Human apolipoprotein A-I polymorphism. Identification of amino acid substitutions in three electrophoretic variants of the Münster-3 type.

Variant forms of apolipoprotein A-I (apo-A-I) have been shown to exist in the human population. One mutant form, referred to as apo-A-I-Münster-3, is one charge unit more basic than normal apo-A-I on isoelectric focusing gels. This variant has the same immunologic characteristics and molecular weight as normal apo-A-I. The apo-A-I-Münster-3 from subjects in three unrelated families (in two of which the trait has been shown to be transmitted as an autosomal co-dominant) has been analyzed by partial amino acid sequencing to define the cause of the electrophoretic abnormality. In the apo-A-I of family A, the abnormality was shown to occur in the smallest cyanogen bromide fragment, CB-2 (residues 87-112), and amino acid sequencing revealed asparagine instead of the usual aspartic acid at residue 103. Subjects with this mutant form have shown no signs of dyslipoproteinemia. The NH2-terminal cyanogen bromide fragment (CB-1, residues 1-86) from the apo-A-I of family B was shown to differ electrophoretically from normal CB-1, and amino acid sequencing revealed that a substitution of arginine for proline at residue 4 was responsible for this variant form. Analysis of the plasma lipids of one affected family B member demonstrated that the percentage of the total cholesterol that was esterified was somewhat lower than that normally observed. In a third family, family C, a variant having the same electrophoretic abnormality as the other two was determined to have an amino acid substitution at yet a different position. In this variant, histidine was found at residue 3 in the apo-A-I sequence, rather than the usual proline. In all three cases, the substitution could account for the electrophoretic abnormality. It is proposed that these three apo-A-I-Münster-3 variants be designated apo-A-I(Asp103----Asn), apo-A-I(Pro4----Arg), and apo-A-I(Pro3----His), respectively, to indicate the substitution that accounts for the abnormality in isoelectric focusing gels.

Variant forms of apolipoprotein A-I (apo-A-I) have been shown to exist in the human population. One mutant form, referred to as apo-A-I-Munster-3, is one charge unit more basic than norma1 apo-A-I on isoelectric focusing gels. This variant has the same immunologic characteristics and molecular weight as normal apo-A-I. The apo-A-I-Munster-3 from subjects in three unrelated families (in two of which the trait has been shown to be transmitted as an autosomal co-dominant) has been analyzed by partial amino acid sequencing to define the cause of the electrophoretic abnormality. In the apo-A-I of family A, the abnormality was shown to occur in the smallest cyanogen bromide fragment, CB-2 (residues 87-112), and amino acid sequencing revealed asparagine instead of the usual aspartic acid at residue 103. Subjects with this mutant form have shown no signs of dyslipoproteinemia. The NHz-terminal cyanogen bromide fragment (CB-1, residues 1-86) from the apo-A-I of family B was shown to differ electrophoretically from normal CB-1, and amino acid sequencing revealed that a substitution of arginine for proline at residue 4 was responsible for this variant form. Analysis of the plasma lipids of one affected family B member demonstrated that the percentage of the total cholesterol that was esterified was somewhat lower than that normally observed. In a third family, family C, a variant having the same electrophoretic abnormality as the other two was determined to have an amino acid substitution at yet a different position. In this variant, histidine was found at residue 3 in the apo-A-I sequence, rather than the usual proline. In all three cases, the substitution could account for the electrophoretic abnormality. It is proposed that these three apo-A-I-Munster-3 variants be designated apo-A-I(Asp,,, -+ Asn), apo-A-I(Pro4 4 Arg), and apo-A-I(Pro3 + His), respectively, to indicate the substitution that accounts for the abnormality in isoelectric focusing gels.
*This work was supported in part by Grant HL27455 from the United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This ance with 18 U.S.C. Section 1734 solely to indicate this fact. article must therefore be hereby marked "advertisement" in accord-$To whom correspondence should be addressed at The Gladstone Foundation Laboratories, P. 0. Box 40608, San Francisco, CA 94140.
Because of the possibility of metabolic defects caused by various apolipoprotein mutations, as has been demonstrated for apo-E (11,12), it is of particular interest to establish which specific regions of the apolipoproteins are affected by mutations and to study their possible impact on lipoprotein metabolism. In this paper, three different mutations that are responsible for the occurrence of three distinct apo-A-I-Miinster-3 variants are reported.

