George Lyman Duff Memorial Lecture. Role of the Liver in Atherosclerosis

Transcription

1 George Lyman Duff Memorial Lecture Role of the Liver in Atherosclerosis Richard J. Havel George Lyman Duff was a pioneer in relating the quality as well as the quantity of plasma lipoproteins to experimental atherosclerosis. His investigations and those of John Gofman foreshadowed much contemporary research on the interactions of lipoproteins with cells and other components of the arterial wall. This current research has, I believe, put to rest past criticisms of the meetings of our Council, at which it seemed to some that the artery is hardly involved in the pathogenesis of atherosclerosis. Perhaps I can then be forgiven for taking as the subject of this lecture the role of the liver in lipoprotein metabolism and atherosclerosis, and for daring to suggest that if the liver would only do correctly its job of secreting the "right" kinds of lipoproteins and of taking up, catabolizing, and excreting lipoprotein components, it would be unnecessary for us to be so concerned about the mechanisms by which lipoprotein-cholesterol accumulates in the arterial intima. It is a truism that laboratory animals fed standard chow diets have low levels of atherogenic lipoproteins and do not develop atherosclerosis, whereas lipid accumulates in arteries and proliferative, spaceoccupying lesions develop when these animals are fed unnatural diets, sometimes aided by auxiliary measures that perturb hepatic cholesterol metabolism. My introduction to this field was in Joseph Bragdon's laboratory at the National Heart Institute in Bragdon, a pathologist, had read a report that Alaskan ground squirrels have rather high serum lipid levels and he undertook a systematic investigation of another species of ground squirrel native to the northern Rocky Mountains, Citellus colombianuss These animals did not develop hypercholesterolemia with cholesterol-feeding. However, in the rather Dr. Richard J. Havel Is at the Cardiovascular Research Institute and the Department of Medicine, School of Medicine, University of California, San Francisco, California. This lecture was given on November 12, 1984, at the 57th Scientific Session of the American Heart Association at Miami, Florida. The author's research since 1971 has been supported by U.S. Public Health Service Grant HL Arteriosclerosis SCOR. Address for reprints: Richard J. Havel, M.D., 1327-M, University of California, San Francisco, California (Arteriosclerosis 5: , November/December 1985) 569 dark and frequently cool basement laboratory that we occupied, they seemed to think that winter was coming and many of them rapidly gained weight and had visibly lipemic serum, whether they were eating low-fat or high-fat vegetable diets. In some of them, triglyceride levels reached several thousand mg/dl and they had creamy serum. When the animals were caused to hibernate by placing them in the cold room, they lost weight and their serum triglyceride levels fell rapidly. Some of the lipemic animals had oil red O-positive extracellular deposits in the aorta, coronary arteries, and heart valves, and two of them had masses of foam cells beneath the endothelium, even though they had quite high high density lipoprotein (HDL) levels, as estimated by analytical ultracentrifugation. What struck me was Bragdon's observation that serum triglyceride levels were correlated with the extent of fatty infiltration of the liver, which in one case reached a level of 44% of gross liver weight, and also that both the severity of the fatty liver and the hypertriglyceridemia were correlated with the rate of body weight gain. 1 Joe and I talked then about a junket to Alsace-Lorraine to study the fatty liver and (we assumed) the hyperlipemia of the force-fed goose. Lipoprotein Synthesis and Secretion Some may not fully appreciate that in 1953 there was little understanding of the sites or mechanisms of lipoprotein biosynthesis, and we knew essentially nothing about apolipoproteins. Much later, when we did know something about lipoprotein structure and metabolism, I encountered in fetal guinea pigs a phenomenon reminiscent of Bragdon's early observations in ground squirrels. In these fetal animals, which rapidly gain weight during the last part of gestation and are born sufficiently mature that they do not need to suckle to survive, Thomas Bahmer and I studied the transplacental transfer of free fatty acids. 2 We showed that a large fraction of fatty acids derived from maternal adipose tissue is transferred to the fetal liver and esterified to form triglycerides, causing fatty liver and hyperlipemia. By electron microscopy, the secretory vesicles of the Golgi apparatus of hepatocytes were seen to be filled with giant, very low density lipoproteins (VLDL) (the size of chy-

2 ARTERIOSCLEROSIS VOL 5, No 6, NOVEMBER/DECEMBER 1985 Figure 1. Transmission electron photomicrographs of fetal guinea pig liver at 68 days of gestation. A. Golgi region of hepatocyte, showing dilated Golgi cistemae and secretory vesicles containing lipoprotein particles up to 2000 A in diameter, x 35,000. B. Portions of two hepatocytes, with extracellular regions between the cells and at the sinusoidal front filled with lipoprotein particles of similar size to those in the Golgi apparatus. Several large fat droplets are present in the cell on the left. Arrow indicates possible point of fusion of secretory vesicle with plasma lemma, x 16,500. (Reprinted from reference 3, with permission of the publisher.) lomicrons normally produced after a fat-rich meal) and similar VLDL were seen in the space of Disse (Figure 1); both were comparable to the animals' serum VLDL. 3 We concluded that the hyperlipemia in these animals, in which no large chylomicrons could have been formed, arose from overloading of the liver with maternal fatty acids, which were subsequently transported, as VLDL-triglycerides, to adipose tissue for use by the animals after birth. I was by then convinced that the liver has a tremendous capacity to secrete triglycerides and, like the gut, can make bigger VLDL when this is necessitated by the fat load. It seems to make no difference whether the fat comes directly from the diet, from preformed free fatty acids, or from hepatic lipogenesis when large amounts of carbohydrates are ingested. In the late 1960s, Robert Hamilton, Albert Jones, and others defined the pathway of synthesis and secretion of VLDL in hepatocytes (Figure 2). The nonpolar cores of the VLDL that the liver secretes usually contain mainly triglycerides. In species that store large amounts of cholesteryl esters in the liver during cholesterol feeding, including rabbits, guinea pigs, some primates, and (with auxiliary measures) rats, the liver then secretes VLDL with cholesteryl ester-enriched cores 56 (Figure 3). Evidently, the liver has the capacity to package and secrete in VLDL whichever nonpolar lipid is available.

