Expression Profiling Identifies 147 Genes Contributing to a Unique Primate Neointimal Smooth Muscle Cell Phenotype
Objective— This study represents the first in an effort to systematically characterize different intimas by using expression array analysis.
Methods and Results— We compared smooth muscle cells (SMCs) of the neointima formed 4 weeks after aortic grafting with those from normal aorta and vena cava from cynomolgus monkeys. Hybridization to cDNA arrays identified subsets of 147 and 45 genes differentially expressed in the neointima versus the aorta and vena cava, respectively. The expression pattern differentiating neointima from aortic SMCs was characterized largely by suppression. Only 13 genes were induced in the neointima: 7 encoded matrix proteins (6 collagens and 1 versican) and 2 encoded inducers of matrix synthesis (osteoblast-specific factor-2/Cbfa1 and connective tissue growth factor). The genes suppressed most in the neointima included the regulator of G-protein signaling-5, SPARClike-1/hevin, and nonmuscle myosin heavy chain-B. A smaller gene set differentiated the neointima from the vena cava. Most were induced (39 of 45 genes), and overlap with the neointima-aorta set was significant (10 of 13 genes). Array results were validated with Northern analysis, in situ hybridization, or immunohistochemistry.
Conclusions— These data underscore the importance of matrix synthesis in neointimal maturation, and novel genes, newly associated with neointimal SMCs (regulator of G-protein signaling-5 and osteoblast-specific factor-2/Cbfa1), have raised new hypotheses regarding the pathogenesis of intimal hyperplasia.
Smooth muscle cell (SMC) “phenotype” has been central to hypotheses about the pathogenesis of atherosclerosis, restenosis, vasospasm, and transplant arteriopathies, diseases in which the characteristic pathology is an accumulation of excess intimal mass.1 However, defining intimal phenotypes has largely been a piecemeal discovery of individual genes through hypothesis-driven experiments or subtractive approaches comparing expression before and after balloon injury of normal arteries. These studies have led to the assertion, largely unsupported, that observed changes in gene expression are generic to other forms of intima. The present study represents the first in an effort to systematically characterize different intimas.
See page 1946
We began a characterization of SMC populations by profiling gene expression from normal aorta and vena cava of cynomolgus monkeys.2 Despite their distinct origins, these SMCs showed a remarkably consistent differential-expression profile, with only 68 of 4048 genes assayed marking arterial media from venous media. More intriguing was that all 68 genes were more highly expressed in aortic SMCs.2
We chose bypass graft neointima as our first intima for analysis because it is well defined morphologically, facilitating a clean dissection, and because it has been carefully modeled and characterized structurally.3 Histologically, graft lesions are homogeneous and are composed almost entirely of SMCs, with few inflammatory cells, and we have begun to learn about molecules that control their extent and composition.3,4⇓ Bypass grafts are particularly vulnerable to occlusion from the pathological accumulation of neointima.5 Characterizing how neointimal SMCs resemble or differ from those of normal vessels may further our understanding of their origin and contribution to graft failure.
Using human cDNA arrays, we identified a small gene set (147 of 4048 genes assayed) that consistently distinguished graft neointima at 4 weeks from the aorta; of these, only 13 were upregulated in the neointima. Forty-five genes distinguished the neointima from vena cava; 39 of these genes were upregulated. Comparisons between data sets identified genes unique or common among the 3 SMC populations, further refining the molecular signature of the mature graft neointima. These data confirm the importance of persistent matrix synthesis to neointimal maturation and extend our understanding of the molecular basis of intimal hyperplasia.
