50 structures (data not shown). The average structure, resulting from superposition of the 50 lowest energy structures, is a long curved helix spanning residues Glu 131 to Arg 178 , according to a DSSP analysis (39). The hydrophobic amino acids are, for the most part, oriented toward the interior of the curvature as expected from the amide secondary chemical shift behavior for an amphipathic helix, and as expected for a helix bound to the surface of a lipid micelle.
Superposition of the final 50 structures along the length of the helix revealed a rather poor convergence. Since the peptide is bound to DPC micelles, at first glance this did not appear to make sense. A relaxation analysis (discussed below) revealed that residues 134 -149 are quite rigid whereas residues 150 -178 appear to show slightly increased internal motions in addition to increased slow motions. Superposition of 15 of the most similar 50 structures over residues 134 -149 reveals a backbone RMSD of 0.55 Ϯ 0.11 Å and a heavy atom RMSD of 1.4 Ϯ 0.11 Å (Fig. 4a). As may be observed, while the entire helix remains intact, the relative position of residues 150 -178 with respect to residues 134 -149 changes. Because of this, we tried to extract conformationally related structure subfamilies using the program NMRCLUST (33). For clustering, the backbone heavy atoms of residues 134 -149 of the 50 structures were superimposed, and clustering was based on residues 133-177. Nine subfamilies were obtained with four structures considered as outliers. Among the subfamilies, the first three contain more than 50% of the structures. Fig. 4b shows the representative structures of the three most populated subfamilies in complex with a modeled DPC micelle composed of 54 DPC molecules (40). While the DPC micelle is only a model and is slightly bigger than what we would expect based on the overall correlation time, the complex gives a conceptual idea of how this peptide may adopt different conformations when bound to DPC. It suggests that the C-terminal portion of the peptide (residues 150 -183) has a lower affinity for the micelle than residues 134 -149. As will be discussed later, this has important implications for the interaction of apoE-(126 -183) with the receptor. The structures in each subfamily are well defined over the entire helical length, with RMSDs about the mean coordinate positions of 2.45 Å for subfamily 1, 2.04 Å for subfamily 2, and 1.94 Å for subfamily 3 for backbone atoms of residues 137-173. However, for the most stable helical region of the peptide (residues 134 -149, see below), the RMSDs about the mean coordinate positions for backbone atoms are as follows: 0.73 Å for subfamily 1, 0.71 Å for subfamily 2, and 0.66 Å for subfamily 3. This shows that the structure is well defined locally. As observed previously in TFE, the structure subfamilies display differing degrees of helix curvature.
Relaxation  3. Amide proton secondary shift plot of apoE3-(126 -183) in presence of 34 mM DPC at pH 6.0. In blue are represented the amino acids with a secondary shift above 0.14 ppm (downfield deviation), and those in red have an upfield deviation below Ϫ0.14 ppm.
(1/T 1 ) and R 2 (1/T 2 ) relaxation rates as well as the heteronuclear NOEs at field strengths of 500 and 600 MHz are shown in Fig. 5. Of the 58 residues in apoE-(126 -183), the three missing and twelve overlapping resonances were not used in the backbone dynamics analysis. In general terms, apoE-(126 -183) bound to an isotropic DPC micelle of ϳ50 molecules/ micelle (aggregation number for DPC is 50 -60, Ref. 41) should result in an overall rotational correlation time of ϳ13 ns, characteristic of an isotropically tumbling molecule of ϳ26 kDa. Consistent with this concept, the average rotational correlation time for DPC bound apoE-(126 -183) was determined to be ϳ12.5 ns. Fig. 5 reveals the measured relaxation characteristics of apoE-(126 -183) bound to DPC micelles. Typically, in a rigid globular protein, the R 1 and R 2 relaxation rates as well as the heteronuclear 1 H-15 N-NOE remain somewhat constant throughout the sequence except for the flexible ends and flexible loop regions. Generally, flexibility is expressed in these relaxation parameters as increased R 1 and decreased R 2 relaxation rates as well as decreased heteronuclear 1 H-15 N-NOE. For apoE-(126 -183) bound to DPC micelles, residues 126 -133 at the N terminus of the peptide as well as residue 182 at the C