Proc. Nat. Acad. Sci. USA Vol. 72, No. 6, pp. 2347-2351, June 1975
A Genetic Determinant of the Phenotypic Variance of the Molecular Weight of Low Density Lipoprotein (hyperlipemia/atherosclerosis/Ag antigen of low density lipoprotein/lipoprotein metabolism)
WALDO R. FISHER, MARY G. HAMMOND, MARVIN C. MENGEL, AND GERMAINE L. WARMKE Departments of Medicine and Biochemistry, University of Florida, Gainesville, Florida 32610
Communicated by C. B. Anfinsen, AMarch 31, 1975 The molecular weight of monodisperse ABSTRACT human plasma low density lipoprotein has been measured in 69 individuals and found to vary over the range of 2.4 to 3.9 X 106. By contrast, the molecular weight of low density lipoprotein measured on two separate occasions for specific individuals shows a mean difference of 0.07 X 106 and a standard deviation of 0.08 X 106; hence low density lipoprotein differing in molecular weight by >0.2 X 106 may be considered different macromolecules. The distribution of the molecular weight of low density lipoprotein does not differ as a function of age or sex. Hyperlipemic subjects having monodisperse low density lipoprotein show similar molecular weight distribution to normal subjects, as do subjects with premature coronary artery disease. Family studies reveal a correlation coefficient of 0.82 between average molecular weights of parents and offspring, with significance at 0.01. In order to assess the influence of environment on molecular weight of low density lipoprotein, the correlation coefficient between the fathers' and mothers' low density lipoprotein was measured and no statistically significant correlation was found. These data are interpreted as strong evidence for a genetic determination of molecular weight of low density lipoprotein. A study of individuals in five families yields molecular weight data consistent with a single gene locus genetic mode of inheritance without dominance. The regression coefficient of the mean low density lipoprotein parental molecular weight on the offspring molecular weight is 0.30. If the variability of molecular weight is considered as an expression of phenotypic variance, then the regression analysis identifies 30% of this phenotypic variance as arising from additive gene action presumably at a single locus. Segregation in the family data is consistent. Since the differences in molecular weight of low density lipoprotein arise from differences in the amount of lipid bound to the apoprotein, it is likely that an additional portion of the phenotypic variance of the molecular weight results from individual variations in the metabolism of low density lipoprotein, which yield differences in lipid content. The individual variation in molecular weight is only approximately 5%; hence those metabolic sequences that influence molecular weight of low density lipoproteins must be precisely controlled.
Human plasma low density lipoproteins (LDL) from individual subjects exist in solution in either a inonodisperse or polydisperse state (1, 2). The present study concerns those subjects with monodisperse LDL, that is, LDL that is found to be present as a single, essentially homogeneous population of macromolecules when analyzed by sedimentation velocity ultracentrifugation or equilibrium banding in a density gradient. It has been shown in this laboratory that LDL in the monodisperse state is found most commonly in nonlipemic in-
dividuals or in subjects with hyperlipemias characterized by increased plasma concentrations of LDL (hyper-f3lipoproteinemias). In a previous study of a small sample of subjects with monodisperse LDL, the molecular weight (MW) of LDL was found to vary over a range of from 2.5 to 3.5 X 106; however, for a given subject, the MW remained constant (1). Analyses revealed that the weight of apoprotein per mole of lipoprotein was constant, irrespective of MW, and the differences in LDL MW were shown to result from different amounts of lipid associated with a constant quantity of
apoprotein. The present study was designed to explore further some of the physiologic variables that are instrumental in determining LDL MW. METHODS
The preparation of LDL from human serum has been described (1, 3). Serum from individual subjects was always obtained after an overnight fast. In brief, the serum was centrifuged in a Beckman L2-50 ultracentrifuge with a Ti-50 rotor at 45,000 rpm for 22 hr in order to remove the very low density lipoprotein (VLDL) which accumulated at the meniscus. The density of the infranatant was adjusted to 1.06 g/ml with solid KBr, and the solution was recentrifuged to float the LDL fraction, which was recovered at the meniscus of the tube. The methods used in measuring the MW of LDL have been reported in detail (1, 3). Briefly, isolated LDL was dialyzed against solutions of KBr buffered to pH 7.0 and having densities of 1.006, 1.06, and 1.20 g/ml. The protein content of the dialyzed LDL solutions was then adjusted to 1.5 or 3.0 mg/ml by dilution with solvent. The sedimentation coefficients of these LDL samples were measured in the analytical ultracentrifuge using double sector cells and centrifuging at 42,040 rpm and 250. The buoyant density of the lipoprotein was determined from a plot of -qS against solution density, and the flotation rate (SOPI.20) was measured by extrapolating to infinite dilution. From these data molecular weights of LDL were calculated as reported, using a frictional ratio of 1.11 (3). Protein concentration of the LDL samples was measured by the method of Lowry (4). Paper electrophoresis of the lipoprotein was performed by the technique of Hatch (5), and serum cholesterol and triglycerides were measured by standard methods. *
* Measurements of serum cholesterol and triglycerides were performed by the Clinical Laboratories at the University of Florida Teaching Hospital.
Abbreviations: LDL, plasma low density lipoprotein; VLDL, plasma very low density lipoprotein; MW, molecular weight. 2347
2348
Medical Sciences: Fisher et al.
Proc. Nat. Acad. Sci. USA 72 (1975)
12 11
0
0 ~~~
o~~
10
00
0
9
8 .1ft
4%
0
0
0
3R
0o
02
l
0
2.6 2.8 3.0 3.2 3.4 MLBCULAR WEIGHT x 1i6
0
00oo0 0
.17 6
*@0o0
5
0
*000 0 0 0
4
0
3
0
2
0
@
0
*
0
0
0
00
0000000 0 * *
0000- 00 r 000 -00, * o
0000000 - -
1
0
* 0
o
0
2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9
MOLECULAR WEIGHT x 10-6 FIG. 1. Molecular weight distribution of monodisperse LDL. 0, LDL from hyperlipemic subjects; 0, LDL from normal subjects. Inset: Molecular weight distribution of LDL for subjects with premature coronary artery disease.
