Published ahead of print on August 8, 2008, doi:10.1164/rccm.200712-1890OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200712-1890OC
Hypoxia Selectively Activates the CREB Family of Transcription Factors in the In Vivo Lung1 University College Dublin, School of Medicine and Medical Science, and Conway Institute of Biomolecular and Biomedical Research, Dublin, Ireland Correspondence and requests for reprints should be addressed to Paul McLoughlin, M.B., Ph.D., University College Dublin, School of Medicine and Medical Sciences, Conway Institute, Belfield, Dublin 4, Ireland. E-mail: paul.mcloughlin{at}ucd.ie
Rationale: Pulmonary hypertension is a common complication of chronic hypoxic lung diseases and is associated with increased morbidity and reduced survival. The pulmonary vascular changes in response to hypoxia, both structural and functional, are unique to this circulation. Objectives: To identify transcription factor pathways uniquely activated in the lung in response to hypoxia. Methods: After exposure to environmental hypoxia (10% O2) for varying periods (3 h to 2 wk), lungs and systemic organs were isolated from groups of adult male mice. Bioinformatic examination of genes the expression of which changed in the hypoxic lung (assessed using microarray analysis) identified potential lung-selective transcription factors controlling these changes in gene expression. In separate further experiments, lung-selective activation of these candidate transcription factors was tested in hypoxic mice and by comparing hypoxic responses of primary human pulmonary and cardiac microvascular endothelial cells in vitro. Measurements and Main Results: Bioinformatic analysis identified cAMP response element binding (CREB) family members as candidate lung-selective hypoxia-responsive transcription factors. Further in vivo experiments demonstrated activation of CREB and activating transcription factor (ATF)1 and up-regulation of CREB family–responsive genes in the hypoxic lung, but not in other organs. Hypoxia-dependent CREB activation and CREB-responsive gene expression was observed in human primary lung, but not cardiac microvascular endothelial cells. Conclusions: These findings suggest that activation of CREB and AFT1 plays a key role in the lung-specific responses to hypoxia, and that lung microvascular endothelial cells are important, proximal effector cells in the specific responses of the pulmonary circulation to hypoxia.
Key Words: hypoxia • cAMP response element binding • pulmonary hypertension • transcription factor binding site
Pulmonary hypertension is a common complication of chronic hypoxic lung diseases, which is associated with increased morbidity and reduced survival. Moreover, cor pulmonale is an independent predictor of increased mortality, suggesting that pulmonary hypertension contributes directly to reduced survival (1, 2). The increased pulmonary vascular resistance underlying chronic hypoxic pulmonary hypertension is accompanied by characteristic changes in the blood vessels, which include remodeling and thickening of the walls of precapillary vessels and sustained vasoconstriction (3–8). These long-term structural and functional changes in the pulmonary vessels in response to hypoxia are unique to this circulation, and are not observed in the vasculature of other organs of the body. The important roles of the vascular endothelium in the local control of vascular smooth muscle tone and in modulating changes in the structure of the vessel wall are now well recognized (9). It is also well recognized that endothelial cells from different organs show considerable heterogeneity, and that their roles in the control of the vasculature differ from organ to organ (10–12). These observations suggest that the pulmonary endothelium probably plays a central role in producing the distinct responses of the pulmonary circulation to hypoxia. The lung displays a further unique aspect in its responses to reductions in alveolar oxygen, that is, marked responses are observed at higher partial pressures of oxygen than would elicit responses in any other organ. For example, hypoxic pulmonary vasoconstriction and pulmonary hypertension develop when alveolar PO2 is 40–60 mm Hg (2, 5, 13), a value that is substantially higher than that in other organs under conditions of normal oxygenation (14–16). This implies that there are specific mechanisms within the lung that are responsible for sensing and responding to changes in PO2, which are different from those found in other organs and tissues. At present, these mechanisms are poorly understood. Given that the lung shows distinct phenotypic responses to hypoxia, at uniquely high partial pressures of oxygen, we postulated that there must be activation of specific transcription factors within the lung that mediate these responses. The aim of this study was to identify transcription factors that are activated in the lung in response to alveolar hypoxia similar to that seen in lung disease, but are not activated in the systemic organs by the same stimulus. In addition, given the important role of endothelial cells in pulmonary vascular remodeling and vasoconstriction, we wished to determine whether hypoxic activation of lung-specific transcription factors was observed in these cells.
