Transgenic Mice Carrying the Apolipoprotein E3-Leiden Gene Exhibit Hyperlipoproteinemia”

Apolipoprotein (apo) E3-Leiden, described in a large Dutch family, is associated with a dominantly inherited form of familial dysbetalipoproteinemia. To study the effect of the APOE*3-Leiden mutation in uiuo, transgenic mice were generated using a genomic 27-kilobase DNA construct isolated from the APOE*3-Leiden proband. This construct carried the APOE gene, the APOCl gene, and all known regulatory elements including an element that mediates liver expression. Three strains were generated that showed human APOE and APOCl expression. All strains had significantly elevated levels of total plasma cholesterol and triglycerides on a regular diet. When mice of one strain were fed a semisynthetic cholesterol-rich diet, total plasma cholesterol and triglyceride levels increased dramatically. This increase was observed mainly in the very low density lipoprotein (VLDL)and low density lipoprotein (LDL)-sized fractions. In cholesterolfed mice, the apoE3-Leiden protein became equally distributed between the VLDL/LDL and HDL-sized fractions, while in mice kept on a regular diet, apoE3Leiden protein was mainly associated with HDL-sized fractions. The presence of hyperlipoproteinemia in the APOE*3-Leiden-expressing transgenic mice supports our finding that the apoE3-Leiden variant behaves like a dominant trait in the expression f familial dysbetalipoproteinemia. ApoE3-Leiden transgenic mice may serve as a model to elucidate additional factors involved in the metabolism of apoE containing remnant lipoproteins in general and the etiology of familial dysbetalipoproteinemia in particular.


Transgenic Mice Carrying the Apolipoprotein E3-Leiden Gene Exhibit
Hyperlipoproteinemia" (Received for publication, November 3, 1992, and  Apolipoprotein (apo) E3-Leiden, described in a large Dutch family, is associated with a dominantly inherited form of familial dysbetalipoproteinemia. To study the effect of the APOE*3-Leiden mutation in uiuo, transgenic mice were generated using a genomic 27-kilobase DNA construct isolated from the APOE*3-Leiden proband. This construct carried the APOE gene, the APOCl gene, and all known regulatory elements including an element that mediates liver expression. Three strains were generated that showed human APOE and APOCl expression. All strains had significantly elevated levels of total plasma cholesterol and triglycerides on a regular diet. When mice of one strain were fed a semisynthetic cholesterol-rich diet, total plasma cholesterol and triglyceride levels increased dramatically. This increase was observed mainly in the very low density lipoprotein (VLDL)-and low density lipoprotein (LDL)-sized fractions. In cholesterolfed mice, the apoE3-Leiden protein became equally distributed between the VLDL/LDL and HDL-sized fractions, while in mice kept on a regular diet, apoE3-Leiden protein was mainly associated with HDL-sized fractions. The presence of hyperlipoproteinemia in the APOE*3-Leiden-expressing transgenic mice supports our finding that the apoE3-Leiden variant behaves like a dominant trait in the expression of familial dysbetalipoproteinemia. ApoE3-Leiden transgenic mice may serve as a model to elucidate additional factors involved in the metabolism of apoE containing remnant lipoproteins in general and the etiology of familial dysbetalipoproteinemia in particular.
Apolipoprotein (apo)' E is one of the major structural components of chylomicron and very low density lipoprotein (VLDL) remnants (1). ApoE functions as a ligand in the receptor-mediated uptake of these remnant lipoprotein particles from the blood by the liver. Mutant forms of apoE have been described leading to an impaired clearance of remnant particles by the liver and thus resulting in familial dysbetalipoproteinemia (FD) or type I11 hyperlipoproteinemia (for a review, see Ref. 2). More than 90% of FD patients are homo-* This research was supported by the Netherlands Heart Foundation, the Dutch Royal Academy of Arts and Science, and the Dutch Foundation of Scientific Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The APOE*3-Leiden mutation is characterized by a tandem duplication of codons 120-126 and yields a mature protein of 306 amino acid residues (4, 5 ) . Family studies have demonstrated that this mutation is associated with a dominantly inherited form of FD (6). Recently, we have shown that the expression of FD in apoE3-Leiden carriers was influenced by other factors, such as body mass index and the second APOE allele (7). T o further investigate the effect of APOE mutants on lipoprotein metabolism, these genes can be introduced in mice, allowing control of both genetic and environmental factors. Recent studies already demonstrated the feasibility of manipulating the mouse lipid metabolism by introducing human genes and studying the effect of overexpression (8-12). Overexpression of rat apoE in mice resulted in a reduction in plasma lipoproteins and resistance against diet-induced hypercholesterolemia (13).
