Structure of human plasma low-density lipoproteins: Molecular

Transcription

1 Pro. Natl. Aad. Si. USA Vol. 74, No. 3, pp , Marh 1977 Biophysis Struture of human plasma low-density lipoproteins: Moleular organization of the entral ore (small-angle x-ray sattering/holesterol ester organization/thermal transition) DAVID ATKINSON, RIHARD J. DEKELBAUM*, DONALD M. SMALL, AND G. GRAHAM SHIPLEY Biophysis Division, Department of Mediine, Boston University Shool of Mediine, Boston, Massahusetts 2118 ommuniated by F. A. otton, January 6, 1977 ABSTRAT Human plasma low density lipoprotein (LDL) exhibits a thermal transition over the temperature range 2 4. This transition is assoiated with a strutural hange within the lipoprotein partile and is refleted in the small-angle x-ray sattering profiles from LDL. The sattering profile of the quasispherial LDL partile at 1-shows a relatively intense maximum at ia-' whih is absent from the sattering of LDL at 45. Theoretial alulations, using model eletron density distributions, have been arried out to desribe the paking of holesterol esters in the ore of LDL. At 1, a radially oriented arrangement of the holesterol esters, based on perturbations of the moleular paking of rystalline holesteryl myristate, adequately reprodues the high relative intensity of the x-ray sattering maximum at Yh A-1. The perturbations of the paking in the rystal struture of holesteryl myristate involve "melting" of the hydroarbon hains of the esters together with translations of pairs of moleules parallel to the moleular long axis. The interation of opposing steroid moieties, with 18 and 19 angular methyl groups interloked, exhibited in the rystal struture is retained in the perturbed arrangement. At 45, thermally indued disorder of this arrangement averages the eletron density of the entral ore. The x-ray sattering profiles of partiles with a homogeneous eletron density in the ore region do not show a high relative intensity of the subsidiary maxima in the IAa A-' region, in agreement with experimental observation. The results of these alulations support the onept that the thermal transition observed for LDL is due to a smeti -- disordered transition of the holesterol esters in the ore of the LDL partile. A knowledge of the role played by the various lipid and protein onstituents of the plasma lipoproteins in determining the overall strutural organization is a prerequisite to a full understanding of the funtion of these maromoleular assemblies. Small-angle x-ray sattering has been used extensively to study both the high density (1-7) and low density (LDL) (5-7) lasses of plasma lipoproteins and, in general terms, a ommon moleular arhiteture for the two systems has been demonstrated. Plasma lipoproteins are quasispherial partiles in whih the neutral lipid is sequestered in a entral ore whose surfae is stabilized by phospholipid polar groups and proteins that are in ontat with the aqueous environment. However, the preise arrangement of the neutral lipid (predominantly holesterol esters) within the ore of both high density and low density lipoproteins has been open to speulation. Dekelbaum et al. (8) have demonstrated a thermal transition in LDL between 2 and 45 whih is assoiated with a strutural hange within the lipoprotein partile. hanges in the x-ray sattering profiles from LDL at 1 and 45 led these authors to postulate that the holesterol esters in the entral ore of LDL undergo an order-disorder transition similar to the smeti - isotropi phase hange exhibited by isolated holesterol esters. Abbreviation: LDL, low density lipoprotein. * Present address: Hadassah University Hospital, Jerusalem, Israel. 142 In this ommuniation, we desribe theoretial alulations based on model eletron density distributions aimed at desribing the ore organization of LDL, with partiular referene to the moleular arrangement of the holesterol esters. MATERIALS AND METHODS LDL from the plasma of fasting male subjets was isolated by ultraentrifugation as desribed (8, 9). X-ray sattering profiles of onentrated (15-25 mg/ml) preparations of LDL were reorded using a fousing amera with Franks double mirror optis (1) and Ni filtered uk, radiation from an Elliot GX-6 rotating anode generator. Speimens were sealed in 1-mm diameter apillary tubes. X-ray films were examined with