Isolation and Characterization of High Density Lipoprotein Apoproteins in the Non-human Primate (Vervet)*

Six different apoproteins have been isolated and characterized from vervet high density lipoprotein (HDL). The apoproteins were isolated and purified by a combination of gel and ion exchange chromatography along with preparative isoelectric focusing. Measured properties of the apoproteins included: relative mobil- ity on urea-polyacrylamide gel electrophoresis (PAGE), isoelectric point, molecular weight, amino acid com- position, sialic acid content, ability to activate purified lipoprotein lipase, and relative content in HDL. Based on these characteristics, several analogies were seen between human and vervet HDL apoproteins.

A-I (apo-A-I) was polymorphic, had a molecular weight of 28,000, and was the major apoprotein of vervet HDL (69% of total HDL protein).
Vervet apo-A-II, a monomeric protein with a molecular weight of 9,900, was the second most abundant HDL apoprotein (11% of total HDL protein).
Vervet apo-C-111 was a glycoprotein (2 mol of sialic acid/m01 of protein) with a molecular weight of 9,500. Apo-C-11 of the vervet was a potent activator of lipoprotein lipase, but it differed chemically (sialic acid and amino acid analysis) from human apo-C-II. The vervet threonine-poor apoproteins were small (iI& = 13,900 and 11,500) monomeric proteins that were variable in amount among HDL of individual animals. APO-C-I was not present in quantities sufficient for isolation and no a&nine-rich apoprotein was detected in vervet HDL. It was concluded that: 1) vervet and human HDL apoproteins are similar based on chemical, physical, and functional characteristics, and 2) the relative amounts of the threonine-poor apoproteins in vervet HDL are greater and more variable than in normal human HDL. and decrease with dietary modification such as a high cholesterol or high polyunsaturated fat diet, or both (2)(3)(4). HDL concentration also has been inversely correlated with the incidence of coronary heart disease (5) and increased HDL concentration is now thought to afford protection from atherosclerosis (6). These HDL responses have prompted the study of metabolism of the HDL apoproteins; presumably availability of at least some of the HDL apoproteins helps determine the plasma HDL concentration (7). However, studies of environmental influences on HDL metabolism are difficult to conduct in human beings since absolute control of the subjects' environment is hard to achieve. We have sought to use nonhuman primates as animal models for study of HDL metabolism in atherosclerosis because of the close phylogenetic relationship of nonhuman and human primates. We were encouraged by the reported similarities in apo-A-I and apo-A-II of two nonhuman primate species compared to those of humans (8,9). In order to find the most appropriate model we have examined the dietaryinduced modifications of plasma lipoproteins of several nonhuman primate species. One of the species which has appeared promising is the African green monkey (Cercopithecus aethiops) of the vervet subspecies. In a study of male vervets, plasma LDL and HDL cholesterol were lowered in response to a diet rich in polyunsaturated fat,' a finding similar to that found in humans by Shepherd et al. (3). Vervets ingesting a high fat diet have a lipoprotein composition and distribution similar to that of North American human beings (4). Earlier studies have shown that the African green monkey develops diet-induced atherosclerotic lesions with topography and morphology similar to those seen in human beings (10,11).
The present study was undertaken to characterize the apoproteins of HDL in vervet monkeys in preparation for studies on diet-induced modifications of HDL metabolism. We report the isolation and some chemical, physical, and functional characteristics of HDL apoproteins in male vervet monkeys fed diets containing 40% of calories from fat and cholesterol levels covering the range of those in the diet of the American human primate population.

6717
After the run was completed, the gels were stained overnight in 0.05% Coomassie blue in 5% sulfosalicylic acid, 10% trichloroacetic acid and destained with 7.5% acetic acid. RF values were determined as the distance from the gel interface that the protein migrated divided by the distance that the bromphenol blue dye band migrated. A nomenclature system for the proteins found in the three peaks eluted from the DEAE-cellulose column was assigned by RF values. If more than one protein from a peak was observed on a urea-PAGE gel, the protein with the lowest RF value was numbered 1 and the other faster migrating proteins on the gel were consecutively numbered, i.e. DI-1, DI-2, DII-1, and DII-2 (Fig. 5 Winter et al. (16), with slight modification.