Polytetrafluoroethylene grafts (WL Gore) were inserted between the distal aorta and iliac arteries of 4 adult monkeys (Macaca fascicularis) as previously described.3,4,6⇓⇓ Animals were anesthetized with ketamine (15 mg/kg IM) and butorphanol (0.05 mg/kg IM), and anastomoses were created end to side with polypropylene sutures (Ethicon). After 4 weeks, the animals were anesthetized with pentobarbital (100 mg/kg IV) and heparinized (300 U/kg IV), and the grafts were immediately flushed with saline. Rings cut from each were fixed in formalin for paraffin embedding. Grafts were then opened longitudinally, the endothelium was denuded, and the neointima was stripped and snap-frozen in nitrogen. Normal aorta and vena cava were harvested proximal to the graft. The media was snap-frozen after removing the adventitia and endothelium, and adjacent rings were fixed and processed into paraffin. Procedures were approved by the Animal Care and Use Committee of Wake Forest University and conformed to the Guide for the Care and Use of Laboratory Animals (7th Edition, National Institutes of Health, 1996).
Array Hybridization and Data Analysis
Neointimas were paired with aorta and vena cava from the same animal for analysis. Frozen specimens were pulverized, RNA was prepared, and lack of degradation was confirmed by intact 18S and 28S bands.2 33P-labeled cDNA probes were synthesized from 2 μg total RNA from each tissue, and equivalent counts were hybridized against DNA arrays of 4048 known human sequences according to the manufacturer’s directions (GF211, Research Genetics). Arrays were washed as described,2 and expression was quantified from phosphorimager scans of similar intensity and background by using Pathways software (Research Genetics).
Individual genes were classified as “consistently expressed” when spot labeling intensity was ≥1.5-fold average background calculated for all arrays of that tissue type. Only genes consistently expressed in at least 1 tissue type were further analyzed. Expression was normalized to average spot intensity before comparing tissues from individual animals. Thus, intra-animal comparisons yielded 3 normalized ratios for each consistently expressed gene: neointima/aorta, neointima/vena cava, and aorta/vena cava.
Differential expression was then defined as an average ratio of at least ±1.5-fold in each tissue comparison (n=4 animals) and at least ±1.3-fold (same sign direction) for any single comparison. Genes were excluded if the coefficient of variation among the 4 expression ratios exceeded 2/3.2 Differentially expressed genes were subsequently categorized hierarchically into classes as previously described2 (please see online Table I, available at http://atvb.ahajournals.org).
Additional analyses were performed to investigate the consistency of our results: (1) mock data sets were created by systematically inverting neointima/aorta ratios in 2 of 4 animals to determine whether comparable numbers of genes appeared significant; (2) aortic array data sets from each animal were paired in all possible combinations of 4 ratios, with each aorta used no more than twice; (3) our selection criteria were applied to bootstrapped neointima/aorta data sets, for which paired ratio data were resampled from the 4 animals at the gene level with replacement, 200 times, to examine the statistical properties of the procedure; and (4) statistically based differential-expression approaches were applied, with adjustment for multiple comparisons, including significance analysis of microarrays (SAM).7–9⇓⇓ Please refer to the online supplement (http://atvb.ahajournals.org).
Northern Analysis and In Situ Hybridization
RNA (15 μg) prepared from neointima, aorta, and vena cava and also from carotid media and liver was run, transferred, and hybridized as described.2 Multiprime probes (Amersham) were synthesized from sequence-verified clones of human regulator of G-protein signaling-5 (RGS5),2 nonmuscle myosin heavy chain (NMMHC)-B,2 (α2)collagen-I,10 versican,11 and a 28S rRNA primer.12
In situ hybridization was performed on 5-μm paraffin sections of aorta, vena cava, and graft neointima adjacent to the tissues used for array analyses, as previously described.13 In addition, archived sections of monkey iliac arteries removed 4 weeks after angioplasty were included in the analysis of the neointima.13 Sections were probed with 35S-labeled riboprobes (sense and antisense) transcribed from human RGS5, (α2)collagen-I, and versican cDNAs as described.2,13⇓ Immunohistochemistry was performed in adjacent sections by using antibodies for α-actin (1:1000, Dako) and NMMHC-B (1:500).13
Array Hybridization Efficiency and Classification of Differential Gene Expression
Array analysis began with measurement of the average signal intensity for each filter after application of neointimal, aortic, or vena caval probes. Of 4048 genes arrayed, our criteria for consistent expression was met by 3620±168 genes in the neointima, 3338±114 genes in the aorta, and 3497±428 genes in the vena cava. These were only slightly less than those for human aorta on the same filters,2 suggesting that interspecies differences in sequence homology had little impact on hybridization affinity.