terminus of the peptide appear rather flexible. This is expected since secondary structural information indicates that the peptide forms a contiguous ␣-helix starting from residue 131 and ending at residue 179 (Fig. 2), with little secondary structural characteristics outside these regions. Interestingly, residues 134 -149 exhibit the lowest R 1 , and the highest R 2 relaxation rates in addition to the highest 1 H-15 N-NOE. Start-ing at residue 150, the relaxation rates start to vary indicating changes in flexibility. These changes are exaggerated in the R 2 /R 1 ratio. This ratio reflects the overall rotational correlation time, assuming the value of R 2 in not dominated by exchange (42,43). Due to a lack of significant field dependence on R 2 , no significant exchange contributions can explain the relaxation characteristics. Notably, changes in the relaxation rates appear to coincide with the appearance of aspartic or glutamic acid residues in the sequence. At the N terminus, residues Glu 131 and Glu 132 act to break up the long-chain hydrophobic amphipathic character of residues 134 -149. Asp 151 , Asp 153 , and Asp 154 appear to accomplish a similar task after residue 149. A small long-chain hydrophobic patch of amino acids (Leu 159 -Val 161 ) appears to increase the R 2 /R 1 ratio slightly, which is subsequently decreased again due to a string of polar and glycine residues (Tyr 162 , Gln 163 , Gly 165 ). Residues Glu 168 and Glu 171 appear to decrease R 2 /R 1 again followed by a slight increase involving residues Leu 174 and Ile 177 . Finally, Glu 178 terminates the helix.
The time scales of motions present may be represented by the calculation of the reduced spectral density functions. Fig. 6 illustrates the reduced spectral density functions J(0), J( N ), and J(0.87 H ) that describe the motion of the H-N bonds derived from the relaxation parameters R 1 , R 2 , and the 1 H- 15   These residues exhibit a J(0.87 H ) consistently lower than 7.5 ps/rad, which indicates a lack of internal flexibility. Not surprisingly, residues 126 -130 and residue 182 exhibit J(0.87 H ) spectral densities greater than 15 ps/rad indicating marked flexibility. Interestingly, residues 150 -180 exhibit J(0.87 H ) spectral densities higher than for residues 134 -149. At 500 MHz, these spectral densities are higher than 7.5 ps/rad. It appears that residues 150 -154 experience slightly elevated J(0.87 H ), which decreases for residues 156 -160. Between residues 164 and 180, J(0.87 H ) increases even more. J(0) data follows the opposite trend for both 500 and 600 MHz data.  (126 -183) is a peptide that includes a lipid binding region and the LDL receptor binding motif of apoE. We have previously demonstrated that this 58-residue peptide conserves some major functions of the NT domain of apoE (19). An earlier study of the DPC bound structure of a 23-residue peptide, apoE-(130 -152), revealed a nearly linear, completely helical conformation that displayed some flexibility (45). Whereas our results agree with the helical character of this region of apoE, we observe a pronounced curvature that, as discussed below, may have functional consequences. Furthermore, although apoE-(130 -152) includes residues 136 -150, it does not contain other residues known to be required for proper apoE-LDL receptor recognition. For instance, it is known that resi-dues 171-183 are critical for receptor binding, as illustrated by studies of truncated variants of apoE. Whereas apoE-(1-183) has nearly full receptor binding activity apoE-(1-174) has only 19% activity with further truncation reducing this activity to 1% (46). Recently, Arg 172 was found to play a key role in receptor binding (47). We previously demonstrated that, when complexed with phospholipid, apoE-(126 -183) binds LDL receptors on the surface of human skin fibroblasts (19). Since apoE-(126 -183) contains all the residues necessary for full receptor binding, it provides a useful working model for studies designed to elucidate the LDL receptor active conformation of apoE. Extending this concept, we determined the structure of apoE-(126 -183) in association with DPC, a zwitterionic single acyl chain phospholipid that possesses a polar head group identical to that of phosphatidylcholine (PC). Unlike PC, however, DPC forms micelles whose size is amenable to high resolution NMR (41). NMR relaxation measurements on DPC have shown that it mimics a PC bilayer environment (48).
