| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
and ß Mediate Distinct Pathways of Vascular Gene Expression, Including Genes Involved in Mitochondrial Electron Transport and Generation of Reactive Oxygen SpeciesDepartment of Biology (R.O., U.H.), Program in Bioinformatics (M.E.S., U.H.), and Department of Biomedical Engineering (J.B.), Boston University, Boston, Massachusetts 02215; Molecular Cardiology Research Institute (K.K., I.Z.J., R.H.K., M.E.M.), Tufts-New England Medical Center, Boston, Massachusetts 02111; CSAIL (J.B.), Massachusetts Institute of Technology, Cambridge, Massachusetts 02139; and Institute of Endocrinology (P.G.V.M.), University of Milan, 20133 Milan, Italy
Address all correspondence and requests for reprints to: Dr. Ulla Hansen, 5 Cummington Street, Boston Massachusetts 02215. E-mail: uhansen{at}bu.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(ER) and ß (ERß), which are both expressed in vascular smooth muscle and endothelial cells. However, the molecular pathways mediating estrogen effects in blood vessels are not well defined. We have performed gene expression profiling in the mouse aorta to identify comprehensive gene sets the expression of which is regulated by long-term (1 wk) estrogen treatment. The ER subtype dependence of the alterations in gene expression was characterized by parallel gene expression profiling experiments in ER-deficient [ER knockout (ERKO)] and ERß-deficient (ERßKO) mice. Importantly, these data revealed that ER- and ERß-dependent pathways regulate distinct and largely nonoverlapping sets of genes. Whereas ER is essential for most of the estrogen-mediated increase in gene expression in wild-type aortas, ERß mediates the large majority (nearly 90%) of estrogen-mediated decreases in gene expression. Biological functions of the estrogen-regulated genes include extracellular matrix synthesis, in addition to electron transport in the mitochondrion and reactive oxygen species pathways. Of note, the estrogen/ERß pathway mediates down-regulation of mRNAs for nuclear-encoded subunits in each of the major complexes of the mitochondrial respiratory chain. Several estrogen-regulated genes also encode transcription factors. Computational analysis of promoters from coexpressed genes revealed overrepresentation of binding sites for such factors, lending support for an estrogen-regulatory transcriptional network in the vasculature. Overall, these findings provide a foundation for understanding the molecular basis for estrogen effects on vasculature gene expression. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
and ß (ER, ERß), members of the steroid hormone receptor family of ligand-activated transcription factors. Upon estrogen binding, these factors interact with specific DNA response elements, resulting in modulation of gene transcription. ER and ERß can also regulate gene expression by altering the function of other classes of transcription factors through protein-protein interactions in the nucleus and/or by activation of signal transduction pathways that modulate the activity of target transcription factors (reviewed in Refs. 1 and 8). In addition, ER and ERß mediate nongenomic actions of estrogen via activation of signal transduction pathways that affect cardiovascular function including the rapid activation of endothelial nitric oxide synthase (reviewed in Ref. 9). ERs are expressed in both vascular endothelial cells and vascular smooth muscle cells (1). These two cell types demonstrate differential effects in response to estrogen exposure, which promotes proliferation and inhibits apoptosis in endothelial cells, while inhibiting proliferation and migration of vascular smooth muscle cells (2, 10, 11, 12, 13, 14, 15). A cohort of genes regulated in the vasculature in response to estrogen have been identified, including genes involved in the regulation of the vascular tone, vascular matrix formation and remodeling, cell adhesion, vascular inflammation, angiogenesis, and coagulation (reviewed in Refs. 1 and 2).
The distinct physiological roles of the two ER subtypes in vascular function have been investigated in ER and ERß knockout (ERKO and ERßKO) mice. These mouse models reveal a role for ER in mediating the vascular protective effect of estrogen in the mouse carotid injury model (3, 4). ERß has a distinct role in controlling systemic blood pressure, in part by modulating endothelial-independent vasodilation and potassium ion channel function in vascular smooth muscle cells (4). However, specific target genes regulated by ER vs. ERß, as well as their distinct functions in vascular tissues, have yet to be fully investigated. A more complete understanding of the molecular mechanisms by which ER and ERß alter vascular physiology requires understanding of the gene-regulatory pathways that they instigate. Such information may permit design of subtype- or pathway-specific ER ligands to be used in the prevention and treatment of common vascular diseases.
The purpose of the present study is to identify a comprehensive set of genes and biological pathways regulated downstream of ER and ERß in the vasculature after steady-state estrogen treatment. We therefore investigated the patterns of gene expression in aortas from mice treated with and without 17ß-estradiol (E2) for 1 wk. Gene expression profiles from wild-type (WT) mice were also compared with the profiles of aortas from complete ERKO or ERßKO mice. Here we report profound and fundamental differences in gene regulation by the two ERs, ER and ERß, which begin to elucidate the molecular mechanisms involved in the important but distinct roles of these two receptors in vascular function and disease.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
vs. ERß in Regulation of Aortic Gene Expression
. In WT mice aortas, approximately twice as many estrogen-responsive genes are up-regulated vs. down-regulated by hormone treatment. Importantly, results in ERßKO mice revealed a similar ratio. These largely reflect ER-mediated gene-regulatory pathways (see below). In contrast, most of the estrogen-responsive genes in ERKO mice are down-regulated, largely representing ERß-mediated pathways (see below). Despite significant differences in the direction of gene regulation, similar numbers of genes are controlled downstream of ER vs. ERß (Fig. 1A). When E2-regulated genes are identified under less stringent threshold criteria, thereby including more genes, the relative proportions of genes induced or inhibited by estrogen in the different genotypes remain similar (Table S1 published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org).
|
|
). A total of 15 representative genes, both up-regulated and down-regulated by estrogen and in different genotypes, displayed qualitatively and quantitatively comparable levels of regulation as in the global gene expression analyses.