With institutional ethics approval, adult male–specific, pathogen-free C57BL/6J mice (20–25 g) were exposed to normoxic (FIO2 = 0.21) or hypoxic conditions (FIO2 = 0.10) for varying periods, from 3 hours to 2 weeks (n = 4–8 per group) using a servo-controlled environmental chamber (Biospherix, Lacona, NY), as previously described (13). After exposure, mice were killed under anesthesia and organs were immediately removed for isolation of RNA. Gene expression within lung tissue was analyzed using microarray (U430Av2 array; Affymetrix, Santa Clara, CA) and Taqman real-time polymerase chain reaction (PCR) techniques. For the microarray analysis, RNA from mice at each time point of normoxia and hypoxia was pooled (n = 6) before cRNA synthesis. Genespring software (Agilent Technologies, Stockport, UK) was used to identify genes, the expression of which was changed by more than 1.5-fold from control normoxic values after hypoxic exposure. These genes were then clustered using k-means analysis. Candidate lung-selective transcription factors were identified based on the rationale that binding sites for these transcription factors would occur with higher-than-expected frequency in the promoters of the genes contained in the hypoxia-regulated clusters. These binding sites were identified using the approach of Ho Sui and colleagues (17) and a recently developed database of transcription factor binding sites that are conserved within the promoters of orthologous genes across the human, mouse, rat, and dog genomes (18, 19). The Fischer exact test (P < 0.001) and the z score statistic (>10) were used to identify overrepresented binding sites within a specific cluster. In separate groups of control and hypoxic mice, nuclear protein was extracted and analyzed for selected transcription factors from the cAMP response element binding (CREB) family (CREB, activating transcription factor [ATF]1, and ATF2) by Western blotting of both total and serine-133 phosphorylated forms. For in vitro experiments, human primary lung (human lung microvascular endothelial cells [HMVECs]-L) and cardiac HMVECs (HMVEC-Cs) were cultured according to the supplier's recommendations (Lonza Bioscience, Basel, Switzerland). Cells between passage six and nine were maintained under hypoxic (5% O2,) or normoxic (21% O2) conditions for varying periods of time up to 24 hours. Nuclear extracts from cells were analyzed for activated, ser133 phosphorylated CREB using a specific CREB response element DNA binding assay (TransAm; Active Motif, Carlsbad, CA). Cells were also harvested, processed for RNA extraction, and analyzed for changes in gene expression by Taqman real-time PCR. All data were presented as mean (±SEM) for n independent experiments. The statistical significance of the differences between means was evaluated using Student's unpaired t test or analysis of variance with a Dunnett post hoc test (GraphPad, San Diego, CA). Where data were nonnormally distributed, Mann-Whitney U test with the Bonferroni correction was used. Further details of experimental procedures can be found in the online supplement.
Identification of Overrepresented Transcription Factor Binding Sites After isolation of RNA from hypoxic lungs, its integrity was checked and only samples that met the highest-quality standard were used in the subsequent microarray analysis. Initial analysis of the microarray data confirmed that well known hypoxia-responsive genes were up-regulated, including solute carrier family 2 (facilitated glucose transporter) member 1, adrenomedullin, hexokinase 2, and endothelin (EDN)-1. K-means cluster analysis (K = 24) was then used to group genes, the expression of which was altered by hypoxia (total no. = 1,534) based on similar temporal expression profiles. One cluster showed prominent early up-regulation, suggesting that these genes were central to initiating the pulmonary hypoxic responses (Figure 1). In this cluster, there was significant overrepresentation of 20 specific transcription factor binding sites (Table 1). Note that only known transcription factors are shown; six overrepresented unknown (predicted) transcription bindings sites were also found, but are not shown here. The most significantly overrepresented binding site was a binding site for the hypoxia-inducible factor (HIF)-1β (aryl hydrocarbon receptor nuclear translocator) homodimer (Table 1). However, of major interest was the presence in this list of specific binding sites for the cAMP response element binding protein (CREB) family of transcription factors including CREB, ATF1/ATF3, ATF3, and CREBp1CJUN. Because these were the most abundant of the transcription factor binding sites present (five motifs identified) and were associated with genes, the expression of which was up-regulated early in the course of hypoxic exposure, we focused our further investigations on this family. We next asked which of the 137 genes in this cluster had promoters that contained putative CREB and CREBP1CJUN motifs, and identified 21 of these (Table 2). Literature review revealed previously published direct biological evidence of hypoxia-induced changes in gene expression for nine of these genes. CREB-dependent control of expression had been previously documented for seven of these 21 genes, while five of this group were known to be HIF-dependent (i.e., responsive to the HIF-1  /HIF-β heterodimer) (Table 2).