It has previously been shown that liver expression of APOE can be obtained with constructs including 23 kb of 3' sequences, which harbor the APOCl gene and a DNA element mediating liver expression (14). No abnormalities in lipid phenotype were observed in these mice, which express both APOE and APOCl at high levels.
In the present study, we have examined the effect of the expression of the APOE*3-Leiden gene in transgenic mice. We used a 27-kb fragment containing the APOE*3-Leiden gene in addition to the APOCl gene and the liver-positive element. ApoE3-Leiden transgenic mice display hyperlipoproteinemia, which becomes more prominent during cholesterol feeding. Since the apoE3-Leiden-expressing transgenic mice have a hyperlipoproteinemic phenotype, we have generated a mouse model to study lipoprotein remnant metabolism.

MATERIALS AND METHODS
Cosmid Library-Genomic DNA used for the construction of a cosmid library was prepared from proband C.V. of the apoE3-Leiden pedigree (7). Proband C.V. is a heterozygous carrier for both the mutant APOE*3-Leiden allele and the common APOEt2(Arg-158 + Cys) allele. For the construction of the library, DNA was partially digested with MboI and cloned into the Supercosl vector (Stratagene). After transformation into DH5e-MCR cells (Bethesda Research Laboratories), approximately 8 X lo5 colonies were obtained and screened using a human APOCl cDNA probe (15). Recombinant clones were characterized using standard techniques (16).
Generation and Analysis of Transgenic Mice-Preparation of the DNA fragment for microinjection and the production of transgenic mice was performed according to standard procedures (17). In brief, a DNA solution (2.5 pg/ml) was microinjected into male pronuclei of fertilized mouse eggs taken from superovulated (C57BWJ x CBA/ J)F1 females. Thirty-six pups were born. Genomic DNA was isolated from 1-cm portions of their tails by standard procedures (17). Polymerase chain reaction analysis was performed on genomic tail DNA using a primer set, spanning the APOE*3-Leiden mutation (4). Transgenic founder mice were bred with C57B1/6J mice to establish transgenic strains. Transgenic mice of strains 2, 181, and 195 and nontransgenic littermates of generations F1 and F2 (between 6 and 10 weeks of age) were used for experiments. For Southern blot analysis, 7 pg of genomic DNA was digested with the appropriate restriction enzyme, fractionated on 0.7% agarose gels, and transferred to Hybond N+ nylon membrane filters (Amersham). Filters were hybridized with a 32P-labeled human APOCl cDNA probe, according to Church and Gilbert (18). Hybridization signals were scanned with a PhosphorImager using the software program Imagequant (Molecular Dynamics) and compared to that of a human control sample.
Isolation of Total RNA and Northern Blot Analysis-Total RNA was isolated from kidney and liver tissue, according to Cromczynski and Sacchi (19). Some 10 pg of total RNA was separated through formaldehyde-containing agarose gels and blotted to Hybond N+ nylon membranes, according to the manufacturer's recommendations. Blots were hybridized using 32P-labeled probes of human APOCl cDNA, human APOE cDNA (20), mouse Apoe cDNA (21), or 28 S rRNA (human genomic 7.3-kb EcoRI-fragment cloned in pAT153). Hybridizations were performed at 55 "C in a solution containing 50% formamide according to Krumlauf (22). Hybridization signals were scanned using the PhosphorImager as above.
