The Study of Serum Proteins and Lipids with the Aid of the Quantity Ultracentrifuge (the Yale Journal Vol 24 No. 6 June 1952; Att: Progress Report Transport and Metabolism of Lipids)

Page 1

THE STUDY OF SERUM PROTEINS AND LIPIDS WITH VII. SOMEFEATURESOF A SYSTEMOF LIPOPROTEINSWHICH We propose in the present paper to describe a system of lipoproteins which contain phospholipid but no free cholesterol and little or no esterified cholesterol. W e utilize data which we have accumulated during the past four years during which time we have studied serum proteins and lipids by making multiple analyses of ten samples of each ultracentrifugate of untreated serum. The samples o€ centrifugate from standardized positions in the column are designated by numbers beginning with number one at the top, which is the cream layer sample.’ The detailed data have been published elsewhere,lV8or are in process of publication.””.”.’* The accumulated evidence from many sources, as reviewed by Edsall,’ indicates that serum lipids are in the main associated with proteins. Our data concerning the distribution of lipids and proteins in the ultracentrifugate provide both general and special support for that concept. Our finding, in low density centrifugate, of a concentration of globulin in excess of that expected from the albumin: globulin ratio of whole serum, along with high concentrations of lipids, strongly suggests that the globulin is there in increased concentration because of buoyancy provided by lipids attached in some fashion to the globulin molecule.’ Conversely, the finding of high concentrations of phospholipid and neutral fat in high density centrifugate indicates the attachment of these low density lipids in small proportions to many high density molecules. More specific information bearing upon this subject is provided by the systematic relationship between total globulin concentration and lipids at certain levels in the upper half of the column which we interpret as indicating stoichiometric relationships between proteins and lipids.” * From the Department of Medicine, School of Medicine, and the Biophysics Laboratory, Tulane University of Louisiana, New Orleans, Louisiana. This work was done in part under a contract from the Office of The Surgeon General, U. S. Army, and under the sponsorship of the Commissioii on Liver Disease, Armed Forces Epidemiot Some of the data in this paper were presented to the Association of American Physicians on May 6, 1952. $ Interne in Medicine, New Haven Hospital, 1924-1925. ** Present address : U.S. Naval Radiological Defense Laboratory, San Francisco, Received for publication May 28, 1952.

Page 2

PHOSPHOLIPID \VITHOUT FREE CHOLESTEROL 45 1 eIt is necessary that we define some of the ternis used in this paper, especially that we make clear which lipids seem in a strict sense to be a part of protein molecules. Only in the cream layer sample have we observed any systematic relationship between neutral fat and any other lipid component, while there is abundant evidence for such relationship among the other three lipid components we have measured in the top five samples of normal centrifugate. Because it is difficult to imagine that so much neutral fat is incorporated into the few protein molecules present in the cream layer sample, we are inclined to discount the stoichiometrical data for neutral fat even there and have adopted tentatively the generalization that neutral fat is attached to lipoprotein molecules by some physical bond rather than by incorporation into the molecule. We have come to use a system of terminology based on the assumption that the two forms of cholesterol and phospholipid are incorporated in different combinations in protein molecules and that these same molecules establish some physical union with neutral fat to Our evidence has led us to the hypothesis that there are two major systems of lipoproteins according to their lipid composition, apart from fat, and one minor system to which we assign a small fraction of cholesterol ester which has a special distribution in the centrifugate. We imagine that the lipoproteins in all of these systems may transport fat and we have evidence that the two major ones do so. The proportion of fat per unit of lipoprotein is believed to be highly variable, and in disease states and under ferences in fat load. The lipoprotein of greatest complexity of lipid composition, according to our evidence, is one which contains globulin, phospholipid, and both forms of cholesterol in stoichiometric proportions, and is found in normal centrifugates in two major concentrations, the cream layer sample and the mid-zone. In this latter zone the distribution curve along the column has the general shape of a normal frequency distribution curve, that of free cholesterol.' The proportion of cholesterol ester relative to free cholesterol differs in the two principal concentrations, that in the mid-zone being twice that in the top sample. The relationship between free cholesterol and phospholipid, however, does not vary significantly from one part of the column to another.' W e refer to the protein of one major system as jor system as phospholipid protein ( PL-protein). We emphasize the absence of free cholesterol from the protein of the second system and for the time being assume that this system contains no ester as well. The second system is therefore conveniently referred to also as that which has phospholipid without cholesterol. There is a possibility that two or more components of this system may actually contain the small fraction of cholesterol ester which has a distinctive distribution, but, partly for convenience in terminology and until more definite evidence is available, we assign this fraction dietary stress sedimentation characteristics may vary because of large dif- 4

