Further Characterization of Apolipoproteins from the Human Plasma Very Low Density Lipoproteins

SUMMARY Valine apolipoprotein, glutamic acid apolipoprotein, and alanine apolipoprotein (apoLP-Ala) (designated by their COOH-terminal amino acids) are three small apolipoproteins which constitute more than half of the protein in human plasma very low density lipoprotein. These proteins were isolated free of lipid by techniques previously described. As determined by sedimentation equilibrium analysis and estimated from amino acid analyses, the approximate molecular weights of the apolipoproteins were: valine glutamic apolipoprotein, and apoLP-Ala, Circular dichroism and optical rotatory dispersion spectra while the obtained to be mainly of random

Previously reported polymorphism of apoLP-Ala was found to be attributed to differences in content of sialic acid.
The very low density lipoproteins of human plasma contain no less than five different apolipoproteins (1, 2). These include the apolipoproteins characteristic of low density and high density lipoproteins (3). Three other apolipoproteins constitute over half of the total protein component of VLlY lipoproteins (2). These three proteins are usually designated by their COOHterminal amino acids as apoLP-Val, apoLP-Glu, and apoLP--4la. The amino-terminal residue of both apoLP-Val alld apoLP-Glu is threonine and of apoLP-Ala is serine. NeTv data obtained in the further characterization of these t'hree apolipoproteins are the subject of this report.
Molecular weights for all three proteins have been determined by sedimentation equilibrium analysis in the ultracentrifuge, and these values hare been compared to molecular weights estimated from amino acid analyses. nleasurements of ultraviolet absorption, optical rotatory dispersion, and circular dichroism have been used to study the st,ructure of apoLP-Val and apoLP-*Yla. A4 basis for apparent 1 The abbreviations used are: VLD, very low density lipoproteins; apoLP-Val, valine apolipoprotein; apoLP-Glu, glutamic acid apolipoprotein; apoLP-Ala, alanine apolipoprotein; CD, circular dichroism; ORD, optical rotatory dispersion. polymorphism of apoLP--4la has also been est,ablished. This protein is present in at least t,wo forms that are immunochemically identical and have the same amino acid composition, NH*-and COOH-terminal groups, and Cl) and ORD spectra. They are separable into two bands on polyacrylamide electrophoresis and two fract,ions by ion exchange chromatography. This chromatographic behavior has been shown to result from differences in carbohydrate composition.
The sources of the remainder of the materials were those preCously reported (I).
Source and Preparation of Apolipoprotains-Tile VLl> lipoproteins were obtained from subjects rrith types IV and V hyperlipoproteinemia (4). The techniques of isolat,ion, rcmovnl of the lipid, and solubilization mere identical with those prcviously described (1). The VLD apolipoproteins were separated chromatographically according to a revised xcheme which permitted a better yield of all three apolipoproteins and a more satisfactory purification of apoLP-Glu (2). The criteria for purity of the apolipoproteins were the presence of :I single band on polyacrylamide electrophoreais, a single precipitin line 01). tained with specific ant,isera by the Ouchterlony technique, and a charact'eristic amino acid composition after acid hydrolysis. The techniques of electrophoresis in polyacrylnmidc gels and immunochemical analysis rrere employed as previously described (1).
Optical &u&e-Samples of apoLP-Ala and al~oIll'-V:~l were lyophilized once in 0.1 JI KH4HC03 and three times in distilled lvater, and stored over PZOS under vacuum to constant weight. Absorbance was determined in 0.02 M potassium phosl)lrate buffer (pH 7.5) in a Beckman I1U spectrophotomcter at :I light path of 1 cm. Linear plots of absorbance against concentration were obtained for solutions containing 0.0625 to 1.0 mg l)er ml. The Cl) spectra were recorded at 23 to 27" with a ('ary model 60 spectropolarimeter, equipped with a Pockels cell (5). The mean residue ellipticity [8], in degrees of square centimeters per decimole, was calculated from the following equation: Linear plots of In absorbance (280 nm) against the square of the distance (T) from the axis of rotation were obtained in each experiment.
The mean residue weight (MRW) was 111 g per mole for npoLP-T'al and 115 g per mole for apoLP-A& The path length (1) v-as 0.5 mm in all experiments. The concentration of the protein (c) varied from 0.25 x 10e3 to 1 .O x 1OF g per ml in 0.05 RI potassium phosphate buffer (~13 7.5). ORD spectra were also recorded in the Cary model 60 with the equipment in the ORD mode (6). The conditions were otherwise identical with those used for CD measurements. The data were plotted as mean residue rotation [m] where: and MRW, 1, and c are the same as those used in the relationship defining molar ellipticity (0). O(X is the observed rotation in degrees at waT:e length X.

