Quantitative trait locus analysis of neointimal formation in an intercross between C57BL/6 and C3H/HeJ apolipoprotein E-deficient mice

Abstract

Background

Inbred mouse strains C57BL/6J (B6) and C3H/HeJ (C3H) exhibit marked differences in neointimal formation after arterial injury when deficient in apolipoprotein E (apoE^−/−) and fed a Western diet. Quantitative trait locus (QTL) analysis was performed on an intercross between B6.apoE^−/− and C3H.apoE^−/− mice to determine genetic factors contributing to the phenotype.

Methods and Results

Female B6.apoE^−/− mice were crossed with male C3H.apoE^−/− mice to generate F₁s, which were intercrossed to generate 204 male F₂ progeny. At 10 weeks of age, F₂s underwent endothelium denudation injury to the left common carotid artery. Mice were fed a Western diet for 1 week before and 4 weeks after injury and analyzed for neointimal lesion size, plasma lipid and MCP-1 levels. One significant QTL, named Nih1 (61cM, LOD score: 5.02), on chromosome 12 and a suggestive locus on chromosome 13 (35cM, LOD: 2.67) were identified to influence lesion size. One significant QTL on distal chromosome 1 accounted for major variations in plasma non-HDL cholesterol and triglyceride levels. Four suggestive QTLs on chromosomes 1, 2, and 3 were detected for circulating MCP-1 levels. No correlations were observed between neointimal lesion size and plasma lipid levels or between lesion size and plasma MCP-1 levels.

Conclusions

Neointimal formation is controlled by genetic factors independent of those affecting plasma lipid levels and circulating MCP-1 levels in the B6 and C3H mouse model.

Introduction

Restenosis remains the most significant challenge limiting the success of angioplasty treatment to atherosclerotic cardiovascular disease. The incidence of restenosis after arterial interventions varies from 10% to 80% within 6 months, depending on the location of treatment and the type of intervention.¹ The placement of intravascular stents has proven effective in reducing post-interventional restenosis, but a fraction of patients receiving stents develop late instent restenosis. Neointimal hyperplasia is the primary cause for instent restenosis.²

Accumulating evidence indicates that restenosis is not a random phenomenon but a multi-factorial disorder with a strong heritable component.^3–6 Association studies have revealed an array of genes to be potentially involved in the restenotic process.^5,7–9 However, the results should be interpreted with caution because of small sample sizes and the heterogeneity of the patients. The difficulties inherent in human genetic studies strongly suggest that parallel approaches should be undertaken to identify genes for neointimal formation by using animal models.

The availability of numerous inbred mouse strains that differ in injury-induced neointimal thickening provides an experimental tool for identifying genetic factors involved in the restenostic process. C57BL/6(B6) and C3H mice are two inbred strains that exhibit dramatic differences in neointimal formation when deficient in apolipoprotein E (apoE^−/−) and fed a Western diet.¹⁰ B6.apoE^−/− mice readily develop neointimal lesions following carotid arterial endothelium denudation injury, whereas C3H.apoE^−/− mice are almost totally resistant to lesion formation. The dramatic difference between the two apoE^−/− strains in neointimal formation is ideal for conducting linkage analysis to dissect genetic factors contributing to the phenotype. In the present study, we performed quantitative trait locus (QTL) analysis of neointimal lesions and associated traits using an intercross between B6.apoE^−/−and C3H.apoE^−/−mouse strains.

Methods

Mice

Female B6.apoE^−/−mice were purchased from the Jackson Laboratory, Bar Harbor, ME. C3H.apoE^−/−mice at the N₁₂ generation were generated in our laboratory. Female B6.apoE mice were crossed with male C3H.apoE^−/−mice to generate F₁s, which were intercrossed by brother sister mating to generate 204 male F₂s. Mice were weaned onto a rodent chow diet and at 9 weeks of age, were switched onto a Western diet containing 21% fat, 0.2% cholesterol, and 19.5% casein (TD 88137; Teklad, Madison, WI) and maintained on the diet throughout the entire experiment. All procedures were carried out in accordance with the National Institutes of Health guidelines and approved by the institutional Animal Care and Use Committee.

