Novel Strategy Using F1-Congenic Mice for Validation of QTLs
Studies at the Proximal Chromosome 10 Atherosclerosis Susceptibility Locus
Objective— We have previously identified a quantitative trait locus (QTL) for atherosclerosis susceptibility on proximal chromosome 10 (Chr10) (Ath11) in independent crosses of FVB and C57BL/6 (B6) mice on the apolipoprotein E (ApoE−/−) and LDL receptor (LDLR−/−) deficient backgrounds. The aims of the current study were to (1) test a novel strategy for validating QTLs using interval-specific congenic strains that were heterozygous (F1) across the genome, (2) validate the Chr10 QTL, and (3) to assess whether the phenotype is transferable by bone marrow transplantation.
Methods and Results— We generated Chr10 (0 to 21 cM) interval-specific mice on the F1.ApoE−/− background by crossing congenic FVB.ApoE−/−Chr10B6/FVB with B6.ApoE−/−, and B6.ApoE−/−Chr10B6/FVB with FVB.ApoE−/− mice. Lesion size was significantly larger in the resultant F1.ApoE−/−Chr10FVB/FVB mice compared to F1.ApoE−/− Chr10B6/FVB and F1.ApoE−/−Chr10B6/B6 mice, validating the Chr10 QTL. The effect of the congenic interval was more robust on the F1.ApoE−/− than on the FVB.ApoE−/− and B6.ApoE−/− backgrounds. Bone marrow transplantation in congenic mice showed that the effect of the proximal Chr10 interval was not transferable by bone marrow–derived cells.
Conclusions— A novel strategy of congenic strains on an F1 background proved useful to validate an atherosclerosis susceptibility QTL on mouse proximal Chr10.
Atherosclerosis is a complex genetic disorder with many genes involved and significant gene–gene and gene–environment interactions. Mouse models have proven useful in genetic studies of atherosclerosis in two general ways, (1) in a hypothesis-driven approach by altering expression of a candidate gene in vivo by creating transgenic or knockout mouse models and (2) in a hypothesis-free approach to identify new genes affecting atherosclerosis susceptibility by QTL mapping. The latter takes advantage of natural differences in atherosclerosis susceptibility between mouse strains. These differences were first noted when feeding mice atherogenic diets1,2 and could also be observed when various mouse strains were backcrossed onto the atherosclerosis-prone ApoE−/− and LDLR−/− backgrounds.3,4
We and others have used the QTL approach to identify genetic loci affecting atherosclerosis lesion size (for a review see5). Our laboratory has focused on atherosclerosis-resistant FVB/N (FVB) and atherosclerosis-susceptible C57BL/6 (B6) mice on the ApoE−/− and LDLR−/− backgrounds.3,4 QTL mapping of mice generated by intercrosses between these strains consistently showed a highly significant locus for atherosclerotic lesion area at the aortic root on proximal Chr10, designated Ath11.6,7 This was the only QTL independent of the animals’ gender, lineage, and genetic background (ApoE−/− and LDLR−/−). Interestingly, and contrary to expectations, the Chr10 QTL suggested a dominant antiatherogenic B6 allele.6,7
One important strategy for validating a QTL as well as identifying the culprit gene(s) is creation of an interval-specific congenic strain.8 This usually involves backcrossing the interval of interest from one strain onto the other strain over 10 generations. The QTL is regarded as validated if mice with the interval of interest on the second background continue to manifest the phenotype. Interval specific congenic mice also define the universe of possible culprit genes, and can be used as a starting point for creation of subcongenic strains to narrow this list to a small number of genes.8 In only a few cases have attempts been made to validate atherosclerosis susceptibility QTLs by creation and study of interval specific congenics.9–11 In each case, phenotyping was carried out by examining mice with the region of interest from one parental strain on the background of the other parental strain. This genetic background is quite different from the average genetic background of the F2 mice in the original cross that led to the discovery of the QTL in the first place. As a result, important interactions with the rest of the genome might be missed that might be necessary to bring out the phenotype of the culprit gene(s) responsible for the QTL.