In published studies using this filter system, there is a high level of reproducibility within a filter lot.14,15⇓ We have confirmed this repeatedly, demonstrating a high level of reproducibility when multiple GF211 filters from 1 lot were probed. However, the same experiment performed with different filter lots demonstrated variability attributable to the lots themselves (data not shown). To eliminate this variability, we paired samples from each animal with filters of the same lot.
Differential expression between tissues within each animal was then determined for expressed genes, comparing normalized labeling intensity and establishing ratios.2 To address potential interanimal variability in gene expression, comparisons were first made within individual animals. Differentially expressed genes were then categorized on the basis of hybridization magnitude and ratio variance as previously described (please see online Table I).2
Genes Marking Neointima From Aorta
We studied mature neointima after 4 weeks to avoid genes associated transiently with inflammation, migration, and proliferation. Graft neointima is nearly quiescent and largely free of inflammatory cells at this time point.3,4⇓ We identified 147 genes distinguishing neointima from aorta: 13 were expressed more highly in the neointima, and 134 were expressed more highly in the aorta (class I genes, Table 1. For all gene classes, please see online Table II available at http://atvb. ahajournals.org). The surprisingly small gene set more highly expressed in the neointima was dominated by matrix genes (7 of 13), including versican and 6 distinct collagen chains. Three genes were expressed >5-fold higher by neointimal SMCs, including osteoblast-specific factor-2 (OSF2)/Cbfa1, an osteogenic transcription factor, and 2 markers of injury-induced neointima, (α2)collagen-I and (α1)collagen-III.10,13,16⇓⇓ Of 23 collagen genes arrayed, 19 were expressed above background in every neointima or aorta. Of these, only 6 were differentially expressed; all were higher in the neointima (Table 2). Although it is well established that the neointima is composed largely of matrix (60% to 80% by volume),6 the dominance of specific collagen subtypes is intriguing.
The gene set downregulated in the neointima relative to the aorta was 10-fold larger (134 genes). Two genes were expressed >5-fold higher in the aorta: SPARClike-1 (SPARCL1)/hevin and the known arterial SMC marker RGS5.2 Array hybridization images for several class IA genes are shown in Figure 1. All genes with expression ratios >2-fold(±) are shown in online Figure I (available at http://atvb.ahajournals.org).
Genes Marking Neointima From Vena Cava
We have previously identified 68 genes that consistently distinguish the aorta from the vena cava; all are expressed more highly in the aorta.2 We speculated that the graft neointima would resemble the aorta more than the vena cava because it is formed within an arterial environment, and neointimal SMCs are thought to be derived, in part, from the artery wall at anastomoses.3 Surprisingly, only 45 genes consistently differentiated the neointima from the vena cava, one third the difference between the neointima and aorta. As with comparing the aorta with the vena cava,2 39 of 45 genes were more highly expressed in the neointima, and this gene set was also dominated by matrix genes, including (α2)collagen-I, (α1)collagen-III, and versican (please see online Table III, available at http://atvb.ahajournals.org).
Three-Way Comparison of Neointima, Aorta, and Vena Cava
To further characterize the neointimal expression phenotype, we surveyed the 3 gene sets for those more highly expressed in the neointima and aorta than in the vena cava, in the neointima and vena cava than in the aorta, or in the neointima alone (please see online Table IV, available at http://atvb. ahajournals.org). No genes were expressed more highly in the vena cava than in the aorta,2 so there were no common markers of the neointima and vena cava from the aorta. Six genes were more highly expressed in the neointima and aorta relative to the vena cava, potentially representing the signature arising from an arterial origin of neointimal SMCs or hemodynamic influences. Five (elastin, SPARC, vimentin, fibromodulin, and NMMHC-B) were among 10 genes most highly expressed in the aorta versus the vena cava.2 Comparing genes upregulated in the neointima versus the aorta and in the neointima versus the vena cava, we found 11 common genes, including all matrix genes identified in the neointima/aorta comparison. This set of 11 genes underscores the key role of specific matrix molecules in neointimal formation and maturation. The osteogenic transcription factor, OSF2/Cbfa1, and connective tissue growth factor (CTGF) further emphasized the representation of extracellular matrix in the neointima, inasmuch as both are known inducers of matrix gene expression.17–19⇓⇓ Genes underrepresented in the neointima versus both the aorta and vena cava consisted of only SPARCL1/hevin and complement component-7.
