INTRODUCTION

Apolipoprotein (apo)¹ B mRNA editing is a process by which apoB-100 mRNA is converted to apoB-48 mRNA (2, 3). It involves the deamination of the first base of the codon CAA, encoding glutamine 2153, converting it to UAA, a stop codon. ApoB-100 and apoB-48, the translation products of the unedited and edited apoB mRNAs, respectively, have quite different biochemical properties and physiological functions (reviewed in Ref. 4). ApoB-100 is required for the production of very low density lipoprotein (VLDL) and its metabolic products, intermediate density lipoprotein (IDL) and low density lipoprotein (LDL). It is also an essential structural component of lipoprotein(a). ApoB-48, on the other hand, is needed for chylomicron production in the small intestine.

The mechanism of apoB mRNA editing is an area of intense investigation (reviewed in Refs. 5, 6, 7). The recent cloning of Apobec-1, a cytidine deaminase-like enzyme that has the capacity to edit apoB mRNA under appropriate conditions in vitro (8, 9, 10), in cultured cells (11), and in transgenic animals in vivo (12), represents a major advance in this area. Apobec-1 exists as a spontaneous homodimer (11) and is expressed exclusively in the small intestine in humans (11, 13), but in numerous tissues, including liver and small intestine, in mice (14). Apobec-1 has apoB mRNA editing activity only in the presence of complementation factors (8).

ApoB mRNA editing plays an important role in the control of lipoprotein metabolism because it determines which species of apoB is synthesized in a particular tissue (6, 7). In most mammals, including humans and mice, the small intestine produces essentially only apoB-48. In contrast, while the human liver produces only apoB-100, mouse and rat liver produce both apoB-100 and apoB-48. Therefore, mouse and rat VLDL contain both apoB-100 and apoB-48. However, only apoB-100-containing VLDL serves as a precursor to LDL (15).

Mice have been used as a model for lipoprotein metabolism and atherosclerosis (reviewed in Refs. 16 and 17). They are genetically well defined, and many transgenic and knockout mouse lines are available. One drawback in the use of mice in studying lipoprotein metabolism is the presence of fairly high levels (60-70%) of apoB mRNA editing in the liver (14). It is one factor that limits the plasma concentration of apoB-100 in mice. It is unclear, however, whether and how elimination of apoB mRNA editing affects lipoprotein biogenesis in mice especially with respect to apoB-containing lipoproteins.

To address this issue, we have created apobec-1 knockout mice by gene targeting. As we were preparing this paper for publication, Hirano et al. (18) reported the production of apobec-1^/ mice and noted the absence of apoB mRNA editing in these animals. Our results corroborate theirs in showing that these animals do not edit apoB mRNA. Here we present the lipoprotein phenotypic effects of apobec-1 gene disruption that were not examined by Hirano et al. (18). We found that apobec-1^/ mice have substantially decreased high density lipoproteins (HDL) as determined by fast protein liquid chromatography (FPLC) as well as by ultracentrifugal flotation. Furthermore, they have marked elevation of their plasma apoB-100 but no change in their total plasma apoE, cholesterol, or triglyceride. Their LDL cholesterol was also apparently unchanged. We next examined the phenotype of apobec-1^/ animals cross-bred with apoB transgenic mice. In these animals, we found that the absence of apoB mRNA editing was associated with a marked increase in total plasma cholesterol and LDL cholesterol as well as in total plasma triglyceride and VLDL triglyceride, changes that would be expected for an elevated plasma apoB-100. Finally, we showed that the absence of apoB mRNA editing in these animals was not an indirect secondary effect of absence of functional Apobec-1, because editing was acutely restored by adenovirus-mediated transfer of apobec-1.

MATERIALS AND METHODS

Cloned Apobec-1 genomic fragments were isolated from a mouse strain 129 DNA library, and a restriction map and exon-intron organization were established as described previously (14). A replacement-type targeting vector was constructed (Fig. 1). The 5 arm of the vector consists of a 3.5-kb XbaI/HindIII fragment containing exon 6 flanked by a portion of introns 5 and 6. The 3 arm subclone is a 2.3-kb fragment spanning from a HindIII site in exon 7 to an XbaI fragment in exon 8. The 0.15-kb deletion containing 0.1 kb of intron 6 and the first 47 nucleotides (up to 490, or the second base of the codon for glutamine 164 of Apobec-1) of exon 7 was replaced by a PGKneobpA expression cassette (19) inserted in the opposite orientation. A pMC-tk-poly(A) cassette (20) was attached to both the 5 and 3 end of the targeting vector.

