Proc. Natl. Acad. Sci. USAVol. 76, No. 11, pp. 5934-5938, November 1979Medical Sciences

Changes in plasma lipoprotein distribution and formation of twounusual particles after heparin-induced lipolysis inhypertriglyceridemic subjects

(very low density lipoproteins/low density lipoproteins/high density lipoproteins/electron microscopy)

TRUDY M. FORTE, RONALD M. KRAUSS, FRANK T. LINDGREN, AND ALEX V. NICHOLSDonner Laboratory, Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720

Communicated by Earl P. Benditt, August 24, 1979

ABSTRACI Lipoprotein morphology and distribution werestudied in three moderately to severely hypertriglyceridemicpatients after heparin-induced lipolysis. Lipoproteins with aflotation rate of S; 12-20 in p 1.063-g/ml NaCI solution in-creased substantially in mass 2.5-10 min after heparin admin-istration. This fraction contained 40- to 120-nm flattened par-ticles and increased concentrations of phospholipid and freecholesterol. At 2.5 and 10 min after heparin, the high densitylipoproteins (HDL) of p, 1.126-1.21 g/ml contained small (5.8-nmdiameter) spherical particles. Both types of particles disap-peared 30-60 min after heparin. Results indicated that withlipolysis HDL3 may be transformed into HDL2a by incorpora-tion of chylomicron and very low density lipoprotein constitu-ents. It is suggested that the 40- to 120-nm particles representsurface fragments of chylomicrons and very low density lipo-proteins generated during lipolysis, whereas the 5.8-nm particlesare produced either by fragmentation of the large surface con-stituents or by loss of lipid from existing HDL3.

Triglyceride-rich lipoproteins are cleared in vivo primarilythrough the action of lipoprotein lipase (EC 3.1.1.34). Thisenzyme is released from endothelial surfaces by intravenousadministration of heparin (1-4). In vivo and in vitro studieshave shown that chylomicrons and very low density lipoproteins(VLDL) exposed to lipolytic activity lose core material andbecome smaller and more dense. As triglyceride is removedfrom the particles a cholesteryl ester-rich remnant is produced(5); this remnant is thought to be cleared by the liver. Reductionof the chylomicron and VLDL cores would produce an excessof surface components-i.e., apoprotein, phospholipid, andunesterified cholesterol. Blanchette-Mackie and Scow (6) havedemonstrated the lamellar nature of this surface material afterexposure of chylomicrons to lipase activity. Recently, Chajekand Eisenberg (7) described constituents released from ratVLDL perfused through rat heart preparations; the surfacematerial was isolated at a density of 1.04-1.21 g/ml and wascomposed of apolipoprotein C, phospholipid, and unesterifiedcholesterol. Morphologically, the surface constituents weredisc-like particles.The present investigation was carried out to determine

changes in lipoprotein morphology and ultracentrifugal dis-tribution during the early stages of heparin-induced lipolysisin hypertriglyceridemic subjects. Electron microscopy was usedto determine the structure of the major lipoprotein sub-classes.

EXPERIMENTALSubjects. Three male subjects with moderate to severe hy-

pertriglyceridemia were selected for study (Table 1). G.E. hadhypertension controlled with hydrochlorothiazide and meth-yldopa, and H.G. had maturity-onset type diabetes mellituscontrolled with insulin. None of the subjects had clinical orlaboratory evidence of thyroid, hepatic, or renal disease.Heparin Studies. Blood samples were taken after an over-

night fast, with omission of usual morning medication. Heparin,10 units/kg, was administered intravenously and additionalblood samples were withdrawn at 2.5, 10, 30, and 60 min afterheparin administration. Blood samples were placed in ice anderythrocytes were sedimented at 800 X g for 30 min at 4°C.Plasma samples were stored at 40C up to 1 week before pre-parative and analytic ultracentrifugation.

Lipoprotein lipase and hepatic lipase activities were mea-sured in the 10-min postheparin samples; these were within thenormal range (9).