Comparison of ApoE-(126 -183) in DPC and TFE-ApoE-(126 -183) bound to DPC forms a curved amphipathic helix from residue Glu 131 to Arg 178 , in agreement with the structure of this peptide in 50% TFE (20). As with the TFE structure, the helix content of the DPC-bound peptide determined by NMR (83%) is higher than that calculated on the basis of far UV CD spectroscopy (70%). While the secondary structure content of apoE-(126 -183) is similar, differences exist with respect to the dynamic behavior of the molecule in TFE versus DPC. In TFE, apoE-(126 -183) exhibits more rigidity in the region between residues 149 and 159. Outside this region, residues 131-143 and 163-176 display a plasticity of motion such that the further away from the central region, the greater the movement. This type of dynamic behavior is consistent with a helix tumbling in solution.
When bound to DPC, however, apoE-(126 -183) behaves more isotropically as might be expected for a peptide bound to the surface of a micelle. R 1 , R 2 , and 1 H-15 N NOE have characteristics that, at first glance, appear to follow conventional relaxation behavior for a globular protein. Residues 134 -149 comprise a more rigid region while residues 150 -179 constitute a more flexible segment of the polypeptide. Between positions 134 and 149 there are six residues with aliphatic side chains (1 Val and 5 Leu). At residue 151, a series of aspartic acid residues (Asp 151 , Asp 153 , and Asp 154 ) appear to disrupt binding of apoE-(126 -183) to the micelle. Between residues 159 and 161, the peptide is a bit less flexible yet increases dramatically in elasticity after Tyr 162 , presumably due to the presence of polar and acidic residues, including Glu 168 and Glu 171 . Flexibility in this region may facilitate receptor interaction by permitting structural adaptation to constraints imposed by individual ligand binding repeats. PhD secondary structure prediction (49) analysis of apoE-(126 -183) yields results that are in agreement with the present NMR structure. PhD predicts two helical segments (Thr 130 -Glu 168 and Arg 171 -Arg 180 ). In the first predicted helix there is a small decrease in the reliability index for residues Leu 149 -Arg 150 . This region and the predicted helix interruption from residues 169 to 170, generally coincide with the present NMR-derived flexibility results.
Curvature and flexibility in this peptide likely facilitate interaction with lipid surfaces such as lipoproteins or phospholipid disc complexes. Consistent with these structural characteristics, intact apoE is known to bind different sized lipoprotein particles, from small HDL to large chylomicron remnants. Similar physical properties have also been observed for other apolipoproteins. For example, NMR structures of apoC-I and apoC-II in a lipid-mimicking environment revealed curved helical regions and/or linker regions between helices with a loosely defined conformation that confers a flexible curvature to the structures (50,51). The X-ray structure of an N-terminal deletion mutant of apoA-I, (⌬1-43)apoA-I, is comprised of a series of curved amphipathic helices (52) that, conceivably, could circumscribe the perimeter of an HDL particle. These examples suggest that the features observed for DPC-bound apoE-(126 -183) may represent typical characteristics of exchangeable apolipoproteins.
Potential Implications of the ApoE-(126 -183) Structure on ApoE-LDL Receptor Interactions-The structure of DPC-bound apoE-(126 -183) has implications with respect to apoE ability to bind to the LDL receptor. For example, Lys 143 and Lys 147 are known to display increased solvent accessibility in the presence of lipid, thereby increasing their electrostatic interaction potential with elements of the LDL receptor (13). This increase in solvent accessibility appears to be a consequence of helix curvature in this region induced by the presence of lipid. Likewise, structuring of residues 165-179 in the presence of lipid most likely facilitates interaction with the LDL receptor since it is known Arg 172 is required for optimal binding (47). In the case of DPC micelles it is evident that residues 134 -149 are anchored at the lipid interface while the region between residues 165 and 179 appears to bind less tightly. It is conceivable that the smaller size and higher radius of curvature of DPC micelles compared with natural lipoproteins affects the ability of apoE-(126 -183) to retain tight contact with the lipid surface. Although residues 165-179 are not as tightly associated with the micelle surface, they are nonetheless helical, suggesting induction of helix in this region of the molecule may be a cooperative event triggered and/or maintained by lipid interaction of another region of the peptide (e.g. residues 134 -149).
