Uptake of endogenous cholesterol by a synthetic lipoprotein

183

Biochimica et Bioplrysica Acta 875 (1986) 183-194 Eisevier

BBA 52096

Uptake of endogenous cholesterol by a synthetic lipoprotein

Kevin Jon Williams * and Angelo M. Scanu Departments

ofMedicine,Biochemistry,

and Molecular Bioiogv, The University of Chicago, Pritzker School of Medicine Chicago, IL (U.S.A.) (Received

Key words:

Hypercholesterolemia;

Cholesterol

transfer;

June 2&h, 1985)

Liposome

disintegration;

Lipoprotein-lipid

interaction;

(Dog)

The addition of cholesterol-poor phospholipid liposomes to canine plasma in vivo and in vitro substantially alters the distribution of phospholipids, apoproteins, and, especially, cholesterol. In vivo, intravenously injected phospholipid liposomes remain discrete particles, which are readily distinguished from the normally occurring lipoproteins by their buoyant density and electrophoretic mobility. They acquire unesterified cholesterol from endogenous sources, thereby producing an acute rise in the concentration of this sterol in pllasma. The liposomes also accumuiate endogenous proteins, one of which is identified as apolipoprotein A-I. In vitro, p~~~ipid liposomes incubated with plasma acquire unesterified cholesterol and a~li~~tein A-I at the expense of hi~~nsi~ li~protei~ (HDL), the major carrier of cholesterol in normal canine plasma. In exchange, the HDL particles are enriched in phosp~lipi~ and become larger. At sufficiently high concentrations, the liposomes nearly completely deplete HDL of its unesterified cholesterol. Thus, there are generated two types of particles, both rich in apolipoprotein A-I and phospholipid, but one (modified HDL) containing mainly esterified cholesterol in its core and the other (modified liposomes) containing mainly unesterified cholesterol at its surface. It is concluded that phospholipid liposomes produce important changes in the distribution of lipids and protein in canine plasma, particularly at the expense of HDL. These changes appear to favor the mobilization of tissue cholesterol into the plasma, and may have application to atherosclerosis.

In~~uction

Phospholipid liposomes have been shown to interact with lipoproteins in vivo and in vitro, particularly by donating phospholipid to the lipoproteins and acquiring apoproteins in exchange [l-4]. However, little attention has been directed towards possible modifications by liposomes in the transport and metabolism of cholesterol.

* To whom correspondence should be addressed at: Division of Gastroenterology, Department of Medicine, Coiumbia University, College of Physicians and Surgeons, 630 West 168th Street, New York, NY 10032, U.S.A.

000%2760/86/%03.50

0 1986 Elsevier Science Publishers

Several lines of evidence recently reviewed [5] suggest that liposomes alter the way cholesterol is handled. In vitro, liposomes extract cholesterol from cells cultured to retain lipid [5]. The fact that labeled, unesterified cholesterol freely exchanges in vitro between cholesterol-phospholipid liposomes and native lipoproteins suggests that cholesterol-poor liposomes would siphon cholesterol from lipoproteins [5], although net transfer of cholesterol mass from lipoproteins to liposomes has not previously been shown. Liposomal phospholipid can be taken up by pre-existing HDL particles during co-incubation; the phospholipidenriched HDL then accumulate unesterified cholesterol at the expense of very-low-density

B.V. (Biomedical

Division)

IS4

(VLDL) and low-density (LDL) lipoproteins [4]. Although liposomes contain very little cholesteryl ester, they have been shown in vitro to participate in cholesterol esterificatlon and transfer. The surface lipids of apolipoprotein A-I-containing liposomes are suitable substrates for lecithin : cholesterol acyltransferase. Essentially all cholesteryl ester produced by lecithin : cholesterol acyltransferase is rapidly transferred from the liposomes to acceIjtor particles by cholesteryl ester transfer protein, thereby relieving end-product inhibition of lecithin : cholesterol acyltransferase. Fur&her esterification and transfer then occur [6]. In vivo, studies on the metabolism of injected liposomes that contain solutes entrapped within their aqueous cores indicate that liposomes can continue to exist as intact, vesicular particles in the circulation [2,5,7-121. Stable phospholipid vesicles also appear in the circulation following infusion of Intralipid [5.73-161 or other emulsions [17]. The triacylglycerol in the cores of the Intralipid particles is rapidly hydrolyzed, leaving surface phospholipid behind as vesicles [13-l 61: triacylglycerol-free phospholipid vesicles in the infused emulsions may also contribute to the plasma vesicle pool 15,171. These circulating vesicles appear to acquire endogenous cholesterol and thereby become transporters of cholesterol. Other studies in vivo indicate that small amounts of liposomal phospholipid can be incorporated into discoidal particles. which are converted by lecithin : cholesterol acyltransferase into spheres that resemble HDL and that transport cholesterol and cholesteryl ester [ 1,4,18]. Some liposomal phospholipid can also be taken up in vivo by pre-existing HDL particles [2], which may enhance the capacity of this MDL to carry choiesterol [5,19]. The most physiologic evidence that liposomes significantly alter cholesterol metabolism in whole animals is the observation that single infusions of cholesterol-free phospholipid produce a transient hypercholesterolemia [20-231 and that repeated infusions produce regression of experimental atherosclerosis [21,24]. It was recently hypothesized [S] that these two observed effects are caused by the uptake of tissue and lipoprotein cholesterol by apoprotein-containing liposomes in the circulation. Uptake of cholesterol by phospholipid-en-

