Identification and DNA Sequence of a Human Apolipoprotein E cDNA Clone*

cDNA clones encoding human apolipoprotein E were identified by screening an adult human liver cDNA library with an oligonucleotide probe. The probe was a mixture of synthetic 14-base long DNA oligomers constructed to correspond to all possible codons for apo-E amino acids 218-222. Plasmids from four of the 20 clones selected by this screening procedure were digested with PstI and all had five internal PstI sites with a total length of the cDNA insert of approximately 900 base pairs. DNA sequence analysis of one of these clones, designated pE-301, revealed that it corresponded to apo-E amino acids 81-299, and contained a standard termination codon, polyadenylation signal, and poly A tail. The DNA sequence examined included the known apo-E polymorphic sites at amino acids 112, 145, and 158, and the mutant apo-E phenotypes can all be explained on the basis of a single base substitution in the first position of each of these codons. This work supports the hypothesis that the apo-E polymorphism is due to mutations in the region of DNA coding for the apo-E structural gene. Apo-E was fiist identified in 1973 as a component of normal human VLDL’ (1). This apoprotein has subsequently been demonstrated in chylomicrons, LDL and HDL, and exists in all mammalian species studied thus far (2-5). Apo-E synthesis occurs in liver and reticuloendothelial cells (6-11). Apo-E is secreted from its 82-mm 1,OOO colonies/dish. After growth and chloramphenicol amplification, the colonies were transfered to nitrocellulose filters and hybridized to the oligonucleotides probe which had been labeled at its 5' terminus (30). The hybridization mixture contained 0.75 M NaCI, 0.15 M Tris-C1 (pH 10 mM EDTA, 0.1% bovine serum albumin, 0.1% polyvinyl pyrrol- idone, 0.1% Ficoll, 0.1% sodium pyrophosphate, 0.1% SDS, 100 pg/ml yeast t-RNA (carrier), and 0.65 pg 5' oligonucleotide lo8 0.05% Luria supplemented with 20 pg/ml tetracycline. Plasmid DNA was isolated using the alkaline lysis and further on CsCl gradients. PstI pur- and the enzymatic the

cDNA clones encoding human apolipoprotein E were identified by screening an adult human liver cDNA library with an oligonucleotide probe. The probe was a mixture of synthetic 14-base long DNA oligomers constructed to correspond to all possible codons for apo-E amino acids 218-222. Plasmids from four of the 20 clones selected by this screening procedure were digested with PstI and all had five internal PstI sites with a total length of the cDNA insert of approximately 900 base pairs. DNA sequence analysis of one of these clones, designated pE-301, revealed that it corresponded to apo-E amino acids 81-299, and contained a standard termination codon, polyadenylation signal, and poly A tail. The DNA sequence examined included the known apo-E polymorphic sites at amino acids 112, 145, and 158, and the mutant apo-E phenotypes can all be explained on the basis of a single base substitution in the first position of each of these codons. This work supports the hypothesis that the apo-E polymorphism is due to mutations in the region of DNA coding for the apo-E structural gene.
Apo-E was fiist identified in 1973 as a component of normal human VLDL' (1). This apoprotein has subsequently been demonstrated in chylomicrons, LDL and HDL, and exists in all mammalian species studied thus far (2)(3)(4)(5). Apo-E synthesis occurs in liver and reticuloendothelial cells (6)(7)(8)(9)(10)(11). Apo-E is secreted from its sites of synthesis as sialo apo-E and subsequently desialated in plasma (10,12). Plasma apo-E is involved in receptor-mediated recognition of lipoprotein particles by tissues (13,14). Extrahepatic and hepatic nonreticuloendothelial cells possess a receptor which recognizes both apo-B and apo-E (13,14). Hepatic tissues also exhibit a receptor which only recognizes apo-E (14). In humans, studies utilizing two-dimensional polyacrylamide gel electrophoresis have re-* This work was supported by Grant HL15895 from the National Institutes of Health and by a grant from the Hood Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Established Investigator of the American Heart Association. The abbreviations used are: VLDL, very low density lipoprotein; LDL, low density lipoprotein; HDL, high density lipoprotein; DMtrT, 5-0-dimethoxytrityl-thymidine; DMtrAp-(MeO)-NEta, 5"dimethoxytrityladenosine-3'-(methoxy)-diethylaminophosphine; HLP, hyperlipoproteinemia; HPLC, high pressure liquid chromatography; bp, base pairs; SDS, sodium dodecyl sulfate. vealed six apo-E phenotypes (15)(16)(17)(18). Through family studies it was shown that these apo-E phenotypes are the result of a single structural gene locus for apo-E with three common alleles designated €4, €3, and €2 (16)(17)(18)(19). According to a new uniform nomenclature system, these phenotypes and the corresponding genotypes are E4/4 = €4, €4; E3/3 = €3, €3; E2/2 = €2, €2; E4/3 = €4, €3; E3/2 = €3, €2; and E4/2 = €4, €2 (19). Studies of apo-E phenotypes in the general population have shown that the most common allele, €3, occurs with a frequency of 0.74, whereas the other two alleles each occur with frequencies of from 0.11 to 0.15 (18,20). Further studies have shown that familial type I11 HLP, a disease characterized by hyperlipidemia, xanthomatosis, and premature atherosclerosis is associated in over 90% of cases with the apo-E phenotype E2/2 (18). In in vitro receptor binding studies, apo-E of the E2/2 phenotype isolated from patients with type I11 HLP has been shown to bind poorly to apo-B/E receptors (21). Thus, it appears that a mutation in an allele specifying an apo-E structural gene may be the underlying defect that leads to altered lipoprotein metabolism and the clinical symptomatology in this disease.  have recently studied the protein sequence of apo-E from individuals with different apo-E phenotypes. Through these studies, they have identified three polymorphic amino acid sites. In the most common apo-E polypeptide specified by the €3 allele, these sites are as follows: Cys 112, Arg 145, and Arg 158. The apo-E allele, €4, differs from the €3 allele by having a Cys 112 + Arg substitution (22-24). In the case of the apo-E allele, €2, individuals with type I11 HLP and the apo-E phenotype E2/2 have been found with either an Arg 145 4 Cys, or an Arg 158 "-f cys substitution (22-24). Furthermore, Ghiselli, et al. (25) have described an individual with type 111 HLP who has no detectable apo-E. To better understand the genetic determinants of functional apo-E in humans, as well as the regulation of apo-E synthesis, it will be necessary to study the genomic organization and sequence of the apo-E gene and its transcripts. As a first step towards this goal, we have isolated and characterized bacterial clones containing portions of human apo-E cDNA.