Analysis of Human ApoE and ApoCl Proteins-Lipoproteins were isolated by centrifugation of sera for 1.5 h (at 28 p.s.i.g.) in an Airfuge, using an A-100 fixed-angle rotor (Beckman), followed by dialysis overnight at 4 "C against phosphate-buffered saline. Samples were subjected to SDS-polyacrylamide gel electrophoresis, using 5-20% gradient gels. Proteins were transferred to nitrocellulose membranes, which were incubated with either polyclonal goat anti-human apoE antiserum or rabbit antiserum recognizing human apoC1, apoC2, and apoC3. As a second antibody, rabbit anti-goat IgG or goat anti-rabbit IgG antibodies, conjugated to horseradish peroxidase, were used. Human apoE concentrations in serum were measured by enzymelinked immunosorbent assay as described earlier by Bury et al. (23).
Measurement of Lipids and Lipoproteins-After an overnight fasting period, approximately 300 pI of blood was obtained through tail bleeding. Cholesterol and triglyceride (without free glycerol) were measured enzymatically, using commercially available kits, CHOD-PAP (No. 236691) and No. 877557, respectively (Boehringer Mannheim).
Lipoprotein patterns were determined by gel filtration chromatography through a Superose 6B column (25 ml) (Pharmacia, Uppsala, Sweden), using fast protein liquid chromatography equipment. Fasted sera of three mice of the F2 generation were pooled. Some 200 pI of pooled serum was injected into the column and eluted with phosphatebuffered saline at a rate of 0.5 ml/min. The effluent was collected in 0.5-ml fractions.
Diets-Mice were housed under standard conditions in conventional cages and given free access to water and food. Mice were given regular mouse diet (SRM-A, Hope Farms, Woerden, The Netherlands). In one experiment, mice of 6 weeks of age were given a semisynthetic cholesterol-rich diet, containing 40.5% sucrose (w/w), 15% cacao butter (w/w), 1% cholesterol (w/w), and 0.5% sodium cholate (w/w) (Hope Farms, Woerden, The Netherlands), essentially according to Nishina et al. (24), for a period of 9 weeks.

RESULTS
Generation of Transgenic Mice Carrying the Human APOE*3-Leiden Gene-A cosmid library prepared from genomic DNA of an APOE*3-Leiden carrier was screened with a human APOCl cDNA probe. Some 20 positive clones were identified. Three cosmids were found to carry both the APOE*3-Leiden gene and the APOCl gene. A 27-kb gene fragment was obtained from one of these cosmids by digestion at the KpnI site located 4 kb 5' of the APOE gene and the ClaI site located in the cosmid vector adjacent to the cloning site (Fig. 1). This fragment contains the APOE*3-Leiden gene, the APOCl gene, and the complete intergenic region between APOCl and the APOCl pseudogene, harboring the postulated liver regulatory element (25).
The gene fragment was injected into fertilized eggs of (C57B1/6J x CBA/J)Fl mice. Six out of thirty-six newborn mice were found to carry the APOE*3-Leiden transgene, as could be detected by polymerase chain reaction analysis, using primers for the APOE gene (not shown). Three transgenic strains 2, 181, and 195 were established by breeding the founder mice with C57B1/6J mice and used for further study.
The mode of integration of the APOE*3-Leiden/APOCl construct in the three strains was examined more closely by Southern blot analysis using a human APOCl cDNA probe. Correct bands were observed in HindIII-digested DNA for every strain (Fig. 2). Digestion with HindIII yielded a 4.8-kb internal fragment and a junction fragment of approximately 21 kb, which will arise when multiple copies have integrated in a head-to-tail manner. In all strains, single unique 3' border fragments were observed, suggesting that the gene construct is integrated at a single chromosomal site in all cases. The inheritance patterns of the transgenes were in agreement with this observation. Scanning of the hybridization signals indicated that strains 2 and 181 contained 4 and 3 copies, respectively, and strain 195 contained 2 copies (data not shown).