riched or newly created. non-vesicular lipoproteins may also play a role [5]. However, the actual particle or particles responsible for the acute r&e in blood cholesterol ~oncentrati(3n or the mobilization of atherosclerotic deposits have not been identified. Therefore. we sought to characterize (i) the particle that carries the rise in blood cholesterol concentration that follows single infusions of egg phosphatidylcholine: (ii) the movements of cholesterol between liposomes and native components of blood in vivo and in vitro; and (iii) the fluxes of protein and phospholipid between liposomes and native lipoproteins in vitro. To allow measurement of phospholipid and cholesterol fluxes independent of changes in lipid concentration or rate of cholesteryl ester formation. the experiments in vitro were conducted in the absence of lecithin : cholesterol acyltransferase. ‘The results of these studies. conducted in whole dngs or their plasma, are the subject of this report.

Materials and Methods Prepamtion of ~ipos~t?~ul ~i~~per.~i~~ns of p~l~,~ph~~ipi~~ A commercial preparation of egg yolk phosphatidyl~h~~iine (Cal Biochem. catalogue No. 429405) was chosen with specifications matching the reported composition of the phospholipid component of Intralipid [25]. These specifications (> 6Oq, phosphatidylcholine. > 10% phosphatidylethanolamine. < 4’% lysophosphatidylcholine) were verified by silica gel thin-layer chromatography 1261. Trace amounts of unesterified cholesterol and protein were detected by standard assays 127.281 in proportions of 58.5 : 1.0 : 0.48 (phospholipid/ unesterified cholesterol/ protein, by weight) or 27.8 : 1.f) (phospholipid/unesterified cholesterol, molar ratio). No cholesteryl ester was detected. A 5% (w/v) dispersion of this phospholipid in physiologic saline (0.9%) was prepared using ultrasonic irradiation at 0°C under N, atmosphere for a total of 40 min, followed by ultracentrifugation for 1 h at 100000 X g to remove fragments of titanium shed by the sonicator probe. This procedure was a modification of the method of Huang et al. [29]. with the omission of their final step of molecular sieve chromatography. The 5% disper-

185

sion was then sterilized by passage through a 0.22 pm Millex-GS filter. This procedure for dispersing and sterilizing phospholipid produced no degradation detectable by thin-layer chromatography. Dispersions were always used within 24 h. Experiments in vivo Two 16-kg dogs (one male, one female) were fasted overnight then injected intravenously over 5 min with 200 mg phospholipid per kg body weight. This dose was chosen for three reasons. First, it is within the range of daily doses of phospholipid commonly administered to patients during parenteral feeding with lipid emulsions [5]. Second, it is comparable to the daily doses fo phospholipid administered intravenously to animals during treatment of experimental atherosclerosis [5,21,24]. Finally, the trace contaminants of cholesterol in this dose would only trivially affect plasma cholesterol concentration (by, at most, 0.174 mmol/l). Blood was drawn from the dogs just before (t = 0 h) and 1, 2, 4, 8 and 25 h after the injection of phospholipid. Selected samples of plasma were subjected to ultracentrifugation, either on a continuous density gradient [30] or by the standard preparative technique [31]. The amount of protein associated with particles floating at different densities was estimated by measuring the absorbance at 280 nm of fractions as they were collected from the continuous density gradient. Plasma samples were also subjected to heparin-manganese precipitation, to precipitate lipoproteins other than HDL [32]. Whole plasma and selected fractions from ultracentrifugation or polyanionic precipitation were analyzed for total and unesterified cholesterol, using the enzymatic assay [27,33] with and without cholesteryl ester hydrolase (EC 3.1.1.13); esterified cholesterol was calculated as the difference between total and free cholesterol. Whole plasma and selected pooled fractions, from which excess salt had been removed by dialysis, were also analyzed for particle types, using lipoprotein electrophoresis on thin gels of 1.0% agarose [34], and for protein types, using polyacrylamide gel electrophoresis of samples delipidated in SDS [35]. Following electrophoresis, agarose gels were stained with Amido black and polyacrylamide gels were stained with Coomassie blue.

