Cholesterol and Cholate Components of an Atherogenic Diet Induce Distinct Stages of Hepatic Inflammatory Gene Expression

Atherosclerosis in inbred mouse strains has been widely studied by using an atherogenic (Ath) diet containing cholesterol, cholic acid, and fat, but the effect of these components on gene expression has not been systematically examined. We employed DNA microarrays to interrogate gene expression levels in liver of C57BL/6J mice fed the following five diets: mouse chow, the Ath diet, or modified versions of the Ath diet in which either cholesterol, cholate, or fat were omitted. Dietary cholesterol and cholate produced discrete gene expression patterns. Cholesterol was required for induction of genes involved in acute inflammation, including three genes of the serum amyloid A family, three major histocompatibility class II antigen genes, and various cytokine-related genes. In contrast, cholate induced expression of genes involved in extracellular matrix deposition in hepatic fibrosis, including five collagen family members, collagen-interacting proteins, and connective tissue growth factor. The gene expression findings were confirmed by biochemical measurements showing that cholesterol was required for elevation of circulating serum amyloid A, and cholate was required for accumulation of collagen in the liver. The possibility that these gene expression changes are relevant to atherogenesis in C57BL/6J mice was supported by the observation that the closely related, yet atherosclerosis-resistant, C57BL/6ByJ strain was largely resistant to dietary induction of the inflammatory and fibrotic response genes. These results establish that cholesterol and cholate components of the Ath diet have distinct proatherogenic effects on gene expression and suggest a strategy to study the contribution of acute inflammatory response and fibrogenesis independently through dietary manipulation. Atherosclerosis in inbred mouse strains has been widely studied by using an atherogenic (Ath) diet containing cholesterol, cholic acid, and fat, but the effect of these components on gene expression has not been systematically examined. We employed DNA microarrays to interrogate gene expression levels in liver of C57BL/6J mice fed the following five diets: mouse chow, the Ath diet, or modified versions of the Ath diet in which either cholesterol, cholate, or fat were omitted. Dietary cholesterol and cholate produced discrete gene expression patterns. Cholesterol was required for induction of genes involved in acute inflammation, including three genes of the serum amyloid A family, three major histocompatibility class II antigen genes, and various cytokine-related genes. In contrast, cholate induced expression of genes involved in extracellular matrix deposition in hepatic fibrosis, including five collagen family members, collagen-interacting proteins, and connective tissue growth factor. The gene expression findings were confirmed by biochemical measurements showing that cholesterol was required for elevation of circulating serum amyloid A, and cholate was required for accumulation of collagen in the liver. The possibility that these gene expression changes are relevant to atherogenesis in C57BL/6J mice was supported by the observation that the closely related, yet atherosclerosis-resistant, C57BL/6ByJ strain was largely resistant to dietary induction of the inflammatory and fibrotic response genes. These results establish that cholesterol and cholate components of the Ath diet have distinct proatherogenic effects on gene expression and suggest a strategy to study the contribution of acute inflammatory response and fibrogenesis independently through dietary manipulation. The mouse has become established as a key animal model for studies of lipid metabolism and atherosclerosis, due to the development of techniques for genetic manipulation and tools for gene discovery in this species (reviewed in Refs. 1Knowles J.W. Maeda N. Arterioscler. Thromb. Vasc. Biol. 2000; 20: 2336-2345Crossref PubMed Scopus (130) Google Scholar, 2Fazio S. Linton M.F. Front. Biosci. 2001; 6: D515-D525Crossref PubMed Google Scholar, 3Reardon C.A. Getz G.S. Curr. Opin. Lipidol. 2001; 12: 167-173Crossref PubMed Scopus (68) Google Scholar, 4Sheth S.S. Deluna A. Allayee H. Lusis A.J. Curr. Opin. Lipidol. 