DNA Microarray Profiling to Identify Angiotensin-Responsive Genes in Vascular Smooth Muscle Cells
Potential Mediators of Vascular Disease
Angiotensin II (Ang II) induces changes in vessel structure by its capacity to activate genes that are coupled to signaling pathways such as extracellular signal–regulated kinase (ERK), p38, and phosphatidylinositol 3-kinase (PI3K). Using a DNA microarray containing 5088 genes and expressed sequence tags, we initially established a database of replicated experiments (n=4) to define the variances in mRNA expression in response to Ang II versus vehicle treatment. We observed a wide range of values for the coefficients of variation in a gene-specific manner. Guided by power calculations, we used statistical inference on a sufficient number of experimental replicates to minimize the number of false-negatives and define a subset of Ang II–responsive genes (P<0.05). To further characterize the molecular circuitry that couples Ang II stimulation with mRNA expression, we assessed expression profiles in the presence and absence of inhibitors of ERK, p38, and PI3K. Using two different methods of computational cluster analysis, we identified a subset of six matricellular proteins (eg, osteopontin and plasminogen activator inhibitor-1) that are coordinately upregulated by Ang II via an ERK/p38-dependent pathway. In addition, these cluster analyses identified calpactins I and II as novel Ang II–responsive genes. Given that Ang II promotes vascular lesion formation, we examined whether this matricellular gene cluster was also coordinately regulated in vivo. Indeed, we demonstrate that both calpactin I and osteopontin are upregulated in response to vascular injury. Taken together, the combined use of DNA microarrays, statistical inference, and cluster analysis identified novel, coordinately regulated Ang II–responsive genes that may mediate vascular lesion formation.
A growing body of evidence indicates that vasoactive substances that exert acute effects on vessel tone also induce long-term effects on vascular structure. Angiotensin II (Ang II) promotes vascular lesion formation by inhibiting vascular smooth muscle cell (VSMC) apoptosis and stimulating VSMC growth, migration, and matrix production.1–4 The angiotensin type I receptor (AT-1R) is pleiotropic in its capacity to engage a number of signaling cascades including extracellular signal–regulated kinase (ERK), p38, phosphatidylinositol 3-kinase (PI3K)/Akt, and the JAK/STAT pathways.5–8 It is postulated that the long-term effects of Ang II on vascular structure reflect this capacity to stimulate signaling cascades that are coupled to the coordinated expression of genes that govern VSMC functions in the context of lesion formation.
Conventional experimental approaches have used the candidate gene analysis to characterize downstream mediator genes that are modulated by AT-1R activation.9–12 However, it is recognized that the molecular circuitry that regulates cell fate decisions involves an intricate network of links that coordinately regulate multiple genes. The technology of DNA microarrays provides a means of defining these gene regulatory networks that mediate pathological changes in vessel structure.
The present study begins to test the hypothesis that Ang II activates a distinctive gene expression profile that promotes pathological changes in VSMC function and vessel structure. The experimental strategy combines the use of DNA microarrays and bioinformatic tools of cluster analysis to define certain subsets of Ang II–responsive genes that are coordinately regulated. It is anticipated that this approach will identify novel downstream target genes of AT-1R activation that have implications for understanding the molecular mediators of vascular disease.
Materials and Methods
Quiescent primary cultures from adult rat aortic smooth muscle cells (RASMCs) were analyzed at a postconfluent state between passages 4 and 16. RASMCs were incubated with either Ang II (300 nmol/L) or vehicle for 6 hours. The mediator role of ERK, p38, and PI3K pathways was evaluated with the following inhibitors, as previously described: U0126 (10 μmol/L, Biomol Research Laboratory, Inc), SB 202190 (25 μmol/L, Biomol Research Laboratory), or LY 294002 (50 μmol/L, Sigma-Aldrich Co), respectively.13–15
Ang II or vehicle was added to RASMCs for 20 hours in the absence or presence of pathway inhibitors or their vehicle as described. [3H]Leucine (1 μCi/mL) was incubated for the last 4 hours of Ang II treatment. Trichloroacetic acid precipitated cells were solubilized in 0.25 N NaOH. Samples were diluted in scintillation liquid, and radioactivity was measured in a β-counter.