Mouse ES cell (J1) culture, transfection, and selection of mutant ES cells were performed as described previously (21). Briefly, J1 cells were grown on feeder layers of -irradiated embryonic fibroblast (EF) cells with leukemia inhibitory factor (500 units/ml) (Life Technologies, Inc.). 2 × 10⁷ cells at passage 10 were electroporated in the presence of 30 µg/ml linearized DNA at 400 V, 25 microfarads in a BTX 300 electroporator. After positive selection with G418 at 200 µg/ml and negative selection with FIAU (Bristol Myers) at 0.2 µM for 8-10 days, recombinant clones were identified by Southern blot analysis after HindIII digestion and probing sequentially with a 0.6-kb HindIII/BamHI intron 4 genomic fragment and then with a 0.5-kb XbaI/HindIII exon 8 fragment (Fig. 1). One of five independent ES cell clones was injected into C57BL/6J blastocysts as described previously (21), yielding three chimeric males with agouti coat color indicating essentially 100% contribution of ES cells. Chimeric males were bred with C57BL/6J females, and germline transmission of the mutant allele was detected by Southern blot analysis of tail DNA from offspring. All experiments were performed with animals from F3, F4, or F5 generations, obtained by cross-breeding with C57BL/6J mice. All animal experiments were conducted in accordance with the guidelines of the Animal Protocol Review Committee of Baylor College of Medicine. Human apoB transgenic mice (22) were a generous gift of Dr. E. Rubin of the University of California, Berkeley. The transgene consists of an 80-kb P1 genomic clone containing the normal human apoB gene.

Tail DNA was prepared by proteinase K digestion, phenol/chloroform extraction, and ethanol precipitation (21). Ten µg of purified DNA was digested with the indicated restriction endonuclease, fractionated by electrophoresis on 0.8% agarose gels, blotted with 0.4 M NaOH onto GeneScreen Plus (DuPont NEN), and hybridized with a randomly ³²P-labeled probe (Fig. 1, probe A or B) at 65 °C. Blots were washed with 0.5 × SSC and 0.1% SDS at 65 °C. The inheritance of the human apoB transgene in the cross-breeding experiments was also followed by Southern blotting of mouse tail DNA.

Total mouse RNA was isolated from the small intestine and the liver by Ultraspec RNA Isolation System (Biotecx). Ten µg of total RNA treated with RNase-free DNase (Promega) was used for primer extension assay as described (23). Briefly, ³²P-end-labeled mouse apoB primer (BBT9, 5- AGTCATGTGGATCATAATTATCTTTAATATACTGA) was annealed overnight at 45 °C with RNA. The annealed mixture was extended in the presence of 0.5 mM each dATP, dCTP, dTTP and 0.5 mM dideoxyGTP by the addition of 10 units of Superscript II (Life Technologies, Inc.). The extension products were resolved on 6% polyacrylamide-urea gels.

For qualitative analysis of plasma apoBs, mouse plasma was subjected either to air-driven ultracentrifugation (Airfuge^TM, Beckman) for 4 h, or to ultracentrifugal flotation as described below before the samples were fractionated on 4-15% SDS gel. For quantitative determination of plasma apoBs, apoB-containing lipoproteins were isolated by density gradient ultracentrifugation (Beckman Ti-50.1 rotor, 40,000 rpm, 10 °C, 18 h) after adjusting 500-µl plasma samples to a density of 1.063 g/ml with a KBr solution (d 1.365 g/ml) in tubes containing a KBr overlay solution (d 1.063 g/ml). The top 500-µl lipoprotein fractions were dialyzed against 150 mM NaCl, 1 mM EDTA, 1 mM NaN₃, 10 µM phenylmethylsulfonyl fluoride (pH 7.5) and enzymatically assayed for cholesterol. To quantify the apoB content, purified lipoproteins (1-10 µg of cholesterol) were solubilized in 2 × SDS sample loading buffer at 60 °C for 30 min. ApoB-48 and B-100 were resolved in a 4-15% polyacrylamide gel containing four dilutions of purified human LDL as internal standards. The gels were fixed in 40% methanol, 10% acetic acid and stained with 0.2% Coomassie Brilliant Blue R250. After destaining, the bands were quantified by densitometer scanning (OmniMedia Scanner) using SepraScan 2001^TM one-dimensional densitometry software (Integrated Separation System, Natick, MA). The chromogenicity was linear with respect to the range of lipoproteins loaded. Total plasma apoE was quantitated by a radial immunodiffusion assay (24).