Experiments in vitro Plasma obtained from a fasted dog was heated at 56°C for 30 mm to inactivate lecithin : cholesterol acyltransferase [36]. The heat-treated plasma was then mixed with the phospholipid dispersion, in proportions simulating doses in vivo of 200 mg phospholipid per kg body weight (80 ~1 of the dispersion were added to each ml of plasma), 400 mg phospholipid/kg (160 pi/ml plasma), 800 mg phospholipid/kg (320 pi/ml plasma), and 1200 mg phospholipid/kg (480 pi/ml plasma). Physiologic saline was added to experimental and control samples to minimize differences in the dilution of plasma components. Samples were then incubated at 37°C for 1 h. As in the experiments in vivo, the lipid carriers were fractionated using density-gradient ultracentrifugation. Selected fractions were analyzed for free and esterified cholesterol content, as described above, for protein content by the method of Markwell et al. [28], and for lipid phosphorus by the method of Bartlett [37]. Fractions corresponding to peaks in the distributions of these constituents with respect to density were pooled and excess salt was removed by dialysis. These pooled fractions were analyzed for particle size, using gradient gel electrophoresis [38], and for protein species, using SDS-polyacrylamide gel electrophoresis [35]. Both types of gels were stained with Coomassie blue. Results Experiments in vivo The excess cholesterol that accumulated in plasma following the single infusion of phospholipid liposomes was almost entirely unesterified and carried by particles that were precipitated by heparin and manganese (Fig. 1). Thus, the particles were not HDL. These particles contained little, if any, cholesteryl ester (Fig. 1). Thus, they were not VLDL or LDL. Between 8 and 25 h, roughly half of the excess unesterified cholesterol disappeared from the plasma, indicating that these particles are gradually removed from the circulation. The particles that carried the excess in plasma unesterified cholesterol were entirely confined to the density range corresponding to LDL (Fig. 2A,

186

A

t=Oh

200

0”

’ I

1 2

TIME

’

’ 4

AFTER

’

’

INJECTION

’

’ 8

.:’

PZ IHDL)

I 25

(h)

Fig. 1. Concentrations of unesterified (open symbols) and esterified (filled symbols) cholesterol in plasma and plasma fractions from dogs infused with phospholipid. Whole plasma (0, 0); fraction precipitated by heparin and manganese (A, A). Vertical bars denote S.E. of the mean for four determinations (duplicates for each of the two dogs). Unesterified cholesterol not precipitated by heparin-manganese (i.e.. on HDL, as verified by agarose gel electrophoresis) was 0.918 kO.017 mmol/l at 0 h and 0.911 +0.022 mmol/l at 4 h; ester was 2.07 kO.056 mmol/l at 0 h and 2.15 kO.068 mmol/l at 4 h.

FRACTION

NUMBER

IO

P2 E = 06

I / I

w t

06

Y 5

peak 1). These particles also contained small amounts of protein (Fig. 2B). These proteins were apolipoprotein A-I and three unidentified proteins, the relative molecular weights of which were 67, 57 and 46 kDa (Fig. 3). The 67 kDa protein was possibly albumin, and the 46 kDa protein was possibly apolipoprotein A-IV. The apolipoprotein B of peak 1 presumably remained associated with LDL. None of these proteins originated from the injected phospholipid dispersion, as the dispersion produced no bands of its own on SDS-polyacrylamide gel electrophoresis (not shown). Injection of phospholipid was associated with the appearance of a new band on agarose gel electrophoresis. This band extended from below the p band (LDL) to beyond the pre-P band (VLDL), with its bulk between the p and pre-P bands (Fig. 4). The least mobile edge of this new band exhibited the same electrophoretic mobility as did the liposomes used for the injections (the electrophoresed liposomes stained off-color with Amido black on the agarose gel). Presumably, when the injected liposomes picked up endogenous

04

s 5: m

0.2

a L

0

II I

I

I

I

I

I

I

IO

FRACTION

I

I

I

/

20

I

I

I

I

30

II bottom

NUMBER

Fig. 2. (A) Distributions of unesterified cholesterol with respect to density. in samples of plasma from dogs infused with phospholipid. Female dog, 0; male dog, + Peak 1 (Pl) refers to peaks approximately in the LDL density range; peak 2 (P2) refers to peaks approximately in the HDL density range. (B) Distributions of protein concentration with respect to density. in samples of plasma from dogs infused with phospholipid. As the plasma samples described in A were separated into fractions by density, absorbance at 280 nm was continuously monitored by the collection apparatus. Displayed here are the tracings from the male dog only; those from the female dog were qualitatively similar. The bold curve is from t = 0 h: the thin curve is from t = 4 h: and the dotted curve is from I = 25 h.

proteins in the circulation, they increased their density to that of LDL and enhanced their electrophoretic mobility. They could be separated from

187

BSA

albumm

A

C

B

b

a

c

HDL

-

-

:_

apoA-

I

:

_& Fig. 3. SDS-polyacrylamide gel electrophoresis of proteins from the peaks 1 of Fig. 2. For each curve in Fig. 2A, 6-8 fractions centered about fractions lo-12 were pooled, delipidated, then electrophoresed on 10% polyacrylamide gels. Standards were bovine serum albumin (BSA) on the far left and canine HDL (i.e., apolipoprotein (apo) A-I) on the far right. Upper case letters mark gels that received samples from the female dog; lower case letters mark gels with samples from the male dog. A, a: r = 0 h; B, b: t = 4 h; C, c: t = 25 h. The arrow on gel ‘A marks the apolipoprotein B band.

