Proc. Natl. Acad. Sci. USA Vol. 90, pp. 2389-2393, March 1993 Medical Sciences Targeted modification of the apolipoprotein B gene results in hypobetalipoproteinemia and developmental abnormalities in mice (embryonic stem cells/gene targeting/hydrocephalus/exencephalus/atherosclerosis) GREGG E. HoMANICS*, TERRY J. SMITH*t, SUNNY H. ZHANG*, DENISE LEE*, STEPHEN G. YOUNGf, AND NOBUYO MAEDA*§ *Department of Pathology, The University of North Carolina at Chapel Hill, CB #7525, Chapel Hill, NC 27599-7525; and tGladstone Foundation Laboratories for Cardiovascular Disease, Cardiovascular Research Institute, Department of Medicine, The University of California, San Francisco, CA 94110-0608 Communicated by Oliver Smithies, December 17, 1992 (receivedfor review November 10, 1992)

ABSTRACT Familial hypobetalipoproteinemia is an auto- proteins (for reviews, see refs. 7 and 8). Such mutations cause somal codominant disorder resulting in a dramatic reduction in familial hypobetalipoproteinemia (HBL), a condition char- plasma concentrations of apolipoprotein (apo) B, cholesterol, acterized by a reduction in circulating apoB, cholesterol, and and ,-migrating lipoproteins. A benefit of hypobetalipopro- ,-lipoproteins. Humans homozygous for "null alleles" have teinemia is that mildly affected individuals may be protected a complete absence of apoB-containing lipoproteins and can from coronary vascular disease. We have used gene targeting be severely affected with symptoms of intestinal fat malab- to generate mice with a modified Apob allele. Mice containing sorption, fat-soluble vitamin deficiency, and neurological this allele display all of the hallmarks of human hypobetali- problems. The phenotypes of individuals homozygous for poproteinemia: they produce a truncated apoB protein, mutant APOB alleles leading to the synthesis of truncated apoB70, and have markedly decreased plasma concentrations apoB proteins are variable (7-10). Heterozygotes have one- of apoB, ,B-lipoproteins, and total cholesterol. In addition, the third to one-half of normal apoB and LDL-cholesterol levels mice manifest several characteristics that are occasionally and are almost always asymptomatic. Because of their low observed in human hypobetalipoproteinemia, including re- cholesterol levels, heterozygotes appear to be protected from duced plasma triglyceride concentrations, fasting chylomicro- coronary vascular disease and have a longer life expectancy nemia, and reduced high density lipoprotein cholesterol. An (10). unexpected finding is that the modified Apob allele is strongly Progress is being made toward understanding atherogen- associated with exencephalus and hydrocephalus. These mice esis by investigating humans and animals that are genetically should help increase our understanding of hypobetalipopro- prone to hyperlipidemia. A valid alternative is to study teinemia, atherogenesis, and the eitiology of exencephalus and humans and animals that are genetically protected from hydrocephalus. cardiovascular disease. With this in mind, we have generated a mouse model of HBL. To do this, we used a sequence insertion vector to disrupt exon 26 ofthe Apob gene in mouse Apolipoprotein (apo) B is a major structural component of embryonic stem (ES) cells. Mice carrying this disrupted gene very low density lipoprotein (VLDL), intermediate density synthesize apoB48 and a truncated apoB (apoB70) but no lipoprotein (IDL), low density lipoprotein (LDL), chylomi- apoBlOO. The lipoprotein phenotype of the mice is remark- crons, and lipoprotein(a). High plasma levels of apoB- ably similar to familial HBL in humans. Quite unexpectedly, containing lipoproteins are associated with an increased risk however, some of the mice also exhibit exencephalus and of coronary artery disease (1). apoB normally exists in two hydrocephalus, pathologic features that have never, to our forms, apoB100 and apoB48; both are the product ofthe same knowledge, been reported in human HBL. gene (2). The human gene spans 43 kb of DNA and contains 29 exons (3). MATERIALS AND METHODS apoBlOO, which contains 4536 amino acids, is synthesized Preparation of Targeting Constructs. An =l11-kb Bgl II exclusively in the liver and is secreted into the circulation as fragment containing exon 26 of the Apob gene was isolated a surface