Rat Carotid Artery Balloon Injury
Male Sprague-Dawley rats (8 animals, 300 to 400 g; Harlan Sprague-Dawley, Indianapolis, Ind) were submitted to balloon injury according to previously described methods,16 in accordance with a protocol approved by the Standing Committee on Animals, Morehouse School of Medicine. Vessels were harvested at different time points for mRNA analysis. Injured vessels were compared with their contralateral controls.
DNA Microarrays and Cluster Analysis
Analysis of rat mRNA expression levels was performed using DNA microarray membranes (GeneFilters, GF 300, Invitrogen Corp). [33P]dCTP (New England Nuclear, 3000 μCi/mol, 100 μCi/sample) radiolabeled cDNA was obtained from 3 μg of RASMCs total RNA by conventional reverse-transcriptase reaction (Superscript II, GLT). Membrane hybridizations were performed according to the manufacturer’s protocol. Intensifying screens were exposed to the membranes and images were acquired in the Phosphorimager Cyclone. Images were processed through the Pathways 3 software (Invitrogen). This software generated normalized intensity values for each gene in each microarray, using the overall mean intensity of the array as the normalization standard. These values were averaged in each experimental group. Ratios between averages for the same gene in two different groups were then calculated, representing the relative expression levels of a certain gene under specific conditions. Hierarchical cluster analysis was performed by the software Cluster.17,18 Ratios were log2-transformed and filtered to focus on genes that were significantly regulated compared with baseline. In this process, genes that did not have their expression modified by 1.5- or 2.0-fold by any of the perturbations were eliminated from the clustering process. Numerical data were processed by the software TreeView18 and presented as colored images. A baseline value of 0 was introduced as a reference. To confirm hierarchical cluster analysis data, self-organizing map (SOM) analysis was also performed utilizing the same matrix of data. To include the potential influence of gene expression variability across different experiments, cluster analysis was also performed with the log10 normalized intensity values for each gene in each microarray (total=40, including controls), subtracted from the average normalized intensity values for the control group and subjected to the filtering process described above.
Quantitative Real-Time RT-PCR (QRTPCR)
Total RNA from cell pellets or pulverized arteries was extracted (Rneasy kit, Qiagen Inc), and reverse-transcriptase reaction (RT-PCR kit, Clontech Laboratory, Inc) was performed with 0.5 to 1 μg of DNAse I (Ambion)–treated RNA. QRTPCR was carried out using the LightCycler thermocycler and the SYBR green I kit (Roche Diagnostics Corp), according to the manufacturer’s recommendations. Specific primer sequences can be found in the online data supplement, available at http://www.circresaha.org. Cycle numbers obtained at the log-linear phase of the reaction were plotted against a standard curve prepared with serially diluted control samples. Expression of target genes was normalized by GAPDH levels.
In [3H]leucine incorporation and QTRPCR experiments, comparisons were analyzed via unpaired Student’s t test (for two groups; P<0.05) or ANOVA followed by Student-Newman-Keuls test (for three or more groups; P<0.05). Paired t test was used for comparisons between normal and injured vessels. Results were presented as mean±SEM. At least 4 different samples were analyzed in each experimental group. Microarray data statistical analysis between the control and Ang II groups was performed through unpaired t test with unequal variance (using log10 normalized intensity values) or Wilcoxon-Mann-Whitney rank sum test. Probability values less than 0.05 were considered statistically significant. Power analysis for t test based on averages and standard deviations obtained from an initial set of experiments was used to define the necessary number of replicates for specific gene regulation detection. A statistical power threshold (1−β) of 0.8 was used.
Ang II Induces VSMC Growth: Signal Transduction Pathways
Initial studies confirmed that Ang II stimulates an increase in protein synthesis in RASMCs. This growth stimulatory effect is significantly attenuated by pharmacological blockade of the ERK, PI3K, or p38 signaling pathways (data not shown). These findings suggested that these interconnected pathways are all coupled to the overall trophic effects of Ang II on VSMC biology. Given the similarity of the effect of each blocker on the Ang II–stimulated growth response, we examined whether this is reflected in coordinate changes in the global pattern of gene expression using DNA microarrays.