Epuilibrium
Experiments-M studies were performed in the Beckman-Spinco model E analytical ultracentrifuge equipped with electronic speed control, absorption optics, photoelectric scanning, and multiplex syst.ems. d titanium rotor (type IN-G) containing six compartments was used. Fire of the compartments contained double sector cells with sapphire windows, and the sixth held a counterbalance with calibrat,ed reference windorrs 1.570 cm apart.
Protein samples were Iyoph- ilized in 0.05 JZ NHJ-ICOB and dissolved at collcclltrations of 1 mg per ml in 0.1 M NaCI, 0.02 nz Tris-TIC'1 at pII 8.6, and guanidine-HU in concentrations of 3, 5, or 7 M (7). The samples were dialyzed in dialysis casing against 50 volumes of the buffer at 4", the dialysates being changed three times at intervals of 24 hours. The density of the final dialysate was tlctermined by pycnometry at 24" and corrected by standard tables to 20". Immediately prior to centrifugation, the l)rotein solutions lvere diluted with the final dialysate to give absorbanccs of 0.85 to 0.45 at 280 nm. The solution sector contained 0.20 ml and the solvent sector 0.25 ml of the final diulysate.
li:quilibri~lm leas judged to be present and the final scans takeu when the plot of absorbance against distance from the axis of rotation did not change for at least 10 hours. The length of most experiments was 44 hours, the rotor speed was 28,000 rpm, and the tcmperature 20".
Amino Acid Analyses--Uiquots (0.5 mg) of the same protein preparations ut.ilized for sedimentation equilibrium were hydralyzed in constant boiling glass-distilled I-ICl at 110" for 12, 24, 48, and 72 hours. Norleucine and homoarginine were added as internal standards before the aliquots were taken for hydrolysis. Other aliquots of protein were subjected to performic acid and hydrolyzed as above for 24 hours.
The amino acid content was determined on a Beckman 120 B automatic analyzer equipped with a high sensitivity-cuvette, an expanded range recorder, and a digital integrator (8,9). Tryptophan was determined spectrophotometrically by the method of Edelhoch (10). Sialic Acid DeterminuttinsSialic acid was released by incubation of the protein in 0.1 M H&SO4 for 1 hour at 80" (11) or by hydrolysis with neuraminidase (V. chobrm) at 25" in 0.08 M ammonium acetate at pH 5.0 (12). It was also necessary to add 0.002 M sodium decyl sulfate to maintain the solubility of apoLP-Ala at this pH. Enzyme (10 units) was added for each mg of glycoprotein substrate.
Free sialic acid was determined by t'he thiobarbituric acid method with N-acetylneuraminic acid as standard (13).

Sedimentation
Equilibrium Experiments-When apoLP-Val, apoLP-Glu, or apoLP-AAla were centrifuged in the absence of guanidine, the slope of the curve relating absorbance to the square of the distance from the axis of rotation (9) increased as measurements were made from the meniscus to the bottom of the cell. This phenomenon was considered most likely due to intramolecular association since each of the protein preparations appeared as a single band on polyacrylamide gel electrophoresis (15% gel, 8 ivr urea).
Accordingly, guanidine-HCl was added to the buffers to offset apparent aggregation.
The technique used by Schachman and Edelstein (7) for analysis of interacting systems to obtain molecular weight was then applied.
Linear plots of In absorbance versus r2 were obtained for apoLP-Glu and Ala in 3, 5, and 7 M guanidine-HCl These values were then plotted against the density of the respective guanidine-HCl solutions. P, is obtained from the intercept on the abscissa, which gives a value of p, and the relation l/p = P,. Extrapolati_on to the intercept on the ordinate gives a value for M, (1 -V,p) in the theoretical solution with density one and thus allows calculation of a value for M, for apoLP-Glu and its preferentially bound small molecular weight substances.
A straight line relation was obtained when M,(l -V,J) was plotted against the density of the respective solution (Figs. 4 and 5) for apoLP-Glu and -Ala. The partial specific volume of the complexes (8,) was calculated from the intercept of these plots on the abscissa for each of the proteins studied and is given in Table I. Since only two points were obtained for apoLP-Val, V, was not calculat'ed for this protein.