Subjecis Subjects for this study were limited to those with monodisperse LDL distributions, as determined by the appearance of the schlieren pattern obtained during analytical ultracentrifugation. Normal subjects comprised a group of individuals who were in apparent good health and had normal serum lipid values. Subjects with hyper-p-lipoproteinemia were individuals with elevated serum cholesterol concentrations and either normal or moderately elevated triglyceride levels. 80
In each case, lipoprotein electrophoresis was performed, and these subjects were subclassified as either Type IIA or IIB (6). The family studies included both normal and hyper-#lipoproteinemic individuals; however, families having individuals with polydisperse LDL were excluded from the study. RESULTS
Previously reported data from this laboratory have demonstrated that the MW of LDL isolated on two or more occasions 0
75
70
0~~~~~~~ 0
65
0
60
00 oS
55 50
60
45
Y 40 35
0 0
I
000
0
0 .0
0
~~~0
30
0
0
0
0
25 0
20
0
so
15 10
5 0
0
0 0
0
0~
2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 MOLECULAR WEIGHT x 10-6
FIG. 2. Molecular weight distribution of LDL as a function of age and sex. 0, Male subjects; 0, female subjects.
Proc. Nat. Acad. Sci. USA 72
Genetic Determinant of LDL Molecular Weight
(1975)
TABLE 2. LDL molecular weight of parents and offspring (X 10 ) Offspring Father Mother 3.6 3.3 2.8 3.5 3.7 2.8 3.1 2.4 2.8 2.7
TABLE 1. The variation of LDL molecular uveight of individual subjects measured on two occasions Subject 1 2 3
4
5
Date of determination
MW X 10^
11/68 3/70 10/68 11/70 7/69 11/70 11/70 6/71 10/74 12/74
2.36 2.46 2.71 2.71 2.72 2.88 2.82 2.76 3.02 2.94
Date of Subdeterject mination 6 7/68 10/70 7 6/68 5/70 8 5/68 5/70 9 3/70 6/71 10 3/69 3/70
MW X 10^ 3.09 2.95 3.01 3.01 3.00 3.06 3.37
3.33 3.47 3.50
Mean of differences of paired determinations from individual subjects = 0.07 X 106. Standard deviation of differences of paired determinations from individual subjects = 0.08 X 106.
from a single individual appears to be relatively constant (1). In order to define the magnitude of the difference in the MW of LDL isolated from the same individual on two separate occasions, a study of 10 subjects was performed. Two samples of LDL were isolated from each fasting subject, and the measured MW are recorded in Table 1. The mean and the standard deviation of the differences between the two MW values for each subject were calculated, and are 0.07 X 106 and 0.08 X 106, respectively. Accordingly, in comparing subjects, one may assume that differences in MW of >0.2 X 106 define lipoproteins that truly differ in their MW, since these differences are greater than two standard deviations. The MW data reported in this study should be interpreted within these statistical limitations. The MW of LDL isolated in the fasting state has been measured in 69 individuals with monodisperse LDL, and is recorded in Fig. 1. The MW range is from 2.4 to 3.9 X 106, with the major concentration of values between 2.7 and 3.2 X 106. In this figure, the MW values for the hyperlipemic and normal subjects are identified separately, and the distribution of these populations is similar. Fig. 2 records the distribution of LDL MW as a function of age and sex. The mean MW for females is 3.1 X 106, and for males, 3.0 X 106. These values do not differ significantly. The data with respect to age fail to show any trend towards increasing or decreasing MW with advancing age, and the MW of LDL from one subject was measured on two occasions separated by six years with values of 3.5 and 3.4 X 106, which constituted no change. An assessment was made of the possible relationship between LDL MW and coronary artery disease. Ten subjects were chosen between the ages of 38 and 59 years (mean age, 47 years) who had survived at least one documented myocardial infarction and who had coronary angiographic evidence of atherosclerosis involving the three major coronary arteries. Their LDL MW are shown in the inset of Fig. 1, which reveals a distribution similar to that for the total
population. In order to evaluate the possibility that the MW of LDL might be genetically determined, family studies were undertaken. The parental and offspring MW data from 11 matings
2349
3.2 2.7 3.0 2.5 3.0 2.9 3.0 3.4
2.4 2.8 3.0 2.8 2.7 2.8 3.2 2.5
2.9 2.5 3.2 2.6 3.2 3.5 3.5 3.1
2.9 3.0 3.9
3.0
Correlation coefficient between Mother and Father: r = 0.32, 0.99 (not significant). Correlation coefficient between Parents and Offspring: r = 0.82, t = 4.3, P < 0.01. Regression coefficient of average Offspring on average Parent: b 0.30, t 4.3, P < 0.01. t
=
=
=
are recorded in Table 2. To determine the degree of resemblance of the offspring to their parents, a regression coefficient of the mean MW of the offspring on the mean parental MW was calculated. The regression coefficient was 0.30, and by t-test this value was significantly different from zero at the 0.01 level. A correlation coefficient between the mean parental and offspring MW was equally significant, with, a value of 0.82. To provide data on the influence of the environment on LDL MW in these families, a correlation coefficient between the MW of the fathers' and the mothers' LDL was calculated. A low value of 0.32 was determined, which statistically did not differ significantly from zero. This lack of correlation between parents implies that the external environment within which a family lives is not a determinant of LDL MW. Together, these sets of analysis provide strong evidence for the genetic determination of LDL MW. In order to explore further a genetic relationship, pedigrees from five families are shown in Fig. 3. A single gene locus with two alleles, one a determinant for high, the other for low MW, has been postulated as a model. Based on the LDL distribu-
tion of the total study population (Fig. 1), a MW of 3.3 X 106 greater is defined as an expression of a homozygous high MW genotype, while an LDL of 2.6 X 101 or less expresses a or
low MW homozygote. The intermediate values would constitute the heterozygous genotype. The data in Fig. 3 and also the one generation observations in Table 2 are consistent with this model, and there is no suggestion of allelic dominance. Further interpretation of these results will be considered in the Discussion. DISCUSSION
The variability of the MW of monodisperse LDL among 69 individual subjects has been documented to range between 2.4 and 3.9 million. LDL is a macromolecule that contains approximately four times as much lipid as protein, and yet for an individual subject repeated determinations of LDL MW reveal a variation of only 5%. Clearly this lipoprotein must be precisely structured, and the structure must differ from one
individual to the next.
