Proc. Natl. Acad. Sci. USA Vol. 92, pp. 8453-8457, August 1995 Medical Sciences

Normal plasma and fertility in gene-targeted mice homozygous for a disruption in the gene encoding very low density receptor (low density lipoprotein receptor gene family/triacylglycerol metabolism/body weight//homologous recombination) PHILIP K. FRYKMAN*, MICHAEL S. BROWN*, ToKuo YAMAMOTOt, JOSEPH L. GOLDSTEIN*, AND JOACHIM HERZ* *Department of Molecular Genetics, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75235-9046; and tTohoku University Gene Research Center, Sendai 981, Japan Contributed by Joseph L. Goldstein, June 1, 1995

ABSTRACT The very low density lipoprotein (VLDL) homologous recombination in embryonic stem (ES) cells (8, 9) receptor is a recently cloned member of the low density to produce mice that lack VLDLR. Males and females ho- lipoprotein (LDL) receptor family that mediates the binding mozygous for the disrupted VLDLR allele were viable and and uptake of VLDL when overexpressed in animal cells. Its fertile, producing litters of normal size. The lipoprotein pro- sequence is 94% identical in humans and rabbits and 84% files of the VLDLR knockout mice showed no difference from identical in humans and chickens, implying a conserved wild-type littermates. Extensive studies failed to reveal any function. Its high level expression in muscle and adipose tissue significant phenotype in homozygous knockout mice, with the suggests a role in VLDL triacylglycerol delivery. Mutations in sole exception that the animals were somewhat smaller and the chicken homologue cause female sterility, owing to im- leaner than their wild-type littermates. paired VLDL and vitellogenin uptake during egg yolk forma- tion. We used homologous recombination in mouse embryonic stem cells to produce homozygous knockout mice that lack METHODS immunodetectable VLDL receptors. Homozygous mice ofboth Assays and Materials. Lipoprotein analysis was performed sexes were viable and normally fertile. Plasma levels of cho- by FPLC on a Superose 6 column (10). and lesterol, triacylglycerol, and lipoproteins were normal when triacylglycerols were determined with assay kits from Boeh- the mice were fed normal, high-carbohydrate, or high-fat diets. ringer Mannheim and Sigma, respectively. , glucose, and The sole abnormality detected was a modest decrease in body free fatty acids were measured in the laboratory of Dr. J. D. weight, body mass index, and adipose tissue mass as deter- McGarry of this institution by use of a Linco rat insulin RIA mined by the weights ofepididymal fat pads. We conclude that kit (no. RI-13K; Linco Research, St. Louis), Beckman Glucose the VLDL receptor is not required for VLDL clearance from Analyzer II, and free Half Micro test kit (Boehringer plasma or for ovulation in mice. Mannheim), respectively. A peptide corresponding to the carboxyl-terminal 10 amino acids of the mouse VLDLR (Ser- The very low density lipoprotein (VLDL) receptor (VLDLR) Val-Val-Ser-Thr-Asp-Asp-Asp-Leu-Ala) was coupled to key- is the member of the low density lipoprotein (LDL) receptor hole limpet hemocyanin (5) and used to immunize rabbits by (LDLR) gene family that most closely resembles the LDLR (1, subcutaneous injection with Freund's adjuvant. A rabbit poly- 2). The VLDLR received its name because its overexpression clonal antibody directed against amino acids 1-724 of the enhances the binding and uptake of apoE-containing lipopro- human VLDLR was kindly provided by Kittie Wyne and Helen teins (VLDL and ,B-VLDL), but not apoB-100-containing Hobbs of this Department. lipoproteins (LDL), in cultured cells (1, 3, 4). A tissue survey Diets. The normal mouse diet (Teklad 6% mouse/rat diet showed widespread expression of VLDLR mRNA, with high- 7002 