component of triglyceride-rich VLDL. It is a ligand from STO cell DNA (11). Two oligonucleotides which encode for the LDL receptor and is responsible for receptor- peptides of human f3s-globin (amino acids 1-12 and 120-131) mediated clearance of LDL by the liver and other organs. were used to replace nucleotides located in exon 26 of the Amino acids 3146-3159 and amino acids 3357-3368 are Apob gene (see Fig. 1) that encode two short amino acid thought to be important components of the LDL receptor- sequences thought to be important in binding to the LDL binding domain of apoBlOO (4). receptor (4). An in-frame stop codon was inadvertently apoB48 is formed as a result of post-transcriptional editing inserted into the 5' 3s-globin sequence while making the ofapoB mRNA, which changes codon 2153 into a stop codon constructs, but its presence later proved beneficial to the (5, 6). apoB48 is synthesized in small intestine and is required outcome of the experiments. 0-type (insertion) targeting for the packaging of dietary lipids into chylomicrons. A large plasmids were constructed with an 8.3-kb HindIII fragment percentage of mouse hepatic apoB mRNA is edited, indicat- ofthe Apob gene which includes the ,35-globin modifications, ing that the mouse probably makes apoB48 in the liver (D. F. an HPRT minigene [pnI2(IlS) of Reid et al. (12)], and Johnson, personal communication). apoB48 lacks the puta- pBluescript (Stratagene) (Fig. 1B). Constructs were made tive LDL receptor-binding domain of apoBlOO and does not interact with the LDL receptor. Abbreviations: HBL, hypobetalipoproteinemia; apo, apolipoprotein; In humans, many mutations in the APOB gene have been HDL, high density lipoprotein; LDL, low density lipoprotein; VLDL, identified that prevent the translation of full-length apoB very low density lipoprotein; IDL, intermediate density lipoprotein; ES cells, embryonic stem cells; HPRT, hypoxanthine phosphoribo- syltransferase. The publication costs of this article were defrayed in part by page charge tPresent address: BioResearch Ireland, National Diagnostic Centre, payment. This article must therefore be hereby marked "advertisement" University College, Galway, Ireland. in accordance with 18 U.S.C. §1734 solely to indicate this fact. §To whom reprint requests should be addressed. 2389 Downloaded by guest on September 30, 2021 2390 Medical Sciences: Homanics et al. Proc. Natl. Acad. Sci. USA 90 (1993)

A - 11.0 kb _ retro-orbital bleeding into tubes containing 0.3 mg of EDTA, 25 mg of gentamycin sulfate, and 11.5 milliunits of aprotinin (Sigma). Plasma was collected by centrifugation at 14,000 x probe B probe A g for 10 min at 4°C. Agarose gel electrophoresis of whole plasma and determination of total cholesterol, high density lipoprotein (HDL) cholesterol, and triglyceride were as pre- B **+ viously described (17, 18). Lipoproteins were fractionated by sequential density ul- tracentrifugation (18). Fractions were concentrated and de- salted by using a Centricon-10 (Amicon) apparatus according probe C to the manufacturer's protocol, and 20 ,ug of protein from each fraction was electrophoresed on an SDS/3-20% poly- C 16I16.5 kb 17.5 kb 12/12.5 kb acrylamide gradient gel (16 x 16 x 1.5 mm) at 150 V for 6 hr (18). The size of lipoproteins in the d (density) < 1.006 g/ml fraction of mouse plasma was assessed by electron micros- copy (19). Whole plasma apoAI concentrations were quan- term. P1 P2 tified by nonreducing gel electrophoresis as described by France et al. (20). FIG. 1. (A) Endogenous Apob locus, showing relative location of Northern Blot Analysis. Total RNA was prepared, electro- sequences coding for LDL receptor-binding domains (white boxes) phoresed, transferred to Hybond-N (Amersham), and hy- within exon 26 (solid black box). The horizontal arrow illustrates the bridized as described (21). Probes used include probes A, B, size of the endogenous Bgl II fragment. Probe A is a 1.1-kb C (see Fig. 1), apoAl (22), and human j3actin (Clontech). HindIII/Bgl II fragment. Probe B is a 2.35-kb EcoRI fragment. (B) Insertion type targeting construct. The sequences that encode the RESULTS AND DISCUSSION LDL receptor-binding domains of the endogenous Apob locus have Targeting of the been replaced with 1s-globin sequences as indicated by asterisks. Apob Locus in