2350
.~ ~C
Medical Sciences: Fisher et al.
B
Proc. Nat. Acad. Sci. USA 72
..........~~~~~~~~~~~~~~~~~~I. a....
FIG. 3.
LDL molecular
weights
of individual members of five
families. The data have been fitted to
genetic model, assuming MW values of
a
single-gene, two-allele
2.6
X
106
or
less
homozygote (11111II1); values of 3.3 greater, the high MW homozygote (~~); and values 3.2 X 1065, the heterozygote (~.0 Females; 0 males.
con-
stitute the low MW
X
or
of 2.7
to
106
What are the determinants of LDL MW? Age and sex of the subjects are not determinants.t Likewise there is no evident difference in the LDL MW distribution among normal and hyperlipemic subjects with monodisperse LDL. The family studies, however, provide strong evidence implicating genetic determination of LDL MW. In analyzing the family studies it is helpful to draw upon the concepts of quantitative genetics (8). If one looks upon LDL MW as a phenotype, then the LDL distribution is a sample of the phenotypic variance of the population (Vp). Vp is determined by the sum of the environmental variance (VE) plus the genetic variance (VG). Of the various factors contributing to VG, the additive variance (VA) is the one that is quantitatively measured by the regression analysis of the offspring on the mean parental LDL. The measured value of 0.30 signifies that approximately 30% of the observed LDL MW variance is due to additive gene action and, judging from t Lindgren et al. have previously studied 32 subjects and reported a slightly higher mean MW for females (7). The present study incorporates more individuals, and the subjects are limited to those with monodisperse LDL.
(1975)
the family segregation data, possibly arises from additive alleles at one locus. Though these family MW data are consistent with two alleles, as suggested, they do not exclude the possibility of a multiple allelic system or of a polygenic or multiple factor inheritance. This distinction, however, should become possible when the apoprotein of LDL has been sufficiently well characterized to determine whether LDL contains a single class of subunit proteins, with limited differences in primary structure reflecting allelic variation, or whether several different classes of proteins comprise the subunit structure of LDL. At the present time, the literature abounds with conflicting reports on the protein components of LDL. Workers in a number of laboratories have reported finding one or several lower MW proteins in LDL (9-14). By contrast, Smith, Dawson, and Tanford have reported the isolation of a single class of subunit proteins from LDL with a MW of around 250,000 (15), and this finding has now been confirmed.t If LDL contains only a single class of protein subunits of about 250,000 MW, then there must be two subunits per native LDL macromolecule, since the protein content of LDL is about 0.6 X 106 g/mole (1). Such a finding presents a compelling case for a single gene locus and supports a model having two alleles, one a determinant for high, the other for low MW LDL.§ In the past decade a number of immunologic studies have been reported describing LDL polymorphism and investigating the inheritance of the immunologically defined Ag antigens (16-20). The genetic data apparently do not permit the differentiation between a model consisting of closely linked but separate genes or a single locus model with multiple alleles. A possible relationship between the immunologically described Ag polymorphism and the MW data reported in this study has yet to be investigated; however, if only a single class of protein subunits is present in LDL, then it seems very probable that these two genetic systems are measuring different properties determined by the same set of alleles.
t R. B. Triplett and W. R. Fisher (1975) "The structural role of the apoprotein in low density lipoproteins of differing molecular weights," manuscript submitted for publication. § Assuming a single locus inheritance with two alleles, one may calculate gene frequency ratios from the data of Fig. 1, thus: 6 subjects low MW homozygote (mm) < 2.6 X 106 48 subjects heterozygote (mM) 2.7 - 3.2 X 106 15 subjects high MW homozygote (MM) > 3.3 X 106 69 subjects Total The calculated frequency of the m allele is 0.4347 and of the M allele, 0.5653. If the population mates at random, one would expect an allelic distribution of m2 + 2 mM + M2, and the predicted genotypic frequencies would be thus: Observed Predicted mm
mM MM
13 34 22
6 48 15
The agreement between observed and predicted values is close enough to provide additional support for the proposed genetic model. The observed data also raise the possibility of heterozygote advantage suggested by the increased heterozygote frequency.
