293
Clinica Chimicn Acta, 71 (1976) 293-301 @ Elsevier Scientific Publishing Company,
Amsterdam
- Printed in The Netherlands
CCA 8025
THE SEPARATION OF PLASMA LIPOPROTEINS ELECTROFOCUSING AND POLYACRYLAMIDE GEL ELECTROPHORESIS
USING GEL GRADIENT
ALAN V. EMES **, ALBERT L. LATNER *, MOHAMMAD RAHBANI-NOBAR BENG HEAN A. TAN *** University Department Tyne, Nel 4LP (U.K.) (Received
of Clinical Biochemistry,
and
Royal Victoria Infirmary, Newcastle upon
April lst, 1976)
Summary Polyacrylamide gel electrofocusing and gradient electrophoresis have been used to separate the lipoproteins in whole plasma and in fractions prepared by sequential flotation in the ultracentrifuge and by precipitation with dextran sulphate and manganous chloride. After the two-dimensional separation, high density lipoproteins appear as a zone showing noticeable heterogeneity with respect to both isoelectric point and molecular weight. Low density lipoproteins are resolved as a compact spot while very low density lipoproteins are visible as a long horizontal streak across the top of the electrophoresis gel. The implications of the technique for the analysis of lipoprotein patterns in pathological plasmas are discussed.
Introduction Plasma lipoproteins have been separated by electrophoresis using a wide variety of supporting media including polyacrylamide gel [ 1,2]. Separations have also been achieved by employing polyacrylamide gradient gels [3,4] and gel electrofocusing [ 5,6]. Electrophoresis in agarose gel is commonly used for routine diagnostic purposes but does not always provide an unequivocal result. We have found that gel electrofocusing followed by polyacrylamide gradient gel electrophoresis [ 71 separates the lipoproteins in plasma into clearly-defined, reproducible zones and that, in certain cases, these zones can be shown to be hetA.L. Latner, Department of Clinical Biochemistry. Royal Victoria Infirmary. Newcastle upon Tyne. Nel 4LP (U.K.) ** Present address: School of Pharmacy, Sunderland Polytechnic, Sundrrland, SRl 3SD, U.K. *** Present address: Department of Pathology, North Manchester General Hospital, Crumpsall, Manchester, MS 6RB. U.K.
* Correspondence should be addressed to: Professor
294
erogeneous. This paper describes the use of the two-dimensional technique in the separation of lipoproteins from whole plasma and also from fractions prepared in the ultracentrifuge and by precipitation with dextran sulphate and manganous chloride. Materials and methods Electrofocusing and electrophoresis Agarose gel electrophoresis was performed using the method described by Noble [8]. Gel electrofocusing followed by polyacrylamide gradient gel electrophoresis has been described previously [ 71. The electrophoresis step was also used alone. The gels were stained for protein with Coomassie Brilliant Blue G [ 91 and for lipid with Sudan Black B in 60% (v/v) ethanol [B] or an acetone/acetic acid/water mixture [lo]. The positions of certain of the lipoproteins in gels after two-dimensional separation were established immunologically [ 7,111 with specific antisera (Hoechst Pharmaceuticals, Hounslow, Middlesex, U.K.). In order to obtain satisfactory migration of the lipoproteins out of the polyacrylamide gel discs, it was necessary to use immunoelectrophoresis [12] rather than double diffusion. Fractionation of lipoproteins Plasma lipoproteins were fractionated by sequential flotation in the ultracentrifuge. Solutions of density 1.0063 g/ml, 1.1816 g/ml and 1.4744 g/ml were prepared and the separations carried out essentially as described by Lindgren et al. [13]. A 40.3 rotor (Beckman Instruments Ltd., Glenrothes, Fife, Scotland) was used at a speed of 3.5 000 rev./min (100 000 X g) and the centrifugation steps allowed to run for 23.5 h (to prepare the VLDL* and LDL fractions) or 26 h (to prepare the HDL fraction) at 16-18°C. Densities were routinely monitored by refractometry and reference to the International Critical Tables [ 141. Where it was necessary to remove chylomicrons before separation of the other lipoproteins, 7 ml of the plasma was centrifuged for 105 min in a Beckman SW39 rotor at 25 000 rev./min (78 000 X g) and the 6.5 ml subnatant taken for further fractionation [ 131. All separated fractions were dialysed against two l-l volumes of 0.05 M KH2P04/Na2HP04 buffer, pH 7.0, and the dialysis residue stored at 4°C. All subsequent analyses were carried out on the lipoprotein fractions as soon as possible after preparation. The HDL fraction was subfractionated in an NaBr density gradient [ 131. Two fractions, HDL2 and HDL,, were obtained after two centrifugation stages on a 3 X 6.5 ml titanium swing-out rotor (Measuring and Scientific Equipment Ltd., Crawley, Sussex, U.K.), used at a speed of 50 000 rev./min (200 000 X g) for 20 h and 24 h, respectively. The separated fractions were dialysed and stored as above. The lipoproteins from 250 ml of plasma were also fractionated by precipitation with dextran sulphate (Pharmacia (Great Britain) Ltd., London W5, U.K.) and MnCI, [ 161. The precipitation procedure yielded a mixture of VLDL and * Abbreviations: HDL, high density lipoproteins: LDL. low density lipoproteins; VLDL, very low density
lipoproteins
[151.