a Joye Loebl model IIIS mirodensitometer. The sattering from LDL preparations was orreted for bakground sattering by subtrating the sattering of the sample apillary tube filled with aqueous solvent. Before subtration the sample and bakground sattering profiles were normalized at an angle orresponding to five times the angle where useful sattering data from the LDL preparation eased to be observed. RESULTS Experimental analysis The small-angle x-ray sattering profiles obtained from LDL at 1 and 45' are shown in Fig. 1. The urve from LDL at 1 (Fig. 1, urve a) is essentially idential to the urves reported by Tardieu et al. (6) and Laggner et al. (7). The series of sattering maxima at angles orresponding to s <Yo A'-I are typial of the sattering from a quasispherial partile. The intense band at s -zii A'- observed at 1 is absent from the sattering profile from LDL at 45 (Fig. 1, urve b). The maxima in the sattering at s < 1/o A-i are, however, essentially idential at the two temperatures, demonstrating that the spherial morphology of LDL remains intat at 45. holesterol esters isolated from LDL exhibit a smeti phase haraterized at low angle by a single sharp diffration maximum at 'A6 A-1. This sharp diffration maximum is not observed at higher temperatures where the holesteri and isotropi phases are present. This parallel with the behavior of the maximum at 1A6 A-1 in the sattering from LDL led Dekelbaum et al. (8) to suggest that the holesterol esters in LDL may undergo a similar smeti to isotropi phase transition. However, the preise origin of the IAa A-' maximum in the x-ray sattering of LDL is unertain. This band may arise from either separate diffration effets from organized mirodomains of holesterol ester in the ore of the LDL partile or it may be part of the fringe pattern due to.the spherial morphology "modified" by the internal organization. In the latter ase the enhaned relative intensity of this maximum may be related to a radially repeating moleular organization.

2 Biophysis: 1; 'E A.2 31% 11 1.o Atkinson et al. 22ki2 3 (A7') FIG. 1. X-ray sattering urves of LDL in.19 M Nal (ph 8.5) after subtration of the bakground sattering from the solvent and sample apillary tube. s = 2/A sin, where 2 is the sattering angle and X = 1.54 A (uk(,). urve a, 1; urve b, 45. Model analysis We have used models of the moleular paking of the holesterol esters within the ore of LDL to examine the origin of the sattering maximum at %6 A-1. We use as our starting point for the holesterol ester paking, the rystal struture of holesteryl myristate determined by raven and DeTitta (11), appropriately modified to "represent" the major LDL holesterol esters, holesteryl linoleate and holesteryl oleate (9) for whih there is no rystallographi information available. The important features of this struture are the formation of holesteryl myristate moleules into pairs with the fatty aid hains assoiated, and the paking of these moleular pairs into a bilayer organization with the is and 19 methyl groups of opposing steroid nulei interloked (Fig. 2A). Extrapolation of the paking of holesteryl myristate to the longer hain length esters (,8) of LDL is straightforward with this hain-hain assoiation. The addition of four methylene groups to the myristate hain would not alter the hain paking provided an all trans onformation is retained. The presene of is-unsaturated double bonds in the hydroarbon hains of the,8 LDL holesterol esters may, however, modify the rystal paking. The wide-angle diffration pattern (V1o < lid < Y2 A-') from holesterol esters in both the smeti and isotropi phases shows a single broad diffuse band at about 5 A (12), indiating a lak of short-range intermoleular order. For the smeti phase we simulate this thermally indued disorder of the holesterol ester struture by a derease in the length of the hydroarbon region of the moleular pair, assoiated with melting of the hydroarbon hains. The derease was alulated using- the ratio of the fully extended to ontrated length observed for the lipid layer of egg yolk leithin bilayers on passing from the gel to the liquid rystal phase (13). A similar relative derease is onsidered in the length of the short hydroarbon hain at the 17 position on the holesterol. Thus, for example, the fully extended length of the,8 fatty aid hain was