A Study of the Heparin-Manganese Chloride Method for Deter- mination of Plasma e-Lipoprotein Cholesterol Concentration T.T. ISHIKAWA, J.B. BRAZIER, P.M. STEINER, L.E. STEWART, P.S. GARTSlDE, and C.J. GLUECK, Chromatography Division, Lipoprotein Research Laboratory and General Clinical Research Center, University of Cincinnati, College of Medicine, Cincinnati, Ohio 45267

ABSTRACT

To assess the limits of the heparin- MnC12 precipitation method for quantita- t i o n of a - l i p o p r o t e i n c h o l e s t e r o l (C-HDL), effects of varying final MnC12 and heparin concentrations were studied, and the precipitation method was com- pared to preparative ultracentrifugation. I n 65 parallel plasma aliquots, C-HDL (X + SE) determined by ultracentrifuga- tion (54.3 + 1.8 mg/dl) correlated signifi- cantly ( r= 0.98, P < 0.001) with the precipitation method (56.0 + 1.9 mg/dl). C-HDL by ultracentrifugal and precipita- tion methods were also similar in 16 sub- jects with triglycerides ranging from 150 to 312 mg/dl (41.4 + 2.6, 43.4 + 2.8, r = 0.97, P < .001). A constant amount of c h o l e s t e r o l in the supernatant was measured over a final MnC12 range of 0.046-0.23 M, and cholesterol values in the supernatant at final MnC12 concentra- tions of 0.046, 0.05, and 0.055 M did not differ from each other, P > O. 1. However, cholesterol levels in the supernatant at final MnC12 concentration of 0.042 M differed from those at concentrations of 0.046, 0.05, and 0.055 M, P < 0.05, and the amount of supernatant cholesterol increased as the final MnC12 concentra- tion was reduced from 0.042 to 0.02 M. A constant amount of cholesterol in the supernatant was measured over a heparin c o n c e n t r a t i o n range of 92-734 USP units/ml. The final MnC12 and heparin concentrations of 0.046 M and 184 USP u n i t s / m l , which are incorporated in widely used procedures, gave C-HDL values for the precipitation method which were in close agreement with the ultra- centrifugal method. There is no evidence for a heparin-Mn++ precipitation of HDL and systematic underestimation of HDL by the precipitation method. However, the final MnC12 concentration is very

1A portion of this work was supported by General Clinical Research Center Grant No. RR00068-14, and a portion of it was done during Dr. Glueck's tenure as an Established Investigator of the American Heart Association, 1971-1976.

near the minimum required for accurate measurement of C-HDL. To preclude in- complete precipitation of low and very low density lipoproteins by insufficient manganese concentration, an increase of the manganese concentration should be considered.

I NTRODUCTION

A number of recent reports reinforce the i m p o r t a n c e of accurate measurement of C-HDL. In several early studies of lipoproteins (1-4), it was realized that low levels of a-lipo- protein cholesterol (C-HDL) and elevated levels of ~-lipoprotein cholesterol (C-LDL) were a s s o c i a t e d with premature atherosclerosis. Miller and Miller (5) and Castelli et al. (6) pro- posed that low plasma HDL concentration, independent of C-LDL or total cholesterol con- centration, may be associated with an increased incidence rate of coronary heart disease. In familial hyperalpha-lipoproteinemia, probands and affected relatives had distinctive elevations of C-HDL with no elevations of C-LDL; kindred members demonstrated elongated life expec- tancy and apparent rarity of premature cardiac events (7,8).

C-HDL has generally been measured using e i t h e r the heparin-manganese precipitation method of Burstein and Samaille (9) or as the lipoprotein cholesterol after ultracentrifugation of plasma at density 1.063g/ml (10). The heparin-manganese method measures the cho- lesterol content of the supernatant following precipitation of low and very low density lipo- proteins, and this quantity is often referred to as a-lipoprotein cholesterol (an electrophoretic term) or as high density lipoprotein cholesterol (an ultracentrifugal term). Measurement of the cholesterol content of the supernatant is the most commonly used method of quantifying (this supernatant), although the apolipoprotein concentration could also be quantitated as another approach. The ultracentrifugal fraction (d>1.063 g/ml) includes not only high density lipoprotein but also very high density lipopro- teins (V-HDL). The cholesterol content of both HDL and V-HDL is measured by the ultra- centrifugal methods (10) used comparatively in this study.