3 ROLE OF THE LIVER IN ATHEROSCLEROSIS Havel 571 Bile Conaliculus Figure 2. Pathway of assembly and secretion of VLDL in hepatocytes. The lipid particle, which is assembled in the smooth endoplasmic reticulum (SER), is thought to acquire its protein components, including apo B which is synthesized in the rough endoplasmic reticulum (RER), at the smooth-surfaced termini of the latter. The nascent VLDL are transported to the Golgi apparatus, where glycosylation of the proteins is completed, and concentrated in secretory vesicles. The latter migrate to the cell surface and fuse with the plasmalemma, releasing the particles into the hepatic sinusoid. (Reprinted from reference 4, with permission of the publisher.) 80 % of Total Lipoprotein Mass 40 VLDL IDL Days Fed 1% Cholesterol Diet Figure 3. The composition of the core of VLDL and IDL isolated from blood plasma of guinea pigs is drastically altered by feeding the animals a diet containing 1 % cholesterol by weight. In animals fed a low-fat chow, these lipoproteins contain very small amounts of cholesteryl esters (CE). In cholesterol-fed animals, cholesteryl esters replace much of the normally predominant triglycerides (TG) in the core of the particles. The size of VLDL and IDL is not appreciably changed by cholesterol-feeding. (Graph prepared from data in reference 6.) CE Hepatic Uptake of Lipoprotelns It has been known since the early 1960s that free fatty acids derived from chylomicron triglycerides, like free fatty acids derived from adipose tissue, can be esterified to form triglycerides in the liver and then exported in VLDL. Dietary cholesterol, however, enters the liver with chylomicron remnants. The early work of Paul Nestel, working with me in San Francisco, 7 and the later studies of Emmett Bergman in sheep and dogs 8 and those of Trevor Redgrave, who first isolated chylomicron remnants from functionally eviscerated rats, 9 defined the two-step pathway of chylomicron metabolism and showed that the liver is the major and obligatory site of uptake of dietary cholesterol. Subsequently, it was found that hepatogenous VLDL are also metabolized by an analogous pathway, the outlines of which had become apparent by (Figure 4). At the same time, it became apparent that lipoproteins can be catabolized by receptor-dependent endocytosis as a result of the discovery of the LDL receptor by Michael Brown and Joseph Goldstein. 12 We also knew that not all VLDL remnants enter the liver, but that some are further processed to form low density lipoproteins (LDL). Finally, we knew that one of the major differences between VLDL remnants, which Ole Mjos isolated from functionally eviscerated rats, 13 and LDL was that LDL lack the arginine-rich protein discovered by Virgie and Bernard Shore, 14 now known as apolipoprotein E. Some of the other features of VLDL metabolism, which are analogous to those of chylomicrons, including the acquisition of the cofactor protein for lipoprotein lipase (apo C-ll) and other C apoproteins from HDL, are also shown in this 1974 model. RER Figure 4. This 1974 model of VLDL metabolism has proven to be generally correct. Nascent VLDL contain apo B (o), apo E (a, here called hi ARG), and C apoproteins (A). The role of apo C-ll as a cofactor for lipoprotein lipase was then known, as was that of HDL as a reservoir for C apoproteins, but the general role of C apoproteins in preventing the premature uptake of nascent VLDL, prior to formation of remnant particles, 11 was discovered later; so was the function of apo E as a ligand for lipoprotein receptors and the definitive demonstration that VLDL remnants are taken up into hepatocytes by receptor-mediated endocytosis. (Reprinted from reference 10, with permission of the publisher.)

4 572 ARTERIOSCLEROSIS VOL 5, No 6, NOVEMBER/DECEMBER 1985 It is now appreciated that the uptake of chylomicron remnants by the liver is a receptor-mediated, endocytic event 15 and that apo E has a crucial role in this process. We know that: 1. HDL containing only apo E compete for uptake of chylomicron remnants in perfused rat livers The addition of apo E to chylomicrons from estradiol-treated rats increases their uptake into perfused rat livers Particles resembling chylomicron remnants (which contain apo B-48) accumulate in the blood plasma of persons with familial dysbetalipoproteinemia The removal of apo B-48 from plasma is grossly impaired in persons with familial dysbetalipoproteinemia caused by at least two mutations of apo E. 20 A modern version of chylomicron metabolism is shown in Figure Several years ago in his Duff lecture, Donald Zilversmit presented evidence that chylomicron remnants, enriched in cholesterol, are responsible for the hypercholesterolemia of cholesterol-fed rabbits. 22 His subsequent studies with Katherine Thompson 23 have shown that the particles that accumulate probably represent remnants of VLDL, secreted from the cholesterol-loaded livers of these animals. They found that chylomicron remnants are removed normally by perfused livers of cholesterolfed rabbits, whereas the cholesterol-rich particles that accumulate in their plasma are removed slowly. The nature of the particles that accumulate in cholesterol-fed animals can also be evaluated by determining the species of apo B that they contain. Jean- Louis Vigne and I (Vigne JL, Havel RJ, unpublished data) have found that mesenteric lymph chylomicrons from both chow-fed and cholesterol-fed rabbits contain apo B-48, the form of apo B secreted from the intestine of rats and humans. 24 By contrast, the cholesterol-enriched VLDL of nonfasted, cholesterol-fed rabbits, like the VLDL secreted from their perfused livers, contain predominantly apo B-100 at all times after cholesterol feeding (Figure 6). The accumulation of cholesterol in the liver evidently has two effects: 1) secretion of cholesterol-enriched VLDL; and 2) down-regulation of hepatic LDL receptors. 25 Both of these contribute to the accumulation of cholesterol in the blood. Only in the alloxan-diabetic, cholesterol-fed rabbit, a model introduced by Duff and McMillan 26 do triglyceride-rich particles that contain mainly apo B-48 accumulate in appreciable amounts (Vigne JL, Havel RJ, unpublished data). These large particles are probably those that Duff INTESTINAL MUCOSAL CELL HDL Blood Capillary Llpo protein Lipau FFA Figure 5. Chylomicron pathway. Chylomicrons secreted from absorptive cells of the small intestine tranport dietary triglycerides (gray region of particle) and cholesterol, as cholesteryl esters (darker region of particle). Most of the triglycerides are hydrolyzed in extrahepatic tissues by lipoprotein lipase. The resulting remnant particles, depleted of trigycerides and most surface lipids and proteins (but retaining the cholesteryl esters, apo B-48, and apo E), are taken up into hepatocytes after binding to a chylomicron remnant receptor. Lysosomal catabolism releases the cholesterol, which thereby regulates a number of key processes in this cell that influence lipoprotein biosynthesis and catabolism, including cholesterol synthesis, cholesterol esterification, bile acid synthesis, and the activity of LDL receptors. (Reprinted from reference 21, with permission of the publisher.)