DNA Microarray Analysis in VSMCs
To define the necessary sample size for a specific and sensitive analysis, we chose to perform an initial series of experiments designed to identify the variation of gene expression in our experimental model system. Thus, we examined the variance of mRNA expression levels determined by the DNA microarray technique in 4 sets of control versus Ang II samples. We then performed power analysis for t test based on averages and standard deviations obtained from those samples. Because there was a relatively good correlation between the coefficients of variation obtained for individual genes in both the control and Ang II–treated groups (r=0.77; data not shown), we considered similar degrees of variance for samples before and after Ang II stimulation. The power analysis demonstrated that with this particular sample of 4 sets of microarrays we would be able to detect 75% of genes regulated by ≥2.0-fold in the presence of Ang II, at P<0.05. In addition, only 31% of genes regulated would be identified if we set the threshold for a ≥1.5-fold change. To identify all genes possibly modified by ≥2.0-fold at P<0.05, we would need to have 28 control–Ang II sets of microarrays. In an attempt to balance both statistical power and experimental practicality, our final collection of experimental data represents a total of 20 microarrays (11 control and 9 Ang II). With this sample size, it was predicted that we could identify 96% and 61% of genes regulated by ≥2.0- or ≥1.5-fold after Ang II administration, respectively. One fundamental issue about microarray statistical analysis relates to the application of multiple tests and the probability value threshold. Different strategies have been used to correct probability value threshold, generating different levels of statistical stringency. Bonferroni correction, for instance, utilizes effective probability value=nominal probability value/number of tests (in our case, 0.05/5088=9.6e−10). According to the results of power analysis, if we applied that criterion, we would miss 99% and 99.8% of genes regulated by ≥2.0- or ≥1.5-fold, respectively, even with a sample size of 11 controls and 9 Ang II samples.
Statistical analysis by t test indicated that 91 genes were significantly regulated by Ang II. The rank sum test defined a set of 97 genes that were significantly regulated (P<0.05). Seventy-five of the Ang II–modulated genes defined by the rank sum test were also significantly altered as determined by the t test. Tables 1 and 2⇓ present the genes significantly modified by ≥1.5-fold after Ang II treatment, according to t and rank sum tests, respectively (to see the complete lists of regulated genes, please refer to the online data supplement).
To confirm the gene regulation determined by the statistical analysis of the microarray data, we performed QRTPCR of 9 selected genes representing the range of regulation listed in Tables 1 and 2⇑. As shown in Table 3, we observed strikingly similar patterns of mRNA regulation as determined by both microarray and QRTPCR techniques. Moreover, it is notable that the modest 1.5- and −1.5-fold changes in calpactin II and metallothionein mRNA levels, respectively, detected by microarray were confirmed as 1.8- and −1.8-fold changes by QRTPCR. Although there appears to be a close correlation between quantitative changes in mRNA expression levels on the DNA microarray and QRTPCR, we also found two genes (Jagged-1 and ID3) for which the magnitude of the regulation according to the last method was considerably more intense. This finding possibly relates to the fact that downregulated genes expressed at the lower limits of detection on microarrays may actually exhibit more substantial decreases in mRNA levels when assessed by the more sensitive method of QRTPCR.
