Overexpression of Human Apolipoprotein C-I11 in Transgenic Mice Results in an Accumulation of Apolipoprotein B48 Remnants That Is Corrected by Excess Apolipoprotein E* “

Overexpression of human apolipoprotein (apo) (3-111 in the plasma of transgenic mice results in hypertriglyc- eridemia, with up to a 20-fold elevation in plasma triglyceride. Nearly all of the triglyceride accumulates in the d c 1.006 g/ml lipoprotein fraction, which consists predominantly of apoB48-containing particles having a low apoE:apoB48 ratio in contrast to normal mice. The transgenic and nontransgenic d e 1.006 g/ml lipoproteins are similar in size, and they are equivalent substrates for lipoprotein lipase in vitro. Total apoBlOO levels are similar in transgenic and normal plasma, but apoB48 levels are increased in transgenic mice. The transgenic d e 1.006 glml particles are poor competitors for the binding of low density lipoproteins to the low density lipoprotein receptor in vitro, which is corrected by the addition of exogenous apoE. The rate of clearance of labeled chylomicron remnants in apoC-111-transgenic mice was about half that in nontransgenic mice. The lipoprotein alterations are accompanied by up to a 5-fold also in the indicate that the key

Overexpression of human apolipoprotein (apo)  in the plasma of transgenic mice results in hypertriglyceridemia, with up to a 20-fold elevation in plasma triglyceride. Nearly all of the triglyceride accumulates in the d c 1.006 g/ml lipoprotein fraction, which consists predominantly of apoB48-containing particles having a low apoE:apoB48 ratio in contrast to normal mice. The transgenic and nontransgenic d e 1.006 g/ml lipoproteins are similar in size, and they are equivalent substrates for lipoprotein lipase in vitro. Total apoBlOO levels are similar in transgenic and normal plasma, but apoB48 levels are increased in transgenic mice. The transgenic d e 1.006 glml particles are poor competitors for the binding of low density lipoproteins to the low density lipoprotein receptor in vitro, which is corrected by the addition of exogenous apoE. The rate of clearance of labeled chylomicron remnants in apoC-111-transgenic mice was about half that in nontransgenic mice. The lipoprotein alterations are accompanied by up to a 5-fold increase in circulating nonesterified fatty acids, which may be the cause of fatty livers and increased liver triglyceride production also observed in the transgenic mice.
These observations indicate that the primary defect leading to hypertriglyceridemia in apoC-I11 overexpressers is an impaired clearance of apoB48 remnants due to apoE insufficiency. Therefore, transgenic mice that overexpressed human apoE were cross-bred with the apoC-I11 overexpressers. Transgenic progeny that produced both human apoE and human apoC-I11 had normal levels of plasma triglyceride and normal amounts of apoB48 remnants. Thus, our studies suggest that a function of apoC-I11 is to modulate the apoE-mediated clearance of lipoproteins, and that the concentration of apoC-111 relative to apoE is a key determinant of triglyceride levels in plasma.
Human apolipoprotein (apo)' C-I11 is a M, =  high density lipoproteins (HDL) (1). It is present in normal human plasma at 7-12 mg/dl (2-4). While its normal function in plasma is not understood, several possibilities have been suggested.
Analysis of human clinical populations have implicated apoC-I11 in plasma triglyceride metabolism. For example, Kashap et al. (4) have reported that hypertriglyceridemic patients have elevated levels of plasma apoC-111. In addition, a Sac1 restriction fragment length polymorphism (the S2 allele), resulting from a cytosine to guanosine substitution in the 3'untranslated region of apoC-III (5), is associated with elevated plasma triglyceride levels (6). Even among randomly selected young healthy English subjects, Shoulders et al. (7) determined recently that the S2 allele is correlated directly with plasma apoC-I11 levels and an increase in plasma triglyceride levels in both men and women. These observations suggest that increased amounts of plasma apoC-I11 may predispose or be a consequence of the development of high plasma triglyceride levels.
Evidence that apoC-I11 modulates the uptake of triglyceriderich lipoproteins by the liver was obtained by Windler et al. (8)(9)(10). Using an isolated perfused rat liver model, they showed that the addition of human or rat apoC-I11 to small chylomicrons or VLDL effectively impaired the clearance of these particles by the liver. Similar results were obtained by Shelburne et al. (ll), who demonstrated that the addition of human apoC-111 inhibited the apoE-mediated increase in the hepatic uptake of large rat lymph chylomicrons and triglyceride emulsions in a nonrecycling isolated rat liver perfusion system.
A potential role for apoC-I11 in the modulation of lipoprotein lipase activity was proposed by Brown et al. (121, who reported that apoC-I11 inhibited the hydrolysis of triglyceride by lipoprotein lipase in vitro. In addition, Wang et al. (13) determined that apoC-111-2, a disialylated form of apoC-111, was a noncompetitive inhibitor of lipoprotein lipase. They also found that lipoprotein lipase inhibitory activity in the plasma of hypertriglyceridemic patients who had normal amounts of lipoprotein lipase correlated directly with plasma apoC-I11 levels. Two female siblings with combined apoC-111-apoA-I deficiency (14) resulting from a 6.5-kb translocation from the apoC-I11 gene to the apoA-I gene (15) had an increased fractional catabolic rate of VLDL and an efficient conversion of VLDL and intermediate density lipoproteins (IDL) to low density lipoproteins (LDL) (16). Consequently, it was suggested that the major role of apoC-I11 is to modulate triglyceride hydrolysis rather than remnant clearance; however, the lack of HDL in these patients complicates this interpretation. The results of Kinnunen and Ehnholm (17) suggest that apoC-I11 also may impair hepatic lipase-mediated hydrolysis of triglyceride, and that the preferred substrates for hepatic lipase may be those that are relatively deficient in apoC-111. Ito et al. (181, using a transgenic mouse model, found a clear relationship between increased levels of apoC-I11 in vivo and hypertriglyceridemia. Two transgenic lines were established that incorporated either -2 or -100 copies of a 6.7-kb genomic fragment containing the human apoC-I11 gene. The high copy transgene line was severely hypertriglyceridemic, whereas the low copy transgene line was mildly hypertriglyceridemic. Based on the absorbance profile a t 280 nm of plasma fractions collected by gel filtration, the authors showed an increase in plasma VLDL levels and a decrease in plasma HDL levels in the severely hypertriglyceridemic transgenic line. These findings were extended further by Aalto-Setala et al. (191, who also found evidence for a decreased fractional catabolic rate of VLDL, accompanied by a n increase in triglyceride production in the apoC-111-transgenic mice. Recently von Eckardstein et al. (20) described two human subjects carrying a lysine 58 to glycine mutation of apoC-111, with decreased plasma apoC-I11 levels. These subjects had the curious plasma profiles of decreased levels of triglyceride and increased levels of HDL cholesterol. In addition, the presence of large, atypical, apoE-rich HDL were observed. This observation of hyperalphalipoproteinemia associated with a variant apoC-I11 suggests that the role of apoC-I11 may not be confined only to the metabolism of triglyceride-rich lipoproteins.
In the present study, we investigated the metabolic consequences of the overexpression of apoC-I11 in transgenic mice to examine the functions of this apolipoprotein. Plasma lipoproteins have been separated by a sequential density centrifugation protocol that permits a more accurate assessment of the metabolic defect in these mice than has been possible previously. The results suggest that apoC-I11 has a regulatory role in the metabolism of chylomicron remnants and HDL1, and that excess apoC-I11 results in a decrease in the rate of clearance of these lipoproteins, leading to hypertriglyceridemia. Our data are not consistent with a decrease in the rate of clearance of apoB100-VLDL caused by overexpression of apoC-III (19). The regulatory function of apoC-I11 suggests that it may have a role in the etiology of some hyperlipidemias.