All mice were maintained on a normal chow diet (Teklad 4% mouse/rat diet 7001, Harlan Teklad Premier Laboratory Diets) that contained 4% (w/v) animal fat and <0.04% (w/v) cholesterol. They were fasted 4-5 h before blood was removed by retro-orbital puncture under anesthesia. Lipoprotein fractions were isolated by gel filtration chromatography using a Beckman System Gold HPLC/FPLC with two Superose 6 columns (Pharmacia Biotech Inc.) connected in series (25). For each analysis, we applied 200 µl of mouse plasma onto the FPLC and 0.5-ml fractions were eluted with 1 mM EDTA, 154 mM NaCl, and 0.02% NaN₃ (pH 8.2). Lipid contents in individual fractions were determined with enzymatic assay kits (Sigma Diagnostics).

Mice were fasted for 4 h, then anesthetized for blood collection by retro-orbital puncture. Plasma samples were pooled from 4 mice, and lipoproteins were isolated from 2 ml of plasma by sequentially adjusting the densities with potassium bromide (26). We isolated six different density lipoprotein fractions (d < 1.006, 1.006-1.02, 1.02-1.05, 1.05-1.08, 1.08-1.10 and 1.10-1.21 g/ml) by ultracentrifugation at 40,000 rpm in a Beckman Ti-70.1 rotor at 10 °C for 20, 22, 24, 24, 36, and 48 h, respectively. A 20-µl aliquot from each fraction was assayed for total cholesterol and triglyceride content using an enzymatic assay (Sigma Diagnostics). The apolipoprotein profiles of the lipoprotein fractions were analyzed by SDS-polyacrylamide gel electrophoresis (4-20%) and Coomassie R250 staining. Total protein content of each fraction was measured using a Bio-Rad protein assay. Mean particle diameters of VLDL (<1.006 g/ml), LDL (1.006-1.063 g/ml), and HDL (1.063-1.21 g/ml) were determined by nondenaturing gradient gel electrophoresis performed on 3-27% polyacrylamide gels (27). Gels were stained with Coomassie G-250.

The cDNA containing the full-length mouse Apobec-1 (14) was excised from the pBluescript II KS (Stratagene) with BamHI/ClaI digestion and subcloned into the BglII/ClaI site of pAvCvSv shuttle vector (28). The recombinant adenovirus was prepared by cotransfection of pAvCvSv containing full-length mouse apobec-1 cDNA (14) and pJM17 (29) into 293 cells (30) as described previously (28). Two weeks after transfection, infectious recombinant adenoviral vector plaques were picked, propagated, and screened for apobec-1 sequences by polymerase chain reaction. Adenoviral vectors that contained Apobec-1 cDNA were purified on 293 cells. Large scale production of high titer recombinant adenovirus was performed as described (28). An adenovirus containing luciferase cDNA (31) instead of the Apobec-1 cDNA was used as a control.

Recombinant adenovirus stock was diluted with phosphate-buffered saline to the appropriate concentration, and 0.2 ml of diluted recombinant adenovirus was injected into the external jugular vein. At different times following adenoviral transduction, blood was collected from animals that had been fasted for 4 h.

RESULTS

Targeted Disruption of Apobec-1 in Mouse ES Cells and Generation of Apobec-1^/ Mice

To disrupt the apobec-1 gene in mouse ES cells, we constructed a replacement type targeting vector by inserting the neo cassette into partially deleted exon 7. The disrupted locus is located 5 to the region encoding a leucine-rich domain of Apobec-1 (14). Homologous recombination was verified by digestion of genomic ES cell DNA with HindIII and Southern blot analysis using probe A or probe B located outside the targeting vector (Fig. 1). The presence of a diagnostic 12.1-kb fragment in addition to the wild-type 7.5-kb or 3.0-kb band indicates insertion of the targeted vector by homologous recombination. Five of 74 clones (7%) that were G418- and FIAU-resistant exhibited a pattern consistent with homologous recombination. One of five independent ES cell clones was injected into C57BL/6J blastocysts, yielding three chimeric males with agouti coat color indicating essentially 100% contribution of ES cells. Chimeric males were bred with C57BL/6J females. Germline transmission of the mutant allele was detected in 33% (11/33) of the progeny by Southern blot analysis of tail DNA.