redued from 22.5 A to approximately 15 A. With these redutions in the dimensions of the hydroarbon regions, the overall length of the moleular pair of holesterol esters is approximately 38 A (see Fig. 2B), a value lose to the dimension of 36 A determined from the single diffration line observed for LDL holesterol esters in the smeti phase (8). Eletron density alulations for the holesterol ester moleule, based on the moleular volume determined from the ) X- () -S.42., P 38 w i 55 < rnn np F, nn n n nr l jl ULJU U U <-L S X 12 (A-) S X 12 (A-) Pro. Natl. Aad. Si. USA 74 (1977) 143 FIG. 2. A. rystal paking of holesteryl myristate (from ref. 11). Note the assoiation of holesteryl myristate moleules into pairs in a bilayer arrangement with the 18 and 19 methyl groups of the steroid moiety opposed. B. Shemati representation of the proposed thermally perturbed struture. Note that the hydroarbon hains are in, a liquid-like onformation. The moleular pairing and bilayer struture are retained.. Eletron density profile of the moleular paking shown in B. The eletron density is represented by two levels: (i).42 e/a: representative of the steroid nuleus and (ii).29 elk/ representative of liquid hydroarbon hains (two regions). partial speifi volume and the volumes of the hydroarbon hain H, H2, and H3 groups, indiate that at low resolution the moleule may be desribed by two eletron density levels (14, 15). The two liquid hydroarbon hain regions would have an average eletron density of.29 e/a3 idential with that of the hydroarbon region of a liquid rystalline phospholipid bilayer (13). The fused ring region of the moleule has an average eletron density.42 e/a3. These eletron density levels are illustrated in Fig. 2. This eletron density profile was used as a trial model for the organization of the holesterol esters in LDL. We have made the assumption that within the orp of the LDL partile the holesterol esters are radially orientated. Therefore, we have alulated the sattering profiles for spherial partiles with inreasing numbers of repeating holesterol ester bilayers omposed of moleular pairs. Fig. 3 illustrates the radial eletron density distribution of partiles with two and four bilayer radial repeating units together with the FIG. 3. Model radial eletron density profile and alulated x-ray sattering urve for spherial partiles ontaining two and four radially repeating bilayer units of holesterol ester. Note the high relative intensities of the sattering fringes in the angular region.2 < s <.3 A-'. In the model eletron density profile, Ap = p(r) - A, the exess eletron density between the partile [eletron density p(r)] and the solvent (eletron density po). The eletron density peak at the surfae of the partile [radius = A and A ] represents the loation of the phospholipid polar groups and protein.

3 144 Biophysis: Atkinson et al. sattering profiles alulated for these distributions. The eletron density peak loated at the outer surfae of the partile [radius = p (Fig. 3A) and A (Fig. 3B)] is due to the polar lipid head groups and protein. The eletron density (p =.38 e4/ai) of this outer shell was based on that obtained from satteripg studies of high density lipoproteins (2-A). The size hanges of the model partile resulting from an inreasing number of holesterol ester bilayer units are'simply refleted by a hange in angular position of the subsidiary fringes due to the spherial morphology. Inreasing the partile diameter'results in a more ondensed fringe distribution. More important, however, is the observation that 'the sattering'profile of the p~artile is sampled at an angular position orresponding to the bilayei repeat spaing of 38 A. The effet of the radially repeating.strutural organization within the partile is to produe, a-'bid relative intensity in the fringe pattern at the angular region orresponding to the angular region where diffration would be observed from an infinite linearly repeating- struture. Thus, in tihe sattering urve for the model with four bilayer theeighth and ninth fringes have a high relative intensity units, while the sattering urve for the partile with two repeats shows Si strong rqjati've intensity at the fourth and fifth subsidiary, mnaxima. For extremely large numbers of radial repeats resulting in extremely large partiles, the diffration