MATERIALS AND METHODS
Preparation of the Oligonucleotide Probe-To probe a human adult liver cDNA library for apo-E cDNA clones, an oligonucleotide mixture was synthesized which corresponds to apo-E amino acids 218 to 222 (Met-Glu-Glu-Met-Gly) (22-24). This region was selected because these amino acids are specified by relatively unambiguous codons. Fig. 1 shows the sequence of mRNA from 5' to 3' specifying apo-E amino acids 218 to 222, as well as the cDNA sequence from 3' to 5'. This cDNA sequence was synthesized by a solid phase phosphate triester method using the reaction conditions and procedures of Matteucci and Camthers (26,27) and Beaucage and Caruthers (28). Briefly, a sample of 25 mg of functionalized silica gel (29), charged with 1.3 pmol of DMTrT attached by its 3"OH group through ester linkage to the solid phase, was unblocked a t the 5'-OH group with a Lewis acid (saturated ZnBr,/aqueous CH:3N02). The sample was condensed with 10 mg of the protected nucleoside phosphoramidite of adenosine (DMtrAp-(MeO)-NEt2) activated with tetrazole.
The nucleoside phosphoramidites used throughout this procedure were obtained from ChemGenes Corporation, Waltham. MA, Any unreacted 5'-OH of the silica gel-bound nucleoside was blocked subse- The phosphites were next oxidized with iodine to phosphates under quently by reaction with a large excess of a very reactive phosphate.
previously described conditions (26). The cycle was repeated with the appropriate nucleoside phosphoramidite(s) until the last condensa- tion was performed, subsequent to which the 5"OH was not unblocked. At each point where an ambiguity in the DNA sequence existed, a mixture of derivatized nucleosides was employed as indicated in Fig. 1. By this procedure, the resultant probe was a mixture of 4 oligomers. All reactions were carried out in a fitration device fashioned from Teflon and stainless steel. After the synthesis was completed, the methyl groups of the phosphodiesters were removed by treatment with thiophenol (26) and the ester bond joining the oligonucleotide to the support was cleaved by treatment with concentrated ammonia as were the base-protecting groups (26). The reaction products were then fractionated by preparative HPLC using a Waters C-8 column. The sample was loaded in 0.1 M triethylammonium bicarbonate, pH 7.0, and eluted by a linear gradient up to 25% acetonitrile over 40 min. This procedure separates failure sequences from the desired trityl-oligomers, which emerge at the top of the gradient. Detritylation in 80% acetic acid for 20 min at room temperature was then followed by a second preparative HPLC under the same conditions. A dominant peak emerging approximately halfway through the gradient proved to be the desired tetradecamer mixture. This peak was subjected to polynucleotide kinase labeling at the 5' terminus with [y3'P]ATP and run on a 20% polyacrylamide gel. The gel was then subjected to autoradiography, and over 95% of the oligomer ran as a 14-base long nucleotide.
Screening of the Adult Human Liver cDNA Library-The cDNA library used in these studies was generously provided by Drs. D.
Plasmid DNA Preparations and Restriction Analysis-Bacterial clones were grown in Luria broth supplemented with 20 pg/ml of tetracycline. Plasmid DNA was isolated using the alkaline lysis method (31) and further purified on CsCl gradients. PstI was purchased from New England Biolabs and the conditions for enzymatic digestion were those recommended by the vendor.
DNA Sequencing Methods-DNA sequencing was performed both by the chemical method of Maxam and Gilbert (32) and the enzymatic method described by Sanger et al. (33). In the latter case, appropriate restriction fragments were excised from low melt agarose and cloned in M13mp7 (34). The recombinant single stranded phage DNA was used as a template for the sequencing reactions with the M13 sequencing primer purchased from New England Biolabs.

