[Frontiers in Bioscience S4, 1556-1567, June 1, 2012]
Transcriptomic analysis reveals pH-responsive antioxidant gene networks
Rodrigo Juliani Siqueira Dalmolin1, Daniel Pens Gelain1, Fabio Klamt1, Mauro Antonio Alves Castro1, Jose Claudio Fonseca Moreira1
1
TABLE OF CONTENTS
1. ABSTRACT
Reactive oxygen species (ROS) are produced in different physiological conditions. In response to ROS imbalance cells activate oxidative stress defenses, which include more than 60 antioxidant genes. It has been suggested that gene products associated with ROS detoxification can work coordinately, acting as an antioxidant-defense network. However, the functional overlap among oxidative stress defenses and other related cell functions makes difficult the characterization of this network. We previously described a network-based model to characterize the interactions existing among different antioxidant gene products and their substrates. Here, we test whether this network-based model of human antioxidant genes can respond to different physiological conditions. We used a systems biology approach applied to the analysis of two independent gene expression datasets: transcriptomes from HeLa cells and primary astrocytes maintained under hypoxic conditions and transcriptomes from SKGT4 cells exposed to low pH environment. We found that the proposed gene network model responds selectively to both hypoxia and acidosis. We anticipate that this antioxidant gene network model can be helpful to describe stress-responsive expression profiles in different cell types.
2. INTRODUCTION
Reactive oxygen species (ROS) are produced during normal cellular metabolism. ROS mediate diverse physiological responses by acting in several signaling pathways (1-3). In addition, ROS overproduction may lead to cell oxidative damage, playing a causative role in numerous diseases (4). To maintain the homeostasis between ROS production and detoxification, cells developed a complex antioxidant apparatus including endogenous and exogenous components.
The most classical enzymatic antioxidant defense includes catalase, superoxide dismutase and glutathione peroxidase (5-6). They are representative members of enzyme families that have many other constituents (e.g. GPX family members). In addition, non-enzymatic components (such as thiol-containing proteins) play essential role in endogenous antioxidant defenses (7-8).
Although gene products associated with antioxidant defenses are in general investigated individually, increasing evidences indicate that they work coordinately as a network (9-12). However, the functional overlap among antioxidant defenses and a number of related cell functions - together with cell compartmentalization of antioxidant key elements - makes difficult the precise characterization of such network. One possible approach to study complex and entangled networks is the use of gene expression data from high-throughput experiments (13).
Gene expression networks are frequently linked with the traditional concept of regulatory pathways that describe cascade of regulatory events and provide the basis for genetic network reconstruction (14). However, such traditional concept has its limitations, since it aims to study properties of individual network components rather than consider all components and their interactions simultaneously (15-16). In order to evaluate the systemic properties of antioxidant gene networks we previously described a network-based model of human antioxidant genes characterizing the interactions existing among different antioxidant key components and its substrates (Figure 1) (7-8). In this model, all genes with prime antioxidant function are considered together with their substrate/product reactions to construct the network interactions.
Here, we accessed the gene expression activity of the antioxidant gene network components in HeLa cells and primary astrocytes submitted to hypoxic condition to test our network-based model. We also analyzed the gene expression activity of the antioxidant gene network in esophageal adenocarcinoma SKGT4 cells exposed to low pH environment. We have found that the proposed antioxidant gene network responds selectively to hypoxia in HeLa cells and low pH treatment in SKGT4 cells.
3. METHODS
3.1. Network-based model of human antioxidant genes
The network-based model of human antioxidant genes has been previously described (7-8) and is provided at the Human Antioxidant Genes (HAG) database (www.ufrgs.br/icbs/hag). The HAG website shows a manually curated set of genes involved in antioxidant defenses. The gene list is classified as following: (1) genes whose products are widely accepted as being primarily antioxidant enzymes; (2) genes whose products may not be antioxidant enzymes, but are also recognized as having a primary antioxidant function; (3) genes whose products are recognized as having a primary role in other physiologic processes, but are known to physically interact with reactive species, also converting them into less or non-toxic products, through redox reactions carried by functional groups; and (4) genes whose products were described to maintain exclusive, specific interactions with the gene products described above, being essential for the antioxidant function of such proteins. In total, 63 antioxidant genes are listed in the gene network, which is subdivided according to three categories: i) peroxidases (17 genes), ii) superoxide dismutases (4 genes), and iii) thiol redox (42 genes). These gene sets hereby are referred to as Groups of Functionally Associated Genes (GFAG). Each gene in the network is identified according to the approved HGNC gene symbol (Figure 1) (17). The network showing edge relationships was handled in Medusa software (18). Functional categories are discriminated by different colors and protein data are crossed with substrates in order to integrate the network. For this purpose we considered the substrates annotated for each gene product in KEGG database (19).