5 ROLE OF THE LIVER IN ATHEROSCLEROSIS Havel 573 B-100 < B-48- B100 B 48 E - Cholesterol Triglycerides no Figure 6. Potyacrylamide gel electrophoretograms run in sodium dodecyl sulfate at a low concentration of polyacrylamide (3.5%) permit the B apoproteins that accumulate in VLDL of cholesterol-fed rabbits to be visualized. The gels, from left to right, are of VLDL obtained from seven rabbits, 1 to 90 days after they were placed on diet of pelleted rabbit chow to which 1% by weight of cholesterol was added by evaporating an ether solution of cholesterol over the pellets. In all cases, apo B-100 predominated and only small amounts of apo B-48 were present. found to be less atherogenic and which Peter Brecher and his associates 27 have shown to have less capacity to induce foam cell formation when incubated with macrophages than the remnant-like VLDL that accumulate in nondiabetic, cholesterolfed rabbits (in whom lipoprotein lipase is normally active). In the alloxan-treated animals, apo B-100 remains the predominant form of apo B secreted by the liver (Figure 7). These data in cholesterol-fed rabbits are consistent with those that my colleagues and I in San Francisco have obtained in Watanabe heritable hyperlipidemic (WHHL) rabbits in collaboration with Brown, Goldstein, and their colleagues in Dallas. We found that apo B-100 accumulates in VLDL, intermediate density lipoproteins (IDL), and LDL of WHHL homozygotes, but little apo B-48 is found in these lipoproteins. 23 Although these animals essentially lack functional hepatic LDL receptors, their livers take up chylomicron remnants normally. 29 By contrast, the Dallas group has shown that the uptake of VLDLprotein by the liver is greatly retarded. 30 From these studies and those of others, 31 it is now evident that chylomicron and VLDL remnants are removed by distinct hepatic receptors. Uptake of both forms of remnants seems to depend upon apo E. Anton Stalenhoef, working with our group in San Francisco, 20 has studied the metabolism of apo B-48 and apo B- 100 of large triglyceride-rich lipoproteins ( A in diameter) in patients with familial dysbetalipopro- Figure 7. Polyacrylamide gradient gel electrophoretogram (3% to 15% polyacrylamide) of VLDL (d< g/ml) isolated from the perfusate of a liver of a cholesterol-fed, alloxan-diabetic rabbit. The liver was perfused for 3 hours. In this overloaded gel, apo B-100 predominates among the proteins of high molecular weight. A component with an apparent molecular weight slightly greater than 200,000 could represent apo B-48. teinemia, who have dysfunctional mutant forms of the protein. He found the removal of both B apoproteins from the blood to be grossly impaired. The defect in the structure of apo E in this genetic disorder thus accounts for the accumulation of both types of remnant in the blood. As I have noted, the lipoproteins that accumulate in WHHL homozygotes include not only LDL, but appreciable amounts of VLDL and IDL as well. The concentration of apo E, like that of apo B-100, is increased severalfold. These findings, taken together with the metabolic studies that I have described in cholesterol-fed and WHHL rabbits, strongly suggest that the hepatic uptake of VLDL remnants, as well as LDL, is mediated by the LDL receptor. In WHHL homozygotes, LDL accumulate not only because their clearance from the blood is impaired, but also because their formation is increased. 32 Toru Kita and David Bilheimer, working with Brown and Goldstein, 30 have suggested that LDL formation is increased because retained VLDL remnants are

6 574 ARTERIOSCLEROSIS VOL 5, No 6, NOVEMBER/DECEMBER 1985 gradually converted to LDL. This conclusion is supported by studies of lipoprotein secretion from the liver of WHHL homozygotes carried out by Conrad Hornick and Kita et al. 33 They found that the rate of accumulation of apo B-100 in perfusates of livers from the mutant animals is not increased; furthermore, virtually all the accumulating apo B-100 was in VLDL and almost none in LDL. It therefore appears that the number of VLDL particles secreted is normal but that fewer VLDL remnants are taken up directly by the liver and more are converted to LDL in the absence of a functional LDL receptor. In humans, Stalenhoef et al. 34 have found that apo B-100 of large VLDL is removed from the blood almost as rapidly as apo B-48 of chylomicrons; little or none of this apo B-100 is converted to LDL. These observations have led us to conclude that the extent to which VLDL remnants are converted to LDL is a function of particle size. The rapid removal of large VLDL by the liver presumably gives little opportunity for further metabolism to yield LDL. The rapid removal of large VLDL may be related to the presence of more than one molecule of apo E in each particle (Table 1). 35 A number of years ago, Thomas Innerarity, Robert Pitas, and Robert Mahley 36 concluded that HDL particles that contain several apo E molecules bind to multiple LDL receptor sites on cultured human fibroblasts, leading to higher binding affinity of these particles for the receptor than LDL, which presumably bind to the receptor via a single site on apo B. Later, we observed similar high affinity binding of rat VLDL particles containing several molecules of apo E to LDL receptors on liver membranes. 37 The high affinity of such particles can now be understood in terms of the recently described structure of the receptor protein: it contains seven repeated segments of about 40 amino acids in its cysteine-rich N-terminal region. Each of these segments has a highly negatively charged region that probably constitutes a receptor-binding domain. 