pattern of the linear struture would ultimately be observed (16). T'he model partile with two radial bilayer'units has a radius similar t6 that of the native LDL partile (11 A). Therefore, the sattering urve alulated for this model may be ompared with the experimental urve obtained from LDL at 1 (ompare Figs. 1, urve a and 3A). The omparison shows that, whilst~be sattering pattern for the native partile exhibits a high rejative intensity of only the fifth subsidiary maximum at A6 A-', the relative intensities of both the fourth ('Ao A-'1) and the fifth ('A8 A-') maxima are inreased in the sattering urve for the model partile. This differene demonstrates that, arthough this model for the organization of theholesterol esters miay explain in general terms the origin of the '/16 A-l maximum, the model'does not provide an exat desription of LDL. hispetion of the model together with the density funtion (17) alulated by Fourier transformation of the theoretial sattered -intensity shows that the high relative intensity of the fourth fringe in the alulated sattering urve is due' to eletron density peaks approximately 5 A apart. These eletron density peaks are illustrated in Fig. 2. Therefore, additional perturbations of the struture were soug'ht that would remove these 5 A orrelations, but still leave the organization dominated by the length of the moleular pair. These perturbations involve translation of the holesterol ester pairs parallel to their long axis. Fig. 4Ashows a moleular organization based on the translation of two moleular pairs in unison. An organization of this type allows many of the domiilant features of the moleular paking in thee holesteryl myristate rystal struture to persist in the perturbed arrangement. Apart from retaining the moleular pair organization, the interloking of the bulky s and i9 methyl groups on the steroid moiety (11) is retained. The low-resolution one-dimensional eletron density distribution for this moleular arrangement it illustrated in Fig. 4A. The overlapping region of hydroarbon hain and steroid moieties results in an average eletron density (p =.31 e/a3) higher than that of pure hydroarbon. The overlapping sterol ring regions, however, produe an eletron density distribution that ontains only single peaks with 38 A separation. The sattering urve alulated for this eletron density profile radially orientated in a spherial partile (radius 11 A) Pro. Natl. Aad. Si. USA 74 (1977) is shown in Fig. 4B. We have positioned the inner edge of the first eletron density peak at a radius r = 28 A, a radial distane equivalent to one ontrated moleular length. The paking in this inner shell is undoulte'dly perturbed due to the small radius of urvature; however,' only 7% of the volume of the partile oupied by holesterol esters is ontained in this inner shell. There is good qualitative agreement between the alulated sattering urve and the urve obtained from LDL at 1 (Fig. 1 urve a). B3eause we have sought to reprodue in the model sattering urve the overall features of the sattering from LDL, we have not further refined the model to improve the agreement, for example, by small variations in the positions of the inner two maxima. X-ray sattering studies of high density lipoprotein (2-4)'and also soniated lipid dispersions (16) have shown that the relative intensities of the subsidiary maxima from spherial partiles with an internal strutural organization are partiularly sensitive to the 'eletron density levels desribing the radial struture. Therefore, adjustment of the eletron'density levels' inluding the eletron density of theouter shell ontainirqg the phospholipid polar groups and protein, would improve the agreement in the relative intensities of the subsidiary maxima. The effet of the order disorder transition of an organized arrangement of the holesterol esters at the enter'of the partile may be simulated in. the model. Random translations of the holesterol esters would-average the eletron density of this region (Fig. 4 and D). Replaement of the radially repeating eletron density of the ore with a ontinuous eletron density equal to the average eletron density of the holesterol ester moleule gives the sattering urve shown in Fig. 4D. As an be seen, the fifth subsi4iary maximum has a relative intensity lower than the fourth, inr