295
LDL. These were separated by diluting the mixture with an equal volume of NaCl solution of density 1.0063 g/ml [13] and centrifuging for 32 h at 1618°C in a Beckman 40.2 rotor at 35 000 rev./min (89 000 X g). The VLDL formed a turbid band at the top of the tube while LDL sedimented to the bottom. The fraction containing HDL was contaminated with other serum proteins. After adjusting the density of the fraction to 1.22 g/ml with NaCl and NaBr [16] and removing the precipitated dextran sulphate by centrifugation, the supernatant was centrifuged for 34.5 h at 35 000 rev./min (89 000 X g) in the Beckman 40.2 rotor at 16-18°C. HDL floated to the top of the tube and were removed. All prepared, fractions were dialysed against 2 1 of 0.02 PIITris/ HCl buffer, pH 7.7, containing 1% (w/v) NaCl. The dialysis residues were stored at 4°C and all subsequent analyses carried out as soon as possible. Densities were monitored by refractometry as before. Samples Small specimens of venous blood were collected using EDTA as anticoagulant after an overnight fast and the plasma stored at 4°C. Normal samples were obtained from apparently healthy laboratory staff. The larger volume of plasma required for the precipitation of lipoproteins was prepared by centrifugation from blood containing citrate which had been stored at 4°C until no longer suitable for transfusion. Results Lipoproteins in normal plasma A sample of normal fasting plasma was subjected to electrofocusing followed by gel gradient electrophoresis. After staining with Coomassie Brilliant Blue G and Sudan Black B a number of proteins containing lipid-staining material were visible (Fig. 1). In this figure, labels have been assigned to certain lipoproteins and lipid-binding proteins on the basis of work to be reported later in this paper (Fig. 2), although HDL and albumin were easily identified by their molecular weights and by reference to the gel slice stained for protein. It was possible to confirm the location of LDL and HDL using immunological methods after the two-dimensional separation. Immunoelectrophoresis using specific anti-p- and anti-c+lipoprotein sera showed that LDL and HDL were located at the positions shown in Fig. 1. Lipoproteins in fractions prepared in the ultracen trifuge The lipoproteins in a sample of the normal fasting plasma shown in Fig. 1 were fractionated by sequential flotation. The densities of the reference solutions [ 131 obtained after each of the three centrifugation steps were 1.0062 g/ml, 1.0559 g/ml and 1.1720 g/ml. After dialysis, the three lipoprotein fractions were subjected to agarose gel electrophoresis and the gel stained with Sudan Black B in ethanol. The fractions showed single bands corresponding to VLDL, LDL and HDL for reference densities of 1.0062 g/ml, 1.0559 g/ml and 1.1720 g/ml, respectively. After gel electrofocusing followed by gel gradient electrophoresis, and staining
296
Fig. 1. Eirctrofocusing followed by gel gradient rlectrophoresis of 5 ~1 of normal plasma, obtained using EDTA as anticoagulant. The gel was cut to give two identical slices; one was stained with Coomassie Brilliant Blue G (I) and the other with Sudan Black B in acetone/acetic acid/water (II). The elrrtrofocusing gel was positioned at the top so that the pH gradient increased from left to right. Electrophoresis proceeded in a downward direction towards the anode. A, albumin; H, HDL; L, LDL; V. VLDL. A full list of identified proteins appears elsewhere [71.