|
vs. ERß in regulating individual gene expression, the estrogen-regulated genes (at least 2-fold change in expression, P < 0.01) in each genotype were plotted by rank order; genes regulated in WT, ERKO, and ERßKO aortas are represented as the monotonically descending sets of points in Fig. 1
, panels B, C, and D, respectively. In these plots, the estrogen-mediated gene expression changes over a wide range (up to 20-fold either up or down) are readily visualized by expressing fold changes in log base 2. The corresponding degree of estrogen responsiveness of each of these genes in the two other genotypes was then plotted in the same figures. When all three genotypes were compared, the estrogen responsiveness of only a single gene (Plod2) remained the same irrespective of the presence or absence of ER or ERß. Thus, estrogen regulation of gene expression in the vasculature is definitely mediated through the classical ERs. Furthermore, for most of the genes in which estrogen responses in WT and ERßKO aortas are correspondingly elevated (Fig. 1, B and D; blue diamonds and green triangles, respectively), there is little, if any, regulation by estrogen in the ERKO background (red squares), demonstrating that ER is the dominant receptor under these conditions. It must be noted that the presence of ERß does influence roughly half of these genes up-regulated by estrogen, however; whereas half of the genes show virtually identical up-regulation plus or minus ERß, the other half of the gene are diminished in their estrogen response when ERß is absent (Fig. 1, B and D, compare diamonds and triangles).
Conversely to the ER dependence of up-regulated genes, the large number of aortic genes significantly down-regulated by estrogen in ERKO mice (Fig. 1C) are in general unresponsive to estrogen in ERßKO animals, indicating that ERß is critical for reducing gene expression in the presence of estrogen. These genes also tend to be down-regulated to a lesser extent in WT mice than in ERKO mice, implicating opposing functions for ER (in enhancing expression) and ERß (in reducing expression) for this particular pathway, discussed further below.
As a quantitative measure of the relative importance of ER vs. ERß in mediating estrogen effects in normal aorta, the degree of similarity between gene expression changes in the different genotypes was statistically analyzed. Considering the set of all 852 genes that were significantly regulated (greater than 2-fold; P < 0.01) in the entire microarray data set (see Materials and Methods), the WT and ERßKO responses to estrogen treatment are highly similar (Pearson correlation coefficient of 0.85), whereas the WT and ERKO responses are not (0.27). Finally, the ERKO and ERßKO responses are even more distinct (0.12), showing markedly nonoverlapping consequences of the two ER subtypes.
We systematically grouped gene expression patterns in response to estrogen in the three genotypes using cluster analysis (16). Based on the patterns of estrogen regulation in ERKO, ERßKO, and WT animals, four overall sets of regulated genes were distinguished (Table 3). The coregulated clusters included: 1) genes up-regulated by estrogen in both WT and ERßKO aortas with little or no change in ERKO aortas; therefore, gene expression enhanced by ER, but less affected by ERß; 2) genes down-regulated by estrogen in both WT and ERßKO aortas with little or no change in ERKO aortas; therefore, gene expression diminished by ER, but less affected by ERß; 3) genes down-regulated in both ERKO and WT aortas with no change in ERßKO aortas; therefore, gene expression diminished by ERß, but relatively unaffected by ER; and 4) genes up-regulated by estrogen in ERKO animals, with no change in either of the other genotypes; therefore, gene expression enhanced by ERß, and apparently countered by ER. The cluster analysis is therefore consistent with the qualitative patterns and conclusions inferred from Fig. 1, with most estrogen-responsive genes in the ER pathway being induced (compare clusters 1 and 2) and most estrogen-responsive genes in the ERß pathway being inhibited (compare clusters 3 and 4). Complete lists of genes comprising clusters 1–4 are provided in supplemental Tables S3–S6.
|
KO and ERßKO mice were also compared with WT gene profiles in the absence of E2. Interestingly, the numbers of genes with elevated expression levels exceeded the numbers of genes with diminished levels in the knockout mice, relative to WT levels (supplemental Table S2) mainly due to up-regulation of gene expression in ERKO mice. Figure 2 depicts the number of genes with stringently altered expression levels in ERKO or ERßKO mice as compared with WT mice (at least 2-fold differences in expression levels; P < 0.01). Most gene expression differences were evident in the ERKO aortas (42 genes), with only a few in the ERßKO aortas (five genes); there was no overlap between these gene sets. When genes differentially expressed in ERKO and ERßKO aortas in the absence of estrogen were directly compared, a much larger set of 208 genes (including the 47 genes mentioned above) were identified, where the difference was at least 2-fold with P < 0.01 (supplemental Fig. S1). Given that many more genes are differentially expressed when the two ER subtype-deficient mice are contrasted than when either individual ER-deficient strain with WT mice are contrasted, these data further support that the two ER subtypes regulate genes in opposite directions.
|
) characterize functionally distinct groups of genes, each gene set was probed for statistical overrepresentation of Gene Ontology (GO) biological and molecular categories (17). ER pathway up-regulated genes (cluster 1) displayed significant overrepresentation in 22 GO categories (supplemental Table S7), with the greatest numbers of genes per category included in the broad categories of development and cell differentiation, and the more specific categories of extracellular space and matrix (Table 4). Representative genes in such categories encode proteins such as IGF binding protein 6, very-low-density lipoprotein receptor, GH receptor, and frizzled-related protein. ER pathway down-regulated genes (cluster 2) were overrepresented in 14 GO categories also including extracellular region as well as the subcategory collagen (Table 4), of which genes encoding procollagen type I 1 and 2 are included (supplemental Table S8). The cluster of 13 genes up-regulated downstream of ERß (cluster 4) does not show GO overrepresentation by such analyses. Of all the clusters, functional analysis was most significant and striking in the group of genes down-regulated in the ERß pathway (cluster 3), which proved to be overrepresented in 49 GO categories (supplemental Table S9), with a concentration in mitochondrial genes involved in oxidative phosphorylation and electron and hydrogen transport (Table 4). These gene sets are characterized further below. In general, and consistent with the minimal overlap between genes regulated in each set (Fig. 1), the GO clustering revealed an inverse functional relationship between genes in the ER-regulated vascular gene set (in WT and ERßKO mice) vs. the ERß-regulated gene set (in ERKO mice).
|
KO aortas, 10 overrepresented GO categories were identified, including carbohydrate catabolism (supplemental Table S10). However, for the group of five genes with altered expression levels in ERßKO mice, no GO categories were overrepresented.