The gel slurry consisted of 5 g of Ultradex, 39 ml of 10 M urea, 5 ml of Ampholines (LKB Instruments, Inc.), and 74 ml of H20. Lyophilized, lipid-free apoprotein samples (4 to 18 mg) were added to the gel slurry and the slurry was evaporated to 60% of the evaporation limit stated by the manufacturer for the Ultradex gel. This procedure was adopted so that the gel matrix would remain fully hydrated in the presence of urea. All isoelectric focusing runs were done at 0°C with a pH 4 to 7 gradient. Gradients were made by using an equal volume (2.5 ml)  an apparent molecular weight of 27,800 k 1020 (mean f standard error). This protein co-migrated with and appeared to be analogous to human apo-A-I (see below). It subsequently will be referred to as apo-A-I. Vervet apo-A-I was polymorphic and had at least four major bands as determined by urea-PAGE and IEF (Figs. 3 and 4, respectively).
Apo-HDL, apo-A-I, and AcA34-III, the pooled material cusing was used to purify the small proteins. Peak ZZZfrom the AcA34 column (Fig. l), were included for reference.
From Peak I (Fig. 2), DI-1 had a mobility on urea-PAGE of Designations are based on the DEAE-column elution and urea-PAGE RF, as described under "Experimental Procedures." RF = 0.38 and accounted for about half of the protein of the DI peak (Fig. 3). The position of DI-1 on urea-PAGE over-not activate purified lipoprotein lipase ( Table Ill). The lapped with that of apo-A-I (Figs. 3 and 5). The DI-1 protein amount of DI-1 in HDL was much more variable than that had a p1 = 6.94. After final purification by IEF, the DI-1 for the other apoproteins of HDL. The relative content of DIprotein migrated as a single band on urea-PAGE (Fig. 5) and 1 in vervet HDL ranged from 1 to 81% of the apo-A-I content IEF (Fig. 6); its mobility in these systems was not modified by in individual HDL samples (Table IV). Shown in Figs. 5 and /3-mercaptoethanol.
A molecular weight of 13,900 f 376 was 6 are representative gels of apo-HDL that illustrate the varifound for DI-1 by SDS-PAGE on 20% acrylamide gels (Fig. ability in DI-1 content; the sample on the right contained very 7). A low content of tbreonine, valine, and leucine, an absence little DI-1, while that on the left had substantial amounts. of tryptophan, and a high content of aspartic acid, glycine, The characteristics of this apoprotein were similar to those of and alanine was found by amino acid analysis of the protein one of the threonine-poor apoproteins of human HDL ( Table  (Table II). DI-1 did not contain detectable sialic acid and did V).

Isolation and Characterization of Vervet HDL Apoproteins
The second protein (DI-2) eluting from the DEAE column Peak I (Fig. 2) had a faster migration on urea-PAGE gels (RF = 0.46; Fig. 3) and was more acidic (p1 = 5.17; Fig. 4). The DI-2 purified by IEF was a single band on urea-PAGE (Fig. 5) and IEF (Fig. 6)  affected by ,&mercaptoethanol. A molecular weight of 9,900 + 308 was obtained by SDS-PAGE (Fig. 7). DI-2 did not contain the amino acids isoleucine, histidine, and tryptophan, while the amino acid in highest content was glutamic acid (Table II). The protein had no detectable sialic acid and did not activate purified lipoprotein lipase (Table III). DI-2 was the second most abundant protein in vervet HDL and represented 8 to 13% of the total apo-HDL. The characteristics of this apoprotein indicated it represented the vervet equivalent to human apo-A-II, and it will subsequently be termed apo-A-II (Table V). The range of individual animal variability of apo-A-II content in vervet HDL relative to apo-A-I was 0.189 to 0.121 (Table IV), with an apparent molar ratio of 2:l (A-I: A-II), a ratio similar to that reported for humans (35).
On urea-PAGE DII-1 was the slower migrating of the two proteins of DEAE Peak II (RF = 0.44). Most of the DII-1 protein eluted in the back region of Peak II of the DEAE column (Figs. 3 and 4). DII-1 had a p1 = 6.44 (Fig. 4). After IEF purification, this protein migrated as a single band on urea-PAGE (Fig. 5) and IEF (Fig. 6); its mobility was not altered in either system when run in the presence of /3-mercaptoethanol.
