The Expression of Intact and Mutant Human apoAI/CIII/AIV/AV Gene Cluster in Transgenic Mice*

The apoAI/CIII/AIV gene cluster is involved in lipid metabolism and has a complex pattern of gene expression modulated by a common regulatory element, the apoCIII enhancer. A new member of this cluster, apolipoprotein (apo) AV, has recently been discovered as a novel modifier in triglyceride metabolism. To determine the expression of all four apo genes in combination and, most importantly, whether the transcription of apoAV is coregulated by the apoCIII enhancer in the cluster, we generated an intact transgenic line carrying the 116-kb human apoAI/CIII/AIV/AV gene cluster and a mutant transgenic line in which the apoCIII enhancer was deleted from the 116-kb structure. We demonstrated that the apoCIII enhancer regulated hepatic and intestinal apoAI, apoCIII, and apoAIV expression; however, it did not direct the newly identified apoAV in the cluster. Furthermore, human apo genes displayed integrated position-independent expression and a closer approximation of copy number-dependent expression in the intact transgenic mice. Because apoCIII and apoAV play opposite roles in triglyceride homeostasis, we analyzed the lipid profiles in our transgenic mice to assess the effects of human apoAI gene cluster expression on lipid metabolism. The triglyceride level was elevated in intact transgenic mice but decreased in mutant ones compared with nontransgenic mice. In addition, the expression of human apoAI and apoAIV elevated high density lipoprotein cholesterol in transgenic mice fed an atherogenic diet. In conclusion, our studies with human apoAI/CIII/AIV/AV gene cluster transgenic models showed that the apoCIII enhancer regulated expression of apoAI, apo-CIII, and apoAIV but not apoAV in vivo and showed the influences of expression of the entire cluster on lipid metabolism.

apoAI is the major protein component of HDL. apoAI levels correlate positively with HDL cholesterol and negatively with atherosclerotic cardiovascular diseases (3). Transgenic mice overexpressing human apoAI are protected from diet-related or apoE deficiency-related atherosclerotic lesions (4,5). apoCIII is the major component of VLDL and a minor component of HDL, and mice carrying human apoCIII develop severe hypertriglyceridemia (6,7). HDL is a major carrier of plasma apoAIV, and overexpression of human apoAIV decreases aortic lesions in transgenic mice (8,9). apoAV is found mainly in HDL and VLDL, and human apoAV transgenic mice display ϳ30% triglyceride level compared with wild-type mice (2). In humans, several important single-nucleotide polymorphisms within the apoAI cluster genes have been shown to be strongly associated with dyslipidema and to increase susceptibility to atherosclerosis (10,11). Recently, such research has focused on association analysis of these single-nucleotide polymorphisms in the whole cluster from general population (12,13) and familial combined hyperlipidemia (14,15).
Besides its important functions, the apoAI cluster is a useful model with which to study the transcriptional regulation and tissue-specific expression of clustered genes. The apo genes in this cluster are expressed at different levels in the liver and intestine: apoAI is expressed at high levels in the liver and intestine, apoCIII synthesis occurs predominantly in the liver and to a much lesser extent in the intestine, and apoAIV is expressed mainly in the intestine and to a lesser extent in the liver (16), whereas apoAV is expressed only in the liver (2). Previous transgenic mice carrying one or two apo genes displayed different expression levels, and these short genome fragments may lack much of the genetic information contained in the large flanking sequence for proper expression of the transgenes (16). With the increasing interest in research on the gene cluster, more and more important cis-acting elements have been identified to be involved in the transcriptional regulation of the gene cluster in the long distance, and this mechanism differs from that in a single gene (17). In the present study, we generated transgenic mice carrying a 116-kb intact genomic fragment containing the entire human apoAI, apoCIII, apoAIV, and apoAV to investigate the expression pattern of these genes in the cluster containing complete flanking sequences in vivo.