for protein and lipid, the separated lipoproteins could be clearly seen. The VLDL formed a streak close to the top and towards the more alkaline end of the electrofocusing gel (Fig. 2a). LDL appeared as a well-defined spot (Fig. 2b) while HDL formed a zone in which some microheterogeneity was visible (Fig. 2~). Albumin was essentially the sole contaminant. The results obtained with the same preparations after gradient gel electrophoresis alone are shown in Fig. 3 in which the main lipoprotein fractions can be seen as separate zones. Lipoproteins in fractions prepared by precipitation Plasma lipoproteins were also fractionated by precipitation with dextran sulphate and MnClz. After removal of chylomicrons, the total low density lipoprotein fraction was separated into VLDL and LDL by ultracentrifugation, the density of the solution between the lipoprotein zones [16] being 1.0062 g/ml. Similarly, the HDL were purified finally by ultracentrifugation giving a reference solution [13] density of 1.1887 g/ml. Agarose gel electrophoresis, and electrofocusing followed by electrophoresis, of the iipopro~in fractions gave results almost identical to those shown by the fractions prepared by sequential flotation. Each fraction produced a single lipidstaining band of the expected mobility on agarose gel electrophoresis and the two-dimensional separations were the same as those in Fig. 2. There was much less contamination with albumin. Lipoproteins in a~~or~a~ plasma The two-dimensional procedure has been applied to abnormal plasma. Fig. 4 shows the results of two such separations. In both cases there was a marked increase in VLDL and a noticeable increase in HDL, especially when compared with Fig. 1. Separation by agarose gel electrophoresis of the lipoproteins in the plasma used in Fig. 4a, and comparison with a similar separation from normal
297
Fig. 2. Separated proteins from 10 jd of lipoprotein fractions prepared by sequential flotation. Experimental details as for Fig. 1. a, VLDL fraction (reference solution density 1.0062 g/ml); b, LDL fraction (reference solution density 1.0559 g/ml); c, HDL fraction (reference solution density 1.1720 g/ml). A, albumin: H, HDL; L, LDL; V, VLDL. In each case, protein staining (I) on left and lipoprotein staining (II) on right.
298
1
2
3
4
1
2
3
4
r’ig. 3. Polyacrylamide gel gradient electrophoresis of normal plasma and lipoprotein fractions prepared from this plasma by sequential flotation. Electrophoresis proceeded in a downward direction towards the anode. The gel was sliced and stained as for Fig. 1. 1, 5 ~1 of whole plasma; 2, 10 ~1 of the VLDL fraction (reference solution density 1.0062 g/ml); 3. 10 ~1 of the LDL fraction (reference solution density 1.0559 g/ml); 4, 10 ~1 of the HDL fraction (reference solution density 1.1720 g/ml). A, albumin. H, HDL; L. LDL: V, VLDL. Protein staining (I) on left and lipoprotein staining (II) on right.
plasma showed HDL to be increased and also the presence of a ‘broad beta’ band. An identical separation of the lipoproteins in the plasma used in Fig. 4b indicated that HDL and VLDL were both increased and that chylomicrons were also present. Subfractionation of high density lipoproteins Because of the observation of heterogeneity within the HDL fraction (see especially Figs. 1 and 4b) an attempt was made to subfractionate the HDL using .- _^_ b
P
“,
,I
.n
-A
-A
Fig. 4. Separated proteins from 7.5 ~1 of abnormal plasmas. Experimental details as for Fig. 1. except that the whole gels were stained with Sudan Black B in ethanol. a, from a patient with xanthomatosis and hyperlipoproteinaemia; b, from a patient with myxoedema. A, albumin: H, HDL; L. LDL; V, VLDL.
299
Fig. 5. Polyacrylamide gel gradient electrophoresis of HDL subfractions. 1. HDL3; 2, HDL2. 3, total HDL. The arrow denotes the low molecular Protein staining (I) on left and lipoprotein staining (II) on right.