38 Thus, rat VLDL particles may bind polyvalently to a single receptor, accounting for the high binding affinity and, presumably, the more efficient removal of such particles from the blood. 39 The rapid removal of human chylomicron remnants from the blood may Table 1. Apo E Content of Human Plasma Very Low Density Llpoprotelns Fraction (diameter, A) Normal (% of apo B mass) Calculated from data in reference 35. Endogenous hyperlipemia (% of apo B mass) also be a function of polyvalent binding via multiple molecules of apo E on these particles (Table 1). Nobuhiro Yamada, working with David Shames and me in San Francisco, has used immunoaffinity chromatography on columns containing anti-apo E to characterize VLDL, their remnants, and LDL in normal rabbits and WHHL homozygotes. We reasoned that VLDL and their remnants, which should contain apo E as well as apo B-100, would be retained on the affinity columns, whereas LDL, which contain only apo B-100, would not. We found that more than one-half of the apo B-100 in plasma of normal rabbits binds to such columns. All fractions (VLDL, IDL, and LDL) contain some particles that do not bind to the columns. The unbound fraction increases with density, but even LDL contain an appreciable portion of particles that contain apo E. Similar findings were obtained with lipoprotein fractions from WHHL homozygotes. We have also used immunoaffinity chromatography and multicompartmental analysis to study the conversion of radioiodinated apo B-100 of VLDL to remnants and LDL in normal and mutant rabbits. In normal rabbits, only a small fraction of VLDL particles (containing apo E) appears to be converted to LDL particles that lack apo E; however, all LDL seem to be derived from VLDL. The fraction of the VLDL converted to LDL is increased in WHHL homozygotes, as predicted from the observations of Kita and his associates. 30 However, the overall rate of removal of apo B-100 of VLDL from the blood considerably exceeds that of IDL or LDL. We have therefore considered the possibility that some VLDL particles in WHHL rabbits, which contain more than one molecule of apo E, are removed by their chylomicron remnant receptor, but we cannot exclude the possibility that a small number of LDL receptors, present on hepatocytes of the mutant animals, mediate the clearance of such particles. Our kinetic studies in normal rabbits also show that IDL particles, which appear to contain only a single molecule of apo E, are removed slowly from the blood, perhaps as slowly as LDL that contain no apo E. Evidently, particles that contain apo E are not invariably removed from the blood more efficiently than LDL. In normal rabbits, the concentration of remnant-like particles, isolated in the density ranges of IDL and LDL, is almost as great as that of particles that contain apo B-100 but no apo E. The apparently large concentration of VLDL remnants in rabbits (relative to "true" LDL) could be related to the very low activity of the hepatic, heparin-releasable lipase in this species 40 (Frost PH, Havel RJ, unpublished data). This enzyme, which hydrolyzes triglycerides and phospholipids in small triglyceride-rich lipoproteins efficiently in vitro, may be involved in the conversion of VLDL remnants to LDL, accompanied by loss of apo E 39 (Figure 8). In this light, the high concentration of putative VLDL remnants in WHHL homozygotes 28 is not surprising. However, humans with homozygous familial hypercholesterolemia also

7 ROLE OF THE LIVER IN ATHEROSCLEROSIS Havel 575 _ ;% PERIPHERAL CELLS Dietary Cholesterol VLDL REMNANT \ / LDL B-10G E Receptor Figure 8. The formation of LDL from VLDL may involve two lipolytic steps. The first extrahepatic step, catalyzed by lipoprotein lipase, is well recognized. The second, which dearty involves further loss of triglycerides and phospholipids, as well as proteins other than apo B-100, may be catalyzed by hepatic lipase. The removal of both VLDL remnants and LDL from the blood depends mainly upon the activity of LDL receptors. According to this scheme, LDL formation will be promoted by increased activity of hepatic lipase or reduced activity of hepatic LDL receptors. have increased levels of IDL and remnant-like VLDL, and elevated concentrations of apo E (up to 30 mg/dl) which are proportional to the concentration of plasma total cholesterol (Havel RJ, Bilheimer DW, unpublished data). Moreover, the conversion of IDL to LDL is slowed in human homozygotes. 41 VLDL remnants may contribute to atherogenesis in receptor-deficient humans and rabbits; therefore, it cannot be assumed that LDL are the sole atherogenic lipoprotein in this situation. Besides causing elevated concentrations of atherogenic Iipoproteins (which can be taken up into the arterial intima), LDL receptor deficiency may contribute to atherogenesis by impeding the process of reverse cholesterol transport. In mammals that have an active cholesteryl ester transfer protein in blood plasma, such as rabbits and humans, the LDL receptor may be a key participant in reverse cholesterol transport. The cholesteryl esters that are carried by VLDL remnants and LDL in these species are derived largely from the action of lecithin-cholesterol acyltransferase. The findings of many investigators, including those of Christopher and Phoebe Fielding, 42 are consistent with the process of reverse cholesterol transport shown in Figure 9. Cholesterol, derived from cell surfaces or plasma Iipoproteins, is esterified on a minor species of HDL, from which it is transferred to other Iipoproteins, mainly remnants and LDL, that are taken up primarily into the liver. After lysosomal catabolism, the cholesterol can be excreted into the bile. This process seems to be impeded in many hyperlipidemic states, including familial hypercholesterolemia, 