agreement with the sattering from LDL at 45 in whih the fifth maximum is not observed experimentally (Fig..1, urve b). An additional meas'ure of the agreement between the sattering alulated for the model and that observed experimentally an be obtained by omparing the radial eletron density distributions obtained by Fourier transformation of 11/2 (s). The eletron density profiles alulated from the experimental and theoretial sattering data, terminated at the same sattering angle, are shown in. Fig. 5. Of partiular importane are the eletron density peaks at 3 and 6 A in the distribution for LDL at 1. A'similar eletron density profile has been reported by Laggner et al.(7). These peaks demonstrate experimentally the radial distribution of eletron density in the ore of the partile and are onsistent with a radial arrangement of holesterol ester as shown in Fig. 4A and B. The transform of the sattering data from LDL at 45, however, exhibits only one low peak at about 45 A, indiating that the ore has a less ordered-struture. The region of the eletron density profile for r <2 A is influened strongly by termination effets'due'to the limited angular range of the experimental satteringurve and no strudural onlusions may be made from this region of the eletron density profile. Thus, all the strutural features exhibited by the experimental transform and the hange in the eletron density-distribution between 1 and 45 (Fig. 5) are qualitatively reprodued by the model transforms. DISUSSION The results of these studies provide evidene that the thermal transition of LDL between 2 and 4 is related to haznges in the strutural organization of the entral holesterol ester-rih ore. Furfhermore, the model alulations show that speifi moleular arrangements of the holesterol esters within the

4 Biophysis: Atkinson et al. -A42 9II Pro. Natl. Aad. Si. USA 74 (1977) 145 () ~~w w~~~ (a a'i 3 ) 6- - I -38 A-.1 -.I n, no l: I. I{ 2' o 6 8' 1 12o ol (D) r~~a b2 R >e n ) )._ y1 I S X 12 (A-,) X 12 (A-1) FIG. 4. A. Perturbation of the regular bilayer arrangement of holesterol esters shown in Fig The perturbation involves translation of pairs of holesterol ester moleules whilst retaining the interloking of the s and II) methyl groups of the steroid huleus. The eletron density distribution of this paking is shown below and onsists of eletron-dense peaks about,38 A apart. B. Eletron density profile and alulated x-ray sattering urve for a spherial partile with a radial eletron density distribution orresponding to that shown in A.. Further perturbation of the holesterol ester paking in whih only parallel orientation of the long moleular axes is retained, giving a homogeneous eletron density (p =.31 e/1a). D. Eletron density profile and alulated x-ray sattering urve ftr the spherial partile with homogeneous eletron density in the entral ore. spherial partile will adequately aount for the observed hanges in the sattering profile. In addition, the model. alulations illustrate that a radial organization of the holesterol esters withinthe partile produes features in the sattering that orrelate with diffration effets observed from a similar but planar strutural arrangement of holesterol esters in isolation. Thus, the parallel between the thermal behavior of the sattering maximum at 'A6 A-' in the sattering from LDL and the diffration pattern of the isolated holesterol esters suggests a smeti-like layered organization for the holesterol esters of LDL at 1. The transformation of the linear paking of the smeti phase of holesterol esters in the isolated system to a L radial organization in LDL may be aided by the triglyeride omponent of the lipoprotein, the triglyerides also being loated in the entral ore. The transition temperature of the holesterol esters in LDL hag been shown to be dependent on the ratio of triglyerides to holesterol esters (9). Reent nulear magneti resonane studies (18) have demonstrated that in the smeti phase the steroid moiety of holesterol esters has a greatly redued mobility ompared to the isotropi phase. Similar observations were made for the holesterol esters in LDL below the thermal transition. Above the transition, the holesterol esters in LDL appear more onstrained than in the true isotropi phase, suggesting that the esters within the partile do not attain omplete mobility.