2002; 13: 181-189Crossref PubMed Scopus (10) Google Scholar). The first studies to demonstrate that the mouse might provide a useful model for characterization of genetic factors affecting atherosclerosis susceptibility appeared more than 30 years ago. These studies surveyed several inbred laboratory mouse strains and demonstrated that some strains develop early atheromatous lesions when fed experimental diets. These diets contained high concentrations of cholesterol (5%) and fat (30%) supplemented either with cholic acid (2%) (5Thompson J.S. J. Atheroscler. Res. 1969; 10: 113-122Abstract Full Text PDF PubMed Scopus (35) Google Scholar) or fed in combination with irradiation treatments (6Vesselinovitch D. Wissler R.W. J. Atheroscler. Res. 1968; 8: 497-523Abstract Full Text PDF PubMed Scopus (32) Google Scholar). These early diets produced high mortality and were subsequently modified to reduce the concentrations of cholesterol (1.25%), fat (15%), and cholate (0.5%). By using this modified atherogenic (Ath) 1The abbreviations used are: Ath, atherogenic diet; LDL/VLDL, low density/very low density lipoprotein; HDL, high density lipoprotein; EST, expressed sequence tag; SAA, serum amyloid A; RT, reverse transcriptase. diet, Paigen et al. (7Paigen B. Morrow A. Brandon C. Mitchell D. Holmes P. Atherosclerosis. 1985; 57: 65-73Abstract Full Text PDF PubMed Scopus (490) Google Scholar, 8Paigen B. Mitchell D. Reue K. Morrow A. Lusis A.J. LeBoeuf R.C. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 3763-3767Crossref PubMed Scopus (229) Google Scholar) demonstrated that fatty streak lesion formation is reproducible within a strain, and that strains differ in their susceptibility. The C57BL/6J strain was among the most susceptible and has been extensively used as a model for dietinduced atherosclerosis. Although the Ath diet has been widely used to study atherogenesis in C57BL/6J and other mouse strains, the effects of the individual dietary components have not been well characterized. In susceptible mouse strains, the Ath diet produces an atherogenic lipoprotein profile and induces inflammatory gene expression in the liver. For example, C57BL/6J mice fed the Ath diet exhibit dramatically elevated low density/very low density lipoprotein (LDL/VLDL) and reduced high density lipoprotein (HDL) cholesterol levels (8Paigen B. Mitchell D. Reue K. Morrow A. Lusis A.J. LeBoeuf R.C. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 3763-3767Crossref PubMed Scopus (229) Google Scholar, 9LeBoeuf R.C. Doolittle M.H. Montcalm A. Martin D.C. Reue K. Lusis A.J. J. Lipid Res. 1990; 31: 91-101Abstract Full Text PDF PubMed Google Scholar). In this strain, the Ath diet also induces inflammatory and oxidative stress genes such as serum amyloid A, monocyte chemotactic protein-1, colony-stimulating factors, and heme oxygenase (10Liao F. Andalibi A. deBeer F.C. Fogelman A.M. Lusis A.J. J. Clin. Investig. 1993; 91: 2572-2579Crossref PubMed Scopus (253) Google Scholar). It is unclear at present which component(s) of the atherogenic diet produce the observed effects on lipoproteins and inflammation. Furthermore, previous work (11Nishina P.M. Verstuyft J. Paigen B. J. Lipid Res. 1990; 31: 859-869Abstract Full Text PDF PubMed Google Scholar, 12Nishina P.M. Lowe S. Verstuyft J. Naggert J.K. Kuypers F.A. Paigen B. J. Lipid Res. 1993; 34: 1413-1422Abstract Full Text PDF PubMed Google Scholar) indicates that reduction in the concentration of either cholesterol or cholate in the Ath diet decreases the rate of aortic lesion formation and that the two components differentially affect gallstone formation and lipid accumulation in liver. This suggests that cholesterol and cholate may have independent pro-atherogenic effects. To test this hypothesis, we undertook a systematic analysis of plasma lipid and gene expression changes that occur in response to the cholesterol, cholate, and fat components of the Ath diet. C57BL/6J mice were fed one of five diets: mouse chow, the Ath diet, or modified versions of the Ath diet in which either cholesterol, cholate, or fat were omitted. We examined plasma lipid profiles and used DNA microarrays to screen the response of more than 11,000 mouse genes and expressed sequence tags (ESTs). A comparison of gene expression levels across all five diets allowed the identification of genes that were activated or repressed specifically by dietary cholesterol, cholate, or fat. We identified more than 300 genes that were activated or repressed by one of the three diet components, with more than a quarter of those (89:316) exhibiting at least a 10-fold response to either cholesterol, cholate, or fat. Cholesterol and cholate were found to induce expression of genes involved in different aspects of the inflammatory response, with cholesterol being required for the acute inflammatory response, whereas cholate was responsible for activating genes associated with hepatic fibrosis. Biochemical measurements of representative proteins from the acute inflammatory and fibrotic responses confirmed the gene expression data. We further investigated the potential role of these gene expression changes in atherogenesis by examining their expression in an atherosclerosis-resistant substrain of C57BL/6 mice. The activation of both the inflammatory and fibrotic genes was dramatically attenuated in the resistant mice, suggesting that their response to the Ath diet at the level of gene transcription may be one mechanism contributing to their resistance to diet-induced atherosclerosis. Overall, our results establish that cholesterol and cholate components