Identification of Differentially Regulated Genes in Mildly Hyperlipidemic ApoE3-Leiden Mice by Use of Serial Analysis of Gene Expression
Although genes determining lipoprotein homeostasis and atherosclerosis are the subject of intensive investigation, only a subset of these genes is known at present. Hence, we do not have sufficient knowledge to explain the genetic basis of hyperlipidemia in the majority of subjects. Our aim was to identify novel genes and pathways underlying lipoprotein homeostasis by using serial analysis of gene expression. The liver expression profile of mild hyperlipidemic apolipoprotein E3-Leiden (E3L) transgenic mice was compared with that of the wild-type C57BL/6JIco (B6) mice. Over 18 000 liver transcripts of B6 as well as E3L mice were analyzed, representing >9400 unique genes. One hundred seventy-five genes showed altered expression between the strains (P<0.05). Although several of these genes belonged to known metabolic pathways, such as lipoprotein metabolism, detoxification processes, glycolysis, and the acute-phase response, most were novel. Differential gene expression of 8 of 10 genes tested could be confirmed by Northern blot analysis. This inventory of differentially expressed genes will provide a unique basis for detailed studies to gain more insight into their role in lipoprotein homeostasis and atherosclerosis.
Atherosclerosis is a disease of the large arteries and is the primary cause of heart disease and stroke. Epidemiological studies have revealed numerous genetic and environmental risk factors.1 The genetic risk involves multiple interacting genes, which contribute to the complexity of the disease. One of the main risk factors is an elevated level of atherogenic plasma lipoproteins (hyperlipidemia).1 At present, several genetic defects associated with impaired lipoprotein metabolism have been described. However, most of the genes remain elusive, and it is still not possible to explain the majority of (mildly) hyperlipidemic cases.2 Knowledge of novel genes may suggest new strategies to limit hyperlipidemia and the development of atherosclerosis.
Studying genetic factors involved in hyperlipidemia is difficult in humans because the hyperlipidemic phenotype of an individual is the result of a complex interaction of a heterogeneous genetic background and environmental challenges, such as smoking, diet, and exercise.1 To facilitate the study of these factors under well-defined genetic and environmental conditions, the transgenic apoE3-Leiden (E3L) mouse model has been developed. E3L mice carry the human apoE3-Leiden and apoC1 gene cluster on an inbred C57BL/6J (B6) background. In humans, the apoE3-Leiden mutation is associated with a dominantly inherited form of familial dysbetalipoproteinemia.3 Additionally, E3L mice have defects comparable to those of patients with familial dysbetalipoproteinemia in the uptake of chylomicron and VLDL remnants by the liver, resulting in high plasma lipid levels. They show a profound diet-induced hyperlipidemia and susceptibility to atherosclerosis.4 It is likely that the expression of many genes will be altered in the E3L mouse and will contribute and/or respond to the hyperlipidemic phenotype. Hence, E3L mice serve as a useful model to identify genes and pathways associated with lipoprotein homeostasis and atherosclerosis. To identify these genes and pathways, serial analysis of gene expression (SAGE) was applied to the livers of B6 and E3L mice on standard chow diets.
SAGE is a powerful technique that allows a detailed analysis of thousands of transcripts simultaneously, without prior knowledge of potentially involved genes. It has the unique property of identifying novel expressed genes and providing quantitative information on the abundance of genes.5 In SAGE, short cDNA sequence tags (14 bp) are isolated from mRNA ≈260 bases from the poly-A+ tail. Tags are ligated to long concatamers, cloned, and sequenced. The frequency of each tag reflects transcript abundance, and tags are specific enough to uniquely identify the corresponding transcript. Thus far, many SAGE transcript profiles have been reported, but only a few concern atherosclerosis.6,7
The present study provides an accurate inventory of gene expression in the livers of B6 and E3L mice on standard diets with the use of SAGE. In addition, analysis of >18 000 transcripts per strain revealed altered expression of 175 genes in a comparison between B6 and E3L mice. A number of these genes are known to be involved in a variety of metabolic processes. However, the major part of these genes consisted of novel genes. Further characterization of these known and novel genes will provide more insight in their biological function and facilitate elucidation of metabolic pathways involved in hyperlipidemia.
Transgenic E3L mice (line 2), expressing the human apoE3-Leiden and human apoC1 genes, have been described previously.8 Ten female mice of the N16th generation (>99% on a C57BL/6JIco [B6] genetic background) and 10 female B6 mice, aged 10 weeks, were used in the experiments. Mice were given a standard mouse diet (Chow, Hope Farms) and housed under standard conditions in conventional cages with free access to water and food.