A total of 11 matings between heterozygous apobec-1^+/ mice yielded 88 offspring, comprising 22 apobec-1^+/+, 42 apobec-1^+/, and 24 apobec-1^/ animals, as expected for Mendelian segregation of the targeted gene. Homozygous male and female animals (apobec-1^/) were fertile and produced normal-sized litters.

The proportion of edited apoB mRNA in the liver and small intestine of wild-type and apobec-1^/ animals was assayed by primer extension (Fig. 2). In wild-type animals (apobec-1^+/+), more than 95% of intestinal apoB mRNA and ~60% of hepatic apoB mRNA consist of the edited species. In their apobec-1^/ littermates, edited apoB mRNA was undetectable in either tissue. Our results corroborate and confirm those reported very recently by Hirano et al. (18), namely the inactivation of a single gene locus, that of apobec-1, by itself is sufficient to totally abolish apoB mRNA editing in vivo. We note, however, that neither we nor Hirano et al. (18) could exclude an indirect effect of apobec-1 disruption as the cause of the loss of editing activity because these apobec-1^/ animals are chronically deficient in Apobec-1 and may develop many secondary changes that may be responsible for the observed phenotype.

Proportion of edited apoB mRNA in the liver and intestine of apobec-1^/, ^+/, and ^+/+ mice.

We performed a detailed analysis of the plasma lipids and lipoproteins in the control and knockout animals (F4 and F5 generations). There was no significant difference in total plasma cholesterol or triglyceride among the apobec-1^/, ^+/, or ^+/+ animals (Table I). When the plasma lipoproteins were fractionated by FPLC, we note that the apobec-1^/ mice consistently had significantly reduced amounts of HDL cholesterol compared with the apobec-1^+/ and ^+/+ mice (Fig. 3). Interestingly, the VLDL and LDL fractions were not different among the three groups.

Table I. Plasma cholesterol and triglyceride concentrations in apobec-1+/+, +/, and / mice Total plasma cholesterol and triglyceride concentrations were from the indicated number (n) of mice following a 4-h fast. The genotypes are homozygous knockout (/), heterozygous knockout (+/), or wild-type (+/+) for the apobec-1 gene. Total plasma cholesterol level Total plasma triglyceride level  /  /+ +/+  /  /+ +/+ mg/dl mg/dl 71  ± 17 70  ± 15 76  ± 12 61  ± 19 65  ± 23 56  ± 11 (n  = 34) (n  = 42) (n  = 16) (n  = 34) (n  = 42) (n  = 16)

FPLC profile of total plasma cholesterol in apobec-1^/, ^+/, and ^+/+ mice.

We next prepared the standard lipoprotein fractions by ultracentrifugal flotation and analyzed the relative size of the fractions by gradient gel electrophoresis (Fig. 4). There is no difference in the apparent size of the VLDL (d 1.006 g/ml), LDL (d 1.006-1.063 g/ml), or HDL (d 1.063-1.21 g/ml) fractions when we compared the samples from the apobec-1^/ or ^+/+ animals. By this technique, the sizes of the LDL were 27.25 ± 1.67 and 28.06 ± 0.98 nm and the sizes of the HDL were 13.18 ± 0.96 and 13.09 ± 0.97 nm for the apobec-1^+/+ and apobec-1^/ samples, respectively.

/

To further characterize the plasma lipoproteins, we sequentially floated them to six different density fractions, using plasma samples pooled from groups of 4 mice from each genotype. It is evident in Table II that the major difference between the different groups of animals is in the HDL lipids and proteins. Compared with apobec-1^+/+ mice, apobec-1^/ mice have a 44% and 35% reduction in their HDL cholesterol in the density ranges 1.10-1.21 and 1.08-1.10 g/ml, respectively. The corresponding reductions in protein are 12% and 39%, respectively. The apobec-1^+/ samples showed an intermediate lipid and protein content. This observation confirms the lower HDL levels detected by FPLC fractionation (Fig. 3). Less marked reductions in protein and cholesterol contents were also noted in the next density (d = 1.05-1.08) fractions. When the different density lipoprotein fractions were analyzed by SDS-gel electrophoresis, we noted the absence of apoB-48 and an increase in the intensity of the apoB-100 band in apobec-1^/ mice (Fig. 5). The lower protein content (Table II) in the d 1.10-1.21 and 1.08-1.10 fractions was reflected by a slightly less intense apoA-I band in these samples.