EXPERIMENTAL PROCEDURES
Subjects-Subjects with apo-A-I electrophoretic variants were detected by screening large populations (4).
Materials-Plasma for preparative isolation of lipoprotein fractions was obtained by plasmapheresis. Acrylamide, N,N,N',N'-tetramethylethylenediamine, and N,N'-methylenebisacrylamide were purchased from Bio-Rad. Sodium decyl sulfate was obtained from Eastman Kodak, and ampholytes were acquired from Serva (Heidelberg, W. Germany) and LKB (Bromma, Sweden).
Electrofocusing and Two-dimensional Gel Electrophoresis Techniques-Isoelectric focusing of serum was performed as previously described (1). Isoelectric focusing of apo-HDL or apo-A-I was performed as described (13), except that 0.75-mm thick vertical slab gels were used rather than rod gels. For two-dimensional electrophoresis, the discontinuous system of Neville was used (14). Immunological detection of apo-A-I after isoelectric focusing was conducted by immunoelectrophoresis in the second dimension (4). Lipoproteins were isolated by ultracentrifugation by the method of Have1 et al. (15). After dialysis against 0.9% NaCl, the HDL were delipidated with ethanokether (3:l) and pure ether (16). The apo-HDL were solubilized in 1% decyl sulfate in 0.01 M Tris-HCl, pH 8.2, and subjected to flatbed preparative isoelectric focusing in a pH gradient of 5 to 7 in 6 M urea (17). Protein bands were eluted from the gel with 1% decyl sulfate in 0.01 M Tris-HC1, pH 8.2, and checked for purity by isoelectric focusing. An alternative preparative isoelectric focusing procedure was performed using the LKB Immobiline system, pH 4.9 to 5.9 (LKB Application Note 321). Modifications included the substitution of an isokinetic gradient for a linear gradient, the inclusion of 6 M urea in the gel solutions, and the use of vertical slab gels. Purified apo-A-I or delipidated HDL were solubilized in 0.1 M Tris-HCl, pH 8.0, containing 1% decyl sulfate, 20% sucrose, 6 M urea, and loaded onto the top of the gel, which was previously rinsed with the same buffer. The samples were then overlaid with 10% sucrose followed by a layer of 0.01 M NaOH (upper electrolyte buffer). The lower electrolyte buffer consisted of 0.01 M glutamic acid (Sigma). The gels were electrophoresed overnight at 4 "C at a constant power of 5 watts. The focused apoproteins were visualized by soaking the gels in water (18). Individual visualized bands were then excised and eluted with 0.1 M Tris-HCl, pH 7.4, containing 4 M guanidine and 1 mM EDTA. Eluted proteins were exhaustively dialyzed against 5 mM NH,HC03, lyophilized, and resolubilized in 0.1 M NH,HCOa.
Cyarwgen Bromide Hydrolysis-For analytical purposes, the pure apo-A-I isoforms were precipitated with 10% trichloroacetic acid to remove the ampholytes and washed 3 times with 10% trichloroacetic acid at 0 'C. The trichloroacetic acid was removed from the precipitated protein by ice-cold acetone. The protein was solubilized in 1% formic acid and 1% decyl sulfate, and CNBr was added in a 500-fold molar excess to the methionine content (1,500 molar ratio to apo-A-I) and the mixture kept for 24 h at 25 "C. The CNBr peptides were lyophilized and subjected to isoelectric focusing in a pH gradient of 3.5 to 10. After separation by isoelectric focusing, the peptides were analyzed in the second dimension by sodium dodecyl sulfate gel electrophoresis (4) to determine their molecular weight.
For preparative purposes, the isoforms were dissolved in 70% formic acid and reacted 24 h at 25 "C with a 30-fold weight excess (7,900 molar ratio to apo-A-I) of CNBr (Pierce). The peptides were then chromatographed on a Sephadex G-50 column (2.2 x 190 cm) with 0.02 N HCl at room temperature (flow rate was 15 ml/h). The fractions were lyophilized and used for amino acid analysis and sequencing.
Amino Acid Analysis-The sample was hydrolyzed in 0.45 ml of 6 N HCl for 20 h at 110 "C in a sealed, evacuated tube. After the tube was brought to room temperature and opened, the contents were dried under vacuum at 40 "C. The sample was dissolved in 0.2 N sodium citrate, pH 2.2, and analyzed on a Beckman 121MB analyzer equipped with a model 126 data system. No corrections were made for hydrolytic destruction or incomplete release.
Sequence Analysis--Sequence analysis was performed on a Beckman 890C Sequencer equipped with a cold trap accessory. The lyophilized samples were dissolved in 0.5 ml of 50% acetic acid and applied along with 2 mg of Polybrene (Sigma). A standard 0.1 M Quadrol program (No. 122974) was used. After conversion in 1 N HCl at 80 "C for 10 min and extraction with ethyl acetate, PTH amino acids were identified and quantified on a Beckman model 332 liquid chromatograph system that was equipped with a CRlA integrator-recorder. The mobile phase, chromatography parameters, and criteria for identification have been described (10).
Lipid Analysis-Levels of cholesterol, triglycerides, and phospholipids were determined with a commercially available test kit (Boehringer Mannheim). Cholesteryl ester levels were determined by thin layer chromatography.