Statistical Controls for Possible False Discovery
Statistical methods used to determine the significance of comparisons between large data sets are still evolving at this early stage of array analysis. For that reason, we used a paired data analysis, minimizing animal-animal variations and problems with comparisons made with different filter lots. We also insisted on consistency by only accepting ratios valid in all 4 experiments. Nonetheless, it is critical to point out the possibility of false discovery whenever making such large numbers of comparisons.
As a first approach to this issue, the neointima/aorta data were reanalyzed to determine whether comparable numbers of genes would appear significant if we had no knowledge of whether the tissue was of intimal or aortic origin. For this purpose, the paired relationship was maintained, and the ratios were systematically inverted and reanalyzed in sets of 4, with 2 pairs having intima as the numerator and 2 having the aorta as the numerator. There are 3 possible scrambled data sets (please see online Table V, available at http://atvb.ahajournals.org). The number of genes surviving our criteria for differential expression in each was 1, 6, and 13, respectively; this is far less than the 147 that were found with the use of valid ratios. Thus, we would expect a 4.5% average false discovery rate. Next, in response to issues raised during review, we estimated the possibility of a comparable discovery rate when expression data from the same tissue type were compared. Ratios were determined from the expressed genes from the 4 aortas, with the selection of all possible combinations of 4 (see Methods). This analysis showed no combinations for which the genes reached significance by our criteria for differential expression.
Neointima/aorta data sets were also analyzed by using bootstrap methods and SAM. These methods confirmed the size of the difference between the expression patterns, although as we will discuss, some genes were lost when they were subjected to these criteria. The SAM analysis, which focuses more on the level of variation of each gene than on the expression ratio, also found some genes that were missed by our criteria. Expression ratios for neointimal genes identified by our analysis but not by SAM (and vice versa) are shown in Table 1 and the online supplement (Tables VI and VII, available at http://atvb.ahajournals.org).
Northern Analysis and In Situ Hybridization
Array data were verified by measuring expression in macaque tissues by Northern analysis (Figure 2), confirming high neointimal expression of (α2)collagen-I and versican and high aortic expression of RGS5 and NMMHC-B. Message sizes were consistent with published human sequences.2,10,11,20⇓⇓⇓
In situ hybridization or immunohistochemistry further validated the expression of (α2)collagen-I and versican in graft neointima and expression of NMMHC-B and RGS5 in the aorta. Extending these observations to another form of neointima, we also probed sections of monkey iliac arteries removed 4 weeks after angioplasty and found a similar induction of (α2)collagen-I (Figure 3) and versican in the neointima compared with uninjured iliac arteries from the same animals. This suggested overlap in the program of gene expression in neointimas formed in response to very different mechanical stimuli. Expression was localized predominantly to SMCs, as demonstrated by α-actin immunohistochemistry in adjacent tissue sections (not shown).
NMMHC-B expression was higher in aortic media than in the neointima of grafts or after angioplasty (Figure 3). Consistent with the array images (Figure 1), RGS5 expression was high in aortic media but variable in graft neointima. RGS5 expression was more consistently elevated in the neointima formed after angioplasty than among the 4 graft neointimas, suggesting differences in maturation between the 2 intimas at the 4-week time point. Expression was colocalized to actin-positive SMCs and was uniformly absent in the endothelium and actin-negative adventitial fibroblasts (not shown).
