Computational Identification of Key Biological Modules and Transcription Factors in Acute Lung Injury
http://www.100md.com
《美国呼吸和危急护理医学》
Departments of Medicine, Physiology and Biophysics, and Pathology, University of Washington
the Veterans Affairs Puget Sound Healthcare System Medical Research Service, Seattle, Washington
ABSTRACT
Rationale: Mechanical ventilation augments the acute lung injury (ALI) caused by bacterial products. The molecular pathogenesis of this synergistic interaction remains incompletely understood.
Objective: We sought to develop a computational framework to systematically identify gene regulatory networks activated in ALI.
Methods: We have developed a mouse model in which the combination of mechanical ventilation and intratracheal LPS produces significantly more injury to the lung than either insult alone. We used global gene ontology analysis to determine overrepresented biological modules and computational transcription factor analysis to identify putative regulatory factors involved in this model of ALI.
Results: By integrating expression profiling with gene ontology and promoter analysis, we constructed a large-scale regulatory modular map of the important processes activated in ALI. This map assigned differentially expressed genes to highly overrepresented biological modules, including "defense response," "immune response," and "oxidoreductase activity." These modules were then systematically incorporated into a gene regulatory network that consisted of putative transcription factors, such as IFN-stimulated response element, IRF7, and Sp1, that may regulate critical processes involved in the pathogenesis of ALI.
Conclusions: We present a novel, unbiased, and powerful computational approach to investigate the synergistic effects of mechanical ventilation and LPS in promoting ALI. Our methodology is applicable to any expression profiling experiment involving eukaryotic organisms.
Key Words: acute lung injury gene network microarray transcription factor
Acute lung injury (ALI) is an important cause of in-hospital morbidity and mortality in critically ill patients. Infectious etiologies, such as sepsis and pneumonia, are among the most common risk factors for developing ALI (1). Mechanical ventilation (MV) is a life-saving maneuver in ALI, but in patients with established lung injury, MV strategy can affect mortality and respiratory failure (2). Despite recent advances, many molecular mechanisms in ALI remain unexplored, and few therapeutic interventions targeting identified molecular pathways have been clinically effective (3, 4). A critical first step in discovering novel therapies in ALI is to gain a deeper understanding of the complex pathophysiologic processes involved in the interaction between MV and infectious insults that lead to lung injury.
We have recently developed a mouse model in which ALI results from the combination of mechanical ventilation and low-dose intratracheal LPS (5). Our model captures two crucial perturbations, mechanical ventilation and infectious insult, that are frequently associated with ALI in the clinical setting. An important feature of this animal model is the observation that neither MV nor LPS alone causes significant lung injury; ALI develops only when both interventions are combined, suggesting a synergistic role for MV in modulating the effects of LPS. We hypothesized that the synergism between MV and LPS depends on the activation of distinct transcription factors (TFs) that trigger key biological processes leading to ALI. Understanding the molecular mechanisms involved in this "double hit" paradigm can shed light on the pathogenesis of ALI in humans.
In this article, we outline a systematic approach to integrate transcriptional profiling with global gene annotation and TF analysis to obtain a large-scale regulatory modular map of the key processes involved in our model of ALI. Our computational analyses determine important biological modules in the context of their putative regulatory networks and identify TFs that may regulate critical biological processes involved in the development of ALI, thereby highlighting potential targets for therapeutic intervention. Some of the results of these studies have been previously reported in the form of an abstract (6).
METHODS
Animal Experiments
All experiments were approved by University of Washington's Animal Care Committee. Twenty-four C57Bl/6 8-wk-old male mice (Harlan; Indianapolis, IN) were randomly assigned to four groups (n = 6/group): (1) phosphate-buffered saline (PBS) aspiration and spontaneous breathing (control), (2) PBS with MV, (3) LPS aspiration and spontaneous breathing (LPS), or (4) LPS with mechanical ventilation (MV+LPS).
After each mouse was anesthetized, 5 ng/g body weight of LPS (Escherichia coli serotype 0111:B4; List Laboratories, Campbell, CA) in the LPS and MV+LPS groups, or an equal volume of sterile PBS (in the control and MV groups) was deposited in the posterior oropharynx. Thirty minutes after aspiration, MV and MV+LPS mice were reanesthetized, intubated, and mechanically ventilated with a tidal volume of 10 ml/kg, respiratory rate of 150/min, FIO2 of 0.21, and 0 cm H2O end-expiratory pressure (MiniVent; Harvard Biosciences, Holliston, MA) for 4 h. The control and LPS mice breathed spontaneously during this 4-h period. At the conclusion of the intervention period, mice were killed by exsanguination, lungs were flushed with RNAse-free PBS via the right ventricle and homogenized, and total RNA was isolated using the RNeasy Midi Kit (Qiagen; Valencia, CA). RNA integrity was confirmed using the Bioanalyzer 2100 (Agilent; Palo Alto, CA).
Histology
For each group, two additional mice were killed at the end of the intervention period, and their lungs removed and fixed via intratracheal instillation of 4% formalin at a transmural pressure of 20 cm H2O. After fixation, the lungs were embedded in paraffin, and 4-μm sections were stained with hematoxylin and eosin (H&E).
Microarray Data Analysis
For each mouse, labeled cRNA was prepared from total RNA, hybridized to an Affymetrix MOE430A microarray, and scanned. Image analysis performed using Affymetrix MAS 5.0 software (Affymetrix, Santa Clara, CA). Background adjustment and quantile normalization across all 24 microarrays were performed using the Robust Multichip Average algorithm (7). Statistically significant differential gene expression was determined using Significance Analysis of Microarrays (8). The problem of multiple comparisons was addressed by choosing a false-discovery rate of 5%. Significantly differentially expressed genes during MV, LPS, or MV+LPS relative to control were clustered using the partitioning around medoids (PAM) algorithm (9). All microarray data, in compliance with Minimum Information about a Microarray Experiment, are available at the GEO website (http://www.ncbi.nlm.nih.gov/projects/geo/, query GSE 2411).
Gene Ontology Analysis
Gene annotation of all probesets present on each GeneChip was obtained from the Gene Ontology database (10). Highly overrepresented biological modules in our experimental conditions were determined using two different software programs, Gene MicroArray Pathway Profiler (GenMAPP) and Expression Analysis Systematic Explorer (EASE) (11, 12). To improve statistical confidence in the results, all identified biological modules had to be statistically significant using both approaches.
TF Analysis
Putative overrepresented TFs within subsets of differentially expressed genes were identified using the PRIMA (PRomoter Integration in Microarray Analysis) algorithm (13, 14) and based on the position weight matrices of approximately 500 known eukaryotic TFs (TRANSFAC database) (15). Multiple comparisons problem was addressed by selecting only overrepresented TFs with Bonferroni-corrected p values of less than 0.05.