Ang II–Responsive Gene Expression Profiles: Cluster Analysis
Having defined the Ang II–stimulated transcriptome based on statistical inference from multiple replications of DNA microarrays, we began to determine whether there were certain patterns of gene expression that correspond to the activation of ERK, PI3K, and p38. Thus, we used cluster analysis of DNA microarray data obtained from samples of RASMCs exposed to 8 different treatments: vehicle (n=11), Ang II (n=9), Ang II in the presence of ERK (n=4), PI3K (n=4), or p38 (n=4) blockers or each blocker alone in the absence of Ang II stimulation (3, 3, and 2 filters under PI3K, p38, and ERK inhibition, respectively). After filtering the matrix of data (5088 genes) with a threshold cutoff of 2-fold change, a group of 1572 genes and expressed sequence tags (ESTs) was available for cluster analysis. From the various clusters identified by the computerized algorithm, we were particularly intrigued by one defined by a profile in which Ang II stimulates an upregulation of mRNA expression that is mediated primarily by ERK and p38 pathways (Figure 1A). The 6 genes within this cluster appear to function as components of the apparatus that links the cytoskeleton to and/or modulates the composition of the extracellular matrix. To further minimize the possibility that this cytoskeleton/matricellular cluster was generated by arbitrary prespecifications, we reevaluated the dataset with a lower filtering threshold to include genes that are modulated by 1.5-fold instead 2-fold (sample of 3243 genes and ESTs). As shown in Figure 1B, this strategy resulted in the inclusion of a few more genes (total=10) but failed to disrupt the original elements of that particular cluster. To further define the determinants of this pattern, we removed the elements in the matrix related to the blockade of the ERK pathway. It is noteworthy that in this new cluster, 5 of the 6 branches of the original one remained intact (Figure 1C). We then examined whether the grouping of these 6 genes within the universe of 1572 genes was artificially generated based on the inherent assumptions of that particular clustering algorithm. To address this issue, we reexamined the same dataset using SOM analysis.19 We observed a similar cluster of this subset of genes, where 5 genes of the cluster depicted in Figure 1A were maintained and 1 gene, annexin I, was replaced by another matrix-related gene named Gla (data not shown). In addition, to introduce gene expression variance as a variable for the cluster analysis, we also performed hierarchical cluster analysis using a matrix containing the normalized intensity values for each gene (instead of ratios between groups of experiments). After an initial data-filtering step, we analyzed 1567 genes and obtained a very similar cluster, containing plasminogen activator inhibitor-1 (PAI-1), osteopontin, calpactin I light chain (LC), tissue inhibitor of metalloproteinase-1 (TIMP-1), and matrix Gla protein (see online data supplement).
The microarray analysis suggested that the mitogen-activated protein kinases (MAPKs)—ERK and p38—may play a mediator role in modulating PAI-1 and osteopontin expression in VSMCs. To address this hypothesis, a second series of QRTPCR experiments were performed with samples from cells treated with Ang II and/or pharmacological blockers to clarify the mediators involved in this phenomenon. As shown in Figure 2A, Ang II stimulated a 9-fold increase in osteopontin mRNA levels that was abolished by pretreatment with either the p38 or ERK blockers (P<0.001). Similarly, Ang II stimulated a 3-fold increase in PAI-1 mRNA levels (Figure 2B). However, blockade of either p38 or ERK resulted in substantial declines in baseline PAI-1 mRNA levels such that the overall stimulatory effect of Ang II in the setting remained below the baseline values. Thus, the ERK/p38 signaling pathways are important determinants of PAI-1 expression in VSMCs in the presence and absence of Ang II.