Expression of the Transgenes-Total RNA of both kidney and liver tissue was isolated to study the expression of the transgenes. Northern blot analysis showed that all three transgenic strains express human APOE and APOCl mRNA in the liver (Fig. 3). The expression level in liver of human APOE mRNA was lower in strain 195 than in strains 2 and 181. Also for APOC1, mRNA expression levels were lower in strain 195 than in strains 2 and 181. Expression of the transgenes in strains 2 and 195 was found in the liver and not in the kidney. However, in strain 181, considerable expression of human APOE and APOCl mRNA in kidney was observed as well. Northern blot analysis of total RNA isolated from nontransgenic littermates showed that mouse Apoe mRNA was expressed a t high levels in liver and only in very low amounts in kidney (Fig. 3). Expression of the endogenous mouse Apoe mRNA in the transgenic strains was comparable with that of nontransgenic littermates.
Apolipoprotein Analysis-The size of human apoE3-Leiden produced in the transgenic strains was compared to mouse and human apoE by SDS-polyacrylamide gel electrophoresis, using 5-20% gradient gels, and subsequent Western blotting (Fig. 4). For immunostaining of the blot, a polyclonal antibody was used that showed partial cross-reaction with mouse apoE. Human apoE3-Leiden produced by the transgenic strains was indistinguishable in size from a human apoES/ES-Leiden control sample, but clearly somewhat larger than mouse apoE. With the same method of SDS-polyacrylamide gel electrophoresis and Western blotting using antibodies recognizing human apoC1, apoC2, and apoC3, significant amounts of human apoCl were found in the transgenic mice (Fig. 4). In the human control lanes, all three apoC proteins are visible, as expected. The results show that in strains 2 and 181 the levels of human apoE and apoCl are much higher than in strain 195. These findings agree with the plasma level of human apoE of strains 181 and 195, which is 71 k 29 mg/dl and 3 k 2 mg/dl, respectively (Table I).
Lipoprotein Analysis-On a regular diet, total plasma cholesterol and triglycerides were elevated about 2 and 4 times, respectively, in transgenic animals of strains 2 and 181, when compared to nontransgenic littermates (Table I) ing with anti-human apoE. Note that the antibody cross-reacts with mouse apoE. Bottom panel, immunostaining with antiserum recognizing human apoC1, apoC2, and apoC3. The positions of human and mouse apoE, as well as human apoC1, are indicated. Extra bands in the human lanes correspond to human apoC2 and apoC3. ' p < 0.05, indicating the difference between transgenic and nontransgenic groups of mice on the same diet, using nonparametric Mann-Whitney test.

Effect of human APOE*3-Leiden expression on serum cholesterol and triglyceride levels in transgenic mice
Values are the mean f S.D. of four mice per group. d p < 0.01. 195, total plasma cholesterol and triglycerides were elevated about 1.5 to 2 times. After separation of lipoprotein fractions by gel filtration chromatography of serum using Superose 6B, it is obvious that the increase of cholesterol and triglycerides in transgenic mice of strains 2 and 181 occurred mainly in the VLDL/LDL-sized fractions (Fig. 5, A and B ) . The amount of cholesterol in HDL-sized fractions was comparable to that observed in the nontransgenic littermates (Fig. 50). In the less hyperlipidemic strain 195, the distribution of cholesterol and triglycerides over the lipoprotein fractions was comparable to that of nontransgenic control mice (Fig. 5, C and 0).
Mice of strain 181 and nontransgenic control mice were fed a cholesterol-rich diet for 9 weeks. This diet led to an approximately 15 times increase of total plasma cholesterol in trans- genic mice, whereas in nontransgenic mice only a 2 times increase was observed ( Table I). Total plasma triglycerides in the transgenic mice were almost 2 times increased, whereas in nontransgenic mice triglycerides were decreased to barely detectable levels (<0.1 mmol/liter). The elution pattern, obtained after Superose 6B gel filtration chromatography of serum, showed that the increase in cholesterol upon cholesterol feeding was mainly confined to the VLDL/LDL-sized fractions (Fig. 6). In addition, the HDL fraction has moved to a larger size range, in particular in the transgenic mice. After cholesterol feeding, triglycerides were present only in VLDL/LDL-sized fractions (not shown).
On a regular diet, in strain 181, the majority of human apoE3-Leiden protein was associated with HDL-sized particles (Fig. 7). However, upon cholesterol feeding, about onehalf of the human apoE3-Leiden was associated with lipoproteins in the VLDL/LDL-sized range. Human apoCl co-eluted with human apoE3-Leiden on both diets.