RESULTS
Isoelectric focusing of serum or apo-HDL in a pH gradient of 4 to 6 revealed in some rare cases (-0.1% of the population studied) the existence of an additional band with one more basic charge than normal apo-A-I (Fig. 1). As determined by two-dimensional electrophoresis, this band had a molecular weight identical with that of normal apo-A-I, M, = 28,000 (Fig. 2). After establishing that the abnormal protein was an apo-A-I isoform, it was designated apo-A-I-Munster-3 (4). After isoelectric focusing, this abnormal band was shown to react with a monospecific antibody to apo-A-I by immunoelectrophoresis (Fig. 3). By visual inspection of the peak (Fig.  3A), a ratio of nearly 1:l for normal and variant apo-A-I was estimated.
This electrophoretic variant has been identified in three families. Pedigree analysis in two of the three families (A and Representative lipid values for affected members of the families are shown in Table I. Values for affected members of family A and C appeared to be within normal limits. However, the analysis of the plasma lipids in family B suggested that the percentage of the total cholesterol that was esterified was lower than that which is normally observed (54% uersus a normal value of 65 to 80%).
To investigate the cause of the electrophoretic abnormality, the variant apo-A-I from the delipidated HDL of one individual (I-l, family A) was isolated by preparative isoelectric

normal apo-A-I (upper) and the apo-A-I-Miinster-3 from individual I-1 of family A (lower).
Included is a molecular weight standard with eight different marker proteins whose molecular weights are indicated. The nomenclature of CNBr peptides is according to Brewer et al. (19), and identification was based on molecular weight determination and isoelectric focusing position. Asterisks mark the abnormal CNBr peptides. focusing (Fig. 5). The purified variant isoform was fragmented with the "analytical" CNBr method described under "Experimental Procedures." This method gave relatively high yields of partial digestion products, which were desirable in this case because the small CNBr peptides of apo-A-I (CB-2 and CB-3) did not fix or stain well using these procedures. Although not conclusive, comparison of two-dimensional gels of normal and variant apo-A-I CNBr fragments suggested that some bands from apo-A-I-Munster-3 apparently focused more cathodically than those from normal A-I (Fig. 6). The tentative peptide designations were made on the basis of both PI (CB-4 is much more basic than CB-1 (19)) and approximate . . normal and variant apo-A-I (family A) apparently occurred not in the large peptides (CB-4 and CB-1) but in the partial digestion fragments that contained CB-2 (CB-1-2* and CB-The CNBr peptides from the variant isoform of family A were prepared and isolated by Sephadex G-50 chromatography (Fig. 7). The amino acid analysis of neither the whole variant apo-A-I-Munster-3,A nor its individual CNBr peptides gave a strong indication of an amino acid substitution 2*-3-4).
(Tables I1 and 111). Although amino acid analysis of CB-2* did not reveal any dissimilarity to normal CB-2, the sequence analysis of CB-2' (Table IV)