from Harlan Teklad Premier Laboratory Diets, Mad- est levels in muscle and adipose tissue and only trace amounts ison, WI) contained 6% fat, 24% protein, and 5% fiber by in the (2, 5). These data suggested that the VLDLR might weight. The high-carbohydrate diet (modified no. 960236; be involved in the transport of triacylglycerol-rich VLDL to its ICN) contained 68% sucrose, 18% casein, and 8% cotton main sites of metabolism, muscle and adipose tissue (1, 5). seed oil by weight. The cholesterol/atherogenic diet (mod- Recently, the chicken homologue of the VLDLR was cloned ified no. 900865; ICN) contained 40% butter, 20% casein, and found to be identical to the oocyte vitellogenin receptor 20% sucrose, 5.3% cholesterol, and 2% sodium cholate by (6, 7). The chicken vitellogenin/VLDL receptor has 84% weight. Animals were weighed on a Sartorius LC 4800 POAC amino acid identity to the human VLDLR, and it transports balance. both VLDL and vitellogenin into the yolk of developing cDNA Cloning. A mouse heart cDNA library in AZapII oocytes. In chickens the VLDLR gene is located on the X (Stratagene) was screened with a 2.6-kb Hindlll fragment of . In birds females are the heterogametic sex and the rabbit VLDLR cDNA (1). The resulting 3-kb mouse cDNA have only one X-equivalent chromosome. Hens hemizygous encoded a sequence that extended from exon 2 to the polya- for a mutated vitellogenin receptor are infertile and hyperlip- denylylation site. The encoded sequence showed 94% identity idemic secondary to the lack of lipoprotein transport into with the human amino acid sequence. oocytes, whereas heterozygous roosters are phenotypically Construction ofTargeting Vector. A 12-kb fragment encod- normal (7). ing introns 3-17 of the mouse VLDLR gene was enriched by In humans the VLDLR is encoded on (2). Its sucrose density ultracentrifugation of Bgl II-digested 129Sv function in the body is unclear. To study the physiological mouse genomic DNA, cloned into ADashII (Stratagene), and function of the VLDLR in mammals, we used the technique of isolated by plaque screening using the 2.6-kb Hindlll fragment

The publication costs of this article were defrayed in part by page charge Abbreviations: BMI, body mass index; ES cells, embryonic stem cells; payment. This article must therefore be hereby marked "advertisement" in LDL, low density lipoprotein; LDLR, LDL receptor; VLDL, very low accordance with 18 U.S.C. §1734 solely to indicate this fact. density lipoprotein; VLDLR, VLDL receptor.

8453 Downloaded by guest on September 25, 2021 8454 Medical Sciences: Frykman et al. Proc. Natl. Acad. Sci. USA 92 (1995) of the rabbit VLDLR cDNA. A replacement-tylpe targeting carried the disrupted VLDLR allele through the germline. All vector (11) was constructed (Fig. 1). The short arm of the experiments were performed with the F2 or F3 generation vector (1 kb) was amplified by PCR using Pfu polymerase descendants, which were hybrids between C57BL/6J and (Stratagene) from the Pst I site in intron 4 to the cysteine-161 129Sv mice. codon in exon 5 of the mouse VLDLR gene. TIhe long arm Blot Hybridization of Genomic DNA. Tail DNA was pre- extending from the third nucleotide position of the proline-262 pared by proteinase K digestion, phenol/chloroform extrac- codon in exon 5 to the Bgl II site (8.8 kb) was pr-epared by a tion, and ethanol precipitation (14). For Southern blotting, 10 combination of PCR cloning and direct subcloniing of a 7-kb ,ug of genomic DNA was loaded onto each lane after digestion Kpn I-Bgl II fragment into pGEM-3Zf(+) (Promelga), deleting with EcoRI. After hybridization with a randomly 32P-labeled residues 162 to 262 of exon 5. The Pol2sneobpAI expression probe (1-2 x 106 cpm/ml; see legend to Fig. 1) at 65°C in cassette (12) was inserted into the deletion created in exon 5, Rapid-hyb buffer (Amersham) for 3-6 hr, the blot was washed and two copies of the herpes simplex virus thymiidine kinase for 15 min in 2x SSC/0.5% SDS (15) at room temperature and gene (12) were inserted in tandem at the 5' end of the short then for 30 min in 0.2x SSC/0.1% SDS at 65°C. arm of the targeting vector. ES Cell Culture. Mouse ES cells (JH-1) were cultured on RESULTS leukemia inhibitory factor-producing STO feeder cells (13). Approximately 2 x 107 cells were electroporated (275 V, 330 To disrupt the VLDLR gene in murine ES cells, we con- ,F; GIBCO/BRL electroporator) with a Sal I-lin earized tar- structed a targeting vector of the replacement type (11) with geting vector (25 ,ug/ml) and seeded onto irradiiated feeder a neo cassette inserted into exon 5 of the VLDLR gene, layers. After selection with G418 (190 jig/ml) and 1-(2-deoxy- disrupting the reading frame (Fig. 1). The linearized targeting 2-fluoro-f3-D-arabinofuranosyl)-5-iodouracil (FIA,U, 0.25 ,uM; construct was electroporated into ES cells, which were then Bristol-Myers Squibb), recombinant clones were iidentified by placed under positive and negative selection (13). Homologous PCR (13) using primers 3 and 4 (primer 3, locatted in the 3' recombinants were detected by PCR and confirmed by South- untranslated region of the neo cassette, 5'-CCT(CGTGCTT- ern analysis ofEcoRI-digested ES cell genomic DNA with the TACGGTATCGCCGCTC-3'; primer 4, located in intron 4, probe shown in Fig. 1. Recombinant ES cell clones were 5 '-CCCTGGAGAAAATCTGCGGGTTAAATA-3'). Ho- expanded and microinjected into C57BL/6J blastocysts by mologous recombination was verified by Southern analysis standard techniques (12). Several male chimeras derived from after EcoRI digestion and probing with a 0.8-kb genomic independently targeted ES cell clones transmitted the dis- fragment 5' of the targeting construct (Fig. 1). Eigght indepen- rupted gene through the germline as determined by Southern dent ES cell clones containing a disrupted VLDLI . allele were blotting (data not shown). injected into C57BL/6J blastocysts, yielding a total of 34 A total of 26 matings between VLDLR +/- mice yielded chimeric males whose coat color (agouti) indicatted a contri- 200 offspring, comprising 50 +/+, 103 +/-, and 47 -/- for bution of ES cells ranging from 20% to 95%. Of the 34 the VLDLR gene, which was very close to the 1:2:1 ratio chimeric males, 27 were fertile, and 8 gave offspring that expected for random Mendelian segregation. Ten litters from homozygous matings between VLDLR(+/+) x VLDLR- (+/+) and VLDLR(-/-) x VLDLR(-/-) mice yielded 9 p K E E ,, average litter sizes of 7.9 and 8.2 pups, respectively, showing that disruption of VLDLR function does not affect fertility in W l -type Allele 1/79 mice. 5 17 Immunoblot analysis was used to confirm that the VLDLR -/- mice lacked immunodetectable VLDLR. Proteins ex- X x tracted from heart muscle membranes of VLDLR + / +, + / -, E K E E and -/- mice were subjected to SDS/PAGE, transferred to TGerting Vector I HWne | IvIn nitrocellulose filters, and probed with a rabbit polyclonal 5 5 17 antibody directed against the carboxyl-terminus of the mouse VLDLR (Fig. 2A). In the heart the VLDLR appears as two E a P E K E 1-4 bands: a precursor and a fully processed mature form (Fig. 2A, Disrupted Al"leleI I1 E lane 1). Heterozygotes (+/-, lane 2) produced reduced 17 amounts of protein, and homozygotes (-/-) produced no EcoRI Dkits: detectable protein (lane 3) even when a 2-fold excess of E E membranes was used (lane 4). Fig. 2B is a Coomassie blue- Disrupted Allele stained gel showing that comparable amounts of protein were loaded in lanes 1-3. We performed another immunoblot locus in the