We thus devised a novel strategy of testing genetic effects using interval-specific congenic strains on an F1 background. These mice are heterozygous for both parental strains across the genome which allows for possible genetic interactions not present on a pure background of either parental strain. This new method was tested by investigating atherosclerotic lesion size in congenic strains for the proximal Chr10 QTL interval on each parental as well as an F1 background. In addition, we performed bone marrow transplantation between congenic strains to determine whether bone marrow–derived cells were responsible for the effect of the Chr10 locus on atherosclerosis susceptibility.
Generation of Congenic Mice
ApoE−/− deficient mice on the C57BL/6 (B6.ApoE−/−) and FVB/N (FVB.ApoE−/−) backgrounds were taken from our own colony.3 Marker-assisted backcrossing12 using a genome-wide set of 167 polymorphic microsatellite markers (see “Genotyping” and supplemental Table I, available online at http://atvb.ahajournals.org) was used to generate interval-specific congenic animals on the FVB.ApoE−/− background carrying the 21-cM proximal portion of the Chr10 B6 genome between the microsatellite marker closest to the centromere (D10Mit49) and the microsatellite marker at ≈21 cM (D10Mit60). These mice, designated FVB.ApoE−/−Chr10B6/FVB, were intercrossed to generate mice homozygous for B6 at the proximal Chr10 interval (FVB.ApoE−/−Chr10B6/B6). In a similar manner, marker-assisted backcrossing was used to generate interval-specific congenic mice on the B6.ApoE−/− background carrying the 21-cM proximal portion of the Chr10 FVB genome. These mice, designated B6.ApoE−/−Chr10B6/FVB, were intercrossed to generate mice homozygous for FVB at the proximal Chr10 interval (B6.ApoE−/−Chr10FVB/FVB).
The following strategies were used to generate the various proximal Chr10 congenic F1.ApoE−/− mice for atherosclerosis studies: (1) For studies on the FVB.ApoE−/− background FVB.ApoE−/−Chr10B6/FVB mice were intercrossed to generate FVB.ApoE−/−Chr10B6/B6, FVB.ApoE−/−Chr10B6/FVB, and FVB.ApoE−/−Chr10FVB/FVB mice. (2) For studies on the B6.ApoE−/− background B6.ApoE−/−Chr10B6/FVB mice were intercrossed to generate B6.ApoE−/−Chr10B6/B6, B6. ApoE−/−Chr10B6/FVB, and B6.ApoE−/−Chr10FVB/FVB mice. (3) For studies on the F1.ApoE−/− background B6.ApoE−/−Chr10B6/FVB and FVB.ApoE−/− mice were crossed to produce F1.ApoE−/− Chr10FVB/FVB and F1.ApoE−/−Chr10B6/FVB mice (Figure 1A); and B6.ApoE−/− and FVB.ApoE−/−Chr10B6/FVB mice were crossed to produce F1.ApoE−/−Chr10B6/FVB and F1.ApoE−/−Chr10B6/B6 mice (Figure 1B). Offspring of these crosses that were recombinant in the proximal Chr10 interval were not used.
To generate mice carrying the B6 proximal Chr10 interval on the FVB.LDLR−/− background (FVB.LDLR−/−Chr10B6/FVB), FVB.ApoE−/−Chr10B6/FVB mice were backcrossed twice with FVB.LDLR−/− mice to substitute the LDLR−/− for the ApoE−/− trait. A PCR test was used to assess the presence or absence of each of the knockout traits (see below). B6.LDLR−/−Chr10B6/FVB and FVB.LDLR−/− mice were crossed to produce F1.LDLR−/− Chr10FVB/FVB and F1.LDLRChr10B6/FVB mice for atherosclerosis studies.