Endothelium denudation

The procedure for removing the endothelium of the left common carotid artery was performed as previously described.^10,11 Briefly, mice were anesthetized by intramuscular injection of ketamine (80mg/kg; Aveco Inc.) and xylazine (8mg/kg; Lloyd Laboratories). Under a dissecting microscope, the left external carotid artery was ligated distally with a 6-0 silk suture and additional 6-0 sutures were looped around the common and internal carotid arteries. A transverse arteriotomy was made in the left external carotid artery, and an epon-resin probe was introduced and advanced ~1cm toward the aortic arch. Endothelial denudation of the artery was achieved by repeated withdrawal for three passes. After removal of the probe, the left external carotid artery was ligated and the skin incision was closed with surgical glue (Ethicon, Inc.).

Plasma lipid measurements

Mice were fasted overnight before blood was collected through retro-orbital puncture under ketamine and xylazine anesthesia. Plasma total cholesterol, non-HDL cholesterol, and triglyceride levels were measuredwith enzymatic assays as we previously described.¹¹

Measurements of circulating MCP-1

The same plasma samples used for lipid analysis were assessed for circulating MCP-1 levels. MCP-1 was quantified with a sandwich ELISA technique according to the manufacturer’s instructions (R&D Systems).

Tissue preparation and lesion quantification

The procedure was performed as previously described.^10,11 Briefly, the carotid arteries were perfusion-fixed via the left ventricle with 10% formalin. After fixation in 10% formalin for over 24 h, the front soft tissues of the neck were dissected out, and processed by using the standard histological technique. Serial 10-μm-thick sections were collected, starting from the disappearance of the left common carotid artery bifurcation, and mounted on poly-D-lysine-coated slides. On average, 400 sections were collected for each mouse. All sections were subjected to van Gieson staining for elastic laminas. Morphometric measurements of the injured common carotid artery were made in every 10 sections using Image ProPlus 3.0 software (Media Cybernetics). Luminal, internal, and external elastic areas were measured, as previously described.¹²

Genotyping

DNA was isolated from the tails of mice by using the standard phenol/chloroform extraction and ethanol precipitation method. A total of 125 microsatellite markers distinguishing B6 from C3H mice and covering all 19 autosomes and the X chromosome at an average interval of 13cM were typed by PCR. The parental and F₁ DNA served as controls for each marker.

Statistical analysis

Linkage analyses of neointimal lesions, plasma lipid and MCP-1 levels were performed by using MapManager QTXb20 (http://mapmgr.roswellpark.org/) and J/qtl (http://research.jax.org/faculty/churchill/software/Jqtl/index.html) software. The distributions of trait values in F₂ mice were assessed for normality with the SPSS program (SPSS, Chicago) by examining skewness, kurtosis, Kolmogorov-Smirnov test, and Q-Q plots. Neointimal lesion sizes were square root transformed before QTL analysis was performed as they were not normally distributed. Non-HDL cholesterol, triglyceride, and MCP-1 levels were directly used for QTL analysis. One thousand permutations of the trait values were used to define the genome-wide LOD score threshold required to be significant or suggestive for each specific trait. The support interval (SI) for each QTL was determined by using a 1-LOD drop from the QTL peak. Variance and mode of inheritance of the traits were determined with MapManager and further confirmation was made with linear regression analysis. ANOVA was used to determine whether the mean phenotype values of progeny with different genotypes at a specific marker were significantly different. Differences were considered statistically significant at P < 0.05.

Candidate gene analysis

Sequence comparison was made to ascertain which genes within the confidence interval of Nih1 were polymorphic between B6 and C3H mice by querying the NCBI Database (http://www.ncbi.nlm.nih.gov/SNP/MouseSNP.cgi) and Mouse Phenome Database (http://www.informatics.jax.org/javawi2/servlet/WIFetch?page=snpQF). The genes containing sequence variations in coding regions leading to amino acid substitutions or in regulatory regions were further examined for their expression in the arterial wall of two parental strains by real-time PCR. Total RNA was exacted from the aorta and carotid arteries of 6~10-week-old B6.apoE^−/−and C3H.apoE^−/− mice fed a chow diet using Trizol, digested by DNase I, and then reverse transcribed into cDNA using the ThermoScript RT-PCR kit (Invitrogen). cDNA was mixed with SYBR Green supermix reagent (Bio-Rad) and gene-specific primers (see supplementary data). Real-time PCR on each sample was run in triplicate on an iCycler iQ5 machine (Bio-Rad) under the condition of 50oC for 2 min, 95oC for 2 min, then 95oC 30 sec, 60oC 30 sec, and 72oC 30 sec for 40 cycles. The expression of each gene was determined in 3 or 4 biologically independent samples for each strain and was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