RESULTS
Combination of MV and LPS Results in Severe Lung Injury
Figure 1 shows H&E stains of mice lungs exposed to each of the four experimental conditions (control, MV, LPS, and MV+LPS) for 6 h. Lungs from the MV-only group were not histologically different from the control group and had no evidence of inflammation or injury. In the LPS-only group, there was modest cellular infiltration in the alveolar spaces, but no edema or septal thickening was seen. In contrast, the lungs from mice in MV+LPS group showed a profound neutrophilic infiltration, septal thickening, and perivascular edema consistent with ALI.
Combination of MV and LPS Results in Global Augmentation of the Transcriptional Response in the Lung
Differential gene expression during each perturbation (MV, LPS, MV+LPS) relative to control was determined for n = 6 mice in each group using significance analysis of microarrays (SAM) and choosing a false-discovery rate of 5%. A total of 7,510 genes were differentially expressed during at least one of the three perturbations relative to the unperturbed control condition. There were 1,415 differentially expressed genes in the MV group (838 up-regulated, 577 down-regulated), 4,536 genes in the LPS group (1,366 up-regulated, 3,170 down-regulated), and 6,837 genes in the MV+LPS group (2,418 up-regulated, 4,419 down-regulated; Figure 2).
We next applied PAM algorithm to cluster all 7,510 differentially expressed genes across their perturbation states (MV, LPS, MV+LPS). Using PAM's quality index and silhouette plot, we determined that this dataset could be most parsimoniously divided into two clusters. This finding was confirmed using principal component analysis, which revealed only two distinct groups of genes in perturbation space (see online supplement). Figure 2 displays these two clusters of gene expression, demonstrating the global augmentation of transcriptional response with the combination of mechanical ventilation and LPS (MV+LPS). The MV+LPS group contains the largest number of differentially expressed genes and, on average, the greatest level of differential expression.
Global Gene Ontology Analysis Reveals Important Biological Modules in ALI
To identify important molecular pathways activated in lung injury due to the synergistic effect of MV and LPS, we focused our attention on the subset of genes that were most likely to differentiate the ALI phenotype (MV+LPS) from the other perturbations. We determined that 2,567 genes were differentially expressed (809 up-regulated, 1,758 down-regulated) in the MV+LPS group relative to all three other conditions (control, MV, LPS) at a false-discovery rate of 5% (see online supplement). Next, we performed global gene ontology analysis using two independent statistical methods, GenMAPP and EASE, to identify overrepresented biological themes within these differentially expressed genes. Tables 1 and 2 list highly significant biological modules determined by GenMAPP and EASE among differentially up-regulated and down-regulated genes, respectively. Among up-regulated genes, response to biotic stimulus, defense response, immune response, RNA and protein processing, and metabolism were among the most overrepresented modules. Highly overabundant modules among down-regulated genes included oxidoreductase and catalytic activity, glutathione transferase activity, extracellular matrix constituents, and organic and fatty acid metabolic pathways.
In Silico Promoter Analysis Identifies Putative TFs Activated during MV, LPS, and Lung Injury due to MV+LPS
To discover putative TFs regulating the lung's response to mechanical ventilation, LPS, and injury from the combination of MV+LPS, we undertook a computational search for overrepresented promoter sites among genes differentially expressed during each of these perturbations. Overrepresentation of these binding sites in differentially regulated genes suggests a possible role for their associated TFs. Table 3 lists highly significant putative TFs identified based on genes differentially expressed during MV, LPS, or MV+LPS relative to Control. In general, under each condition, a different repertoire of TFs is highly enriched. For example, the LPS-only group included TFs known to be involved in the cellular response to LPS, such as nuclear factor-B (NF-B), cRel, and interferon regulatory factor (IRF) (16). None of these TFs was present in the MV-only group, which included ETF, E2F, and Nrf1. The MV+LPS group shared several TFs with LPS-only and MV-only conditions, implying that gene regulation in this condition was strongly influenced by mechanical ventilation and LPS. Among down-regulated genes, all three perturbations shared Sp1 as the most overrepresented TF.
We next focused our attention on the subset of 2,567 genes differentially expressed during MV+LPS relative to all other conditions (control, MV, and LPS); these genes differentiate the transcriptional response between the lung injury phenotype seen in MV+LPS compared with the other perturbations. Table 4 lists highly overrepresented TFs among these genes. This list includes TFs, such as NF-B and ETF, that were previously identified in LPS-only and MV-only conditions, respectively, suggesting that common regulatory factors are, in part, responsible for the severe lung injury phenotype seen in MV+LPS. However, we also identified several TFs, such as Myc, USF2, and Elk1, that were not enriched among differentially expressed genes during MV or LPS. These factors may represent novel sites of gene regulation responsible for the molecular pathways leading to ALI.
Integration of Global Gene Annotation and TF Analysis: Constructing a Gene Regulatory Map of ALI
Promoter analysis of 2,567 differentially expressed genes in the MV+LPS group relative to Control, MV, and LPS identified overrepresented putative TFs (Table 4). Each of these TFs recognizes specific binding site(s) in the 5' flanking region of a subset of these 2,567 differentially expressed genes. We systematically performed gene ontology analysis on subsets of differentially expressed genes that possess binding sites for the same putative TF. In this manner, overrepresented biological modules regulated by each enriched TF were identified. The results of this analysis were integrated with our original gene ontology analysis of all 2,567 differentially expressed genes in the MV+LPS group to construct a modular gene regulatory map. Figure 3 displays a graphical representation of the hierarchical and modular structure of this regulatory map, highlighting important biological processes and their putative regulators in this model of ALI. For example, the "immune response" module is composed of differentially up-regulated genes, many of which share common binding motifs for IFN-stimulated response element (ISRE), whereas genes assigned to the "programmed cell death" module primarily recognize IRF7 in their upstream promoter sites. Several biological modules, such as "RNA processing," "RNA metabolism," and "intracellular signaling" recognize multiple TFs, suggesting complex interactions among these factors and the genes they regulate.
DISCUSSION
The main finding of this study is that the combination of a noninjurious ventilatory strategy with low-dose intratracheal LPS induces a form of ALI that is characterized by the activation of distinct sets of biological modules and TFs when compared with MV or LPS alone. We identified these processes by systematically integrating expression profiling with gene ontology and TF analysis. The regulatory modular map resulting from our computational methods provides information on which genes are differentially expressed during lung injury from MV+LPS, which biological modules are activated in this complex process, and which putative TFs are regulating the genes within these modules (Figure 3).