Animal care and experimental procedures conformed to the guidelines of the American Heart Association. Research animals were housed in the Rockefeller University’s Laboratory Animal Research Center in a specific pathogen-free environment in rooms with a 7 am to 7 pm light/dark cycle. The Rockefeller University’s Institutional Animal Care and Use Committee approved all procedures.
Bone Marrow Transplantation
Bone marrow transplantation was performed by transplanting F1.ApoE−/− Chr10B6/FVB bone marrow into F1.ApoE−/−Chr10FVB/FVB recipients and vice versa. In addition as controls, F1.ApoE−/−Chr10B6/FVB and F1.ApoE−/−Chr10FVB/FVB bone marrow was transplanted onto recipients of the same genotype. Recipient mice were irradiated with a lethal dose (9 Gy) one day before transplantation. Donor animals were euthanized, and femurs and pelvic bones were removed under sterile conditions. Bone marrow was flushed with 10-mL BSS solution (5.6 mmol/L D-Glucose, 0.4 mmol/L KH2PO4, 1.1 mmol/L Na2HPO4, 1.3 mmol/L CaCl2, 5.4 mmol/L KCl, 137 mmol/L NaCl, 1.0 mmol/L MgCl2, 0.8 mmol/L MgSO4) supplemented with 2% fetal calf serum (FCS). Cells were put through a cell strainer, washed twice, and resuspended in RPMI containing 2% FCS at a concentration of 5×107 cells per mL. Animals were transplanted by injecting 200 μL (107 cells) into the tail vein. After transplantation, animals received antibiotic treatment by adding gentamicin (200 μL, Sigma G1522), neomycin (30 mg, Sigma N1876), and kanamycin (30 mg, Sigma K4000) to the drinking water. Engraftment of bone marrow was confirmed at sacrifice by testing the genotype of Chr10 microsatellite markers in whole blood from transplanted animals.
Dissection of Mice
For atherosclerosis studies, mice were weaned at 28 days of age and fed a semisynthetic modified AIN76A diet containing 0.02% cholesterol. Animals on the ApoE−/− background were euthanized at 16 weeks of age; animals on the LDLR−/− background were euthanized at 20 weeks of age. Transplanted F1.ApoE−/− mice were euthanized 10 weeks after bone marrow transplantation at 18 weeks of age. At sacrifice, blood was removed by cardiac puncture and plasma isolated. The heart was removed for quantification of atherosclerosis at the aortic root as previously described.4
Genomic DNA was isolated as described.6 Genotyping for the ApoE-knockout trait was performed in a multiplex PCR containing primers for exon 2 (sense 5′-ctctgtgggccgtgctgttggtcacattgctgaca-3′), neomycin (sense 5′-cgccgctcccgattcgcagcgcatcgc-3′), and the common primer for exon 3 (antisense 5′-ctcgagctgatctgtcacctccggctctccc-3′). Genotyping for LDL-receptor knockout trait was performed as described on the Jackson Laboratory website (www.jax.org). Genotyping of the proximal Chr10 interval was performed using a panel of 11 microsatellite markers (d10mit49hex, 0 cM; d10mit80hex, 2.2 cM; d10mit213ned, 6.6 cM; d10mit16fam, 9.8 cM; d10mit106fam, 12 cM; d10mit214hex, 15.3 cM; d10mit3hex, 16.4 cM; d10mit147ned, 20.8 cM; d10mit60fam, 21.9 cM; d10mit20fam, 25.1 cM; d10mit92ned, 36.1 cM). Forward primers were fluorescently labeled with the dyes FAM, HEX, and NED as indicated. PCR reactions were carried out at the Genomics Resource Center of the Rockefeller University. Markers were resolved by capillary electrophoresis on an 3700 DNA-sequencer and analyzed with the Genotyper software (Applied Biosystems) as previously described.6 Marker assisted backcrossing was performed by testing the genome for 167 polymorphic microsatellite markers (supplemental Table I). PCR reactions were resolved on slab-gels as previously described.7
Plasma analyses were performed as previously described.13
Values are given as mean and standard deviation unless noted otherwise. Statistical analysis was done by nonparametric testing implemented in the Prism software, version 4.0. The Mann–Whitney test was used to compare 2 groups of mice, and the Kruskal-Wallis test with the Dunn multiple comparison test as a post test was used to compare more than 2 groups of mice.