Immunohistochemical analysis

Paraffin-embedded sections were stained with the following primary antibodies: H-196, rabbit anti-human α-smooth muscle actin IgG (Santa Cruz); MCA519, rat anti-mouse macrophage/monocyte IgG (Serotec); H-414, rabbit anti-human Yy1 IgG (Santa Cruz); and rabbit anti-human Pacs2 IgG (provided by Dr. Gary Thomas at Vollum Institute, Portland, Or). Sections were deparaffinized with xylene, followed by incubationwith the primary antibodies. After a thorough wash, the slides were incubated with a biotinylated secondary antibody (Vector Laboratories). The reactions were visualized with a peroxidase chromogen kit (Vector Laboratories).

Sequence analysis

The promoter region of the Yy1 gene of B6 and C3H mice was PCR-amplified using primers 5 ’-ACCTGGTCTATCGAAAGGAAGCAC-3’/5’-TTGCCTTTACTCGTTACTCGGG-3’ and 5’-TAACGGACACACTTCCAACGGTGA-3’/5’-TCCGCGGTGCAACAGTGACAAA-3’ and purified PCR products sequenced on a 3730 DNA Analyzer (Applied Biosystems).

Results

B6.apoE^−/− mice developed prominent neointimal lesions in the common carotid artery 4 weeks after endothelial denudation injury when fed the Western diet (Figure 1). In contrast, neointimal lesions were barely detectable in C3H.apoE^−/− mice. Immunohistochemical analysis with specific antibodies against smooth muscle cells and macrophages revealed that smooth muscle cells were the primary cellular component of neointimal lesions and a few focally accumulated macrophages were present in the peripheral region of the lesions. To determine genetic factors contributing to variation in neointimal lesion size, we performed QTL analysis using an intercross between the two apoE^−/− strains. 204 male F₂ mice were analyzed for such phenotypes as neointimal lesion size, plasma lipid and MCP-1 levels 4 weeks after arterial injury and were genotyped for microsatellite markers spanning the entire genome. As shown in Figure 2, the values of plasma non-HDL cholesterol, triglyceride, and MCP-1 levels in F₂ mice are approximately normally distributed. The normal distribution parameters obtained by examining skewness, kurtosis, Kolmogorov-Smirnov test, and Q-Q plot indicated that the values of square root-transformed lesion sizes of F₂ mice approached a normal distribution.

Figure 1.

Representative cross-sections of the injured common carotid artery of B6.apoE^−/− and C3H.apoE^−/− mice fed a Western diet. Paraffin-embedded tissues were stained with van Gieson stain or specific antibodies against smooth muscle cells and macrophages. Note the distinct difference between the two strains in neointimal lesion size, the strong -actin positive arterial wall and neointimal lesions, and the focal accumulation of macrophages (arrow). Original magnification: 10 or 20×. Insets display magnified views of representative areas.

Figure 2.

Distributions of square root-transformed neointimal lesion size, untransformed non-HDL cholesterol, triglyceride, and MCP-1 levels in 204 male F₂ mice derived from B6.apoE^−/−and C3H.apoE^−/− mice. Mice were subject to endothelium denudation injury to the left common carotid artery and fed a Western diet for 1 week before and 4 weeks after injury.

A genome-wide scan with F₂ mice revealed that a significant QTL on chromosome 12 and a suggestive QTL on chromosome 13 influenced neointimal lesion size (Figure 3). Details of the QTL detected, including peak marker locus, LOD score, SI, variance, and mode of inheritance, are presented in Table 1. The chromosome 12 locus peaked near marker D12Mit144 (61cM, LOD: 5.02) and explained 15% of the variance in lesion size of the cross (Figure 4). We designated this QTL as Nih1 to represent the first QTL detected that affects neointimal lesion size in the mouse. This locus exhibited a dominant inheritance pattern in that F₂s with the heterozygous BC genotype at peak marker D12Mit144 had a lesion size comparable to F₂s with the homozygous BB genotype but larger than those homozygous in the CC genotype (P=1.27x10⁻⁵; Table 2). The chromosome 13 locus peaked near D13Mit250 (35cM) and had a suggestive LOD score of 2.67 (Figure 4, Table 1). This locus explained 8% of the variance in lesion size and exhibited a dominant effect from the B6 allele on the trait (Table 2).