Measurement of Serum Lipids
Approximately 200 μL of blood was collected from each individual mouse in a Microvette tube (Sarstedt) through tail bleeding, after a 4-hour fasting period. Total serum cholesterol (kit No.236691, Boehringer-Mannheim), triglycerides without free glycerol (kit No. 337-B, Sigma Chemical Co), free fatty acids (FFAs, NEFA-C kit, WAKO Chemicals), and ketone bodies (β-hydroxybutyrate, kit No. 310-A, Sigma) were measured enzymatically.
The day after bleeding, mice were euthanized, and the total liver was removed and immediately deep-frozen in liquid nitrogen and stored at −80°C. A small portion was kept behind and fixed in 4% neutral-buffered formalin, processed, and embedded in paraffin. Sections (3 μm) were stained with hematoxylin-phloxine-saffron and examined microscopically.
Serial Analysis of Gene Expression
Total RNA from the caudate liver lobe was isolated by using the Instapure method according to the manufacturer’s protocol (Eurogentec). RNA from each group of 10 animals was pooled. Subsequently, poly-A+ RNA was isolated by using an Oligotex mRNA Midi/Maxi kit (Qiagen). Poly-A+ RNA (5 μg) was used in the SAGE procedure, which was performed essentially as described previously, with some modifications.5 Poly-A+ RNA was converted to first-strand cDNA primed from the bound oligo(dT)20, followed by second-strand cDNA synthesis (GIBCO-BRL). cDNA was cleaved with the restriction enzyme NlaIII, and 3′ cDNA fragments were bound to streptavidin-coated magnetic beads (Dynabeads M-280, Dynal) and divided into 2 pools. Each pool was ligated to a linker containing a recognition site for the tagging enzyme BsmFI. After ligation, SAGE tags were released by digestion with BsmFI, blunted, combined, and ligated to form ditags. A dilution (1:100) of the ligation mixture was used as a template in a polymerase chain reaction (PCR). PCR products of 200 separate reactions (100 μL) were pooled. After PAGE, the ditag-linker product (102 bp) was purified from gel. To separate ditags from linkers, the product was digested with NlaIII. After precipitation, the ditags were stabilized by the addition of 50 mmol/L NaCl before PAGE. The produced ditags (28 bp) were purified from gel and were ligated to concatamers. Before electrophoresis, concatamers were heated 10 minutes at 65°C, preventing contamination of larger fragments with aggregates of smaller ditags.9 After PAGE, gel regions between 400 and 800 bp were excised, and purified concatamers were subsequently cloned in the SphI site of pZero (Invitrogen). Clones containing an expected minimum of 28 tags were selected for sequence analysis. Sequencing was performed by using the Big Dye Primer Kit (Perkin Elmer) and analyzed on a 377 ABI automated sequencer (Perkin Elmer). Sequence files were analyzed, and statistical analysis of the data was performed by the use of SAGE software (Monte Carlo test), version 3.04.5
Confirmation of Differences in Expression Levels
Total RNA (10 μg) from the individual livers of B6 or E3L mice (same as in SAGE) was run on a 1% (wt/vol) agarose-formaldehyde gel and blotted onto Hybond-N nylon membranes, according to the manufacturer’s protocol (Amersham). Probes of genes or expressed sequence tags (ESTs) corresponding to SAGE tags were generated by reverse-transcription PCR on liver cDNA. PCR products were purified by using a gel extraction kit (Qiagen). PCR conditions were as follows: 5 minutes at 94°C and 30 cycles for 30 seconds at 94°C, 30 seconds at 59°C, 30 seconds at 72°C, and 7 minutes at 72°C. Hybridizations with 32P-labeled PCR products, labeled by use of a random priming labeling kit (GIBCO-BRL), were performed as described.10 Hybridized and washed blots were analyzed by using a PhosphorImager (Molecular Dynamics), and hybridization signals were quantified by using Image Quant (Molecular Dynamics).