The position of DII-1 on urea-PAGE gels overlapped with that of apo-A-II (DI-2) and the two proteins appeared as a single wide band on urea-PAGE in all of the individual apo-HDL samples (Fig. 5). The molecular weight of DII-1 was found to be 11,500 f 351 (Fig. 7). The protein was low in threonine, valine, and leucine, while high in aspartic acid, glycine, and alanine (Table II). Methionine and tryptophan were not present in DII-1. The protein did not contain sialic acid and did not activate purified lipoprotein lipase (Table III). The amount of DII-1 protein in individual HDL samples ranged from 4 to 12% of that amount of protein present as apo-A-I (Table IV). This protein had many similarities to the threonine-poor apoproteins of human HDL (Table V). Ratios were obtained by densitometrlc scanning at 635 nm of analytical isoelectric focusing gels. Apo-HDL samples were obtained from individual animals (n = 51) consuming a variety of dietary regimens. The content of small proteins was normalized to A-I content by using a protein/A-I ratio, as the amount of apo-A-I in HDL varied little from sample to sample. Characteristics summarized are relative mobility on urea polyacrylamide gel electrophoresis (RF), isoelectric point (PI), molecular weight (M,), moles of sialic acid/m01 in protein (SA), abllit.y to activate purified linonrotein lipase (LPL). and protein/ape-A-I ratio.  The second major protein (DII-2) which eluted from DEAE Peak II (Fig. 2) had a migration on urea-PAGE of RF = 0.51 ( Fig. 3). It was more acidic (p1 = 5.20; Fig. 4) than DII-1. DII-2 was present in equal amounts in the front and back regions of DEAE column Peak II (Figs. 2 and 3). Complete purification of DII-2 was not accomplished even after repeated ion exchange chromatography and preparative isoelectric focusing; DII-2 contained a minor faster migrating contaminant band which was approximately 7% of the stained protein as determined by densitometric scanning (Fig. 5). However, a single band was obtained for "purified" DII-2 on IEF gels (Fig. 6). The mobility of DII-2 on urea-PAGE gels was not affected by ,&mercaptoethanol.
The molecular weight of DII-2 was estimated to be 8,000 f 335 by SDS-PAGE (Fig. 7). The amino acids methionine, histidine, and tryptophan were not present in DII-2, while glutamic acid was present in the highest amount. DII-2 was found to be a glycoprotein containing 31.2 * 2.6 pg of sialic acid/mg of protein (five determinations) or 1 mol of sialic acid/m01 of protein. The activity of purified lipoprotein lipase was stimulated 'I-fold above control levels by DII-2 (Table III). The content of DII-2 in individual HDL samples was proportional to the content of apo-A-I (Table IV) and showed minimal variation. Based on its property of lipoprotein lipase activation, this protein would seem to be the monkey's equivalent to apo-C-II (Table V).
DIII was the only protein isolated in significant quantities from DEAE Peak III (Fig. 2). It was the fastest migrating protein on urea-PAGE gels (RF = 0.59; Fig. 3) and the most acidic protein of the apo-HDL proteins as seen on IEF gels (p1 = 5.05; Fig. 4). The purified protein migrated as a single band on urea-PAGE (Fig. 5) and IEF gels (Fig. 6) and its mobility was not affected by fi-mercaptoethanol.
DIII had an apparent molecular weight of 9,500 + 255 (Table II). Isoleutine was not present in DIII, and glutamic acid and alanine were the two most abundant amino acids of the protein (Table  II). DIII was also a glycoprotein and contained 52.3 + 3.0 pg of sialic acid/mg of protein (three determinations) or 2 mol of sialic acid/m01 of protein. A slight activation (2-fold) of purified lipoprotein lipase was found with DIII (Table III). Of the isolated and purified HDL proteins, DIII was found to be the least abundant among individual HDL samples, and the amount of DIII was proportional to the amount of apo-A-I and relatively constant (Table IV). Vervet DIII appeared analogous to human apo-C-III. DISCUSSION We have reported the isolation and characterization of six different apoproteins from vervet HDL. Comparison of our purified vervet HDL apoproteins with known human HDL apoproteins has indicated that similarities exist in some of the chemical, physical, and functional characteristics.
Vervet apo-A-I and -A-II closely parallel their human counterparts in all measured characteristics (amino acid analysis, mobility on urea-PAGE gels, sialic acid content, molar ratio, lipoprotein lipase activation, and molecular weight) except one; human apo-A-II (36) can exist as a dimer (two identical monomers linked by a disulfide bond), while vervet apo-A-II does not. Remarkable similarity between vervet DI-1, DII-1, and the two human threonine-poor apoproteins and vervet DIII and human apo-C-III also was found for all measured characteristics. However, vervet DIII was not as heterogeneous as human C-III; DIII was present only as one glycoprotein species (2 mol of sialic acid/m01 of protein). DII-2 was the only vervet apoprotein which differed from its presumed human counterpart, apo-C-II, based on chemical and physical characteristics. However, functional similarity was indicated for vervet DII-2 and human apo-C-II by their mutual ability to activate lipoprotein lipase. The many similarities between human and vervet HDL apoproteins indicate that the vervet is a useful animal model for the study of HDL apoprotein metabolism.