Previous studies have shown that a common regulatory element, the apoCIII enhancer, located 490 -890 bp upstream of apoCIII, regulates tissue-specific expression of apoAI, apoCIII, * This work was supported by Grant TG2000056902 from the Major State Basic Research Development Program of China (973 Program). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Both authors contributed equally to this work. § To whom correspondence should be addressed. Tel.: 8610-65296415; Fax: 8610-65133086; E-mail: liudp@pumc.edu.cn. 1 The abbreviations used are: apo, apolipoprotein; BAC, bacterial artificial chromosome; FISH, fluorescence in situ hybridization; RPA, RNase protection analysis; HDL, high density lipoprotein; VLDL, very low density lipoprotein; LDL, low density lipoprotein; IDL, inter-and apoAIV in vitro and in vivo (16) (Fig. 1). Although the proximal promoter of apoAV has been studied, transcriptional regulation of this newly discovered gene is largely unknown (18,19). Whether apoAV transcription is also regulated by the apoCIII enhancer or by some other element is an interesting question, the answer to which will deepen understanding of the mechanism underlying transcriptional regulation of the apoAI gene cluster. To study these issues, we created a mutant transgenic mouse line containing the 116-kb human apoAI gene cluster in which the apoCIII enhancer core region was deleted.
Transgenic lines containing a 33-kb human apoAI/CIII/AIV gene cluster construct were recently generated, in which the expression of transgenes induced hyperlipidema and reduced atherosclerosis (20). This model carried only a relatively short genomic fragment and did not contain the newly identified apoAV. Because apoCIII and apoAV have predominant but opposite roles in triglyceride homeostasis, to explore the relationship between these two genes and altered triglyceride, Baroukh et al. (21) very recently established another apoAI/CIII/ AIV/AV transgenic mouse line and found no differences in triglyceride concentrations between the transgenic and control mice (21). Here we investigate the action of apoCIII and apoAV in triglyceride metabolism via our intact and mutant apoAI gene cluster transgenic mice. Furthermore, we used these models to observe the action of apoAI clustered genes as an integrated unit in the change of lipid profiles induced by atherogenic diets.

EXPERIMENTAL PROCEDURES
Human Bacterial Artificial Chromosome (BAC) Library Screening and Fluorescence in Situ Hybridization (FISH) Analysis-High density hybrid membranes for the BAC library (Research Genetics, Huntsville, AL) were screened by [␣-32 P]dCTP-labeled apoAI probe (PCR from human genome; primers: hapoAI-exon-1, 5Ј-GACCTGCAAGCCTGCA-GCACT-3Ј; hapoAI-exon-2, 5Ј-CCGATGGTTGGCTCCTAGGTT-3Ј). Selected BAC DNA was purified by alkaline lysis and digested with different restriction endonucleases including NotI, KpnI, SnaBI, SalI, and XholI. The digested BAC DNA was separated by pulse field gel electrophoresis on 1% agarose (Roche Applied Science), using the 10 -250-kb autoprogram of CHEF MAPPER (Bio-Rad). Usually, SnaBI digestion of BAC DNA was transferred to nylon membranes for Southern blot analysis. A physical map of the BAC clone was drawn according to the pulse field gel electrophoresis and Southern blot results. The two terminals of selected positive BAC clone, 286I4, were sequenced, and FISH analysis of the BAC clone was performed as described previously (22).
Deletion of the apoCIII Enhancer from BAC DNA-Mutant BAC clones were derived from the wild-type 286I4 BAC clone by temperature-sensitive shuttle vector-mediated homologous recombination. Briefly, two fragments (F1 and F2) located at the two terminals of the target sequences were directly cloned by PCR from BAC 286I4 (primers: 5Ј-ACCAGGACTGATTCGCTCGTT-3Ј (FP1) and 5Ј-TGTTAGAGGCTC-CTTCTGCCT-3Ј (RP1) for F1; 5Ј-GGAGCCACTGATGCCTGGTCT-3 (FP2) and 5Ј-TTACCTGGAGCAGCTGCCTCT-3Ј (RP2) for F2), and the length of each fragment was more than 500 bp. The two fragments were cloned together into the multi-clone sites of constructive vector EGFP to produce plasmid pEGFP-(F1ϩF2). The (F1ϩF2) box was then released by SalI and XhoI digestion and inserted into the SalI linearized shuttle vector pSV-RecA. The recombinant shuttle vector pSV-RecA-(F1ϩF2) was used to transform Escherichia coli DH10B, which harbors a wild-type 286I4 BAC clone. After two rounds of heat-induced homologous recombination, mutant BAC clones were screened by positive (Tet) and counter (FA) selections and verified by Southern blotting with amplified F1 as probe and by PCR with primers FP1 and RP2 (Fig. 2).