VLDL (Fig. 4), LDL (Fig. 4) and HDL (Figs. 1 and 2A). Their minimal content of cholesteryl ester (Fig. 1) indicates that their basic structure remained vesicular: discoidal particles that form from liposomes disrupted by HDL apoproteins are rapidly converted by lecithin : cholesterol acyltransferase into spherical particles with cores of cholesteryl ester [4,18], whereas intact vesicles in

Origin b

a

B

l-pa

Fig. 4. Agarose gel electrophoresis of particles of density less than 1.063 g/ml. Lane a, normal fasted dog (t = 0 h, male dog). Lane b, t = 4 h, male dog. Lane B, I = 4 h, female dog. Lane Lps, raw liposomes. Samples for lanes a, b, and B were isolated by preparative ultracentrifugation. The upper arrow on lane a marks the pre-/3 band; the lower arrow makrs the /3 band. The anode was toward the top of the gel.

the presence of lecithin : cholesterol acyltransferase accumulate very little cholesteryl ester, owing to their inability to develop a lipid core [6]. The uptake of apoproteins by the phospholipid liposomes in our study may have made their lipid bilayers transiently permeable to solutes [3], until their uptake of cholesterol made them impermeable again [2,3,7]. These effects may account for any albumin the vesicles may have accumulated. In density, lipid composition, electrophoretic mobility through agarose gel, and precipitability with heparin and divalent cations, these circulating liposomes resembled lipoprotein-X [39], the abnormal lipoprotein of obstructive jaundice. The circulating liposomes also shared these characteristics with the vesicular particles that accumulate in human plasma following infusion of phospholipid in the form of Intralipid [5]. However, the protein species carried by the circulating liposomes were similar, but not identical, to those carried by human lipoprotein-X. Apolipoprotein A-I and albumin, which the liposomes appear to have acquired, are present in lipoprotein-X, but lipoprotein-x also contains apolipoprotein E and Capoproteins, which were not detected here, and lipoprotein-X has no proteins of weights 57 and 45 kDa [39]. The naturally occurring lipoproteins were only slightly affected by the infusion of liposomes. HDL shifted to a. slightly lower density following the infusion (Figs. 2A, B), presumably reflecting loss of apolipoprotein A-I to the liposomes and gain of phospholipid. Apolipoprotein A-I remained the only protein associated with peak 2 (not shown). The amount of unesterified cholesterol in HDL did not change (Fig. 1). Esterified cholesterol in whole plasma increased slightly, particularly at 8 and 25 h following the infusion (Fig. 1). LDL remained associated with apolipoprotein B (Fig. 3) and migrated slightly less rapidly on agarose gel electrophoresis (Fig. 4). The plasma concentration of LDL may have declined following the phospholipid infusion: the concentration of cholesteryl ester precipitated by heparin and manganese decreased between 0 and 4 h (Fig. l), and the concentration of protein associated with peak 1 was slightly less at 25 h than before the injection of liposomes (Fig. 2B). VLDL, which occurs in small quantities in the dog, did not

188

appear to be appreciably pholipid infusion.

affected

by the phos-

Experiments in vitro Incubation of phospholipid liposomes with heat-treated plasma produced a shift of large amounts of unesterified cholestrol and protein, as well as some cholesteryl ester, into the LDL density range. The bulk of the added phospholipid also accumulated in this density range (Fig. 5 and Table I). As in vivo, it appears that the liposomes increased their density by picking up protein and losing some phospholipid. As in vivo, the liposomes accumulated unesterified cholesterol. Density-gradient ultracentrifugation was able to produce a relatively pure isolate of these modified liposomes only from the incubated mix with the highest ratio of liposomes to plasma (480 ~1 dispersion/ml plasma). Peak 1 b from this incubation contained no apolipoprotein B (Fig. 6) and virtually no cholesteryl ester (Fig. 5 and Table I), and hence virtually no contaminating LDL. The modified liposomes in peak lb were rich in unesterified cholesterol and phospholipid (Table 1). Apolipoprotein A-I was essentially their only protein (Fig. 6). They smoothly distributed in size in the same range as did the unincubated liposomes (Fig. 7; the unincubated liposomes stained faintly and off-color with Coomassie blue). Ultracentrifugation was unable to similarly isolate LDL from the incubations. Nevertheless. several observations about the modifications of LDL during incubation with phospholipid liposomes can be made. First, LDL remained a discrete particle, and could be identified by its content of cholesteryl ester (peak la in the lowest panel of Fig. 5) and apolipoprotein B (Fig. 6). Second, it ultimately shifted to a slightly lower density (Fig. 5) presumably because of its gains in phospholipid and cholesteryl ester (Table I), which were unaccompanied by any loss of protein. The protein of LDL, apolipoprotein B, was neither lost to the HDL in the incubated mixes (peaks 2) nor to the modified liposomes (peak lb from the incubation of 480 ~1 dispersion/ml plasma; see Fig. 6). The size distribution of the incubated LDL is represented by the closely spaced bands in Fig. 7. These bands were absent from the electrophoretic

P2

No liposomes

400

4 a

-E

P2

80~1 /ml plasma

400

(HDL)

i\

r

i’

C

480 PI /ml

plosmo L

200

100

0

20

FRACTION

30

bottom

NUMBER

Fig. 5. Distributions of rhterified ( X) and unesterified (0 J cholesterol with respect to density, in the incubated mixes oi liposomes with canine plasma. Displayed here are the distrlbutiona for the incubations with no liposomes (top panel). the incubation of 80 ~1 of the phospholipid dispersion per ml of plasma (middle panel), and the incubation of 480 IJ.I of the dispersion per ml of plasma (bottom panel).