Lipoprotein Preparation. Chylomicrons of Sf > 500 (flo-tation rate, expressed in Svedberg units, in a NaCl solution ofdensity 1.063 g/ml) were removed from the serum of subjectsG.C. and H.G. by a preliminary ultracentrifugation in aBeckman SW 25.3 swinging-bucket rotor at 20,000 rpm for 96min. Lipoproteins of density

Proc. Natl. Acad. Sci. USA 76 (1979) 5935

Table 1. Clinical information on subjectsSerum lipid, Lipo-

Age, Height, Weight, mg/dl proteinSubject yr m kg TG Cho pattern*

G.E. 55 1.87 99.3 408 300 Type 4G.C. 33 1.83 109.3 981 277 Type 5H.G. 37 1.83 116.1 3703 781 Type 5

TG, triglyceride; Cho, total cholesterol.* Determined by quantitative agarose electrophoresis of lipoproteins

(8).

Electron Microscopy. Lipoprotein fractions were negativelystained with 1% sodium phosphotungstate, pH 7.4, and wereexamined with the JEM 100C electron microscope (JEOL,Tokyo, Japan) at instrumental magnifications between X40,000and X80,000. Particle size was determined by measuring100-150 freestanding particles.

RESULTSAdministration of heparin resulted in a rapid decrease in plasmatriglyceride levels, with maximal decreases of 29%, 47%, and87% observed at 30 min for subjects H.G., G.C., and G.E., re-spectively. These decreases were paralleled by decreases in li-poproteins of So 100-400 and concomitant increases in lipo-proteins of slower flotation rates, particularly those of S' 0-20(LDL) and F0.o 0-20 (HDL) (F1.20 is the flotation rate, ex-pressed in Svedberg units, in a NaBr/NaCl solution of density1.20 g/ml). Comparisons of LDL mass distribution in Fig. 1indicate that the major changes in LDL mass in all three subjectsoccurred in the S* 12-20 region, which showed substantial in-creases during the first 10 min. Between 30 and 60 min afterheparin, So 12-20 lipoprotein concentrations decreased to ap-proximately the preheparin levels. Changes in concentrationof lipoproteins of Sf 0-12 were smaller and more variable (Fig.1B), and these lipoproteins returned to preheparin levels within60 min.

Heparin-induced lipolysis affected both mass and flotationrate of HDL (F1.20 0-20). In subjects G.E. and G.C. (Table 2)there were increases in total HDL concentration that appearedwithin 2.5 min and were maximal at 10 min after heparin. Insubject H.G., the preheparin HDL concentration was extremelylow; unlike the situation in other subjects, total HDL mass de-creased after 2.5 min and then progressively increased toachieve preheparin levels at 30 min. The changes in lipoprotein

. 600A B

550

.160EC500

0140

812

&_

450

FIG12. Chne ncnetain fliortnsf 02be

C. 0

H.G.~~~~~~~~~~~a

-200

EE-80E~~~~~~~~CqI150

CYJ60

j.40~~~~~~~020

0 020 40 60 0 20 40 60

Time after heparin, min

FIG. 1. Changes in concentrations of lipoproteins of 5* 0-20 be-fore and after heparin administration as determined by analytic ul-

tracentrifugation. (A) Changes in lipoproteins of 5; 12-20. (B)Changes in lipoproteins of 5' 0-12. Subjects: E, G.E.; A, G.C.; 0,H.G.

Table 2. Analytic ultracentrifugal changes after heparinadministration

Time after F1.20 of HDL, mg/dlheparin, major Total

Subject min peak HDL HDL3 HDL2a HDL2b

G.E. 0 1.6 256 201 40 152.5 2.7 294 89 201 4

10.0 2.8 293 84 204 530.0 1.7 280 222 58 060.0 1.7 216 172 44 0

G.C. 0 1.8 228 168 61 02.5 2.7 281 98 165 18

10.0 2.3 332 81 208 4430.0 2.1 276 140 129 760.0 2.1 253 138 114 1

H.G. 0 1.9 91 53 38 02.5 1.7 19 19 0 0

10.0 1.5 35 35 0 030.0 1.6 97 77 20 060.0 1.7 113 76 37 0

mass were accompanied by parallel changes in peak HDLflotation rates (Table 2).Changes in mass and flotation properties of the HDL after