628

a-LIPOPROTEIN CHOLESTEROL 629

The accurate measurement of C-HDL is essential for the indirect determination of C-LDL by several ultracentrifugation tech- niques (10-12) and by the Friedewald formula (C-LDL = total cholesterol- C-HDL- triglyc- eride/5). Recently Srinivasan et al. (13)have suggested that heparin and high density lipo- proteins interact in the presence of Mn ++, with an overall effect that would lead to the precipi- tation of some C-HDL and a resultant under- e s t ima t ion of C-HDL after heparin-MnC12 precipitation. An additional problem with the hepa r in -MnC12 precipitation method also appears when the plasma triglycerides are elevated (usually ~300 mg/dl), and a "floating precipitate" appears in the supernate after addi- tion of heparin and manganese, providing an overestimation of C-HDL by the precipitation method as compared to d~1.063 g/ml ultra- centrifugation.

In the development of a micro-gas liquid chromatography method to measure C-HDL (14), it was noted that minute reductions of the final MnC12 concentrations below 0.046 M led to incomplete precipitation of LDL and VLDL. The current study was designed to assess effects of varying heparin and manganese chloride con- centrations on C-HDL quantitation and to provide a comparison of C-HDL measured by ultracentrifugation and precipitation (9-11 ).

MATERIALS AND METHODS

Subjects

Fasting plasma samples from 65 subjects studied in the outpatient section of the General Clinical Research Center were obtained using v a c u t a i n e r s containing crystalline disodium ethylenediaminetetraacetic acid (EDTA), and plasma was separated following the Lipid Research Clinic Protocols (11). Plasma choles- terol and triglyceride levels ranged, respectively, from 137 to 325 and from 43 to 312 mg/dl.

To assess the effects of varying the final con- centration of MnC12 and/or heparin on the completeness of the precipitation of LDL and VLDL, seven normal plasma pools with tri- glycerides ~200 mg/dl and five lipemic plasma pools with triglyceride ]>200 mg/dl were used (Fig, 1-4).

Comparison of C-HDL Ouantitation by Heparin-MnCI 2 Precipitation and Ultracentrifugation

To assess the accuracy of the heparin-MnCl 2 precipitation method (9-11), duplicate aliquots of 65 plasma samples were processed in parallel, with quantitation of C-HDL by precipitation and ultracentrifugation (1I) (Table I). The ultracentrifugal method was arbitrarily desig-

< [--

o

0

o

e- .o

o

,-1

6

0 0

o o o

o. o. o o o o v v v

o o o

~ ~ d d d

~ ~ 0 O 0

A ~ ~A "~, ~ = �9 ~ o A "."

E ~

~ N N

, - k

O

+

~

~_

p a g

~2

LIPIDS, VOL. 11, NO. 8

630 T.T. ISHIKAWA ET AL.

t I o ;

FIG. 1. Concentrations of cholesterol in the super- natant after precipitation of LDL and VLDL by heparin:manganese. Final heparin concentration con- stant at 184 USP units/ml, f'mal MnCI2 reagent con- centration varied from 0 to 0.23 M; (A) represents the MnC12 concentration commonly reported, 0.046 M.

1 2 3 4 : 5 7

FIG. 2. Agarose gel eleetrophoresis of the total plasma (1) and the supernatant after the precipitation of LDL and VLDL using a final concentration of heparin, 184 USP units/ml, with the following molar concentrations of MnC12: (2) 0.018, (3) 0.037, (4) 0.042, (5) 0.046, (6) 0.055, and (7) 0.23.

. . . . A A

,

g-i ! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I ~ E ~ I i N C ~ I ~ I A T ~ ~USP ~ I T S / M L )

FIG. 3. Concentrations of cholesterol in the super- natant after precipitation of LDL and VLDL by heparin:manganese. Final MnCI2 concentration (B) constant at 0.046 M, final heparin reagent concentra- tion varied from 0 to 1,468 USP units/ml; (B) repre- sents the heparin concentration commonly reported.

nated as the reference method in this compari- son.