CREB/ATF1 Activation in the Hypoxic Lung Activation of CREB requires protein phosphorylation at serine-133; ATF1 is also activated in a similar manner. Western blotting of lung nuclear lysates revealed CREB phosphorylation after hypoxia when compared with normoxic tissues (Figure 2A). This activation was observed at 3 hours, was more marked at 1 day of hypoxia, and had returned to control levels by 1 week. Total CREB was unchanged by hypoxic exposure. Levels of the constitutively expressed nuclear protein, octamer-binding transcription factor (OCT)1, were analyzed as a control for protein loading (Figure 2A). Densitometric analysis of phosphorylated CREB expression over total CREB levels demonstrated a biphasic pattern of statistically significant increase thereafter 3 hours, and again after 1 day, of hypoxic exposure as compared with normoxic controls (Figure 2B). We also observed a significant increase in phosphorylation of a second member of the CREB family, ATF1, after 3 hours of exposure (Figure 2). The phosphorylation-dependent activation of a further CREB family transcription factor, ATF2, was also assessed, but did not show any significant hypoxia-induced alterations (data not shown). Interestingly, there were no significant changes in CREB or ATF1 phosphorylation in three systemic organs (kidney, liver, and spleen) isolated from the same hypoxic animals (Figure 3).
Confirmation of CREB-dependent Gene Expression within the Hypoxic Lung From the list of 21 potentially CREB-dependent genes, we selected four genes in which to confirm increased expression using real-time PCR; namely, ATF3, EDN-1, vascular endothelial growth factor receptor (VEGFR)-1, and follistatin (FST). Three of these, ATF3, EDN-1, and FST, showed significant up-regulation early in hypoxia (6 h and/or 1 d), and had returned to baselines values by 1 week (Figure 4). In the case of VEGFR1, a significant change in gene expression was also observed over time (P < 0.025, analysis of variance), with the highest values seen after 1 day of hypoxic exposure. However, on post hoc testing, the difference between the peak and baseline values was not statistically significant. Three of the four genes demonstrated progressive increases between 6 and 24 hours, suggesting that peak expression had occurred at some point between the first day and 1 week of hypoxic exposure.
Lung-selective CREB-dependent Gene Expression To confirm that the lung-selective activation of CREB family transcription factors caused an organ selective pattern of gene expression in hypoxia, we undertook a separate, independent series of experiments in which we examined ATF3, EDN-1, FST, and VEGRF1 mRNA in the lung, and a panel of five systemic organs in groups of control normoxic and hypoxic mice (n = 4). This showed up-regulation of these genes in the lung without significant alterations in any other organ examined (Figure 5).
Selective Activation of CREB and ATF-1 in the Pulmonary Microvascular Endothelium Given the central role that the endothelium plays in hypoxia-induced pulmonary vascular remodeling and the development of pulmonary hypertension, we compared responses in primary HMVECs to those in cardiac microvascular endothelial cells. Hypoxia (5% O2) treatment resulted in the activation of CREB in a time-dependent manner in lung, but not cardiac cells (Figure 6A), showing a biphasic pattern of activation similar to that observed within the in vivo lung. Examination of CREB-dependent gene expression revealed an up-regulation of ATF3, EDN-1, and VEGRF-1 that occurred only in the pulmonary, but not in the cardiac cells (Figure 6B). FST mRNA was undetectable in the pulmonary endothelial cells, both under normoxic conditions and after hypoxic exposure.