Figure 3.

A genome-wide scan with F₂ mice to search for loci influencing neointimal lesion size. Chromosomes 1 through X are represented numerically on the X-axis. The relative width of the space allotted for each chromosome reflects the relative length of each chromosome. The Y-axis represents LOD score. Three dash lines on the plot represent LOD score thresholds for “suggestive (P=0.63)”, “significant (P=0.05)”, or “highly significant (P=0.01)” QTLs as calculated by permutation tests.

Table 1.

Significant and suggestive QTLs for neointimal lesion size, plasma non-HDL cholesterol, triglyceride, and MCP-1 levels identified in an intercross between B6.apoE^−/− and C3H.apoE^−/− Mice.

Chromosome marker (cM)	Trait	LODa	SI (cM)b	Variance (%)c	Model of inheritanced
D12mit144(61)	Lesion size	5.02	51.0–61.0	15	B6 Dominant
D13mit250(35)	Lesion size	2.67	27–42.5	8	B6 Dominant
D1mit270(92.3)	Non-HDL	3.31	83.0–104.0	7	Additive
D12mit97(47)	Non-HDL	4.00	27.0–59.0	9	B6 Dominant
D4mit192(6.3)	Non-HDL	2.18	6.3–40.8	5	Additive
D7mit330(57.5)	Non-HDL	2.13	49.0–65.2	5	Additive
D15mit161(69.2)	Non-HDL	2.08	47.9–65.2	4	B6 Dominant
D1mit270(92.3)	Triglyceride	4.92	82.0–103.5	11	Additive
D16mit165(10.3)	Triglyceride	2.70	10.3–15.8	6	Additive
D1mit102(73)	MCP-1	2.59	63.0–83.5	5	C3H dominant
D2mit263 (92.0)	MCP-1	2.06	57.9–101.8	4	Heterosis
D3mit21(19.2)	MCP-1	2.21	11.2–26.4	4	B6 dominant
D3mit230(38.3)	MCP-1	2.17	31.7–56.46	4	B6 dominant

^aLOD scores were generated with J/qtl software. “Suggestive” or “significant” QTLs were calculated by 1000 permutation tests and the corresponding genome-wide significance thresholds were P=0.63 and P=0.05, respectively. The significant loci are underlined to easily distinguish them from suggestive loci.

^bSI, support intervals, were defined by a 1-unit decrease in LOD score on either side of the peak.

^cVariance (%), which indicates the percentage of the phenotypic variance detected in the F₂ cohort under the peak marker, were generated with MapManager QTX program.

^dModel of inheritance was determined by using the MapManager QTX program and further confirmation was made for loci at which heterozygotes exhibited significantly higher or lower trait values than both homozygotes by performing linear regression using the additive and dominant/recessive models.

Figure 4.

Likelihood plots for neointimal lesion size on chromosome 12 and chromosome 13. Plots were generated using the interval mapping function of Map Manager QTX, including a bootstrap test shown as a histogram estimating the confidence interval of a QTL. Vertical green lines on the plot represent significance thresholds for the likelihood-ratio statistic, indicating “suggestive (P=0.63)”, “significant (P=0.05)”, or “highly significant (P=0.01)” peaks as calculated by permutation analysis for the genome-wide significance thresholds. Black plots reflect the likelihood-ratio statistic calculated at 1-cM intervals. The red plot represents the additive regression coefficient and the blue plot represents the dominant regression coefficient, indicating effect of the B6 allele.

Table 2.

Effects of B6 (B) and C3H (C) alleles in different QTLs on neointimal lesion size, plasma lipid and MCP-1 levels in the B6.apoE^−/− and C3H.apoE^−/− intercross.