The present study also defined TIMP-1, calpactin I LC (annexin II ligand, p11), and calpactin II (annexin I, lipocortin) as Ang II–responsive genes in VSMCs. Indeed, we are unaware of previous reports that have documented calpactin I LC and calpactin II expression in VSMCs. To verify these DNA microarray data, we performed QRTPCR analysis as described above. As shown in Figure 2C, Ang II stimulated a 2-fold increase in TIMP-1 mRNA levels that is abolished by either ERK or p38 blockade (P<0.05). Similarly, QRTPCR analysis confirmed that Ang II stimulated 2.5-fold and 1.8-fold increases in calpactin I LC and calpactin II mRNA levels, respectively, that were selectively abolished by ERK blockade (P<0.05) (Figures 2D and 2E). To corroborate these findings, another series of experiments was performed using a chemically dissimilar ERK inhibitor, PD 98059 (Promega). As observed in the previous series of studies, Ang II induced an upregulation of calpactin I LC and calpactin II that was abolished by ERK blockade with PD 98059 (n=4, P<0.05; data not shown). Osteopontin mRNA levels were also analyzed in this new set of experiments as a positive control, and a response similar to that observed in the presence of U0126 compound was obtained (data not shown). Additionally, to better define the response profile in RASMCs, we evaluated mRNA expression levels for calpactins in a time-course fashion after Ang II stimulation. We demonstrated that the significant upregulation of calpactin I LC observed at 6 hours was maintained after 12 hours (2.0- and 2.2-fold change at 6 and 12 hours, respectively; n=7 to 10, P<0.001), with mRNA levels returning to baseline after 24 hours (n=9). A similar profile was observed for osteopontin mRNA levels (data not shown). Calpactin II mRNA levels demonstrated a nonstatistically significant 1.3-fold change (n=7 to 8) after 12 hours of Ang II administration, in contrast to the significant 1.6-fold change demonstrated at 6 hours (n=10, P<0.005). Of interest, calpactin II mRNA levels were significantly decreased at 24 hours compared with basal values (−2.1-fold change, n=9, P<0.005). Thus, these independent series of experiments using QRTPCR confirmed that a common ERK-dependent signal transduction pathway modulates this novel cluster of Ang II–responsive genes. Taken together, these studies support the selection of 6 hours as an optimal time point for the analysis of this particular set of genes.
Ang II–Responsive Gene Expression Profiles: Implications for Vascular Lesion Formation
We hypothesized that the cluster analysis of gene expression profiles would define a subset of genes that is coordinately regulated to exert trophic effects on VSMCs and modulate vessel structure. Three of the 6 known genes in the matricellular cluster (ie, osteopontin, PAI-1, and TIMP-I) had been previously described as genes that are upregulated in the context of vascular lesion formation. To begin to explore whether this relationship to vascular injury is a characteristic feature shared by other members of this cluster, we examined the mRNA expression of calpactin I LC in specimens derived from the rat carotid artery injury model. In accord with this hypothesis, Figure 3 documents that both calpactin I LC and osteopontin exhibit a significant upregulation in mRNA expression within 1 week after vascular injury (P<0.05).
A large body of evidence derived from both animal models as well as clinical trials indicates that Ang II promotes vascular lesion formation.20,21 In addition, Ang II promotes alterations in vessel structure by inhibiting VSMC apoptosis and stimulating VSMC growth, migration, and matrix production.1–4 This study confirmed that Ang II–stimulated activation of the AT-1R promotes VSMC growth via the activation of signaling cascades that include ERK, p38, and PI3K/Akt. Although each of these signaling pathways is coupled to the activation of transcription factors, Ang II–responsive genes downstream of these pathways remained to be defined. The objective of this study was to utilize DNA microarray technology coupled with bioinformatic tools of cluster analysis to define novel Ang II–responsive genes that may mediate its effects on vascular lesion formation.
A commonly used approach to use DNA microarrays is to define the transcriptome by an arbitrary fold change in mRNA expression. In addition, most analyses appear to presume a common level of variance across the thousands of genes being assayed and generally do not perform more than a few experimental replicates to verify that these changes in gene expression are statistically significant. We performed power analysis with our initial dataset of 4 arrays to guide the design of replicate experiments that would optimize gene discovery while minimizing both false-positive and false-negative results. Using this approach, we determined that with the number of arrays we performed, a 2-fold change in mRNA levels could be detected at the P<0.05 level for the majority of the genes analyzed. In addition, our analysis of the number of genes identified as Ang II–responsive by an arbitrary fold-change criterion versus statistical inference suggests that the nonstatistical approach is vulnerable to generation of both false-positives and false-negatives in the context of limited replications. As is the case with all experimental data, independent confirmatory analyses, such as QRTPCR, need to be performed, and the final arbiter of Ang II–responsive relevance is biology.