of the Ath diet have distinct proatherogenic effects on gene expression, which correlate with genetic differences in atherosclerosis susceptibility in two closely related C57BL/6 mouse strains. Mice and Diets—Male C57BL/6J and C57BL/6ByJ mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Mice were fed ad libitum Purina Mouse Chow 5001, or one of four experimental diets (Teklad Research Diets, Madison, WI) containing various combinations of cholesterol, sodium cholate, and fat (in the form of cocoa butter). The Ath diet (Teklad TD90221) contained (by weight) 75% Purina Mouse Chow, 7.5% cocoa butter, 1.25% cholesterol, and 0.5% sodium cholate. The other three diets were equivalent to the Ath diet, with the omission of cholesterol (TD98232), cholate (TD94059), or fat (TD98233). At 3 months of age, mice were housed individually and fed the specified diet for 3 weeks before harvesting tissue and blood. The care of the mice and all procedures used in this study were conducted in accordance with the National Institutes of Health animal care guidelines. Lipid Determinations—Blood was obtained after a 16-h fast. Enzymatic assays for total cholesterol, HDL cholesterol, unesterified cholesterol, triglyceride, and free fatty acids were performed using enzymatic assays (13Hedrick C.C. Castellani L.W. Warden C.H. Puppione D.L. Lusis A.J. J. Biol. Chem. 1993; 268: 20676-20682Abstract Full Text PDF PubMed Google Scholar). LDL/VLDL cholesterol levels were determined as the difference between total and HDL cholesterol levels. For hepatic cholesterol and triglyceride determinations, lipids were extracted from 100 mg of tissue as described (14Carr T.P. Andresen C.J. Rudel L.L. Clin. Biochem. 1993; 26: 39-42Crossref PubMed Scopus (484) Google Scholar). Oligonucleotide Microarray Hybridization—Liver samples comprising the identical lobe were harvested from each mouse and flash-frozen in liquid nitrogen. Total RNA was prepared from mouse liver using Trizol reagent (Invitrogen). Each microarray hybridization was performed using 10 μg of total liver RNA pooled from four mice. Oligonucleotide microarrays (MU11K) were from Affymetix (Santa Clara, CA) and contained representations of more than 11,000 full-length mouse genes and EST clusters. cRNA synthesis, hybridization, washing, and scanning were performed according to standard Affymetrix protocols. Fluorometric data were generated by Affymetrix software, and the gene chips were globally scaled to all the probe sets with an identical target intensity value. Transformation of the fluorescent signals into numerical values and filtering of the data were accomplished as described (15Kersten S. Mandard S. Escher P. Gonzalez F.J. Tafuri S. Desvergne B. Wahli W. FASEB J. 2001; 15: 1971-1978Crossref PubMed Scopus (180) Google Scholar). Only genes with an absolute expression level (expressed as average difference, or Avg Diff, value from Affymetrix software output) above a threshold of 30 were analyzed. Identification of genes that are activated or repressed by specific diet components was accomplished using Microsoft EXCEL. The full set of microarray data for strains C57BL/6J and C57BL/6ByJ is available in supplemental data online. mRNA Quantitation—Confirmation of mRNA expression differences observed on microarrays was performed by Northern blot and RT-PCR. Total liver RNA was isolated using the Trizol reagent (Invitrogen). Poly(A)+ RNA was prepared from total RNA using the Poly(A)Tract mRNA isolation system (Promega, Madison, WI) and 2 μg loaded per lane for Northern analysis. Hybridizations were performed as described (16Cohen R.D. Castellani L.W. Qiao J.-H. Van Lenten B.J. Lusis A.J. Reue K. J. Clin. Investig. 1997; 99: 1906-1916Crossref PubMed Scopus (149) Google Scholar) with cDNA probes generated by RT-PCR. RT-PCR was performed using 2 μg of total liver RNA (cDNA Cycle Kit, Invitrogen). Primer sequences for examples shown in Fig. 2 were as follows: Saa3-f, agagacatgtggcgagcctac, and Saa3-r, cagcacattgggatgtttagg; W34845-f, gccaggccttcacctttcag, and W34845-r, acagttcagtcacccttacaag; Col3a1-f, cccatgactgtcccacgtaag, and Col3a1-r, cagggccaatgtccacaccaa; Mup1-f, ggcatactattatcctggcctc, and Mup1-r, gatggtggagtcctggtgaga; Igfbp-f, ttctcatctctctcgtacatg, and Igfbp-r, acgcagctttccacgttcag; and Gck-f, gtggccacaatgatctcctgc, and Gck-r, tcggcgacagagggtcgaaggc. Expression levels for activated hepatic stellate cell transcripts were examined by RT-PCR using previously published primer sets for platelet-derived growth factor receptor β, transforming growth factor β1, collagen 1α1, tissue inhibitor of metalloproteinases-1, α-smooth muscle actin (17Canbay A. Higuchi H. Bronk S.F. Taniai M. Sebo T.J. Gores G.J. Gastroenterology. 