Genes Encoding Mitochondrially Localized Proteins and Oxidoreductases Are Regulated by Estrogen in Mouse Aorta
As indicated above, the cluster of genes down-regulated in response to estrogen in both ERKO and WT mouse aortas (cluster 3, consisting of 92 genes), and more specifically, the set of genes most extensively down-regulated (by 2-fold or more) in ERKO mice (Fig. 1C, consisting of 71 genes) were significantly enriched in nuclear-encoded, mitochondrion-related genes (16 genes) and mitochondrial membrane genes (seven genes; Table 5). In addition, within the gene set down-regulated in ERKO mice by estrogen, a molecular function category that was notably overrepresented was that of the oxidoreductases (Table 5 and supplemental Table S11). Of the 11 oxidoreductase-encoding genes down-regulated by estrogen in ERKO mice, seven overlap the set of mitochondrial-related genes [Ndufa7: reduced nicotinamide adenine dinucleotide (NADH) dehydrogenase (ubiquinone) 1 subcomplex, 7 (B14.5a); Ndufs8: NADH dehydrogenase (ubiquinone) Fe-S protein 8; Ndufc1: NADH dehydrogenase (ubiquinone) 1; Sdhb: Succinate dehydrogenase complex, subunit B, Fe-S (Ip); Qcr9: RIKEN cDNA 1110020P15 gene (Ortholog: ubiquinol-cytochrome c reductase complex); Cox8b: cytochrome c oxidase, subunit VIIIb; and Hsdl2: hydroxysteroid dehydrogenase like 2]. In contrast, none of the oxidoreductases in the WT up-regulated gene set are grouped within the mitochondrial category. Oxidoreductases that are differentially expressed between genotypes in the absence of estrogen were also identified (supplemental Table S12).
|
(in WT and ERßKO mice) represent proteins involved in a wide variety of metabolic pathways. In contrast, six of the 11 oxidoreductases the mRNA levels of which are down-regulated by estrogen via ERß (in ERKO mice) are involved in oxidative phosphorylation. These oxidoreductases comprise subunits within four of the five mitochondrial respiratory chain (MRC) complexes embedded within the inner mitochondrial membrane (Table 6; genes in bold with superscript letter b). This gene set also included two additional genes in the oxidative phosphorylation category: Atp5o (ATP synthase, H+ transporting, mitochondrial F1 complex, O subunit), a member of the fifth complex of the MRC (Table 6), and Immp21 (inner mitochondrial membrane peptidase 2-like).
|
KO mice. At less stringent classification criteria (no fold threshold; P < 0.01 either upon estrogen stimulation or comparing two genotypes), nine additional genes representing subunits of Complexes I, III, IV, and V prove to be down-regulated in response to estrogen in ERKO mice (Table 6; genes not in bold). Therefore, mRNA levels encoding 28% of the subunits of the MRC, including subunits in every major complex, are down-regulated by estrogen treatment of ERKO aortas. All MRC-related genes initially identified under stringent criteria, as well as these additional genes, are also down-regulated to a lesser extent in response to estrogen in WT mouse aortas (Table 6; note that many did not meet the more stringent threshold criteria). Thus, ERß mediates coordinate down-regulation of subunits in each of the MRC complexes in response to estrogen treatment in aortic tissue.
Other Reactive Oxygen Species (ROS)-Regulating Genes Affected by Estrogen Treatment in Mouse Aorta
ROS are produced in the blood vessel wall and, at high levels, can promote vascular dysfunction by inactivating nitric oxide, the endothelial-derived relaxation factor. Because the MRC is a major source of cellular ROS (19), down-regulation of MRC gene expression by estrogen would be predicted to decrease overall levels of cellular ROS, thereby improving endothelial function. Examination of the significantly estrogen-regulated expression profiles for other genes related to ROS regulation yielded additional instances of estrogen responsiveness (see supplemental Table S13 for specifics), including a glutathione peroxidase (see below). In an ERß-dependent manner, estrogen down-regulates expression of a scavenger receptor (Scarb1; 0.3-fold in ERKO mice), demonstrated in one report to be essential for ROS production in response to stress in neuronal tissue (20). Furthermore, ERß mediates down-regulation of two different subunits of cytosolic glutathione-S-transferase (0.3- to 0.4-fold in ERKO mice), an enzyme that can both mediate the oxidative stress response and reduce ROS by covalent addition of glutathione (21). Finally, through an ER-mediated pathway, estrogen also down-regulates gene expression of cytochrome P450 2E1 (Cyp2e1; 0.3- and 0.4-fold in WT and ERßKO mice, respectively), which generates ROS as a product of xenobiotic metabolism, and perhaps of normal pathways (22).
Whereas both ERß and ER pathways reduce gene expression for these ROS-producing proteins, ER pathways conversely enhance gene expression for two ROS-reducing proteins (Fig. 3): glutathione peroxidase 3 (Gpx3, 2.3- and 3.6-fold in WT and ERßKO mice, respectively) and synuclein (Snca, 3.0-fold in WT mice). Gpx3 reduces cytoplasmic hydrogen peroxide, a major ROS product generated by mitochondria as a product of ATP biosynthesis (23). Although Gpx3 mRNA is up-regulated in WT and ERßKO mice, it is down-regulated in ERKO mice (0.4-fold), suggesting opposing gene-regulatory effects of ER and ERß for its expression. The molecular function of Snca remains elusive; however, it can cause cells to become resistant to hydrogen peroxide, in part via down-regulation of oxidative stress signaling through the JNK pathway (24). Furthermore, Snca-deficient mice demonstrate mitochondrial membrane disruptions, and reduction in Complex I and III activities (25). Therefore, the convergence of ER pathways on regulation of oxidative species production and oxidative stress pathways suggests that this is an important biological system influenced by estrogen in vasculature.