Trait	Peak Marker	BB	BC	CC	P-value
Lesion size	D12mit144	58134.0±3888.8	63368.0±7330.8	28399.0±9647.2	1.27×10−5
Lesion size	D13mit250	27707.0±6085.8	36030.0±4129.0	19292.2±4964.4	0.002
Non-HDL	D1mit270	735.7±353.9	894.3±373.3	990.5±338.1	0.001
Non-HDL	D12mit97	966.9±304.1	934.6±371.0	712.7±327.6	0.0002
Non-HDL	D4mit192	960.9±380.4	909.7±363.5	759.1±344.1	0.008
Non-HDL	D7mit330	783.2±334.0	851.0±362.6	1007.8±377.0	0.007
Non-HDL	D15mit161	859.5±340.3	819.5±367.6	1005.6±369.4	0.012
Triglyceride	D1mit270	108.2±26.1	130.8±40.1	144.3±48.5	2×10−5
Triglyceride	D16mit165	126.3±48.9	139.6±42.3	114.3±29.6	0.002
MCP-1	D1mit250	79.3±32.5	98.1±38.2	93.9±30.1	0.006
MCP-1	D2mit263	100.8±40.6	85.0±28.3	94.8±38.5	0.02
MCP-1	D3mit21	89.5±34.9	85.9±31.9	104.2±41.6	0.009
MCP-1	Demit230	86.5±40.5	87.8±27.4	103.0±39.3	0.01

Measurements are means±SD. ANOVA was used to determine P values. The units of measurements are: lesion size: μm²/section; non-HDL cholesterol and triglyceride: mg/dl; MCP-1: pg/ml. BB, homozygous for B6 alleles at a linked marker; CC, homozygous for C3H alleles; BC, heterozygous for B6 and C3H alleles. The traits with significant linkage are underlined.

As shown in Figure 5 and Table 1, plasma non-HDL cholesterol and triglyceride levels were each influenced by two or more QTL. A QTL on distal chromosome 1 near D1Mit270 (92.3cM) had significant LOD scores of 3.31 for non-HDL and 4.92 for triglyceride and explained 9 and 11% of the variance in the traits, respectively. A significant QTL on chromosome 12 near D12Mit97 (47cM) had a significant LOD of 4.0 and accounted for 9% of the variance in non-HDL cholesterol levels. This QTL overlapped with Nhdlq12, recently mapped in a female BXH F₂ cross.¹³ A suggestive QTL for non-HDL near D4Mit192 (6.3cM; LOD: 2.18) corresponded to Chol8, identified in a 129S1/SvImJXCAST/Ei intercross,¹⁴ and a QTL near D7Mit330 (57.5cM, LOD: 2.13) corresponded to Chldq4, mapped in a MRL/MpJXSJL/J intercross.¹⁵ The QTL near D15Mit161 for non-HDL (69.2cM, LOD: 2.08) and the QTL near D16Mit165 (10.3cM, LOD: 2.7) for triglyceride are novel.

Figure 5.

Genome-wide scans with F₂ mice to search for loci affecting plasma levels of non-HDL cholesterol and triglyceride. Chromosomes 1 through X are numerically represented on the X-axis and the LOD scores represented on the Y-axis. The LOD score thresholds are shown on the figure.

Four suggestive loci, located on chromosomes 1, 2, and 3, were detected to affect plasma MCP-1 levels of F₂ mice (Table 1). The chromosome 1 locus near D1Mit102 (73cM) had a suggestive LOD score of 2.59 and explained 5% of the variance in MCP-1 levels. The chromosome 2 locus near D2Mit263 (92cM) had a LOD score of 2.06 and explained 4% of the variance. The chromosome 3 loci, peaked near D3Mit21 (19.2cM, LOD: 2.21) and D3Mit230 (38.3cM, LOD: 2.17), respectively, each accounted for 4% of the variance in MCP-1 levels.