RESULTS
Identification of apo-E cDNA CZones"10,OOO cDNA clones were transferred to nitrocellulose filters and hybridized to the oligonucleotide probe as described under "Materials and Methods." Initial washing of the filters at 23 "C showed significant nonspecific hybridization, while washings at 30 or 40 "C substantially reduced this background, and positive clones could be identified. About 30 positive clones were observed after the 30 "C wash, 20 of which maintained the hybridization signal after the 40 "C wash. A 50 "C wash totally  abolished the signal from all clones. Four out of the 20 clones selected after the 40 "C wash were used for further analysis. The plasmid preparations from these four clones were digested with PstI and all four plasmids had five internal PstI sites with a total length of the cDNA inserts of approximately 900 bp. DNA sequence analysis of one of these clones was undertaken to verify that it corresponded to apo-E cDNA.
Sequencing of Clone pE-301"Clone pE-301 has been mapped by a combination of PstI digestion and DNA sequencing of the fragments obtained. Fig. 2 shows the location of the PstI sites in pE-301 and their relationship to the amino acid sequence of the apo-E polypeptide as reported by Rall et al. (22,24) and Weisgraber et al. (23). Fig. 3 shows the complete DNA base sequence of this clone. As can be seen, clone pE-301 corresponds to apo-E amino acids 81-299 and contains a TGA termination codon, a 158-bp long 3' untranslated region and a 44-bp long portion of the poly A tail of the apo-E mRNA. At the polymorphic sites 112,145, and 158, this cDNA clone specifies cysteine, arginine, and arginine, respectively.

DISCUSSION
In previous studies, we have shown that apo-E is a relatively abundant secretory product of human fetal and adult liver in organ culture and of human hepatoma cells in tissue culture (10,12). In addition, we have recently shown that approximately 0.2% of the protein synthesized in a cell-free translation system of human liver cytoplasmic poly A containing RNA is apo-E (35). This implies that apo-E mRNA is abundant in a total liver mRNA preparation and, therefore, human liver cDNA libraries should be enriched in clones containing the apo-E cDNA sequence. Based on the amino acid sequence for apo-E, we designed and synthesized a mixture of oligonucleotides and used these as a probe to screen an adult human liver cDNA library in an attempt to isolate these apo-E cDNA clones. The region that spans apo-E amino acids 218-222 is specified by relatively unambiguous codons and was selected for the construction of the 14-base long oligonucleotide probe. It has been shown earlier that oligonucleotides 13-15 bases in length are sufficient for the detection of a unique gene in a genomic yeast DNA library (36). -To establish the hybridization condition for the screening, we used empirical formulae (37) to estimate that the T, of a perfectly matched 14-base long DNA oligomer with a 50% content of GC should be 51 "C. This indicated that hybridization at 1 M salt at 31-36 "C should satisfy the opposing requirements for sensitivity (low stringency) and specificity (high stringency). As predicted, screening of the cDNA bank with the mixture of labeled 14-base long DNA oligomers at room temperature (low stringency) showed a high background of nonspecific hybridization. However, hybridizations at 30-40 "C (high stringency) increased the specificity without significantly affecting the sensitivity of the hybridization signal. In this manner, 10, OOO clones of the cDNA library were screened using the probe under high stringency hybridization conditions and 20 clones were identified as strongly hybridizing colonies. Further examination of 4 of these 20 positive clones showed that they contained five internal PstI sites at similar positions and that their total length was approximately 900 bp. To verify that the DNA sequence contained in these clones was indeed the apo-E cDNA sequence, clone pE-301 was digested with PstI and the fragments isolated and used for DNA sequence analysis. The DNA base sequence coincides with what one would predict from the amino acid sequence reported by Rall et al. (22,24) and Weisgraber et al. (23).
The DNA sequence examined includes the apo-E polymorphic sites at amino acids 112, 145, and 158. This particular DNA clone specifies cysteine, arginine, and arginine, respectively, in these positions. Thus, the apo-E corresponding to this DNA clone would be of the phenotype E3 and this clone represents the €3 gene. Rall, et al. have demonstrated amino acid changes deviating from wild type (€3) of Cys 112 + Arg, Arg 145 + Cys, and Arg 158 + Cys. As shown in Fig. 4, based on the DNA sequence of our clone, each of these amino acid changes could be the result of a single base substitution in the first position of each of the codons. This work adds support to the hypothesis that the apo-E polymorphism is due to mutations in the region of DNA coding for the apo-E structural gene (22-24). Such structural gene mutations, it appears, are of great physiological importance and probably underlie a condition in humans called type I11 HLP which leads to hyperlipidemia, xanthomatosis, and premature atherosclerosis.