Table II. Cholesterol, triglyceride, and protein concentrations of plasma lipoproteins isolated by sequential ultracentrifugal flotation Each sample represents plasma pooled from four animals and prepared as described under ``Materials and Methods.'' The /, +/, and +/+ refer to the apobec-1 genotype of the mice. Density Cholesterol Triglyceride Protein  /  /+ +/+  /  /+ +/+  /  /+ +/+ mg/dl mg/dl µg/ml d < 1.006 6.5 4.5 7.7 30.9 23.2 42.4 119.2 86.0 134.1 1.006 -1.02 5.3 3.7 6.0 10.9 5.7 12.6 97.6 81.0 97.6 1.02 -1.05 20.6 20.8 19.8 14.2 13.6 13.5 227.2 247.2 182.4 1.05 -1.08 8.2 11.9 10.9 5.5 5.8 5.6 127.5 152.5 149.1 1.08 -1.10 4.7 7.1 8.4 3.0 3.3 3.6 132.5 197.3 217.3 1.10 -1.21 47.7 59.1 72.6 4.1 4.4 4.8 3578.3 4058.4 4066.8

SDS-polyacrylamide gel electrophoresis of lipoprotein fractions isolated by sequential ultracentrifugal flotation of plasma of apobec-1^/, ^+/, and ^+/+ mice.

We characterized the plasma apoBs in apobec-1^+/+, ^+/, and ^/ animals by SDS-gel electrophoresis. We found that apobec-1^+/+ and ^+/ mouse plasma contains both apoB-48 and apoB-100, whereas apobec-1^/ plasma contains only apoB-100 but no detectable apoB-48 (Fig. 6A). The intensity of the plasma apoB-100 band was substantially higher in the apobec-1^/ mice compared to the other two genotypes. Direct quantitation indicates that the plasma apoB-100 concentration in the apobec-1^/ animals was 4.96 ± 1.39 mg/dl, which is 276% of the level (1.80 ± 1.24 mg/dl) in the apobec-1^+/+ littermates; the apobec-1^+/ animals displayed an intermediate apoB-100 level of 2.69 ± 0.61 mg/dl (Fig. 6B). The plasma apoB-48 levels were similar (0.62 ± 0.44 and 0.58 ± 0.20 mg/dl, respectively) in wild-type and heterozygous animals, and apoB-48 was undetectable in apobec-1^/ animals. Therefore, inactivation of apobec-1 abolishes apoB mRNA editing, diverting the apoB-48 biosynthetic capacity of both the liver and small intestine to the production of apoB-100 exclusively, resulting in the disappearance of plasma apoB-48 and a marked accumulation of plasma apoB-100 in apobec-1 knockout mice. The plasma apoE is the same in wild-type (4.68 ± 1.06 mg/dl) and apobec-1 knockout (4.54 ± 0.97 mg/dl) animals.

SDS-polyacrylamide gel electrophoresis of plasma apoBs in apobec-1^/, ^+/, and ^+/+ mouse and quantitation of plasma level of apoB-100 and apoB-48.

/

+/

/

The apobec-1 Gene Disruption Markedly Raises Plasma Lipids, VLDL, and LDL in apobec-1^/ Mice Cross-bred with Human ApoB Transgenic (TgB) Mice