Atherosclerosis in Chr10 Congenic Mice on the ApoE−/− Background
In previous work, we had identified a QTL for atherosclerosis susceptibility on proximal Chr10 in independent crosses of FVB and B6 mice on the ApoE−/− and LDLR−/− backgrounds. The QTL suggested a dominant antiatherogenic B6 allele. To validate this QTL, atherosclerosis studies were carried out in various Chr10 congenic mice. Initially, experiments were performed using “classical” congenic mice with the proximal Chr10 B6 interval on the FVB.ApoE−/− background. The genotypes of the mice studied are shown in Figure 2A. Because FVB.ApoE−/− is an atherosclerosis resistant strain, these mice had relatively small lesions (Figure 2B and 2C). Lesion size in the strain completely FVB across the genome (FVB.ApoE−/−Chr10FVB/FVB) was 4×104 μm2 and 5×104 μm2 in females and males, respectively. Based on the Chr10 QTL, one would expect an atheroprotective effect of the B6 allele and thus a further reduction of lesion size in congenic mice heterozygous (FVB.ApoE−/− Chr10B6/FVB) or homozygous (FVB.ApoE−/−Chr10B6/B6) for B6 at this locus, compared to FVB.ApoE−/−. Indeed, atherosclerotic lesion size was smaller in FVB.ApoE−/−Chr10B6/B6 and FVB.ApoE−/−Chr10B6/FVB mice than in FVB.ApoE−/−, but the reductions did not reach statistical significance in all comparisons (Figure 2B and 2C). These results are supportive but do not constitute a complete validation of the properties of the Chr10 alleles suggested by the initial QTL study.
We also tested the effect of the proximal Chr10 FVB interval on the B6.ApoE−/− background. The genotypes of the mice studied are shown in Figure 2D. Because B6 is an atherosclerosis susceptible strain, these mice had relatively large lesions (Figure 2E and 2F) despite significantly lower cholesterol levels than mice on the FVB background (Table 1). Based on the Chr10 QTL, one would expect a further increase in lesion size in congenic mice homozygous for the FVB allele (B6.ApoE−/−Chr10FVB/FVB) compared to heterozygotes (B6.ApoE−/−Chr10B6/FVB) and homozygotes (FVB.ApoE−/−Chr10B6/B6) for the C57 allele. However, this was not observed (Figure 2E and 2F), and thus the atherosclerosis susceptibility QTL on proximal Chr10 could not be validated on the B6.ApoE−/− background.
Atherosclerosis studies were then repeated using the novel approach of studying effects of the B6 and FVB proximal Chr10 congenic segments in F1 mice. The genetic background of these animals is shown in Figure 2G. Lesion size in F1.ApoE−/− mice averaged 19×104 μm2 and 13×104 μm2 in females and males, respectively, and was considerably larger than in FVB.ApoE−/− mice. As shown in Figure 2H and 2I, atherosclerotic lesion size was roughly equal in F1.ApoE−/− Chr10B6/FVB and F1.ApoE−/−Chr10B6/B6 mice but in each case significantly smaller than in F1.ApoE−/−Chr10FVB/FVB mice. This was exactly as predicted from our original QTL study in which mice either heterozygous or homozygous for B6 at the Chr10 locus had about equal atherosclerosis and both had less atherosclerosis than mice homozygous for FVB.7 These results validate the proximal Chr10 atherosclerosis QTL and confirm that this region contains a B6 allele diminishing atherosclerosis or a FVB allele increasing atherosclerosis.
The differences in lesion sizes between FVB.ApoE−/−, B6.ApoE−/−, and F1.ApoE−/− mice containing the various combinations of B6 and FVB proximal Chr10 congenic segments were not explainable by their differences in total or HDL cholesterol levels as shown in Table 1.