heparin were evaluated by using the method of Anderson etal. (16) whereby the analytic ultracentrifuge pattern is resolvedinto three components, HDL3, HDL2, and HDL2b. By thisanalysis, a reduction in concentration of the slower floatingcomponent (HDL3) was apparent at 2.5 and 10 min afterheparin in all subjects (Table 2). The increment in HDL massobserved in G.E. and G.C. during this time interval had theflotation characteristics of HDL2a, whereas very little lipo-protein mass was found in the fastest floating component,equivalent to HDL2b. In H.G., there was a decrease rather thanan increase in the region corresponding to HDL2a. By 30 minafter heparin, levels of HDL3 had returned to, or surpassed, thepreheparin values and HDL2a levels were also approachingbaseline.The morphology of the patients' lipoproteins was analyzed

in order to determine whether structural changes had occurredin lipoproteins of p 1.006-1.063 g/ml and p 1.063-1.21 g/ml.Electron microscopy revealed that the preheparin p 1.006-1.063 g/ml fraction of all patients contained round particles,19-29 nm in diameter (Fig. 2A). After heparin administration,two types of particles were seen (Fig. 2B); one resembled pre-heparin particles and the other consisted of large,40- to 120-nm,flattened structures. These latter structures were numerous andappeared by 2.5 min after heparin; however, their numbersdeclined by 60 min after heparin. The large structures wereenriched in the p 1.006-1.019 g/ml fraction (Fig. 2C) butconstituted only a small fraction of the particles of p 1.019-1.063 g/ml. Their appearance in the p 1.006-1.019 g/mlfraction together with the observed decrease in So > 100 ma-terial suggested that these structures might have originatedfrom the surfaces of chylomicrons or large VLDL. The chylo-micron fraction was examined by electron microscopy andsurface alterations were in fact noted. The large triglyceride-rich particles were smooth and round before heparin treatmentbut after heparin they possessed lamellar structures, 5 nm thick,at their periphery (Fig. 2D).

Electron microscopic examination of lipoproteins in sixdensity subfractions of preheparin HDL showed decreasingparticle size with increasing density. The mAjor morphologicalchanges after heparin occurred in the top fraction of p1.063-1.102 g/ml, and in the bottom fraction of p 1.126-1.21

Medical Sciences: Forte et al.

Dow

nloa

ded

by g

uest

on

July

6, 2

021

5936 Medical Sciences: Forte et al.

II

FIG. 2. Electron micrographs of negatively stained lipoproteins before and after heparin administration. (A) Density 1.006-1.063 g/mlfraction from preheparin sample. Freestanding particles are round with diameters ranging from 19 to 29 nm; contiguous particles are generallypolygonal. (B) Density 1.006-1.063 g/ml fraction 10 min after heparin treatment. Two types of particles are evident; one similar in size and shapeto those in A and the other (arrows) consisting of large (40- to 120-nm) flattened structures. (C) Density 1.006-1.019 g/ml subfraction 10min after heparin shows enrichment of flattened structures in this density range. The small particles are 25-29 nm in diameter and they probablyrepresent intermediate density lipoproteins. (D) Chylomicron-like particles isolated in S; > 500 fraction 10 min after heparin. Note lamellarstructures on the surface of these large particles. Bar markers represent 100 nm.

g/ml. The preheparin top fraction contained spherical particles10.3 + 1.0 nm (mean + SD) in diameter (Fig. 3A). After hep-arin, LDL-like particles appeared in the top fraction (Fig. 3B);however, smaller spherical particles 9.4 + 1.2 nm in diameter

were also present and predominated. HDL particles isolatedin the bottom fraction of 1.126-1.21 g/ml before heparintreatment were homogeneous round structures 8.2 4 0.8 nmin diameter (Fig. 3C). After lipolysis, particles 5.8 4 0.5 nm in

FIG. 3. Electron micrographs of negatively stained HDL fractions from subject G.C. (A) Top fraction (p 1.063-1.102 g/ml) from preheparinpreparation. HDL particles are uniform round particles approximately 10.3 nm in diameter. (B) Top fraction, 10 min after heparin. Two typesof particles are clearly visible: small ones approximately 9.4 nm in diameter and large ones 17-19 nm in diameter. (C) Bottom fraction (p 1.126-1.21g/ml) before heparin. This fraction contains 8.2-nm round particles. (D) Bottom fraction, 10 min after heparin; lipoproteins are extremely smallparticles 5.8 nm in diameter. Bar marker represents 100 nm and applies to all micrographs.