EXPERIMENTAL PROCEDURES
Generation of l'bansgenic Mice Transgenic mice were generated as previously described (21) with ICR mice, obtained from Charles River Laboratories (Wilmington, MA), by using two apoC-I11 genomic constructs ( Fig. 1) derived from the ACA2.1 cosmid (22). The C3.PP construct was a 4.4-kb PstI-PstI fragment that contained 200 base pairs of 5"flanking sequence and 780 base pairs of 3'-flanking sequence, and C 3 . m was a 10.5-kb XbaI-XbaI fragment that contained 5.9 kb of 5"flanking sequence and 1.1 kb of 3"flanking sequence. All analyses of transgenic mice were carried out with mice heterozygous for the transgene, and analyses were performed on at least four different mice from each construct line. RNase protection analysis (23) of the total RNA collected from 12 different tissues from F1 transgenic progeny from each founder showed that the C3.PP construct was expressed only in the liver, whereas the C 3 . m construct was expressed both in liver and intestine (Fig. 1).

Lipoprotein Isolation and Characterization
Preparation of Lipoproteins-Blood was collected under Metofane anesthesia (Pitman-Moore, Mundelein, IL) by heart puncture or from the tails of mice that had unrestricted access to water and Purina 5001 Mouse Chow containing 4.5% fat, as well as mice that were fasted for 8 h. Mice were maintained on a lighvdark cycle of 7:OO a.m./7:00 p.m., and blood was collected from fed mice between 8:30 a.m. and 9:00 a.m., and from fasted mice between 4:30 p.m. and 5:OO p.m. The blood was collected into a final concentration of 1 mg/ml EDTA and 50 unitdml   (23). For each tissue, 2 pg of RNA were hybridized to a radioactively labeled human apoC-I11 antisense probe, then digested with RNase as described elsewhere (23). Remaining labeled hybrids were resolved by gel electrophoresis and detected by autoradiography (23).
Trasylol (Miles, New Haven, CT) to prevent coagulation and to minimize proteolysis. Plasma was obtained by centrifugation at 3,000 x g for 15 min at 4 "C. The plasma from single mice, or pooled from two littermates of the same sex, was adjusted to different densities with KBr, and lipoprotein fractions were collected by sequential density ultracentrifugation using the TLA100.2 fixed angle rotor in a Beckman TLlOO ultracentrifuge. The lipoproteins in the density ranges of d < 1.006, d = 1.006-1.02, d = 1.02-1.04, d = 1.04-1.06, d = 1.06-1.08, and d = 1.08-1.10 g/ml were obtained following successive centrifugations at 100,000 rpm, 4 "C for 2.5 h. The d = 1.10-1.21 g/ml fraction was isolated following centrifugation a t 100,000 rpm, 4 "C for 4 h. The lipoproteins in each density range were recovered by tube slicing and dialyzed against 10 m Tris a t pH 7.4, 150 m NaC1, 1 m EDTA. The recovery of triglyceride and cholesterol in the isolated lipoprotein fractions was >801.
Agarose Gel Electrophoresis of Lipoproteins-Isolated density fractions were resolved by electrophoresis in a 1% agarose gel in barbital buffer for 35 min, then neutral lipid-containing lipoproteins were detected by staining the gels with Fat Red 7B according to the manufacturer's directions (Ciba Coming, Palo Alto, CA). To detect the apolipoprotein components, the separated lipoproteins in the agarose gels were transferred to nitrocellulose by blotting, followed by reaction of the nitrocellulose blots with apolipoprotein-specific antibodies.
Polyacrylamide Gel Electrophoresis of Lipoproteins-Apolipoprotein components of isolated lipoprotein fractions were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of 30-111 aliquots in 3-15% or 10-201 gradient gels as described elsewhere (26). The apolipoproteins were stained with 0.5% Coomassie Brilliant Blue or they were blotted to nitrocellulose and reacted with apolipoproteinspecific antibodies as previously described (27).
Antibodies-Apolipoprotein antibodies generated against human apoA-I or apoB (Calbiochem, San Diego, CA) that cross-reacted with mouse apolipoproteins were used. The mouse apoC-I, apoC-11, and apoC-111 antibodies were raised against purified mouse apolipoproteins. The anti-human apoC-I11 antibody used in this study did not cross-react with mouse apoC-111, and the anti-mouse apoC-I11 antibody did not cross-react with human apoC-111. An anti-rat antibody was used to detect mouse apoE. The primary antibodies were detected by reaction with an 1251-labeled secondary antibody, followed by autoradiography, or a horseradish peroxidase-conjugated secondary antibody followed by chemiluminescent detection.