Atherosclerosis in Chr10 Congenic Mice on the LDLR−/− Background
In our previous work, we have identified the atherosclerosis QTL at Chr10 not only in an intercross of FVB and B6 on the ApoE−/− background7 but also in an independent intercross of the same strains on the LDLR−/− background.6 To test whether the QTL could also be confirmed in LDLR−/− mice, we backcrossed B6.ApoE−/−Chr10B6/FVB mice for 2 generations onto the LDLR−/− background to obtain B6.LDLR−/−Chr10B6/FVB animals. Because of the expected recessive effect of the FVB allele on increasing lesion size, it was sufficient to compare mice that were homozygous FVB at Chr10 with those that were heterozygous at this locus. Thus, B6.LDLR−/−Chr10B6/FVB animals were intercrossed with FVB.LDLR−/− mice to obtain F1.LDLR−/−Chr10B6/FVB and F1.LDLR−/−Chr10FVB/FVB mice (Figure 3A). As shown in Figure 3B and 3C, lesion size was significantly reduced in F1.LDLR−/−Chr10B6/FVB compared to F1.LDLR−/−Chr10FVB/FVB mice (44% and 42% in females and males, each P<0.001, respectively). These results confirmed the expected effects of the Chr10 QTL on atherosclerosis susceptibility on the LDLR−/− background. The influence on atherosclerosis was independent of lipid levels because no significant differences between strains were observed for plasma cholesterol and lipoprotein levels (Table 2).
Bone Marrow Transplantation of Chr10 Congenic Mice
Bone marrow transplantation between congenic strains was carried out to provide possible insight into the cellular origin of the proximal Chr10 interval effect on atherosclerosis susceptibility. If the proximal Chr10 interval effect was mediated by bone marrow–derived cells, one would expect an increase of lesion size when transplanting bone marrow from congenic mice that are FVB homozygous at the Chr10 interval into mice that are heterozygous at that interval. Conversely, one would expect a decrease of lesion size when transplanting bone marrow from congenic mice that are heterozygous at the Chr10 interval onto mice that are FVB homozygous at the interval. All mice used in these experiments were females on the F1.ApoE−/− background. As shown in Figure 4, there was no difference in lesion size between transplanting either F1.ApoE−/−Chr10B6/FVB or F1.ApoE−/−Chr10FVB/FVB marrow into F1.ApoE−/−Chr10B6/FVB mice, nor was there a difference in lesion size between transplanting either F1.ApoE−/−Chr10FVB/FVB or F1.ApoE−/−Chr10B6/FVB marrow into F1.ApoE−/−Chr10FVB/FVB mice. Regardless of the genotype of the transplant donor, atherosclerotic lesion size was smaller in F1.ApoE−/−Chr10B6/FVB compared to F1.ApoE−/−Chr10FVB/FVB recipients (P<0.01). Thus, the genotype of the bone marrow transplantation donor had no significant effect on lesion size, strongly suggesting the proximal Chr10 interval effect is not mediated by bone marrow–derived cells. The values for total and HDL cholesterol for each group of transplanted mice are given in supplemental Table II.
The initial aim of this study was to use congenic strains to validate the atherosclerosis QTL on proximal Chr10. In the process we tested the novel strategy of studying the effect of the interval in congenic strains heterozygous (F1) rather than homozygous for one or the other parental strains across the rest of the genome. This new strategy allowed us to validate the atherosclerosis QTL on proximal Chr10 in both ApoE−/− and LDLR−/− mice. We showed that the effect of the interval was more robust when tested in F1-congenic compared to B6-congenic or FVB-congenic mice. The method presented here might be more useful than the classical approach in validating atherosclerosis susceptibility QTLs. In addition, we used bone marrow transplantation in F1-congenic mice to show the proximal Chr10 interval effect was independent of bone marrow–derived cells.