pattern of the one sample that contained liposomes but no LDL (peak 1 b from the incubation of 480 ~1 dispersion/ml plasma). The furthest migrating of these bands represent particles that penetrated the gel slightly beyond unmodified LDL (Fig. 7. lower gel). Presumably, these particles are LDL that lost small amounts of unesterified cholesterol to the liposomes, and thereby became

189

TABLE

I

COMPOSITION OF THE PEAKS GENERATED BY DENSITY-GRADIENT INCUBATED MIXES OF LIPOSOMES WITH CANINE PLASMA

ULTRACENTRIFUGATION

OF

THE

Displayed are the sums of measurements from fractions that correspond to peaks in the distributions of lipid and protein with respect to density (see Fig. 5). Molecular weights of 386.67 for free cholesterol and 650 for cholesteryl ester were assumed. In Fig. 5, peaks 1 and 2 each appear to have at least two components; thus, data are given for the less dense (Pla and P2a) and more dense (Plb and P2b) portions of each peak separately. Total lipid recoveries from ultracentrifugation were 97.7-102.5%. Chol, cholesterol: PPL, phospholipid; n.d., not determined. Peak

Incubated mix (pi phospholipid ml plasma)

dispersion/

Fraction numbers of peak

Composition of peak (mg/peak per dl plasma) free choi.

chol. ester

PPL

protein

Pl

0 80 160 480

9-15 9-15 6-13 4-13

5.23 21.40 nd. 48.56

8.46 12.51 n.d. 18.81

50.34 n.d. 676.57 1854.4

25.49 n.d. 78.04 171.03

Pla

0 80 160 480

9-12 9-12 6-10 4-10

2.94 12.11 n.d. 27.36

4.42 7.70 n.d. 15.50

26.07 n.d. 425.25 1211.5

14.46 n.d. 47.80 96.80

Plb

0 80 160 480

13-15 13-15 11-13 11-13

2.29 9.29 n.d. 21.21

4.05 4.81 n.d. 3.33

24.21 n.d. 251.32 636.86

11.03 n.d. 30.24 74.23

P2

0 80 160 480

19-26 17-24 16-23 14-20

23.82 13.69 n.d. 2.78

96.27 97.28 nd. 87.15

354.61 n.d. 573.92 661.21

280.68 n.d. 254.68 206.03

P2a

0 80 160 480

19-21 17-19 16-18 14-16

5.01 3.71 n.d. 2.06

16.29 20.73 n.d. 47.61

64.29

n.d. 164.21 307.94

35.78 n.d. 58.07 91.89

0 80 160 480

22-26 20-24 19-23 17-20

18.81 9.98 n.d. 0.72

79.98 76.56 n.d. 39.54

n.d. 409.71 353.27

244.91 n.d. 196.60 114.14

29.05 35.09 n.d. 51.34

104.73 109.79 n.d. 105.96

404.95 n.d. 1250.5 2515.6

306.17 n.d. 332.72 377.06

P2b

Pl+P2

0 80 160 480

-

slightly smaller. A similar loss of surface components by LDL has been hypothesized to account for the lecithin : cholesterol acyltransferase-dependent shrinkage of LDL that occurs during incubation of normal human plasma [40]. The remaining bands represent particles whose sizes differ by discrete increments. Apparently, during the incubations, LDL accumulated lipid or

290.32

protein in discrete amounts. This uptake of discrete amounts of material by LDL is unlikely to have been the result of simple fusion with liposomes. The liposomes were too large (M, > 106) to account for the close spacing of the bands (the bands were approx. 10’ daltons apart). and the liposomes had an initial size distribution that was too broad to allow sharp bands after fusion.

190

Pl

0

LDL

Pla 160

320

480

STD

Lps

-

669

K

440K

Fig. 6. SDS-polyacrylamide gel electrophoresis of proteins from the different portions of the peaks 1 of Table 1. Standards were bovine serum albumin (BSA) on the far left and canine HDL (i.e., apolipoprotein (apo) A-I) on the far right. peaks (Pl. Pla, Pl b) are defined in Table I; 2-3 fractions from each peak were pooled for use here. The number above each gel (0. 160, 320, or 480) is the ~1 of phospholipid dispersion per ml of plasma in the incubated mix from which the pooled fractions applied to that gel originated. Normal canine LDL was applied to the second gel from the left; the arrow on this gel marks the apolipoprotein B band.