We initially showed that there is a global augmentation of transcriptional response in the lung when a noninjurious mechanical ventilation strategy is combined with low-dose LPS administration (Figure 2). We used gene ontology analysis to identify important biological modules activated during the ALI that results from this "double hit" perturbation (MV+LPS). Initially, we annotated gene products based on their molecular function, biological process, or cellular component using the Gene Ontology Consortium's database (http://www.geneontology.org/). Next, functional categories (biological modules) with overrepresentation of differentially expressed genes were identified by calculating the probability of this enrichment occurring by chance (see Tables 2 and 3, and online supplement). The presence of several statistically significant modules involved in RNA and protein metabolism and biosynthesis implies a general activation of these pathways during ALI. Another highly significant biological process overrepresented among differentially up- regulated genes in the MV+LPS group relative to all other conditions (Control, MV, LPS) was "immune response." Genes assigned to this module included CD14, TLR4, NF-B, MD-1, and MD-2, all of which are crucial in recognizing and mediating the innate immune response to LPS (16). Other differentially up-regulated genes in this module included cytokines and chemokines (IL15, Cxcl5, Cxcl16, Ccr1, Ccr2), other Toll-like receptors (TLR2, TLR6, TLR7), and several IFN-induced molecules (Ifi16, Ifi35, Ifi203, Ifi205, Isgf3 g, Ifitm3). These genes highlight activated pathways that are likely to result in the severe lung injury seen during MV+LPS.
The most significant biological module among genes down-regulated during MV+LPS relative to Control, MV, and LPS was "oxidoreductase activity." Several members of the cytochrome P450 family were assigned to this module, including Cyp1a1, Cyp2d22, Cyp2f2, Cyp4f13, Cyp4v3, and Cyp27a1. Alteration in the activity of cytochrome P450 family of enzymes has been reported during various models of organ injury, including chemical and viral hepatitis, ischemia/reperfusion injury of the heart, and inhalation/chemical injury to the lungs (17–19). Recently, the activity of cytochrome P450 was shown to be significantly suppressed in critically ill trauma patients and strongly correlated with the degree of organ failure (20).
Another significant biological module among genes down-regulated in the MV+LPS group was "glutathione transferase activity," which consisted of 14 different members of the glutathione S-transferase family. Glutathione transferases perform a crucial role in protecting cells from oxidative stress and other toxic insults (21). Down-regulation of members of this module during MV+LPS suggests a novel mechanism in the pathogenesis of ALI.
To our knowledge, this is the first study to undertake a systematic computational search for putative TFs involved in the lung's response to mechanical ventilation, LPS, and acute injury due to MV+LPS. We and others have shown that factors regulating inflammation, such as NF-B and AP-1, are activated when lungs are mechanically ventilated at high tidal volumes that independently result in stretch injury (22, 23). However, we now report several putative TFs during a noninjurious mechanical ventilation strategy that is much more representative of current clinical practice. Highly significant regulators of gene expression in the MV-only group that were computationally identified included ETF, E2F, Nrf1, and CREB (cAMP-responsive element binding protein). CREB has been shown to be activated during in vitro mechanical stretching of lung bronchial epithelial cells, and more recently, during mechanical ventilation of isolated perfused rabbit lungs (23, 24).
Several TFs known to regulate cellular response to LPS, such as NF-B and IRF families, were among the most significantly enriched factors in the LPS-only group, thereby confirming the robustness of our computational approach. Although there was minimal histologic evidence of lung injury in mice exposed to low-dose LPS, our analysis revealed a profound inflammatory transcriptional and regulatory response during this perturbation.
We identified several putative TFs that were highly overrepresented during ALI from the combination of MV and LPS. By focusing on the subset of genes differentially expressed in the MV+LPS group relative to all other conditions (Control, MV, LPS), we searched for TFs that were most likely to regulate the severe injury seen in MV+LPS. We then systematically integrated these results with global gene ontology analysis of the same subset of genes to construct a large-scale modular regulatory map of ALI (Figure 3).
Myc and USF2 were among the most overrepresented putative TFs identified. Figure 3 reveals that they regulated primarily genes involved in RNA metabolism and processing. IRF7 was another highly overrepresented TF in the MV+LPS group and primarily regulated apoptotic and protein transport/targeting modules. ISRE was a highly enriched binding motif regulating several biological modules involved in immune and defense response. The activation of genes within these modules is likely to be crucial in development of the lung injury seen during MV+LPS. ISRE is a key binding site for several TFs, such as IRFs (IRF1, IRF3, IRF4, IRF7, IRF9), STAT1, and STAT2, that regulate IFN-mediated cellular signaling during viral/bacterial infection, cytokine stimulation, and innate immune response (25–27). Therefore, it was not surprising that ISRE was highly overrepresented in the LPS-only group; however, its enrichment among genes that are differentially up-regulated in MV+LPS relative to LPS implies that ISRE remains a central regulatory motif during the development of ALI. Our computational analysis predicts that intervention at this regulatory site may have a profound impact on the expression of genes responsible for the severe phenotype seen in MV+LPS.
Sp1 was the most significantly overrepresented TF among down-regulated genes during MV+LPS relative to control, MV, and LPS. This TF is ubiquitously expressed in mammalian cells and regulates functionally diverse groups of genes (28). Sp1 binding activity has been shown to be down-regulated in lungs of mice exposed to systemic LPS; furthermore, this down-regulation was accompanied by a reduced expression of several Sp1-dependent genes (29). In our model of lung injury, Sp1 binds to upstream motifs of genes within several biological modules, including many genes assigned to the "oxidoreductase activity," "catalytic activity," and "glutathione transferase activity" modules. In fact, seven out of 14 members of the glutathione S-transferase family down-regulated during MV+LPS have putative upstream binding motifs for Sp1, suggesting a novel role for this TF in regulating oxidative stress-related pathways during lung injury.
There are several limitations in this study. We used transcriptional profiling of whole lungs to obtain gene expression values and therefore have not determined the location of differentially expressed genes. Our study was limited to only one time point during a temporally dynamic and complex pathophysiologic process. Earlier and later time points in this model of ALI may have different gene expression profiles and subsequently involve different biological modules and regulatory mechanisms. We chose a relatively permissive false-discovery rate to maximize our discovery of truly differentially expressed genes while accepting that approximately 5% of these genes were chosen in error. Although we could narrow the candidate gene list by using more restrictive criteria, such as lower false-discovery rate or a fold-change cutoff, we believed that our gene ontology and TF analyses would best capture a "global snapshot" of lung injury by including as many truly differentially expressed genes as statistically acceptable.