In undertaking this study, we set out to identify specific transcription factors, the activation of which could be responsible for the changes in protein expression that underpin the well known lung-specific responses to hypoxia, including hypoxic pulmonary vascular remodeling and pulmonary hypertension. From the genes identified as hypoxia responsive in the lung in our initial gene array studies, we selected for transcription factor binding site analysis one cluster in which genes showed early increases in response to hypoxia. This early response suggested that genes within this cluster played a key role in initiating the pulmonary hypoxic response and, for this reason, we focused our further investigations on these. Bioinformatic analysis identified a list of transcription factor binding sites that had a putative role in the pulmonary responses to hypoxia; among them, some that were not previously known to be hypoxia responsive (Table 1). The most significantly overrepresented binding site was one that binds the HIF-1β (aryl hydrocarbon receptor nuclear translocator) homodimer (Table 1). Although it is likely that this homodimeric transcription factor plays a role in controlling gene expression (20), there is little evidence for its involvement in hypoxia-induced responses. Interestingly, five separate binding sites for members of the CREB family were found (Table 1). To test the putative role of CREB during pulmonary hypoxia suggested by our bioinformatics analysis, we examined CREB and ATF1 phosphorylation in a separate series of experiments. Both transcription factors showed a basal level of phosphorylation under normoxic conditions that was increased after hypoxic exposure (Figure 2), with peak values observed after 24 hours. This finding directly validated the central role of CREB family members suggested by the initial bioinformatics analysis. Moreover, in three systemic organs harvested from the same mice (liver, kidney, and spleen), neither of these transcription factors showed increased phosphorylation, suggesting that the response was lung specific.
Within the early responding cluster, 21 of the 137 genes had putative CREB family binding sites (Table 2), of which nine were previously reported to be hypoxia responsive, including EDN-1, which plays a key role in the development of hypoxic pulmonary hypertension (21). For seven of these genes, there was previously published evidence indicating CREB responsiveness (Table 2). We chose four genes from this list for validation by real-time PCR; two of these, EDN-1 and VEGFR-1, were chosen because they were known to be HIF responsive (HIF-1 Our initial hypothesis predicted that the CREB-dependent genes that we had identified should show selective, or at the least preferential, up-regulation in the hypoxic lung in vivo. To test this prediction independently of the data set (and mRNA) that had initially identified the CREB family of transcription factors as important, we undertook further independent experiments using separate groups of mice (i.e., not those used in the initial gene expression studies) and confirmed hypoxia-induced up-regulation in the lung in the absence of changes in expression in five systemic organs: the heart, liver, kidney, spleen, and thymus (Figure 5). These data strongly support the central role of this family of transcription factors in mediating the lung-specific hypoxia responses. Given the key role of the endothelium in hypoxic pulmonary vascular remodeling, hypoxic vasoconstriction, and the development of pulmonary hypertension, we examined CREB and ATF1 activation and the expression of CREB-dependent genes in human primary microvascular endothelial cells in response to hypoxia and compared these responses to those observed in cardiac microvascular endothelial cells. Our finding that CREB and ATF1 activation were observed only in the pulmonary endothelial cells supports the in vivo observation that this response is specific to the pulmonary tissue. Moreover, three of the four exemplary CREB-responsive genes were selectively up-regulated by hypoxia in the pulmonary endothelial cells (Figure 6), suggesting that CREB family activation was functionally important. These findings in isolated endothelial cells show that the responses were not secondary effects in vivo, but that they were due to a direct effect of hypoxia on the cells.
Lung-selective CREB Activation The data presented here expand in another important way upon previously published work. Previous in vivo work has shown hypoxia-induced CREB phosphorylation in nonpulmonary tissues, but this has only been shown in response to the extreme hypoxia caused by vascular occlusion (ischemia) (27). The present findings are markedly different in that we have demonstrated lung-selective phosphorylation of CREB on serine-133 at very modest levels of hypoxia (40–60 mm Hg), typical of those seen in the pulmonary alveoli during the course of lung diseases (2). In addition, our data also demonstrate, for the first time in any in vivo tissue, phosphorylation of ATF1 induced by hypoxia. These findings show that there are lung-specific mechanisms responsible for CREB and ATF1 activation in the conditions of relatively mild hypoxia characteristic of lung disease.
Conclusions These findings suggest that the CREB family of transcription factors plays a key role in mediating the specific responses of the pulmonary circulation to hypoxia. These data also suggest that pulmonary microvascular endothelial cells are important, proximal effector cells in the specific hypoxic responses of the pulmonary circulation.
Supported by grants from the Health Research Board of Ireland (P.M.), an unrestricted grant from Actelion (P.M.), the Higher Education Authority of Ireland, PRTLI (P.M.), the Wellcome Trust (C.T.T.), and the Science Foundation of Ireland (C.T.T.). This article has an online supplement, which is available from the issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200712-1890OC on August 8, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form December 25, 2007; accepted in final form July 17, 2008
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
P < 0.05 and 