The NCBI and the Mouse Phenome databases were queried to identify sequence differences underlying Nih1 that lead to changes in either the structure or quantity of a gene product between B6 and C3H mice. 22 non-synonymous SNPs were detected in 10 genes and 4 gene sequences (RIKEN cDNA or expressed sequences) and 41 SNPs were detected in the regulatory regions of 12 genes and 7 gene sequences. The genes containing non-synonymous SNP or/and SNP in regulatory regions were further examined for their expression in the arterial wall of the two strains. Three functional candidates, including Tnfaip2, Crip2, and Itgb8, were also evaluated by real-time PCR, although their sequence information was unavailable for C3H. As shown in Table 3, Yy1 was highly expressed in C3H, Pacs2 highly expressed in B6, Traf3, Cdc42bp, Tnfaip2, Klc1, Kif26a, Inf2, Pld4, Akt1, Gpr132, Jag2, Mta1, Crip2, and Itgb8 similarly expressed in both strains, and several other genes, including Chga, Bdkrb2, Adam6, Ptprn2, and Cdca7l, were barely detectable. Bag5 and Siva1 were also differentially expressed in the two strains when the original ct values (target gene-Gapdh cycle threshold) were analyzed (supplementary Table). Yy1, Bag5, Siva1, and Pacs2 were further evaluated by real-time PCR using RNA extracted from the uninjured carotid arteries (Table 4). The trend of differences between the two strains in gene expression levels was comparable in the aorta and carotid arteries. A search of the NCBI Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) reveals that Yy1, Bag5, Siva1 and Pacs2 are expressed in the aortic wall of C57BL/6 and C3H mice (accession # GDS735, GSE14854, GSE1560). Interestingly, a microarray study using RNA extracted from the aorta of B6.apoE^−/− and C3H.apoE^−/− mice also shows higher Yy1 and Bag5 expression in C3H (GDS735).

Table 3.

Real-time PCR assessment of expression of candidate genes for Nih1 in the aortic wall of B6.apoE^−/− and C3H.apoE^−/− mice.

Gene symbol	Gene name	Location (Mb)	B6	C3H	P value	SNP type
Chga	Chromogranin A	103.79	NQ	NQ		Reg
Bdkrb2	Bradykinin receptor, β2	106.80	NQ	NQ		?
Yy1	YY1 transcription factor	110.03	26.36±9.12	78.38±34.14	0.03	Reg
Traf3	Tnf receptor-associated factor 3	112.49	6.38±2.10	10.71±6.43	0.33	Reg
Cdc42bp	Cdc42 binding protein kinase beta	112.56	36.37±11.81	34.01±17.22	0.83	CN
Tnfaip2	Tumor necrosis factor, α-induced protein 2	112.68	32.08±21.08	50.28±57.56	0.86	?
Bag5	BCL2-associated athanogene 5	112.95	12.40±3.92	34.29±21.44	0.09	CN
Klc1	Kinesin light chain 1	113.03	42.04±6.01	47.31±14.38	0.52	Reg
Kif26a	Kinesin family member 26A	113.41	0.91±0.38	0.89±0.72	0.97	CN, Reg
Inf2	Inverted formin, FH2 and WH2 domain containing	113.85	33.59±13.08	51.10±21.46	0.21	CN, Reg
Siva1	SIVA1, apoptosis-inducing factor	113.87	8.25±2.66	23.91±14.78	0.08	CN
Akt1	Thymoma viral proto- oncogene 1	113.89	29.14±4.90	56.28±24.12	0.073	Reg
Pld4	Phospholipase D family, member 4	114.01	2.09±0.34	1.47±0.89	0.24	Reg
Gpr132	G protein-coupled receptor 132	114.10	NQ	NQ		Reg
Jag2	Jagged 2	114.15	5.39±3.17	5.84±2.41	0.83	CN
Pacs2	Phosphofurin acidic cluster sorting protein 2	114.31	8.11±3.36	2.61±1.36	0.02	CN, Reg
Mta1	Metastasis associated 1	114.37	30.79±2.60	29.21±9.48	0.76	CN
Crip2	Cysteine rich protein 2	114.38	71.76±29.90	89.21±54.86	0.6	?
Adam6	A disintegrin and metallopeptidase domain 6	115.24	NQ	NQ		Reg
Ptprn2	Protein tyrosine phosphatase, receptor type, N polypeptide 2	118.09	NQ	NQ		CN
Cdca7l	Cell division cycle associated 7 like	119.11	NQ	NQ		CN, Reg
Itgb8	Integrin beta 8	120.47	6.91±3.53	7.44±3.54	0.86	?

Gene expression levels are expressed as copy number of candidate gene mRNA relative to 1000 copies of GAPDH mRNA. Results are means ± SD of 3~4 biologically independent samples for each strain. Total RNA was prepared from the aorta of 6~10 week-old mice fed a chow diet. NQ: expression level too low to accurately quantify; CN: Coding nonsynonymous polymorphism resulting in amino acid substitution; Reg: regulatory region; ?: sequence information unavailable.