demonstrated a single substitution at cycle 17 (asparagine in apo-A-I-Munster-3,A as opposed to aspartic acid in normal apo-A-I). This corresponds to residue number 103 in apo-A-I (19).
When the methods described above were applied to the isolated apo-A-I-Miinster-3 of individual 1-1 of family B, it was observed that the CB-1 fragment of this variant form had a more basic PI value than that of normal CB-1 (Fig. 8).
Furthermore, amino acid analysis of the variant apo-A-1-Munster-3,B (Table 11) and its CNBr peptides (Table V) suggested that there was probably one less proline but one more arginine residue compared to either normal apo-A-I or the normal apo-A-I isoform from this subject. Therefore, sequence analysis of peptide CB-1* was undertaken. This analysis revealed that the occurrence of arginine instead of proline at residue 4 was the probable cause of the different PI value of this mutant (Table VI).
Sequence analysis of the first six residues of the intact apo-A-I-Munster-3,B variant isoform (2000 pmol) from the same subject confirmed the substitution at residue 4: step 1, Asp (850 pmol); step 2, Glu (1200 pmol); step 3, Pro (910 pmol); step 4, Arg (<200 pmol; Pro carryover to cycle 4 was 11%); step 5, Gln (530 pmol); step 6, Ser (trace). The amino acid sequence of the isolated peptide CB-2 (family B) was identical with that of normal apo-A-I, i.e. aspartic acid at residue 103 (Table VII), while the sequence of the first five residues of peptide CB-1 (family A) revealed proline at residue 4, the same as in normal A-I (data not shown). Therefore, it is definite that these two families do not share the same mutation.
A third subject (family C) was identified who also had the apo-A-I-Miinster-3 variant (as well as normal apo-A-I). Amino acid analysis of the variant apo-A-I isoform isolated from this individual suggested that there was probably one less proline but one more histidine residue compared to nor-   Present as homoserine lactone. The partial digestion fragment CB-1,2 (see Fig. 7) was recovered in 14% yield.

Residue
No."   19. The amino acid at residue 103 in normal apo-A-I is aspartic acid, as determined from both protein sequence (19) and apo-A-I cDNA sequence (25.26). mal apo-A-I (Table 11). Because there are three proline residues in the first seven amino acid residues of the normal apo-A-I polypeptide (19), the NH2-terminal sequence of the intact apo-A-I-Munster-3,C variant isoform was determined first. At residue 3, where there is normally proline in apo-A-I, histidine was found (Table VIII). This histidine substitution was the probable cause of the electrophoretic abnormality in this isoform.

FIG. 8. Isoelectric focusing of the CNBr peptides from the apo-A-I-Miinster-3 from individual 1-1 of family B (left) and from normal apo-A-I (right).
Isoelectric focusing was performed in a pH gradient of 3.5 to 10. The nomenclature of CNBr peptides is according to Brewer et al. (19), and identification was based on molecular weight determination and isoelectric focusing position. Asterisks mark the abnormal CNBr peptides. 'Data are given in residues/mol and are the average from two preparations. Numbers in parentheses are the number of residues in the corresponding peptides of normal A-I (19); ND = not determined. Boxed residues highlight the difference from normal A-I that was confirmed by sequence analysis.
No further sequence analysis was undertaken with any of these variants to check for the possibility of additional substitutions. Although we have identified the substitution in each case that can account for the electrophoretic abnormality, other compensating substitutions or isomorphic replacements cannot be ruled out. However, based on amino acid analyses of the variant isoforms and/or their CNBr fragments, no obvious further substitutions are apparent. There is a low probability that such substitutions exist in these variants. The amino acid at residue 4 in normal apo-A-I is proline, as determined from both protein sequence (19) and apo-A-I cDNA sequence (26).