Proc. Natl. Acad. Sci. USA 76 (1979)

;,I"

Dow

nloa

ded

by g

uest

on

July

6, 2

021

Proc. Natl. Acad. Sci. USA 76 (1979) 5937

Table 3. Composition and mass changes in lipoproteins from subject G.C. after heparin administrationDensity Time after Proteintnd lipid, mg/mI % total lipoprotein weightfraction, heparin, Free Cho Free Chog/ml min Protein PL Cho ester TG Protein PL Cho ester TG

1.006-1.019 0 0.48 0.52 0.16 0.92 0.71 17.1 18.7 5.7 33.1 25.310 1.72 5.30 2.91 3.39 2.39 10.9 33.8 18.5 21.6 15.2

1.019-1.063 0 5.00 3.47 0.80 5.90 2.70 27.8 19.3 4.4 32.8 15.010 10.70 5.89 2.13 8.90 2.63 35.4 19.5 7.0 29.4 8.7

1.063-1.102 0 0.62 0.67 0.04 0.28 0.40 30.8 33.3 2.2 13.8 19.910 3.14 2.16 0.20 0.45 0.19 51.0 35.1 3.3 7.4 3.2

1.102-1.126 0 1.99 1.59 0.08 0.46 0.78 40.5 32.4 1.7 9.5 15.910 3.63 1.68 0.16 0.28 0.07 62.2 28.9 2.8 4.9 1.2

1.126-1.210 0 1.49 0.56 0.02 0.18 0.20 61.0 22.8 0.9 7.1 8.110 0.85 0.27 0.03 0.04 0 71.4 22.4 2.5 3.5 0

PL, phospholipid; free Cho, free cholesterol; Cho ester, cholesteryl ester; TG, triglyceride.

diameter appeared in this fraction (Fig. 3D); moreover, insubject H.G., these small lipoprotein particles were readilyidentified in the unfractionated postheparin HDL.

All subjects showed similar changes in composition of lipo-proteins of p 1.006-1.063 g/ml and p 1.063-1.21 g/ml afterheparin treatment. Table 3 summarizes compositional data forthe major lipoprotein fractions from subject G.C. before and10 min after heparin. After 10 min, phospholipid and freecholesterol mass increased 10- and 18-fold, respectively, in thep 1.006-1.019 g/ml fraction, while cholesteryl ester and tri-glyceride mass increased only 3.5-fold.

Lipoproteins of p 1.019-1.063 g/ml prior to heparin ad-ministration possessed relatively high levels of triglyceride (15%of total composition). After heparin treatment, the mass of allsurface components increased substantially (Table 3), whereascholesteryl ester was only slightly elevated and triglyceride massdid not change. As a result there was an apparent decrease inpercentage triglyceride and an increase in percentage of bothfree cholesterol and protein.The HDL top fraction had a significant increase (3- to 5-fold)

in mass of all surface components while the mass of core com-ponents increased only slightly or even diminished 10 min afterheparin (Table 3). The intermediate fraction showed increasesin mass of both protein and free cholesterol; however, corecomponents were significantly decreased. Unlike the top andintermediate fractions, the bottom fraction of p 1.126-1.21g/ml showed a 50% loss of lipoprotein mass. The initial tri-glyceride levels were high and decreased rapidly after heparin:84%, 92%, and 100% for top, intermediate, and bottom frac-tions, respectively (Table 3).

DISCUSSIONAdministration of heparin to hypertriglyceridemic subjectsaffords the opportunity to study changes in serum lipoproteinstructure associated with accelerated lipolysis. Ordinarily thelipolytic enzymes are thought to act at the endothelial surfacerather than intravascularly; however, when released into bloodby heparin, they stimulate lipolysis so that triglyceride hy-drolysis far exceeds that seen during the course of chylomicronor VLDL metabolism under nonpharmacologic conditions.Free fatty acids reach high levels that are likely to exceed thebinding capacity of albumin and may alter lipoprotein structureby detergent or other effects. Also, it is likely that lipolysiscontinued in vitro, albeit at a slower rate, so that the lipoproteinchanges observed after heparin may not have been the sameas those occurring in vivo. In addition, lecithin-cholesterolacyltransferase (EC 2.3.1.43) activity has been reported to be

suppressed for 30 min after heparin (17), and, as describedbelow, this may have affected the early changes seen afterheparin in the present study.