Table 4.

Real-time PCR assessment of four potential candidate genes for Nih1 in the carotid arteries of B6.apoE^−/− and C3H.apoE^−/− mice.

Gene symbol	B6	C3H	P value
Yy1	15.50±1.53	30.45±4.57		0.0008
Bag5	10.74±1.70	25.42±16.22		0.12
Siva1	5.14±1.84	14.37±2.77		0.0015
Pacs2	11.61±2.92	5.19±1.89		0.01

Yy1, Bag5, Siva1, and Pacs2 were differentially expressed in the aorta of two parental strains when the original ct values (target gene-Gapdh cycle threshold) were analyzed (supplementary Table). These 4 genes were further evaluated by real-time PCR using RNA extracted from the uninjured carotid arteries of 6~10 week-old mice fed a chow diet. Gene expression levels are expressed as copy number of candidate gene mRNA relative to 1000 copies of GAPDH mRNA. Results are means ± SD of 4 biologically independent samples for each strain.

Immunohistochemical analysis confirmed the expression of Yy1 and Pacs2 in the arterial wall and neointimal lesions (Figure 6). For Yy1, the uninjured carotid artery was more intensely stained in C3H than in B6 while the inner layer of neointimal lesions was more intensely stained in B6 than in C3H. Pacs2 showed endothelial staining in the uninjured artery but diffuse staining of the medial layer and neointimal lesions in the injured artery. We sequenced the promoter region of the Yy1 gene for B6 and C3H mice and found 5 SNPs (-A1445G, -A1443C, -C1435A, -A1354C, and -G1304T) and a 10-bp deletion (−1351 to −1342: TTTTTAAATA) in C3H (Figure 7).

Figure 6.

Representative light photographs of immunohistochemical analysis of Yy1 and Pacs2 expression in the carotid arteries with or without endothelium denudation injury. Sections were stained with rabbit polyclonal antibodies against Yy1 and Pacs2. Original magnification: 20×. Insets display magnified views of representative areas.

Figure 7.

Selected sequence traces of the promoter region of the Yy1 gene for B6 and C3H mice. Differences between the two strains in nucleotides are highlighted. Partial sequences of the Yy1 promoter region are not presented because no sequence difference has been found between the two strains.

Discussion

B6.apoE^−/− and C3H.apoE^−/− mice exhibit marked differences in neointimal formation following the carotid endothelium denudation injury when fed a Western diet. In the present study, we performed QTL analysis using an intercross derived from the two apoE^−/− strains to search for genetic factors contributing to neointimal formation and associated traits. We identified one significant QTL on chromosome 12 and one suggestive QTL on chromosome 13 for neointimal lesions, replicated 7 QTLs for plasma lipids, and detected 4 suggestive QTLs for circulating MCP-1.

Inbred strains B6 and C3H are prototype mouse models for genetic studies of atherosclerosis. B6 mice readily develop atherosclerosis whereas C3H mice are resistant when fed an atherogenic diet or deficient in apoE.^16,17 Using intercrosses derived from the two strains, we and others have identified several QTLs contributing to the development of atherosclerosis.^12,18–20 However, the study of various mouse strains has shown that genetic susceptibility to injury-induced neointimal hyperplasia is distinct from susceptibility to atherosclerosis as strains susceptible to atherosclerosis are resistant to neointimal hyperplasia and vice versa.²¹ Thus, genetic factors identified for atherosclerosis might not reflect genetic control of neointimal hyperplasia. ApoE^−/− mice represent an animal model in which spontaneous hyperlipidemia and atherosclerosis occur on a low fat, low cholesterol diet. Because most restenosis patients have a history of hyperlipidemia and atherosclerosis, apoE^−/− mice are obviously a more suitable model for the study of restenosis than wild-type mice. In this study, using a F₂ cross derived from the two apoE^−/− strains, we identified a significant QTL, named Nih1, on distal chromosome 12 and a suggestive QTL on chromosome 13 that affected neointimal lesion size. The chromosome 12 QTL mapped to the distal region (61 cM), and the B6 allele conferred the increased neointimal lesion formation at the locus. This locus did not overlap with any atherosclerosis QTLs mapped in genetic crosses derived from B6 and C3H mice,^{12,13,18–20} suggesting that neointimal hyperplasia and atherosclerosis are controlled by separate genetic factors. Indeed, the pathology of these two vascular disorders is quite different: the neointimal lesion consists largely of vascular smooth muscle cells, although macrophages are also present (Figure 1), whereas macrophages are the major cellular component ofatherosclerotic lesions at all stages and smooth muscle cellsonly become prominent in the advanced stage.²²