Negative Staining Electron Microscopy oflipoproteins-Plasma lipoprotein density fractions were diluted to 0.1 mg of proteidml in 10 m Tris, pH 7.4, 150 m NaCI, and 1 m EDTA, then adjusted to 1% phosphotungstic acid, pH 7.0. Sucrose was added to a concentration of 0.11, then the samples were dried on Formvar carbon-coated grids and   1596 459 193 f 51 examined using a JEOL 100 CXII transmission electron microscope. Quantitation of the particle size distribution of each lipoprotein fraction was carried out by examination of the diameters of at least 200 particles, using an Image-VAT analysis system (Universal Imaging Corp., West Chester, PA). Analytical Methods-Cholesterol and triglyceride determinations were camed out using either enzymatic assays developed for the semiautomated Abbott Spectrum Analyzer (Abbott) or a quantitative colorimetric plate assay. In this latter assay, specific dilutions of cholesterol (Abbott) or triglyceride (Boehringer Mannheim) standards, and samples for lipid determination, were pipetted into a 96-well microtiter plate in 100-pl aliquots. Cholesterol or triglyceride-color developing reagent (Abbott and Boehringer Mannheim, respectively) was added to the samples in 100-pl aliquots and incubated for 15 min at 37 "C. Color intensity of the samples was read at 490 nm. Nonesterified fatty acid levels in plasma were determined using an analytical kit from WACO Chemicals USA, Inc. (Richmond, VA) according to the manufacturer's directions.
The protein content of lipoprotein samples was determined using the Lowry method (28). The content of human apoC-111 in transgenic mouse plasma was determined by quantitative immunoblot analysis, using purified human apoC-I11 as the standard.

Thin Section Light Microscopy of Livers
Mice that were 8-10 weeks of age were anesthetized with sodium pentobarbital (750 pgf10 g of body weight) and perfused with saline via the left ventricle of the heart. After perfusion with 4% paraformaldehyde, the livers were excised and placed in 4% paraformaldehyde at room temperature overnight. Thin sections (1 pm) of the livers were embedded in Epon, then stained with toluidine blue.
In Vitro Lipolysis of Piglyceride in the d < 1.006glml

Plasma Fraction
Purified lipoprotein lipase from bovine milk was added to increasing concentrations of the d < 1.006 gfml plasma fraction in the presence of excess fatty acid-free albumin. The reaction mixture was incubated at 37 "C for 1 h, and the released nonesterified fatty acids were quantitated using an analytical kit (WACO Chemicals USA, Inc.) according to the manufacturer's directions. The assay was performed on three different preparations of d < 1.006 g/ml fractions from normal and transgenic mice. The variation between determinations was <0.05 mmoVl of fatty acid released.

Fibroblast LDL Receptor Binding Assay
The ability of d < 1.006 gfml lipoproteins from normal and human apoC-111 overexpressing mice to compete for 1261-labeled LDL binding to the LDL receptor was determined using a human foreskin fibroblast LDL receptor binding assay (29). To determine if added apoE enhanced the ability of these lipoproteins to compete for '261-labeled LDL binding to the LDL receptor, 50 pg of purified human apoE was incubated with 300 pg of normal or transgenic d < 1.006 gfml triglyceride at 37 "C for 1 h. The lipoprotein-associated apoE was separated from free apoE by Superose 6 HR10/30 gel filtration using fast protein liquid chromatography (Pharmacia LKB Biotechnology Inc.). The apoE-d < 1.006 gfml lipoprotein mixture (200 PI) was injected onto the Superose 6 column, equilibrated in phosphate-buffered saline (10 ~l l~ sodium phosphate, pH 7.4, 150 n m NaCl, and 1 l l l~ EDTA), and eluted at a flow rate of 0.5 mI/min. Competition for '251-labeled LDL binding by the mouse d < 1.006 gfml fraction was determined as described (29).

In Vivo Chylomicron a n d Chylomicron Remnant Clearance
Chylomicrons were isolated from dog lymph and the remnants were prepared by injection of the dog chylomicrons into hepatectomized rabbits as described (30, 31). Normal (n = 4) and transgenic (n = 3) mice overexpressing apoC-111 were injected via the tail vein with [14C]cholesterol and PHIretinol dual-labeled chylomicron remnants (3 mg of triglyceride). Blood was obtained from the mice 0, 5, and 10 min after injection of the chylomicron remnants. At 20 min, the mice were exsanguinated, and tissues (liver, spleen, and kidney) were collected to determine the uptake of the radioactive remnants. Plasma samples were collected 20 min after injection, adjusted to d = 1.02 g/ml with KBr, and separated by ultracentrifugation in a TLlOO tabletop Beckman ultracentrifuge for 2.5 h at 100,000 rpm, 4 "C. At collection time, >85% of the radioactivity was recovered in the d < 1.02 gfml plasma fraction and the tissues. The radioactivity present in tissues was determined as described previously (31).
Canine chylomicrons (10 mg of triglyceride) were injected into the tail veins of normal (n = 10) and apoE-transgenic (n = 8) mice. Plasma samples were collected 0, 5, and 15 min after injection. The mice were exsanguinated 30 min after injection, and the liver, spleen, and kidneys were removed to determine the clearance and uptake of radioactivity.