There are many potential reasons that might explain the more robust validation of the Chr10 atherosclerosis QTL using F1-congenics compared to B6 and FVB congenics. (1) A rather trivial explanation for the lack of validation in B6 congenics might have been that smaller numbers of mice were used. However, these mice were difficult to breed, and given the trends observed we did not expend further efforts on this group. (2) We also considered that the Chr10 QTL might have been a false-positive and not verifiable in congenics. However, this QTL was originally identified in 2 independent intercrosses between FVB.ApoE−/− and B6.ApoE−/− mice7 and confirmed in a third independent cross between FVB.LDLR−/− and B6.LDLR−/− mice. In each case and in the aggregate the LOD score for this QTL was extremely high.6 Thus, the likelihood of the Chr10 QTL being false-positive must be very low. (3) The FVB.ApoE−/− background is atherosclerosis resistant, and lesion sizes might have been too low to appreciate the further reductions brought about by introducing one or two copies of the atheroprotective Chr10 B6 alleles on this background. (4) Similarly, the B6.ApoE−/− background is atherosclerosis-susceptible, and lesions may have been too advanced to see any additional effect of 2 proatherosclerotic FVB Chr10 alleles. (5) Finally, there is considerable evidence of the complexity of the atherosclerosis trait with many genes involved and it is presumed significant gene–gene interactions. In the current intercross, even though B6 is the atherosclerosis susceptible strain and FVB the atherosclerosis resistant strain, based on the QTL mapping study at the Chr10 locus, the FVB allele is proatherogenic and the B6 allele is antiatherogenic. The QTL result is based on measurements in F2 mice in which across the genome 25% of genes were FVB.FVB, 50% FVB.B6, and 25% B6.B6. Therefore, it is possible and perhaps even likely that the proatherogenic FVB allele or the antiatherogenic B6 allele at the Chr10 interval may require interaction with one or more FVB and B6 alleles across the genome to manifest its phenotype. This might not occur when examining the FVB interval on the B6 background or the B6 interval on the FVB background as in the classical method. We believe that the issues outlined above as (3), (4), and (5) are most likely to have caused the difficulty in validating the atherosclerosis QTL at proximal Chr10 by the classical method.
The F1 congenic strategy overcomes these difficulties because mice are heterozygous B6.FVB. Thus, across the genome these mice have both FVB and B6 alleles providing opportunities for gene–gene interactions not available on either the FVB or the B6 background. Therefore, the F1 background more closely approximates the situation in the original F2 mice which served as the basis for the QTL analysis. We believe this strategy has general applicability and should be tried in cases where confirmation of a locus in interval-specific “classical” congenics has failed.
The Chr10 interval specific congenic mice described in this study were also used to perform bone marrow transplantation studies. The result was clear and strongly suggests the phenotype of atherosclerosis susceptibility of the Chr10 interval is not conveyed by bone marrow–derived cells. Because the phenotype is not attributable to altered traditional cardiovascular disease risk factors, such as lipid levels, blood pressure, or glucose/insulin metabolism, transplantation studies suggest that the altered phenotype probably relates to preexisting cells in the vessel wall such as vascular smooth muscle or endothelial cells.
One interesting aspect of the Ath11 locus is the paradox that the atheroprotective allele at Chr10 is derived from B6 and the susceptible allele from FVB,6,7 while as a strain B6 is atherosclerosis susceptible and FVB is atherosclerosis resistant.3,4 The detection of a resistance allele in a susceptible strain has been elegantly dissected in a yeast model, where END3 was identified as a contributor to high temperature growth in heat-susceptible S288c yeast.14 A similar phenomenon might be responsible for the Ath11 locus, even though the responsible gene is not yet known. Atherosclerosis candidate genes at proximal Chr10 include genes such as interferon gamma receptor and arginase 1, for which we and others have shown an association with atherosclerosis susceptibility.15–17 However, sequencing and expression analyses led to the exclusion of these candidates (data not shown). Therefore, we believe that ultimately, only systematic narrowing of the QTL using congenic strains with reduced portions of the Chr10 interval will lead to the identification of responsible genes at this locus. In this regard, the work presented in this article will serve as an important first step.