43 as a result of impairment of the transfer step. 44 At least in some cases, this abnormality is related to altered composition of the acceptor particles, VLDL and LDL. 45 LI.LK CF_L Figure 9. The primary pathway for reverse cholesterol transport in those species that have an active mechanism for transfer of cholesteryl esters among lipoprotein particles is shown here. It involves the movement of cholesteryl esters synthesized by lecithin-cholesterol acyltransferase (LCAT) from a fraction of HDL (to which the enzyme is bound) to acceptor Iipoproteins, which are mainly those that contain apo B. This produces a gradient that permits more cholesterol to move to the site of esterification from the surfaces of cells or other plasma Iipoproteins. The process of reverse transport is completed when the acceptor Iipoproteins are taken up by the liver via receptor-mediated endocytosis. In many hyperlipoproteinemic states, the ability of the acceptor Iipoproteins to receive the newly synthesized esters is impaired. This altered property seems to result from an alteration of the composition of the particle surface (see text). (Reprinted from reference 21, with permission of the publisher.) Intracellular Processing of Llpoproteins in Hepatocytes For several years, my associates and I have been studying the intrahepatic processing of remnants and LDL, especially that mediated by the LDL receptor of hepatocytes. Much of this work has been carried out in ethinyl estradiol-treated rats which, as shown by Petri Kovanen and others in Dallas 46 and by Yu-sheng Chao and several associates in San Francisco, 47 express up to 20 times the normal number of LDL receptors on hepatocytes. The intracellular pathway, defined originally by autoradiographic studies carried out in collaboration with Albert Jones et al., 48 seems to be the same as that found for other ligands taken up into hepatocytes by receptor-mediated endocytosis. 49 In the course of this work, we found that radioiodinated apo B of LDL accumulates in multivesicular bodies at the bile canalicular pole of the cell, close to the Golgi apparatus. We pursued this observation, with a view to determining the possible significance of this topo-

8 576 ARTERIOSCLEROSIS VOL 5, No 6, NOVEMBER/DECEMBER 1985 Figure 10. A negatively stained multivesicular body is shown here at high magnification (x 180,000). Rupture of its membrane (black and white arrow) has permitted the phosphotungstate stain to penetrate and outline the contents. These include collapsed internal vesicles (white arrowheads); numerous electron-lucent particles ( A in diameter), which represent remnants of triglyceride-rich lipoproteins; a few particles about 200 A in diameter, which may represent LDL, that were injected into the rat from which the organelles were isolated (black arrowheads); and occasional discoidal structures that may represent early stages of lipoprotein degradation (white arrows). This multivesicular body, from the liver of an estradiol-treated rat, is larger than most observed in normal rats (see text). (Reprinted from reference 50, with permission of the publisher.)

9 ROLE OF THE LIVER IN ATHEROSCLEROSIS Havel 577 graphic localization for recycling of the receptor and degradation of the lipoproteins. We have now 50 isolated multivesicular bodies from hepatocytes of estradiol-treated rats and have determined some of their properties. In these animals, the organelles are filled with triglyceride-rich lipoproteins with the properties of remnants (Figure 10). This observation is consistent with autoradiographic studies in untreated rats, in which we have shown 15 that chylomicrons and VLDL are processed in the liver by the same pathway found for LDL. The content lipoproteins of multivesicular bodies are largely undegraded; indeed, this organelle seems to represent the last prelysosomal way-station in which endocytosed lipoproteins accumulate just before acid hydrolases are delivered from the nearby Golgi apparatus within primary lysosomes, which fuse with the multivesicular bodies. 51 The traffic of chylomicron and VLDL remnants into multivesicular bodies of rat hepatocytes seems at first glance to be extraordinary, but our studies of this organelle are consistent with current knowledge of lipoprotein catabolism. In the rat, not only virtually all chylomicron remnants, but more than 90% of VLDL remnants, are taken up into the liver within minutes of their secretion M The ability to localize intact lipoproteins in multivesicular bodies and to isolate these structures is also clarifying our understanding of the role of the Golgi apparatus in processing endocytosed macromolecules and of the properties of nascent lipoproteins that are delivered to the Golgi apparatus from the endoplasmic reticulum and are finally concentrated in secretory vesicles. Although the role of the Golgi apparatus in receptor-recycling remains to be defined, ligands destined for lysosomal degradation do not themselves enter the Golgi compartment of the cell. Lipoprotein-filled structures in the Golgi region of hepatocytes, described by many investigators, may represent elements of distinct secretory or endocytic pathways, and we now recognize that Golgi-rich fractions of livers, as isolated by common methods, may be seriously contaminated by multivesicular bodies. It is difficult to distinguish secretory vesicles filled with nascent VLDL from multivesicular bodies filled with remnant lipoproteins by ordinary electron microscopic methods. Robert Hamilton et al. 50 have used modified staining techniques that aid this distinction. Multivesicular bodies, as the name implies, contain a number of internal bilayer vesicles, about 600 A in diameter. In