Structural characteristics of the LDL receptor have been discussed in a preceding study (20). Here, we briefly outline features that are pertinent to this discussion. The presence of highly conserved acidic residues within ligand binding modules of the receptor has led to the hypothesis that ligand-receptor recognition may be due to electrostatic interactions. Although ligand binding repeat 5 (LR5) was demonstrated to be the most important (53), experiments also suggest a role for other repeats, located upstream of LR5 (53,54). The well characterized receptor binding region of apoE (136 -150) displays a highly positive electrostatic surface potential and thus, may interact with LR5, the most negatively charged ligand binding repeat. Residues 165-178 become structured in the presence of lipid and present another positively charged surface that could interact with a second ligand binding repeat. Glutamic acid residues punctuating the largely positive surface might be important for proper orientation of the peptide at the surface of the receptor, interacting with the few arginine and lysine residues present in almost every ligand binding repeat. Once a ligandbinding repeat, presumably LR5, has recognized the most basic region of apoE (residues 136 -150), adjacent repeat(s) might interact with another region(s) of apoE (notably around Arg 172 ). This interaction may orient the apolipoprotein and, by extension, the entire lipoprotein particle, in a favorable manner with respect to the receptor. When bound to DPC or in TFE (20), apoE-(126 -183) adopts an elongated structure ϳ70 Å in length. By the same token, a single ligand-binding repeat has a maximum diameter of 25 Å. These distances are consistent with the possibility that apoE interacts with more than one repeat at the surface of the receptor in a process that would be facilitated by the structural independence of adjoining ligand binding repeats (55,56). A recently reported crystal structure of the entire extracytosolic domain of the human LDL receptor provides support for this view (57). It is known that, after lipoprotein internalization, at endosomal pH (pH Ͻ 6), the LDL receptor discharges its ligand before recycling to the cell surface. Because this crystal structure was obtained at pH 5.3, it should represent the endosomal conformation of the LDL receptor. The most remarkable feature of this structure is that ligand binding repeats LR4 and LR5 are held in place through interactions with the ␤-propeller motif of the receptor, while other repeats seem to be fixed mainly by intermolecular crystal contacts. The authors suggest that, at endosomal pH, following release of the lipoprotein ligand, the ␤-propeller can serve as an alternative substrate for the two ligand binding repeats, and this might be a pH-regulated, reversible process. These conclusions favor a predominant role for LR4 and LR5 in ligand binding activity of the receptor, in agreement with concepts presented here and elsewhere (20).

FIG. 7. Schellman C-cap motif termination of helix 4 in lipidfree apoE.
Panel A, sequence of human apoE around Ala164, corresponding to the putative C-capping motif terminating helix 4 (Line 1). Line 2 indicates conventional C-capping nomenclature, C 1 , C 2 etc . . . refer to residues within the helix while CЈ, CЉ refer to residues in the loops. Cc refers to the last residue whose ␣ carbon lies approximately along the helical path (60). Line 3 depicts the Schellman motif consensus sequence where h represents a hydrophobic residue, p a polar residue, G a glycine, and X is indifferent. Below the consensus sequence are displayed other structural aspects necessary for the C-cap stabilization, including main-chain hydrogen bonds between residues CЈ and C 2 , and CЉ and C 3 (59). Panel B, alignment of 15 apoE sequences around two regions of interest (see text); asterisk indicates identity, colon indicates strong homology, and dots indicate weaker homology. Panel C, at left is a ribbon representation of the end of the 4-helix bundle (1LPE) of apoE in the region of the 80s loop (11). Backbone atoms in helix 4 have been represented to visualize helix hydrogen bonds (cyan dotted lines). Other important residues (see text) have been displayed in a stick representation. The red dotted line represents the C-capping CЈ-C 2 (Gly 165 -Tyr 162 ) H-bond while the yellow dotted line represents the stabilizing main chain H-bond between Thr 83 and Ala 164 . The right hand side depicts the same structure after 90°rotation along the vertical axis. To better visualize the Tyr 162 side chain, Leu 93 is not displayed, and the ribbon is represented with some transparency.