232

-

0

LDL

Nevertheless, it is likely that the liposomes were the source of much of the material accumulated by LDL. LDL incubated in plasma without liposomes does not enlarge [40]. In contrast, LDL incubated with only the bottom fraction (d > 1.20 g/ml) and a low concentration of phospholipid liposomes becomes phospholipid-enriched, enlarges, and developes a single, extra, discretely separated peak in its size distribution [40]. Presumably, the separate peaks in the size distribution of LDL in Fig. 7 were largely the result of a similar, but far more pronounced, uptake of discrete amounts of liposomal phospholipid. Enhancement of cholesteryl ester transfer from HDL [41] appears to have also played a role (Fig. 5, Table I). HDL was also significantly modified during the incubation with liposomes. At a liposomal concentration simulating the phospholipid dose administered in vivo, HDL lost almost half its unesterified cholesterol; at higher concentrations of liposomes, it lost virtually all its unesterified cholesterol. At these higher concentrations of liposomes, HDL also lost about 10% of its cholesteryl ester and over a quarter of its protein content (apolipoprotein A-I), while its phospholipid content almost doubled (peak 2 in Fig. 5 and Table I). These modifications shifted the entire HDL peak

PI

K

Plb 160

320

480

STD

Lps

-

-

440K

-

232K

Fig. 7. Gradient gel electrophoresis of particles from the different portions of the peaks 1 of Table I. Gels had polyacrylamide gradients of 2-16%. Molecular weight standards (STD) were thyroglobulin (669000), ferritin (440000). catalase (232000). lactic dehydrogenase (140000). and bovine serum albumin (67000). The last two standards were too small for these gels. The lanes marked Lps received the 5% phospholipid dispersion. The notation for samples is otherwise identical to that used in Fig. 6.

to a lower density (Fig. 5). All forms of HDL (i.e., peaks 2a and 2b) became larger (Fig. S), presumably reflecting gain of phospholipid [5] and perhaps a lesser degree of packing following the loss of protein [42]. The size increased gradually with increasing liposomal concentration, arguing against particle fusion (cf. Ref. 42). All forms of HDL still contained only one protein, apolipoprotein A-I (not shown).

191

P2a 0

160

320

P2b 480

ST0

320

480

669

K

440K HDLa

-

232K

P2a

P2b 0

160

320

480

ST0

0

400

-

669K

-440K

-

232K

-

140K

-

67 K

HDL ,3 -

Fig. 8. Gradient gel electrophoresis of particles from the different portions of the peaks 2 of Table 1. The polya~~Ia~de gradient of the top gei was 2-X&; the gradient of the bottom gel was 4-30’5~ Standards (STD} were the same as in Fig. 7. The notation for incubations and peaks is defined in Table 1 and Fig. 6.

Discussion

The data presented support the previously unconfirmed hypothesis [5] that the transient rise in plasma unesterified cholesterol concentration that follows intravenous injection of phospho~pid is carried p~n~pally by a synthetic lipid particle, the phosphatidyl~holine liposome. For several reasons, the unesterified cholesterol carried by these circulating liposomes appears to

have been extracted mainly from pre-existing tissue stores. First, phospho~pid dispersions in vitro are able to extract cholesterol from cellular membranes and from whole cells in culture IS]. Second, hypercholesterolemia produced by injection of phospholipid in vivo is associated with an initial drop in the cholesterol content of several organs [S], especially the (highly vascular) lungs [22]. Finally, the accumulation of cholesterol in the bloodstream following an infusion of phospholipid occurs independently from changes in the rates of cholesterol excretion [20] or synthesis [22,23]. The accumulation of vesicular particles in the circulation following infusion of phospholipid appears to depend on the dose of phospholipid administered. At relatively low doses (2.0-17 mg phospholipid/kg body weight), intravenously administered phospholipid vesicles are rapidly incorporated into particles resembling HDL in density and chemical composition [1,2]. Similarly, following the infusion of chylomicrons containing 1.5-3.6 mg phospho~pid/kg body weight, the rapid hydrolysis of their t~acyl~y~rol cores liberated surface phospholipid, which formed vesicles that were also rapidly incorporated into HDL [43]. Thus, the total assimilation of small amounts of vesicular phospholipid into HDL may be a step in the normal metabolism of the phospholipid of chylomicrons. Higher doses of intravenously administered phospholipid result in persistent vesicles. At 200-800 mg phospholipid/kg body weight, phospholipid vesicles with solutes entrapped in their aqueous cores partially retain these solutes after infusion 171, even when the infused vesicles were made without any cholesterol to stabilize their structure [8-121. By implication, the vesicles not only retained their vesicular structure, but remained relatively impermeable to the entrapped solutes. Similarly, following the infusion of Intralipid containing 73-480 mg phospholipid/kg body weight into human subjects, the rapid hydrolysis of the triacylglycerol cores liberated surface phospholipid, which formed vesicles 113-161; triacylglycerol-free phospho~pid vesicles in the infused Intra~pid may also have contributed [5,17]. At this high dose of phospho~pid, the vesicles persisted in the circulation for days. During this time, they accumulated endogenous