Page 3

452 YALE JOURS AL OF BIOLOGY AND MEDICISE of ester to a third and separate system of lipoproteins which we call the cholesterol ester proteins (CE-proteins). This fraction of ester occurs in two concentrations, the larger deep in the column along and beyond the high density tail of the free cholesterol distribution curve, and the smaller one high in the column, chiefly in the third sample. In these two regions there is cholesterol ester in excess of that expected by the strict proportionality with free cholesterol exhibited in other parts of the column.' According to our terminology, union of any of these lipoproteins with fat FIG.1. Mean curves of distribution of phospholipid assigned to two systems in centrifugates of 18 healthy young men. The abscissa represents volume of centrifugate according to position in the centrifuge tube. Figures 1-10 indicate the centers of each of the ten standard samples. T h e centrifuge tube may be imagined as lying horizontally with the bottom toward the right and the top at the left. T h e vertical lines on either side of the curve for the PLprotein system indicate one standard deviation. Large values for S.D. in number 6 sample are possibly attributable to the fact that this sample was frequently collected in two parts with consequent increase in error. gives a protein-fat complex characterized by the specific lipoprotein and by localization in the column. The phospholipid proteins devoid of free cholesterol are the principal subject of the present paper. We shall give descriptive data for the PL-Cprotein system only when comparison is especially helpful in describing the system without cholesterol. Proof for the existence of lipoproteins which contain phospholipid but not free cholesterol is provided by the fact that about thirty-five per cent of lipid phosphorus is found in the high density one-third of the column of centrifugate where insignificant quantities of free cholesterol are found (Fig. 1). At higher levels where significant concentrations of free cholesterol occur, differentiation of the two systems of phospholipid-containing lipoproteins is entirely a mathematical procedure.

Page 4

PHOSPHOLIPID I VITH OUT FREE CIiOLESTEROL 453 \ve have found a linear relationship between concentration of free cholesterol and phospholipid for a group of samples from each of the upper five positions in the column of normal centrifugates.O In no instance does the line of regression pass through the origin, indicating phospholipid in excess of that necessary to meet the requirements set by slope of the line (Fig. 2 ) . This excess varies widely with position in the column and seems to be related to the mean concentration of phospholipid without free cholesterol. The general features of the graph for a given level in the column of centrifugate are similar to those published by Albrink, Man, and Peters for SAMPLE 2 FREE CHOCESTEROL FIG.2. Spot graphs in which values for free cholesterol are plotted against those for phospholipid (in milligrams per cent) for 18 samples from each of the three upper positions in the column, centrifugates of 18 healthy young men. (Transactions of the Tenth Conference on Liver Injury of the Josiah Macy, Jr. Foundation, X e w York, May 1951, Fig. 2, p. 19.) whole serum.' Our graph for a group of samples from some levels, for example the third from the top, shows points with such small abscissae and large ordinates as to exclude the possibility that the line might curve and pass through the origin (Fig. 2 ) .' The slope of the line differs somewhat for whole serum and for groups of samples representing different levels in the column but not significantly as evaluated by statistical methods. This demonstration of linear relationships with limited variation in slope at several levels after radical rearrangement of lipoproteins and with wide differences in intercept values is strong evidence for the existence in the upper five samples of centrifugate of phospholipid in two states, in one combined with free cholesterol in stoichiometric proportion and in the other uncombined. The concentration of phospholipid in the second state tends toward constancy at a given level in the column. The variation in concentration of phospholipid without free cholesterol in normal centrifugates at any one