Mice are HDL animals and have very low plasma levels of apoBs. In order to examine the effect of elevation of basal plasma apoB on the response of these animals to abolition of apoB mRNA editing, we crossed the apobec-1^/ mice with human apoB transgenic (TgB) mice, which have an integrated human apoB transgene driven by its natural promoter (22). By repeated cross-breeding between apobec-1 knockout mice and TgB mice, we obtained siblings that were apobec-1^//TgB^+/ and apobec-1^+/+/TgB^+/. These animals had a moderate increase in basal plasma apoB-100 of 9.8 ± 3.9 mg/dl in the apobec-1^+/+/TgB^+/ mice, which was increased further to 26.6 ± 18.3 mg/dl in the apobec-1^//TgB^+/ mice. The total plasma apoE levels in these animals and their nontransgenic littermates were not different statistically, being 3.83 ± 0.79 mg/dl (n = 6), 3.88 ± 0.74 mg/dl (n = 5), 4.25 ± 0.47 mg/dl (n = 4), and 5.05 ± 0.26 mg/dl (n = 4) for the apobec-1^//TgB^/, apobec-1^+/+/TgB^/, apobec-1^//TgB^+/, and apobec-1^+/+/TgB^+/ animals, respectively. The total plasma lipids were significantly elevated in the TgB mice, which were also homozygous for the disrupted apobec-1 gene (Table III). The total plasma cholesterol was 80% higher (p < 0.05) and the triglyceride 58% higher (p < 0.05) in the apobec-1^//TgB^+/ mice compared with the apobec-1^+/+/TgB^+/ mice. We next fractionated the plasma lipoproteins of animals with the two genotypes by FPLC (Fig. 7). It is evident that the higher total plasma cholesterol in apobec-1^//TgB^+/ mice was accounted for almost exclusively by a marked elevation in LDL; their LDL cholesterol was 373% that of the apobec-1^+/+/TgB^+/ animals (Fig. 7A, Table III). There was also a slight increase in VLDL cholesterol. The increased plasma triglyceride was caused by a marked (2.0-fold) increase in VLDL triglyceride and a substantial (1.8-fold) increase in LDL triglyceride (Fig. 7B). There was no difference in the HDL cholesterol or triglyceride between the two groups of animals, and the LDL/HDL cholesterol ratio was ~3.3-fold higher in the apobec-1^//TgB^+/ mice compared with the apobec-1^+/+/TgB^+/ mice (Table III).

Table III. Lipid content of plasma and lipoprotein fractions and LDL/HDL ratio in apobec-1//TgB+/ and apobec-1+/+/TgB+/ mice VLDL, LDL, and HDL fractions were isolated by FPLC fractionation of 200 µl of total mouse plasma (Fig. 7). The number of animals studied in each group (n) is shown after each experimental value. TG, triglyceride. *, p < 0.05 compared to apobec-1+/+/TgB+/ value. Plasma cholesterol Total LDL cholesterol LDL/HDL ratio Plasma TG Total VLDL TG Total LDL TG mg/dl µg mg/dl µg µg apobec-1//TgB+/ 176  ± 28* (5) 103  ± 66* (4) 0.67  ± 0.45 106  ± 31* (5) 63  ± 17* (4) 110  ± 29 (4) apobec-1+/+/TgB+/ 94  ± 21 (4) 28  ± 17 (3) 0.20  ± 0.09 67  ± 11 (4) 32  ± 5 (3) 61  ± 21 (3)

FPLC profile of plasma lipoproteins in apobec-1^//TgB^+/ and apobec-1^+/+/TgB^+/ littermates.

/

+/

+/

/

/

In order to determine whether restoration of mouse apobec-1 expression can acutely reverse the loss of editing activity, we injected 2 × 10⁹ plaque-forming units of recombinant virus containing either the Apobec-1 cDNA (Ad-Apobec-1) or the luciferase cDNA (Ad-Luc) into apobec-1^/ mice. Four days and 8 days after administration of the recombinant viruses, we prepared total RNAs from the liver and small intestine and measured the relative amounts of edited apoB mRNA by primer extension (Fig. 8). It is clear that adenovirus-mediated transfer of apobec-1 efficiently restored apoB mRNA editing in the liver of these animals. There was no apparent effect of the gene transfer in the small intestine because little of the vector was taken up by the small intestine; Northern blots showed that the Apobec-1 transgene mRNA was detectable in the liver but not in the small intestine (data not shown). The control virus, Ad-Luc, had no effect on apoB mRNA editing in either the liver or the small intestine. We examined the plasma apoB by SDS-gel electrophoresis (Fig. 9) in apobec-1^/ animals 4 days and 8 days after treatment with Ad-Apobec-1 or Ad-Luc. Ad-Apobec-1 treatment caused the reappearance of plasma apoB-48 in these animals, accompanied by a marked reduction in the apoB-100 band, indicating that apoB-48 production and secretion was restored at the expense of apoB-100 by apobec-1 gene transfer in the liver of these animals. Ad-Luc had no effect on the plasma apoB species (Fig. 9).