Determination of in Vivo Diglyceride Synthesis following
Injection of Diton WR-1339 with Triton WR-1339 in saline at 20 mgfg of body weight via the tail Normal (n = 3) and apoC-111-transgenic (n = 3) mice were injected vein. Blood was collected under anesthesia just prior to injection, and at 60, 120, and 200 min after injection, and triglyceride concentrations in plasma were determined.

RESULTS
Characterization of the Hyperlipidemia in the ApoC-IZItransgenic Mice-Three transgenic mouse lines that expressed human apoC-I11 were established, with the amounts in mouse plasma ranging from about 2-fold to 12-fold above normal human plasma levels ( Table I). The concentrations of human apoC-I11 in transgenic mouse plasma did not correspond closely with gene copy number, probably reflecting the influence of neighboring endogenous mouse sequences at the different sites of integration of the human transgene. The transgenic mice had moderate to severe hypertriglyceridemia, with elevations in total plasma triglycerides corresponding to the amount of human apoC-I11 in the plasma, in agreement with the observations of It0 et al. (18) and Aalto-Setala et al. (19). The in- struct) were sources of human apoC-111. The high expresser line also had a 2-fold elevation in total plasma cholesterol levels.
The elevation in triglyceride levels in transgenic mice was up to 1.7-fold greater in males than in females (Table I). Plasma cholesterol levels also were increased as much as 1.6-fold in transgenic males compared to females. For both triglyceride and cholesterol, these differences were greater in mice with higher levels of apoC-111. In nontransgenic mice, plasma cholesterol was 1.4-fold greater in males than in females, but no significant difference was observed between male and female triglyceride levels.

Alterations in Plasma Lipoproteins in the ApoC-ZZZ-transgenic Mice-
To determine the effect of overexpression of apoC-I11 on plasma lipoproteins, plasma fractions were collected by sequential density centrifugation using density ranges (described under "Experimental Procedures") that had been optimized to separate major classes of mouse lipoproteins.2 The isolated density fractions were analyzed by agarose gel electrophoresis followed by staining with Fat Red 7B to identify lipoproteins that contained neutral lipid. In normal mouse plasma (Fig. 2, upper left panel), the d < 1.006 g/ml fraction contained pre-P-migrating VLDL, and the d = 1.006-1.02 g/ml fraction contained IDL. The d = 1.02-1.04 and 1.04-1.06 g/ml fractions, however, contained both a-migrating HDLl lipoproteins (band c) and pre-P-migrating apoB48-containing remnants (band b), as well as P-migrating LDL (band a ) . Fractions of higher density contained mainly a-migrating HDL. The identity of the lipoprotein classes in each density fraction had been determined previously by sequential density separation of lipoproteins from large volumes of pooled mouse plasma, followed by agarose gel electrophoresis, elution of individual lipoprotein classes from the gel, and analysis of their apoiipoprotein and lipid components. 2 Transgenic mice overexpressing human apoC-I11 had profound alterations in plasma lipoproteins (Fig. 2,

panel).
A large increase in the amount of the d < 1.006 g/ml VLDL fraction was apparent. The d = 1.006-1.02 g/ml fraction contained substantial amounts of HDLl in addition to IDL. Large increases in the amount of HDLl and apoB48 remnants were observed in the d = 1.02-1.04 and 1.04-1.06 g/ml fractions. The distribution of apoB48 remnants extended into the d = 1.06-1.08 g/ml fraction, where they are not found normally. A decreased amount of apoBlOO LDL was observed in the d = 1.02-1.04 and 1.04-1.06 g/ml fractions. A decrease in the quantity of HDL in the d = 1.08-1.10 g/ml fraction was also noted. Analysis of the lipid content of the individual fractions (Table  11) revealed that essentially all of the excess triglyceride in apoC-111-transgenic mice was in the d < 1.006 g/ml VLDL fraction. An increase also was observed in the cholesterol content of the d < 1.006 g/ml fraction of the transgenic mice, in contrast to nontransgenic mice, where the majority of the plasma cholesterol was recovered in the d = 1.08-1.10 and 1.10-1.21 g/ml HDL fractions. In the transgenic mice, the d = 1.02-1.04 and d = 1.04-1.08 g/ml fractions were enriched in total cholesterol, much of i t esterified. This finding was consistent with the observed increase in the number of HDLl and apoB48 lipoproteins in these fractions (Fig. 2). In contrast, the cholesterol content in d = 1.08-1.10 and 1.10-1.21 g/ml HDL fractions was decreased in the transgenic mice, consistent with a slight decrease in plasma HDL content.