experiment like the one in Fig. 2, except that a rabbit poly- FIG. 1. Strategy for targeted disruption of the VLDL,R Rlocu m clonal antibody directed against the extracellular domain mouse genome. A targeting vector of the replacement Itype (11) was (amino acids 1-724) of the human VLDLR was used. With this constructed as described in Methods. The neomycin-resistance (ned) antibody we were also unable to detect any truncated protein gene is driven by the murine RNA polymerase II p followed by the 3' untranslated region of the bovine growth hormone product (data not shown). gene containing the polyadenylylation signal. The tiranscriptional To screen for effects on lipoprotein metabolism, VLDLR direction of the neo gene is antiparallel to that of the \(LDLR gene. +/+ and -/- mice were fed various diets, and the plasma Two copies of the herpes simplex virus thymidine kinaste gene (HSV- lipoprotein profiles were analyzed by FPLC. Mice were fed TK) flank the 5' homology segment. In the event off homologous either a normal diet (Fig. 3A), a 68% sucrose diet that increases recombination, the disrupted allele will have acquired an additional hepatic synthesis of VLDL (Fig. 3B), or a diet high in choles- restriction site for EcoRI. The expected EcoRI digestion pattern terol and saturated fat that increases chylomicrons (Fig. 3C). resulting from a targeting event is shown at the bottoi f.The DNA In no case did we observe a detectable difference between the probe used for Southern blotting is a 0.8-kb PCR produ4 portion of exon 4 to the Pst I site in intron 4. The positiccts5 of the two plasma lipoprotein cholesterol profiles of the VLDLR +/+ oligonucleotide primers used for PCR diagnosis of honmologous re- and -/- mice. We also produced double-knockout mice that combination are indicated by the arrowheads (primer 1, 3' end of neo were homozygous for LDLR deficiency (17) as well as VLDLR cassette; primer 2, upstream of Pst I site in intron 4). B, Bgl II; E, deficiency. These animals had the same plasma cholesterol EcoRI; K, Kpn I; P, Pst I. profile as the LDLR -/- single-knockout mice (Fig. 4). Downloaded by guest on September 25, 2021 Medical Sciences: Frykman et al. Proc. Natl. Acad. Sci. USA 92 (1995) 8455

i kDa A111213|4i BF11213 ~~~~kDa 15 - l O LDLR -4- |HDL - 200 oc | LDLR A4; VLDLR 1-|- .0 °. 10 LDL Mature- -. o 0 *0 Precursor---w <,>2 5 -95 VLDL *- -69.. 0 10 20 30 Fraction Number 46 Genotype ['-tbI71-/-I-fI +E6+/ I e-; FIG. 4. FPLC profiles of mouse lipoproteins from LDLR -/- single-knockout and LDLR -/-, VLDLR -/- double-knockout FIG. 2. Immunoblot analysis of VLDLR in heart muscle mem- mice. Mice (3 months of age) with the indicated genotype (three males branes (100,000 x g fraction) from mice that were wild-type (+/+, per group) were fed a normal diet. Plasma was processed for lipopro- lane 1), heterozygous (+ / -, lane 2), or homozygous (- / -, lanes 3 and tein profiles as described in Fig. 3. Each profile shown represents mean 4) for the disrupted VLDLR allele. Membranes were solubilized with values of individual profiles from 3 mice. 1x Laemmli buffer (16) as described (5). An aliquot of each sample (150 ,tg of protein in lanes 1-3 and 300 ,tg in lane 4) was subjected to (BMI, weight divided by length squared) was measured before SDS/6.5% PAGE under reducing conditions. (A) The proteins were and after the animals were on a 68% sucrose which transferred to nitrocellulose filters for immunoblot analysis (5). The placed diet, filter was incubated with rabbit polyclonal antibody (2 ,tg/ml, IgG is reported to produce mild obesity (18) (Fig. 6). BMI is fraction) directed against a synthetic peptide corresponding to the last frequently used as an indicator of adiposity in humans (19) and 10 amino acids of