The present study updates concepts of neointimal pathology proposed in 1995 by one of the authors.1 That review compiled a set of 77 “intimal” SMC genes identified by subtractive cloning or hypothesis-based studies. The putative set of intimal genes has increased substantially since then.21 The most important observation in the present study is the small number of genes that consistently marked SMCs within the neointima 4 weeks after lesion initiation and the limited overlap with genes previously associated with intimal pathology.1,21⇓ Only 13 genes were upregulated in the neointima relative to normal aorta, whereas 134 genes were downregulated. If it is assumed that these arrays represent 12% of the expressed human genome (4048 of ≈35 000 genes22), it is unlikely that the mature neointima is distinguished from the aorta by upregulation of >110 genes and downregulation of >1100 genes.
These data suggest that the literature overestimates the number of genes constituting a neointimal SMC phenotype. The reasons for this discrepancy are obvious. Previous reports focused on gene expression underlying early events in intimal formation (eg, replication, migration, and response to inflammation), often in the complicated atherosclerotic milieu. Although critical to lesion initiation, these variables may have little bearing on gene expression by SMCs constituting mature neointimal lesions. Moreover, the significance of any single gene expression ratio is difficult to determine unless something is known about how other genes vary between 2 tissues. Array-based analyses solve this problem by sampling a large data set, but as we will discuss, the large number of genes studied in each experiment raises its own set of statistical issues.
In the present study, we used a simple experimental design by performing paired analyses of specimens from individual animals and by sorting data by a statistic that identified expression ratios as significant on the basis of their magnitude and consistency, adjusting for the observed variance. This approach also controlled for variability across array lots, inasmuch as all tissues from a given animal were assayed with the use of the same lot of filters. Expression ratios were repeated in multiple animals, and only those genes meeting criteria for altered expression in every animal were included.
A problem inherent to large array analyses is the possibility that some values will appear significant by chance at the usual confidence levels. For example, using a probability value of 0.05, one would expect ≈200 observations to appear significant in a single experiment in which 4000 genes were studied. To address this valid concern, we repeated each measurement in 4 individual animals and required that the data survive our criteria for significance in all. We then used additional analytic methods to interrogate the data for consistency. One approach is to use the t test on multiple sets of data for which the identity of the specimen source is randomized. Our application of related multiple comparison adjustments based on the paired t test resulted in between 0 and 3 genes being flagged as differentially expressed. This low number reflects the small number of data sets (4) and illustrates a key issue for array studies: how many data sets (experiments) can be performed for a reasonable price? This issue is particularly relevant to primate studies, for which costs and resource utilization make large experiments impractical. As an alternative evaluation of the stochastic nature of the criteria for gene selection, we performed a bootstrap statistical analysis in which the neointima/aorta data were resampled by gene (with replacement from the 4 animals) 200 times, allowing for the possibility of using the same value 4 times. To those simulated data sets, we applied our original criteria for differential expression and found 43 genes, all of which were among the original data set of 147 genes, including 9 of the 13 class IA neointimal genes (Table 1). Thus, within the limits of these statistical considerations, our analyses demonstrate a surprisingly small cohort of genes that consistently characterize graft neointima, including limited overlap with genes previously associated with intimal pathology.1,21⇓ We acknowledge that a number of important genes may have been excluded, either because they fall below the sensitivity of measurement or because low expression values will have a relatively lower signal-to-noise ratio.
Surprisingly, 90% of the genes with altered expression in the neointima versus the aorta were downregulated in the neointima (134 of 147). This suggests that neointimal pathology results primarily from a loss of gene function (perhaps inhibitory genes), particularly if neointimal SMCs are derived from the adjacent aortic media, as proposed.3 RGS5, a definitive “arterial” gene,2 was variably expressed in the neointima (Figure 1) and was intermediate between the aorta and vena cava. Our previous observation that RGS5 is highly expressed by SMCs of arteries but not veins2 suggests that RGS5 may provide chronic inhibition of G-protein–mediated hypertrophic responses peculiar to the arterial environment. RGS5 is a GTPase for Gαi and Gαq, promoting reformation of an inactive G-protein heterotrimer.23 Cardiac overexpression of Gαq has been shown to induce hypertrophy,24 suggesting that regulators of G-protein signaling counter hypertrophic signals in normal cardiovascular tissues. In fact, RGS5 and closely related regulators of G-protein signaling have been directly implicated in the control of hypertrophic responses.23,25,26⇓⇓ Our observation that RGS5 expression is lower in graft neointima than in the normal artery wall is the first association of altered RGS expression with SMC pathology, and by convention, downregulation implies loss of inhibition in G-protein signaling. We speculate that RGS5 expression increases in neointimal SMCs as the lesions mature. Future experiments will test the hypothesis that RGS5 expression increases within the neointima at later time points (beyond 4 weeks), when it reaches a quiescent state through enhanced growth-inhibitory signals.