Effect of ApoC-ZZZ Overexpression on Apolipoprotein Distribution-
The effect of the high expression levels of transgenic human apoC-I11 on the distribution of endogenous mouse apoC-I11 in plasma lipoprotein fractions was investigated by immunoblot analysis (Fig. 2, upper middle and lower middlepanels). In nontransgenic mice, apoC-I11 was distributed among several lipoprotein classes, with most of it found in dense HDL and d < 1.006 g/ml VLDL. In transgenic animals, mouse apoC-I11 was decreased in lipoproteins of higher densities, and substantial amounts of mouse apoC-I11 were detected in apoB48 remnant lipoproteins. Human apoC-I11 in transgenic mice (Fig. 2, lower  right panel) had a similar distribution as mouse apoC-I11 in plasma lipoproteins.
The apolipoprotein content of the individual density frac-  tions was examined by immunoblot analysis (Fig. 3). While most of the apoB was in the VLDL fraction, the distribution of apoB was different between normal and transgenic mice a t higher densities. In normal mice, the majority of apoB in the d = 1.006-1.02, the d = 1.02-1.04, and the d = 1.04-1.06 g/ml fractions was in a slow moving band (band a), characteristic of the electrophoretic mobility of apoBlOO remnants and LDL.2 In contrast, the apoB in transgenic plasma was primarily in faster migrating apoB48 particles (band b), which were predominant in the d = 1.006-1.06 g/ml fractions. Compared to its distribution in normal plasma, apoE content increased slightly in transgenic mice in the d c 1.006 g/ml fraction, in apoB48 remnants, and in large HDL1. This distribution reflected the accumulation of these lipoproteins in transgenic mice. Apolipoprotein E also accumulated in a P-HDL particle in the d = 1.10-1.21 g/ml fraction in transgenic mice (Fig. 3). This particle was present in much lower amounts in normal mice. 2 The distribution of apoC-I (Fig. 3) among plasma lipoproteins in normal and transgenic mice was essentially the same as that of mouse apoC-111 (Fig. 2) in the corresponding animals. Apolipoprotein C-I1 had a similar distribution in normal and transgenic mice but showed a redistribution from d = 1.06-1.10 g/ml fractions to lower densities, with some apoC-I1 found on apoB48 remnants in the transgenic mice.

Effects of Overexpression of ApoC-111 in Dansgenic Mice
The accumulation of apoB48 remnants in the transgenic mice was reflected by an increase in the total amount of apoB48 in plasma, as determined by SDS-PAGE followed by electroimmunoblot analysis (Fig. 4, left panel). Compared to normal mice, the transgenic mice had relatively little change in the content of apoB100, but the amount of apoB48 increased substantially. Examination of the apolipoprotein content of the various lipoprotein density fractions by SDS-PAGE followed by protein staining or electroimmunoblot analysis showed that nearly all of the increased apoB48 in the transgenic mice was found in the d e 1.006 g/ml fraction (Fig. 4, right panel). Densitometric scanning of these stained gels showed that the apoB48:apoE ratio was increased up to &fold in the transgenic mouse compared to the normal mouse (data not shown). Three major electrophoretic isoforms of transgenic human apoC-I11 were observed, similar to those found in normal human plasma, and digestion with neuraminidase resulted in a shift of these isoforms to a single species (data not shown).
Particle Size Distribution among Lipoprotein Density Fractions-Lipoprotein sizes in the various density fractions were measured by negative-staining electron microscopy (Fig. 5). The d < 1.006 g/ml VLDL fraction of normal and transgenic plasma contained particles that ranged in size from -100 to 800 A, with a n average diameter of 335 and 397 b; in the normal and transgenic particles, respectively. The d = 1.006-1.02 g/ml IDL fraction of transgenic plasma showed a shift in average particle diameter from 194 A in normal plasma to 215 b; in transgenic plasma. The lipoproteins of the d = 1.02-1.04 g/ml fraction in both normal and transgenic plasma were similar in size, even though two distinctly different types of particles (HDL1 and apoB48 remnants) were present (Fig. 2). The sizes of lipoproteins in the higher density fractions showed significant differences between normal and transgenic mice, reflecting apparent alterations in HDL metabolism.
Lipolysis of Diglyceride-rich Lipoproteins-The lack of accumulation of large particles in the d < 1.006 g/ml fraction suggested that lipoprotein lipolysis occurred at similar rates in both transgenic and nontransgenic mice. To examine this possibility, mice were fasted for 8 h with free access to water, resulting in 64% and 54% decreases in total plasma triglyceride levels in normal and transgenic mice, respectively (data not shown). Total cholesterol levels decreased about 40% in both groups of animals. These results suggested that lipase activity in the transgenic mice might not be affected significantly by the high levels of circulating apoC-111. Heparin administration released additional lipase activity into plasma, resulting in a dramatic decrease in plasma triglyceride levels (Table 111). In post-heparin plasma, nonfasting transgenic mice showed an 85% decrease in plasma triglycerides compared to the 70% decrease in normal mice.
To determine if impaired hydrolysis of plasma triglyceride might contribute to the hypertriglyceridemia in apoC-111-transgenic mice, the d c 1.006 g/ml lipoproteins were isolated from transgenic and normal plasma, and the ability of bovine lipoprotein lipase to hydrolyze the triglyceride in this fraction was determined (Fig. 6). The triglyceride in the transgenic d < 1.006 g/ml fraction was hydrolyzed efficiently, although the rate and extent of hydrolysis were slightly less than that observed for the normal d < 1.006 g/ml fraction. Analysis of three different  preparations of the normal and transgenic d < 1.006 g/ml fractions showed that the maximum hydrolytic rate of the transgenic fraction was -8% less than the corresponding fraction from normal mice. Thus, the earlier contention (12,13,16) that apoC-I11 may inhibit lipoprotein lipase activity appears not to be a significant factor in the hyperlipidemia resulting from apoC-I11 overexpression in transgenic mice.
The small difference in hydrolysis between the normal and transgenic d < 1.006 g/ml lipoproteins may be due to differences in lipid and protein composition. With respect to this possibility, isolated transgenic d < 1.006 dm1 lipoproteins were found to contain apoC-I1 (Fig. 31, a necessary cofactor for lipoprotein lipase, but in amounts decreased by up to -75% in the transgenic compared to the normal plasma fraction (data not shown). This remaining apoC-I1 in the d e 1.006 g/ml fraction, however, appeared sufficient to catalyze the hydrolysis of triglyceride in transgenic mice at nearly normal levels, as demonstrated in Fig. 6.