hepatocytes these vesicles are obscured by the content lipoproteins unless the bilayer is heavily stained. The distinction is easier to make in livers of estradiol-treated rats because the multivesicular bodies are larger and the Golgi secretory vesicles smaller than in untreated animals. By use of a procedure for isolation of "intact" Golgi apparatus in which the secretory vesicles remain attached to the flattened cisternae, this organelle can be separated from multivesicular bodies. It is thus possible to isolate and characterize nascent and remnant lipoprotein populations from the same liver. Application of these methods promises to provide a new dimension to our understanding of the processes of hepatic lipoprotein secretion and catabolism. A current concept of the latter, based upon our studies and those of investigators who have studied the processing of other ligands by hepatocytes, is shown in Figure 11. Role of VLDL Remnants in Atherogenesls Until recently, it was generally considered that few VLDL remnants are taken up into the liver as such in humans. Rather, it was thought that the remnants (more or less equivalent to intermediate density lipoproteins) were first processed to LDL, which are taken up slowly by the liver and other tissues that express LDL receptors. Recent kinetic studies, however, suggest that humans may not differ so drastically in this respect from all other mammals, and that perhaps one-half or more of VLDL may normally be taken up into the human liver as remnants. 39 The efficiency of this uptake could be a major determinant of the concentration of atherogenic lipoproteins. Epidemiologic and other clinical observations have yielded conflicting data concerning the atherogenicity of VLDL (the concentration of which is closely related to plasma triglyceride levels). In Table 2,1 have listed some of the structural and metabolic properties of triglyceride-rich lipoproteins that could promote their atherogenieity. The last, but certainly not the least, is the activity of hepatic LDL receptors, which to a substantial extent determines whether VLDL remnants as well as LDL accumulate in the blood. Many years ago, Gofman proposed that small VLDL and IDL may be more atherogenic than LDL. 55 This idea was discounted by a cooperative study of lipoproteins in patients with coronary heart disease. 58 I do not believe, however, that subsequent studies, which in the main have measured as "LDLcholesterol" the fraction that includes IDL, have clearly demonstrated that "true LDL" (particles that contain apo B-100 as the sole protein) are the major atherogenic lipoprotein in human populations. In an ongoing intervention study, John Kane, Mary Malloy, Thomas Ports and I have found that the majority of asymptomatic familial hypercholesterolemia heterozygotes, whose average age is about 40 years, have little or no coronary atherosclerosis demonstrable by angiography. The determinants of premature atherosclerosis in these individuals must include factors other than the concentration of LDL. These could include additional lipoprotein determinants, such as the concentration of VLDL remnants. The role of hepatic LDL receptors in human lipoprotein catabolism has recently been shown dramatically by Bilheimer and his associates 57 in a young girl with homozygous familial hypercholesterolemia, in whom the desperate clinical situation required cardiac transplantation. The child also re-

10 ARTERIOSCLEROSIS VOL 5, No 6, NOVEMBER/DECEMBER 1985 Sinusoid Figure 11. This diagram, prepared by Robert Hamilton, depicts current concepts of the intracellular processing of endocytosed lipoproteins in hepatocytes. The initial binding occurs at sites on the sinusoidal surface of the cell, from which the receptor-ligand complex moves to a coated pit. Endocytosis ensues and the clathrin coat is rapidly lost. The primary endosomes appear to fuse and then migrate to the bile canalicular pole of the cell, where they are seen as multivesicular bodies (MVB). Primary lysosomes (L 1 ), derived from the nearby Golgi apparatus, fuse with the multivesicular bodies, converting them to secondary lysosomes (L 2 ), releasing enzymes that hydrolyze the lipid and protein components. The soluble products can then diffuse into the cytoplasm. The membrane of the endosomes, including multivesicular bodies, contains an ATP-drive protein pump, which acidifies the interior of these organelles to a ph of about 5.5. M The reduced ph serves to dissociate lipoproteins from the receptors and provides a suitable environment for the action of lysosomal enzymes. After dissociation, the receptor is recycled to the cell surface. The endosomal compartment from which dissociation occurs has been named CURL (compartment of uncoupling of receptor and ligand) and the process of dissociation has been visualized by immunocytochemical techniques for some nonlipoproteln receptors. 49 The precise site of dissociation and the pathway of recycling are poorly defined. BC = bile canaliculus. Table 2. Trlglyceride-Rich Lipoproteins: Potential for Atherogenlclty Property Content of cholesteryl esters Potential for receptor-mediated endocytosis Potential to form LDL Capacity to accept LCAT-derived cholesteryl esters Activity of hepatic vs extrahepatic LDL receptors Determinants Composition of nascent particles, residence time in the blood Loss of C apoproteins, surface lipid composition, content and reactivity of apo E Species of apo B, size/cholesteryl ester content, hepatic lipase activity Surface cholesterol concentration Cells' need for cholesterol