For example, vervet HDL contains higher amounts of the threonine-poor apoproteins than human HDL (37), and vervet HDL concentrations on the average are twice the concentration of the human population, 5602 versus 279 mg/dl(38).
Study of vervet threonine-poor apoprotein metabolism may afford the opportunity to determine if these apoproteins have a role in the control of plasma HDL concentrations.
The most abundant protein of vervet HDL (69% of total HDL protein) had chemical and physical characteristics similar to human apo-A-I. The similarities between vervet and human apo-A-I included molecular weight, amino acid composition, elution profile from gel chromatography column, percentage of total HDL protein, and heterogeneity on urea-PAGE and IEF gels (39). Both vervet and human apo-A-I had an apparent molecular weight of 28,000 and neither contained sialic acid. Minor differences were seen in the content of aspartic acid, glutamic acid, alanine, valine, lysine, and arginine (Table II). There was also close agreement in the amino acid composition of apo-A-I from the vervet and that from the baboon and rhesus monkey (8,9). Charge heterogeneity on urea-PAGE and IEF gels has been described for human apo-A-I (39) and was observed for vervet apo-A-I samples (Figs. 5 and 6). Immunological microheterogeneity of vervet apo-A-I using an antiserum prepared from purified apo-A-I also was detected.3 However, the immunological microheterogeneity was not due to different antigenic sites on each polymorphic form of apo-A-I (as visualized on IEF gels) but, rather, was due to different antigenic regions within the primary sequence (NH? versus COOH-terminal), similar to that noted for human apo-A-I (40). IEF gels of purified apo-A-I appeared more heterogeneous than the apo-A-I region from gels of HDL samples (Fig. 6). Osborne and Brewer have suggested that apo-A-I heterogeneity might be due to protein concentration steps which utilize lyophilization or dialysis (34). Our results would support their findings as the apo-HDL samples were concentrated only once by lyophilization, while the purified apo-A-I required several concentration steps during isolation. We have ruled out the possibility that the apo-A-I charge heterogeneity was due to carbamylation by the fact that homocitrulline was not detected in apo-A-I samples by amino acid analysis.
DI-2, a small monomeric protein (Mr = 9,900), was the second most abundant HDL protein for most vervet monkeys. Its chemical characteristics, including amino acid analysis, molecular weight, percentage of total HDL protein, and mobility on urea-PAGE gels, were similar to those of apo-A-II from human, baboon, and rhesus monkeys (8,9,41). The amino acid analysis of apo-A-II for all four species (human, rhesus, baboon, vervet) is remarkably similar. Vervet, rhesus, and baboon apo-A-II were all missing the amino acids isoleutine, histidine, tryptophan, and arginine. APO-A-II was slightly smaller (Mr = 8500) in the human, baboon, and rhesus monkey than we have found using SDS-PAGE for vervet DI-2 (Mr = 9900). Vervet apo-A-II was found to be a monomeric protein by SDS-PAGE; human apo-A-II, however, can exist as two identical subunits (Mr = 8,500) connected by an interchain disulfide bond (36,41). The monomeric nature of nonhuman primate apo-A-II also has been described in the baboon and rhesus monkey (8,9). Although apo-A-II was the second most abundant HDL protein in most animals, there were some animals which had greater amounts of DI-1 relative to apo-A-

Isolation
and Characterization of Vervet HDL Apoproteins II. This resulted primarily from the variability in the DI-1 level because the variability in the relative amount of apo-A-II was minimal (Table IV).
Vervet DII-2 was compared with human apo-C-II on the basis of chemical, physical, and functional characteristics. Both DII-2 and human apo-C-II have similar mobility on urea-PAGE gels and the molecular weight estimate for DII-2 on SDS-PAGE gels was 8,000 compared to 8,837 for human apo-C-II determined from the primary structure (42). Major differences exist for the amino acid composition of DII-2 and human apo-C-II (Table II); the content of threonine, serine, and glutamic acid was the most different of the amino acids. DII-2 was found to be a glycoprotein (Table V), although human apo-C-II does not contain sialic acid (43,44). DII-2 and human apo-C-II have functional similarity in that they both are activators of lipoprotein lipase (Table III; Refs. 45 and 46). Perhaps DII-2 is more effective than human apo-C-II in the activation of lipoprotein lipase due to the apo-C-II versus DII-2 chemical differences (amino and sialic acid). This could be a possible explanation for the low plasma triglyceride and VLDL concentrations in the vervet monkey (23 mg/dl)' consuming a variety of dietary regimens (4). Based on the functional similarities, we have concluded that DII-2 is the vervet's equivalent of human apo-C-II.
