Proc. Natl. Acad. Sci. USA
Vol. 76, No. 9, pp. 4646-4649, September 1979
Apolipoproteins in human cerebrospinal fluid (lipoproteins/electroimmunoassay/interstitial fluid/capillary permeability)
PAUL S. ROHEIM*, MICHAEL CAREYt, TRUDY FORTEt, AND GLORIA L. VEGA* *Department of Physiology and tDepartment of Neurosurgery, Louisiana State University Medical Center, New Orleans, Louisiana 70119; and *Donner Laboratory, University of California, Berkeley, California 94720
Communicated by Alex B. Novikoff, June 8, 1979
ABSTRACT The presence of apolipoproteins A-I, E, C-II, and GIII and the absence of apolipoprotein B was demonstrated in human cerebrospinal fluid. The concentration of apolipoproteins was measured by electroimmunoassay. Apolipoproteins E, C-I1, and C-Il were present in cerebrospinal fluid at 3-5% of their concentration in plasma; the cerebrospinal fluid level of apolipoprotein A-I was 0.4%. Most of the cerebrospinal fluid apolipoproteins were present in the p < 1.21 g/ml i protein fraction. The major apoli proteins of cerebrospinal fluid are E and A-I. The possible mechanism of transfer and the physiological and pathophysiological role of apolipoproteins in cerebrospinal fluid are postulated.
Lipoproteins are macromolecular complexes with Mrs ranging from 107 for very low density lipoproteins (VLDL), 2-3 X 106 for low density lipoproteins (LDL), and 2-4 X 105 for high density lipoproteins (HDL) (1, 2). The protein moieties of the lipoproteins, the apolipoproteins (apo), have specific physiological functions, and some of their alterations are associated with pathological conditions (1-3). The enzyme lecithin cholesteryl acyltransferase requires the presence, as a cofactor, of apo A-I which has a Mr of 28,000 (1, 2). Recently, a cholesteryl ester transfer protein has been described and it is thought to be one of the minor apos, apo D (Mr, 22,000) (4). The action of lipoprotein lipase is modulated by the apo Cs which have Mrs of approximately 7000-10,000 (1, 2). Cellular cholesterol metabolism is regulated by apo B (Mr, 260,000) and apo E (Mr) 33,000) (1, 2,5, 6). Because interstitial fluid is in direct contact with the cell, knowledge of its apo composition and concentration is important in understanding the regulation of cellular lipid metabolism. Lipoprotein and apo profiles of plasma are being extensively investigated, but information on the apo composition of interstitial fluid is sparse. Apo B concentration in human peripheral lymph is 10% of its concentration in plasma (7) and it is biologically active at that concentration (8). It has been postulated that apos are present in the circulation not only as part of the lipoprotein particle but also as "free" apos-i.e., not associated with a conventional lipoprotein particle (9). Studies with rat renal lymph have shown that its apo composition is different from that in plasma but it is similar to the apo composition of the lipoprotein-free p > 1.21 g/ml fraction (10). The dominant process in the formation of interstitial fluid is filtration of plasma. The concentrations of macromolecules in the interstitial fluids, lymph, and cerebrospinal fluid are inversely related to their Mr (11). Because the permeability of the blood cerebrospinal fluid barrier is much more restricted than the permeability of peripheral capillaries, we chose to