The partial specific volumes of the proteins calculated by the method of Cohn and Edsall (14) from the amino acid composition are also given in Table I as v'p. If it is assumed that the difference between Y, and v'p is due entirely to preferentially bound water, the portion of the weight of the complex due to bound water (x) can be used to obtain a molecular weight for the protein (MJ by the relation M, = X,/(1 -8,). M,, t,he molecular n-eight of the complex, was obtained from 7, and the quantjity M,(lvcp) derived from the intercept on the ordinate where p = 1 (7) (Figs. 4 and 5). M, was also calculated for each concentration of guanidine from the equation    (7) are considerably higher than the corresponding values for VP. If this difference is assumed to be due entirely to bound water, the quantities of water present may be calculated and are shown below. An appropriate reduction in molecular weight is obtained by subtracting the bound water from the molecular weight of the complex (M,) to obtain Mp. The molecular weight obtained from the amino acid analyses of presented in Table II. The -?p calculated for apoLP-Glu ignored the tryptophan content since the amount of protein available was insufficient for accurate det.ermination. Estimates from the amino acid analyses suggested that no more than 2 tryptophan residues were present per molecule of protein. This would introduce a negligible error into the calculation of Pp The solid portion of the bar represents the measured release of sialic acid at the time given. A third more slowly migrating band on electrophoresis appeared in the first aliquot after 30 min of incubation, and after 24 hours, when 90% of the theoretical total had been released, only this band was prominent.
(0.725) obtained by the use of the remaining amino acids alone, since the v of kyptophan itself (0.74) is so close to this value.
Amino Acid Composition-The composition of each of the polypeptides differed in almost every amino acid (Table III). "Cysteine + cystine" values were shown not to be present by the absence of cysteic acid in hydrolysates of performic acid-oxidized protein.
All analyses were performed on samples from the same individual.
The internal standards were norleucine and homoarginine. Moles per mole of protein were calculated as described under Tyrosine and histidinc were not present in apoLP-Val, apoLP-Glu contained no histidine, and isoleucine was missing from apoLP-hla.
No cysteine or cystine was found in any of the proteins.
Assuming t,he presence of 1 31 of amino acid per mole of protein for glycine in the case of apoLP-Val, isoleucine in apoLP-Glu, and lristidine for apoLP-Ala suggests that npoLP-Val contains about 62 amino acids, apoLP-Glu contains about 95 amino acids, and apoLP-Ala contain approsimately 81 amino acids. Molecular weights calculated from the amino acid compositions with the use of the nearest integers were 7,200 g per mole for apoLP-Val, and 9,100 for apoLP-Ala. A comparable value was not calculated for apoLP-Glu in the absence of tryptophan determinations.
Sialic Acid Deternlinations-ApoLP-Ala is obtained in two distinct peaks on DIUE-cellulose chromatography (1). These were preriouslg designated Da and D4 (1). On polgacrylamide gel electrophoresis mixtures of Dg and Dg produce two discrete bands with D4 having slightly greater mobility.
These two forms of apoLP-Ala are identical by immunochemical and amino acid analyses and have the same X&-terminal (serine) and COOH-terminal (alanine) amino acids. The sialic acid contents of D3 and D4 were determined after hydrolysis in HzS04. The amount released from D4 was consistently tn-ice that from Da. The values obtained were appror;imately 0.6 and 1.2 moles of possible that the experimentally determined values were low due to the destruct.ion of sialic acid during hydrolysis.
It should also be noted that, if S-glycolylneuraminic acid &ould prove to be present, its low color yield comparetl to the X-acct,gl ncuraminic acid used as standard (11) would also lead to uutlrrcsti-m&ion of the total sialic acid content.
Equal amounts of DZ and D4 were mixed and incubated with neuraminidase.
Timed aliquots were removed for clectrophoresis on polyacrylamide gels and for determination of sialic acid released. Aft,er 10yc of the total sialic acid rrns released, a third less rapidly migrat,ing band appea.red in the gel and became more prominent as increasing quantities of cnrl~ohydr:ale were released (Fig. 6). ilfter 24 hours of incubation, this bnutl was predominant.
Only a trace of the seconcl band (I),), and none of the most rapidly migratin, w band (Dd) was present. These observations strongly suggest that D, and D, ldh represent molecules of apoLP-Ala differing only in content of rarbolrydrate.
Ko sialic acid was detected after a.cid hydrolysis of apoI,p-Val (D1) from two subjects.
Optical Studies-The ultraviolet absorption spectrum of apoLP-Val is very similar to that of trgptophnn or -\--acctyl-upt,rypt.ophanamide (lo), reflect.ing the absence of tyrosiue (Fig. 7). At 280 nm a solution containing 1 mg per ml of apoLP-Val gives an absorbance of 0.91 as compared to 1.61 obtainetl with npoLPsialic acid per mole of protein in Da and Dd, respectively.
It is Ala in the same concentration. This high value for the latter protein is due to the presence of the 3 moles of tryptophan and 2 moles of tyrosine per mole of protein.