Proc. Nat. Acad. Sci. USA 72
Genetic Determinant of LDL Molecular Weight
(1975)
Since the phenotypic variance is determined by both the genetic and environmental variance, it is necessary to consider the environmental variance as a major contributing factor in the 70% of phenotypic MW variance that is unexplained by additive genetic inheritance. LDL contains approximately 80% lipid, and it has previously been shown that MW differences in LDL result from variability in the quantity of lipid associated with the apoprotein (1). On the assumption that husband and wife live in a similar nutritional and physical environment, the finding that there is no correlation between their LDL MW suggests that the external environment is not a significant determinant. Current thinking envisions LDL as a remnant released during the catabolism of VLDL, a process which is accomplished enzymatically (21). Clearly, differences in the metabolic sequence involving the synthesis or catabolism of VLDL could give rise to differences in the lipid content of LDL. These differences would be reflected in LDL MW and would be recognized as environmental variance. The experimental observation that the MW of LDL for a given individual remains constant over many months to years, however, indicates that such metabolic sequences must be precisely controlled, showing little variation from day to day, when measured with the subject in the fasting state. It seems quite possible that through a genetic mechanism LDL is structured into high, low, and intermediate MW classes; however, within these classes modulating metabolic processes can alter the lipid content of LDL of a given individual producing additional lesser shifts in LDL MW. In summary, it has been shown that individual differences in LDL MW are determined in large part by an additive genetic mechanism. Family pedigree data are less extensive than desired, but the results are completely consistent with a single gene locus determination, and reports of two similar MW protein subunits in LDL strongly support this interpretation. It seems probable that differences in the metabolic sequence by which LDL is synthesized and degraded may well produce individual variability in the lipid content of LDL which is manifest as small but constant MW differences within groups of LDL having identical apoprotein constituents. The interpretation of the genetic data has been accomplished with the advice and counsel of Dr. F. C. Johnson of the Department of Zoology, The University of Florida. We are most appreciative of his help. Statistical analysis of the genetic data was performed by Dr. Ronald Marks of the Department of Statistics, The University of Florida. 1. Fisher, W. R., Hammond, M. G. & Warmke, G. L. (1972) "Measurements of the molecular weight variability of plasma low density lipoproteins among normals and subjects with hyper-fl-lipoproteinemia. Demonstration of macromolecular heterogeneity," Biochemistry 11, 519-525. 2. Hammond, M. G. & Fisher, W. R. (1971) "The characterization of a discrete series of low density lipoproteins in the dis-
3.
4.
5. 6.
7.
8.
9. 10. 11. 12.
13.
14.
15.
2351
ease, hyper-pre-fl-lipoproteinemia," J. Biol. Chem. 246, 5454-5465. Fisher, W. R., Granade, M. E. & Mauldin, J. L. (1971) "Hydrodynamic studies of human low density lipoproteins. Evaluation of the diffusion coefficient and the preferential hydration," Biochemistry 10, 1622-1629. Bailey, J. L. (1967) in Techniques in Protein Chemistry (Elsevier, New York), 2nd Ed., pp. 340-341. Hatch, F. T. (1964) in Serum Proteins and the Dysproteinemias, eds. Sunderman, F. W. & Sunderman, F. W., Jr. (Lippincott, Philadelphia), pp. 232-237. Beaumont, J. L., Carlson, L. A., Cooper, G. R., Fejfar, Z., Frederickson, D. S. & Strasser, T. (1970) "Classification of hyperlipidaemias and hyperlipoproteinaemias," Bull. W. H. 0. 43, 891-915. Lindgren, F. T., Jensen, L. C., Wills, R. 1). & Freeman, N. K. (1969) "Flotation rates, molecular weights and hydrated densities of the low-density lipoproteins," Lipids 4, 337-344. Falconer, D. S. (1960) Introduction to Quantitative Genetics (Ronald Press, New York), 36.5 pp. Shore, B. & Shore, V. (1967) "The protein moiety of human serum fl-lipoproteins," Biochem. Biophys. Res. Commun. 28, 1003-1007. Scanu, A., Pollard, H. & Reader, W. (1968) "Properties of human serum low density lipoproteins after modification by succinic anhydride," J. Lipid Res. 9, 342-349. Gotto, A. M., Levy, R. I. & Fredrickson, 1). S. (1968) "Preparation and properties of an apoprotein derivative of human serum 8-lipoprotein," Lipids 3, 463-470. Kane, J. P., Richards, E. G. & Havel, R. J. (1970) "Subunit heterogeneity in human serum beta lipoprotein," Proc. Nat. Acad. Sci. USA 66, 1075-1082. Lee, D. M. & Alaupovic, P. (1970) "Studies of the composition and structure of plasma lipoproteins. Isolation, composition, and immunochemical characterization of low density lipoprotein subfractions of human plasma," Biochemistry 9, 2244-22.52. Chen, C. H. & Aladjem, F. (1974) "Subunit structure of the apoprotein of human serum low density lipoproteins," Biochem. Biophys. Res. Commun. 60, 549-55)4. Smith, R., Dawson, J. R. & Tanford, C. (1972) "The size and number of polypeptide chains in human serum low density
lipoprotein," J. Biol. Chem. 247, 3376-3381. 16. Blumberg, B. S., Bernanke, D. & Allison, A. C. (1962) "A human lipoprotein polymorphism," J. Clin. Invest. 41,
1936-1944. 17. Hirschfeld, J. & Rittner, C. (1969) "Inheritance of the Ag(x), Ag(y), (al) and Ag(z) antigens," Vox Sang. 16, 146-
154. 18. Morganti, G., Beolchini, P. E., Butler, R., Brunner, E. & Vierucci, A. (1970) "Contribution to the genetics of serum 04-lipoproteins in man. IV. Evidence for the existence of the Ag,al/d and Ag/Ig, loci, closely linked to the AgXIY locus," Humangenetik 10, 244-253. 19. Albers, J. J. & Dray, S. (1968) "Identification and genetic control of two rabbit low-density lipoprotein allotypes," Biochem. Genet. 2, 25-35. 20. Rapacz, J., Grummer, R. H., Hasler, J. & Shackelford, R. M. (1970) "Allotype polymorphism of low density ,8-lipoproteins in pig serum (LDLpp 1, LDLpp 2)," Nature 225, 941-942. 21. Levy, R. I., Bilheimer, D. W. & Eisenberg, S. (1971) in Plasma Lipoproteins, ed. Smellie, R. M. S. (Academic Press, New York), pp. 3-17.