In summary, we have outlined a novel computational framework for the systematic integration of transcriptional profiling with gene ontology and TF analysis to understand the pathogenesis of ALI in a clinically relevant mouse model. Our approach creates a global map of the important biological modules activated during lung injury in the context of their putative regulatory networks and identifies several novel molecular mechanisms and regulatory sites that may be targets for future therapeutic intervention. An important feature of this map is its integration of known functional classification of gene products with regulatory motifs that are identified using an unbiased search algorithm. In this manner, putative TFs can be given biologically plausible regulatory roles based on their enrichment during specific perturbations. Therefore, our computational methodology is generally applicable to any gene expression profiling experiment involving eukaryotic organisms.
Acknowledgments
The authors thank Dowon An, Shen-Sheng Wang, and Wilson Slaid Jones for their assistance with animal experiments and RNA isolation. They also thank the staff at University of Washington's Center for Expression Arrays for labeling and hybridizing the oligonucleotide microarrays.
FOOTNOTES
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200509-1473OC on December 30, 2005
Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
Ware LB, Matthay MA. The acute respiratory distress syndrome. N Engl J Med 2000;342:1334–1349.
The Acute Respiratory Distress Syndrome Network. N Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301–1308.
Cheng IW, Matthay MA. Acute lung injury and the acute respiratory distress syndrome. Crit Care Clin 2003;19:693–712.
Matthay MA, Zimmerman GA, Esmon C, Bhattacharya J, Coller B, Doerschuk CM, Floros J, Gimbrone MA Jr, Hoffman E, Hubmayr RD, et al. Future research directions in acute lung injury: summary of a National Heart, Lung, and Blood Institute working group. Am J Respir Crit Care Med 2003;167:1027–1035.
Altemeier WA, Matute-Bello G, Gharib SA, Glenny RW, Martin TR, Liles WC. Modulation of LPS-induced gene transcription and promotion of lung injury by mechanical ventilation. J Immunol 2005;175:3369–3376.
Gharib SA, Matute-Bello G, Glenny RW, Martin TR, Liles WC, Altemeier WA. Identification of key molecular pathways and transcription factors in a mouse model of acute lung injury . Proc Am Thorac Soc 2005;2:A886.
Bolstad BM, Irizarry RA, Astrand M, Speed TP. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2003;19:185–193.
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001;98:5116–5121.
Kaufman L, Rousseeuw PJ. Finding groups in data: an introduction to cluster analysis. New York, NY: Wiley; 1990.
Gene Ontology Consortium. Creating the gene ontology resource: design and implementation. Genome Res 2001;11:1425–1433.
Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC, Conklin BR. MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol 2003;4:R7.
Hosack DA, Dennis G Jr, Sherman BT, Lane HC, Lempicki RA. Identifying biological themes within lists of genes with EASE. Genome Biol 2003;4:R70.
Sharan R, Maron-Katz A, Shamir R. CLICK and EXPANDER: a system for clustering and visualizing gene expression data. Bioinformatics 2003;19:1787–1799.
Elkon R, Linhart C, Sharan R, Shamir R, Shiloh Y. Genome-wide in silico identification of transcriptional regulators controlling the cell cycle in human cells. Genome Res 2003;13:773–780.
Matys V, Fricke E, Geffers R, Gossling E, Haubrock M, Hehl R, Hornischer K, Karas D, Kel AE, Kel-Margoulis OV, et al. TRANSFAC: transcriptional regulation, from patterns to profiles. Nucleic Acids Res 2003;31:374–378.
Janeway CA Jr, Medzhitov R. Innate immune recognition. Annu Rev Immunol 2002;20:197–216.
Simmonds AC, Ghanayem BI, Sharma A, Reilly CA, Millen B, Yost GS, Forkert PG. Bioactivation of 1,1-dichloroethylene by CYP2E1 and CYP2F2 in murine lung. J Pharmacol Exp Ther 2004;310:855–864.
Bradford BU, Kono H, Isayama F, Kosyk O, Wheeler MD, Akiyama TE, Bleye L, Krausz KW, Gonzalez FJ, Koop DR, et al. Cytochrome P450 CYP2E1, but not nicotinamide adenine dinucleotide phosphate oxidase, is required for ethanol-induced oxidative DNA damage in rodent liver. Hepatology 2005;41:336–344.
Granville DJ, Tashakkor B, Takeuchi C, Gustafsson AB, Huang C, Sayen MR, Wentworth P Jr, Yeager M, Gottlieb RA. Reduction of ischemia and reperfusion-induced myocardial damage by cytochrome P450 inhibitors. Proc Natl Acad Sci USA 2004;101:1321–1326.
Harbrecht BG, Frye RF, Zenati MS, Branch RA, Peitzman AB. Cytochrome P-450 activity is differentially altered in severely injured patients. Crit Care Med 2005;33:541–546.
Hayes JD, Flanagan JU, Jowsey IR. Glutathione transferases. Annu Rev Pharmacol Toxicol 2005;45:51–88.
Altemeier WA, Matute-Bello G, Frevert CW, Kawata Y, Kajikawa O, Martin TR, Glenny RW. Mechanical ventilation with moderate tidal volumes synergistically increases lung cytokine response to systemic endotoxin. Am J Physiol Lung Cell Mol Physiol 2004;287:L533–L542.
Kirchner EA, Mols G, Hermle G, Muehlschlegel JD, Geiger KK, Guttmann J, Pahl HL. Reduced activation of immunomodulatory transcription factors during positive end-expiratory pressure adjustment based on volume-dependent compliance in isolated perfused rabbit lungs. Br J Anaesth 2005;94:530–535.
Oudin S, Pugin J. Role of MAP kinase activation in interleukin-8 production by human BEAS-2B bronchial epithelial cells submitted to cyclic stretch. Am J Respir Cell Mol Biol 2002;27:107–114.
Ning S, Huye LE, Pagano JS. Regulation of the transcriptional activity of the IRF7 promoter by a pathway independent of interferon signaling. J Biol Chem 2005;280:12262–12270.
Brierley MM, Fish EN. Functional relevance of the conserved DNA-binding domain of STAT2. J Biol Chem 2005;280:13029–13036.
Taniguchi T, Ogasawara K, Takaoka A, Tanaka N. IRF family of transcription factors as regulators of host defense. Annu Rev Immunol 2001;19:623–655.
Li L, He S, Sun JM, Davie JR. Gene regulation by Sp1 and Sp3. Biochem Cell Biol 2004;82:460–471.