Of the 13 genes expressed more highly in the neointima than in the aorta, nearly all were associated with extracellular matrix production. This is consistent with many previous reports implicating an overproduction of matrix genes in other forms of the intima.13,27,28⇓⇓ The 6 nonmatrix genes included several that were known to have an impact on matrix biology in vascular tissues, including CTGF, OSF2/Cbfa1, and lysyl oxidase–like-2.29–31⇓⇓ CTGF is expressed in the shoulder of atherosclerotic plaques32 and is known to stimulate collagen-I synthesis33 and matrix accumulation.17 Knockout mice demonstrate that OSF2/Cbfa1 is required for bone formation,34 and transgenic mice expressing mutant OSF2/Cbfa1 show reduced matrix expression in bone, including collagen-I.18 Lysyl oxidase–like-2 belongs to a gene family that promotes collagen cross-linking.35
Curiously, our analysis did not detect transforming growth factor-β, elastin, osteopontin, or NMMHC-B as upregulated in the neointima relative to the aorta, although they are commonly reported in neointimas.36–38⇓⇓ The absence of NMMHC-B upregulation was particularly unexpected inasmuch as increased NMMHC-B expression has been reported as a defining feature in postangioplasty and spontaneous intimas.39 These differences suggest that graft neointima is fundamentally different from other forms of intima or, as noted above, that we studied a mature neointimal phenotype in contrast to the earlier intimal events studied in many models of intimal hyperplasia.
In summary, we have systematically characterized gene expression profiles of graft neointima versus normal aorta in a nonhuman primate. Compared with the normal artery, after 4 weeks, 13 genes were upregulated in the neointima, and 134 were downregulated. RGS5, a marker characteristic of normal arteries, was present but downregulated in this intima, suggesting a role for this G-protein regulator in SMC hyperplasia. Limited overlap with genes identified in previous studies of intimal hyperplasia suggests that further experiments will be required to define the extent of diversity among distinct intimal SMC populations.
This work was supported by National Institutes of Health grants RO1 HL-57557 (R.L.G.), RO1 HL-58083 (S.M.S.), and PO1 HL-03174 (S.M.S.). Dr Geary is a Brook’s Scholar of the Wake Forest University School of Medicine. The authors thank Deanna Brown, Isa Werny, Jonathan McBride, and Colette Norby-Slycord for excellent technical assistance and Michelle Gammons, Karen Greenough, and Tami Daley for manuscript preparation.
Received August 20, 2002; revision accepted August 28, 2002.
- ↵Schwartz SM, deBlois D, O’Brien ER. The intima: soil for atherosclerosis and restenosis. Circ Res. 1995; 77: 445–465.
- ↵Adams LD, Geary RL, McManus B, Schwartz SM. A comparison of aorta and vena cava medial message expression by cDNA array analysis identifies a set of 68 consistently differentially expressed genes, all in aortic media. Circ Res. 2000; 87: 623–631.
- ↵Geary RL, Kohler TR, Vergel S, Kirkman TR, Clowes AW. Time course of flow-induced smooth muscle cell proliferation and intimal thickening in endothelialized baboon vascular grafts. Circ Res. 1994; 74: 14–23.
- ↵Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A. 2001; 98: 5116–5121.
- ↵Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, Speed TP. Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res. 2002; 30: e15.