A Genetic Determinant of the Phenotypic Variance of the Molecular Weight of Low Density Lipoprotein (hyperlipemia/atherosclerosis/Ag antigen of low density lipoprotein/lipoprotein metabolism)
WALDO R. FISHER, MARY G. HAMMOND, MARVIN C. MENGEL, AND GERMAINE L. WARMKE Departments of Medicine and Biochemistry, University of Florida, Gainesville, Florida 32610
Communicated by C. B. Anfinsen, AMarch 31, 1975 The molecular weight of monodisperse ABSTRACT human plasma low density lipoprotein has been measured in 69 individuals and found to vary over the range of 2.4 to 3.9 X 106. By contrast, the molecular weight of low density lipoprotein measured on two separate occasions for specific individuals shows a mean difference of 0.07 X 106 and a standard deviation of 0.08 X 106; hence low density lipoprotein differing in molecular weight by >0.2 X 106 may be considered different macromolecules. The distribution of the molecular weight of low density lipoprotein does not differ as a function of age or sex. Hyperlipemic subjects having monodisperse low density lipoprotein show similar molecular weight distribution to normal subjects, as do subjects with premature coronary artery disease. Family studies reveal a correlation coefficient of 0.82 between average molecular weights of parents and offspring, with significance at 0.01. In order to assess the influence of environment on molecular weight of low density lipoprotein, the correlation coefficient between the fathers' and mothers' low density lipoprotein was measured and no statistically significant correlation was found. These data are interpreted as strong evidence for a genetic determination of molecular weight of low density lipoprotein. A study of individuals in five families yields molecular weight data consistent with a single gene locus genetic mode of inheritance without dominance. The regression coefficient of the mean low density lipoprotein parental molecular weight on the offspring molecular weight is 0.30. If the variability of molecular weight is considered as an expression of phenotypic variance, then the regression analysis identifies 30% of this phenotypic variance as arising from additive gene action presumably at a single locus. Segregation in the family data is consistent. Since the differences in molecular weight of low density lipoprotein arise from differences in the amount of lipid bound to the apoprotein, it is likely that an additional portion of the phenotypic variance of the molecular weight results from individual variations in the metabolism of low density lipoprotein, which yield differences in lipid content. The individual variation in molecular weight is only approximately 5%; hence those metabolic sequences that influence molecular weight of low density lipoproteins must be precisely controlled.
Human plasma low density lipoproteins (LDL) from individual subjects exist in solution in either a inonodisperse or polydisperse state (1, 2). The present study concerns those subjects with monodisperse LDL, that is, LDL that is found to be present as a single, essentially homogeneous population of macromolecules when analyzed by sedimentation velocity ultracentrifugation or equilibrium banding in a density gradient. It has been shown in this laboratory that LDL in the monodisperse state is found most commonly in nonlipemic in-
dividuals or in subjects with hyperlipemias characterized by increased plasma concentrations of LDL (hyper-f3lipoproteinemias). In a previous study of a small sample of subjects with monodisperse LDL, the molecular weight (MW) of LDL was found to vary over a range of from 2.5 to 3.5 X 106; however, for a given subject, the MW remained constant (1). Analyses revealed that the weight of apoprotein per mole of lipoprotein was constant, irrespective of MW, and the differences in LDL MW were shown to result from different amounts of lipid associated with a constant quantity of
apoprotein. The present study was designed to explore further some of the physiologic variables that are instrumental in determining LDL MW. METHODS
The preparation of LDL from human serum has been described (1, 3). Serum from individual subjects was always obtained after an overnight fast. In brief, the serum was centrifuged in a Beckman L2-50 ultracentrifuge with a Ti-50 rotor at 45,000 rpm for 22 hr in order to remove the very low density lipoprotein (VLDL) which accumulated at the meniscus. The density of the infranatant was adjusted to 1.06 g/ml with solid KBr, and the solution was recentrifuged to float the LDL fraction, which was recovered at the meniscus of the tube. The methods used in measuring the MW of LDL have been reported in detail (1, 3). Briefly, isolated LDL was dialyzed against solutions of KBr buffered to pH 7.0 and having densities of 1.006, 1.06, and 1.20 g/ml. The protein content of the dialyzed LDL solutions was then adjusted to 1.5 or 3.0 mg/ml by dilution with solvent. The sedimentation coefficients of these LDL samples were measured in the analytical ultracentrifuge using double sector cells and centrifuging at 42,040 rpm and 250. The buoyant density of the lipoprotein was determined from a plot of -qS against solution density, and the flotation rate (SOPI.20) was measured by extrapolating to infinite dilution. From these data molecular weights of LDL were calculated as reported, using a frictional ratio of 1.11 (3). Protein concentration of the LDL samples was measured by the method of Lowry (4). Paper electrophoresis of the lipoprotein was performed by the technique of Hatch (5), and serum cholesterol and triglycerides were measured by standard methods. *
* Measurements of serum cholesterol and triglycerides were performed by the Clinical Laboratories at the University of Florida Teaching Hospital.
Abbreviations: LDL, plasma low density lipoprotein; VLDL, plasma very low density lipoprotein; MW, molecular weight. 2347
2348
Medical Sciences: Fisher et al.
Proc. Nat. Acad. Sci. USA 72 (1975)
12 11
0
0 ~~~
o~~
10
00
0
9
8 .1ft
4%
0
0
0
3R
0o
02
l
0
2.6 2.8 3.0 3.2 3.4 MLBCULAR WEIGHT x 1i6
0
00oo0 0
.17 6
*@0o0
5
0
*000 0 0 0
4
0
3
0
2
0
@
0
*
0
0
0
00
0000000 0 * *
0000- 00 r 000 -00, * o
0000000 - -
1
0
* 0
o
0
2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9
MOLECULAR WEIGHT x 10-6 FIG. 1. Molecular weight distribution of monodisperse LDL. 0, LDL from hyperlipemic subjects; 0, LDL from normal subjects. Inset: Molecular weight distribution of LDL for subjects with premature coronary artery disease.