Ye X, Liu SF. Lipopolysaccharide down-regulates Sp1 binding activity by promoting Sp1 protein dephosphorylation and degradation. J Biol Chem 2002;277:31863–31870.(Sina A. Gharib, W. Conrad Liles, Gustavo)
the Veterans Affairs Puget Sound Healthcare System Medical Research Service, Seattle, Washington
ABSTRACT
Rationale: Mechanical ventilation augments the acute lung injury (ALI) caused by bacterial products. The molecular pathogenesis of this synergistic interaction remains incompletely understood.
Objective: We sought to develop a computational framework to systematically identify gene regulatory networks activated in ALI.
Methods: We have developed a mouse model in which the combination of mechanical ventilation and intratracheal LPS produces significantly more injury to the lung than either insult alone. We used global gene ontology analysis to determine overrepresented biological modules and computational transcription factor analysis to identify putative regulatory factors involved in this model of ALI.
Results: By integrating expression profiling with gene ontology and promoter analysis, we constructed a large-scale regulatory modular map of the important processes activated in ALI. This map assigned differentially expressed genes to highly overrepresented biological modules, including "defense response," "immune response," and "oxidoreductase activity." These modules were then systematically incorporated into a gene regulatory network that consisted of putative transcription factors, such as IFN-stimulated response element, IRF7, and Sp1, that may regulate critical processes involved in the pathogenesis of ALI.
Conclusions: We present a novel, unbiased, and powerful computational approach to investigate the synergistic effects of mechanical ventilation and LPS in promoting ALI. Our methodology is applicable to any expression profiling experiment involving eukaryotic organisms.
Key Words: acute lung injury gene network microarray transcription factor
Acute lung injury (ALI) is an important cause of in-hospital morbidity and mortality in critically ill patients. Infectious etiologies, such as sepsis and pneumonia, are among the most common risk factors for developing ALI (1). Mechanical ventilation (MV) is a life-saving maneuver in ALI, but in patients with established lung injury, MV strategy can affect mortality and respiratory failure (2). Despite recent advances, many molecular mechanisms in ALI remain unexplored, and few therapeutic interventions targeting identified molecular pathways have been clinically effective (3, 4). A critical first step in discovering novel therapies in ALI is to gain a deeper understanding of the complex pathophysiologic processes involved in the interaction between MV and infectious insults that lead to lung injury.
We have recently developed a mouse model in which ALI results from the combination of mechanical ventilation and low-dose intratracheal LPS (5). Our model captures two crucial perturbations, mechanical ventilation and infectious insult, that are frequently associated with ALI in the clinical setting. An important feature of this animal model is the observation that neither MV nor LPS alone causes significant lung injury; ALI develops only when both interventions are combined, suggesting a synergistic role for MV in modulating the effects of LPS. We hypothesized that the synergism between MV and LPS depends on the activation of distinct transcription factors (TFs) that trigger key biological processes leading to ALI. Understanding the molecular mechanisms involved in this "double hit" paradigm can shed light on the pathogenesis of ALI in humans.
In this article, we outline a systematic approach to integrate transcriptional profiling with global gene annotation and TF analysis to obtain a large-scale regulatory modular map of the key processes involved in our model of ALI. Our computational analyses determine important biological modules in the context of their putative regulatory networks and identify TFs that may regulate critical biological processes involved in the development of ALI, thereby highlighting potential targets for therapeutic intervention. Some of the results of these studies have been previously reported in the form of an abstract (6).
METHODS
Animal Experiments
All experiments were approved by University of Washington's Animal Care Committee. Twenty-four C57Bl/6 8-wk-old male mice (Harlan; Indianapolis, IN) were randomly assigned to four groups (n = 6/group): (1) phosphate-buffered saline (PBS) aspiration and spontaneous breathing (control), (2) PBS with MV, (3) LPS aspiration and spontaneous breathing (LPS), or (4) LPS with mechanical ventilation (MV+LPS).
After each mouse was anesthetized, 5 ng/g body weight of LPS (Escherichia coli serotype 0111:B4; List Laboratories, Campbell, CA) in the LPS and MV+LPS groups, or an equal volume of sterile PBS (in the control and MV groups) was deposited in the posterior oropharynx. Thirty minutes after aspiration, MV and MV+LPS mice were reanesthetized, intubated, and mechanically ventilated with a tidal volume of 10 ml/kg, respiratory rate of 150/min, FIO2 of 0.21, and 0 cm H2O end-expiratory pressure (MiniVent; Harvard Biosciences, Holliston, MA) for 4 h. The control and LPS mice breathed spontaneously during this 4-h period. At the conclusion of the intervention period, mice were killed by exsanguination, lungs were flushed with RNAse-free PBS via the right ventricle and homogenized, and total RNA was isolated using the RNeasy Midi Kit (Qiagen; Valencia, CA). RNA integrity was confirmed using the Bioanalyzer 2100 (Agilent; Palo Alto, CA).
Histology
For each group, two additional mice were killed at the end of the intervention period, and their lungs removed and fixed via intratracheal instillation of 4% formalin at a transmural pressure of 20 cm H2O. After fixation, the lungs were embedded in paraffin, and 4-μm sections were stained with hematoxylin and eosin (H&E).
Microarray Data Analysis
For each mouse, labeled cRNA was prepared from total RNA, hybridized to an Affymetrix MOE430A microarray, and scanned. Image analysis performed using Affymetrix MAS 5.0 software (Affymetrix, Santa Clara, CA). Background adjustment and quantile normalization across all 24 microarrays were performed using the Robust Multichip Average algorithm (7). Statistically significant differential gene expression was determined using Significance Analysis of Microarrays (8). The problem of multiple comparisons was addressed by choosing a false-discovery rate of 5%. Significantly differentially expressed genes during MV, LPS, or MV+LPS relative to control were clustered using the partitioning around medoids (PAM) algorithm (9). All microarray data, in compliance with Minimum Information about a Microarray Experiment, are available at the GEO website (http://www.ncbi.nlm.nih.gov/projects/geo/, query GSE 2411).
Gene Ontology Analysis
Gene annotation of all probesets present on each GeneChip was obtained from the Gene Ontology database (10). Highly overrepresented biological modules in our experimental conditions were determined using two different software programs, Gene MicroArray Pathway Profiler (GenMAPP) and Expression Analysis Systematic Explorer (EASE) (11, 12). To improve statistical confidence in the results, all identified biological modules had to be statistically significant using both approaches.
TF Analysis
Putative overrepresented TFs within subsets of differentially expressed genes were identified using the PRIMA (PRomoter Integration in Microarray Analysis) algorithm (13, 14) and based on the position weight matrices of approximately 500 known eukaryotic TFs (TRANSFAC database) (15). Multiple comparisons problem was addressed by selecting only overrepresented TFs with Bonferroni-corrected p values of less than 0.05.