the mouse VLDLR, after which bound IgG was has been applied successfully to distinguish between lean and detected by an immunoperoxidase procedure using the Amersham obese mice (20). At the beginning of the study the BMIs of the ECL kit (5). The positions of migration of the mature and precursor 7- to 8-week-old VLDLR + / + and - /- mice were similar. On forms of the receptor are indicated. (B) The proteins of the nitrocel- a chow diet, the BMIs increased more extensively in the +/+ lulose filter were stained with Coomassie blue. than in the -/- mice (Fig. 6A). This difference was accen- During the course of these experiments, we noticed that the tuated on the 68% sucrose diet (Fig. 6B). VLDLR -/- animals appeared smaller than their wild-type To determine whether the difference in BMI was reflected or heterozygous littermates. We compared the weights of in adipose tissue mass, we dissected and weighed epididymal VLDLR +/+ and -/- animals as a function of age from fat pads from VLDLR +/+ and -/- male mice that were fed postnatal day 5 to day 50. This revealed a persistent 15-20% the normal or 68% sucrose diet for 3.5 months (Table 1). As reduction in weight of the -/- mice (Fig. 5). The VLDLR a control we also dissected and weighed the kidneys from the -/- mice were also shorter than their wild-type or heterozy- same animals. On a normal chow diet the epididymal fat pads gous littermates. from the normal mice were twice as heavy as those from the To determine whether there was a difference in the adipos- -/- mice. On the 68% sucrose diet, the epididymal fat pads ity of the VLDLR +/ + and - - mice, the body mass index from both groups increased by about 2-fold, the fat pads from the +/+ mice remaining about twice as heavy as those from 10 A Normal HDL the -/- mice. These differences persisted even when we

5 A Males VLDL LDL C 25- 15_ B 68% Sucrose HDL !O5- _ 1 co' 0 Gentp

~~~~LDL o I/~~~ ZVLDL 0 _ . I I I B Females 20 C 5.3% Cholesterol HDL a 5 cO +1 21!O - 4- -&4H 15 .5P 5- a 1,0 VLDRR + (a CD

5- ~~~LDL VLDL 0 10 20 30 40 50 Age (days) ~~10 20 30O Fraction Number FIG. 5. Weight vs. age in wild-type (+/+) and VLDLR knockout (-/-) mice. Ten litters of each genotype resulting from (+/+) X FIG. 3. FPLC profiles of mouse lipoproteins from wild-type (+/+) (+/+) matings (0) or (-/-) x (-/-) matings (0) were weighed and VLDLR knockout (- / -) mice. Mice with the indicated genotype every 5 days from day 5 to day 50 after birth. For days 5, 10, and 15, (five males per group) were fed for 2 weeks on a normal diet (A), a 68% the animals of each gender were weighed as a group, and the mean sucrose diet (B), or a cholesterol/atherogenic diet (C). Pooled plasma weight was recorded; thereafter all animals were weighed individually. from each group of nonfasted animals (0.1 ml) was subjected to FPLC All animals were weaned at day 20 and placed on normal chow. Each gel filtration, and the cholesterol content of each fraction was mea- value represents the mean SD. The number of animals at each age sured. The mice were 6-9 weeks of age. HDL, high density lipoprotein. was 29-34 for males and 42-48 for females. Downloaded by guest on September 25, 2021 8456 Medical Sciences: Frykman et al. Proc. Natl. Acad. Sci. USA 92 (1995) of endurance and histology. In all studies VLDLR +/+ and 0.45 -A Normal B 68% Sucrose -/- mice were essentially indistinguishable. X O VLDLR+/+ la 0.40 -* VLDL.R /-4 T DISCUSSION E 0.35 The current results suggest that the VLDLR of the mouse is ;O%.0.30 not required to maintain normal levels of plasma triacylglyc- erols or cholesterol in the fed or fasted state, nor is it required to 0.25 p=.012 .007 .001 .013 for ovulation, nor for any apparent aspect of gestation. These results are quite surprising