DNA Purification for Microinjection and Generation of Transgenic Mice-BAC DNA was purified by routine alkaline lysis and CsCl gradient ultracentrifugation. After NotI digestion, the linearized BAC DNA was added to a 0.5 ϫ 5 cm pre-equilibrized Sepharose CL-4B column and collected as described previously (22). The purified intact and mutant BAC 286I4 DNA was diluted to 1.2 ng/l for pronuclear microinjection into murine zygotes derived from C57BL/6J-mated KM female mice using previously described procedures (22).
Detection of Transgenic Mice-The transgenic founders were screened by PCR using the apoAI primers 5Ј-GACCTGCAAGCCTGC-AGCACT-3Ј and 5Ј-CCGATGGTTGGCTCCTAGGTT-3Ј, which span a 560-bp fragment containing human apoAI exon 1 and its upstream region. Identification was further confirmed by Southern blotting of mouse tail genomic DNA after PstI digestion and hybridization with three different probes: the human apoAI far downstream sequence (181382-181706 in the human genomic sequence; GenBank TM accession number AC007707), human apoCIII gene (118941-119333 in Gen-Bank TM accession number AC007707), and human apoAV upstream (75412-75853 in GenBank TM accession number AC007707). The copy numbers of the founder mice were estimated by densitometric scanning of autoradiograms using a PhosphorImager (Amersham Biosciences). Positive founder F 0 mice were outbred to generate lines of heterozygous mice.
RNA Preparation and Analysis-Tissue samples (liver, small intestine, brain, heart, lung, stomach, spleen, kidney, and muscle) from 4 -6-week-old mice were collected. Total RNA was extracted from these samples by using TRIzol reagent (Invitrogen), dissolved in diethyl pyrocarbonate-treated water and used for RNase protection analysis (RPA) of human apoAI, apoCIII, apoAIV, and apoAV mRNA. In brief, 10 g of total RNA were hybridized with [␣-32 P]UTP-labeled probes (1 ϫ 10 5 cpm) in hybridization solution at 55°C overnight, followed by RNase A (8 g/ml) and RNase T1 (10 units/l) digestion at 25°C for 30 min and proteinase K treatment. The protected fragments were purified by water-saturated phenol/chloroform and electrophoresed on a 6% polyacrylamide-7 M urea gel. The gel was exposed to an x-ray film for 24 -48 h, and the band intensity was quantified by a scanning densitometer (Amersham Biosciences).
Animals and Diets-Transgenic mice containing the intact and mutant 116-kb apoAI gene cluster were used. Nontransgenic littermates were used as controls. Five-week-old transgenic mice and nontransgenic littermates were housed in a temperature-controlled room with alternating 12-h light (7 a.m. to 7 p.m.) and dark (7 p.m. to 7 a.m.) periods. Animals were kept on a regular mouse chow. Male and female mice were used in approximately equal ratio in all experiments. For blood sampling, mice were fasted overnight and bled the next morning from the retro-orbital plexus while under anesthesia. Blood was collected into a tube containing 6 l of EDTA (0.5 M), separated by centrifugation at 2000 ϫ g for 15 min, and kept at 4°C until analysis.
In an atherogenic food experiment, the positive F 1 mice of intact apoAI gene cluster transgenic line 4, mutant transgenic line 4 mice, and nontransgenic littermates had free access to an atherogenic diet (40% basic food, 15% fat, 1.25% cholesterol, 0.5% sodium cholate) or regular mouse chow for 20 weeks, and then plasma lipids were analyzed.