Mouse ES Cells. The scheme The downward arrow indicates the Sac I site used to linearize the of targeting of the Apob locus is illustrated in Fig. 1. A total construct prior to electroporation. Probe C hybridizes to plasmid of 1967 HPRT+ ES cell colonies were screened by PCR for vector sequences (thin line). H in stippled box indicates hypoxan- homologous recombination at the Apob locus. From 23 thine phosphoribosyltransferase (HPRT) DNA. (C) Targeted Apob PCR-positive colonies, 6 lines of ES cells were established in locus of cell line ESapoB6-3, which contains three repeats of Apob which targeting at the Apob locus was confirmed by Southern sequences that include the ps-globin modifications. In this locus, the blot analysis (Fig. 2). Probe A, which is external to the HPRT gene is in the opposite transcriptional orientation from the targeting DNA, detected an 11-kb Bgl II fragment from the Apob gene. The expected sizes of the Bgl II fragments are illustrated unaltered endogenous Apob gene in all cells. In correctly with horizontal arrows. The in-frame premature translation termi- nation codon introduced into the modified locus is illustrated (term.). targeted ES cells, an additional Bgl II band of 12 or 12.5 kb This will alter the amino acid sequence of the gene product at was detected, depending on the orientation ofthe HPRT gene positions 3142-3145 to Val-His-stop from Leu-Ser-Val-. P1 is a in the targeting constructs (Fig. 2A). In addition to these Ps-globin-specific PCR primer (5'-TTACACCTCCTGTCCAAGCC- bands, we also detected, with probe B, a 16.0- or 16.5-kb 3') and P2 is an apoB-specific PCR primer (5'-TATCAGCCAGT- fragment (depending on the orientation of HPRT) and a TCTTGCACG-3'). 17.5-kb Bgl II fragment in all six targeted cell lines (Fig. 2B). The size of this largest fragment and its intensity indicated having the HPRT gene in the same (pTS5) or opposite (pTS6) that multiple tandem copies of the targeting construct had orientation as the Apob gene. been integrated (see Fig. 1C). Such tandem incorporation of Targeting of the Apob Locus in Mouse ES Cells. Electropo- multiple copies of incoming DNA into a targeted locus has ration ofthe targeting constructs into the HPRT- ES cell line been reported by others (15, 23, 24). E14TG2a (13) and selection ofHPRT+ cells was as previously All Repeats in the Modified Apob Locus Have the 3Ss-Globin described (14, 15). ES cells surviving the selection procedure Modification. One of the targeted cell lines, ESapoB6-3, was were screened for targeted recombination by PCR essentially as described (16), using a human jBs-globin sequence-specific A B primer and an Apob sequence-specific primer (see Fig. 1C). cu ESapoB c ESapoB PCR-positive ES cell colonies were expanded and genomic DNA was prepared for Southern blot analysis. Blots were O.D 0CoX)o 00co hybridized to a 32P-labeled probe from outside the targeting II tr - n M - .1-- construct (probe A, Fig. 1A), to a probe from within the 23 - targeting construct (probe B, Fig. 1A), or to a plasmid probe 16- (probe C, Fig. 1B). Production of Mice with Modified Alleles ofApob. A total of 12 - 120 blastocysts were injected with one of the targeted cell 10- lines, ESapoB6-3. Following transfer to pseudopregnant re- cipient females, 21 pups were born, 13 of which were chimeric. Three male chimeras proved capable of germline 7.5 - transmission of the ES cell genome when mated with C57BL/6J mice. Offspring inheriting the ES cell genome, as FIG. 2. Southern blot analysis of Bgl II-digested DNA from judged by coat color, were screened for the presence of the targeted cell lines. E14TG2a is the parental ES cell line. ESapoB cell modified Apob allele by genomic Southern hybridization lines 6-3, 10-2-3, and 5D2 were derived from targeting construct analysis of tail DNA. pTS6. The ESapoB cell line 19-1 was derived from the targeting construct pTS5. The differences in the size of bands between these Lipid, Lipoprotein, and Apolipoprotein Analysis. Morpho- cell lines reflect differences in the targeting constructs. (A) Blot was logically normal, apparently healthy mice that had survived hybridized with probe A. The sizes ofA phage DNA molecular weight to adulthood were used for lipid and lipoprotein analysis. standards are indicated (in kb) to the left. (B) Same blot as