B6 and C3H are among the inbred mouse strains whose whole genome sequences and genetic variants have been well characterized. Thus, we perused all genes within the confidence interval of Nih1 to detect sequence variants that may contribute to the quantitative trait. Flint et al²³ have analyzed 20 cloned QTL genes in rodents and found that all of them have sequence variations in coding or regulatory regions leading to changes in structure, quantity, or both of a gene product. One probable candidate for Nih1 is Yy1, a ubiquitous and zincfinger transcription factor. Expression of this gene results in inhibition of neointima growth in human, rabbit and rat blood vessels.²⁴ We found that Yy1 expression levels in uninjured arterial walls were higher in C3H than in B6 by real-time PCR and immunohistochemistry. A higher expression of Yy1 by arterial wall cells would lead to a greater inhibition of neointimal formation in C3H. The strong expression of Yy1 by proliferating neointima might also regulate its growth. We identified a 10-bp deletion and 5 SNPs in the promoter region of Yy1, which could be responsible for the higher baseline expression in C3H. Other probable candidates in the region are three apoptosis related genes, including Bag5, Siva1, and Pacs2. Apoptosis or programmed cell death of vascular smooth muscle cells plays a critical role in injury-induced neointimal hyperplasia.²⁵ These three genes exhibited non-synonymous polymorphisms between B6 and C3H and were differentially expressed in the arterial wall of the two strains.

In the present study, we found that QTL on distal chromosome 1 contributed to variations in plasma non-HDL cholesterol and triglyceride levels in the F₂ cross. Thisfinding is consistent with previous observations made with female F₂ crosses between B6.apoE^−/− and C3H.apoE^−/− mice.^12,13 Hyperlipidemia has been considered responsible for aggravated neointimal formation of apoE^−/− or LDLR^−/− mice on a Western diet^10,11. However, the present study of F₂ mice has demonstrated that neointimal formation is independent of plasma lipid levels in that the size of neointimal lesions was not correlated with plasma non-HDL cholesterol or triglyceride levels (data not shown). A previous observation that B6.apoE^−/− mice developed 5-fold larger neointimal lesions than C3H.apoE^−/− mice despite their comparable levels of plasma LDL/LDL and HDL cholesterol on a chow diet¹⁰ also suggests that factors other than plasma lipids contribute to differential neointimal formation of the B6 and C3H strains.

In the present study, we detected four suggestive QTLs that influenced plasma levels of MCP-1 in apoE^−/− mice, although none of them overlapped with neointimal lesion QTLs. MCP-1 is the prototypical CC chemokine that is induced by noxious stimuli such as oxidized LDL, endotoxin, and mechanical forces in a variety of cells, including smooth muscle cells, endothelial cells, macrophages, and fibroblasts. We previously observed a distinct difference between B6 and C3H mice in the induction of MCP-1 by oxidized LDL in arterial wall cells.^17,26 Because MCP-1 functions in the recruitment of monocytes, the expression of this pro-inflammatory molecule by vascular wall cells is expected to lead to monocyte recruitment to injured arterial walls and promote neointimal formation. However, the present study of F₂ mice has suggested a less significant role for this chemokine in differential neointimal growth of the B6 and C3H strains because the size of neointimal lesions was not correlated with MCP-1 levels.

In summary, this study has identified the first significant QTL for neointimal hyperplasia, which does not overlap with loci for atherosclerotic lesions identified from previous crosses derived from the same parental strains. Moreover, we found no correlations between neointimal lesion size and plasma lipid levels or between lesion size and plasma MCP-1 levels. Thus, our findings indicate that neointimal formation is controlled by genetic factors independent of those affecting plasma lipid levels and circulating MCP-1 levels in hyperlipidemic mice.