examine whether apos are present in cerebrospinal fluid as an approach to the molecular form in which apos are transferred. METHODS Cerebrospinal fluid was collected from patients undergoing myelography. Unconcentrated fluid was used for determining the apo concentrations. When the fluid was subjected to ultracentrifugation, it was concentrated 5- to 10-fold by using Millipore membrane filters (Millipore Co., Bedford, MA) prior to ultracentrifugation. Apos were quantitated by the electroimmunoassay (EIA) technique of Laurell (12). Agarose (Seakem, Marine Colloids, Inc., Rockland, ME) was dissolved in 0.025 M barbital pH 8.6 buffer and mixed with the antiserum at 50°C; and this mixture was poured onto 200 X 100 X 1.5 mm plates. Standard curves were obtained by dilution of pooled human serum with buffer. Apo concentrations were determined in the cerebrospinal fluid and plasma of the same patient. The cerebrospinal fluid values are expressed as percentage of the corresponding apo concentration in plasma because one of the requirements (12) for absolute quantitation by EIA is that the sample and standards be in the same physical state. Every sample was incubated at 52°C for 3 hr (13) to decrease possible differences in the availability of antigenic sites. The plates were stained with 0.5% Coomassie brilliant blue R-250 and destained in ethanol/acetic acid/H20, 4.5:1.0:4.5 (vol/vol). The following antisera were used: anti-apo B, anti-apo E, anti-apo A-I, anti-apo C-II, anti-apo C-Ill, and anti-albumin. Apo B was prepared by injecting LDL (p = 1.0-1.050 g/ml), apo E was prepared by sodium dodecyl sulfate/polyacrylamide gel electrophoresis of VLDL. Apo A-I was obtained by Sephadex G-150 gel filtration (14). The apo C group was isolated from delipidated VLDL by a combination of Sephadex G-200 and DEAE-cellulose ion exchange chromatography (14). Albumin was prepared according to the method of Schwert (15). Antisera were produced in goats according to the method of Vaitukaitis et al. (16). The various antisera were tested against whole serum and pure antigens by the double-diffusion technique (17) and by immunoelectrophoresis (18). Cerebrospinal fluid lipoproteins were isolated in the Beckman L5-50 preparative ultracentrifuge according to the method of Havel et al. (19) using a 40.3 rotor with its 2-ml adapter. The centrifugation was carried out for 44 hr at 114,000 X g. Sodium dodecyl sulfate/polyacrylamide gel electrophoresis, according to Shapiro, et al. (20) as modified by Maizel (21), was used to determine the apo composition of cerebrospinal fluid with human VLDL (p < 1.006 g/ml) and HDL (p 1.063-1.21
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
Abbreviations: VLDL, very low density lipoprotein; LDL, low density lipoprotein; HDL, high density lipoprotein; apo, apolipoprotein; EIA, electroimmunoassay; VHDL, very high density lipoprotein. 4646
Proc. Natl. Acad. Sci. USA 76 (1979)
Medical Sciences: Roheim et al. g/ml) and proteins of known molecular weight used as standards. Protein was determined according to the methid- of Lowry et al. (22) with albumin as the standard. Electron microscopy was carried out according to the method of Forte and Nichols (23). Unconcentrated native cerebrospinal fluid samples and p < 1.21 g/ml fractions were diluted with an equal volume of 2% sodium phosphotungstate (pH 7.4) and immediately examined in the JEM 100C electron microscope at instrument magnifications of X40,000-X80,000. Particle size was determined on free-standing particles with round profiles. RESULTS The presence of apos in cerebrospinal fluid was demonstrated by using the techniques of double diffusion and immunoelectrophoresis. Cerebrospinal fluid was examined with antisera produced against apo VLDL, apo HDL, apo B, apo A-I, apo E, apo C-II, and apo C-III; all except apo B were identified in cerebrospinal fluid. Apo E, apo C-III, and apo C-Il were present in cerebrospinal fluid at 2-5% of their plasma concentrations whereas apo A-I was present at 1.21 g/ml was determined (Table 1; Fig. 1). The distribution of albumin and of total protein was also determined. In the cerebrospinal fluid 80-90% of the apo E and apo A-I and 60% of the apo C-II and apo C-III were present in the p < 1.21 g/ml fraction. Consequently, 10-20% of the apo E and apo A-I and 40% of the apo C-II and apo C-III were found in the p > 1.21 g/ml fraction. About 5% of the total protein was recovered in the p < 1.21 g/ml fraction; this