Plasma Lipoprotein and Lipid Analysis-The individual lipoprotein fractions (VLDL, d Ͻ 1.006 g/ml; IDL ϩ LDL, d ϭ 1.006 -1.063 g/ml; FIG. 1. Scheme of the human apoAI/CIII/AIV/AV gene cluster. Human apoAI, apoCIII, apoAIV, and apoAV genes are arrayed in a cluster on chromosome 11q23, and the transcriptional orientation of apoCIII is reversed to the others. The distal regulatory region of the apoCIII enhancer (Ϫ490/Ϫ890) acts as a common regulatory element for apoAI, apoCIII, and apoAIV in the cluster.
HDL, d ϭ 1.063-1.21 g/ml) were isolated from pooled serum of at least six mice of intact transgenic line 4 and six mice of mutant transgenic line 4, respectively. Human and wild-type mouse serum was used as a positive and negative control. Isolation was achieved by sequential ultracentrifugation at the respective densities at 42,000 rpm in a CP 70 MX ultracentrifuge equipped with a P42AT 0313 rotor (Hitachi) for 3, 12, and 16 h at 10°C. Each lipoprotein fraction was analyzed for human apolipoprotein composition by 3-15% sequential SDS-PAGE. Proteins were transferred to nitrocellulose membranes (Schleicher & Schuell) followed by incubation with polyclonal goat antibody against human apoAI, apoCIII, apoAIV (Santa Cruz Biotechnology), and apoAV (a gift from Dr. Robert O. Ryan and Michael N. Oda), respectively. With regard to apoCIII, the total plasma (5 l) of transgenic mice was also used for Western blot. The antibodies directed against human apolipoproteins were highly specific and did not show any cross-species reactivity. Rabbit anti-goat IgG (Amersham Biosciences) was used as a secondary antibody.
The positive F 1 mice of intact transgenic mouse line 4 and mutant line 4 were selected to analyze lipid profile after consumption of atherogenic and chow diets. Triglyceride concentrations were determined by an enzymatic glycerol oxidase-p-aminophenazone method; total and HDL cholesterol concentrations were determined by enzymatic cholesterol oxidase-p-aminophenazone methods with commercial kits (Beijing Zhongsheng High-Tech company, Beijing, People's Republic of China). All analysis was performed on the HITACHI 7060 Automatic Analyzer (Japan).
Statistical Analysis-Results are reported as means Ϯ S.E. In all experiments, nontransgenic littermates were used as controls. An unpaired Student's t test was used to compare values in different groups. Statistical significance was defined as p Ͻ 0.05.

Screening and Identification of BAC 286I4 and Mutant
Clone-After SnaBI digestion and pulse field gel electrophoresis of BAC DNA, the approximate sizes of the selected clones were estimated by comparison with size markers of multimers of bacteriophage (New England Biolabs) (data not shown). Clone 286I4 is 116-kb long and contains all of the human apoAI, apoCIII, apoAIV, and apoAV genes. Two terminal sequencing further proved that 286I4 was the fragment from 72491-188971 in the human genomic sequence (GenBank TM accession number AC007707). FISH analysis showed that the hybridization signals were located on human chromosome 11q23. No other signal was detected in the FISH results (Fig. 3).
Using temperature-sensitive shuttle vector-mediated homologous recombination, we identified a mutant 286I4 clone. Southern blotting results indicated that a ϳ300-bp core sequence (Ϫ491/Ϫ793 to apoCIII) of the apoCIII enhancer was deleted in the mutant clone (Fig. 2, lane 1). Further sequencing results of the PCR product verified that the core sequence of the apoCIII enhancer was deleted.
Creation of Intact Human apoAI/CIIII/AIV/AV Transgenic Mice-Four founder mice were identified by PCR and further verified by Southern blot. They all had the same hybrid bands with the human genomic DNA control (Fig. 4), which confirmed the structural integration of the human apoAI gene cluster. Compared with human genomic DNA, the copy numbers of transgenes in mice lines T1, T2, T3, and T4 were 1, 2, 3, and 2, respectively (Fig. 4).