2002; 123: 1323-1330Abstract Full Text Full Text PDF PubMed Scopus (251) Google Scholar), and TATA box-binding protein (18Vergnes L. Phan J. Stolz A. Reue K. J. Lipid Res. 2003; 44: 503-511Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). Total RNA was treated with DNase (Ambion, Austin, TX) to remove contaminating genomic DNA, and cDNA was prepared from 2 μg of total liver RNA. 5% of the resulting cDNA sample was amplified for 32 cycles using a Touchdown protocol with a beginning annealing temperature of 63 °C and a final annealing temperature of 53 °C (18Vergnes L. Phan J. Stolz A. Reue K. J. Lipid Res. 2003; 44: 503-511Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). PCR products were analyzed by electrophoresis in agarose, and quantitation of digital images was performed using 1D Image Analysis software (Eastman Kodak Co.). Serum Amyloid A and Collagen Quantitation—Serum amyloid A levels in mouse plasma were determined by enzyme-linked immunosorbent assay (BioSource International, Camarillo, CA). Collagen concentration in liver was determined using the Sircol collagen assay (Accurate Chemicals, Westbury, NY). Briefly, 50 mg of liver was homogenized, and total acid pepsin-soluble collagens were extracted overnight using 5 mg/ml pepsin in 500 μl of 0.5 m acetic acid. One ml of Sircol dye reagent was added to 100 μl of each sample, in duplicate, and incubated at 25 °C for 30 min. After centrifugation, the pellet was suspended in 1 ml of alkali reagent, and absorbance was read at 540 nm. Effects of Ath Diet Components on Circulating Lipid Levels—To investigate the effect of specific components of the Ath diet on circulating lipid levels, C57BL/6J mice were fed a chow diet, the Ath diet, or one of three modified Ath diets in which either cholesterol, cholate, or fat was omitted (Table I). Compared with chow, the Ath diet produced more than a 2-fold increase in plasma cholesterol levels, due to an increase of more than 100 mg/dl in LDL/VLDL cholesterol (Fig. 1), as observed previously (9LeBoeuf R.C. Doolittle M.H. Montcalm A. Martin D.C. Reue K. Lusis A.J. J. Lipid Res. 1990; 31: 91-101Abstract Full Text PDF PubMed Google Scholar). In contrast to previous reports, HDL cholesterol levels were not significantly reduced on the Ath diet. This may be a result of using male, as opposed to female mice, and/or the shorter (3 week) duration of the diet compared with some previous studies. The amount of unesterified cholesterol more than doubled on the Ath diet, whereas triglyceride levels were reduced ∼50%, and free fatty acid levels were unchanged.Table IComposition of the five dietsChowAthNo cholateNo cholesterolNo fatCholate-+-++Cholesterol-++-+Fat-+++- Open table in a new tab The diets lacking cholesterol, cholate, and fat components each produced distinct plasma lipid profiles (see Fig. 1). The omission of fat from the Ath diet did not alter plasma cholesterol, triglyceride, or fatty acid levels compared with the complete Ath diet, indicating that the fat added to this diet has little effect on the circulating lipid levels. In contrast, omission of cholesterol prevented any significant increase in total cholesterol or unesterified cholesterol above the levels on a chow diet. The cholate-free diet produced elevated cholesterol levels that were intermediate between values on the chow and Ath diets and were significantly higher than the chow values. Although total cholesterol levels were not statistically different between Ath and no cholate diets, the distribution of cholesterol among LDL/VLDL and HDL fractions was dramatically affected by dietary cholate. LDL/VLDL cholesterol increased 10-fold on the Ath diet but only 2-fold on the cholate-free diet, whereas HDL cholesterol was at it highest on the cholate-free diet. The omission of cholesterol from the Ath diet also blunted the increase in LDL/VLDL cholesterol levels seen with the complete diet, indicating that cholesterol and cholate act synergistically to elicit the large increase in LDL/VLDL cholesterol that occurs on the Ath diet. A similar diet effect was observed with hepatic cholesterol levels. Thus, although hepatic cholesterol levels increased 6-fold on the Ath diet, this required the inclusion of both cholesterol and cholate. Triglyceride levels were suppressed about 2-fold on the Ath diet (Fig. 1). Omission of either cholate or cholesterol from the Ath diet produced a significant elevation in triglyceride levels above those seen on either chow or Ath diets, suggesting that the two components act together to effect the reduced triglyceride levels observed with the Ath diet. Hepatic triglyceride levels were also repressed on the Ath diet, but omission of cholate prevented this repression. Free fatty acid levels were not significantly affected by the Ath diet components. Thus, cholesterol and cholate components appear to have both independent effects and synergistic effects on plasma and hepatic lipid levels. Cholesterol and Cholate Induce Distinct Sets of Inflammatory Genes—To investigate gene expression changes underlying the diet-induced alterations in lipid levels, we performed microarray hybridization studies. Liver RNA from mice fed each of the five diets was