in light of previous information 00 Ii I I I ~1I I I I Ii 0 0 20 40 60 80 0 20 40 60 80 regarding the VLDLR. Day Day The high degree of conservation of the amino acid sequence of the VLDLR among animal species (94% identity between FIG. 6. BMI of wild-type (+/+) and VLDLR knockout (-/-) human and rabbit; 84% between human and chicken) implies mice on a normal (A) or 68% sucrose (B) diet. Mice (7-8 weeks of age) that a mutation that inactivates the VLDLR would be dele- with the indicated genotype (8-10 males per group) were fed the terious to biologic fitness. In transfected cells the receptor indicated diet for 84 days during which time each mouse had its weight mediates the binding and uptake of VLDL to a much greater and length (nose to anus) measured at the indicated time for calcu- degree than it mediates the binding and uptake of LDL (1, 3). lation of BMI (g/cm2). Statistically significant differences were de- In vivo the receptor is present in tissues with high rates of termined with Student's t test (two-tailed, unpaired). VLDL-triacylglycerol uptake-namely, adipose tissue and muscle. It therefore seemed likely masses -/- mice muscle, including cardiac corrected for the different lean body of the that the VLDLR acted, perhaps in concert with lipoprotein as estimated from the kidney weights (a/b in Table 1). to mediate of VLDL triacylglycerols in these The daily food consumption of 10 male mice of each lipase, uptake tissues. If the VLDLR does play this role, it must make a genotype was determined by weighing the uningested food quantitatively minor contribution to overall triacylglycerol daily over 31 days. The mean daily intake for VLDLR +/+ clearance from the bloodstream in mice. Otherwise, its inac- and -/- mice was 2.99 and 2.96 g/day, respectively. tivation would lead to an elevation in triacylglycerol levels. In addition to the studies detailed here, we compared many Moreover, VLDLR-mediated triacylglycerol delivery does not -I- other parameters in the VLDLR +/+ and mice. None appear to be required for the normal functioning of skeletal or of these parameters showed a consistent difference between cardiac muscle or for the gross development of adipose tissue, the genotypes. These measurements included the following although it may have some effect on adipose tissue mass. constituents of plasma: cholesterol, triacylglycerol, glucose, The lack of a requirement for the VLDLR in VLDL insulin, and free fatty acids (all in fed and fasted states); clearance is supported by previous experiments with the electrolytes; liver-related enzymes (aspartate and alanine ami- receptor-associated protein (RAP) (4, 10). This protein, which notransferases, alkaline phosphatase); creatinine; total pro- was originally isolated in a complex with the LDLR-related tein; albumin; blood urea nitrogen; uric acid; total bilirubin; protein (LRP) and subsequently shown to inhibit binding of lactate dehydrogenase; and thyroxine. A complete blood cell ligands to LRP, also potently inhibits the binding of VLDL to count with differential screen was performed along with the VLDLR in vivo (4). Mice that overexpress RAP as a result extensive histological examination of tissues using hematoxylin of adenoviral gene transfer to the liver exhibited only a minor and eosin or trichrome stains. In addition, NADPH oxidase increase in plasma levels of triacylglycerols despite the pres- and oil red 0 stains were used for muscle. Tissues analyzed ence of circulating levels of RAP that should have been were heart, (quadriceps and soleus), aorta, sufficient to block completely the binding of VLDL to the vena cava, lung, kidney, spleen, thymus, pancreas, liver, larynx, VLDLR (10). esophagus, stomach, jejunum, colon, brain, epididymal fat, The