in A after After an overnight fast, 200-400 ,ul of blood was collected by rehybridization with probe B. Downloaded by guest on September 30, 2021 Medical Sciences: Homanics et al. Proc. Natl. Acad. Sci. USA 90 (1993) 2391 used to generate animals, and the structure of the modified a small amount of apoB70, and about halfthe normal amount locus was determined by using DNA isolated from mice of apoB48. This finding is identical to the situation in humans homozygous for the alteration. By PCR amplification fol- with HBL, where truncated apoB proteins are invariably lowed by restriction enzyme digestion of the PCR products, present in very low concentrations (7). we were able to establish (data not shown) that ESapoB6-3 To investigate the mechanism responsible for the reduced contains three repeats of Apob sequences at the targeted amount ofplasma apoB proteins, we examined the amount of locus, each having LDL receptor-binding domains that have apoB mRNA in small intestine and liver. This analysis been changed to Is-globin sequences (Fig. 1C). showed that wild-type and heterozygous mice have similar Since a stop codon is present in all three ofthe 5' 3s-globin amounts of apoB mRNA (Fig. 4A); in contrast, the amounts sequences, this modified Apob allele should always produce of apoB mRNA in small intestine and liver of homozygous a transcript that contains this premature translation termina- mice are only 25% and 40%, respectively, of the amounts in tion codon at position 3145 (human equivalent) even if wild-type or heterozygous mice, as estimated by densitom- alternative splicing of the mRNA occurs. It should not have etry of the Northern blot. The size of the apoB mRNA (=14 a leaky phenotype such as has been suggested to occur after kb) transcribed from the modified allele is similar to that from disruption of the N-myc (25) and cystic fibrosis (26) genes; the unmodified allele. However, probe A, which is 3' to the rather it should result in an apoB protein approximately 70% site ofinsertion ofthe targeting construct, does not hybridize the size of apoBlOO. The translation termination codon to the apoB message from homozygous mice (Fig. 4B), and should not alter the size of apoB48. probe C, which detects plasmid-related sequences, hybrid- A Truncated ApoB, ApoB70, Is Produced from the Modified izes to mRNA from heterozygous and homozygous mice only Allele. SDS/polyacrylamide gel electrophoresis ofthe VLDL (Fig. 4C). Thus, the transcript of the mutant allele appears to and IDL/LDL fractions from wild-type mice reveals apoB48 terminate in the inserted DNA. Therefore, the modified and and apoBlOO (Fig. 3). In contrast, the VLDL and IDL/LDL unmodified alleles produce distinctly different transcripts fractions from homozygous mutants contain no detectable even though the sizes appear similar. The difference in the apoBlOO; only apoB48 and a truncated apoB are observed. level of the mutant apoB mRNA could be the result of We estimate the size of this truncated apoB protein to be differences in mRNA stability caused by the inserted se- -70% the size ofapoB100 on the basis ofthe observation that quences and/or insertional disruption of (unknown) tran- the truncated apoB migrates slightly faster than human scriptional regulatory sequences. Irrespective ofthe cause of apoB74, a proteolytic product of apoBlOO (data not shown). the reduced quantity of apoB mRNA, it is at least one factor We will refer to the truncated apoB as apoB70. The VLDL that leads to the reduced amount of apoB proteins in mice and IDL/LDL fractions from heterozygous mice contain all homozygous for the modified Apob allele. three apoB proteins, B48, B70, and B100. It should be noted HBL in Mice Homozygous for the Modified Apob Allele. that in the IDL/LDL fraction from heterozygous mice, the Agarose gel electrophoresis of total plasma from wild-type amount of apoB70 is approximately 9% the amount of mice shows that the majority of plasma lipoproteins migrate apoBlOO. Also, in heterozygous mice the amount of apoB48 as three distinct bands at the a, f, and pre-13 positions (Fig. relative to apoBlOO is increased. This parallels similar ob- 5). Electrophoresis of total plasma from the homozygous servations in humans who are heterozygotes for apoB67 and mutants reveals a virtual absence of (3-migrating