hybridized to Affymetrix MU11K oligonucleotide microarrays to assess the relative expression levels of thousands of mouse genes and ESTs. Data were filtered to exclude signals below a defined threshold of absolute expression level and to ensure specificity of hybridization for perfect match versus mismatch oligonucleotides (see "Experimental Procedures") (15Kersten S. Mandard S. Escher P. Gonzalez F.J. Tafuri S. Desvergne B. Wahli W. FASEB J. 2001; 15: 1971-1978Crossref PubMed Scopus (180) Google Scholar). By using these criteria, 7200 of the 13,104 DNA elements on the array were scored as present in at least one diet sample. Comparison of gene expression profiles for the two most extreme diets, chow and Ath, revealed that the combination of cholesterol, cholate, and fat produces widespread changes in hepatic gene expression levels. 839 genes were activated, and 454 genes were repressed by at least 3-fold on the Ath diet compared with basal levels on the chow diet. A systematic comparison of expression levels among all five diets allowed the identification of genes that are regulated by specific dietary components. Approximately 1.4% of genes represented on the array exhibited altered expression specifically in response to cholesterol, cholate, or fat. By comparing the expression levels of genes across all five diets, we defined six gene expression patterns with respect to regulation by specific diet components: cholesterol-activated, cholesterol-repressed, cholate-activated, cholate-repressed, fat-activated, and fat-repressed. For example, "cholesterol-activated genes" were those having at least 2-fold higher levels on all three diets containing cholesterol (Ath, No Cholate, and No Fat diets) than on the two diets lacking cholesterol (Chow and No Cholesterol diets). Representative expression profiles for each of the six groups are shown in Fig. 2. A full list of genes activated or repressed by cholesterol, cholate, or fat is given in Table II, and a summary of each group is given below.Table IIGenes activated or repressed by cholesterol, cholate, or fat Cholesterol-regulated Genes—Expression of 38 of the genes assayed was altered by the presence of dietary cholesterol, including 25 genes that were activated and 13 repressed by cholesterol (Table II). The magnitude of activation by cholesterol was striking, with 30% of the cholesterol-activated genes showing more than 10-fold induction in response to cholesterol. An example of a cholesterol-activated gene, serum amyloid A3 (Saa3), showed equivalent high levels of expression on the complete Ath diet and the No Fat diet, and lowest levels on the No Cholesterol diet (Fig. 2a). The nearly identical expression levels on the Ath and No Fat diets indicate that fat had little effect on expression of this group of genes and also illustrate the consistency of gene expression measurements across individual microarray hybridizations. Most of these genes had expression levels on the No Cholate diet that were intermediate between those on Ath and the No Cholesterol diets. This suggests that whereas cholesterol has the strongest effect on expression of these genes, cholate plays an additive role with cholesterol to achieve the peak expression levels. Notable among the cholesterol-activated genes were 12 genes known to be involved in acute inflammation and the immune response: genes of the serum amyloid A (SAA) family (Saa2, Saa3, and Saa4), histocompatibility antigens (H2-1A-α, H2-1A-β, H2-1E-β, and Ia-associated invariant chain), and additional inflammation/immune-associated genes including interleukin-2 receptor γ (Il2rg), small inducible cytokine B9 (Scyb9), SAM domain and HD domain 1 (Samhd1), paired-Ig-like receptor A5 (Pira5), and galectin-3 (Lgals3) (Table II). SAA gene expression was induced from 7- to 8-fold (Saa2 and Saa4) to 37-fold (Saa3) on the Ath diet. Whereas omission of cholate from the diet diminished the response, cholesterol was absolutely critical for activation of SAA gene expression (Fig. 2a). Similar cholesterol requirements were observed for the other inflammation-related genes shown in Fig. 2a, although magnitude of expression was lower. Genes that were repressed by dietary cholesterol included aquaporin-8, which has been implicated in canalicular bile secretion in liver (19Huebert R.C. Splinter P.L. Garcia F. Marinelli R.A. LaRusso N.F. J. Biol. Chem. 2002; 277: 22710-22717Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar), Cyp17a1, a key enzyme in C-21 steroid biosynthesis, and apolipoprotein A-IV, a component of high density lipoproteins, that has been shown previously (20Baroukh N. Ostos M.A. Vergnes L. Recalde D. Staels B. Fruchart J. Ochoa A. Castro G. Zakin M.M. FEBS Lett. 2001; 502: 16-20Crossref PubMed Scopus (18) Google Scholar) to be repressed by the Ath diet. An additional 7 novel ESTs of unknown function were also repressed by cholesterol (for example see Fig. 2b). Cholate-regulated Genes—Of the three Ath diet components examined here, cholate affected expression levels of the greatest number of genes, with 81 genes induced and 23 repressed by cholate (Table II; see examples in Fig. 2, c and d). Most striking was the induction of numerous genes encoding collagen and non-collagen extracellular matrix components. The expression and excretion of extracellular matrix proteins is indicative of fibrogenesis, a wound healing process that occurs in response to inflammation induced by infectious or metabolic agents (21Neubauer K. Saile B. Ramadori G. Can. J. Gastroenterol. 