|
; see supplemental Table S14 for details). Estrogen regulation of representative transcription factor mRNAs was validated by quantitative RT-PCR [serum response factor (SRF) and Ets1; Table 2]. Activities of transcription factors can also be altered by regulating expression of partner proteins. A search of the expression data yielded three such partner proteins the expression of which was impacted by estrogen treatment: mga, a binding partner of Max that modifies its specificity (26) and two Id proteins (27, 28, 29). Although the latter are also most directly connected with subsets of the E-box family of transcription factors (e.g. myc/max, E2A/E47), Id proteins also interact with Ets1 and other families of factors (e.g. Pax), to inhibit transcription factor activities (30, 31).
|
(32), can regulate gene expression from distant enhancer sequences, essentially all transcription factors also bind sequences within 1000 bp of transcription initiation sites. Therefore, although not comprehensive, these sequences have been widely used as a highly informative basis for relevant transcription factors for gene regulation (33, 34). Initially, six sets of genes were independently analyzed: gene sets either up-regulated or down-regulated by estrogen treatment in WT, ERKO, or ERßKO mice (gene sets shown in Fig. 1; see supplemental Tables S15 and S16, for binding site analyses).
Significantly, recognition sites for all of the estrogen-responsive DNA-binding transcription factors (Table 7) were among the binding sites shown by computational analyses to be overrepresented in promoters of estrogen-regulated genes. In many cases, the connection was specific, in that the ER subtype required for estrogen regulation of a transcription factor mirrored the ER subtype dependence of the set of promoters in which its binding sites were statistically overrepresented. Expression of Stat1 is down-regulated by the estrogen/ERß pathway (in ERKO aortas, Table 7), and its binding sites are only predicted to be enriched in estrogen-inducible promoters in this genotype (Table S16). Fox01 expression is elevated by the estrogen/ER pathway (in ERßKO and WT aortas), and FOX binding sites are enriched only in promoters that are estrogen-responsive in ERßKO aortas. Finally, mothers against DPP homolog 7 and mothers against DPP expression is diminished in an estrogen/ER-dependent manner (in ERßKO aortas), and mothers against DPP homolog 7 and mothers against DPP binding sites are only overrepresented in genes down-regulated by estrogen in this genotype.
Genomic regions upstream of the MRC and oxidoreductase gene sets that are down-regulated in ERKO mice were independently analyzed for statistically overrepresented mammalian transcription factor binding sites (supplemental Table S17). In these gene sets, binding sites were predicted for both Ets1, whose expression is decreased by estrogen pathways downstream of both ER subtypes, and SRF, whose expression is significantly regulated by estrogen only in the ERKO genotype. Finally, when the additional constraint was imposed of evolutionary conservation of binding sites between mouse and human gene upstream sequences (supplemental Table S18), binding sites for NRF1 (nuclear respiratory factor 1) emerged with statistical relevance (P < 0.05). This correlates with the regulation of NRF1 gene expression by estrogen only in the ERKO genotype.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
One critical goal of defining the molecular pathways underlying physiological estrogen effects is to tease apart the relative contributions of the two ER subtypes, ER and ERß. By analyzing ER-deficient mice, we determined aspects of estrogen regulation in the vasculature with general implications to physiological responses to estrogen, including that: 1) the classical ERs, and not other potential estrogen-binding receptors, are important mediators of estrogen function in the vasculature, as is true in other tissues (reviewed in Refs. 35 and 36); 2) ER and ERß are responsible for regulation of distinct sets of genes in the vasculature, with the function of ERß not simply being to oppose ER action; 3) a major role for ERß in the uninjured aorta is to down-regulate pathways that regulate mitochondrial function; and 4) ERs also regulate expression of specific vascular transcription factors, which are likely, in turn, to control subsequent waves of gene expression.
Distinct Roles for ER and ERß in the Vasculature
From gene expression profiling and cluster analysis (Table 3), genes up-regulated by estrogen are largely resistant to estrogen when ER is absent (Fig. 1B). Thus, ER plays the central and critical role for most genes regulated by hormone in WT aortas. This is consistent with previous genetic and pharmacological studies on global estrogen-responsive gene expression patterns in other primary target tissues, such as uterus and kidney, where ER is also thought to be the predominant receptor responsible for biological function (37, 38).
In previous studies of estrogen effects on gene expression in other primary tissues, few, if any, differences were attributable to ERß (37, 38, 39). However, as demonstrated repeatedly in our study, this is not the case in the aorta, which is either qualitatively different in this regard from other tissues, or where the statistical character of our dataset is sufficiently robust to reveal contributions by ERß. As stated in Results, only half of the genes significantly up-regulated by estrogen in WT aorta are regulated to the same extent in the absence of ERß, with the other half responding less avidly (Fig. 1, B and D). Gene clustering analysis, when allowing for a larger number of gene clusters, supports this distinction between two subsets of these estrogen-up-regulated genes (data not shown). This suggests that, whereas ER is both necessary and sufficient for full induction by estrogen for some genes, for another group of genes, ER and ERß act in concert to stimulate gene expression. This ability of the two ER subtypes to act in an additive fashion does not fit the proposed model that the major role for ERß is to oppose the function of ER, at least for the vasculature (38, 40).
A separate vascular role for ERß was initially revealed in the ERKO background, in which a large group of genes are seen to be down-regulated by estrogen, including several nuclear-encoded mitochondrial genes (see below). ERß is required for this decrease in gene expression, because the estrogen response for these genes is abrogated in ERßKO vessels (Fig. 1C and Table 3). The consequences of ERß vs. ER for this gene set are quite distinct from the concerted action discussed above, in that the WT response is intermediate between the responses in the two knockout backgrounds. This phenotype is diagnostic of ERß and ER opposition. However, the data support that ERß is involved in active repression of this subset of genes, not simply in opposing activation by ER. Thus, multiple types of interactions are evident between the two ER subtypes in regulating gene expression in the aorta.