Effect of Diton WR-1339 Administration on the Diglyceride
Content of Plasma-The possibility that overexpression of apoC-I11 might affect the production of triglyceride-rich lipoproteins was investigated. Triton WR-1339 inhibits the lipolysis and clearance of triglyceride in vivo, thus enabling an indirect determination of the rate of liver triglyceride output (32). Therefore, normal and transgenic mice were injected with Triton WR-1339, and the concentration of plasma triglycerides was measured during a 200-min postinjection period (Fig. 7). The initial rate of triglyceride synthesis in high expresser apoC-111-transgenic mice appeared to be -2-fold higher than in nontransgenic mice. At the end of the experiment, the transgenic mice had 1.6-fold higher levels of plasma triglyceride than normal mice. These results suggested that the production of triglyceride by the liver is increased in transgenic mice. The relatively modest difference in triglyceride production, as indicated by the indirect Triton WR-1339 assay, and the minimal differences in triglyceride hydrolysis found here and by Aalto-Setala et al. (19) between normal and transgenic mice seemed insufficient to account for the up to 20-fold increase in mice had liver weights that averaged -8% of body weight, with no significant difference in overall body weights.
Thin sections (1 pm) of the livers were examined by light microscopy. A large number of the hepatocytes in transgenic mouse livers contained an abundant quantity of lipid droplets within the cytosol of widely varying size (Fig. 8B). The lipid droplets were in the cytosol and were not associated with any intracellular structures. Examination of several fields of liver sections (data not shown) revealed that up to half of the hepatocytes in the liver were lipid-laden. While the lipid-packed hepatocytes often appeared in clusters, this irregular distribution was not associated notably with any liver structural features or with the circulatory system of the organ. In contrast, nontransgenic hepatocytes contained few lipid droplets, and they were uniformly small in size (Fig. 8 A ) . The extent of fatty liver development correlated with apoC-I11 expression, being less advanced in the low expresser 621 line and the intermediate-expresser 614 line than in the high expresser 638 line (data not shown).

Effects of Overexpression of
Binding of d < 1.006 g/ml Lipoproteins to the LDL Receptor-To investigate the possibility that apoB48 remnants accumulate in apoC-111-transgenic mice due to insufficient apoE, which is required for their receptor-mediated clearance from plasma, the ability of isolated d < 1.006 g/ml lipoproteins to compete with normal LDL for binding to the LDL receptor on human fibroblasts was determined (Fig. 9). Lipoproteins of the d < 1.006 g/ml fraction from normal mouse plasma competed effectively with 1251-labeled LDL. However, the d < 1.006 g/ml lipoproteins from high expresser apoC-111-transgenic mice were poor competitors with 12%labeled LDL for receptor binding. While the addition of purified human apoE to normal d 1.006 g/ml lipoproteins increased their binding activity, a striking increase in the ability of transgenic d < 1.006 g/ml lipoproteins to bind the LDL receptor occurred when apoE was added. To achieve a 50% inhibition in receptor binding, only 2 pg/ml of the apoE-supplemented transgenic lipoproteins was required, compared to >40 pg/ml for the native transgenic d < 1.006 g/ml lipoproteins. Zn Vivo Clearance of Chylomicron Remnants from the Plasma-To determine if the rate of clearance of apoB48 particles in vivo was impaired, radioactively labeled canine chylomicron remnants were injected intravenously into normal and transgenic mice, and their disappearance from plasma was followed over a 20-min period (Fig. 10, left panel). The uptake of radioactivity by the liver and peripheral tissues was determined 20 min after injection (Fig. 10, right panel). The chylomicron remnants were cleared at a reduced rate in the high expresser apoC-111-transgenic mice compared to normal mice. After 20 min, 40% of the radioactivity remained in the plasma of transgenic mice, whereas only 16% of the radioactivity remained in the plasma of normal mice. The extent of chylomicron remnant removal from the plasma was reflected in the liver uptake of the label. At the 20-min time point, 70% of the injected dose was recovered in the liver of normal mice, while only 37% of the injected radiolabel was found in the liver of transgenic mice. In addition, 8.8% of the radioactivity was found in the spleen of normal mice, compared to 5.5% in transgenic mice; ~3 % of the injected dose was detected in the kidney in both nontransgenic and transgenic mice.
Overexpression of ApoE in Dansgenic Mice-Transgenic mice that overexpressed human apoE were generated using an apoE3 gene construct, HEG.LE2 (241, that contained 5 kb of 5"flanking sequence, the intact apoE gene, 1.7 kb of 3'-flanking sequence, and the hepatic control region of the apoE/C-I gene locus (33). Mice that were heterozygous for the transgene expressed human apoE in the plasma at levels of 25-35 mg/dl, with no apparent reduction in the level o f circulating mouse apoE, which was estimated to be 3-5 mg/dl (25). A preliminary characterization of these mice revealed that they had more than a 2-fold reduction in total plasma triglyceride levels, which was consistent with an equivalent reduction in the d < 1.006 g/ml f r a~t i o n .~ Administration of labeled canine chylomicrons to the apoE overexpressers revealed an increase in the plasma clearance rate that correlated with a corresponding increase in the uptake of the label by the liver (Fig. 11). Thus, transgenic human apoE is likely to have equilibrated rapidly with the exogenous chylomicrons, resulting in an increase in the rate of clearance of the remnants.
Cross-breeding of ApoE Overexpresser and ApoC-ZZZ Overexpresser Dansgenic Mice-To determine the effect of increased plasma apoE on the hypertriglyceridemia resulting from excess apoc-111 production, heterozygous transgenic mice that overexpressed human apoE3 were mated with heterozygous transgenic mice that overexpressed apoC-111 at -18 mg/dl (low expressers, apoC-IIIL) or at -130 mg/dl (high expressers, apoC-IIIH). Total plasma or the d < 1.10 g/ml plasma lipoprotein fraction was isolated from the resulting littermates (nontransgenic mice, apoE overexpressers, apoC-111 overexpressers, and combined apoE/C-I11 overexpressers) and analyzed by gel electrophoresis (Fig. 12, upper panel 1. The amounts of apoB48 in the apoE overexpressers and in the combined apoE/C-IIIH transgenic littermates were similar to those of the nontransgenic littermates, which were substantially lower than those found in apoC-IIIH overexpressers. These results were confirmed by immunoblot analysis of total plasma using apoB antibodies (Fig. 12, lower panel), examining plasma from the apoC-IIIL and apoE/C-IIIL littermates.
The effect of increased apoE production on plasma triglyceride levels in both apoC-111 high and low expressers that had been mated with apoE overexpressers was dramatic (Fig. 13). A marked decrease in plasma triglyceride levels was obtained in both cross-bred lines. Nontransgenic littermates and apoE transgenic mice had triglyceride levels of 77 40 mg/dl and 50 * 35 mg/dl, respectively. Transgenic littermates overexpressing only apoC-I11 had 10-fold (apoC-IIIL) and 20-fold (apoC-IIIH) elevations in plasma triglycerides. In contrast, the combined apoE/C-IIIL transgenic mice had plasma triglyceride values of only 96 2 65 mg/dl, and the apoE/C-IIIH transgenic mice had a 9-fold decrease in plasma triglycerides compared to the apoC-IIIH mice. The increased plasma cholesterol levels of the apoC-I11 high expresser line was abolished by apoE overexpression.