2001; 15: 187-193Crossref PubMed Scopus (85) Google Scholar). Five collagen genes were activated up to 70-fold by the Ath diet: Col1a1 and Col1a2 (which encode procollagen, type I, subunits α1 and α2), Col3a1 (procollagen, type III, α1), Col4a1 (procollagen, type IV, α1), Col6a1 (procollagen, type VI, α1), and nidogen (a glycoprotein that binds type IV collagen). Omission of cholesterol from the diet had little effect on collagen gene expression, but omission of cholate prevented induction (Fig. 2c). Additional cholate-activated genes involved in extracellular matrix synthesis included nidogen and lumican, two proteins that have direct interactions with extracellular collagens, as well as vimentin, a cytoskeletal intermediate filament protein, and connective tissue growth factor (Ctgf), which modulates extracellular matrix secretion. Thus, dietary cholate appears to have a specific effect on activation of genes involved in the response to chronic inflammation. Several additional genes induced by cholate have recognized roles in lipid metabolism. For example, cholate induced phospholipid transfer protein (Pltp), which is involved in lipoprotein remodeling and has been shown recently to be regulated by bile acids through the farnesoid X-activated nuclear hormone receptor (22Urizar N.L. Dowhan D.H. Moore D.D. J. Biol. Chem. 2000; 275: 39313-39317Abstract Full Text Full Text PDF PubMed Scopus (208) Google Scholar), Also induced by cholate was liver X receptor β, an oxysterol-binding nuclear hormone receptor that activates several genes involved in cellular cholesterol efflux. Dietary cholate also induced choline kinase and lipocalin 2, genes involved in phospholipid synthesis and intracellular lipid transport, respectively. The list of genes repressed by cholate was less extensive than those activated by this component (Table II). These included chemokine orphan receptor 1, a choline/ethanolamine kinase (Chk1), a gene implicated in very long chain fatty acid elongation (Elovl3), and cytochrome P450, 7b1 (Cyp7b), a key enzyme in the alternate pathway of bile acid synthesis. Fat-regulated Genes—Of the Ath diet components, fat affected expression of the fewest genes, with 6 fat-activated and 9 fat-repressed genes identified by our criteria of at least 2-fold effects in response to fat across all five diets (Table II). The magnitude of expression of most of these genes was quite modest. A notable exception was insulin growth factor-like binding protein-1 (Igfbp1), which was expressed at high levels on the chow and no fat diets, but repressed on all diets containing fat (Fig. 2f). Confirmation of Independent Cholesterol and Cholate Effects on SAA and Fibrogenic Gene Expression—The data above demonstrated that cholesterol was required for the large induction of SAA gene expression, whereas cholate was required for induction of collagen gene expression. To confirm that these gene expression changes resulted in corresponding increases in protein levels, we quantitated SAA protein levels in blood and collagen levels in liver under the various diet conditions. In agreement with the mRNA expression results, SAA protein levels in the circulation increased about 30-fold on Ath compared with a chow diet. The same high SAA levels were present when cholate was omitted from the Ath diet, but omission of cholesterol prevented any increase above chow values (Fig. 3a). Analogously, hepatic collagen levels were elevated specifically in diets containing cholate, regardless of the other components (Fig. 3b). These results establish that the cholesterol- and cholate-specific changes in SAA and collagen gene expression give rise to altered protein levels as well. Additional confirmation that the fibrotic process is activated by dietary cholate was obtained by examining additional markers of fibrotic gene expression. The major source of collagen and other extracellular matrix proteins in liver fibrosis is hepatic stellate cells, a population of perisinusoidal cells comprising 15% of the resident liver cells (23Kmiec Z. Adv. Anat. Embryol. Cell Biol. 2001; 161: 1-151Crossref Google Scholar, 24Friedman S.L. J. Biol. Chem. 2000; 275: 2247-2250Abstract Full Text Full Text PDF PubMed Scopus (1893) Google Scholar). Hepatic stellate cells typically exist in a quiescent state, serving as the principal storage site for retinoids. In response to stimuli such as bacterial infection or inflammation, the stellate cells become activated and transform into proliferative, fibrogenic cells. To characterize further the fibrogenic response to dietary cholate, we examined established markers of activated stellate cells via RT-PCR (17Canbay A. Higuchi H. Bronk S.F. Taniai M. Sebo T.J. Gores G.J. Gastroenterology. 