Experimental details as for Fig. 3. weight components seen in HDL3.
a NaBr density gradient. While the subfractions appeared identical after agarose gel electrophoresis, Fig. 5 shows the subfractions after gradient gel electrophoresis alone and staining with Sudan Black B. HDL3 contained more components of low molecular weight than HDL2 although there was considerable overlap. In the total HDL fraction the low molecular weight material shown in HDL3 (arrowed in Fig. 5) was not present in a concentration sufficient to show up as a prominent, stained zone. Discussion The separation of lipoproteins in whole plasma using gradient gel electrophoresis alone (Fig. 3) gave a pattern of lipid-staining zones very similar to those obtained elsewhere [ 3,4] although the HDL were more densely stained than in the results shown by Bautovich et al. [ 41. The lipoprotein fractions obtained in the ultracentrifuge each gave a single lipid-staining band after agarose gel electrophoresis. In Fig. 3, VLDL and LDL appear as compact bands although some material in the LDL sample has trailed back through the position occupied by VLDL. HDL obviously contain components with a wide range of molecular weights. Two-dimensional separation of the lipoproteins in whole plasma gave a welldefined HDL zone while the LDL and VLDL were not clearly resolved (Fig. 1). The same separation of the components of the lipoprotein fractions prepared in
300
the ultracentrifuge showed that LDL were confined to the dense lipid-staining spot seen in Figs. 1 and 2b while the long streak in Figs. 1 and 2a consisted entirely of VLDL. The density of the reference solution taken at the same time as this VLDL fraction was 1.0062 g/ml, which is very close to the solvent density used elsewhere to isolate VLDL [l.3,17]. Identical results were seen with lipoprotein fractions obtained by precipitation. The appearance of VLDL as a streak rather than a spot could be due to incomplete electrofocusing. This fraction contains components with molecular weights between 5 X 106 and 10 X lo6 fll] and molecules of this size would be expected to suffer considerable retardation due to molecular sieving effects. However, VLDL did migrate a short distance into gradient gels in which the lowest acrylamide concentration was 4% (see [4] and Fig. 3) although it has recently been reported [6} that certain high molecular weight components in the VLDL fraction will not enter a 3% polya~rylamide gel. For ele~trofo~using we used 3.5% polyac~lamide gels [7]. Al~~atively VLDL, which is composed of more than 90% lipid [ 151, is likely to be adversely affected by the conditions present during electrofocusing. To overcome this, Righetti and Drysdale [ 51 recommend the use of non-ionic detergents and gels containing ethylene glycol have also been used [l.8]. Nevertheless, satisfactory electrofocusing of VLDL has been achieved without using these additives [ 61. It seems more probable that the isoelectric heterogeneity of the VLDL fraction, as demonstrated by electrofocusing [ 181 and implied by the recent separation of VLDL by ion-exchange chromatography [19], is the cause of the streak seen after electrofocusing. As in the case of IgG [7], a number of proteins, each properly focused as a compact spot but lying close together, would give the appearance of a diffuse band. During electro~horesjs~ the more acid constituents with higher electrophoretic mobilities would migrate faster so that, after a two-dimensional separation, a sloping band would result. This can be seen in Figs. 2a and 4 although molecular sieving has held down the actual rnou bilities of the VLDL components. In several of our separations HDL appear as several spots or bands of different molecular weights and isoelectric points. ~ubfra~t~onation and separation of the components of the subgroups (Fig. 5) showed that there was considerable overlap between HDL1 and HDL3. This may have been due to incomplete separation of the subfractions although some coincidence could be expected since Nichols [ZO] has suggested that HDLz may be partially converted to HDL3 during ultra~ent~fugation~ and Sundaram et al. [21] have found