Tissue-specific Expression of apoAI, apoCIII, apoAIV, and apoAV in Intact apoAI Gene Cluster Transgenic Mice-Human apo gene expression in the transgenic mouse lines was determined by RPA. The human apoAI clustered genes were primarily expressed in the liver and intestine of transgenic mice and were hardly detected in the kidney, brain, heart, lung, stomach, spleen, and skeletal muscle (data not shown). Human apoAI is expressed mainly in the liver and to a much lesser extent in the intestine of transgenic mice. Human apoCIII expression was high in the liver and moderate in the intestine. apoAIV is expressed mainly in the intestine and to a lesser extent in the liver. Consistent with previous findings, apoAV expression was detected only in the liver (shown as T1, T2, T3, and T4 in Figs. 5 and 6) (2). The expression pattern of the four apo genes in transgenic mice is similar to that of the human in vivo (2,16).
Besides the expression of apoAI and apoAIV in transgenic mouse line 1 (T1 in Fig. 9a), each transgene in the apoAI cluster transgenic mice showed Ͻ3-fold variation in expression per transgene copy (T1, T2, T3 and T4 in Fig. 9). Normalized apoAI, apoCIII, apoAIV, and apoAV specific values were plotted against the gene copy number of the transgene for each animal. This indicated that the expression of apoAI clustered genes showed a closer approximation of copy number dependence (Fig. 7).
Creation of Mutant Human apoAI/CIIII/AIV/AV Transgenic Mice-PCR and Southern blot verified that five stable mutant transgenic mouse lines were generated by linear mutant 286I4 BAC DNA microinjection. All mutant transgenic mouse lines had the same hybrid bands as the human genomic DNA control. Compared with human genomic DNA, the copy numbers for each strain of the five founders were 3, 20, 1, 2, and 6, respectively (Fig. 8).
Tissue-specific Expression of apoAI, apoCIII, apoAIV, and apoAV in Mutant apoAI Gene Cluster Transgenic Mice-No apoAI clustered gene expression was detected in tissues other than the liver and intestine in mutant transgenic mice (data not shown). Hepatic expression of human apoAI was abolished in mutant transgenic mouse lines 3 and 4 and was reduced significantly in lines 1, 2, and 5. The expression of human apoCIII and apoAIV was also reduced remarkably in the liver of mutant transgenic mice (M1-M5 in Figs. 5, 6, and 9).
However, hepatic expression of human apoAV in mutant mice was not different from that in the intact apoAI cluster transgenic mice (Figs. 5, 6, and 9). This suggested that deletion of the apoCIII enhancer did not influence hepatic apoAV expression in the cluster.
Plasma Apolipoproteins and Lipids in the Transgenic Mice-The distribution of human apolipoproteins among various lipoprotein classes in the plasma of transgenic mice was analyzed by Western blot. As positive control (human plasma), human apoAI and apoAIV were present mainly in HDL (d ϭ 1.063-1.21 g/ml) particles, and human apoAV was present in HDL and VLDL particles of intact and mutant transgenic mice (Fig. 10a). Human apoCIII was detected in the whole plasma of transgenic mice. All four human apolipoproteins were not detected in the lipoproteins of the negative control, wild-type mice plasma (Fig. 10). Table I presents plasma lipid profiles in the intact human apoAI cluster transgenic mice (line 4), mutant apoAI cluster transgenic mice (line 4), and nontransgenic littermates. Plasma triglyceride levels in the intact apoAI cluster transgenic mice increased ϳ3-fold compared with control animals (296 Ϯ 127 versus 81 Ϯ 15, p Ͻ 0.05), whereas in the mutant transgenic mice, the level decreased to 70% compared with control mice (57 Ϯ 12 versus 81 Ϯ 15, p Ͻ 0.05). Total cholesterol in the intact transgenic mice increased about 60% (141 Ϯ 31 versus 87 Ϯ 14, p Ͻ 0.05), mainly due to an increase in the non-HDL subclass, whereas no significant change was observed in mutant transgenic mice compared with control animals (90 Ϯ 15 versus 87 Ϯ 14) (Table. 1).