Estrogen Regulation of the MRC
The biological functional category of greatest statistical significance within the vascular estrogen-responsive genes consisted of nuclear-encoded mitochondrial proteins. Given the central roles of such genes in both energy production and cellular ROS production (see below), we examined this estrogen pathway in greater detail. The MRC complexes are composed of more than 80 proteins, only 13 of which are encoded in mitochondrial DNA (41). Strikingly, genes encoding multiple subunits of all five complexes comprising the MRC, including several enzymatically active components, were regulated by ERß. The down-regulation of this gene set in WT aorta was unexpected, because in other cell types, estrogen enhances expression of specific nuclear-encoded MRC proteins (37, 42, 43, 44). For example, in rat cerebral blood vessel preparations (composed predominantly of vascular endothelial cells), estrogen treatment enhanced protein levels of cytochrome c, Cox4, and the nuclear transcription factor NRF1 (45). Physiological estrogen levels can even enhance the oxidative phosphorylation pathway by up-regulation of mitochondrial-encoded protein subunits of the MRC (41, 46, 47), although the mechanism involved is somewhat controversial and is probably unrelated to that by which nuclear-encoded subunits are regulated (41, 48).
The discrepancy between striking down-regulation of this set of nuclear-encoded MRC genes in the aorta vs. up-regulation of such genes by estrogen in other tissues may reflect the biological consequences of estrogen in the different cell types. Estrogen inhibits proliferation and migration of vascular smooth muscle cells, the major aortic cell type; both of these processes require active energy production in mitochondria. In contrast, for most other cells examined (e.g. breast cancer cell lines, uterus), estrogen activates cell proliferation. Our data in the aorta suggest a novel molecular mechanism for how estrogen could differentially regulate the MRC, because down-regulation by ERß is countered by ER activity. Therefore, whether the mitochondrial respiratory pathway is enhanced or inhibited by estrogen in a given cell could be dictated by the balance of ERß vs. ER activities, which in turn would be dependent on the relative levels of the two ERs and/or cell type-dependent profiles of ER coregulatory molecules. For example, in rat cerebral blood vessels, estrogen has been reported to induce expression of NRF1, a nuclear transcription factor that activates expression of genes encoding mitochondrial proteins (49). In contrast, in our studies in the aorta of WT mice, NRF1 expression is not detectably altered by estrogen treatment, whereas in the ERKO aorta, estrogen diminishes expression of NRF1 (Table 7). This is consistent with a model in which the estrogen transcriptional networks targeting NRF1 reflect the ER/ERß balance, resulting in either up-regulation or down-regulation of MRC activity.
Estrogen Receptors Differentially Regulate Genes Involved in ROS Production
Regulation of the MRC not only affects cellular energy metabolism, but also the levels of cellular ROS. Generation of superoxide and hydrogen peroxide by the electron transport chain provides one major source of cellular ROS (50), with predominantly hydrogen peroxide passing through the mitochondrial membrane into the cytoplasm. Hydrogen peroxide is reduced by catalases and glutathione peroxidases (Gpx) (19). Although high levels of ROS are damaging to vascular cells (51), in part through scavenging of nitric oxide, ROS also function as vascular second messengers (19, 52), leading to cellular proliferation of both aortic smooth muscle (19, 53, 54) and endothelial cells (19, 53, 54).
Physiological estrogen treatment modulates cellular ROS not only by regulating the MRC, but also by regulating activity of enzymes that control ROS levels (45, 55, 56, 57). In the mouse aorta, we have demonstrated differential regulation of ROS-related genes mediated by the two ER subtypes (Fig. 3). The ERß pathway reduces expression of ROS-generating activities (e.g. electron transport complex subunits), and the ER pathway mediates enhanced expression of antioxidants or inhibitors of oxidative stress pathways (e.g. Gpx3 and Snca; supplemental Table S13), and reduced expression of ROS-generating activities (and Cyp2E1), in addition to countering ERß-mediated MRC inhibition. Overall, the variety of genes regulated make a compelling case that the two hormone-liganded ERs, ER and ERß, function in parallel to reduce oxidative stress in aortas and to modulate ROS levels. In the smooth muscle cells, reduction in ROS by long-term estrogen treatment may result in reduced ROS second messenger signaling for cell proliferation, leading to some of the beneficial effects of estrogen on the vasculature. By acting as ROS sensors and regulators, ERs would promote the maintenance of healthy vascular function.
Estrogen-Driven Transcriptional Networks
A complete understanding of how estrogen treatment enhances vascular tone and protects against cardiovascular disease will require delineation of the transcription networks driven by the hormone in both smooth muscle and endothelial cells. Estrogen is expected to initiate a gene-regulatory network, starting with activation of ERs (both in the nucleus as transcription factors and in the cytoplasm as initiators of signal transduction pathways that culminate in activating multiple transcription factors), which would lead to regulation of primary target genes, including some encoding transcription factors. Each ER-responsive transcription factor would cooperate to activate or repress subsequent sets of genes. Estrogen regulation of gene expression in some instances could also occur by influencing stability or degradation of mRNA, rather than directly influencing transcription. Finally, given that the experiments were performed in the context of a whole animal, not all the estrogen-responsive gene expression changes would necessarily be cell autonomous.
Initial computational analyses of promoter proximal regions of estrogen-responsive vascular genes, combined with analysis of transcription factors the expression of which is estrogen-responsive in the aorta (Table 7), predicted key aspects of transcriptional networks driven by ER and ERß. That the estrogen responsiveness of expression of specific transcription factors correlates with the overrepresentation of their binding sites in promoters of genes regulated under similar circumstances strongly supports the hypothesis that specific transcription factors are critical mediators of the ER-induced cellular pathways.