The analysis of microarrays using statistical inference remains a poorly defined area in which further development and investigation are required. It is recognized that there are limitations to the use of the rank sum and t tests that render these approaches less than optimal. One concern is the problem of multiple testing when 5000 genes are being analyzed. However, we observed that conventional approaches to account for this problem such as the Bonferroni correction were far too stringent and would result in many false-negative results. Nevertheless, the validity of the approach used was verified by a series of additional replication experiments in which the findings of the microarray analysis were confirmed by QRTPCR analysis. Indeed, there was a striking correlation between the fold changes in normalized intensities determined by microarray compared with QRTPCR. Furthermore, we were reassured by the detection of genes that were previously defined as Ang II–responsive.22–26 These studies indicated that genes noted to be significantly upregulated by 1.5-fold in response to Ang II on microarray (P<0.05) could be reliably confirmed as Ang II–responsive in subsequent experiments using QRTPCR. Thus, an overly conservative criterion for the transcriptome such as a 3-fold change could have resulted in a substantial number of false-negative candidate mediators worthy of further investigation.
To further characterize the subsets of gene profiles within the Ang II–responsive transcriptome, we used computational cluster analysis. Based on the observation that Ang II–induced growth effects are mediated by ERK, p38, and PI3K/Akt, we used pharmacological blockade of each of these pathways and assessed the effects on the gene expression profile. This study tested the postulate that this matrix of conditions would reveal gene expression profiles that may be coordinately regulated and functionally linked to this Ang II–stimulated signaling cascade. Genes from one particular cluster defined by this algorithm shared features consistent with Ang II–induced upregulation and a dependence on the integrity of the ERK/p38 pathway. Our finding that the same cluster was identified using a distinctly different computational algorithm further verified the integrity of the interrelationships between these genes and the signaling networks. This cluster included genes that were previously defined as Ang II–responsive, such as osteopontin and PAI-1,27,28 as well as the novel Ang II–responsive genes, such as the calpactins I and II. It is noteworthy that many members of this cluster share the functional characteristics of being associated with the cytoskeleton (myosin LC) or matricellular proteins that are part of the link between the extracellular matrix (eg, osteopontin, TIMP-1, and PAI-1) and the cytoskeleton (eg, calpactin I LC and calpactin II). We speculate that these genes may be coordinately regulated by this signaling pathway to promote matrix remodeling by Ang II. Given the role of Ang II and this signaling pathway in the process of lesion formation, we examined whether members of this cluster may also be coordinately regulated in the context of vascular injury. Indeed, we established for the first time that calpactin I LC is expressed in the vasculature. Moreover, these studies documented a coordinate upregulation of both calpactin I LC and osteopontin at 7 days after injury. Additional studies are needed to further clarify the role of calpactins I and II as potential mediators of the effects of Ang II on vascular structure.
Taken together, these studies are consistent with the potential use of microarrays and cluster analysis to identify novel genes that are coordinately regulated and potentially interrelated by function. Despite the recognized utility of microarrays, our findings also demonstrate the importance of experimental replication and an appreciation for variance in the analysis of gene expression profile data. It is anticipated that advances in statistical approaches and bioinformatic tools will further expand the full usefulness of the microarray technology as a tool for functional genomics.
Alexandre H. Campos is supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (São Paulo, Brazil). Dr Gibbons’ laboratory is supported by grants from the National Heart, Lung, and Blood Institute (NHLBI), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Center on Minority Health and Health Disparities, the Research Centers in Minority Institutions (RCMI) program, and the American Heart Association. Dr Pollman’s laboratory is supported by grants from the National Institutes of Health Minority Biomedical Research Support (NIH MBRS) program, the RCMI program, and an NIH-K08 award. We gratefully acknowledge the statistical consultative support provided by Karen Vranizan, Functional Genomics Laboratory, University of California, Berkeley.
Original received October 19, 2001; resubmission received October 9, 2002; accepted November 13, 2002.
Pollman MJ, Yamada T, Horiuchi M, Gibbons GH. Vasoactive substances regulate vascular smooth muscle cell apoptosis: countervailing influences of nitric oxide and angiotensin II. Circ Res. 1996; 79: 748–756.
deBlois D, Lombardi DM, Su EJ, Clowes AW, Schwartz SM, Giachelli CM. Angiotensin II induction of osteopontin expression and DNA replication in rat arteries. Hypertension. 1996; 28: 1055–1063.