DIII was compared with human apo-C-III on the basis of chemical, physical, and functional characteristics.
DIII and apo-C-111 were similar in mobility on urea-PAGE gels in which both were the fastest migrating proteins (Fig 5; Ref. 47) and amino acid analysis (Table II). Both DIII and human apo-C-III were glycoproteins; however, apo-C-III contained 0, 1, or 2 mol of sialic acid/m01 of protein (47), while DIII contained 2 mol of sialic acid/m01 of protein (Table V). Similar heterogeneity due to different numbers of sialic acid residues apparently did not exist for vervet DIII. DIII had some potential for activating lipoprotein lipase (Table III); a similar finding has been reported for human apo-C-III (45,46,48).
Based on our findings we would conclude that DIII is analogous to human apo-C-III.
Recently, Shore et al. have described two "new" threoninepoor apoproteins that appear to be minor protein components in HDL of normal human beings (37). Many similarities exist between the two human threonine-poor apoproteins and vervet DI-1 and DII-1. Their appearance on urea-PAGE and IEF gels was remarkably similar; p1 values were 6.94 and 6.44 for DI-1 and DII-1, respectively, while the values for the human threonine-poor apoproteins were 6.5 and 6.0. DI-1 and one of the human threonine-poor apoproteins were found to migrate in the approximate position of apo-A-I on urea-PAGE, while DII-1 was found to overlap the migration of DI-2 (vervet apo-A-II).
The overlapping migration on urea-PAGE gels of the threonine-poor apoproteins with apo-A-I and -A-II may explain why the threonine-poor apoproteins were not detected in human HDL until amphotericin B treatment increased their content relative to other HDL apoproteins (37). The amino acid composition of DI-1 and DII-1 agreed reasonably well with that of the human threonine-poor apoproteins (Table II). These similarities suggest that vervet DI-1 and DII-1 are analogous to the human threonine-poor apoproteins (Table V). However, there was disparity in the molecular weight data between human threonine-poor apoproteins and vervet DI-1 and DII-1. Shore et al. reported that the less acidic threonine-poor apoprotein (PI= 6.5) existed mainly as dimeric protein (&Zr = 40,000) linked by a disulfide bond (37). Upon reduction of the disulfide bond a M, = 25,000 monomer was found. The more acidic threonine-poor apoprotein (p1 = 6.0) had a molecular weight of 10,000. The vervet DI-1 and DII-1 were found to be monomeric and had molecular weight values of 13,900 and 11,500, respectively ( Table V).
The amount of the threonine-poor apoproteins in normal human HDL is low. Treatment of humans with amphotericin B resulted in HDL that contained 25% or more of the total protein as the threonine-poor apoproteins (37). Normal vervet HDL contained 1 to 56% of the total HDL protein as DI-1 and 8 to 13% as DII-1 (Table IV). When this higher percentage is considered together with the higher concentration of total HDL protein in vervets, one concludes that much higher concentrations of these apoproteins are normally present in these monkeys compared to threonine-poor apoprotein levels in human beings. Shore et al. reported that the threoninepoor apoproteins were found in higher amounts in monkeys hyporesponsive to dietary cholesterol (37). In preliminary studies, we have not found any correlation between the amount of DI-1 or DII-1 in HDL and total serum cholesterol or a-lipoprotein cholesterol. We have not seen a consistent relationship of the levels of these apoproteins with the type of dietary fat or the level of dietary cholesterol.
No apparent difference in fasted versus fed plasma samples was found. Further study is needed to define the metabolic role of these HDL apoproteins and to learn the significance of higher plasma levels of these apoproteins.
Two apoproteins which have been detected in human HDL, apo-C-I and apo-E, were not present in vervet HDL in amounts sufficient for isolation. HDL samples from individual animals had a protein whose p1 was more basic than DI-1 (see HDL gels in Fig. 6); apo-C-I from human HDL samples has a similar appearance on IEF gels (37). However, the very low content of this protein in HDL would have made its isolation extremely difficult. We can only speculate that this protein may be a vervet analog for human apo-C-I. We have found apo-E in the size region between LDL and HDL peaks of the agarose column fractionation of plasma lipoproteins, but apo-E was not present in the HDL peak which was used for apoprotein isolation. It appears that the vervet has a larger sized HDL subfraction present in minor amounts that contain apo-E. This may be analogous to the situation in human beings* (49).