Lipid Levels and Histology of the Liver
On a standard mouse diet (chow), E3L mice, compared with B6 mice, were mildly hyperlipidemic. Moreover, the β-oxidation of FFAs by the liver is significantly increased in E3L mice. Compared with B6 mice, E3L mice showed a significant increase in total serum cholesterol, total triglycerides, FFAs, and ketone body levels (marker for the rate of β-oxidation); see Table 1. The E3L liver is characterized by some minor steatosis (triglyceride deposition) and the characteristic E3L protein bodies.11 The presence of the E3L protein bodies result in some more multifocal infiltration (by lymphocytes and macrophages) that is due to single-cell necrosis. However, the cellular morphology is highly comparable in the 2 strains: the majority of liver cells consist of hepatocytes and Kupffer cells, whereas the multifocal infiltrate constitutes only a very small part. The highly similar cellular morphology allows comparison of the gene expression profiles of livers from the 2 strains.
Gene Expression Profile of the Liver
Expression profiles from the livers of B6 and E3L mice were obtained by using SAGE. The B6 library consisted of a total of 18 861 sequenced tags, whereas the E3L library consisted of 19 219 sequenced tags; after removal of the tags representing linker sequences, 18 743 and 18 423 tags, respectively, remained. The number of unique tags, each representing a different gene, was highly similar in both libraries (Figure 1A). The number of unique genes identified was directly proportional to the number of tags sequenced, with a total of 5804 unique genes in the B6 library and 5719 unique genes in the E3L library. Combining data sets of the 2 libraries resulted in 9427 unique genes (data not shown), of which 40% corresponded with a known mouse sequence from GenBank (December 2000).
Twenty percent of the tags occurred >2 times, whereas the vast majority of tags (80%) occurred once or twice (Figure 1B). The most abundant genes in the liver were plasma protein serum albumin (4.5%) and apoE (2.6%); see Table 2. Other highly abundant genes were transcripts for plasma proteins involved in binding and transport, protease inhibitors, complement factors, and genes involved in several metabolic processes, such as the synthesis of lipoprotein particles, glycolysis, gluconeogenesis, and detoxification processes. For a complete overview, please see Table I (which can be accessed online at www.ahajournals.org).
Comparison of Liver Expression Profiles Between E3L and B6 Mice
Of the 9427 unique genes, 175 genes (tags) were differentially expressed (P<0.05). Twenty-nine percent of these gave a hit with a known mouse sequence in GenBank, 15% had an EST retrieved from the mouse Unigene map, and 56% were unknown (Table 3). For the complete data set, please Table II (which can be accessed online at www.ahajournals.org).
The majority of differentially expressed genes were upregulated (21%) or downregulated (37%) in the range of 1- to 5-fold in E3L mice. However, a small percentage of genes was at least 20-fold upregulated (3%) or downregulated (2%); see Figure 2. As expected, the human apoE3-Leiden and apoC1 transgenes were highly represented in the E3L liver, with 345 and 1045 tags, respectively, and they were completely absent in the B6 liver (Table 3). Another tag displaying a large difference in expression was matched with an EST homologous to apoA-II (ApoAII), an apolipoprotein involved in HDL metabolism (Table 3). Additionally, among the differentially expressed genes, some were known to be involved in lipoprotein homeostasis, such as apoA1 (ApoAI), vitronectin, liver fatty acid–binding protein (L-fabp), and ceruloplasmin (Cp); see Table 3. Moreover, differentially expressed genes were involved in all metabolic processes as described in the previous section. However, the majority of these genes consisted of novel transcripts with unknown biological functions.
Confirmation of Differential Gene Expression Detected by SAGE
A selection of differentially expressed genes or ESTs detected by SAGE was validated by Northern blot analysis (Table 3, underlined). Selection of these genes was based on involvement in lipid metabolism and/or large differences in tag abundance. mRNA levels of the selected genes were correlated with mRNA levels of cyclophilin, which showed no difference in tag abundance (Table 2). Differential expression of α-tubulin isotype 4 (≈7-fold), vitronectin (1.5-fold), an EST homologous to cytochrome P-450 steroid-inducible 3a11 (25-fold), prothymosin β4 (≈7-fold), an EST matching ApoAII (≈24-fold), and the human apoE3-Leiden (≈354-fold) and apoC1 (≈1045-fold) transgenes could be confirmed by Northern blot analysis. For the extensive data set, please see Figure I (which can be accessed online at www.ahajournals.org). One gene, cytochrome P-450 naphthalene hydroxylase (≈8-fold), showed no significant difference in mRNA abundance; however, a trend was visible in the direction indicated by SAGE (online Figure I). However, in the case of ornithine aminotransferase (Oat) and ApoAI, the difference in expression detected by SAGE could not be confirmed by Northern blot analysis (data not shown).