Page 5

454 YALE JOL RNA L OF BIOLOGY ASD5fED ICI SE level, hoLvever, is believed to be sufficiently great to prevent highly accurate estimation of the constant which expresses the stoichiometrical relationship betiveen free cholesterol and phospholipid in the PL-C-protein. In the data used in the present paper we have assigned phospholipid to two systems, level by level, in centrifugates according to the formula for each level. After extensive application of this simplest possible interpretation of the systematic relationship between free cholesterol and phospholipid to centrifugate data from a wide variety of sources we have gained confidence in its truth, without any assumption of final accuracy for the mathematical procedure. Just as considerable reproducibility of concentration of phospholipid without free cholesterol in a group of samples from a given position in the column appears to be essential to demonstration of linear relationship between free cholesterol and phospholipid in centrifugate (the line of regression not passing through the origin), a similar constancy of concentration in whole serum might be essential for the relationship shown by Albrink, Man, and Peters.' Evidence that such is the case mill be presented in this paper. -4 constant multiplied by the concentration of free cholesterol gave concentration of phospholipid in the PL-C-protein system in a given sample of centrifugate. That assigned to the PL-protein system was obtained by subtraction from total concentration of phospholipid estimated in that sample. Phospholipid in each system in whole serum was calculated from data for the centrifugate. The two curves in Figure 1 show mean distribution of phospholipid in 18 centrifugates of healthy young men assigned to the two major systems of lipoproteins. The size of the area beneath one of these curves is proportional to the total concentration of phospholipids in that system in whole serum. The shape of the area describes the constituents of the system mainly according to density of lipoprotein complexes of which it is constituted. In the centrifugates of healthy young men there is considerable reproducibility of both size and shape of the areas representing both systems. As a result of stress of a large meal or of disease a system pattern may deviate significantly from the normal, either as to size or shape or in both ways. The mean concentration of phospholipid in the PL-C-proteins of whole serum of IS healthy young men was found to be SO mgs. per cent with a standard deviation of 21.1 and that in the PL-proteins (without free cholesterol) 150 mgs. per cent with a standard deviation of 16.4. When standard deviation is expressed as percentage of the mean, that for the PL-C-protein system is found to be more than twice that for the system without cholesterol, being 26% as compared to 11%. The tendency toward constancy of concentration of phospholipid in whole serum assigned to the system without free cholesterol as compared to the variability in the cholesterol-containing system is even more impressive when the3era of a group of miscellaneous patients are considered. Such a

Page 6

PHOSPHOLIPID WITHOUT FREE CHOLESTEROL 455 comparison is shown in Figure 3 which represents the results of analyses of centrifugates of 47 sera expressed as phospholipid assigned to the two systems. This group of patients includes a disproportionate number, 13, with diseases of the liver and biliary tract, two or three each with diabetes, nephrosis. and thyroid disease, and the remainder with miscellaneous diagnoses without particular reference to known disturbances in serum lipids. I n general, the graph shows a strong tendency toward wide scatter and increased concentration of phospholipid in the PL-C-protein system and a PATIENTS CO WIRED FIG.3. Comparison of phospholipid concentration assigned to the PL-Cprotein system (with cholesterol) with that assigned to the P L system (without cholesterol) as determined in centrifugates of 47 patients with tendency toward small scatter and diminished concentration of phospholipid in the system without free cholesterol. The normal ranges indicated by the cross-hatched areas were established by the maximum and minimum values obtained from study of 30 apparently healthy individuals, including a number of sera with concentration of neutral fat which exceeded our standards for normal and therefore were not included in the strictly normal series of 18 referred to above. In summary, our comparison of the two serum phospholipid compartments as to size indicates that the one without free cholesterol is the larger and more constant in the post-absorptive state in health, and that in disease it also varies little from normal size. When it does, it is predominantly toward diminution in the diseases we have studied. O n the other hand, phospholipid in the system with free cholesterol vanes more widely both in health and in disease, principally toward increase, containing occasionally in disease more than ten times as much phospholipid as 1 0 5 2 7 5 11 the system without cholesterol.

Page 7

456 YALE JOUR XAL OF BIOLIJGY AS D MEDICI KE As shown graphically in Figure 1, there is considerable reproducibility of distribution in the centrifugate of the different components of the system of lipoprotein which contains phospholipid n-ithout free cholesterol. As we have reported elsewhere,’ the reproducibility of specific gravity for our standard ten samples of centrifugate of the serum of healthy young men is the greatest of any measurement we have made. There is abundant evidence that the lipoprotein-containing particles tend to localize in the column from top to bottom according to increasing density of complexes. W e have FIG.4. Graph showing average position of a hypothetical plane which separates phospholipid in the PL-protein system into two equal cornpartments, 18 normal centrifugates. …