lipoproteins. apoB83 mutations (27, 28). We estimate that apoB48 levels in homozygous mice are reduced by one-halfrelative to normal. small The intensities of staining of apoB48 in the IDL/LDL frac- intestine liver tions of wild-type and homozygous mice are nearly equal in +/+ +1- -1 +/+ +I -1- Fig. 3, but twice the amount of sample was loaded in the latter. Thus, our apoB mutation results in the production of 9.5- ..X . VLDL IDL/LDL HDL 7.5- --+l +1- -X- +1 +X- /X +1+ +1- /X B *w .a'... apoB100-_ apo B70 _ apo B48 --_ C ;AAAL.. -- Awdlig T. :1i,fip -140 -67 -43 apo E -_ apoAl -_ -30 -20 FIG. 4. Northern blot analysis of total RNA from small intestine (20 ug of RNA per lane) and liver (10 ,ug of RNA per lane) from wild-type (+/+), heterozygous (+/-), and homozygous (-/-) mice. (A) Northern blot hybridized with probe B. The size (in kb) of RNA standards is illustrated to the left. (B) Same blot as in A, after FIG. 3. SDS/PAGE of apolipoproteins isolated from VLDL, rehybridization with probe A (hybridizes to apoB sequences that are IDL/LDL, and HDL fractions of wild-type (+/+), heterozygous 3' of the insertion site of the targeting construct). The sizes of the (+/-), and homozygous (-/-) mouse plasma. Molecular mass is bands are the same as those in A. (C) Same blot as in A, after indicated on the right in kDa. The positions of apoBlOO, apoB70, rehybridization with probe C (recognizes plasmid sequences). The apoB48, apoE, and apoAl are indicated to the left. An amount ofeach sizes of the bands are the same as those in A. (D) Same blot as in A, lipoprotein fraction was loaded that is equivalent to the amount of after rehybridization with an apoAl probe. (E) Same blot as in A, plasma (in Al) shown below each lane. The identity of the apoB after rehybridization with a human 3-actin probe. This probe was proteins was confirmed by Western blotting using two different used to assess the integrity of the RNA preparation and to serve as apoB-specific antibodies (data not shown). a control for the amount of RNA loaded. Downloaded by guest on September 30, 2021 2392 Medical Sciences: Homanics et al. Proc. Natl. Acad. Sci. USA 90 (1993) Table 1. Plasma concentrations of total cholesterol, HDL

a- cholesterol, and triglyceride Conc., mg/dl Pre-p- :3 C- ., -Pre-P Total HDL Total Genotype (n) cholesterol cholesterol triglyceride c- -C Wild type (17) 111.2 ± 24.1 76.9 ± 13.1 56.9 ± 29.4 +/+ +1- / normal I human human Heterozygous (16) 87.6 ± 14.7 63.1 ± 11.6 41.3 ± 17.0 B37/B86 P < 0.005* P < 0.005* P < 0.05* Homozygous (16) 50.6 ± 12.8 41.4 ± 12.6 38.5 ± 13.8 FIG. 5. Agarose gel electrophoresis of total plasma. The gel was P < 0.0005*t P < O.OOO5*t P < 0.025* stained with fat red 7B. The samples analyzed are from mice that are All values are expressed ±SD. Concentrations were compared wild type (+/+), heterozygous (+/-), or homozygous (-/-) with between genotypes by Student's t test. *, Values differ from wild respect to the targeted Apob locus, from a normal human, and from type; t, values differ from heterozygous. a human who is an apoB37/apoB86 compound heterozygote (9, 29). Mouse plasma was obtained after a 20-hr fast; the human plasma was tabolism of HDL (and apoAI) may be responsible for the obtained after a 14-hr fast. The locations of a-lipoproteins, pre-,- lipoproteins, 3-lipoproteins, and chylomicrons (C) are indicated to reductions in HDL cholesterol and apoAI. In human HBL the left and to the right for mouse plasma and human plasma, homozygotes, HDL concentrations have been variable, rang- respectively. ing from low to elevated (7-10). The concentration oftriglyceride in plasma is also reduced The f3lipoproteins are also reduced in heterozygous mice, in a dominant manner by the modified Apob allele (Table 1). but less markedly. A reduction in p-lipoproteins is a defining A reduction in plasma triglycerides in human heterozygotes characteristic of HBL in humans (Fig. 5). has been observed in many studies (27, 30, 31). Agarose gel electrophoresis of the plasma also reveals that Exencephalus and Hydrocephalus Are Associated with the fasted homozygous mice have chylomicronemia, as evidenced Modified Apob Allele. During our analysis ofthe genotypes of by accumulation of stained lipids at the origin of the gel (Fig. 21-day-old mice from heterozygote