To learn more about genetic factors and metabolic pathways linked to (mild) hyperlipidemia, liver expression profiles of B6 and E3L mice on standard diets were generated by SAGE. Compared with other techniques, such as microarraying,12,13 SAGE has the unique advantage of identifying novel genes and elucidating the absolute levels of gene expression. In the present study, a total of 37 166 tags were obtained, providing insight in the expression of >9427 unique highly/moderately abundant genes in the liver. As expected, the apoE3-Leiden and apoC1 transgenes were most abundant in the E3L liver and completely absent in the B6 liver. The majority of highly abundant genes encode for various plasma proteins, such as serum albumin, haptoglobin, and ferritin, and for several apolipoproteins, apoA-I, apoA-II, apoC-I, apoC-III, and apoE, in both strains. Other highly expressed genes in the mouse liver were genes encoding protease inhibitors, cytoplasmic proteins, complement factors, coagulation factors, and enzymes involved in several metabolic processes, such as amino acid metabolism, lipid metabolism, glycolysis, and detoxification processes.
To our knowledge, the present study is the first describing a SAGE expression profile of the mouse liver. Interestingly, a human SAGE expression profile of a normal liver has been described, although that study was limited to only 1 SAGE profile and included no comparisons with other SAGE profiles.14 Yamashita et al14 detected 8597 unique genes and showed that the same functional classes as described in the present study for mice were also highly expressed in the normal human liver.
The differences in expression levels from 8 of 10 genes detected by SAGE could be confirmed by Northern blot analysis. These genes were selected on the basis of involvement in lipid metabolism and/or large differences in tag abundance. Biases in the observed results may be due to differences between the techniques. SAGE and Northern blot analysis are complementary techniques, PCR-based and hybridization-based, respectively, each with its own specific sensitivity and consequences for quantifying differences in expression levels. This becomes most clear when validating the human apoE3-Leiden and apoC1 transgenes by Northern blot analysis. SAGE revealed 354 tags for apoE3-Leiden and 1045 tags for apoC1 in the E3L liver and, as expected, no tags in the B6 liver. Subsequently, Northern blot analysis revealed smaller differences in expression for the transgenes, giving an indication of the difference in sensitivity between the techniques. However, the difference indicated by Northern blot analysis for the human apoE3-Leiden transgene is underestimated because of some cross-hybridization between human apoE and endogenous mouse ApoE. This demonstrates the sensitivity of Northern blot analysis for cross-hybridization between homologous genes. In some cases, Northern blot analysis does not allow us to distinguish between members of (large) gene families. Other biases may be caused by sequencing errors and the lack of uniqueness and randomness in tag sequences.15
Interestingly, a tag displaying a large difference in expression between E3L versus B6 was matched with an EST homologous to ApoAII, an apolipoprotein involved in HDL metabolism. Obtaining more sequence of this EST may reveal a novel gene with possible involvement in lipoprotein metabolism. Moreover, other differentially expressed genes were known to determine lipoprotein homeostasis and atherosclerosis, such as Cp and L-fabp. Cp, a copper-binding plasma protein, was more highly expressed in E3L mice than in B6 mice. Cp has been reported to play a role in LDL oxidation and in the progression of atherosclerosis.16 Therefore, elevated levels of Cp may be associated with enhanced atherosclerosis susceptibility in E3L mice.
Additionally, upregulation of L-fabp in E3L mice may be the result of elevated serum levels and increased β-oxidation of FFAs by the liver.17 FFAs induce gene expression of L-fabp through the peroxisome proliferator activated receptor-α (PPARα), a nuclear receptor.18 Activation of PPARα by FFA results in an increased uptake and β-oxidation of FFAs by the liver. Hence, these actions of PPARα enhance degradation of FFA-derived inflammatory mediators.19
Interestingly, clustering by function of differentially expressed genes revealed several inflammatory/acute-phase (AP) response genes. The AP response is the physiological response of the liver to infections and injuries involving local inflammation and initiation of events leading to a systemic response.20 The AP response induces changes in the concentration of specific plasma proteins, which help to protect the host from further injury and facilitate the repair process. Levels of positive AP proteins (serum amyloid A, α1-antitrypsin, complement C3, haptoglobin, and ceruloplasmin) increase during the AP response, whereas levels of negative AP proteins (serum albumin, L-fabp, and apoA1) decrease.20 However, most of these AP response genes were altered in a manner different from that during an AP response (complement C3, α1-antitrypsin, haptoglobin, serum albumin, and L-fabp). PPARα may control the expression of these AP response genes; it was shown that PPARα inhibits the activation of inflammatory response genes by negatively interfering with the nuclear factor-κB, the signal transducer(s) and activator(s) of transcription, and the activator protein-1 signaling pathways.19,21 Further studies should provide more insight in the biological relevance of the differentially expressed AP response genes and other known and novel genes detected by SAGE.