Subjecis Subjects for this study were limited to those with monodisperse LDL distributions, as determined by the appearance of the schlieren pattern obtained during analytical ultracentrifugation. Normal subjects comprised a group of individuals who were in apparent good health and had normal serum lipid values. Subjects with hyper-p-lipoproteinemia were individuals with elevated serum cholesterol concentrations and either normal or moderately elevated triglyceride levels. 80
In each case, lipoprotein electrophoresis was performed, and these subjects were subclassified as either Type IIA or IIB (6). The family studies included both normal and hyper-#lipoproteinemic individuals; however, families having individuals with polydisperse LDL were excluded from the study. RESULTS
Previously reported data from this laboratory have demonstrated that the MW of LDL isolated on two or more occasions 0
75
70
0~~~~~~~ 0
65
0
60
00 oS
55 50
60
45
Y 40 35
0 0
I
000
0
0 .0
0
~~~0
30
0
0
0
0
25 0
20
0
so
15 10
5 0
0
0 0
0
0~
2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 MOLECULAR WEIGHT x 10-6
FIG. 2. Molecular weight distribution of LDL as a function of age and sex. 0, Male subjects; 0, female subjects.
Proc. Nat. Acad. Sci. USA 72
Genetic Determinant of LDL Molecular Weight
(1975)
TABLE 2. LDL molecular weight of parents and offspring (X 10 ) Offspring Father Mother 3.6 3.3 2.8 3.5 3.7 2.8 3.1 2.4 2.8 2.7
TABLE 1. The variation of LDL molecular uveight of individual subjects measured on two occasions Subject 1 2 3
4
5
Date of determination
MW X 10^
11/68 3/70 10/68 11/70 7/69 11/70 11/70 6/71 10/74 12/74
2.36 2.46 2.71 2.71 2.72 2.88 2.82 2.76 3.02 2.94
Date of Subdeterject mination 6 7/68 10/70 7 6/68 5/70 8 5/68 5/70 9 3/70 6/71 10 3/69 3/70
MW X 10^ 3.09 2.95 3.01 3.01 3.00 3.06 3.37
3.33 3.47 3.50
Mean of differences of paired determinations from individual subjects = 0.07 X 106. Standard deviation of differences of paired determinations from individual subjects = 0.08 X 106.
from a single individual appears to be relatively constant (1). In order to define the magnitude of the difference in the MW of LDL isolated from the same individual on two separate occasions, a study of 10 subjects was performed. Two samples of LDL were isolated from each fasting subject, and the measured MW are recorded in Table 1. The mean and the standard deviation of the differences between the two MW values for each subject were calculated, and are 0.07 X 106 and 0.08 X 106, respectively. Accordingly, in comparing subjects, one may assume that differences in MW of >0.2 X 106 define lipoproteins that truly differ in their MW, since these differences are greater than two standard deviations. The MW data reported in this study should be interpreted within these statistical limitations. The MW of LDL isolated in the fasting state has been measured in 69 individuals with monodisperse LDL, and is recorded in Fig. 1. The MW range is from 2.4 to 3.9 X 106, with the major concentration of values between 2.7 and 3.2 X 106. In this figure, the MW values for the hyperlipemic and normal subjects are identified separately, and the distribution of these populations is similar. Fig. 2 records the distribution of LDL MW as a function of age and sex. The mean MW for females is 3.1 X 106, and for males, 3.0 X 106. These values do not differ significantly. The data with respect to age fail to show any trend towards increasing or decreasing MW with advancing age, and the MW of LDL from one subject was measured on two occasions separated by six years with values of 3.5 and 3.4 X 106, which constituted no change. An assessment was made of the possible relationship between LDL MW and coronary artery disease. Ten subjects were chosen between the ages of 38 and 59 years (mean age, 47 years) who had survived at least one documented myocardial infarction and who had coronary angiographic evidence of atherosclerosis involving the three major coronary arteries. Their LDL MW are shown in the inset of Fig. 1, which reveals a distribution similar to that for the total
population. In order to evaluate the possibility that the MW of LDL might be genetically determined, family studies were undertaken. The parental and offspring MW data from 11 matings
2349
3.2 2.7 3.0 2.5 3.0 2.9 3.0 3.4
2.4 2.8 3.0 2.8 2.7 2.8 3.2 2.5
2.9 2.5 3.2 2.6 3.2 3.5 3.5 3.1
2.9 3.0 3.9
3.0
Correlation coefficient between Mother and Father: r = 0.32, 0.99 (not significant). Correlation coefficient between Parents and Offspring: r = 0.82, t = 4.3, P < 0.01. Regression coefficient of average Offspring on average Parent: b 0.30, t 4.3, P < 0.01. t
=
=
=
are recorded in Table 2. To determine the degree of resemblance of the offspring to their parents, a regression coefficient of the mean MW of the offspring on the mean parental MW was calculated. The regression coefficient was 0.30, and by t-test this value was significantly different from zero at the 0.01 level. A correlation coefficient between the mean parental and offspring MW was equally significant, with, a value of 0.82. To provide data on the influence of the environment on LDL MW in these families, a correlation coefficient between the MW of the fathers' and the mothers' LDL was calculated. A low value of 0.32 was determined, which statistically did not differ significantly from zero. This lack of correlation between parents implies that the external environment within which a family lives is not a determinant of LDL MW. Together, these sets of analysis provide strong evidence for the genetic determination of LDL MW. In order to explore further a genetic relationship, pedigrees from five families are shown in Fig. 3. A single gene locus with two alleles, one a determinant for high, the other for low MW, has been postulated as a model. Based on the LDL distribu-
tion of the total study population (Fig. 1), a MW of 3.3 X 106 greater is defined as an expression of a homozygous high MW genotype, while an LDL of 2.6 X 101 or less expresses a or
low MW homozygote. The intermediate values would constitute the heterozygous genotype. The data in Fig. 3 and also the one generation observations in Table 2 are consistent with this model, and there is no suggestion of allelic dominance. Further interpretation of these results will be considered in the Discussion. DISCUSSION
The variability of the MW of monodisperse LDL among 69 individual subjects has been documented to range between 2.4 and 3.9 million. LDL is a macromolecule that contains approximately four times as much lipid as protein, and yet for an individual subject repeated determinations of LDL MW reveal a variation of only 5%. Clearly this lipoprotein must be precisely structured, and the structure must differ from one
individual to the next.