Preparation of Ultracentrifugal Fractions of Plasma A 5 ml sample of plasma was adjusted to

background density 1.063 g/ml with solid KBr according to the following calculation (15):

X = V i (df - di)

1 - V df

in which X is the g solid KBr to be added, Vi is the initial volume of the plasma, df is the final density after adjustment, di is the initial density, and Vr is the partial specific volume of KBr. In a cellulose nitrate tube, the adjusted sample is over!aid with KBr solution, d = 1.063 g/ml, to fill the tube and then centri- fuged at 105,000 x g for 18 hr at 10 C. Lipo- protein fractions of d>1.063 g/ml were re- covered using a tube slicer to cut the cellulose nitrate tube (11). The fraction >1.063 g/ml was then assayed for cholesterol levels, using both the Technicon AA-II (11) and gas liquid chromatographic (GLC) methods (16).

Precipitation of LDL and VLDL

The following conditions were used to assess the effects of changes in final MnC12 and heparin concentrations on the precipitation of LDL and VLDL and the subsequent measure- ments of C-HDL:

1) Mn ++ concentration was varied while heparin concentration was held constant.

2) Heparin concentration was varied while Mn ++ concentration was held constant.

To determine the effects of varying MnC12 concentrations, the final heparin concentration was maintained at the commonly used 184 USP units/ml (9,17,18), while the final Mn ++ con- centration was varied from 0 to 0.23 M using MnC12 tetrahydrate (Mallinckrodt No. 6126, St. Louis, MO) (Figs. 1 and 2). Concentrations of Mn ++ in various MnC12 solutions were verified by titration with EDTA (19).

To determine the effects of heparin concen- tration variation on the precipitation of LDL and VLDL, analysis of the cholesterol in the resul t ing supernatant was performed at a constant final MnC12 concentration of 0.046 M, a final concentration which has been shown to be necessary for complete precipitation of VLDL and LDL (9,10). The heparin (Liquae- min Sodium "400," Organon, Inc., West Orange, NJ) final concentration was varied from 0 to 1,468 USP units/ml (Figs. 3 and 4). The potency of the sodium heparin preparation was 149 USP units/mg. The weight of heparin added to plasma aliquots for C-HDL (precipita- tion method) was 1.34 mg/ml plasma for the experiment comparing quantitation by precipi- tation and ultracentrifugation. Heparin from different sources does not always have a constant equivalency factor which can be used

LIPIDS, VOL. 11, NO. 8

c~-LIPOPROTEIN

to translate from USP units/ml to mg/ml. There may be considerable tool wt variation of com- mercial sodium heparins (20). In commercial hepann preparations, there can be a 6% varia- tion in USP units/mg between different lots and sou rces . Each individual heparin lot was checked for its ability to completely precipitate VLDL and LDL by agarose gel electrophoresis (11). These minor variations in heparin potency are probably not of great practical significance, however, since the commonly used concentra- tion of heparin (184 USP units/ml) is quite far removed from the low or the high concentra- tion limits.

In all precipitation reactions, volumes were held constant as follows: a) 1 ml plasma, b) 0.04 ml heparin, and c) 0.05 ml MnC12 solu- tion.

Measurement of C-HDL All samples were kept on ice throughout the

procedure. Heparin was added to the plasma and the sample vortexed. This was followed by the addition of MnC12, repeat mixing, and return to the ice bath for 30 rain. After sub- sequent centrifugation (1,600 x g at 4 C for 30 rain), the supernatant was removed with a Pasteur pipet. Cholesterol in the supernatant after precipitation and in the ultracentrifugal fraction d>1.063 g/ml was measured by the Technicon AA-II (11) and gas chromatographic methods (16). The AA-II method of the Lipid Research Clinics (11) uses a serum calibrator supplied by the Lipid Standardization Labora- tory of the Center for Disease Control, Atlanta, GA, whose purpose is to adjust AA-II results so that they are comparable with the manual Abell-Kendall reference method. Conditions for the GLC cholesterol method were the same as previously reported (16). C-HDL quantitated by GLC was essentially identical to C-HDL measured by AA-II, and, for consistency, results summarized in Table I are for the AA-II method.

Lipoprotein Electrophoresis Thin agarose gel electrophoresis of the super-

natant following precipitation of LDL and VLDL was carried out using a commercially available kit (ACI-Corning), which is a modifi- cation of the method of Noble (21). The agarose plates were stained with fat red 7B. Electrophoresis was also used to confirm proper gradient separation in the ultracentrifugal frac- tions of plasma. Ten mg/dl LDL cholesterol or less could be detected by the gel electro- phoresis.