To identify these novel genes, more sequence of the unknown tags should be obtained. For this purpose, several techniques have been developed, such as PCR by rapid amplification of sequenced tags and the generation of longer cDNA fragments from SAGE tags for gene identification.22,23 Identification of novel genes is a major advantage of SAGE compared with other techniques, such as microarray, in which only previously identified transcripts are analyzed.12,13
Studying the expression levels of the differentially regulated genes in microarray experiments under different conditions (eg, diets and mouse models) and during different time points may reveal most interesting candidate genes and unravel the metabolic pathways involved. Integration of these data with quantitative trait loci associated with dyslipidemic phenotypes will facilitate the identification of new candidate genes for hyperlipidemia. In conclusion, expression profiles have been obtained from the livers of E3L and B6 mice by use of SAGE, resulting in a large number of differentially expressed genes linked to hyperlipidemia. Strikingly, most of these genes have not been previously associated with lipoprotein metabolism. They represented a whole spectrum of metabolic processes or were novel. Knowledge of these genes may provide more insight into basic molecular mechanisms modulated by perturbations in lipid metabolism and may suggest potential leads for therapeutic intervention.
This project was supported by a grant from the Dutch Organization for Scientific Research (project NWO 98010001). M.H. Hofker is an Established Investigator of the Dutch Heart Association (project NHS D95022). C.J.A. Moen is a research fellow of the Dutch Organization for Scientific Research (project NWO 903-39-174). We thank K. Willems van Dijk and T. Kooistra for critical reading of the manuscript.
Received February 28, 2001; revision accepted August 24, 2001.
Havel RS, Kane JP. Introduction: structure and metabolism of plasma lipoproteins: familial hypercholesterolemia.In: The Metabolic Basis of Inherited Disease. 7th ed. New York, NY: McGraw-Hill; 1981: 2030.
van Vlijmen BJ, van den Maagdenberg AM, Gijbels MJ, van der Boom H, HogenEsch H, Frants RR, Hofker MH, Havekes LM. Diet-induced hyperlipoproteinemia and atherosclerosis in apolipoprotein E3-Leiden transgenic mice. J Clin Invest. 1994; 4: 1403–1410.
Velculescu VE, Zhang L, Vogelstein B, Kinzler KW. Serial analysis of gene expression. Science. 1995; 270: 484–487.
Suzuki T, Hashimoto Si, Toyoda N, Nagai S, Yamazaki N, Dong HY, Sakai J, Yamashita T, Nukiwa T, Matsushima K. Comprehensive gene expression profile of LPS-stimulated human monocytes by SAGE. Blood. 2000; 96: 2584–2591.
van den Maagdenberg AM, Hofker MH, Krimpenfort PJ, de Bruijn I, van Vlijmen B, van der Boom H, Havekes LM, Frants RR. Transgenic mice carrying the apolipoprotein E3-Leiden gene exhibit hyperlipoproteinemia. J Biol Chem. 1993; 268: 10540–10545.
Kenzelmann M, Muhlemann K. Substantially enhanced cloning efficiency of SAGE (serial analysis of gene expression) by adding a heating step to the original protocol. Nucleic Acids Res. 1999; 27: 917–918.
Anonymous. Current Protocols in Molecular Biology. New York, NY: J Wiley & Sons; 1994.
Mensenkamp AR, van Luyn MJ, van Goor H, Bloks V, Apostel F, Greeve J, Hofker MH, Jong MC, van Vlijmen BJ, Havekes LM, et al. Hepatic lipid accumulation, altered very low density lipoprotein formation and apolipoprotein E deposition in apolipoprotein E3-Leiden transgenic mice. J Hepatol. 2000; 33: 189–198.
Schena M, Shalon D, Davis RW, Brown PO. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science. 1995; 270: 467–470.
Stollberg J, Urschitz J, Urban Z, Boyd CD. A quantitative evaluation of SAGE. Genome Res. 2000; 10: 1241–1248.
Jensen LE, Whitehead AS. Regulation of serum amyloid A protein expression during the acute-phase response. Biochem J. 1998; 334: 489–503.
Chen JJ, Rowley JD, Wang SM. Generation of longer cDNA fragments from serial analysis of gene expression tags for gene identification. Proc Natl Acad Sci U S A. 2000; 97: 349–353.