These considerations operate against direct extrapolation ofthe-present findings to in vivo metabolic events. However, theperturbations in lipoprotein structure and distribution producedunder these experimental conditions are of intrinsic.interest andinterpretable within the context of our current understandingof the metabolism of triglyceride-rich lipoproteins.

Both the lipid composition and morphology of lipoproteinsin the p 1.006-1.019 g/ml fraction were altered after lipolysisand at least two types of lipoprotein particles were formed. Onetype consisted of 25- to 29-nm round particles whose appear-ance coincided with a 3.5-fold increase in cholesteryl ester mass;the other consisted of large flattened structures whose ap-pearance coincided with a 10- to 18-fold increase in phospho-lipid and free cholesterol. It is likely that the latter constitutesurface fragments generated by removal of triglyceride fromVLDL and chylomicrons, with a corresponding production ofexcess surface material. Bilayered or lamellar structures withthe thickness of phospholipid-cholesterol bilayers were seenon the surface of large VLDL and chylomicrons, further sup-porting the hypothesis that the 40- to 120-nm structures in theLDL region could represent excess surface material. Particleswith the same morphology and a corresponding elevatedphospholipid and free cholesterol content have been describedin the LDL fraction of patients deficient in lecithin-cholesterolacyltransferase (18). Glomset et al. (18) have suggested thatthese aberrant particles may constitute "surface remnants" fromtriglyceride-rich particles that persist in the plasma as a con-sequence of the lecithin-cholesterol acyltransferase deficiency.In our subjects the time course of appearance and disappearanceof large flattened structures parallels that reported for thesuppression and return of lecithin-cholesterol acyltransferaseactivity after heparin (17).

Heparin-stimulated lipolysis produced a transient increasein the mass of the p 1.019-1.063 g/ml fraction in the first 30min. This finding corroborates the observations of Reardon etal. (19) on an in vitro system in which human VLDL were se-quentially transformed into LDL2 (p 1.019-1.063 g/ml) via anLDL1 (p 1.006-1.019 g/ml) intermediate. The actual mass oftriglyceride in the p 1.019-1.063 g/ml fraction was unchangedafter 10-min lipolysis, suggesting either that LDL2 triglyceridewas not hydrolyzed or that triglyceride hydrolysis was balancedby an influx of triglyceride in "remnants." A similar phenom-enon was noted by Barter and Conner (3) in their study on tri-glyceride transport in heparin-injected subjects.

Medical Sciences: Forte et al.

Dow

nloa

ded

by g

uest

on

July

6, 2

021

5938 Medical Sciences: Forte et al.

Postheparin lipolytic activity appears to have effected adecrease in HDL triglyceride as well as a shift in HDL analyticultracentrifugal distribution. The mechanism of such a shift isspeculative, but transfer of chylomicron lipids to the HDLdensity class during lipolysis has been demonstrated in vivo (20).Patsch et al. (21) have recently reported the production in vitroof HDL-like particles after incubating VLDL with bovine milklipase in the presence of HDL3, and they have suggested thatthis occurs by incorporation of VLDL constituents into HDL3.However, their HDL3 fraction had a F'.21 peak of 4.1, and theproduct after lipolysis ("HDL2") had a peak of 7.4. Their flo-tation rate values are similar to those of HDL2a and HDL2b asdefined by Anderson et al. (16), and thus the HDL transitionthey observed may not correspond to that seen in the presentstudy.The fact that there was no observable increase in faster

floating HDL in the subject with the lowest basal HDL level(H.G.) might be due to an inability of HDL to take up chylo-micron or VLDL lipids or might be due to a redistribution ofHDL components to other density classes after lipolysis.The appearance of 5.8-nm particles in the HDL fraction after

lipolysis has not been described previously. However, particlesof the same dimensions have been seen in the HDL fraction ofpatients deficient in lecithin-cholesterol acyltransferase (22).The 5.8-nm particles generated during lipolysis could arisethrough loss of triglyceride and surface components frompreexisting HDL or, alternatively, they could have been derivedby fragmentation of excess surface constituents removed fromchylornicrons and VLDL. In all three subjects these particleswere enriched in protein, phospholipid, and cholesterol, andin addition preliminary studies on apoprotein distribution in-dicated an increase (2-fold) in relative content of apoA-II inthese small particles. The small HDL particles disappear 30-60min after lipolysis, which is the approximate time frame fordisappearance of the large structures seen in the LDL region.The composition of the small HDL particles is consistent withtheir acting as acceptors of additional phospholipid and freecholesterol. It has already been demonstrated in vitro thatHDL3 can remove phospholipid from liposomes (23, 24). Itremains to be determined whether 5.8-nm particles are in-volved in HDL metabolism in normal subjects.We thank Mr. Robert Nordhausen, Ms. Chris Giotas, and Mr. Jerry