DISCUSSION
The overexpression of apoC-111 in transgenic mice results in moderate to severe hypertriglyceridemia, with most of the excess triglyceride accumulated in the VLDL fraction of the plasma. This observation in the ICR outbred mouse agrees with the finding of hypertriglyceridemia in the hybrid C57BU6 x A. 6. CBA mouse reported by Breslow et al. (18,19). In both strains of mice, the degree of hyperlipidemia correlates positively with the level of apoC-I11 in the plasma. The ICR mouse seems especially susceptible to the effects of apoC-I11 overexpression. The male ICR mice had higher levels of both triglyceride and cholesterol than the females, and this difference was signifi- in the liver, and the other two lines expressed the transgene in both liver and intestine, but no qualitative difference was apparent in the effects of apoC-I11 overexpression in any of the lines that was attributable to the site of synthesis. We found that the increase in plasma triglyceride in apoC-I11 overexpressers was the result of an accumulation of apoB48 remnants in the d < 1.006 g/ml lipoprotein fraction, with up to a 5-fold increase in the amount of apoB48. No increase in apoB100-containing lipoproteins was evident in this fraction.
While the d < 1.006 g/ml lipoproteins were only -15% larger in diameter in transgenic mice than in normal mice, their volume was increased by -40%. Thus, the increase in lipoprotein volume together with the increase in the quantity of apoB48 remnant particles accounted for the elevation in plasma triglyceride. In comparison to our results, Aalto-Setala et al. (19) did not observe a selective accumulation of apoB48 particles in apoc- The apoB48 lipoproteins in apoC-111-transgenic ICR mice were characterized by relatively high levels of apoC-111, which may have interfered with the acquisition of apoE during circulation. Therefore, the individual lipoproteins in the transgenic d < 1.006 g/ml density fraction were relatively deficient in their content of apoE, even though this fraction as a whole contained slightly more apoE, in comparison to nontransgenic mice. Thus, the ratio of apoE:apoB48 was greatly decreased in the apoB48 remnants as a consequence of excess apoC-111. Because apoE is required for the clearance of chylomicron remnants from plasma (34, 3 9 , i t seems likely that a deficiency of apoE on these particles would lead to their accumulation, resulting in the observed hyperlipidemia.
Excess plasma apoC-I11 might inhibit lipoprotein lipase or  interfere with the interaction of apoC-I1 and lipoprotein lipase, resulting in a n inhibition of plasma triglyceride hydrolysis that would lead to hypertriglyceridemia. However, a lack of fasting chylomicronemia was observed. In addition, the ability of transgenic d < 1.006 g/ml lipoproteins to function as a substrate for lipoprotein lipase in vitro and the lack of accumulation of very large d < 1.006 g/ml particles suggest that impaired lipolysis is not a primary mechanism by which apoC-I11 overexpression causes hypertriglyceridemia. Furthermore, this mechanism would not explain the preferential accumulation of apoB48 remnants with no increase in apoBlOO remnants. In addition, the nonesterified fatty acid level in transgenic mice correlated directly with the triglyceride content of the plasma, suggesting that the available lipase activity was sufficient to accommodate the increase in lipoprotein substrate.
The potential involvement of impaired hepatic lipase activity as a contributing factor in the hypertriglyceridemia of apoC-111-transgenic mice cannot be ruled out. If excess apoC-I11 influences apoE distribution among the plasma lipoproteins, the function of hepatic lipase could be affected, since apoE appears to modulate the activity of this enzyme (36). An impairment in hepatic lipase activity might delay the processing of chylomicron and VLDL remnants, as well as decrease the rates of conversion of IDL to LDL and HDL2 to HDL3 (37-40). However, .heparin administration to transgenic mice resulted in a rapid reduction in plasma triglycerides, indicating the presence of active enzyme (Table 111). Furthermore, no evidence suggests that hepatic lipase acts preferentially on either apoBlOO remnants or apoB48 remnants. The diameters ford < 1.006 dm1 lipoproteins from ICR mice were about 30% less than those of VLDL isolated from C57BIJ6 x CBA mice (19). This variance between the two strains may be a function of differences in plasma lipoprotein metabolism or in lipid metabolism in the liver. Regarding this last possibility, apoC-I11 transgenic ICR mice developed fatty livers. In contrast, the apoC-111-transgenic C57BIJ6 x CBA hybrids were not reported to have fatty livers, although they had up to a 1.7-fold increase in circulating nonesterified fatty acids (19). The composition of HDL also varied between the two strains: the dense HDL from ICR mice contained apoE, whereas dense HDL from the hybrid strain had no apoE (19). Since significant differences were observed in lipoprotein metabolism between the two strains of mice, both qualitative and quantitative differences in the effects of apoC-I11 overexpression might be expected, as observed.