In using such correlations to construct a transcriptional regulatory network, the best predictive value derives from sets of genes that are coregulated, such as the specific ERß-dependent inhibition of expression of the biologically important respiratory chain- and oxidoreductase-encoding genes. For these subsets of mouse genes, statistically significant binding sites were identified for the NRF2 transcription factor (nuclear respiratory factor-2; also known as GABP; supplemental Table S17) (58), which is known to be connected to mitochondrial gene regulation. In addition, phylogenetic analysis indicated conservation in location between human and mouse orthologous promoters of binding sites for NRF1, as well as YY1 (Table S18). Conservation across species particularly highlights these latter sites as biologically significant, although the analysis does not necessarily relate the sites to estrogen regulation. The hypothesis that NRF1 is a potential intermediary of estrogen/ERß signaling derives from its own regulation by this pathway (Table 7). These findings are consistent with the hypothesis (see above) that NRF1 may be a critical target for differentiating the physiological consequences to estrogen, based on the cell type and/or environment.
Expression of Ets1 and SRF, the binding sites of which also figure prominently in the set of MRC and oxidoreductase promoters, are oppositely regulated by estrogen. Ets1 activity is predicted to be reduced even further by interaction with Id family members, the expression of which is elevated by estrogen in the ERKO aortas. Taken together, multiple convergent molecular mechanisms are proposed in a model of the estrogen transcriptional regulatory network for the MRC- and oxidoreductase-encoding genes (Fig. 4). One prediction of this model is that genes encoding Id2, Id4, Ets-1, SRF, and NRF-1 are primary target genes of ERs. To probe this aspect of the model further, we computationally analyzed the 1300 bp upstream of the transcription start sites of genes encoding these particular transcription factor genes for binding sites for ERs (estrogen response elements). As anticipated, estrogen response elements (EREs) were not generally overrepresented in promoters of vascular genes regulated by estrogen in our dataset (data not shown); most genes are not expected to be primary targets of the ERs. However, high-affinity EREs were predicted within the promoter proximal regions of Ets1, SRF, Idb2, and NRF1 genes. Further analysis will be required to determine whether these EREs are responsible for estrogen regulation.
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
–/– (ERKO) (59) and ERß–/– (ERßKO) (60) mice were obtained from propagation of heterozygous mouse colonies; all animals were handled in accordance with NIH standards and procedures approved by the New England Medical Center Institutional Animal Care and Use Committee. For global gene expression profiling experiments, ERKO and ERßKO female mice at 2.5–4.5 months of age were paired with their respective WT female littermates, and all were ovariectomized. After a 1-wk recovery, half the mice from each genotype were implanted with E2 pellets, and the other half were implanted with placebo pellets (60-d release, 0.25 mg/pellet; Innovative Research of America, Sarasota, FL). After 7–8 d of estrogen or placebo treatment, aortas were harvested and immediately frozen in liquid nitrogen. For quantitative RT-PCR experiments, ovariectomized WT (2.5–8.5 months of age), ERßKO (4–6 months), or ERKO (10–11.5 months) mice were similarly subjected to either E2 or placebo pellets for 1 wk, followed by harvesting of the aortas.
RNA Isolation, Microarray Hybridization, and Data Analysis
Total aortic RNA was isolated by homogenization in RNA STAT-60 (Tel-Test, Inc., Friendswood, TX) and purification by chloroform extraction, with a yield of approximately 6 µg total RNA per aorta. Total aortic RNAs were analyzed for quality control with an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Total RNA (5 µg; a separate aortic sample per microarray) was used for labeling (One Cycle Labeling kit; Affymetrix, Santa Clara, CA) according to the manufacturer’s instructions. Unamplified, labeled cRNA (15 µg) was hybridized to whole mouse genome 430 2.0 microarrays (Affymetrix).
The gene expression values for each probe set were first estimated from the hybridization intensities using the corrected for GC content Robust Multichip Analysis (GCRMA) method as implemented in Bioconductor (61, 62, 63). Samples were grouped into six categories according to their genotype status (WT, ERKO, or ERßKO) and whether they were treated with estrogen or placebo after the ovariectomy. Each category (both estrogen treatment and placebo treatment) of each ER knockout strain was represented by three independent biological samples; each category of WT mice was represented by five samples, three of which were ERKO littermates and two of which were ERßKO littermates. Gene expression values for each sample were estimated using the linear model of the data as implemented by the LIMMA package in BioConductor (63). The differential expression of genes was evaluated for the following 11 comparisons: WT_E2 vs. WT, ERKO_E2 vs. ERKO, ERßKO_E2 vs. ERßKO, ERKO vs. WT, ERßKO vs. WT, ERKO vs. ERßKO, ERKO_E2 vs. WT_E2, ERßKO_E2 vs. WT_E2, ERKO_E2 vs. ERßKO_E2, ERKO_E2 vs. WT, and ERßKO_E2 vs. WT. The differences in log (base 2) expression levels were evaluated by the t test as implemented by the LIMMA package in BioConductor by comparing distribution of expression values between different sample categories. The total number of probe sets with P < 0.01 for at least one comparison of conditions included 2925 probe sets. Only genes the expression levels of which were both highly reproducible (P < 0.01), and varied by at least 2-fold (absolute log base 2-fold change > 1) in one or more genotypes, were considered as significantly differentially expressed between two biologically defined categories. These strict threshold criteria were applied to increase the confidence of identifying biologically relevant estrogen-responsive genes with a minimum of false positives. A total of 852 probe sets was included in this stringent identification of statistically regulated genes across all the comparisons of two conditions.
Quantitative RT-PCR Analysis
Total aortic RNA was isolated either as described above or using Trizol Reagent (Invitrogen, Carlsbad, CA) as previously described (64), and treated with DNase I (Ambion, Inc., Austin, TX) and repurified immediately before reverse transcription. RNA (1–1.5 µg) was reverse transcribed using either oligo dT or random hexamers as primers and Superscript II reverse transcriptase (Invitrogen Corp.), except for parallel reactions without reverse transcriptase as controls. Reverse transcriptase reactions were amplified by the indicated primers for each gene using SybrGreen (Applied Biosystems, Foster City, CA) or Qiagen Quantitec SYBR PCR kit (QIAGEN, Valencia, CA) to detect product [ABI Prism 7900 HT Real-Time PCR System (Applied Biosystems, Foster City, CA) or Stratagene Real-Time PCR machine (Stratagene, La Jolla, CA) and MxPro-Stratagene Software]. Each PCR was performed in triplicate. The final products in each case were subjected to thermal denaturation to ensure that the product denatured as a uniform peak at the appropriate temperature. For each gene the RNA level was normalized to that of ß2-microglobulin and is expressed as fold change in estrogen-treated samples compared with placebo-treated samples.