192

cholesterol and protein, much like the vesicular particles in our study. The persistence of vesicular particles in our study in vivo, in which 200 mg phospholipid/kg body weight were administered. is quite consistent with these previous reports. Based on these previous studies and our own, there seems to be a threshold for doses of intravenously administered phospholipid. Below the threshold, phospholipid liposomes are completely incorporated into HDL, possibly as part of a normal metabolic pathway. Above the threshold, the homeostatic mechanisms can not accommodate additional liposomal phospholipid. Possible explanations include saturation of mechanisms to remove liposomes from the circulation (51, such as in the liver or spleen, and inability of native lipoproteins to pick up enough phospholipid or donate enough apoproteins to disrupt the liposomes. Above the threshold, the liposomes may become stabilized by their uptake of cholesterol [2,3,7] faster than the possibly limited amounts of apoproteins can make them leaky or break them apart into discoidal particles. Our data suggest that the accumulation of cholesterol in plasma following infusion of phospholipid liposomes is enhanced by interactions of the liposomes with naturally occurring lipoproteins. For example, the liposomes picked up apoproteins from lipoproteins. Apolipoproteins have been shown to enhance the ability of phospholipid dispersions to extract cellular cholesterol in vitro and, presumably, would enhance this ability in vivo as well [5]. In addition, our studies in vitro showed that liposomes can siphon substantial amounts of unesterified cholesterol from HDL. We presume this effect was not detected in vivo because the HDL rapidly recovered its losses at the expense of the tissues. Thus. in addition to direct liposomal uptake of tissue cholesterol, cholesterol may be shuttled by HDL from the tissues to the liposomes [S]. Finally, the slight rise in the concentration of cholesteryl ester in plasma following the infusion of liposomes may be a consequence of enhanced lecithin : cholesterol acyltransferase activity caused by phospholipidenrichment of pre-existing HDL particles or by the appearance of new substrate particles, such as discoidal phospholipid-apoprotein complexes or intact liposomes that acquired ap~~lipoprotein A-I

[5,6]. Cholesteryl esters produced on these liposomes would be available for transfer to acceptor lipoproteins 161. Alternatively. the circulating liposomes may have caused the rise in plasma cholesteryl ester concentration by slowing the removal from plasma of particles rich in cholesteryl ester, possibly by competing with hepatie and peripheral apoprotein receptors [5]. Our studies in vitro generated a group of particles in the LDL density range that were of similar composition to those previously reported [4] to arise during incubation of liposomes with plasma (3%, unesterified cholesterol, less than 0.5% cholesteryl ester, 87% pllospholipid, 10% protein by weight; see peak lb. Table I). Our data are consistent with the previously described discoidal nature [3,4,18] of these particles. The low content of esterified cholesterol in these particles dictates that their structure be vesicular or discoidal, but, in the absence of lecithin : cholesterol acyltransferase, does not distinguish between the two possibilities. The ability of liposomes in high doses to retain their vesicular structure in vivo but not in vitro may reflect the greater availability of cholesterol from tissue stores in vivo to stabilize the liposomes against the disruptive effects of apoproteins. This difference in structure may also explain the absence from the LDL density range in vitro of the proteins of weights 67. 57 and 46 kDa that were found in the LDL density range in vivo. Some proteins, particularly albumin, associate with vesicular lipoproteins by dissolving in their aqueous cores [39]. The uptake of apolipoprotein A-I by the synthetic particles in vivo and in vitro is consistent with previous demonstrations that the core is a minor determinant of association with apolipoprotein A-l 11’71. We describe three new effects in our incubations of liposomes with plasma. First, the phospholipid-rich particles in the LDL density range accumulated large amounts of cholesterol from HDL, eventually with near-complete depletion of the unesterified cholesterol of HDL. Second, the HDL fraction showed no saturation of its capacity to aqquire phosphotipid. It accumulated ever-increasing amounts of phospholipid in the presence of increasing amounts of added phospholipid, reaching a maximum of 1.09 mg of acquired phos-

193

pholipid per mg of HDL protein in the starting plasma. Third, LDL appears to have become extensively enriched in phospholipid from the liposomes and cholesteryl ester from HDL, in discrete amounts (cf. Ref. 40). The specific absence of the first two of these effects from a previous report of a similar set of incubations of liposomes with plasma [4] is only partly explained by differences in the proportion of added phospholipid to HDL protein, as there was overlap between our study (weight ratios ranged from 0 to 8.55) and theirs (weight ratios ranged from 0 to 2.5). Additional factors that may have led to the differing results include the use of a different species of blood donor (human) and a different means of inactivating lecithin : cholesterol acyltransferase (2 mM 5,5’-dithionitrobenzoic acid) in the prior study. Concerning the third effect, the loss of HDL core lipid (Fig. 5 and Table I), despite an increase in HDL particle size (Fig. 8), is contrary to prior prediction (cf. Ref. S), and may be a result of enhanced capacity for cholesteryl ester in the modified acceptor particles. Overall, our studies provide evidence that single infusions of phospholipid mobilize tissue cholesterol into plasma through the formation of distinct, vesicular lipoproteins and through modifications in the naturally occurring lipoproteins. It is plausible that these mechanisms play a role in the ability of repeated infusions of phospholipid to produce regression of experimental atherosclerosis. Acknowledgements The authors wish to thank Ms. Maria Garcia of the Lipoprotein Core of Program Project USPHSHL 18577 for her technical assistance, Dr. Gunther M. Fless for his helpful advice, and Dr. Arthur Rubenstein for his interest in the study. This work was partially supplemented by USPHS grant HL 18577. References 1 Krupp, L., Chobanian, A.V. and Brecher, PI. (1976) Biothem. Biophys. Res. Commun. 72, 1251-1258 2 Tall, A.R. (1980) J. Lipid Res. 21, 354-363 3 GUO, L.S.S., Hamilton, R.L., Goerke, J., Weinstein, J.N. and Havel, R.J. (1980) J. Lipid Res. 21, 993-1003