DISCUSSION
FD is a genetically heterogeneous disorder of lipoprotein metabolism predisposing to premature coronary and/or peripheral vascular disease (for a review, see Ref. 26). Homozygosity for the APOE*2(Arg-158 -Cys) allele has been recognized as the primary cause in more than 90% of FD  Samples (50 pl) were loaded on 5-20% SDS-polyacrylamide gels. After Western blotting, immunostaining was performed using either goat anti-human apoE antiserum (top panel) or rabbit antiserum against human apoCl (bottom panel).
patients. Interestingly, heterozygosity for some rare apoE mutants, such as APOE*3-Leiden (7), APOE*2(Lys-146 -+ Gln) (27), and APOE*3(Cys-112 + Arg; Arg-142 + Cys) (28) is associated with dominantly inherited forms of FD. In addition to this genetic heterogeneity, FD is characterized by a highly variable expression of the hyperlipidemic phenotype, which has been shown to be influenced by environmental and genetic factors (3, 7). In the present study, we describe a transgenic mouse model carrying APOE*3-Leiden that may serve to elucidate some of these additional factors involved in lipoprotein remnant metabolism.
Three independent transgenic mouse strains were generated carrying two to four copies of a 27-kb APOE*3-Leiden/ APOCl gene construct. Two strains (2 and 195) show liverspecific expression of both APOE and APOC1, which confirms earlier results with a similar gene construct (14). In one strain (181), the transgene is expressed also in the kidney, which might be due to a position effect. However, it has been shown that kidney-produced apoE associates normally with lipoproteins (14, 29). The total human APOE mRNA level is lower in strain 195 than in strains 2 and 181, while the endogenous Apoe mRNA level is comparable to that in control mice. In all transgenic strains, both plasma cholesterol and plasma triglycerides are significantly elevated, which is mainly due to increased levels of VLDL/LDL-sized particles. This increase is most prominent in the strains with the highest apoE3-Leiden plasma levels. During cholesterol feeding, plasma cholesterol increases to extreme values ( X 0 mmol/liter). Also, human subjects with FD respond strongly to dietary changes. In addition, the observation that the majority of apoE3-Leiden was associated with lipoproteins in the size range of VLDL indicates that in these transgenic mice the expression of the APOEL3-Leiden gene causes, like in humans, an impaired clearance of apoE3-Leiden-containing lipoproteins. Interestingly, the difference in protein levels between strains 2 and 195 is much larger than the difference in mRNA levels. Apparently, a small increase of APOE*3-Leiden mRNA level leads to a progressive plasma accumulation of apoE3-Leiden containing VLDL/LDL-sized lipoproteins, which suggests that a certain threshold level is required to cause hyperlipidemia. Despite the presence of the endogenous mouse Apoe gene, the transgenic mice show a hyperlipidemic phenotype, which implies that the effect of the APOE*3-Leiden mutation is dominant in mice as well.
Previous reports demonstrated that transgenic mice overexpressing rat apoE had decreased plasma cholesterol and triglyceride levels in a dose-dependent manner (13). This study confirmed an earlier finding, that cholesterol levels could be lowered in rabbits by intravenous injection of purified apoE (30). Our transgenic mice also express the human APOCl gene. High level expression of APOCl in transgenic mice was shown to lead to elevated cholesterol and triglyceride levels (25) suggesting that the hyperlipidemia in the APOE*3-Leiden mice is caused by APOC1. Although apoE and apoCl influence lipoprotein levels, mice that express high levels of both APOE and APOCl do not show hyperlipidemia (25). The gene construct of the present study closely resembles the construct used in the former study. Therefore, it is likely that APOE*3-Leiden is the major cause of hyperlipidemia in these mice.
In man, mutations in the APOE gene mainly disturb chylomicron remnant VLDL and metabolism, leading to an accumulation of these particles in plasma (FD patients). This study suggests that the severity of the hyperlipidemia depends on the level of expression of APOE*3-Leiden. This transgenic mouse model carrying APOE*3-Leiden may help to study additional factors modulating the metabolism of apoE-containing remnant lipoproteins in general and the etiology of FD in particular.