2350
.~ ~C
Medical Sciences: Fisher et al.
B
Proc. Nat. Acad. Sci. USA 72
..........~~~~~~~~~~~~~~~~~~I. a....
FIG. 3.
LDL molecular
weights
of individual members of five
families. The data have been fitted to
genetic model, assuming MW values of
a
single-gene, two-allele
2.6
X
106
or
less
homozygote (11111II1); values of 3.3 greater, the high MW homozygote (~~); and values 3.2 X 1065, the heterozygote (~.0 Females; 0 males.
con-
stitute the low MW
X
or
of 2.7
to
106
What are the determinants of LDL MW? Age and sex of the subjects are not determinants.t Likewise there is no evident difference in the LDL MW distribution among normal and hyperlipemic subjects with monodisperse LDL. The family studies, however, provide strong evidence implicating genetic determination of LDL MW. In analyzing the family studies it is helpful to draw upon the concepts of quantitative genetics (8). If one looks upon LDL MW as a phenotype, then the LDL distribution is a sample of the phenotypic variance of the population (Vp). Vp is determined by the sum of the environmental variance (VE) plus the genetic variance (VG). Of the various factors contributing to VG, the additive variance (VA) is the one that is quantitatively measured by the regression analysis of the offspring on the mean parental LDL. The measured value of 0.30 signifies that approximately 30% of the observed LDL MW variance is due to additive gene action and, judging from t Lindgren et al. have previously studied 32 subjects and reported a slightly higher mean MW for females (7). The present study incorporates more individuals, and the subjects are limited to those with monodisperse LDL.
(1975)
the family segregation data, possibly arises from additive alleles at one locus. Though these family MW data are consistent with two alleles, as suggested, they do not exclude the possibility of a multiple allelic system or of a polygenic or multiple factor inheritance. This distinction, however, should become possible when the apoprotein of LDL has been sufficiently well characterized to determine whether LDL contains a single class of subunit proteins, with limited differences in primary structure reflecting allelic variation, or whether several different classes of proteins comprise the subunit structure of LDL. At the present time, the literature abounds with conflicting reports on the protein components of LDL. Workers in a number of laboratories have reported finding one or several lower MW proteins in LDL (9-14). By contrast, Smith, Dawson, and Tanford have reported the isolation of a single class of subunit proteins from LDL with a MW of around 250,000 (15), and this finding has now been confirmed.t If LDL contains only a single class of protein subunits of about 250,000 MW, then there must be two subunits per native LDL macromolecule, since the protein content of LDL is about 0.6 X 106 g/mole (1). Such a finding presents a compelling case for a single gene locus and supports a model having two alleles, one a determinant for high, the other for low MW LDL.§ In the past decade a number of immunologic studies have been reported describing LDL polymorphism and investigating the inheritance of the immunologically defined Ag antigens (16-20). The genetic data apparently do not permit the differentiation between a model consisting of closely linked but separate genes or a single locus model with multiple alleles. A possible relationship between the immunologically described Ag polymorphism and the MW data reported in this study has yet to be investigated; however, if only a single class of protein subunits is present in LDL, then it seems very probable that these two genetic systems are measuring different properties determined by the same set of alleles.
t R. B. Triplett and W. R. Fisher (1975) "The structural role of the apoprotein in low density lipoproteins of differing molecular weights," manuscript submitted for publication. § Assuming a single locus inheritance with two alleles, one may calculate gene frequency ratios from the data of Fig. 1, thus: 6 subjects low MW homozygote (mm) < 2.6 X 106 48 subjects heterozygote (mM) 2.7 - 3.2 X 106 15 subjects high MW homozygote (MM) > 3.3 X 106 69 subjects Total The calculated frequency of the m allele is 0.4347 and of the M allele, 0.5653. If the population mates at random, one would expect an allelic distribution of m2 + 2 mM + M2, and the predicted genotypic frequencies would be thus: Observed Predicted mm
mM MM
13 34 22
6 48 15
The agreement between observed and predicted values is close enough to provide additional support for the proposed genetic model. The observed data also raise the possibility of heterozygote advantage suggested by the increased heterozygote frequency.