VLDLR -/- mice were somewhat smaller than their brown fat, thyroid, adrenal, pituitary, skin, bone, testes, epi- wild-type counterparts, both in length and in weight, when didymis, seminal vesicles, uterus, and Fallopian tubes. No studied on normal chow or high-sucrose diets. Their BMI was differences were observed between VLDLR '+/+ and -/- also reduced. Measurements of epididymal fat pads suggested mice. that the mean size of the adipocyte was reduced in the VLDLR During our attempts to define the phenotype of VLDLR -/- mice on a chow diet. Despite this reduction in adipose knockout mice, we challenged the +/+ and -/- mice with tissue mass, the adipocytes appeared to function in a relatively several physiological tests. In separate studies, mice were (a) normal manner. Thus, the adipose tissue mass increased when fasted for 72 hr with measurement of weight loss and blood the animals were placed on a high-sucrose diet (Table 1). glucose, (b) placed at 4°C for 24 hr with temperature mea- Moreover, the adipose tissue was able to supply sufficient surements to test their thermogenesis response, (c) exercised energy to maintain the animal in the same metabolic state as strenuously on a treadmill with measurement of endurance as wild-type animals during a 72-hr fast. well as muscle-derived enzymes in blood, or (d) treated with The significance of the reduced BMI is uncertain. The propylthiouracil or triiodothyronine (4) with subsequent tests difference might be due to either slightly reduced food intake, Table 1. Weight of epididymal fat pads in wild type (+/+) and VLDLR knockout (-/-) mice Organ weight, g Epididymal Genotype n Diet Body mass, g fat pads (a) Kidneys (b) a/b +/+ 8 Normal 31.2 ± 1.8 0.47 ± 0.1 0.47 + 0.03 1.0 3 68% Sucrose 35.1 ± 1.9 1.1 ± 0.1 0.42 ± 0.1 2.6 -/- 9 Normal 24.9 ± 1.1 0.22 ± 0.02 0.37 + 0.01 0.59 5 68% Sucrose 25.0 ± 1.2 0.43 ± 0.05 0.33 + 0.1 1.3 Male animals (5.5 months of age) were fed the indicated diet for 3.5 months prior to organ harvest. Both epididymal fat pads and both kidneys were dissected from each animal and weighed. Values are shown as mean + SEM. Downloaded by guest on September 25, 2021 Medical Sciences: Frykman et al. Proc. Natl. Acad. Sci. USA 92 (1995) 8457

slightly accelerated metabolism, or both. Although measure- (HL20948), the W. M. Keck Foundation, the Perot Family Founda- ments of food intake failed to document a difference between tion, and the Moss Heart Foundation. P.K.F. is supported by Medical J.H. is a Lucille P. Markey Scholar. the +/+ and -/- mice, the accuracy of this technique is Scientist Training Grant GM08014. to of food. We cannot exclude the limited, owing spillage T., Sakai, J. & mice ate less than the +/+ mice. We 1. Takahashi, S., Kawarabayasi, Y., Nakai, possibility that the -/- Yamamoto, T. (1992) Proc. Natl. Acad. Sci. USA 89, 9252-9256. have not yet performed studies of metabolic rate. 2. Sakai, J., Hoshino, A., Takahashi, S., Miura, Y., Ishii, H., Suzuki, Two possible explanations for the reduced adipose tissue H., Kawarabayasi, Y. & Yamamoto, T. (1994) J. Biol. Chem. 269, mass in VLDLR -/- mice must be ruled out: (i) an incom- 2173-2182. plete gene knockout and (ii) the existence of related receptors 3. Suzuki, J., Takahashi, S., Oida, K., Shimada, A., Kohno, M., that assume the function of the VLDLR when its gene is Tamai, T., Miyabo, S., Yamamoto, T. & Nakai, T. (1995) destroyed. The first possibility is excluded by immunoblot Biochem. Biophys. Res. Commun. 