crosses, we noted signif- 5). This finding is not observed in heterozygous or wild-type icantly fewer homozygotes than expected based on the 1:2:1 mice. Steinberg and co-workers (9) have reported fasting Mendelian ratios of wild type:heterozygotes:homozygotes. chylomicronemia in a human with HBL; and plasma from We observed 108 wild-type mice, 239 heterozygotes, and Steinberg's human subject, a compound heterozygote with only 58 homozygotes (X2 = 25.5, P < 0.0001). To investigate one mutant allele yielding apoB37 and another mutant allele the reason for the deficiency of homozygotes, we examined yielding apoB86 (29), is compared with the mouse plasma in 47 fetuses (day 15.5-19.5 of gestation) derived from mating Fig. 5 to illustrate this. The cause of the fasting chylomicro- homozygous males with heterozygous females. Thirteen nemia in the homozygous mice or in the human compound were exencephalic; all proved homozygous for the modified heterozygote is not immediately obvious. Since neither the Apob allele. Thus, the deficiency of homozygotes at weaning targeted modification that we created nor the apoB86 mutation appears to be due to perinatal mortality associated with of Steinberg's patient is expected to interfere with the function exencephalus. of apoB48, it might be expected that chylomicrons would be Some mice that survive to weaning develop hydrocepha- packaged, transported, and metabolized normally. It is im- lus. This abnormality is characterized externally by a large, portant to note that the lipoproteins at the origin ofthe agarose dome-shaped head that is usually detected between 3 and 6 gel may not be "chylomicrons" but instead may be hepatic weeks of age. Gross dissection and histological analysis of lipoproteins or a mixture ofhepatic and intestinal lipoproteins. the brain of several of the hydrocephalic mice indicate that This topic warrants further investigation. fluid accumulation is confined to the lateral and third ven- The finding of chylomicronemia led us to examine, by tricles. Genotype analysis reveals that by 8 weeks of age, electron microscopy, the size distribution of lipoprotein par- hydrocephalus is detected in 32% (24/75) ofhomozygotes, in ticles with densities less than 1.006 g/ml. The fractions oflarge 1% (3/234) of heterozygotes, and in 0o (0/91) of wild-type particles (>70 nm) in wild-type, heterozygous, and homozy- littermates. Unaffected homozygotes usually survive to ma- gous mice were 0.8%, 0.6%, and 1.8%, respectively. The mean turity, appear healthy, and reproduce, but some of their (±SD) diameters of the d < 1.006 g/ml particles in wild-type, offspring can be exencephalic or hydrocephalic. heterozygous, and homozygous mice were virtually identical: Exencephalus and hydrocephalus have not been reported 41.34 ± 9.56, 38.98 ± 9.26, and 39.66 ± 9.64 nm, respectively. in humans with abetalipoproteinemia or HBL and, at this Thus, similar to human apoB37/apoB86 compound heterozy- stage, we have no direct evidence that abnormalities in lipid gotes (9), homozygous mutant mice also have a normal VLDL metabolism are the cause of the morphological abnormalities diameter but a skewing toward larger particles. in our mice. For example, since only one targeted ES cell line The modified Apob allele affects circulating concentrations was used to generate animals, we cannot exclude the possi- of cholesterol in a co-dominant manner (Table 1). The bility that a mutation was introduced in an unidentified gene concentrations of total and HDL cholesterol in the plasma of that is linked to the Apob locus. It is also conceivable that fasted homozygous mice are reduced to =50% of the normal overexpression of the HPRT genes introduced into the Apob levels. The concentrations of total and HDL cholesterol in locus affects purine metabolism during development. It is heterozygous mice are also significantly less than that in also possible that insertion of 35 kb of DNA into the Apob wild-type mice. The reduction of HDL cholesterol levels in locus has affected the function of nearby genes involved in homozygous mice is also reflected in a reduction in apoAI development of the brain. However, the most plausible levels of -45% as determined by nonreducing gel electro- explanation ofthese morphological abnormalities is that they phoresis of