When analyzed in 6 M guanidine, no shift was noted in the position or magnitude of the maximum at 280 nm and shoulder at 290 nm obtained wit,h either protein in the absence of guanidine.
This suggests that the exposure of the aromatic group of tryptophan and tyrosine to the solvent is not changed by exposure to guanidine.
The circular dichroic spectrum of apoLP-Val included strong and well resolved bands at 222 nm and 207 nm (Fig. 8). The band at 222 mn is frequently associated with t,he n + n-transition of the peptide bond in the a! helix.
If the intensity of absorption at this wave length was produced by this configuration of the peptide bond, these studies suggested a significant content of 01 helical structure.
The ORD spectrum of apoLP-Val was also consistent with some content of ar helix, there being a minimum at 232 mn and a shoulder at 212 to 216 nm. The CD and ORD spectra of apoLP-Ala were quite different from those of apoLP-Val (Fig. 9). The major CD band was at 204 nm with a shoulder at 220 to 230 nm. ORD measurements on the same sample revealed a negative Cotten effect with a minimum at 228 to 230 nm. While these spectra suggest that apoI,P-,\la contains mainly disordered structure, they do not exclude small contributions from helical or fi structure. The molecular weight,s det,ermined for the three VLD apolipoproteins, apoLl'-Val, apoLP-Glu, and apoLP-Ala were approximately 7,000, 10,000, and 10,000, respectively.
A previous estimate of 14,000 for apoLP-Ala had been based solely on the yield of terminal residues by carboxypeptidnse (1). These are the smallest of the apolipoproteins thus far isolated from human plasma. The two major high density apolipoproteins, threonine apolipoprotein and glutamine apolipoprotein, have molecular weights of approximately 15,000 (15). All of the apolipoproteins have a strong tendency to aggregate, and measurements of molecular weights by sedimentation equilibrium have required the presence of dissociating substances such as guanidine or urea (15). This requires application of a technique such as that of Sclmchman and Edelstein (7) in the determination of molecular weight to accommodate the preferential interactions that occur between the protein and low molecular weight constituents of the system. It is also recognized that the values of vc obtained in these experiments are high compared to some other proteins, and it is therefore reassuring that the estimates of molecular weight obtained from amino acid composition agree well with those obtained by ultracentrifugation.
could be due to a small content of OL helix or /3 structure or might come from the aromatic chromophores of tyrosine and tryptophan that have strong absorption bands in the wave length range (18). The CD and ORD spectra of apoLP-Val suggest a significant content of helical structure.
It is of interest that the two major apolipoprot.eins of high density lipoprotein from the density range 1.125 to 1.195 g per cm3 contain appreciable quantities of (Y helical configuration (19). The apoprotein of low density lipoprotein (d = 1.019 to 1.063) appears to contain a mixture of /3, a! helical, and random coil structure (20).
Polymorphism of the apoLP-Ala has been shown to be related to a difference in sialic acid content of roughly 1 mole per mole between the two predominant forms as observed on DEAE-cellulose chromatography and polyacrylamide gel electrophoresis. Similar changes in apparent charge and mobility of protein due to the sialic acid content have been shown for alkaline phosphatase and leucine aminopeptidase in chicken plasma (21). The presence of both forms of apoLP-Ala has been observed in each sample obtained from three male and two female normal subjects and three patients each with type IV and type V hyperlipoproteinemia.
Nothing further is known of the carbohydrate moieties of the smaller VLD apolipoproteins.
It is apparent that the VLD lipoprotein complexes must contain a large number of different protein molecules.
For example, the T'LD particle of molecular weight 15 x 106, contains about 107, protein (16) by weight, of which half consists of the smaller apolipoproteins, apoLl'-Vnl, apoLP-Glu, and apoLP-Ala. The average particle would the11 contain approximately 75 to 100 molecules of these npolipoproteins in addition to the contribut.ions of the npolipoproteins which are apparently identical with those of low density lipoprotein and high density lipoprotein (2).
Shore and Shore have recently published studies of the VLD apolipoproteiiis (17). They obtained a total of six different fractions by DEhE-cellulose chromatography. From the amino acid composition alone, none of these can be positively identified as corresponding to the proteins referred to here as apoLP-Val or apoLl'-Glu, but the composition of Peaks 1 and 2 obtained by Shore and Shore is very similar to that, of apoLP-Ala.
Several differences in our technique may be noted.
Shore and Shore utilized plasma from nonfasting patients and a slightly different technique of delipidation.
They also did not, employ an initial separation by gel filtration and used a Tris buffer in 8 in urea for the DEXE-cellulose chromatography. We do not know which of these differences in methods might account for the greater heterogeneity of VLD apolipoproteins encountered by Shore and Shore.