RESULTS
Combination of MV and LPS Results in Severe Lung Injury
Figure 1 shows H&E stains of mice lungs exposed to each of the four experimental conditions (control, MV, LPS, and MV+LPS) for 6 h. Lungs from the MV-only group were not histologically different from the control group and had no evidence of inflammation or injury. In the LPS-only group, there was modest cellular infiltration in the alveolar spaces, but no edema or septal thickening was seen. In contrast, the lungs from mice in MV+LPS group showed a profound neutrophilic infiltration, septal thickening, and perivascular edema consistent with ALI.
Combination of MV and LPS Results in Global Augmentation of the Transcriptional Response in the Lung
Differential gene expression during each perturbation (MV, LPS, MV+LPS) relative to control was determined for n = 6 mice in each group using significance analysis of microarrays (SAM) and choosing a false-discovery rate of 5%. A total of 7,510 genes were differentially expressed during at least one of the three perturbations relative to the unperturbed control condition. There were 1,415 differentially expressed genes in the MV group (838 up-regulated, 577 down-regulated), 4,536 genes in the LPS group (1,366 up-regulated, 3,170 down-regulated), and 6,837 genes in the MV+LPS group (2,418 up-regulated, 4,419 down-regulated; Figure 2).
We next applied PAM algorithm to cluster all 7,510 differentially expressed genes across their perturbation states (MV, LPS, MV+LPS). Using PAM's quality index and silhouette plot, we determined that this dataset could be most parsimoniously divided into two clusters. This finding was confirmed using principal component analysis, which revealed only two distinct groups of genes in perturbation space (see online supplement). Figure 2 displays these two clusters of gene expression, demonstrating the global augmentation of transcriptional response with the combination of mechanical ventilation and LPS (MV+LPS). The MV+LPS group contains the largest number of differentially expressed genes and, on average, the greatest level of differential expression.
Global Gene Ontology Analysis Reveals Important Biological Modules in ALI
To identify important molecular pathways activated in lung injury due to the synergistic effect of MV and LPS, we focused our attention on the subset of genes that were most likely to differentiate the ALI phenotype (MV+LPS) from the other perturbations. We determined that 2,567 genes were differentially expressed (809 up-regulated, 1,758 down-regulated) in the MV+LPS group relative to all three other conditions (control, MV, LPS) at a false-discovery rate of 5% (see online supplement). Next, we performed global gene ontology analysis using two independent statistical methods, GenMAPP and EASE, to identify overrepresented biological themes within these differentially expressed genes. Tables 1 and 2 list highly significant biological modules determined by GenMAPP and EASE among differentially up-regulated and down-regulated genes, respectively. Among up-regulated genes, response to biotic stimulus, defense response, immune response, RNA and protein processing, and metabolism were among the most overrepresented modules. Highly overabundant modules among down-regulated genes included oxidoreductase and catalytic activity, glutathione transferase activity, extracellular matrix constituents, and organic and fatty acid metabolic pathways.
In Silico Promoter Analysis Identifies Putative TFs Activated during MV, LPS, and Lung Injury due to MV+LPS
To discover putative TFs regulating the lung's response to mechanical ventilation, LPS, and injury from the combination of MV+LPS, we undertook a computational search for overrepresented promoter sites among genes differentially expressed during each of these perturbations. Overrepresentation of these binding sites in differentially regulated genes suggests a possible role for their associated TFs. Table 3 lists highly significant putative TFs identified based on genes differentially expressed during MV, LPS, or MV+LPS relative to Control. In general, under each condition, a different repertoire of TFs is highly enriched. For example, the LPS-only group included TFs known to be involved in the cellular response to LPS, such as nuclear factor-B (NF-B), cRel, and interferon regulatory factor (IRF) (16). None of these TFs was present in the MV-only group, which included ETF, E2F, and Nrf1. The MV+LPS group shared several TFs with LPS-only and MV-only conditions, implying that gene regulation in this condition was strongly influenced by mechanical ventilation and LPS. Among down-regulated genes, all three perturbations shared Sp1 as the most overrepresented TF.
We next focused our attention on the subset of 2,567 genes differentially expressed during MV+LPS relative to all other conditions (control, MV, and LPS); these genes differentiate the transcriptional response between the lung injury phenotype seen in MV+LPS compared with the other perturbations. Table 4 lists highly overrepresented TFs among these genes. This list includes TFs, such as NF-B and ETF, that were previously identified in LPS-only and MV-only conditions, respectively, suggesting that common regulatory factors are, in part, responsible for the severe lung injury phenotype seen in MV+LPS. However, we also identified several TFs, such as Myc, USF2, and Elk1, that were not enriched among differentially expressed genes during MV or LPS. These factors may represent novel sites of gene regulation responsible for the molecular pathways leading to ALI.
Integration of Global Gene Annotation and TF Analysis: Constructing a Gene Regulatory Map of ALI
Promoter analysis of 2,567 differentially expressed genes in the MV+LPS group relative to Control, MV, and LPS identified overrepresented putative TFs (Table 4). Each of these TFs recognizes specific binding site(s) in the 5' flanking region of a subset of these 2,567 differentially expressed genes. We systematically performed gene ontology analysis on subsets of differentially expressed genes that possess binding sites for the same putative TF. In this manner, overrepresented biological modules regulated by each enriched TF were identified. The results of this analysis were integrated with our original gene ontology analysis of all 2,567 differentially expressed genes in the MV+LPS group to construct a modular gene regulatory map. Figure 3 displays a graphical representation of the hierarchical and modular structure of this regulatory map, highlighting important biological processes and their putative regulators in this model of ALI. For example, the "immune response" module is composed of differentially up-regulated genes, many of which share common binding motifs for IFN-stimulated response element (ISRE), whereas genes assigned to the "programmed cell death" module primarily recognize IRF7 in their upstream promoter sites. Several biological modules, such as "RNA processing," "RNA metabolism," and "intracellular signaling" recognize multiple TFs, suggesting complex interactions among these factors and the genes they regulate.
DISCUSSION
The main finding of this study is that the combination of a noninjurious ventilatory strategy with low-dose intratracheal LPS induces a form of ALI that is characterized by the activation of distinct sets of biological modules and TFs when compared with MV or LPS alone. We identified these processes by systematically integrating expression profiling with gene ontology and TF analysis. The regulatory modular map resulting from our computational methods provides information on which genes are differentially expressed during lung injury from MV+LPS, which biological modules are activated in this complex process, and which putative TFs are regulating the genes within these modules (Figure 3).