fraction did not contain any detectable albumin. Apos of the cerebrospinal fluid were further characterized by sodium dodecyl sulfate/polyacrylamide gel electrophoresis. In the p < 1.21 g/ml fraction, two major apos were found; the first major band corresponded to apo E and the second, to apo A-I (Fig. 2). In addition, low Mr apo Cs were also present. In the p > 1.21 g/ml fraction, albumin and bands similar in Mr to the apo Cs were present. It was possible that the apos found in the p > 1.21 g/ml fraction are present in the form of very high density lipoproteins (VHDL; p 1.21-1.24 g/ml) (24). Therefore, the apo distribution between VHDL and p > 1.24 g/ml was also determined by subjecting the p > 1.21 g/ml fraction to an additional ultraTable 1. Apo concentration and distribution between fractions in cerebrospinal fluid % of plasma concentration* p < 1.21 g/ml p > 1.21 g/ml Total Compound n 24
(0.8-0.2) 2.7 (6.5-0.8)
(0.6-0.2) 2.2 (5.1-1.0)
(0.1-0.05) 0.2 (0.4-0.1)
2.0 3.3 (6.2-1.0) (9.2-1.2) 5.1 3.8 2.0 Apo C-III 25 (6.0-1.1) (11.7-1.7) (9.6-1.4) 14 Albumin 0.4 0.5 0 (0.6-0.3) (0.8-0.3) (0) * Mean (and range) of values obtained from three different pools of cerebrospinal fluid. In three determinations, mean (and range) for protein was: p < 1.21 g/ml, 1.7 (1.4-2.3) mg/dl; p > 1.21 g/ml, 41.7
FIG. 1. EIA of cerebrospinal fluid (A) Apo E; (B) apo C-III; (C) apo C-II; (D) apo A-I; and (E) albumin. Positions: 1, whole fluid; 2, p < 1.21 g/ml; 3, p > 1.21 g/ml. Two ml of 5-fold concentrated cerehrospinal fluid was subjected to ultracentrifugation; after dialysis, the p < 1.21 g/ml and p > 1.21 g/ml fractions were made up to 2 ml, and thus all three samples are directly comparable.
centrifugation at p = 1.24 g/ml. Only traces were present in VHDL, and practically all of the apos were recovered in the p > 1.24 g/ml fraction. Cerebrospinal fluid lipoproteins isolated at p < 1.21 g/ml were studied by electron microscopy. The majority of these particles were spherical, although some elongated structures were also present. The particles were somewhat heterogeneous
Medical Sciences: Roheim et al.
Proc. Natl. Acad. Sci. USA 76 (1979)
FIG. 2. Sodium dodecyl sulfate/polyacrylamide gel electrophoresis. Lanes: 1, p < 1.21 g/ml; 2, human VLDL; 3, human HDL; 4, p > 1.21 g/ml. Positions of apos are shown at left.
in size; the mean particle diameters ranged from 11 to 13 nm. Particles of similar dimensions were also present in the native
unfractionated cerebrospinal fluid. DISCUSSION It has been established that the relative concentrations of proteins in cerebrospinal fluid, expressed as percentage of their plasma concentrations, decrease in proportion to the increase in their radius and Mr (11). Albumin (radius, 35.8 A; Mr 69,000) is present in the cerebrospinal fluid at 0.5% of its plasma concentration, which is similar to the observation reported here. However, prealbumin which has only a slightly different radius (32.5 A) and Mr (61,000) is present at 6.1% of its plasma concentration. These data suggest that, although the majority of plasma proteins found in the cerebrospinal fluid are transferred by a process that is related to their size, a specific mechanism should be invoked to explain the presence of prealbumin in such a high concentration. Observations from this study show a relatively high concentration of 110 to 130-A particles in cerebrospinal fluid. We have demonstrated the presence of apos A-I, E, C-I, and C-III in cerebrospinal fluid; apo B could not be detected. Previous reports indicated the presence of a lipoproteins in normal cerebrospinal fluid and the appearance of :3 lipoproteins in patients with neurological disorders (25). We have used the absence of apo B and measurements of albumin concentration as an indication that the samples were not contaminated with plasma proteins. Cerebrospinal fluid samples that had detectable apo B (>0.01% of plasma concentration) and increased albumin concentrations were excluded from the study (two cases). Our double-diffusion data suggested the presence of apo A-Il in cerebrospinal fluid; however, these data were not included because the assays are still under development. The cerebrospinal fluid contains cholesterol and phospholipid in concentrations less than 0.2% of their concentrations in plasma (26, 27). When subjected to ultracentrifugation, most of the cerebrospinal fluid apos were found in the form of lipo-