I mmunodiffusion

Immunodiffusion studies were carried out in

CHOLESTEROL 631

FIG. 4. Agarose gel electrophoresis of the total plasma (1) and the supernatant after the precipitation of LDL and VLDL using a final Mna2 concentration of 0.046 M and heparin in USP units/ml: (2) 46, (3) 92, (4) 184, (5) 642, (6) 734, and (7) 1468.

1% agarose (Sigma A6877) in a pH 7.0 buffer, using a center well containing antibeta-lipo- p r o t e i n a n t i b o d y ( B e h r i n g Diagnostics 14-512-301). Diffusion was carried out for 48 hr in a moist chamber at 10 C. Precipitation lines elicited by the unknown samples were compared to those formed by standards which were made by the serial dilution of sera with ultracentnfugally determined LDL concentra- tions. The sensitivity of the immunodiffusion was 5 mg/dl C-LDL was determined by use of preassayed standards.

RESULTS

Comparison of C-HDL Quantitation by Precipitation and Ultracentrifugation

In 65 samples of plasma, with triglycerides ranging from 42 to 312 and cholesterols from 137 to 325 mg/dl, ultracentrifugaUy quanti- tated C-HDL (.X + SE) was 54.3 + 1.8 mg/dl, while C-HDL measured by precipitation in parallel aliquots was 56.0 + 1.9 mg/dl (Table I). C-HDL measured by ultracentrifugal and precipitation methods was highly correlated, r = 0.982, P<.001 (Table I). Comparison of the two methods for C-HDL quantitation was also carried out for 14 subjects with familial hyper- a lpha- l ipoproteinemia (7.8), whose C-HDL levels were />70 mg/dl. In these 14 patients, C-HDL by ultracentrifuge was 74.6 + 2.5, not significantly different from 77.4+ 2.7 by precipitation, and the two methods were again highly correlated, r = 0.941, P<.001. Plasma samples from 16 subjects with tnglycerides ranging from 150-312 mg/dl were also eval- uated using the precipitation and ultracentri- fugal methods to quantitate C-HDL. Mean (+ SE) C-HDL by ultracentrifugation and precipi- tation were, respectively, 41.4 + 2.6 and 43.4 + 2.8 mg/dl, and were closely correlated, r = 0.97, P<.001 (Table I).

Effects of Varying MnCI 2 Concentration with Heparin Constant

A constant amount of cholesterol in the

LIPIDS, VOL. 11, NO. 8

632 T.T. ISHIKAWA ET AL.

supernatant was measured over a final Mn ++ concentration range of 0.046 to 0.23 M using both AA-II and GLC methodologies (Fig. 1). When f i n a l Mn ++ c o n c e n t r a t i o n s were <0.046 M, however, the amount of cholesterol in the supernatant increased, with sharp incre- ments observed as final MnC12 concentration was reduced from 0.042 to 0 .02M (Fig. 1). Agarose gel electrophoresis of the supernatant in samples where Mn ++ concentrations were ~>0.046 M revealed HDL as the sole identifiable lipoprotein (Fig. 2). Immunodiffusion studies of these same supernatants revealed no re- maining LDL when Mn ++ was ~0.046 M.

Two-factor analysis of variance (22) (using the least squares solution of the general linear hypotheses, Method I) confirmed that choles- terol in the supernatant (C-HDL) reached a constant level and that LDL and VLDL were completely precipitated at final Mn ++ concen- trations ~0.046M. Cholesterol levels in the supernatant varied significantly over final con- centrations of 0.03-0.055 M Mn ++ for both the colorimetric method (F = 16.98, df = 4/30, P<0.0001) and for GLC (F = 9.68, df = 4/30, P<0.0001). A t-test probe of the data indicated that cholesterol values in the supernatant con- taining Mn ++ concentrations of 0.046, 0.05, and 0.055 M did not differ significantly from each other for both cholesterol assay methods. However, one sided t-test of the cholesterol level of the supernatant containing 0.042 M Mn ++ differed significantly from the concentra- tions of 0.046, 0.05, and 0.055 M for both AA-II (t = 8.24, df = 50, cz < 0.001) and GLC (t = 5.28, d f= 50, a < 0.001). Five aliquots of plasma were analyzed for C-HDL at the final Mn ++ concentrations of 0.034, 0.042, 0.046, and 0.048 M. Mean (+ SE) C-HDL levels were, respectively, 54.4 + 1.7, 45.6 + 0.9, 44.8 +-05, and 46.0 + 0.0 mg/dl, indicating reproducibility of C-HDL quantitation even at final MnC12 con- centrations (0.034M) which did not provide complete precipitation of VLDL and LDL.