whole plasma (data not shown). Since apoAl is are caused by deficiency of one or more fat-soluble vitamins the major protein component of HDL, the reduced levels of that is secondary to the lack of LDL. For example, vitamin apoAI could possibly account for the reduced HDL choles- E is a lipid-soluble vitamin that is transported in the body as terol levels. However, we noted that the amount of apoAI part of LDL. Humans with abetalipoproteinemia and human mRNA is not reduced in homozygotes and may even be HBL homozygotes often manifest clinical symptoms of vi- increased in small intestine (Fig. 4D). Thus, increased ca- tamin E deficiency (10). This deficiency leads to neuromus- Downloaded by guest on September 30, 2021 Medical Sciences: Homanics et al. Proc. Natl. Acad. Sci. USA 90 (1993) 2393

cular and retinal degeneration and can be alleviated by 3. Blackhart, B. D., Ludwig, E. M., Pierotti, R., Caiati, L., massive oral supplementation with vitamin E. In laboratory Onasch, M. A., Wallis, S. C., Powell, L., Pease, R., Knott, animals, maternal vitamin E deficiency has been shown to T. J., Chu, M.-L., Mahley, R. W., Scott, J., McCarthy, B. J. cause exencephalus and hydrocephalus in offspring (32, 33), & Levy-Wilson, B. (1986) J. Biol. Chem. 261, 15364-15367. 4. Yang, C.-Y., Chen, S.-H., Gianturco, S. H., Bradley, W. A., and we have preliminary data indicating that our adult Sparrow, J. T., Tanimura, M., Li, W.-H., Sparrow, D. A., homozygous mice have plasma levels of a-tocopherol ap- DeLoof, H., Rosseneu, M., Lee, F.-S., Gu, Z.-W., Gotto, proximately 35% of those observed in wild-type mice A. M., Jr., & Chan, L. (1986) Nature (London) 323, 738-742. (G.E.H. and M. Traber, unpublished observations). The 5. Chen, S.-H., Habib, G., Yang, C.-Y., Gu, Z. W., Lee, B. R., effects of the Apob gene modification on the metabolism of Weng, S.-A., Silberman, S. R., Cai, S.-J., Deslypere, J. P., fat-soluble vitamins in these mice bear further investigation. Rosseneu, M., Gotto, A. M., Jr., Li, W.-H. & Chan, L. (1987) Conclusion. We have used gene targeting in ES cells to Science 238, 363-366. create a mouse model of familial HBL. The HBL mice 6. Powell, L. M., Wallis, S. C., Pease, R. J., Edwards, Y. H., manifest the Knott, T. J. & Scott, J. (1987) Cell 50, 831-840. hallmarks that define this condition in humans. 7. Young, S. G. (1990) Circulation 82, 1574-1593. They have very low concentrations of cholesterol and the 8. Farese, R. V., Jr., Linton, M. F. & Young, S. G. (1992) J. j-lipoproteins, and they synthesize a truncated form of apoB. Intern. Med. 231, 643-652. All apoB proteins are present in very low levels in the plasma. 9. Steinberg, D., Grundy, S. M., Mok, H. Y. I., Turner, J. D., The mice also exhibit several characteristics observed in some Weinsterin, D. B., Brown, W. V. & Albers, J. J. (1979) J. Clin. HBL humans: reduced HDL cholesterol levels, reduced Invest. 64, 292-301. plasma triglyceride concentrations, fasting chylomicronemia, 10. Kane, J. P. & Havel, R. J. (1989) in The Metabolic Basis of and symptoms suggesting vitamin E deficiency. Therefore, Inherited Disease, eds. Scriver, C. R., Beaudet, A. L., Sly, these mice represent a useful model for investigating how W. S. & Valle, D. (McGraw-Hill, New York), pp. 1139-1164. 11. Smith, T. J., Hautamaa, D. & Maeda, N. (1990) Gene 87, truncated apoB proteins alter lipoprotein metabolism. 309-310. HBL mice will be a useful model for investigating the 12. Reid, L. H., Gregg, R. G., Smithies, 0. & Koller, B. H. (1990) effects of this condition on the development of atherogenesis. Proc. Natl. Acad. Sci. USA 87, 4299-4303. The lipoprotein profile in these mice is expected to be 13. Hooper, M., Hardy, K., Handyside, A., Hunter, S. & Monk, protective from fatty streak formation and atherosclerosis. M. (1987) Nature (London) 326, 292-295. This mouse model will permit us to test this expectation by 14. Doetschman, T., Gregg, R. G., Maeda, N., Hooper, M. L., using a genetic approach. Mice with HBL can be mated with Melton, D. W., Thompson, S. & Smithies, 0. (1987) Nature mice carrying mutations in additional genes involved in (London) 330, 576-578. lipoprotein metabolism. For example, mice homozygous for 15. Piedrahita, J. A., Zhang, S. H., Hagaman, J. R., Oliver, P. M. & Maeda, N. (1992) Proc. Natl. Acad. Sci. USA 89,4471-4475. a disrupted Apoe allele are hypercholesterolemic and spon- 16. Kim, H.