Potential Molecular Mechanism Involved in Structuration of
Residues 165-178 -The lipid bound structure of apoE-(126 -183) raises questions about why helix 4 does not terminate at the same residue in the presence and absence of lipid? In the crystal structure of apoE3-NT domain, helix 4 ends at Ala 164 (10). The amino acids around this residue, Val-Tyr-Gln-Ala 164 -Gly-Ala correspond to a typical Schellman C-capping terminal motif sequence (Fig. 7), commonly found at the end of ␣-helices (Ref. 58; for a review see Ref. 59). Furthermore, analysis of the x-ray coordinates of apoE3-NT (PDB code: 1LPE; 10) provides evidence for the presence of Schellman motif structural requirements (Fig. 7, panel A). Vadar analysis 2 revealed the possibility of an H-bond between the NH group of Gly 165 and the CO group of Tyr 162 (Fig. 7C) corresponding to the CЈ-C 2 bond in a Schellman motif. Although a CЉ-C 3 H-bond was not detected, Ala 166 (corresponding to CЉ and the last residue observed in the electron density), has a very high B factor suggesting its position and orientation may not be accurately determined. The angle value for the Gly 165 (corresponding to CЈ) is 133°, a positive value, as expected for a left-handed amino acid conformation. The last criterion of a Schellman motif, namely a hydrophobic interaction between CЉ and C 3 (corresponding to Ala 166 and Val 161 ), is more difficult to identify. We used the CSU program (61) to identify potential hydrophobic contacts in the vicinity of Ala 166 and Val 161 . Whereas the side chain of Ala 166 is not oriented correctly to promote hydrophobic interaction between these residues, CSU analysis detected a potential hydrophobic interaction between the side chains of Val 161 and Leu 93 (located at the beginning of helix 3; Fig. 7C). At the same time, Leu 93 makes hydrophobic contact with Ala 166 . Thus, we suspect that the hydrophobic interaction required for stabilization of the C capping motif is provided by a long-range interaction with Leu 93 .
Other interactions seem to exist between the end of helix 4 and (a) residues connecting helices 2 and 3 (the 80s loop; Ref. 11) and (b) residues located at the end of helix 2 and the beginning of helix 3. Tyr 162 forms many hydrophobic interactions through its aromatic ring atoms that orient between helices 2 and 3, locking the end of helix 4 in close proximity to the 80s loop (Fig. 7C). Val 85 has a potential for hydrophobic contact with the side chain of Ala 166 while Thr 83 makes a main chain hydrogen bond with Ala 164 , as detected by Vadar and CSU. This H-bond serves to stabilize the CO group of Ala 164 at the terminus of helix 4. Interestingly, most of the residues described here as involved in stabilizing interactions are conserved among different apoE sequences (Fig. 7B). The only non-conserved residue is Thr 83 . Insofar as Thr 83 is substituted only by amino acids having smaller side-chains, these substitutions are not expected to sterically hinder H-bond formation between the backbone NH group at position 83 and Ala 164 .
The interaction of apoE-NT with lipid is believed to arise by opening the globular 4-helix bundle about a hinge region between helices 2 and 3 (7). It appears that helices 1 and 2 and helices 3 and 4 remain preferentially paired during the first stage of bundle opening (18) with possible subsequent reorganization of helix segments (16 -18). If this view is accurate, then the hinge region, implicated in the first stage of the opening, must include the 80s loop. Opening the bundle in this manner would disrupt interactions between 80s loop residues and the end of helix 4. Subsequent separation of helices 3 and 4 would abolish long-range hydrophobic interactions between Leu 93 and Val 161 , which contribute to stabilization of the Schellman C capping motif. It has been observed that, upon disruption of such hydrophobic interactions (by the presence of a cofactor, for example), helix structure can extend despite the presence of a C-capping consensus sequence or a helix-breaking Gly residue (58).
In summary, we suggest that lipid binding-induced alteration of interactions responsible for termination of helix 4 in lipid-free apoE provides a molecular explanation for the apparent transition of residues 165-178 from unstructured to ␣-helix, thereby conferring LDL receptor recognition properties to the protein. Reorganization of the hinge region is likely to be an early event in apoE NT lipid binding, and one that triggers structure formation within residues 165-178. In addition to its role in receptor recognition, structure formation in this region may be an important adaptation that leads to a favorable orientation of the protein with respect to the lipid surface. This proposal has specific advantages including the following. 1) It accounts for the fact that the region of apoE around Arg 172 adopts a helical structure in the presence of DPC while it is unstructured in absence of lipid. 2) It explains, at the molecular level, how this region can transition from one structure to another (or more exactly to the absence of structure); and 3) it explains why the lipid-bound conformation is required for correct interaction between apoE and the LDL receptor. The lipid surface effectively serves as a molecular switch to modify stabilizing interactions (especially hydrophobic interactions) at one end of the helix bundle, inducing a 15-residue conformational change from unstructured to amphipathic ␣-helix that orients and aligns essential elements of the receptor binding region of this protein.