295, 835-842. analysis with antibodies recognizing different epitopes. We 4. Battey, F. D., Gafvels, M. E., FitzGerald, D. J., Argraves, W. S., found no evidence that the gene-knockout animals could Chappell, D. A., Strauss, J. F. & Strickland, D. K. (1994) J. Bio. produce a fragment of the VLDLR. Even if a truncated Chem. 269, 23268-23273. E. V., K. T., Wyne, K. L., Ho, Y. K., were produced, it would have lacked a large portion 5. Jokinen, Landschultz, fragment Frykman, P. K. & Hobbs, H. H. (1994) J. Biol. Chem. 269, of the putative ligand-binding domain as well as the mem- 26411-26418. brane-spanning region, which should have rendered it inactive. 6. Bujo, H., Hermann, M., Kaderli, M. O., Jacobsen, L., Sugawara, The possibility of a redundant receptor is more difficult to S., Nimpf, J., Yamamoto, T. & Schneider, W. J. (1994) EMBO J. address. The most likely candidate would be the LDLR. 13, 5165-5175. However, the LDLR -/-, VLDLR -/- double-knockout 7. Stifani, S., Barber, D. L., Nimpf, J. & Schneider, W. J. (1990) mice showed the same lipoprotein profile as the LDLR -/- Proc. Natl. Acad. Sci. USA 87, 1955-1959. single-knockout animals, thereby excluding the LDLR as a 8. Capecchi, M. R. (1989) Science 244, 1288-1292. backup receptor for the VLDLR. We cannot exclude the 9. Bronson, S. K. & Smithies, 0. (1994) J. Biol. Chem. 269, 27155- possibility that another member of the LDLR gene family 27158. performed this backup function in the VLDLR -/- mice. 10. Willnow, T. E., Sheng, Z., Ishibashi, S. & Herz, J. (1994) Science 264, 1471-1474. and the one that we consider most likely, A third possibility, 11. Mansour, S. L., Thomas, K. R. & Capecchi, M. R. (1988) Nature is that the function of the VLDLR may be to internalize (London) 336, 348-352. molecules other than the triacylglycerols of VLDL. In addition 12. Soriano, P., Montgomery, C., Geske, R. & Bradley, A. (1991) Cell to triacylglycerols, VLDL contains tocopherols, carotenoids, 64, 693-702. quinones, (3-carotene, and other lipid-soluble molecules (21). 13. Willnow, T. E. & Herz, J. (1994) Methods Cell Biol. 43, 305-334. It is possible that the VLDLR participates in a pathway that 14. Hogan, B., Costantini, F. & Lacy, E. (1986) Manipulating the extracts such molecules from VLDL without extracting tria- Mouse Embryo: A Laboratory Manual (Cold Spring Harbor Lab. cylglycerols and without destroying the entire particle. This Press, Plainview, NY). possibility may be tested by measuring blood and tissue levels 15. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular in VLDLR -/- and + / + mice of all the substances known to Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press, in VLDL. The major ligand for the VLDLR may not Plainview, NY), 2nd Ed. be carried U. K. (1970) Nature (London) 227, 680-685. of the LDLR gene 16. Laemmli, be a lipoprotein at all. Other members 17. Ishibashi, S., Brown, M. S., Goldstein, J. L., Gerard, R. D., family, such as LRP, bind a large variety of ligands, including Hammer, R. E. & Herz, J. (1993) J. Clin. Invest. 92, 883-893. protease/antiprotease complexes such as a2-macroglobulin 18. Sclafani, A. (1992) in Obesity, eds. Bjorntorp, P. & Brodoff, B. N. (22). This question might be answered by a more exhaustive (Lippincott, Philadelphia), pp. 241-248. analysis of the proteins that accumulate in plasma of VLDLR 19. Keys, A., Fidanza, F., Karvonen, M. J., Kimura, N. & Taylor, -/- mice. H. L. (1972) J. Chron. Dis. 25, 329-343. 20. Bahary, N., Leibel, R. L., Joseph, L. & Friedman, J. M. (1990) Scott Clark provided excellent technical assistance. We thank our Proc. Natl. Acad. Sci. USA 87, 8642-8646. colleagues Helen Hobbs, Shun Ishibashi, Denis McGarry, and Dennis 21. Bjornson, L. K., Kayden, H. J., Miller, E. & Moshell, A. N. (1976) Burns for helpful suggestions throughout the study. This work was J. Lipid Res. 17, 343-352. supported by research grants from the National Institutes of Health 22. Krieger, M. & Herz, J. (1994) Annu. Rev. Biochem. 63, 601-637. Downloaded by guest on September 25, 2021