Effects of Atherogenic Diets on apo Gene Expression and
Lipid Profile in Transgenic Mice-Total cholesterol was elevated by 36% in the intact human apoAI gene cluster transgenic mice (from 141 Ϯ 31 to 192 Ϯ 84, p Ͻ 0.05), 66% in the mutant transgenic mice (from 90 Ϯ 15 to 149 Ϯ 24, p Ͻ 0.05), and 44% (from 87 Ϯ 14 to 125 Ϯ 32, p Ͻ 0.05) in control littermates when these three groups of mice were fed the atherogenic diet for 20 weeks, respectively. The atherogenic diet significantly increased the concentration of HDL cholesterol (from 73 Ϯ 21 to 104 Ϯ 31, p Ͻ 0.05) in the intact transgenic mice and mutant transgenic mice (from 67 Ϯ 16 to 92 Ϯ 14, p Ͻ 0.05), whereas no significant change was observed in HDL cholesterol level in control littermates (Table I).
The expression of human apo genes in the liver and intestine of intact transgenic mice was determined by RPA after these animals were fed the atherogenic diet. The atherogenic diet did not seem to influence hepatic expression of human apoAI, apoCIII, apoAIV, and apoAV, whereas up-regulated intestinal expression of human apoCIII and apoAIV was seen in transgenic mice fed the atherogenic diet as compared with those fed the chow diet (Fig. 11).

DISCUSSION
Transgenic animals carrying large genomic fragments provide more complete and accurate information than animals harboring short constructs, and this is especially important for studying clustered genes. In this study, we screened the human BAC library and obtained a 116-kb clone containing the whole human apoAI/CIII/AIV/AV gene cluster. BAC DNA was suc-  9. Comparison of expression of human apoAI, apoCIII, apoAIV, and apoAV per copy number in intact and mutant apoAI cluster transgenic mice. Human apoAI-, apoCIII-, apoAIV-, and apoAV-specific signals and mouse G3PDH-specific signals in different exposures of the experiment shown in Figs. 5 and 6 were subjected to densitometric scanning. The resulting values for the apo-specific signals were then normalized against the values for the G3PDH-specific signals, and these ratios of transgene to endogenous control expression were then divided by the relative copy number to obtain the values presented. Means were obtained from RNA preparations from five different mice for each transgenic mouse line. T, intact apoAI cluster transgenic mice; M, mutant apoAI cluster transgenic nice. a, hepatic expression of apoAI clustered genes in transgenic mice; b, intestinal expression of apoAI clustered genes in transgenic mice.
cessfully introduced into the mouse genome, and four stable transgenic mouse lines containing intact human apoAI/CIII/ AIV/AV cluster constructs were generated. In contrast to the studies on transgenic mice carrying a single apo gene or short 33-kb apoAI/CIII/AIV cluster constructs (20, 23-25), the apo gene expression exhibited an integrated position-independent pattern in our intact apoAI cluster transgenic mice. Moreover, the transgenes showed a closer approximation of copy numberdependent expression in our transgenic mice (Fig. 7). These results suggested that there might exist some cis-acting elements in the flank region that play a major role in protecting apoAI genes from strong regulatory influences of neighboring mouse DNA. Such influences include the well-known insulator or matrix attachment region, which can define a spatial region and therefore constitutes the functional domain for this locus (26,27).