We initially showed that there is a global augmentation of transcriptional response in the lung when a noninjurious mechanical ventilation strategy is combined with low-dose LPS administration (Figure 2). We used gene ontology analysis to identify important biological modules activated during the ALI that results from this "double hit" perturbation (MV+LPS). Initially, we annotated gene products based on their molecular function, biological process, or cellular component using the Gene Ontology Consortium's database (http://www.geneontology.org/). Next, functional categories (biological modules) with overrepresentation of differentially expressed genes were identified by calculating the probability of this enrichment occurring by chance (see Tables 2 and 3, and online supplement). The presence of several statistically significant modules involved in RNA and protein metabolism and biosynthesis implies a general activation of these pathways during ALI. Another highly significant biological process overrepresented among differentially up- regulated genes in the MV+LPS group relative to all other conditions (Control, MV, LPS) was "immune response." Genes assigned to this module included CD14, TLR4, NF-B, MD-1, and MD-2, all of which are crucial in recognizing and mediating the innate immune response to LPS (16). Other differentially up-regulated genes in this module included cytokines and chemokines (IL15, Cxcl5, Cxcl16, Ccr1, Ccr2), other Toll-like receptors (TLR2, TLR6, TLR7), and several IFN-induced molecules (Ifi16, Ifi35, Ifi203, Ifi205, Isgf3 g, Ifitm3). These genes highlight activated pathways that are likely to result in the severe lung injury seen during MV+LPS.
The most significant biological module among genes down-regulated during MV+LPS relative to Control, MV, and LPS was "oxidoreductase activity." Several members of the cytochrome P450 family were assigned to this module, including Cyp1a1, Cyp2d22, Cyp2f2, Cyp4f13, Cyp4v3, and Cyp27a1. Alteration in the activity of cytochrome P450 family of enzymes has been reported during various models of organ injury, including chemical and viral hepatitis, ischemia/reperfusion injury of the heart, and inhalation/chemical injury to the lungs (17–19). Recently, the activity of cytochrome P450 was shown to be significantly suppressed in critically ill trauma patients and strongly correlated with the degree of organ failure (20).
Another significant biological module among genes down-regulated in the MV+LPS group was "glutathione transferase activity," which consisted of 14 different members of the glutathione S-transferase family. Glutathione transferases perform a crucial role in protecting cells from oxidative stress and other toxic insults (21). Down-regulation of members of this module during MV+LPS suggests a novel mechanism in the pathogenesis of ALI.
To our knowledge, this is the first study to undertake a systematic computational search for putative TFs involved in the lung's response to mechanical ventilation, LPS, and acute injury due to MV+LPS. We and others have shown that factors regulating inflammation, such as NF-B and AP-1, are activated when lungs are mechanically ventilated at high tidal volumes that independently result in stretch injury (22, 23). However, we now report several putative TFs during a noninjurious mechanical ventilation strategy that is much more representative of current clinical practice. Highly significant regulators of gene expression in the MV-only group that were computationally identified included ETF, E2F, Nrf1, and CREB (cAMP-responsive element binding protein). CREB has been shown to be activated during in vitro mechanical stretching of lung bronchial epithelial cells, and more recently, during mechanical ventilation of isolated perfused rabbit lungs (23, 24).
Several TFs known to regulate cellular response to LPS, such as NF-B and IRF families, were among the most significantly enriched factors in the LPS-only group, thereby confirming the robustness of our computational approach. Although there was minimal histologic evidence of lung injury in mice exposed to low-dose LPS, our analysis revealed a profound inflammatory transcriptional and regulatory response during this perturbation.
We identified several putative TFs that were highly overrepresented during ALI from the combination of MV and LPS. By focusing on the subset of genes differentially expressed in the MV+LPS group relative to all other conditions (Control, MV, LPS), we searched for TFs that were most likely to regulate the severe injury seen in MV+LPS. We then systematically integrated these results with global gene ontology analysis of the same subset of genes to construct a large-scale modular regulatory map of ALI (Figure 3).
Myc and USF2 were among the most overrepresented putative TFs identified. Figure 3 reveals that they regulated primarily genes involved in RNA metabolism and processing. IRF7 was another highly overrepresented TF in the MV+LPS group and primarily regulated apoptotic and protein transport/targeting modules. ISRE was a highly enriched binding motif regulating several biological modules involved in immune and defense response. The activation of genes within these modules is likely to be crucial in development of the lung injury seen during MV+LPS. ISRE is a key binding site for several TFs, such as IRFs (IRF1, IRF3, IRF4, IRF7, IRF9), STAT1, and STAT2, that regulate IFN-mediated cellular signaling during viral/bacterial infection, cytokine stimulation, and innate immune response (25–27). Therefore, it was not surprising that ISRE was highly overrepresented in the LPS-only group; however, its enrichment among genes that are differentially up-regulated in MV+LPS relative to LPS implies that ISRE remains a central regulatory motif during the development of ALI. Our computational analysis predicts that intervention at this regulatory site may have a profound impact on the expression of genes responsible for the severe phenotype seen in MV+LPS.
Sp1 was the most significantly overrepresented TF among down-regulated genes during MV+LPS relative to control, MV, and LPS. This TF is ubiquitously expressed in mammalian cells and regulates functionally diverse groups of genes (28). Sp1 binding activity has been shown to be down-regulated in lungs of mice exposed to systemic LPS; furthermore, this down-regulation was accompanied by a reduced expression of several Sp1-dependent genes (29). In our model of lung injury, Sp1 binds to upstream motifs of genes within several biological modules, including many genes assigned to the "oxidoreductase activity," "catalytic activity," and "glutathione transferase activity" modules. In fact, seven out of 14 members of the glutathione S-transferase family down-regulated during MV+LPS have putative upstream binding motifs for Sp1, suggesting a novel role for this TF in regulating oxidative stress-related pathways during lung injury.
There are several limitations in this study. We used transcriptional profiling of whole lungs to obtain gene expression values and therefore have not determined the location of differentially expressed genes. Our study was limited to only one time point during a temporally dynamic and complex pathophysiologic process. Earlier and later time points in this model of ALI may have different gene expression profiles and subsequently involve different biological modules and regulatory mechanisms. We chose a relatively permissive false-discovery rate to maximize our discovery of truly differentially expressed genes while accepting that approximately 5% of these genes were chosen in error. Although we could narrow the candidate gene list by using more restrictive criteria, such as lower false-discovery rate or a fold-change cutoff, we believed that our gene ontology and TF analyses would best capture a "global snapshot" of lung injury by including as many truly differentially expressed genes as statistically acceptable.