Geisterfer AA, Peach MJ, Owens GK. Angiotensin II induces hypertrophy, not hyperplasia, of cultured rat aortic smooth muscle cells. Circ Res. 1988; 62: 749–756.
Griendling KK, Ushio-Fukai M, Lassegue B, Alexander RW. Angiotensin II signaling in vascular smooth muscle: new concepts. Hypertension. 1997; 29: 366–373.
Ushio-Fukai M, Alexander RW, Akers M, Griendling KK. p38 Mitogen-activated protein kinase is a critical component of the redox-sensitive signaling pathways activated by angiotensin II: role in vascular smooth muscle cell hypertrophy. J Biol Chem. 1998; 273: 15022–15099.
Ushio-Fukai M, Alexander RW, Akers M, Yin Q, Fujio Y, Walsh K, Griendling KK. Reactive oxygen species mediate the activation of Akt/protein kinase B by angiotensin II in vascular smooth muscle cells. J Biol Chem. 1999; 274: 22699–22704.
Taubman MB, Berk BC, Izumo S, Tsuda T, Alexander RW, Nadal-Ginard B. Angiotensin II induces c-fos mRNA in aortic smooth muscle: role of Ca2+ mobilization and protein kinase C activation. J Biol Chem. 1989; 264: 526–530.
Chen XL, Tummala PE, Olbrych MT, Alexander RW, Medford RM. Angiotensin II induces monocyte chemoattractant protein-1 gene expression in rat vascular smooth muscle cells. Circ Res. 1998; 83: 952–959.
Mitani Y, Zaidi SH, Dufourcq P, Thompson K, Rabinovitch M. Nitric oxide reduces vascular smooth muscle cell elastase activity through cGMP-mediated suppression of ERK phosphorylation and AML1B nuclear partitioning. FASEB J. 2000; 14: 805–814.
Cho A, Graves J, Reidy MA. Mitogen-activated protein kinases mediate matrix metalloproteinase-9 expression in vascular smooth muscle cells. Arterioscler Thromb Vasc Biol. 2000; 20: 2527–2532.
Bai H, Pollman MJ, Inishi Y, Gibbons GH. Regulation of vascular smooth muscle cell apoptosis: modulation of Bad by a phosphatidylinositol 3-kinase–dependent pathway. Circ Res. 1999; 85: 229–237.
Morishita R, Gibbons GH, Ellison KE, Nakajima M, Zhang L, Kaneda Y, Ogihara T, Dzau VJ. Single intraluminal delivery of antisense cdc2 kinase and proliferating-cell nuclear antigen oligonucleotides results in chronic inhibition of neointimal hyperplasia. Proc Natl Acad Sci U S A. 1993; 90: 8474–8478.
Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A. 1998; 95: 14863–14868.
Stanford University software. Available at: http://rana.lbl.gov/ EisenSoftware.htm. Accessed November 20, 2002.
Morishita R, Gibbons GH, Tomita N, Zhang L, Kaneda Y, Ogihara T, Dzau VJ. Antisense oligodeoxynucleotide inhibition of vascular angiotensin-converting enzyme expression attenuates neointimal formation: evidence for tissue angiotensin-converting enzyme function. Arterioscler Thromb Vasc Biol. 2000; 20: 915–922.
Yamashita J, Itoh H, Ogawa Y, Tamura N, Takaya K, Igaki T, Doi K, Chun TH, Inoue M, Masatsugu K, Nakao K. Opposite regulation of Gax homeobox expression by angiotensin II and C-type natriuretic peptide. Hypertension. 1997; 29: 381–387.
Boulanger CM, Heymes C, Benessiano J, Geske RS, Levy BI, Vanhoutte PM. Neuronal nitric oxide synthase is expressed in rat vascular smooth muscle cells: activation by angiotensin II in hypertension. Circ Res. 1998; 83: 1271–1278.
van Leeuwen RT, Kol A, Andreotti F, Kluft C, Maseri A, Sperti G. Angiotensin II increases plasminogen activator inhibitor type 1 and tissue-type plasminogen activator messenger RNA in cultured rat aortic smooth muscle cells. Circulation. 1994; 90: 362–368.