2002; 123: 1323-1330Abstract Full Text Full Text PDF PubMed Scopus (251) Google Scholar). The Ath diet increased expression of several hepatic stellate cell markers including platelet-derived growth factor β-receptor (Pdgfrb), tissue inhibitor of metalloproteinases-1 (Timp1), transforming growth factor β1(Tgfb1) (see Fig. 4), and α-smooth muscle actin (not shown). The induction of hepatic stellate cell genes was attenuated when cholate was omitted from the diet, consistent with the observed induction of collagen and other fibrotic genes specifically by cholate. Attenuated Response to Cholesterol and Cholate in an Atherosclerosis-resistant C57BL/6 Substrain—As described above, the Ath diet induces expression of several inflammatory genes in C57BL/6J mice, primarily through the cholesterol and cholate components. This raises the possibility that induction of these genes contributes to the pro-atherogenic effect of this diet. To address this issue, we investigated whether the same gene expression patterns occur in an atherosclerosis-resistant, but otherwise genetically similar, mouse strain C57BL/6ByJ. C57BL/6ByJ mice were derived from the same original progenitor as C57BL/6J but have been bred independently for many years (25Bailey D.W. Morse H.C.T. Origins of Inbred Mice. Academic Press, New York1978: 197-215Crossref Google Scholar). Although very few DNA polymorphisms have been detected between the two C57BL/6 substrains, the C57BL/6ByJ strain is resistant to hypercholesterolemia and aortic lesion formation in response to the Ath diet (26Mouzeyan A. Choi J.-H. Allayee H. Wang X.W. Sinsheimer J. Phan J. Castellani L.W. Reue K. Lusis A.J. Davis R.C. J. Lipid Res. 2000; 41: 573-582Abstract Full Text Full Text PDF PubMed Google Scholar). To determine whether this strain also differs in gene expression response to cholesterol and cholate, C57BL/6ByJ mice were fed the five diets described earlier, and expression levels of inflammatory genes induced by cholesterol and cholate were compared with those seen for C57BL/6J. C57BL/6ByJ mice were found to differ dramatically from C57BL/6J mice in gene expression response to cholesterol. The inflammatory genes activated by cholesterol in C57BL/6J mice were expressed at similar levels on the chow diet but were not induced significantly in C57BL/6ByJ by the Ath diet or other diets (Fig. 5, a and b). Although the SAA genes were induced 7–37-fold in C57BL/6J, Saa2 and Saa4 were induced only 2–3-fold and Saa3 was not induced at all in C57BL/6ByJ. Many other cholesterol-responsive inflammatory genes showed either no induction at all (Scyb9, Samhd1, Pira5, Lgals, and Ly6), or increased expression slightly on cholesterol-containing diets (H2-Aa, H2-Ab1, H2-Ebi, Ii, and Il2rg) (Fig. 5b). The induction of fibrosis-related gene expression in C57BL/6ByJ was also attenuated compared with C57BL/6J, but less dramatically than the cholesterol-responsive genes (Fig. 5, c and d). Although nidogen, lumican, vimentin, and Ctgf were similarly activated by cholate in both strains, most collagen genes were either not activated at all (Col6a1 and Col1a2) or activated at 25–50% the levels seen in C57BL/6J (Col1a1 and Col3a1). Thus, the two C57BL/6 substrains differ substantially in their gene expression response to dietary cholesterol, and C57BL/6ByJ mice also fail to induce collagen family members to the levels seen in C57BL/6J mice in response to cholate. The gene expression differences between C57BL/6J and C57BL/6ByJ in SAA and collagen were confirmed by biochemical measurements. Circulating SAA levels in C57BL/6J mice fed the Ath diet were 5–8-fold higher than in C57BL/6ByJ mice (Fig. 6a). Hepatic collagen levels also remained significantly lower in the C57BL/6ByJ mice in response to the Ath diet (Fig. 6b), consistent with the gene expression data. Thus, the closely related C57BL/6 substrains exhibit clearly different responses to the cholesterol and cholate components of the Ath diet, at both the transcriptional and protein levels. These findings are consistent with the possibility that attenuated inflammatory and fibrotic gene expression contributes to the atherosclerosis resistance in C57BL/6ByJ compared with C57BL/6J mice. The atherogenic diet containing cholesterol, cholate, and fat has been used for more than 25 years by several investigators to study the pathology and genetics of atherosclerosis in inbred mouse strains. Although it has been shown that both cholesterol and cholate are required to produce aortic lesions in susceptible mouse strains in a practical period of time (12Nishina P.M. Lowe S. Verstuyft J. Naggert J.K. Kuypers