Proportion of edited apoB mRNA in liver and intestinal RNA of apobec-1^/ mice following transduction in vivo by Ad-Apobec-1 or Ad-Luc.

/

SDS-polyacrylamide gel electrophoresis of plasma apoBs in apobec-1^/ mice before and 4 days after administration of Ad-Apobec-1 or Ad-Luc.

/

/

/

/

DISCUSSION

We created apobec-1 knockout mice to investigate the effects of inactivation of apoB mRNA editing on plasma lipoproteins in mice, a popular model for studying lipoprotein metabolism and atherosclerosis (16, 17). While we were preparing this manuscript for publication, Hirano et al. (18) reported that the targeted disruption of the mouse apobec-1 gene abolishes apoB mRNA editing and eliminates apoB-48. Using a different targeting strategy, we produced apobec-1 knockout mice, which similarly displayed a complete elimination of apoB mRNA editing and apoB-48 production. We also detected no difference in the total plasma cholesterol or triglyceride in the apobec-1^/, ^+/, or ^+/+ animals. The foregoing observations are quite consistent with the very recent study by Hirano et al. (18). Superficially, therefore, the complete diversion of apoB production from the apoB-48 to the apoB-100 form seemed not to produce any distinctive phenotypic effect on plasma lipids: a surprise and somewhat unexpected finding. The relative lack of obvious phenotypic effect led us to perform more in-depth experiments that revealed that apobec-1 gene inactivation in the presence of a wild-type genetic background is indeed associated with distinctive changes in lipoprotein profiles that are apparent only upon more detailed analysis. Apart from the phenotypic effects of apobec-1 gene disruption in wild-type animals, we will also discuss its effects in mice cross-bred with human apoB transgenic animals.

The most consistent change in the lipoprotein profile in apobec-1^/ mice with a wild-type genetic background is a 30-40% reduction in plasma HDL. This was evident when the plasma lipoproteins were fractionated by FPLC (Fig. 3) or by ultracentrifugal flotation (Table II) and was especially marked in the d 1.08-1.10 and 1.10-1.21 g/ml fractions. Both cholesterol and protein were reduced in these fractions. These changes in HDL level were not accompanied by any change in the apparent sizes of the HDL or of the other major plasma lipoproteins when they were analyzed by gradient gel electrophoresis (Fig. 4).

The abolition of apoB mRNA editing totally eliminates apoB-48 production, and the liver and small intestine synthesize exclusively apoB-100. It is not clear if there is feedback regulation of apoB-100 production when all the apoB mRNA has been diverted to apoB-100 biogenesis. Direct quantitative measurements revealed that apobec-1^/ mice have plasma apoB-100 levels that are markedly elevated to 276% that of apobec-1^+/+ animals (Fig. 5). The heterozygous apobec-1^+/ animals showed an intermediate level. Therefore, there is accumulation of plasma apoB-100 when all apoB-48 production is shut down and the apoB mRNAs are completely diverted toward apoB-100 production. If there was feedback regulation of plasma apoB-100, it was ineffective in holding the level of the protein constant under these conditions. Furthermore, there is no change in plasma apoE in these animals that might account for the lack of phenotypic differences.

Although there is a marked increase in plasma apoB-100 in apobec-1^/ mice, there is no apparent change in the plasma VLDL that requires apoB as a structural component. Since apoB-48 is an integral part of VLDL in mice, its replacement by apoB-100 may have little effect on the amount of circulating VLDL. We note, however, under the appropriate conditions, i.e. increased basal plasma apoB-100 in TgB mice, the substitution of apoB-48 by apoB-100 does in fact elevate VLDL (see below). In a study of 12 different mammalian species, Greeve et al. (32) found that the presence of hepatic apoB mRNA editing was associated with a low plasma VLDL/LDL. Gene transfer experiments indicate that hepatic apobec-1 overexpression is highly effective in lowering plasma LDL in mice and rabbits (23, 33). However, the initial observation in the apobec-1^/ animals with a wild-type genetic background suggests that perhaps the converse is not true and lack of editing may not alter plasma LDL. We did not believe that this is a plausible conclusion but reasoned that the apparent lack of effect of absence of apoB mRNA editing in mice may be the consequence of the much lower plasma apoB-100 in the basal state in these animals compared to humans. Despite the impressive (176%) increase in plasma apoB-100 (Fig. 6), the plasma concentration of this apoprotein in apobec-1^/ mice is still only about 5% that in humans. It is likely that alterations in plasma LDL may not be evident because apoB-100 and LDL levels are so low in these animals. Possible increases in LDL concentration may be masked by the HDL reduction because there is partial overlap between LDL and HDL₁ in some of these separation methods (34). Therefore, observations on plasma LDL in knockout mice with a wild-type genetic background may not be directly extrapolated to the situation in humans.