In summary, we have outlined a novel computational framework for the systematic integration of transcriptional profiling with gene ontology and TF analysis to understand the pathogenesis of ALI in a clinically relevant mouse model. Our approach creates a global map of the important biological modules activated during lung injury in the context of their putative regulatory networks and identifies several novel molecular mechanisms and regulatory sites that may be targets for future therapeutic intervention. An important feature of this map is its integration of known functional classification of gene products with regulatory motifs that are identified using an unbiased search algorithm. In this manner, putative TFs can be given biologically plausible regulatory roles based on their enrichment during specific perturbations. Therefore, our computational methodology is generally applicable to any gene expression profiling experiment involving eukaryotic organisms.
Acknowledgments
The authors thank Dowon An, Shen-Sheng Wang, and Wilson Slaid Jones for their assistance with animal experiments and RNA isolation. They also thank the staff at University of Washington's Center for Expression Arrays for labeling and hybridizing the oligonucleotide microarrays.
FOOTNOTES
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200509-1473OC on December 30, 2005
Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
Ware LB, Matthay MA. The acute respiratory distress syndrome. N Engl J Med 2000;342:1334–1349.
The Acute Respiratory Distress Syndrome Network. N Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301–1308.
Cheng IW, Matthay MA. Acute lung injury and the acute respiratory distress syndrome. Crit Care Clin 2003;19:693–712.
Matthay MA, Zimmerman GA, Esmon C, Bhattacharya J, Coller B, Doerschuk CM, Floros J, Gimbrone MA Jr, Hoffman E, Hubmayr RD, et al. Future research directions in acute lung injury: summary of a National Heart, Lung, and Blood Institute working group. Am J Respir Crit Care Med 2003;167:1027–1035.
Altemeier WA, Matute-Bello G, Gharib SA, Glenny RW, Martin TR, Liles WC. Modulation of LPS-induced gene transcription and promotion of lung injury by mechanical ventilation. J Immunol 2005;175:3369–3376.
Gharib SA, Matute-Bello G, Glenny RW, Martin TR, Liles WC, Altemeier WA. Identification of key molecular pathways and transcription factors in a mouse model of acute lung injury . Proc Am Thorac Soc 2005;2:A886.
Bolstad BM, Irizarry RA, Astrand M, Speed TP. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2003;19:185–193.
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001;98:5116–5121.
Kaufman L, Rousseeuw PJ. Finding groups in data: an introduction to cluster analysis. New York, NY: Wiley; 1990.
Gene Ontology Consortium. Creating the gene ontology resource: design and implementation. Genome Res 2001;11:1425–1433.
Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC, Conklin BR. MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol 2003;4:R7.
Hosack DA, Dennis G Jr, Sherman BT, Lane HC, Lempicki RA. Identifying biological themes within lists of genes with EASE. Genome Biol 2003;4:R70.
Sharan R, Maron-Katz A, Shamir R. CLICK and EXPANDER: a system for clustering and visualizing gene expression data. Bioinformatics 2003;19:1787–1799.
Elkon R, Linhart C, Sharan R, Shamir R, Shiloh Y. Genome-wide in silico identification of transcriptional regulators controlling the cell cycle in human cells. Genome Res 2003;13:773–780.
Matys V, Fricke E, Geffers R, Gossling E, Haubrock M, Hehl R, Hornischer K, Karas D, Kel AE, Kel-Margoulis OV, et al. TRANSFAC: transcriptional regulation, from patterns to profiles. Nucleic Acids Res 2003;31:374–378.
Janeway CA Jr, Medzhitov R. Innate immune recognition. Annu Rev Immunol 2002;20:197–216.
Simmonds AC, Ghanayem BI, Sharma A, Reilly CA, Millen B, Yost GS, Forkert PG. Bioactivation of 1,1-dichloroethylene by CYP2E1 and CYP2F2 in murine lung. J Pharmacol Exp Ther 2004;310:855–864.
Bradford BU, Kono H, Isayama F, Kosyk O, Wheeler MD, Akiyama TE, Bleye L, Krausz KW, Gonzalez FJ, Koop DR, et al. Cytochrome P450 CYP2E1, but not nicotinamide adenine dinucleotide phosphate oxidase, is required for ethanol-induced oxidative DNA damage in rodent liver. Hepatology 2005;41:336–344.
Granville DJ, Tashakkor B, Takeuchi C, Gustafsson AB, Huang C, Sayen MR, Wentworth P Jr, Yeager M, Gottlieb RA. Reduction of ischemia and reperfusion-induced myocardial damage by cytochrome P450 inhibitors. Proc Natl Acad Sci USA 2004;101:1321–1326.
Harbrecht BG, Frye RF, Zenati MS, Branch RA, Peitzman AB. Cytochrome P-450 activity is differentially altered in severely injured patients. Crit Care Med 2005;33:541–546.
Hayes JD, Flanagan JU, Jowsey IR. Glutathione transferases. Annu Rev Pharmacol Toxicol 2005;45:51–88.
Altemeier WA, Matute-Bello G, Frevert CW, Kawata Y, Kajikawa O, Martin TR, Glenny RW. Mechanical ventilation with moderate tidal volumes synergistically increases lung cytokine response to systemic endotoxin. Am J Physiol Lung Cell Mol Physiol 2004;287:L533–L542.
Kirchner EA, Mols G, Hermle G, Muehlschlegel JD, Geiger KK, Guttmann J, Pahl HL. Reduced activation of immunomodulatory transcription factors during positive end-expiratory pressure adjustment based on volume-dependent compliance in isolated perfused rabbit lungs. Br J Anaesth 2005;94:530–535.
Oudin S, Pugin J. Role of MAP kinase activation in interleukin-8 production by human BEAS-2B bronchial epithelial cells submitted to cyclic stretch. Am J Respir Cell Mol Biol 2002;27:107–114.
Ning S, Huye LE, Pagano JS. Regulation of the transcriptional activity of the IRF7 promoter by a pathway independent of interferon signaling. J Biol Chem 2005;280:12262–12270.
Brierley MM, Fish EN. Functional relevance of the conserved DNA-binding domain of STAT2. J Biol Chem 2005;280:13029–13036.
Taniguchi T, Ogasawara K, Takaoka A, Tanaka N. IRF family of transcription factors as regulators of host defense. Annu Rev Immunol 2001;19:623–655.
Li L, He S, Sun JM, Davie JR. Gene regulation by Sp1 and Sp3. Biochem Cell Biol 2004;82:460–471.
Ye X, Liu SF. Lipopolysaccharide down-regulates Sp1 binding activity by promoting Sp1 protein dephosphorylation and degradation. J Biol Chem 2002;277:31863–31870.(Sina A. Gharib, W. Conrad Liles, Gustavo)