The primers are as follows (designated from 5' to 3')
ß2-microglobulin forward: GCTATCCAGAAAACCCCTCAA,
ß2-microglobulin reverse: CATGTCTCGATCCCAGTAGACGGT,
Mmd2 forward: CGGCGATGTTCACTCTGG,
Mmd2 reverse: TCAGTGGGCTGGTACCTCTT,
IGFBP6 forward: AGGAGAGCAAACCCCAAGGA,
IGFBP6 reverse: TGTGGTTTGTGTCACGAGAGC,
Ghr forward: CCACAGCCACTTTTGAGCAG,
Ghr reverse: AGAGGCGAGTTGGTGGGTT,
Frzb forward: TGAGAAGTGGAAGGATCGGC,
Frzb reverse: TTCATATCCCAGCGCTTGACT,
Lepr forward: TGAGGAGGTACGTGGTGAAGC,
Lepr reverse: TCTGACCACGTCCCATTGTG,
Glp2 forward: AAGGTGTCCAGCGGAACAAG,
Glp2 reverse: CTCGAAGCTCAGCCAGAACTG,
Fxdy6 forward: GGTCCTCTTCTCCGTTGGG,
Fxdy6 reverse: CACTTGCACCTGCGACTGAG,
Vldlr forward: CACTCGACCTTGTCAAAAGCC,
Vldlr reverse: TGTGCAACTTGGAATCCAGC,
Kctd1 forward: AGCCAATGTTGTTGGAGATGG,
Kctd1 reverse: GGCCAGTTTCCCTGTCCTGT,
Ppp1r3c forward: GGAAACCTGACGGAGTGCAG,
Ppp1r3c reverse: GGAATGCACAGTCTTTGGGTG,
Rgs5 forward: ATGAAGAACCTGGTGGAACCG,
Rgs5 reverse: TTTCTGGGCCAAGTCAAAGC,
Wif1 forward: CCCCGATGTATGAACGGTG,
Wif1 reverse: GCAGATGCAGAAGCCAGGAG,
Thbs-1 forward: ACTCGGGGCAGGAAGACTAT,
Thbs-1 reverse: TCTGTGTCTGCTTGGTCAGG,
Ets-1 forward: ACTGTGTGCCCTGGGTAAAG,
Ets-1 reverse: GGGAGGAACACACTGAGCAT,
Srf forward: ACGACCTTCAGCAAGAGGAA,
Srf reverse: GGAGAGTCTGGCGAGTTGAG,
Stc-1 forward: ACGAGGCGGAACAAAATGATT,
Stc-1 reverse: TGCACTTTAAGCTCTCTTTGACA,
Col1a2 forward: GTAACTTCGTGCCTAGCAACA,
Col1a2 reverse: CCTTTGTCAGAATACTGAGCAGC
Bioinformatics Analyses of GO Categorization and Metabolic Pathways
K-Means-clustering was performed on all genes the expression of which was changed at least 2-fold (P < 0.01) by estrogen in WT, ERKO, or ERßKO aortas, with The Institute for Genomic Research MultiExperiment Viewer (16) using Euclidian distance to separate the genes into five groups based on the fold changes in the three genotypes. Two of these clusters showed similar responses to the ER subtypes and were therefore grouped together, generating four groups.
General analysis for enrichment of GO categories was carried out using the Gene Ontology Tree Machine (GOTM) (17). The Affymetrix whole mouse genome 430 2.0 microarray data set was used as a reference gene list. In addition, gene sets were analyzed by GoSurfer using the Affymetrix Gene Chip identifiers (65). Gene sets were analyzed using GO terms associated with any gene in the union of all the genes in the input files.
Metabolic pathways were analyzed by applying Locuslink identifiers of regulated genes into KEGG (http://www.genome.jp/kegg/tool/search_pathway.html) (18).
Transcription Factor Binding Site Analysis within Upstream Regulatory Regions
Genomic regions upstream of the transcription start sites for each gene in the regulated and background gene sets were extracted, when RefSeq numbers were available, from the Cold Spring Harbor Laboratory Mus musculus Promoter Database (MmPd) version 2.33 http://rulai.cshl.edu/cgi-bin/CSHLmpd2/mmpd.pl. For each comparison between two conditions (estrogen vs. placebo treatment), a background set of unregulated genes was identified to compare with the regulated set of genes. The background genes were obtained from the complete group of statistically regulated 852 probe sets, where for the pairwise comparison of interest the absolute value of the log (base 2) fold change was less than 0.1. Promoter sequences (defined in this study as 1000 bp upstream of the transcription start site) were analyzed by Tractor (66) (Schaffer, M. E., and S. Kasif, in preparation), a program that utilizes the Match algorithm (67) and position weight matrices to identify statistically overrepresented binding sites in a set of coexpressed genes relative to a background set of sequences. We tested 546 vertebrate matrices from the TRANSFAC Professional database (version 8.4) (68) using precomputed thresholds of minFN, minFP, and minSUM, designed to reduce false negatives, false positives, or the sum of false negatives and false positives, respectively. For each matrix, the frequencies of predicted sites per gene in the coregulated set were compared with the frequencies and distribution of predictions in the respective background set with a one-sided permutation test. Only transcription factor binding site matrices with statistically overrepresented predicted site frequencies (permutation P value < 0.05) in the coexpressed gene set and at least a 2-fold enrichment of mean sites per gene over background were analyzed further. The transcription factors associated with these matrices were then grouped into families of factors possessing highly similar binding sites.