Effects of Varying Heparin Concentration with MnCI 2 Held Constant

A constant amount of cholesterol in the supernatant was measured over a heparin con- centration range of 92-734 USP units/ml (Fig. 3). At the final heparin concentrations <92 and >734 USP units/ml, precipitation of LDL and V L D L was incomplete; consequently, the amount of cholesterol in the supernatant was increased. Agarose gel electrophoresis of the supernatant obtained after the addition of various heparin concentrations revealed only HDL when the concentration was between 92 and 642 USP units/ml (Fig. 4). For heparin

concentrations <92 and >734 USP units/ml, electrophoresis revealed unprecipitated LDL and VLDL (Fig. 4).

DISCUSSI ON

As initially shown by Burstein and SamaiUe (9) and confirmed by Fredrickson et al. (10), a final concentration of MnCI 2 > 0 . 0 4 6 M is required for complete precipitation of VLDL and LDL for subsequent accurate measurement of C-HDL. The current report indicates, as did a previous pilot study (14), that this final MnCI 2 concentration is very near the minimum con- centration required for accurate measurement of C-HDL (0.042 M). Minute inaccuracies in preparation of the MnC12 solution (often caused by hygroscopic qualities of MnC12 tetra- hydrate) which reduce final MnC12 concentra- tion to <0.046 M result in incomplete precipi- tation of LDL and VLDL (Figs. 1 and 2). At final Mn ++ concentrations ~<0.042 M, choles- terol in the supernatant is reproducibly quanti- tated but is systematically and inaccurately elevated above levels obtained by the use of Mn ++ final concentrations >/0.046 M. When the final concentration of MnC12 is increased over a range from 0.046 to 0.23 M, a constant amount of C-HDL is measured. In sharp contrast, two- fold increases and decreases of the heparin con- centration often used (9) do not adversely affect precipitation of LDL and VLDL. An increase of the final MnC12 concentration (>0 .06M) should be considered to minimize the effects of minor inaccuracies in the prepara- t i o n of MnC12 solution or in volumetric measurements which might affect the final Mn ++ concentration and thereby the C-HDL determination.

Srinivasan et al. (13) have reported that in the presence of 0.025-0.075 M Mn ++, over 60% of HDL is associated with heparin which is attached to Sepharose. This report (13)fur ther indicates that 25% of the total HDL produces an insoluble complex with free heparin in the presence of manganese. This would lead to a systematic underestimation of C-HDL by the precipitation method as compared to the ultra- centrifugal method (which also includes V-HDL cholesterol). In the present study, however, the comparison of C-HDL in 65 plasma samples quantitated by the two methods revealed not only a highly significant correlation but means which were very close for the two procedures. Similar results were previously obtained by Wilson et al. (18).

Another potential problem involves plasma samples containing "sinking prebeta-lipopro- rein," the Lp(a) antigen, an antigenic variant of

LIPIDS, VOL. 11, NO. 8

cx-LIPOPROTEIN CHOLESTEROL 633

LDL with a densi ty range over lapping tha t o f HDL (23,24). A por t ion of Lp(a) would be recovered in the ul t racentr i fugal ly f rac t iona ted plasma at densi ty > 1 . 0 6 3 g/ml. In the occa- sional plasma samples where Lp(a) concent ra- t i o n is high, u l t racentr i fugal ly de t e rmined C-HDL might be cons i s ten t ly higher t han C-HDL de t e rmined by precipi ta t ion. Al though the presence or absence of Lp(a) was no t deter- mined in the 65 samples in this s tudy, the notab le similari ty be tween C-HDL (ul t racentr i - fuge) and C-HDL (prec ip i ta t ion) suggests tha t Lp(a) p robab ly does no t play a major role which would render measuremen t of C-HDL inaccurate .