Proc. Nat. Acad. Sci. USA 72
Genetic Determinant of LDL Molecular Weight
(1975)
Since the phenotypic variance is determined by both the genetic and environmental variance, it is necessary to consider the environmental variance as a major contributing factor in the 70% of phenotypic MW variance that is unexplained by additive genetic inheritance. LDL contains approximately 80% lipid, and it has previously been shown that MW differences in LDL result from variability in the quantity of lipid associated with the apoprotein (1). On the assumption that husband and wife live in a similar nutritional and physical environment, the finding that there is no correlation between their LDL MW suggests that the external environment is not a significant determinant. Current thinking envisions LDL as a remnant released during the catabolism of VLDL, a process which is accomplished enzymatically (21). Clearly, differences in the metabolic sequence involving the synthesis or catabolism of VLDL could give rise to differences in the lipid content of LDL. These differences would be reflected in LDL MW and would be recognized as environmental variance. The experimental observation that the MW of LDL for a given individual remains constant over many months to years, however, indicates that such metabolic sequences must be precisely controlled, showing little variation from day to day, when measured with the subject in the fasting state. It seems quite possible that through a genetic mechanism LDL is structured into high, low, and intermediate MW classes; however, within these classes modulating metabolic processes can alter the lipid content of LDL of a given individual producing additional lesser shifts in LDL MW. In summary, it has been shown that individual differences in LDL MW are determined in large part by an additive genetic mechanism. Family pedigree data are less extensive than desired, but the results are completely consistent with a single gene locus determination, and reports of two similar MW protein subunits in LDL strongly support this interpretation. It seems probable that differences in the metabolic sequence by which LDL is synthesized and degraded may well produce individual variability in the lipid content of LDL which is manifest as small but constant MW differences within groups of LDL having identical apoprotein constituents. The interpretation of the genetic data has been accomplished with the advice and counsel of Dr. F. C. Johnson of the Department of Zoology, The University of Florida. We are most appreciative of his help. Statistical analysis of the genetic data was performed by Dr. Ronald Marks of the Department of Statistics, The University of Florida. 1. Fisher, W. R., Hammond, M. G. & Warmke, G. L. (1972) "Measurements of the molecular weight variability of plasma low density lipoproteins among normals and subjects with hyper-fl-lipoproteinemia. Demonstration of macromolecular heterogeneity," Biochemistry 11, 519-525. 2. Hammond, M. G. & Fisher, W. R. (1971) "The characterization of a discrete series of low density lipoproteins in the dis-
3.
4.
5. 6.
7.
8.
9. 10. 11. 12.
13.
14.
15.
2351
ease, hyper-pre-fl-lipoproteinemia," J. Biol. Chem. 246, 5454-5465. Fisher, W. R., Granade, M. E. & Mauldin, J. L. (1971) "Hydrodynamic studies of human low density lipoproteins. Evaluation of the diffusion coefficient and the preferential hydration," Biochemistry 10, 1622-1629. Bailey, J. L. (1967) in Techniques in Protein Chemistry (Elsevier, New York), 2nd Ed., pp. 340-341. Hatch, F. T. (1964) in Serum Proteins and the Dysproteinemias, eds. Sunderman, F. W. & Sunderman, F. W., Jr. (Lippincott, Philadelphia), pp. 232-237. Beaumont, J. L., Carlson, L. A., Cooper, G. R., Fejfar, Z., Frederickson, D. S. & Strasser, T. (1970) "Classification of hyperlipidaemias and hyperlipoproteinaemias," Bull. W. H. 0. 43, 891-915. Lindgren, F. T., Jensen, L. C., Wills, R. 1). & Freeman, N. K. (1969) "Flotation rates, molecular weights and hydrated densities of the low-density lipoproteins," Lipids 4, 337-344. Falconer, D. S. (1960) Introduction to Quantitative Genetics (Ronald Press, New York), 36.5 pp. Shore, B. & Shore, V. (1967) "The protein moiety of human serum fl-lipoproteins," Biochem. Biophys. Res. Commun. 28, 1003-1007. Scanu, A., Pollard, H. & Reader, W. (1968) "Properties of human serum low density lipoproteins after modification by succinic anhydride," J. Lipid Res. 9, 342-349. Gotto, A. M., Levy, R. I. & Fredrickson, 1). S. (1968) "Preparation and properties of an apoprotein derivative of human serum 8-lipoprotein," Lipids 3, 463-470. Kane, J. P., Richards, E. G. & Havel, R. J. (1970) "Subunit heterogeneity in human serum beta lipoprotein," Proc. Nat. Acad. Sci. USA 66, 1075-1082. Lee, D. M. & Alaupovic, P. (1970) "Studies of the composition and structure of plasma lipoproteins. Isolation, composition, and immunochemical characterization of low density lipoprotein subfractions of human plasma," Biochemistry 9, 2244-22.52. Chen, C. H. & Aladjem, F. (1974) "Subunit structure of the apoprotein of human serum low density lipoproteins," Biochem. Biophys. Res. Commun. 60, 549-55)4. Smith, R., Dawson, J. R. & Tanford, C. (1972) "The size and number of polypeptide chains in human serum low density
lipoprotein," J. Biol. Chem. 247, 3376-3381. 16. Blumberg, B. S., Bernanke, D. & Allison, A. C. (1962) "A human lipoprotein polymorphism," J. Clin. Invest. 41,
1936-1944. 17. Hirschfeld, J. & Rittner, C. (1969) "Inheritance of the Ag(x), Ag(y), (al) and Ag(z) antigens," Vox Sang. 16, 146-
154. 18. Morganti, G., Beolchini, P. E., Butler, R., Brunner, E. & Vierucci, A. (1970) "Contribution to the genetics of serum 04-lipoproteins in man. IV. Evidence for the existence of the Ag,al/d and Ag/Ig, loci, closely linked to the AgXIY locus," Humangenetik 10, 244-253. 19. Albers, J. J. & Dray, S. (1968) "Identification and genetic control of two rabbit low-density lipoprotein allotypes," Biochem. Genet. 2, 25-35. 20. Rapacz, J., Grummer, R. H., Hasler, J. & Shackelford, R. M. (1970) "Allotype polymorphism of low density ,8-lipoproteins in pig serum (LDLpp 1, LDLpp 2)," Nature 225, 941-942. 21. Levy, R. I., Bilheimer, D. W. & Eisenberg, S. (1971) in Plasma Lipoproteins, ed. Smellie, R. M. S. (Academic Press, New York), pp. 3-17.