- ↵Adams LD, Lemire JM, Schwartz SM. A systematic analysis of 40 random genes in cultured vascular smooth muscle subtypes reveals a heterogeneity of gene expression and identifies the tight junction gene zonula occludens 2 as a marker of epithelioid “pup” smooth muscle cells and a participant in carotid neointimal formation. Arterioscler Thromb Vasc Biol. 1999; 19: 2600–2608.
- ↵McCormick SM, Eskin SG, McIntire LV, Teng CL, Lu CM, Russell CG, Chittur KK. DNA microarray reveals changes in gene expression of shear stressed human umbilical vein endothelial cells. Proc Natl Acad Sci U S A. 2001; 98: 8955–8960.
- ↵Mori T, Kawara S, Shinozaki M, Hayashi N, Kakinuma T, Igarashi A, Takigawa M, Nakanishi T, Takehara K. Role and interaction of connective tissue growth factor with transforming growth factor-beta in persistent fibrosis: a mouse fibrosis model. J Cell Physiol. 1999; 181: 153–159.
- ↵Ducy P, Starbuck M, Priemel M, Shen J, Pinero G, Geoffroy V, Amling M, Karsenty G. A Cbfa1-dependent genetic pathway controls bone formation beyond embryonic development. Genes Dev. 1999; 13: 1025–1036.
- ↵Simons M, Wang M, McBride OW, Kawamoto S, Yamakawa K, Gdula D, Adelstein RS, Weir L. Human nonmuscle myosin heavy chains are encoded by two genes located on different chromosomes. Circ Res. 1991; 69: 530–539.
- ↵Schwartz SM. The intima: a new soil. Circ Res. 1999; 85: 877–879.
- ↵D’Angelo DD, Sakata Y, Lorenz JN, Boivin GP, Walsh RA, Liggett SB, Dorn GW. Transgenic Galphaq overexpression induces cardiac contractile failure in mice. Proc Natl Acad Sci U S A. 1997; 94: 8121–8126.
- ↵Grant SL, Lassegue B, Griendling KK, Ushio-Fukai M, Lyons PR, Alexander RW. Specific regulation of RGS2 messenger RNA by angiotensin II in cultured vascular smooth muscle cells. Mol Pharmacol. 2000; 57: 460–467.
- ↵Lin H, Wilson JE, Roberts CR, Horley KJ, Winters GL, Costanzo MR, McManus BM. Biglycan, decorin, and versican protein expression patterns in coronary arteriopathy of human cardiac allograft: distinctness as compared to native atherosclerosis. J Heart Lung Transplant. 1996; 15: 1233–1247.
- ↵Oemar BS, Werner A, Garnier JM, Do DD, Godoy N, Nauck M, Marz W, Rupp J, Pech M, Luscher TF. Human connective tissue growth factor is expressed in advanced atherosclerotic lesions. Circulation. 1997; 95: 831–839.
- ↵Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, Sato M, Okamoto R, Kitamura Y, Yoshiki S, Kishimoto T. Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell. 1997; 89: 755–764.
- ↵Jourdan-Le Saux C, Tronecker H, Bogic L, Bryant-Greenwood GD, Boyd CD, Csiszar K. The LOXL2 gene encodes a new lysyl oxidase-like protein and is expressed at high levels in reproductive tissues. J Biol Chem. 1999; 274: 12939–12944.
- ↵Schwartz SM, Heimark RL, Majesky MW. Developmental mechanisms underlying pathology of arteries. Physiol Rev. 1990; 70: 1177–1209.
- ↵Majesky MW, Giachelli CM, Schwartz SM, Reidy MA. Rat carotid neointimal smooth muscle cells re-express a developmentally regulated phenotype during repair of arterial injury. Circ Res. 1992; 71: 759–768.
- ↵Kuro, Nagai R, Nakahara K, Katoh H, Tsai RC, Tsuchimochi H, Yazaki Y, Ohkubo A, Takaku F. cDNA cloning of a myosin heavy chain isoform in embryonic smooth muscle and its expression during vascular development and in arteriosclerosis. J Biol Chem. 1991; 266: 3768–3773.