4 Tall, A.R. and Green, P.H.R. (1981) J. Biol. Chem. 256, 2035-2044 5 Williams, K.J., Werth, V.P. and Wolff, J.A. (1984) Perspect. Biol. Med. 27, 417-431 6 Chajek, T., Aron, L. and Fielding, C.J. (1980) Biochemistry 19, 3673-3677 7 Finkelstein, M. and Weissmann, G. (1978) J. Lipid Res. 19, 289-303 8 McDougall, I.R., Dunnick, J.K., McNamee, M.G. and Kriss, J.P. (1974) Proc. Nat]. Acad. Sci. USA 71, 3487-3491 9 McDougall, I.R., Dunnick, J.K., Goris, M.L. and Kriss, J.P. (1975) J. Nucl. Med. 16, 488-491 10 New, R.R.C., Chance, M.L., Thomas, S.C. and Peters, W. (1978) Nature 272, 55-56 11 Kirby, C. and Gregoriadis, G. (1980) Life Sci. 27.2223-2230 12 Poste, G., Bucana, C., Raz, A., Bugelski, P., Kirsh, R. and Fidler, I.J. (1982) Cancer Res. 42, 1412-1422 13 Thompson, G.R., Segura, R., Hoff, H. and Gotto, A.M. (1975) Eur. J. Clin. Invest. 5, 373-384 14 Griffin, E., Breckenridge, W.C., Kuksis, A., Bryan, M.H. and Angel, A. (1979) J. Clin. Invest. 64, 1703-1712 15 Untracht, S.H. (1982) B&him. Biophys. Acta 711, 176-192 16 Muckle, T.J., Bradford, J.A. and Aukland, P. (1982) Am. J. Clin. Pathol. 78, 523-526 17 Connelly, P.W. and Kuksis, A. (1981) Biochim. Biophys. Acta 666, 80-89 18 Forte, T.M., Nichols, A.V., Gong, E.L., Lux, S. and Levy, RI. (1971) B&him. Biophys. Acta 248, 381-386 19 Goldberg, C.S., Tall, A.R. and Krumholz, S. (1984) J. Lipid Res. 25, 714-720 20 Byers, S.O. and Friedman, M. (1956) Proc. Sot. Exp. Biol. Med. 92, 459-462 21 Friedman, M., Byers, SO. and Rosenman, R.H. (1957) Proc. Sot. Exp. Biol. Med. 95, 586-588 22 Byers, S.O., Friedman, M. and Sugiyama, T. (1962) J. Biol. Chem. 237, 3375-3380 23 Byers, S.O. and Friedman, M. (1969) Lipids 4, 123-128 24 Stafford, W.W. and Day, C.E. (1975) Artery 1, 106-114 25 Goto, M., Kawada, R., Sekiya, T., Ohki, K., Sakata, K. and Nozawa, Y. (1983) B&hem. Med. 29, 235-247 26 Skipski, V.P., Peterson, R.F. and Barclay, M. (1964) Biothem. J. 90, 374-378 27 Allain, CC., Poon, L.S., Chan, C.S.G., Richmond, W. and Fu, P.C. (1974) Clin. Chem. 20, 470-475 28 Markwell, M.A.K., Haas, SM., Bieber, L.L. and Tolbert, N.E. (1978) Anal. B&hem. 87, 206-210 29 Huang, C. and Thompson, T.E. (1974) Methods Enzymol. 32, 485-489 30 Nilsson, J., Mannickarottu, V., Edelstein, C. and Scanu, A.M. (1981) Anal. B&hem. 110, 342-348 31 Havel, R.J., Eder, H.A. and Bragdon, J.H. (1955) J. Clin. Invest. 34, 1345-1353 32 Warnick, G.R. and Albers, J.J. (1978) J. Lipid Res. 19, 65-76 33 Steele, B. W., Koehler, D.F., Azar, M.M., Blasrkowski, T.P., Kuba, K. and Dempsey, M.E. (1976) Clin. Chem. 22, 98-101 34 Papadopoulos, N.M. and Kintzios, J.A. (1971) Clin. Chem. 17. 427-429

794

35 Weber, K. and Osborn, M. (1969) J. Biol. Chem. 244, 4406-4412 36 Glomset, J.A. (1968) J. Lipid Res. 9, 155-167 37 Bartlett, G.R. (1959) J. Biol. Chem. 234, 466-468 38 Anderson, D.W., Nichols, A.V., Forte, T.M. and Lindgren, F.T. (1977) B&him. Biophys. Acta 493, 55-68 39 Narayanan, S. (1979) CRC Crit. Rev. Clin. Lab. Sci. 11. 31-51

40 Shahrokh, 2. and Nichols, A.V. (1982) Biochem. Biophys. Res. Commun. 108, 888-895 41 Ha, Y.C. and Barter, P.J. (1982) Comp. Biochem. Physiol. 71B, 265-269 42 Nichols, A.V., Gong, E.L., Forte, T.M. and Blanche, P.J. (1978) Lipids 13. 943-950 43 Tall, A.R., Green, P.H.R.. Glickman, R.M. and Riley, J.W. (1979) J. Clin. Invest. 64, 977-989