The tissue-specific expression patterns of apoAI, apoCIII, apoAIV, and apoAV in our intact apoAI cluster transgenic mice were similar to those of humans, but the intestinal apoAI expression level was very low in the intact transgenic mice compared with the relatively high apoAI mRNA levels in human and mouse intestine (2,16,28,29). This apoAI expression level is similar to those in the report of transgenic mice carrying a 33-kb human apoAI/CIII/AIV in which intestinal apoAI mRNA did not exceed 20% of liver apoAI mRNA (20). However, in apoAI/CIII transgenic mice carrying only the apoCIII enhancer, 2.1-kb 5Ј region and 1.6-kb 3Ј region of the human apoAI gene, the apoAI expression level in the intestine was almost as high as that seen in the liver (23). From the abovementioned results, we hypothesize that this discrepancy may be caused by a different reaction to the 5Ј-flanking sequence up to 2.1 kb or the 3Ј-flanking sequence up to 1.6 kb of apoAI in humans and mice. Moreover, overexpression of other apo genes in the cluster and subsequent lipid profile alterations might another possible reason. In addition, human apoAIV was ex-  pressed mainly in the intestine and to a lesser extent in the liver of transgenic mice. This observation is consistent with the tissue-specific expression of apoAIV in humans and in human apoAIV gene transgenic mice (30). In contrast, apoAIV was expressed mainly in the liver and at relatively low levels in the intestine in 33-kb human apoAI/CIII/AIV cluster transgenic mice that had the same 3Ј-flanking sequence as the human apoAIV gene transgenic mouse (20). Excluding possible differences in methods used to measure human apoAIV mRNA and the physiological variability of apoAIV mRNA levels, we think this discrepancy is worth noting, and it suggests that our large fragment used to establish the transgenic mice recapitulates the endogenous expression pattern more authentically.
Previous in vitro and in vivo studies have proved that the apoCIII enhancer, as a common regulatory element, regulates tissue-specific expression of apoAI, apoCIII, and apoAIV (16). In order to gain further insight into the role of this common enhancer in the whole cluster, we engineered five stable mutant transgenic mouse lines that carried the whole human apoAI cluster construct but in which the apoCIII enhancer was deleted. We found that deleting the apoCIII enhancer almost abolished intestinal expression of human apoAI and apoCIII, as compared with the expression level of genes in intact gene cluster transgenic mice. The enhancer-deleted mice also showed a significant decrease in hepatic expression of human apoAI, apoCIII, and apoAIV and intestinal expression of human apoAIV, respectively (Figs. 5, 6, and 9). These results are in accord with studies using transgenic lines for short mutant human apoAI, apoCIII, and apoAIV constructs (23)(24)(25). In contrast with other apo genes in the cluster, hepatic expression of human apoAV in all mutant transgenic mice was not influenced by deletion of the apoCIII enhancer, providing evidence that the newly identified apoAV is not regulated by the apoCIII enhancer in the cluster (Figs. 5 and 9). This conclusion was further supported by the results from lipid profile analysis in the transgenic mice: plasma triglyceride in our mutant apoAI cluster transgenic mice decreased to 70% compared with nontransgenic mice (57 Ϯ 12 versus 81 Ϯ 15, p Ͻ 0.05) ( Table I). apoCIII and apoAV have predominant but opposite roles in plasma triglyceride: compared with nontransgenic controls, the triglyceride level in human apoCIII transgenic mice increases to 200 -2000% but decreases to 30% in human apoAV transgenic mice (2,6). In our mutant transgenic mouse, human apoAV overexpression was not influenced, whereas human apoCIII expression was nearly abolished after deletion of the apoCIII enhancer, which led to the fall of triglyceride level (apoAI and apoAIV were not considered here because these genes do not seem to influence the plasma triglyceride in transgenic mice, respectively) (29,32). In contrast, in our intact apoAI gene cluster transgenic mice, the overexpression of both apoCIII and apoAV induced ϳ2-fold increase of plasma triglyceride (Table I). Our results revealed that the apoCIII enhancer was not a common enhancer for all four apo genes in the cluster and thus raised the following issue: is there any other element that performs coregulation between apoAV and the other three apo genes in the cluster?