In order to examine the effect of apobec-1 gene knockout in a mouse model with apoB levels higher than that in wild-type mice, we cross-bred the apobec-1^/ animals with TgB mice to produce homozygous apobec-1^/ animals carrying the TgB gene. We note that TgB mice express the human apoB transgene in the liver only but not in the small intestine (22). The abolition of editing therefore affects human apoB expression in the liver only; it was, however, highly effective in producing a dramatic phenotype. The TgB mice have a basal apoB-100 level (9.8 ± 3.9 mg/dl) that is about 5 times the wild-type level, but is still only about 10% that in humans. Disruption of apobec-1 raised this level to 26.6 ± 18.3 mg/dl, which is ~30% that in humans. In the presence of the human apoB transgene and the moderately increased plasma apoB-100, the inactivation of apobec-1 produced increases of 80% and 58%, respectively, in total plasma cholesterol and triglyceride (Table III). Furthermore, the increase in cholesterol was accounted for by a 273% increase in plasma LDL cholesterol in apobec-1^//TgB^+/ animals compared with apobec-1^+/+/TgB^+/ controls (Fig. 7A). The increase in plasma triglyceride was distributed both in the VLDL and IDL/LDL fractions (Fig. 7B). We conclude that the higher basal apoB-100 level in the presence of the apoB transgene allowed the mice to further increase their plasma apoB-100 level when they lost their ability to edit apoB mRNA, leading to a rise in VLDL and LDL. There was no difference in the plasma HDL in these TgB mice in the presence or absence of the disrupted apobec-1 gene and the LDL/HDL cholesterol ratio was ~3.3-fold higher in apobec-1^//TgB^+/ than in apobec-1^+/+/TgB^+/ mice (Table III). One explanation for the major difference in the effect of apobec-1 knockout on the plasma lipids and VLDL and LDL lipids in TgB mice and non-TgB mice is the substantially higher basal plasma apoB-100 in the TgB animals. Although the total level of plasma apoB-100 was still relatively low in these animals compared to humans, it was sufficient to unmask a marked LDL response. Another possible explanation is that human apoB-100 behaves differently from mouse apoB-100 in its competence in effecting VLDL production and LDL biogenesis. To address this possibility, one needs to generate animals that genetically overexpress mouse apoB at a comparable level and examine whether apobec-1 knockout allows such animals to marshal VLDL and LDL responses similar to those seen in mice with the integrated human apoB transgene. Finally, the plasma apoE levels are similar in the transgenic animals with and without the inactivated apobec-1 gene, which indicates that the lipoprotein changes are not mediated by changes in apoE expression.

The study of Hirano et al. (18) and the phenotypic effects described above implicate apobec-1 as required for apoB mRNA editing. However, the evidence reviewed thus far does not exclude the possibility that the loss of editing is an indirect effect of apobec-1 inactivation because the lack of editing was observed in animals that were chronically depleted of functional Apobec-1. It is well known that secondary changes could contribute to phenotypic expression in knockout mice produced by gene targeting. To ensure that some secondary effect of apobec-1 disruption was not responsible for the absence of apoB mRNA editing, we attempted to acutely restore mouse apobec-1 expression in the liver by adenovirus-mediated gene transfer. Indeed, this maneuver was highly effective in restoring apoB mRNA editing in this organ (Fig. 8). It also resulted in the reappearance of apoB-48 in the circulation (Fig. 9). We conclude that the abolition of apoB mRNA editing was a direct effect of the absence of Apobec-1caused by apobec-1 inactivation in vivo. This conclusion is important because it indicates that Apobec-1 is an essential and integral component of the apoB mRNA editing machinery.