TABLE VI ADO-A-I-Munster-3, family B, CB-l* sequence analysis
From aqueous phase analysis; ND = not determined.
As a further check to see whether amino acid analysis was truly sensitive to small differences in composition, the proapo-A-I isoform of the Munster-3,C variant (see Fig. 1) was isolated by preparative isoelectric focusing, and its composition was determined. As shown in Table 11, the amino acids known to occur in the propeptide of apo-A-I (20) were detected by amino acid analysis. Sequence analysis (Table IX) confirmed that this isoform was the pro-apo-A-I of the variant.

DISCUSSION
In this study, the nature of the difference between normal apo-A-I and apo-A-I-Munster-3 has been examined. It has been shown that at least three variant forms of the apo-A-1-Munster-3 type exist, which differ from one another in both the nature and sites of amino acid substitutions. For the variant form of family A, a substitution occurs at position 103, where asparagine replaces aspartic acid. This single substitution is sufficient to explain the more basic PI value of this apo-A-I-Munster-3. A single base change (point mutation) in the normal apo-A-I gene corresponding to this position could account for the occurrence of apo-A-I-Miinster-3,A. The substitution apparently has no effect on lipid metabolism in affected individuals, since all lipid values were in the normal range and the variant was present in the same lipoprotein fractions as normal apo-A-I. The asparagine at position 103 occurs in a region with a predicted &-helical conformation (21). Since both asparagine and aspartic acid have a polar side chain, the substitution probably does not alter the nature of the hydrophobic or hydrophilic face of this presumed From Ref. 19. The amino acid at residue 103 in normal apo-A-I is aspartic acid, as determined from both protein sequence (19) and apo-A-I cDNA sequence (25,26). In addition, our own analyses of the comparable peptide from several normal individuals gave the same sequence as above (not shown).
a-helix, which may be important in apo-A-I interaction with lipids (21).
In the apo-A-I variant of family B, the amino acid arginine occurs instead of proline at residue 4, while in the apo-A-I variant of family C, histidine occurs instead of proline at residue 3. These variants can also be explained by a point mutation in the apo-A-I gene, and each substitution is sufficient to explain the electrophoretic abnormality.
No known function has been ascribed to the NHz-terminal segment of apo-A-I. In normal human A-I and in primate A-I, consecutive proline residues are present at positions 3 and 4. In all other species, one of these prolines is absent (reviewed in Ref. 22 Because all individuals so far identified as having one of these three variants are heterozygous for a variant apo-A-I and normal apo-A-I, it is possible that the presence of normal apo-A-I in these subjects may mask or compensate for any lipid abnormality caused by the variant apo-A-I. Therefore, studies of the possible effects of these apo-A-I variants on protein-lipid and protein-protein interaction, and on activation of 1ecithh:cholesterol acyltransferase, must be carried out on the isolated variant isoform in vitro. Such studies are currently in progress. The data presented here show that the identical electrophoretic abnormalities of apo-A-I variants are not necessarily caused by the same mutation. The designation of these var-  "The amino acid at residue 3 in normal apo-A-I is proline, as determined from both protein sequence (19) and apo-A-I cDNA sequence (26).
From aqueous phase analysis; ND = not determined.

NHz-terminal sequence analysis of the pro-apo-A-I form of the Munster-3.C variant DO-A-I
Residue N O .

Cycle
No.

A-I-Munster-3
iants should therefore relate to the differences in amino acid sequences that account for the electrophoretic abnormality. The following nomenclature is proposed for the apo-A-1-Munster-3 of family A, apo-A-I(Asp,,, + Asn); for the apo-A-I-Munster-3 of family B, apo-A-I(Pro4 3 Arg); and for the apo-A-I-Miinster-3 of family C, apo-A-I(Pros + His). In like fashion, the previously documented apo-A-IMilano variant (13) can be described as apo-A-I(ArglV3 + Cys).