5 146 Biophysis: Atkinson et al. A. II. I I I 41.al.jL-k -% I % * T %, e l a.2a i5 I p.) %, a. ' IZ ) 4) -, x { / FIG. 5. Radial eletron density distribution obtained by Fourier inversion of the sattering data for LDL ompared to the model distribution. A. 1. B. 45. The solid urves represent the experimental distributions for LDL. The step profiles (solid lines) are those for the model shown in Fig. 4. The dashed urve is alulated from the sattering data of the model when terminated at the same angular position as the experimental urve. The Fourier transformation was arried out aording to itp = [p(r) - p] = 2 - s1112(s) sin 27rsr ds r In the experimental data for LDL, the minima in the sattering do not go to zero between maxima as expeted for perfetly spherial partiles. The positions of the maxima were used to interpolate the zeros. This treatment assumes that the observation of nonzero minima is due mainly to polydispersity and only to a small degree to deviation from perfet spherial symmetry. Simple translational disordering of the smeti phase would average the eletron density of the entral ore of LDL and result in the observed hanges in the sattering profile. These onstraints may result from a residual radial orientation of the holesterol esters above the transition temperature (see Fig. 4) Ṫhe proposed arrangement aounts adequately for the sattering urve of LDL and demonstrates the importane of the internal strutural organization in the interpretation of x-ray sattering from plasma lipoprotein partiles. We thank Dr. B. raven for valuable disussions. R.J.D. was a Fellow of the Medial Researh ounil of anada. We thank Mrs. Pat Sullivan for preparation of the manusript. This work was supported by U.S. Publi Health Servie Researh Grant HL and Training Grant GM Shipley, G. G., Atkinson, D. & Sanu, A. M. (1972) "Small angle x-ray sattering of human serum high density lipoprotein," J. Supramol. Strut. 2, Pro. Natl. Aad. Si. USA 74 (1977) 2. Atkinson, D., Davis, M. A. F. & Leslie, R. B. (1974) "The struture of a high density lipoprotein (HDL3) from porine plasma," Pro. R. So. London Ser. B 186, Laggner, P., Muller, K., Kratky, O., Kostner, G. & Holasek, A. (1973) "Studies on the struture of lipoprotein A of human high density lipoprotein HDL3: The spherially averaged eletron density distribution," FEBS Lett. 33, Muller, K., Laggner, P., Kratky, O., Kostner, G., Holasek, A. & latter,. (1974) "X-ray small angle sattering of human plasma high density lipoprotein LpA from HDL2: Appliation of a new evaluation method," FEBS Lett. 4, Luzzati, V., Tardieu, A., Mateu, L. & Stuhrmann, H. B. (1976) "Struture of human serum lipoproteins in solution I. Theory and tehniques of an x-ray sattering approah using solvents of variable density," J. Mol. Biol. 11, Tardieu, A., Mateu, L., Sardet,., Weiss, B., Luzzati, V., Aggerbek, L. & Sanu, A. M. (1976) "Struture of human serum lipoprotein in solution II. Small angle x-ray sattering study of HDL3 and LDL," J. Mol. Biol. 11, Laggner, P., Muller, K. & Kostner, G. (1976) "X-ray small angle sattering on human plasma lipoprotein," J. olloid Interfae Si. 55, Dekelbaum, R. J., Shipley, G. G., Small, D. M., Lees, R. S. & George, P. K. (1975) "Thermal transitions in-human plasma low density lipoproteins," Siene 19, Dekelbaum, R. J., Shipley, G. G. & Small, D. M. (1976) "Struture and interations in human plasma low density lipoproteins," J. Biol. hem., 252, Franks, A. (1958) "Some developments and appliations of mirofous x-ray diffration tehniques," Br. J. Appl. Phys. 9, raven, B. M. & DeTitta, G. T. (1976) "holesteryl myristate: Strutures of the rystalline and mesophases," J. hem. So. Perkin Trans. 2, Small, D. M., Loomis,. R., Janiak, M. J. & Shipley, G. G. (1974) in Liquid rystals and Ordered Fluids, eds. Johnson, J. F. & Porter, R. S. (Plenum, New York), pp Luzzati, V., (1968) in Biologial Membranes, ed. hapman, D. (Aademi Press, London), Vol. 1, pp Rand, R. P. & Luzzati, V. (1968) "X-ray diffration study in water of lipid extrated from human erythroytes, the position of holesterol in the lipid lamellae," Biophys. J. 8, Atkinson, D. (1975) "X-ray studies of the struture of natural and re-assembled plasma high density lipoproteins," Ph.D. Dissertation, ounil for National Aademi Awards, London. 16. Atkinson, D., Hauser, H., Shipley, G. G. & Stubbs, J. M. (1974) "Struture and morphology of phosphatidylserine dispersions," Biohim. Biophys. Ata 339, Beeman, W. W., Kaesberg, P., Anderegg, J. W. & Webb, H. B. (1957) in Handbuh der Physik, ed. Flugge, J. (Springer-Verlag, Berlin), Vol. 32, pp Sears, B., Dekelbaum, R. J., Janiak, M. J., Shipley, G. G. & Small, D. M. (1976) "Temperature-dependent 13 nulear magneti resonane studies of human serum low density lipoproteins," Biohemistry 15,