Adamson for their valuable technical help. This work was supportedin part by U.S. Public Health Service Grant HL-18574 from the Na-tional Heart, Lung and Blood Institute, and by the Division of Bio-medical and Environmental Research of the U.S. Department of En-ergy.

1. Robinson, D. S. (1963) Adv. Lipid Res. 1, 133-182.2. Nichols, A. V., Strisower, E. H., Lindgren, F. T., Adamson, G.

L. & Coggiola, E. L. (1968) Clin. Chim. Acta 20,277-283.3. Barter, P. J. & Conner, W. E. (1975) J. Lab. Clin. Med. 85,

260-272.4. Krauss, R. M., Windmueller, H. G., Levy, R. I. & Fredrickson,

D. S. (1973) J. Lipid Res. 14,286-295.5. Redgrave, T. G. (1970) J. Clin. Invest. 49,465-471.6. Blanchette-Mackie, E. J. & Scow, R. 0. (1973) J. Cell Biol. 58,

689-708.7. Chajek, T. & Eisenberg, S. (1978) J. Clin. Invest. 61, 1654-

1665.8. Wong, R. A., Banchero, P. G., Jensen, L. C., Pan, S. S., Adamson,

G. L. & Lindgren, F. T. (1977) J. Lab. Clin. Med. 89, 1341-1348.

9. Krauss, R. M., Levy, R. I & Fredrickson, D. S. (1974) J. Clin.Invest. 59, 1107-1123.

10. Lindgren, F.. T. (1974) in Fundamentals of Lipid Chemistry,eds. Burton, R. M. & Guerra, F. C. (BI-Science Publications Di-vision, Webster Groves, MO), pp. 475-510.

11. Anderson, D. W., Nichols, A. V., Forte, T. M. & Lindgren, F. T.(1977) Biochim. Biophys. Acta 493,55-68.

12. Bucolo, G. & David, H. (1973) Clin. Chem. 19,476-482.13. Allain, C. C., Poon, L. S., Chan, C. S. G., Richmond, W. & Fu,

P. C. (1974) Clin. Chem. 20,470-475.14. Bartlett, G. F. (1959) J. Biol. Chem. 234, 466-468.15. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J.

(1951) J. Biol. Chem. 193,265-275.16. Anderson, D. W., Nichols, A. V., Pan, S. S. & Lindgren, F. T.

(1978) Atherosclerosis 29, 161-179.17. Homma, Y. & Nestel, P. J. (1975) Atherosclerosis 22, 551-

563.18. Glomset, J. A., Nichols, A. V., Norum, K. R., King, W. & Forte,

T. (1973) J. Clin. Invest. 52, 1078-1092.19. Reardon, M. G., Fidge, N. H. & Nestel, P. J. (1976) Artery 2,

540-563.20. Havel, R. J., Kane, J. P. & Kashyap, M. L. (1973) J. Clin. Invest.

52,32-38.21. Patsch, J. R., Gotto, A. M., Jr., Olivecrona, T. & Eisenberg, S.

(1978) Proc. Natl. Acad. Sci. USA 75,4519-4523.22. Norum, K. R., Glomset, J. A., Nichols, A. V., Forte, T., Albers,

J. J., King, W. C.,' Mitch6lI, C. D., Applegate, K. R., Gong, E. L.,Cabana, V. & Gjone, E. (1975) Scand. J. Clin. Lab Invest. 35,31-55.

23. Tall, A. R., Hogan, V., Askinazi, L. & Small, D. M. (1978) Bio-chemistry 17, 322-326.

24. Nichols, A. V., Gong, E. L., Forte, T. M. & Blanche, P. J. (1978)Lipids 13, 943-950.

Proc. Nati. Acad. Sci. USA 76 (1979)D

ownl

oade

d by

gue

st o

n Ju

ly 6

, 202

1