proteins because they had a hydrated density p 1.21 g/ml fraction, more apo C (40%) was present than apo E and apo A-I (10-20%). Studies with sodium dodecyl sulfate/polyacrylamide gel electrophoresis confirmed the observations derived by immunochemical methods and showed that the predominant apos in cerebrospinal fluid are apo E and apo A-I. These data suggest that the apo composition of cerebrospinal fluid lipoproteins is different from that of the known major plasma lipoproteins. There are two plasma lipoproteins reported to have apo E as their major apo constituent; these are HDLC (28) and the nascent or discoidal HDL (29). Electron microscopic studies indicated that the lipoprotein particles in the p < 1.21 g/ml fraction are mainly spherical and range in size from 11 to 13 nm. The diameters of the cerebrospinal fluid lipoprotein particles are larger than reported values for human serum HDL2b (10.5 + 0.8 nm) which are the least dense (p, 1.063-1.100 g/ml) and the largest of the major serum HDL subfractions (30). The lipoprotein structures seen in the p < 1.21 g/ml fractions from concentrated cerebrospinal fluid were also identifiable in native unconcentrated fluid, thus ruling out possible morphological artifacts. Evidence has been presented that the majority of the apos found in the cerebrospinal fluid are present in the form of a p < 1.21 g/ml lipoprotein fraction similar in size to plasma HDL. Some of the apos are present in much higher concentrations than would be expected if an intact HDL molecule had been transferred. At this time we can only postulate the different mechanims that could be responsible for the presence of apos in cerebrospinal fluid. The possibility that free apos (or apo complexes) are transferred from the plasma to cerebrospinal fluid could be considered. It is possible that the apo concentration in cerebrospinal fluid depends on the concentration and the size of the circulating free apos. After the free apos are transferred to the cerebrospinal fluid, they may combine with lipids to form the lipoproteins. It should be noted that no free apo B could be demonstrated in the p > 1.21 fraction of plasma and no apo B could be demonstrated in cerebrospinal fluid. At the present time, however, no direct experimental evidence exists to establish the presence of free apos in plasma. The other possibility is that whole lipoprotein particles are transferred from the plasma to the cerebrospinal fluid. It could be assumed that a special "lipoprotein class" exists in the circulation which has an apo composition similar to that of the cerebrospinal fluid lipoproteins, and this specific lipoprotein could be transferred to the cerebrospinal fluid. In this respect, the unique apo composition of cerebrospinal fluid must be considered. The only known lipoprotein fractions that approximate the apo composition of cerebrospinal fluid is HDLC (28) and the nascent discoidal HDL (29). The mechanism of lipoprotein transfer to cerebrospinal fluid should be different from the mechanism responsible for the transfer of the other plasma proteins because the apo concentration of the p < 1.21 g/ml fraction of cerebrospinal fluid is higher than one would expect from filtration of a protein with high Mr and large size.