Studies with Triton WR-1339 suggested that the rate of production of triglyceride by the liver was increased in transgenic mice compared to normal mice. This increased synthetic rate may be a consequence of the higher amounts of nonesterified fatty acids in the plasma of apoC-111-transgenic mice, which were detected at levels up to 5-fold greater than in normal mouse plasma. Greater amounts of nonesterified fatty acids delivered to the liver through the circulation might stimulate triglyceride and VLDL biosynthesis, as well as increased storage of lipid by the liver (41,42). The apparent increase in the rate of triglyceride synthesis in apoC-I11 transgenic mice may be a consequence and not a cause of the reduced clearance and increased accumulation of triglyceride-rich lipoproteins. Furthermore, the selective accumulation of apoB48 remnants that occurs in apoC-111-transgenic mice argues that an increase in triglyceride synthesis by the liver is not the primary defect responsible for the observed hyperlipidemia in these animals. However, the increase in triglyceride production by the liver may contribute to the hypertriglyceridemia manifested by apoC-I11 overexpression.
The hypertriglyceridemia in apoC-111-transgenic ICR mice is accompanied by changes in HDL. Plasma HDL cholesterol levels and HDL-associated apoA-I levels were decreased slightly. While plasma apoA-I levels were unchanged quantitatively, a modest redistribution of apoA-I from HDL to the d < 1.006 dm1 fraction occurred.
Large HDLl in normal mice extend to d = 1.02 g/ml, and they contain mainly apoE. Apolipoprotein E-rich HDLl are postulated to play a key role in the delivery of cholesterol to peripheral tissues via an apoE-mediated process (43, 44). The HDLl in apoC-111-transgenic mice extend to d = 1.006 dml, and in addition to apoE, they contain large amounts of human apoC-111. The increased apoC-1II:apoE ratio may interfere with the apoE-mediated clearance of these particles from the plasma. Conversely, as suggested by the in vitro data of Kinnunen and Enholm (171, the high concentration of apoC-I11 may interfere with hepatic lipase activity, causing these particles to increase in size and float at lower densities.
The primary defect caused by the overexpression of apoC-I11 is most likely to be impaired lipoprotein clearance due to the relative deficiency of apoE on apoB48 remnants. This deficiency would be expected to reduce the rate of clearance of these lipoproteins, leading to hypertriglyceridemia. Excess apoC-I11 also appears to interfere with the clearance of HDL1. The present study provides direct evidence for this mechanism to account for the hyperlipidemia observed in apoC-111transgenic mice. This mechanism is supported by a decreased apoE:apoB48 ratio in the d c 1.006 dm1 fraction, defective binding of transgenic d c 1.006 dm1 lipoproteins to the LDL receptor, and an impaired rate of clearance of chylomicron remnants. The addition of purified apoE to the d < 1.006 g/ml particles from transgenic mice enhanced their LDL receptorbinding activity in cultured fibroblasts. More conclusively, when apoE was introduced into the apoC-111-transgenic mice by cross-breeding apoE overexpressers with apoC-I11 overexpressers, the levels of apoB48 and triglyceride in plasma were observed to be normal. These results suggest that excess plasma apoC-I11 interferes with the apoE-mediated clearance of lipoproteins. The finding that cross-breeding with apoE overexpressers corrected the hyperlipidemia caused by apoC-I11 overexpression argues that the deficiency of apoE in the accumulating apoB48 remnants is responsible for the observed hypertriglyceridemia.
How does excess apoC-I11 affect the apoE-mediated clearance of lipoproteins from the plasma? As first suggested by Havel and co-workers (8)(9)(10) in their early perfusion studies of lipoprotein clearance in rat liver, apoE simply may be displaced from the lipoprotein surface by excess apoC-111. In the transgenic mouse model, excess apoC-I11 may prevent sufficient apoE from associating with apoB48 remnants and HDLl to permit clearance at a rate necessary to maintain a proper metabolic balance in plasma. Alternatively, apoC-I11 may interfere, directly or indirectly, with the interaction of apoE and the receptods) mediating the clearance of these particles, or it may interfere with the interaction of apoE and hepatic lipase. The absolute amount of apoE or the ratio of apoC-I11 to apoE on a given lipoprotein particle may be a key determinant in the uptake of lipoproteins by the liver. Although the mechanism remains to be resolved, our studies suggest that apoC-I11 functions as a natural modulator of apoE-mediated clearance processes in vivo. Our findings and those ofAalto-Setala et al. (19) suggest that abnormalities in apoC-I11 expression may contribute to some forms of hyperlipidemia.