-S. & Smithies, 0. (1988) Nucleic Acids Res. 16, taneously develop atherosclerotic lesions (18, 34). It will be 8887-8903. of great interest to evaluate whether the dominant effect of 17. Young, S. G., Bertica, S. J., Curtiss, L. K. & Witztum, J. L. the Apob gene modification will overcome the effect of the (1987) J. Clin. Invest. 79, 1831-1841. apoE deficiency in mice with combined defects and protect 18. Zhang, S. H., Reddick, R. L., Piedrahita, J. A. & Maeda, N. them from hyperlipidemia and atherosclerosis. Information (1992) Science 258, 468-471. gained from studying this animal model of HBL could even- 19. Forte, T. M. & Nordhausen, R. W. (1986) Methods Enzymol. tually lead to therapies (drug and/or gene) to inhibit the 128, 442-457. atherosclerosis disease process. 20. France, D. S., Hughes, T. E., Miserendino, R., Spirito, J. A., Babiak, J., Eskesen, J. B., Tapparelli, C. & Paterniti, J. R., Jr. We also observed that the modified Apob allele is strongly (1989) J. Lipid Res. 30, 1997-2004. associated with the morphological abnormalities exenceph- 21. Homanics, G. E. (1991) Dev. Genet. 12, 371-379. alus and hydrocephalus. This may be the first example of a 22. Williamson, R., Lee, D., Hagaman, J. & Maeda, N. (1992) known genetic alteration that causes, probably indirectly, the Proc. Natl. Acad. Sci. USA 89, 7134-7138. congenital abnormalities exencephalus and hydrocephalus. 23. Thomas, K. R. & Capecchi, M. R. (1990) Nature (London) 346, While all forms of these abnormalities are unlikely to be due 847-850. to changes in the Apob gene itself, a portion of them could be 24. Hasty, P., Rivera-Perez, J., Chang, C. & Bradley, A. (1991) due to disturbances in lipid metabolism. The regular devel- Mol. Cell. Biol. 11, 4509-4517. in these mice 25. Moens, C. B., Auerbach, A. B., Conlon, R. A., Joyner, A. L. opment of exencephalus and hydrocephalus & Rossant, J. (1992) Genes Dev. 6, 691-704. provides a model for investigating the development and the 26. Dorin, J. R., Dickinson, P., Alton, E. W. F. W., Smith, S. N., pathogenesis of a disorder that affects a substantial number Geddes, D. M., Stevenson, B. J., Kimber, W. L., Fleming, S., of human newboms. Clarke, A. R., Hooper, M. L., Anderson, L., Beddington, R. S. P. & Porteous, D. J. (1992) Nature (London) 359, 211- We acknowledge John Hagaman, Paula Oliver, and Jeffrey Hodgin 215. for their expert technical assistance, Mike Sandlin for animal care, S. V. R. & S. G. Dr. Kinuko Suzuki for analysis of hydrocephalic mice, Dr. Maret 27. Welty, F. K., Hubl, T., Pierotti, Young, of (1991) J. Clin. Invest. 87, 1748-1754. Traber (New York University Medical Center) for analysis 28. Farese, R. V., Jr., Garg, A., Pierotti, V. R., Vega, G. L. & a-tocopherol concentrations, Dr. D. Sanan for electron microscopy, S. G. J. Res. 569-577. and Drs. Sarah Bronson and Suzanne Kirby for critical reading ofthe Young, (1992) Lipid 33, enthu- 29. Linton, M. F., Pierotti, V. & Young, S. G. (1992) Proc. Natl. manuscript. We especially thank Dr. Oliver Smithies for his Acad. Sci. USA 89, 11431-11435. siastic support and encouragement. This work was supported by 30. Young, S. G., Hunt, S. T., Chappell, D. A., Smith, R. S., National Institutes of Health Grants HL42630 (to N.M.), GM20069 Clairborne, F., Snyder, S. M. & Terdiman, J. F. (1989) N. (to Oliver Smithies), and HL08639 (to G.E.H.) and by an American Med. 1604-1610. T.J.S. was Engl. J. 320, Heart Association California Affiliate Grant (to S.G.Y.). 31. Young, S. G., Hubl, S. T., Smith, R. S., Snyder, S. M. & supported by an EMBO long-term fellowship. Terdiman, J. F. (1990) J. Clin. Invest. 85, 933-942. 1. Sniderman, A., Shapiro, S., Marpole, D., Skinner, B., Teng, B. 32. Cheng, D. W., Chang, L. F. & Bairnson, T. A. (1957) Anat. & Kwiterovich, P. O., Jr. (1980) Proc. Natl. Acad. Sci. USA Rec. 129, 167-168. 77, 604-608. 33. Verma, K. & King, D. W. (1967) Acta Anat. 67, 623-635. 2. Young, S. G., Bertics, S. J., Scott, T. M., Dubois, B. W., 34. Plump, A. S., Smith, J. D., Hayek, T., Aalto-Setala, K., Curtiss, L. K. & Witztum, J. L. (1986) J. Biol. Chem. 261, Walsh, A., Verstuyft, J. G., Rubin, E. M. & Breslow, J. L. 2995-2998. (1992) Cell 71, 343-353. Downloaded by guest on September 30, 2021