Finally, the local synthesis of cerebrospinal fluid cannot be excluded a priori. It has been shown that cerebrospinal fluid contains cholesterol hydrolase activity (31) and it is known that the cholesterol ester concentration of the brain increases during active demyelination (32). Regulation of cholesterol synthesis by LDL in cultured cells of neural origin has also been demonstrated (33). However, there is no information on the possible physiological function and pathophysiological role of the apos present in the cerebrospinal fluid. The relatively high apo concentrations in ce-
Medical Sciences: Roheim et al. rebrospinal fluid and the known relationship between apos and lipid metabolism warrant further study. It is tempting to speculate that the apos of cerebrospinal fluid influence the lipid metabolism of the brain or influence the lipid metabolism through hypothalamic receptors. This work was supported by Grants HL20954, HL18574, and T2 HL07098 from the National Institutes of Health. 1. Jackson, R. L., Morrisett, J. D. & Gotto, A. M. (1976) Physiol; Rev. 56,259-316. 2. Brewer, B. H. & Osborne, J. C. (1977) Adv. Protein Chem. 31, 253-337. 3. Schaefer, E. J., Eisenberg, S. & Levy, R. J. (1978) J. Lipid Res. 19,667-687. 4. Chajek, T. & Fielding, C. J. (1978) Proc. Natl. Acad. Sci. USA 75,3445-3449. 5. Goldstein, J. L. & Brown, M. S. (1977) Annu. Rev. Biochem. 46, 897-930. 6. Mahley, R., Innerarity, T. L., Pitas, R., Weisgraber, K., Brown, J. & Gross, E. (1977) J. Biol. Chem. 252, 7279-7287. 7. Reichl, D., Myant, N. B. & Pflug, J. J. (1977) Biochim. Biophys. Acta 489, 98-105. 8. Reichl, D., Myant, N. B., Brown, M. J. & Goldstein, J. L. (1978) J. Clin. Invest. 61, 64-71. 9. Roheim. P. S., Miller, L. & Eder, H. A. (1965) J. Biol. Chem. 240, 2994-3001. 10. Aoheim, P. S., Edelstein, D. & Pinter, G. G. (1976) Proc. Natl. Acad. Sci. USA 73, 1757-1760. 11. Felgenheuer, C. B. (1974) Kin. Wochenschr 52, 1158-1164. 12. Laurell, C. B. (1972) Scand. J. Clin. Lab. Invest. 29, Suppl. 124, 21-37. 13. Karlin, J. B., Juhn, D. J., Starr, J. I., Scanu, A. M. & Rubinstein, A. H. (1976) J. Lipid Res. 17, 30-37. 14. Brown, W. V., Levy, R. J. & Fredrickson, D. J. (1969) J. Biol. Chem. 244, 5687-5694.
Proc. Natl. Acad. Sci. USA 76 (1979)
15. Schwert, G. W. (1957) J. Am. Chem. Soc. 79, 139-141. 16. Vaitukaitis, J. J., Robbins, B., Nieschlag, E. & Ross, G. T. (1971) J. Clin. Endocrinol. Metab. 33,988-991. 17. Ouchterlony, 0. (1968) in Handbook of Immunodiffusion and Immunoelectrophoresis (Ann Arbor Science Publisher Inc., Ann Arbor, MI), pp. 21-31. 18. Grabar, P. & Williams, C. A. (1955) Biochim. Biophys. Acta 17, 67-74. 19. Havel, R. J., Eder, H. A. & Bragdon, J. H. (1955) J. Clin. Invest. 34, 1345-1353. 20. Shapiro, A. L., Vinuela, E. & Maizel, J. V., Jr. (1967) Biochem. Biophys. Res. Commun. 28, 815-820. 21. Maizel, J. V., Jr. (1971) in Methods of-Virology, eds. Mormorosch, K., Koprowski, H., (Academic, New York), Vol. 5, pp. 179246. 22. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. (1951) J. Biol. Chem. 193,265-275. 23. Forte, T. & Nichols, A. V. (1972) Adv. Lipid Res. 10, 1-41. 24. Alaupovic, P., Sanbar, S. S., Furman, R. H., Sullivan, M. L. & Walroven, S. L. (1966) Biochemistry 5, 4044-4053. 25. Swahn, B., Bronnestam, R. & Dencker, S. J. (1961) Neurology 11, 437-440. 26. Pedersen, H. E. (1973) Acta Neurol. Scand. 49, 626-638. 27. Pedersen, H. E. (1973) Acta Neurol. Scand. 49, 639-648. 28. Mahley, R. W. (1978) in Disturbances in Lipid and Lipoprotein Metabolism, eds. Dietschy, F. M., Gotto, A. M. & Ontko, J. A. (Am. Physiol. Soc., Bethesda, MD), pp. 181-197. 29. Hamilton, R. L., Williams, M. C., Fielding, C. J. & Havel, R. J. (1976) J. Clin. Invest. 58,667-680. 30. Nichols, A. V., Gong, E. L., Forte, T. M. & Blanche, P. J. (1978) Lipids 13, 943-950. 31. Shantibal, N. S. & Johnson, R. C. (1978) Exp. Neurol. 58, 6873. 32. Wender, M., Filipek-Wender, H. & Stanislowska, J. (1974) Clin. Chim. Acta 54,269-275. 33. Volpe, J. J., Hennessy, S. W. & Wong, T. (1978) Biochim. Biophys. Acta 528, 424-435.