This s tudy conf i rms the adequacy of the final hepar in and Mn ++ concen t ra t ions sug- gested for prec ip i ta t ion of LDL and VLDL (9,10) and indicates tha t a conservat ive incre- ment in final Mn ++ concen t r a t ion would give a greater margin of safety for accurate measure- ments . Accura te de te rmina t ion of C-HDL is impor t an t , n o t only for es t imat ion of C-LDL (12), but also in evaluat ion of the role of C-HDL in familial hypera lpha- l ipopro te inemia (6,7) and as an "ant i - r i sk" factor in a thero- sclerosis (6-8).

ACKNOWLEDGMENTS

The authors would like to acknowledge Joseph Lukach, Susan Shaw, Grace Evans, Edward Rooch, Robert Yeakle, Susan Rasche, and Cynthia Beatty for their excellent assistance in the preparation of this manuscript.

REFERENCES

1. Nikkila, E., Scand. J. Clin. Lab. Invest., Suppl. 5:1-101 (1953).

2. Gofman, J.W., H.B. Jones, F.T. Lindgren, T.P. Lyon, H.A. Elliot, and B. Strisower, Circulation 2:161 (1950).

3. BAIT, D.P., E.M. Russ, and H.A. Eder, Am. J. Med. 1 1 : 4 8 0 (1951).

4. Jencks, W.P., M.R. Hyatt, M.R. Jetton, T.W.

Mattingly, and E.L. Durrum, J. Clin. Invest. 35:980 (1956).

5. Miller, G.J., and N.E. Miller, Lancet 1 (7897):16 (1975).

6. Castelli, W.P., J.T. Doyle, T. Gordon, C. Hames, S.B. Hulley, A. Kagan, D. McGee, W.J. Vicic, and W.J. Zukel, Circulation 51,52:Supplement II, I1-96 (1975).

7. Glueck, C.J., R.W. Fallat, F. Millett, P. Gartside, R.C. Elston, and R.C.P. Go, Metabolism 24:1243 (1975).

8. Glueck, C.J., R.W. Fallat, and M. Spadafora, Circulation 51,52: Supplement II, 11-272 (1975).

9. Burstein, M., and J. Samaille, Clin. Chim. Acta 5:609 (1960).

10. Fredrickson, D.S., R.I. Levy, and F.T. Lindgren, J. Clin. Invest. 47:2446 (1968).

11. Manual of Laboratory Operations, Lipid Research Clinics Program, Vol. 1, Lipid and Lipoprotein Analysis, HEW Publication (NIH) 75-628 (1974).

12. Friedewald, W.T., ILl. Levy, and D.S. Fredrick- son, Clin. Chem. 18:499 (1972).

13. Srinivasan, S.R., B. Radhakrishnamurthy, and G.S. Berenson, Arch. Biochem. Biophys. 170:334 (1975).

14. Lutmer, R.F., D. Parsons, C.J. Glueck, J.A. Mor- rison, L. Stewart, J.B. Brazier, C.R. Buncher, and T.T. Ishikawa, J. Lipid Res. 15:611 (1974).

15. Hatch, F.T., and R.S. Lees in "Advances in Lipids Research," Edited by R. Paoletti and D. Kritchevsky, Academic Press, New York, NY, 1968, p. 35.

16. Ishikawa, T.T., J. MacGee, J.A. Morrison, and C.J. Giueck, J. Lipid Res. 15:286 (1974).

17. Burstein, M., and H.R. Scholnick, in "Advances in Lipid Research," Edited by R. Paoletti, and D. Kritchevsky, Academic Press, New York, NY, 1973, pp. 67-108.

18. Wilson, D.E., and M.J. Spiger, J. Lab. Clin. Med. 82:473 (1973).

19. Flaschka, H., and A.M. Amin, Mikrochim. Acta 4141:420 (1953).

20. Nader, H.B., N.H. McDuffie, and C.P. Dietrich, Biochem. Biophys. Res. Commun. 57:2 (1974).

21. Noble, R.P., F.T. Hatch, J.A. Mazrimas, F.T. Lindgren, L.C. Jenson, and G.L. Adamson, Lipids 4:55 (1969).

22. Overall, J.E., and D.tC Spiegel, Psycbol. Bull. 72:311 (1960).

23. Rider, A.K., R.I. Levy, and D.S. Fredriekson, Circulation 42:10 (1970).

24. Ehnholm, C., H. Garoff, K. Simons, and H. Aro, Biochim. Biophys. Acta 236:431 (1971).

[Received March 5, 1976]

LIPIDS, VOL 11, NO. 8