11 ROLE OF THE LIVER IN ATHEROSCLEROSIS Havel 579 ceived a liver homograft, and has since maintained lipoprotein levels only slightly above normal. Preoperatively, she had grossly elevated levels of IDL as well as LDL and cholesterol-enriched VLDL (Havel RJ, Bilheimer DW, unpublished data). Conclusions It is now recognized that the liver has a central role in regulating both the synthesis and the catabolism of the plasma lipoproteins. However, we lack some crucial pieces of information. Although we understand a good deal about the regulation of hepatic triglyceride synthesis and secretion, we still do not know what determines the rate of secretion of VLDL particles (i.e., the rate of synthesis and secretion of apo B). We are aware that hepatic lipoprotein receptors are subject to important regulatory influences related to cholesterol metabolism, but we do not know how the availability of cholesterol to the cell is coupled to receptor activity. We know that hepatic cholesterol homeostasis is regulated in part by biliary secretion of cholesterol and bile acids, but we do not know how these processes are regulated, nor do we understand the basis for the large species differences in the capacity to secrete these substances into the bile in response to the influx of dietary cholesterol in chylomicron remnants. We also know little about the determinants of variation of LDL receptor activity within and among species. Finally, we still do not know whether or how the liver participates in the formation of LDL. These problems pose interesting challenges for investigators of lipid and lipoprotein metabolism. If the accumulation of lipoproteins that contain apo B, with or without accompanying apo E, is as important for atherogenesis as many of us believe, the solution to these problems may be not only interesting, but also quite useful. References 1. Bragdon JH. Hyperiipemla and atheromatosls In a hibemator, Cttellus columbianus. Circ Res 1954;2: Behmer T, Havel RJ. Genesis of fatty liver and hyperiipemla in the fetal guinea pig. J Upid Res 1975; 16: Behmer T, Havel RJ, Long JA. Physiological fatty liver and hyperiipemla In the fetal guinea pig: chemical and ultrastructural characterization. J Upid Res 1972;13: Alexander CA, Hamilton Rl_ Havel RJ. Subcellular localization of "8" apoprolein of plasma lipoproteins in rat liver. J Cell Bio) 1976;69: Hamilton RL Hepatic secretion of nascent plasma lipoproteins. In: Glaumann H, Peters J, Redman C, eds. Plasma protein secretion by the liver. London: Academic Press, 1983: Guo LSS, Hamilton RL, Ostwald R, Havel RJ. Secretion of nascent lipoproteins and apolipoproteins by perfused livers of normal and cholesterol-fed guinea pigs. J Upid Res 1982; 23: NesteJ R, Havel RJ, Bezman A. Sites of initial removal of chylomicron triglyceride fatty acids from blood. J Clin Invest 1962;41: Bergman EN, Havel RJ, Wolfe BM, Behmer T. Quantitative studies of the metabolism of chylomicron triglycerides and cholesterol by liver and extrahepatic tissues of sheep and dogs. J Clin Invest 1971;50: Redgrave TG. Formation of cholesteryl ester-rich paniculate lipid during metabolism of chylomicrons. J Clin Invest 1970; 49: Havel RJ. Upoproteins and lipid transport. Adv Exp Med Blol 1975:63: Havel RJ. Functional activities of hepatic lipoprotein receptors. Annu Rev Physiol, 1985 in press 12. Brown MS, Goldstein JL. Familial hypercholesterolemia: defective binding of lipoproteins to cultured fibroblasts associated with impaired regulation of 3-hydroxy-3-methylglutaryl coenzyme A reductase activity. Proc Natl Acad Sci USA 1974:71: M oa OD, Faergeman O, Hamilton RL, Havel RJ. Characterization of remnants produced during the metabolism of triglyceride-rich lipoproteins of blood plasma and intestinal lymph in the rat. J Clin Invest 1975:56: Shore VG, Shore B. Heterogeneity of human plasma very low density lipoproteins. Separation of species differing In protein components. Biochemistry 1973;12: Jones AL, Hradek GT, Hornlck C, Renaud G, Windier EET, Havel RJ. Uptake and processing of remnants of chylomicrons and very low density lipoproteins by rat liver. J Upid Res : Shenill BC, Innerartty TL, Mahley RW. Rapid hepatic clearance of the canine lipoproteins containing only the E apoprotein by a high affinity receptor. J Biol Chem 1980;255: Windier E, Chao YS, Havel RJ. Regulation of the hepatic uptake of triglyceride-rich lipoproteins In the rat. J Biol Chem 1980:255: Falnaru M, Mahley RW, Hamilton RL, Innerartty TL. Structural and metabolic heterogeneity of beta-very low density lipoproteins from cholesterol-fed dogs and from humans with type III hyperiipoprotelnemia. J Upid Res 1982:23: Kane JP, Chen GC, Hamilton RL, Hardman DA, Malloy MJ, Havel RJ. Remnants of lipoproteins of intestinal and hepatic origin in familial dysbetalipoproteinemla. Arteriosclerosis 1983;3: Stalenhoef AFH, Malloy MJ, Kane JP, Havel RJ. Metabolism of apolipoproteins B-48 and P-100 of triglyceride-rich lipoproteins in patients with familial dysbetalipoproteinemia. Circulation 1984:70:11: Havel RJ. Approach to the patient with hyperiipidemia. Med Clin North Am 1982;66: Zllversmtt DB. Atherogenesis: a postprandial phenomenon. Circulation 1979:60: Thompson KH, Zltversmtt DB. Plasma very low density lipoprotein (VLDL) in cholesterol-fed rabbits: chylomicron remnants or liver lipoproteins? J Nutr 1983;113: Kane JP. Apolipoprotein B: structural and metabolic heterogeneity. Annu Rev Phystol 1983:45: Kovanen PT, Brown MA, Basu SK, Bllhelmer DW, Goldstein JL. Proc Natl Acad Sci USA 1981 ;78: Duff GL, McMillan GC. The effect of albxan diabetes on experimental cholesterol atherosclerosis in the rabbit. I. Inhibition of experimental cholesterol atherosclerosis In alioxan diabetes. J Exp Med 1949:89: Brecher P, Chobanian AV, Small DM, et al. Relationship of abnormal plasma lipoprotein to protection from atherosclerosis in the cholesterol-fed diabetic rabbit. J Clin Invest 1983; 72: Havel RJ, Ktta T, Kottte L, et al. Concentration and composition of lipoproteins in blood plasma of the WHHL rabbit. An animal model of human familial hypercholesterolemia. Arteriosclerosis 1982;2:467^ Ktta T, Goldstein JL, Brown MS, Watanabe Y, Hornlck CA, Havel RJ. Hepatic uptake of chylomicron remnants in WHHL rabbits: a mechanism genetically distinct from the low density lipoprotein receptor. Proc Nati Acad Sci USA 1982;79: Ktta T, Brown MS, Bllhelmer DW, Goldstein JL. Delayed clearance of very low density and intermediate density lipoproteins with enhanced conversion to low density lipoprotein in WHHL rabbits. Proc Nati Acad Sci USA 1982;79:

12 580 ARTERIOSCLEROSIS VOL 5, No 6, NOVEMBER/DECEMBER Angelin B, Raviola CA, Innerarity TL, Mahley RW. Regulation of hepatic lipoprotein receptors in the dog. Rapid regulation of apolipoprotein B, E receptors, but not of apolipoprotein E receptors, by intestinal lipoproteins and bile acids. J Clin Invest 1983;71: Bilhelmer DW, Watanabe Y, Kita T. Impaired receptor-mediated catabolism of low density lipoprotein in the WHHL rabbit, an animal model of familial hypercholesterolemia. Proc Natl Acad Sci USA 1982;79: Hornlck CA, Kita T, Hamilton RL, Kane JP, Havel RJ. Secretion of lipoproteins from the liver of normal and Watanabe heritable hyperlipidemic rabbits. Proc Natl Acad Sci USA 1983;80: Stalenhoef AFH, Malloy MJ, Kane JP, Havel RJ. Metabolism of apolipoproteins B-48 and B-100 of triglyceride-rich lipoproteins in normal and lipoprotein lipase-deficient humans. Proc Natl Acad Sci USA 1984:81: Kane JP, Sata T, Hamilton RL, Havel RJ. Apoprotein composition of very low density lipoproteins in human serum. J Clin Invest 1975:56: Innerarity TL, Pitas RE, Mahley RW. Receptor binding of cholesterol-induced high density lipoproteins containing predominantly apoprotein E in cultured fibroblasts with mutations at the low density lipoprotein receptor locus. Biochemistry 1980:19: Windier EET, Kovanen PT, Chao Ys, Brown MS, Havel RJ, Goldstein JL. The estradiol-stimulated lipoprotein receptor of rat liver: a binding site that mediates the uptake of rat lipoproteins containing apoproteins B and E. J Biol Chem 1980:255: Siidhof TC, Goldstein JL, Brown MS, Russel DW. The LDL receptor gene: a mosaic of exons shared with different proteins. Science 1985:288: Havel RJ. The formation of LDL: mechanisms and regulation. J Lipid Res 1984:25: Solniterova IB, Nikul-cheva NG. Lipoprotein lipase and liver triglyceride lipase in post-heparin blood plasma of rat and rabbit. Vopr Med Khim 1982:28: Soutar AK, Myant NB, Thompson GR. The metabolism of very low density and intermediate density lipoproteins in patients with familial hypercholesterolemia. Atherosclerosis 1982:43: Fielding CJ, Fielding PE. Cholesterol transport between cells and body fluids. Role of plasma lipoproteins and the plasma cholesterol esterification system. Med Clin North Am 1982:66: Fielding PE, Fielding CJ, Havel RJ, Kane JP, Tun P. Cholesterol net transport, esterification, and transfer in human hyperlipidemic plasma. J Clin Invest 1983:71: Fielding CJ. The origin and properties of free cholesterol potential gradients in plasma, and their relation to atherogenesis. J Lipid Res 1984;25: Fielding CJ, Reaven GM, Liu G, Fielding PE. Increased free cholesterol in plasma low and very low density lipoproteins in noninsulin-dependent diabetes mellitus and its role in the inhibition of cholesteryl ester transfer. Proc Natl Acad Sci USA 1984:81: Kovanen PT, Brown MS, Goldstein JL. Increased binding of low density lipoprotein to liver membranes from rats treated with 17-alpha-ethinyl estradiol. J Biol Chem 1979;254: Chao Ys, Windier E, Chen GC, Havel RJ. Hepatic catabolism of rat and human lipoproteins in rats treated with 17- alpha-ethinyl estradiol. J Biol Chem 1979;254: Chao Ys, Jones AL, Hradek GT, Windier EET, Havel RJ. Autoradiographic localization of the sites of uptake, cellular transport and catabolism of low density lipoproteins in the liver of normal and estradiol-treated rats. Proc Natl Acad Sci USA 1981:78: Schwartz A. The hepatic asialoglycoprotein receptor. CRC Crit Rev Biochem 1984:16: Hornick CA, Hamilton RL, Spaziani E, Enders GH, Havel RJ. Isolation and characterization of multivesicular bodies from rat hepatocytes: an organelle distinct from secretory vesicles of the Golgi apparatus. J Cell Biol 1985; 100: Jost-Vu E, Hamilton RL, Hornick CA, Belcher JO, Havel RJ. Fusion between primary lysosomes and multivesicular bodies (MVB) gives birth to secondary lysosomes. J Histochem Cytochem 1985 (in press) 52. Faergeman O, Havel RJ. Metabolism of cholesterol esters of rat very low density lipoproteins. J Clin Invest 1975;55: Faergeman O, Sata T, Kane JP, Havel RJ. Metabolism of apoprotein B of plasma very low density lipoproteins in the rat. J Clin Invest 1975:56: Van Dyke RW, Hornick CA, Belcher J, Scharschmidt BF, Havel RJ. Characterization of ATP-dependent proton transport by multivesicular bodies from rat liver. J Cell Biol 1984:99: Gofman J, Glazier F, Tamplin A, Strisower B, de Lalla O. Lipoproteins, coronary heart disease, and atherosclerosis. Physiol Rev 1954:34: The Technical Group and Committee on Lipoproteins and Atherosclerosis. Evaluation of serum lipoproteins and cholesterol measurements as predictors of complications of atherosclerosis. Circulation 1956:14: Bilheimer DW, Goldstein JL, Grundy SM, Starzl T, Brown MS. Liver transplantation to provide low-density-lipoprotein receptors and lower plasma cholesterol in a child with homozygous familial hypercholesterolemia. N Engl J Med 1984; 311: Index Terms: lipoprotein apoprotein cholesterol triglycerides chylomicrons LDL-receptor