To investigate the relationship of the apoCIII and apoAV genes, Baroukh et al. (21) recently generated independent transgenic lines that either overexpressed or completely lacked both these genes and proposed that apoAV and apoCIII independently influence plasma triglyceride concentrations but in an opposing manner. In our intact apoAI gene cluster transgenic lines, plasma triglyceride level increased ϳ3-fold compared with control mice. This value is higher than that from the apoAI/CIII/AIV/AV transgenic mouse generated by Baroukh et al. (21) but is less than that in single human apoCIII trans-genic mice (5-20-fold) (6) and in the 33-kb human apoAI/CIII/ AIV transgenic mice (4 -10-fold) (20). In addition, in our mutant transgenic mouse, deletion of the apoCIII enhancer did not affect human apoAV expression but significantly reduced human apoCIII expression, thus leading to a lower triglyceride level compared with control mice (Table I). Therefore, triglyceride levels in our intact and mutant transgenic mice further indicated that apoCIII and apoAV independently influence plasma triglyceride homeostasis in an opposing manner, which was supported by Baroukh et al. (21).
The total cholesterol concentration of the intact transgenic mice was significantly elevated, due mostly to an increase in non-HDL cholesterol, a result similar to those seen in the study of the 33-kb human apoAI/CIII/AIV transgenic mice (20) but different from those of studies of the human apoAI transgenic mice, which showed a significant increase in HDL cholesterol but not non-HDL cholesterol (6,29). Cholesterol level in our transgenic mice indicates the complex action of apolipoproteins in the cluster because expression of human apoCIII raised the total cholesterol, mainly due to an increase in non-HDL cholesterol (7,33). Total cholesterol did not change in human apoAIV or apoAV transgenic mice (2,8).
The atherogenic diet elevated total cholesterol levels in control mice as a result of an increase in non-HDL cholesterol fractions, whereas in intact transgenic mice, HDL cholesterol was significantly increased as a result of atherogenic diet feeding, and this change was similar to that seen in the single human apoAI transgenic line (29) and the human 33-kb apoAI/ CIII/AIV gene cluster transgenic mice (20) but different from that in another line of human apoAI transgenic mice (6) and human apoAIV transgenic mice (8). Human apoAI and apoAIV were detected mainly in the HDL particle of transgenic mice plasma (Fig. 10a), and these two well-known agents in promoting reverse cholesterol transport led to the elevation of HDL cholesterol after atherogenic diet feeding (31). This primarily suggested the anti-atherosclerosis property of our transgenic mice because the inverse association between HDL cholesterol and the risk of coronary heart disease is well known (3).
The atherogenic diet increased both HDL cholesterol and non-HDL cholesterol in mutant mice (Table I). This was different from the effect seen in intact transgenic mice and nontransgenic mice. We think that a low level of apoAI and apoAIV in the plasma of mutant transgenic mice may lead to these differences (Fig. 10).
The atherogenic diet did not seem to influence hepatic expression of human apoAI, apoCIII, and apoAIV, whereas upregulated intestinal expression of human apoCIII and apoAIV was seen in transgenic mice fed the atherogenic diet as compared with those fed the chow diet (Fig. 11). These expression changes were similar to the results of studies of the 33-kb apoAI/CIII/AIV transgenic mice (20). Moreover, our results firstly indicated that human apoAV expression did not seem to be influenced by the atherogenic diet (Fig. 11).
In conclusion, we generated an intact transgenic line carrying the 116-kb human apoAI/CIII/AIV/AV gene cluster and a mutant transgenic line in which the apoCIII enhancer was deleted from the 116-kb structure. Comparing expression of the apo genes in the cluster, we demonstrated that the apoCIII enhancer directed the hepatic and intestinal expression of apoAI, apoCIII, and apoAIV in vivo but did not regulate the transcription of apoAV. The triglyceride levels in the plasma of our transgenic mice indicated apoCIII and apoAV independently play opposite roles in plasma triglyceride homeostasis. Moreover, expression of human apoAI and apoAIV led to the elevation of HDL cholesterol in apoAI cluster transgenic mice fed with atherogenic diet. Thus we have generated useful mod-els with which to study apoAI, apoCIII, apoAIV, and apoAV and their effects on lipid metabolism as well as to explore mechanisms underlying transcriptional regulation of these genes in combination.