components of HDLJ with isoelectric points identical to those of some HDI& constituents” HDL3 contained a number of extra components of low molecular weight and these results are in broad agreement with those obtained by disc electrophoresis 1221, where the HDL3 fraction was also shown to contain components of higher mobility. Fig. 4 shows results obtained with plasma from hy~erlipid~em~~ patients. When compared with Fig. 1, the raised level of HDL can be clearly seen in both plasmas as can the increased concentration of VLDL in Fig. 4b. Chylomicrons, of course, remained at the point of application and were not detected. In Fig. 4a, the lipoproteins presumably contributing to the ‘broad beta’ band seen after agarose gel electrophoresis appear as a dense zone extending over most of the
301
VLDL region and probably into the LDL spot. Although there is a wide range of methods available [l] satisfactory analysis of lipoprotein patterns can present considerable problems. The common classi‘operational’ distinctions [ 151 dependent fications based on density employ upon the procedures used in individual laboratories and this, together with other limitations recently outlined by Blaton [23], can make interpretation of results from electrophoretic separations on agarose gel difficult and imprecise. The ‘broad beta’ type of lipoproteinaemia [24] is an obvious example. At the cost of adding a further technique to those already in use, we believe that, in combination with existing methods, gel electrofocusing followed by polyacrylamide gradient gel electrophoresis could provide additional information without recourse to ultracentrifugation. The two-dimensional technique should overcome the objections concerning polyacrylamide gel electrophoresis alone [ 21 and provide more information, especially about heterogeneity within the lipoprotein fractions, than the single-stage methods reported earlier [ 4,6,18]. References 1 Hatch, F.T. and Lees, R.S. (1968) Adv. Lipid Res. 6, 2-68 2 Iammarino. R.M. (1972) in Protides of the Biological Fluids (Peeters, H., ed.), Vol. 19. PP. 91-95. Pergamon, Oxford 3 Pmt. J.J. and Dangerfield. W.G. (1969) Clin. Chim. Acta 23,189-201 4 Bautovich, G.J., Dash, M.J., Hensley, W.J. and Turtle, J.R. (1973) Clin. Chem. 19,415418 5 Righetti, P.G. and Drysdale, J.W. (1974) J. Chromatogr. 98.271521 6 Godolphin, W.J. and Stinson, R.A. (1974) Clin. Chim. Acta 56.97-103 7 Emes, A.V.. Latner, A.L. and Martin. J.A. (1975) Clin. Chim. Acta 64. 69-78 8 Noble, R.P. (1968) J. Lipid Res. 9,693-700 9 Diezel, W., KopperschlPger. G. and Hofmann. E. (1972) Anal. Biochem. 48.617-620 10 Pmt. J.P., Lamy. J.N. and Weill, J.D. (1969) Bull. Sot. Chim. Biol. 51. 1367 11 Dewar, J.H. and Latner, A.L. (1970) Clin. Chim. Acta 28.149-152 12 Williams, C.A. (1971) in Methods in Immunology and Immunochemistry (Williams, C.A. and Chase, M.W., eds.), Vol. 3, pp. 234-294, Academic Press, New York 13 Lindgren, F.T., Jensen, L.C. and Hatch, F.T. (1972) in Blood Lipids and Lipoproteins: Quantitation, Composition and Metabolism (Nelson, G.J., ed.), pp. 181-274. Wiley-Interscience, New York 14 Washburn, E.W. (1933) (ed,), International Critical Tables of Numerical Data, McGraw-Hill. New York 15 16 17 18 19 20 21 22 23 24
Scanu, A.M. and Ritter, M.C. (1973) Adv. Clin. Chem. 16,111-151 Burstein. M. and Scholnick, H.R. (1972) in Protides of the Biological Fluids (Peeters, H., ed.), Vol. 19, PP. 21-28. Pergamon, Oxford Levy, R.I.. Bilheimer, D.W. and Eisenberg, S. (1971) in Plasma Lipoproteins (Smellie, R.M.S., ed.), pp. 3-17, Academic Press, London Kostner, G., Albert. W. and Holasek, A. (1969) Hoppe-Seylers 2. Physiol. Chem. 350, 1347-1352 Ho, W.K.K. (1975) FEBS L&t. 50,175-179 Nichols. A.V. (1967) Adv. Biol. Med. Phys. 11, 109-158 Sundaram, G.S.. Mackenzie, S.L. and Sodhi, H.S. (1974) Biochim. Biophys. Acta 337, 196-203 Narayan, K.A.. Narayan, S. and Kummerow. F.A. (1965) Nature 205, 246-248 Blaton, V. (1975) Clin. Chim. Acta 61, 105 Fredrickson, D.S. and Levy, R.I. (1972) in The Metabolic Basis of Inherited Disease (Stanbury. J.B.. Wyngaarden, J.B. and Fredrickson, D.S., eds.), 3rd edn., pp. 545-614, McGraw-Hill, New York