The method above was extended with a comparative analysis of human promoters. Human orthologs for each mouse gene were identified with National Center for Biotechnology Information (NCBI)’s Homologene database (version 48.1) (69) where available and supplemented with ortholog alignment data from the University of California Santa Cruz (UCSC) Genome Browser (70). Of the 85 coexpressed and background mouse genes, 69 orthologs were identified with Homologene and an additional 12 were found with the UCSC Genome Browser. For each human ortholog, the 1-kb regions upstream of the transcription start site, as defined by NCBI’s Entrez Gene database (71), were obtained from NCBI’s Human genome reference chromosome sequences (version 36.1), and the detection of overrepresented binding sites was performed as previously described. In addition to independent analyses of the human and mouse upstream regions, an analysis of conserved binding sites was performed. Sequence alignments of the human genome to each 1-kb upstream mouse sequence were extracted from MULTIZ alignments provided by the UCSC Genome Browser (http://hgdownload.cse.ucsc.edu/goldenPath/mm8/multiz17way/). To determine overrepresented sites, each sequence was independently scanned with the MATCH algorithm with 546 vertebrate matrices and three thresholds as previously described. Conserved sites were defined as nonoverlapping predictions present at the same position in the sequence alignment. The frequencies of conserved sites per promoter were compared with the background prediction frequencies as before, and P values indicating overrepresentation were calculated with a one-sided permutation test.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Disclosure Statement: There are no conflicts of interest to disclose.
First Published Online March 20, 2007
1 R.O. and K.K. contributed equally to this study. 
Abbreviations: E2, 17ß-Estradiol; ER, estrogen receptor; ERE, estrogen response element; ERKO, ER knockout; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MRC, mitochondrial respiratory chain; NADH, reduced nicotinamide adenine dinucleotide; NRF1, nuclear respiratory factor 1; ROS, reactive oxygen species; SRF, serum response factor; WT, wild type.
Received for publication November 27, 2006. Accepted for publication March 12, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
mediates the protective effects of estrogen against vascular injury. Circ Res 90:1087–1092 gene variation and cardiovascular disease. JAMA 290:2263–2270-Synuclein protects against oxidative stress via inactivation of the c-Jun N-terminal kinase stress-signaling pathway in neuronal cells. J Biol Chem 277:11465–11472-synuclein. Mol Cell Biol 25:10190–10201 regulates gene expression in the kidney. Endocrinology 144:701–710 in the uterus. Proc Natl Acad Sci USA 97:5936–5941 (ER) and ß (ERß) on mouse reproductive phenotypes. Development 127:4277–4291[Abstract]NURSA Molecule Pages Link:
This article has been cited by other articles:
 |
|
 |
|
  K. E. Roberts, M. B. Fallon, M. J. Krowka, R. S. Brown, J. F. Trotter, I. Peter, H. Tighiouart, J. A. Knowles, D. Rabinowitz, R. L. Benza, et al. Genetic Risk Factors for Portopulmonary Hypertension in Patients with Advanced Liver Disease Am. J. Respir. Crit. Care Med., May 1, 2009; 179(9): 835 - 842. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S.-H. Yang, S. N. Sarkar, R. Liu, E. J. Perez, X. Wang, Y. Wen, L.-J. Yan, and J. W. Simpkins Estrogen Receptor {beta} as a Mitochondrial Vulnerability Factor J. Biol. Chem., April 3, 2009; 284(14): 9540 - 9548. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Cignarella, C. Bolego, V. Pelosi, C. Meda, A. Krust, C. Pinna, R. M. Gaion, E. Vegeto, and A. Maggi Distinct Roles of Estrogen Receptor-{alpha} and {beta} in the Modulation of Vascular Inducible Nitric-Oxide Synthase in Diabetes J. Pharmacol. Exp. Ther., January 1, 2009; 328(1): 174 - 182. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. W. Aberdeen, S. J. Wiegand, T. W. Bonagura Jr., G. J. Pepe, and E. D. Albrecht Vascular Endothelial Growth Factor Mediates the Estrogen-Induced Breakdown of Tight Junctions between and Increase in Proliferation of Microvessel Endothelial Cells in the Baboon Endometrium Endocrinology, December 1, 2008; 149(12): 6076 - 6083. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Tsutsumi, X. Zhang, K. Takata, K. Takahashi, R. H Karas, H. Kurachi, and M. E Mendelsohn Differential regulation of the inducible nitric oxide synthase gene by estrogen receptors 1 and 2 J. Endocrinol., November 1, 2008; 199(2): 267 - 273. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. K. Schnoes, I. Z. Jaffe, L. Iyer, A. Dabreo, M. Aronovitz, B. Newfell, U. Hansen, G. Rosano, and M. E. Mendelsohn Research Resource: Rapid Recruitment of Temporally Distinct Vascular Gene Sets by Estrogen Mol. Endocrinol., November 1, 2008; 22(11): 2544 - 2556. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Klinge, N. S. Wickramasinghe, M. M. Ivanova, and S. M. Dougherty Resveratrol stimulates nitric oxide production by increasing estrogen receptor {alpha}-Src-caveolin-1 interaction and phosphorylation in human umbilical vein endothelial cells FASEB J, July 1, 2008; 22(7): 2185 - 2197. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. M. Miller and S. P. Duckles Vascular Actions of Estrogens: Functional Implications Pharmacol. Rev., June 1, 2008; 60(2): 210 - 241. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. A. Mattingly, M. M. Ivanova, K. A. Riggs, N. S. Wickramasinghe, M. J. Barch, and C. M. Klinge Estradiol Stimulates Transcription of Nuclear Respiratory Factor-1 and Increases Mitochondrial Biogenesis Mol. Endocrinol., March 1, 2008; 22(3): 609 - 622. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Hook, A. D. Skillman, B. Gopalan, J. A. Small, and I. R. Schultz Gene Expression Profiles in Rainbow Trout, Onchorynchus mykiss, Exposed to a Simple Chemical Mixture Toxicol. Sci., March 1, 2008; 102(1): 42 - 60. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