F.A. Paigen B. J. Lipid Res. 1993; 34: 1413-1422Abstract Full Text PDF PubMed Google Scholar), it is not clear what effect each component produces. To begin to address this issue, we have used microarrays to quantitate gene expression levels in response to Ath diet components. By comparing expression patterns on the Ath diet with those on diets in which one component was omitted, we identified groups of genes that are activated or repressed specifically by cholesterol, cholate, or fat. Although some genes were regulated by more than one component, many genes were strongly induced or repressed primarily by a single dietary component. Two key groups of genes that are likely to play a role in atherogenesis were found to be induced by the Ath diet: 1) more than 20 genes involved in acute inflammation/immune response, and 2) extracellular matrix proteins, characteristic of hepatic fibrosis in response to chronic injury. Inflammatory gene activation was dependent on the presence of cholesterol in the diet, whereas the collagen gene family members were induced specifically by cholate. These results extend previous observations showing that the Ath diet induces SAA genes in C57BL/6J liver (27Liao F. Andalibi A. Qiao J.H. Allayee H. Fogelman A.M. Lusis A.J. J. Clin. Investig. 1994; 94: 877-884Crossref PubMed Scopus (206) Google Scholar) by demonstrating that activation of SAA and other acute inflammatory genes occurs largely in response to cholesterol. Furthermore, the fact that SAA levels were elevated on the cholate-free diet demonstrates that the dramatic elevation in LDL/VLDL seen on the Ath diet is not required to activate SAA gene expression, nor do elevated HDL levels protect against it (see Fig. 1). Likewise, fibrotic gene expression was induced even when cholesterol was omitted from the Ath diet, indicating that fibrosis is not dependent on elevated plasma or hepatic cholesterol levels (Fig. 1). These results indicate that the cholesterol and cholate components of the Ath diet have distinct proatherogenic effects and suggest a strategy to study the contribution of the acute inflammatory response and fibrogenesis independently through dietary manipulation. To evaluate the potential relationship between the Ath diet-induced expression of inflammatory and fibrogenic genes and susceptibility to atherosclerosis, we compared gene expression in two substrains of C57BL/6 mice, one susceptible and the other resistant to atherosclerosis. We determined previously (26Mouzeyan A. Choi J.-H. Allayee H. Wang X.W. Sinsheimer J. Phan J. Castellani L.W. Reue K. Lusis A.J. Davis R.C. J. Lipid Res. 2000; 41: 573-582Abstract Full Text Full Text PDF PubMed Google Scholar) that C57BL/6ByJ mice fed the Ath diet maintain lower plasma LDL/VLDL cholesterol levels than C57BL/6J. Here we show that an important consequence of this may be attenuated expression in C57BL/6ByJ of cholesterol-responsive inflammatory genes in the liver, resulting in reduced levels of SAA in the circulation. We also observed reduced collagen gene activation in C57BL/6ByJ. This is intriguing in light of our finding that C57BL/6ByJ mice exhibit increased bile acid excretion compared with C57BL/6J mice (28Phan J. Pesaran T. Davis R.C. Reue K. J. Biol. Chem. 2002; 277: 469-477Abstract Full Text Full Text PDF PubMed Scopus (24) Google Scholar). Thus, reduced bile acid accumulation in B6By mice may protect these animals from fibrosis via reduced bile acid-induced stellate cell activation and/or apoptosis (29Brady L.M. Beno D.W.A. Davis B.H. Biochem. J. 1996; 316: 765-769Crossref PubMed Scopus (36) Google Scholar, 30Higuchi H. Gores G.J. Am. J. Physiol. 2003; 284: G734-G738Crossref PubMed Scopus (112) Google Scholar). Inhibition of stellate cell activation has been suggested as a strategy for treatment of conditions characterized by hepatic inflammation and fibrosis, including chronic viral hepatitis, alcoholic liver disease, and other causes of liver cirrhosis (23Kmiec Z. Adv. Anat. Embryol. Cell Biol. 2001; 161: 1-151Crossref Google Scholar, 31Bissell D.M. J. Gastroenterol. Hepatol. 1998; 33: 295-302Google Scholar, 32Li D. Friedman S.L. J. Gastroenterol. Hepatol. 1999; 14: 618-633Crossref PubMed Scopus (334) Google Scholar, 33Marra F. J. Hepatol. 1999; 31: 1120-1130Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, 34Tsukamoto H. Lu S.C. FASEB J. 2001; 15: 1335-1349Crossref PubMed Scopus (326) Google Scholar, 35Knolle P. Lohr H. Treichel U. Dienes H.P. Lohse A. Gerken G. Z. Gastroenterol. (Verh.). 1995; 33: 613-620PubMed Google Scholar). Because C57BL/6ByJ mice are resistant to atherosclerosis and exhibit reduced hepatic fibrosis, they may provide a valuable model to establish whether inhibition of stellate cell activation is a useful strategy for treatment of chronic viral hepatitis, alcoholic liver disease, and other causes of liver cirrhosis. We thank Ping Xu for excellent technical assistance. Download .zip (1.94 MB) Help with zip files