In conclusion, we have presented a novel strategy to validate QTLs for complex traits or diseases using interval-specific F1-congenic mice. This strategy was tested and found to be superior to the classical method in validating the proximal Chr10 atherosclerosis susceptibility QTL. We suggest that our strategy will be useful for validating QTLs for other complex traits.
Sources of Funding
This study was supported by the NIH P01HL54591 Program Project 4/07/2006-3/31/2011 Project 1 (to J.L.B.) and a grant of the Deutsche Forschungsgemeinschaft (to D.T.; DFG Th374/1-1).
Received October 21, 2008; revision accepted February 2, 2009.
Dansky HM, Charlton SA, Sikes JL, Heath SC, Simantov R, Levin LF, Shu P, Moore KJ, Breslow JL, Smith JD. Genetic background determines the extent of atherosclerosis in ApoE-deficient mice. Arterioscler Thromb Vasc Biol. 1999; 19: 1960–1968.
Teupser D, Persky AD, Breslow JL. Induction of atherosclerosis by low-fat, semisynthetic diets in LDL receptor-deficient C57BL/6J and FVB/NJ mice: comparison of lesions of the aortic root, brachiocephalic artery, and whole aorta (en face measurement). Arterioscler Thromb Vasc Biol. 2003; 23: 1907–1913.
Teupser D, Tan M, Persky AD, Breslow JL. Atherosclerosis quantitative trait loci are sex- and lineage-dependent in an intercross of C57BL/6 and FVB/N low-density lipoprotein receptor−/− mice. Proc Natl Acad Sci U S A. 2006; 103: 123–128.
Dansky HM, Shu P, Donavan M, Montagno J, Nagle DL, Smutko JS, Roy N, Whiteing S, Barrios J, McBride TJ, Smith JD, Duyk G, Breslow JL, Moore KJ. A phenotype-sensitizing Apoe-deficient genetic background reveals novel atherosclerosis predisposition loci in the mouse. Genetics. 2002; 160: 1599–1608.
Glazier AM, Nadeau JH, Aitman TJ. Finding genes that underlie complex traits. Science. 2002; 298: 2345–2349.
Mehrabian M, Wong J, Wang X, Jiang Z, Shi W, Fogelman AM, Lusis AJ. Genetic locus in mice that blocks development of atherosclerosis despite extreme hyperlipidemia. Circ Res. 2001; 89: 125–130.
Holdt LM, Thiery J, Breslow JL, Teupser D. Increased ADAM17 mRNA expression and activity is associated with atherosclerosis resistance in LDL-receptor deficient mice. Arterioscler Thromb Vasc Biol. 2008; 28: 1097–1103.
Teupser D, Pavlides S, Tan M, Gutierrez-Ramos JC, Kolbeck R, Breslow JL. Major reduction of atherosclerosis in fractalkine (CX3CL1)-deficient mice is at the brachiocephalic artery, not the aortic root. Proc Natl Acad Sci U S A. 2004; 101: 17795–17800.
Thomas AC, Sala-Newby GB, Ismail Y, Johnson JL, Pasterkamp G, Newby AC. Genomics of foam cells and nonfoamy macrophages from rabbits identifies arginase-I as a differential regulator of nitric oxide production. Arterioscler Thromb Vasc Biol. 2007; 27: 571–577.
Teupser D, Burkhardt R, Wilfert W, Haffner I, Nebendahl K, Thiery J. Identification of macrophage arginase I as a new candidate gene of atherosclerosis resistance. Arterioscler Thromb Vasc Biol. 2006; 26: 365–371.
Getz GS, Reardon CA. Arginine/arginase NO NO NO. Arterioscler Thromb Vasc Biol. 2006; 26: 237–239.