3.2. Data selection
Microarray data of HeLa cells were obtained from GEO database, series GSE3051 (http://www.ncbi.nlm.nih.gov/projects/geo/). This dataset comprises 6 arrays (Affymetrix GeneChip Human Genome U133 Plus 2.0 Array), as previously described (20). The experimental design provides three independent sample-sets grown under normoxic (95% air-5% CO2) and three independent sample-sets grown under hypoxic conditions (1% O2-5% CO2-94% N2) for 24 hours. Microarray data of human primary astrocytes follow the same experimental design (GEO database, series GSE3045). Microarray data of esophageal adenocarcinoma SKGT4 cell line were also obtained from GEO database, series GSE2144 (http://www.ncbi.nlm.nih.gov/projects/geo/). This dataset comprises 10 arrays (Affymetrix GeneChip Human Genome U133 Array Set HG-U133A), as previously described (21). This study was originally designed to identify pathways involved in the progression of gastro-esophageal reflux disease through Barrett's esophagus to adenocarcinoma. The SKGT4 esophageal cells were stimulated with Low pH environment (pH=6.5) for 30, 120, 180 and 240 minutes (n=2 for each treatment time; further details please see (21) and GEO database). Each sample contemplates 50 antioxidant genes listed in HAG website. Microarray data are provided as normalized raw signal (MAS5.0) (20-21).
3.3. Statistical analysis
The statistics used to describe the activity and diversity of a given gene set has been previously described (13). Briefly, to obtain a quantitative measure to characterize the regulation of each GFAG in the samples we accessed the information content using Shannon's entropy (22-23) defined as follows. Consider M as the number of genes in the GFAG a (a = 1,..., n). For a given GFAG, we can define s(i,a ) as being the signal of a given gene i, (i = 1,..., Ma ), whose sum for a given a adds up to Na . The contribution p(i,a ) of the signal of gene i to the total signal of the a -GFAG is
, (1)
such that 
. The normalized entropy function Ha
, hereby called diversity, is defined as
, (2)
where we have divided all terms by the normalization factor ln(Ma ) to guarantee that 0 £ Ha £ 1. In this way we can compare different GFAGs that may present different numbers of genes. Finally, in order to normalize the quantities by groups of genes, taking as reference the diversity of the signal of the control, we define the relative diversity ha for any given GFAG as
, (3)
where 
and 
are, respectively, the diversity of experiment and control (i.e. GFAGs from treatment and control microarrays). In analogy, the relative gene expression activity na
of the a
-GFAG is defined as
, (4)
where 
and 
are, respectively, the gene expression activity of experiment and control. To determine the significantly altered GFAGs we used resampling analysis with replacement (bootstrapping) with 50.000 replications in order to estimate the sampling distribution of both ha
and na
in the microarray dataset (this bootstrap number guarantee convergence without fluctuations). Given that this analysis considers genes in the context of functional groups, the statistical design is set to compare groups of genes (i.e. GFAGs). Therefore, to understand the significance boundaries, it is essential to observe that the bootstrap distribution produces random groups with the same size of the GFAGs (i.e. the number of genes seeded in the bootstrap is the same of the GFAG tested). This procedure can detect other aspects of the transcriptome functioning, possible not detect in the typical single-gene analysis. The raw P-values from the bootstrap analysis were controlled for multiple comparisons by False Discovery Rate (FDR) analysis (24). This procedure was used to identify GFAGs exhibiting significant differential gene expression profiles with a FDR no greater than 5% (i.e. a 5% FDR indicates that among all GFAGs identified as being differentially expressed, 5% of them are truly not significant). All data analysis was carried out in the ViaComplex software (25).
4. RESULTS
4.1. Antioxidant gene network activity in hypoxic conditions
The relative gene expression activity (na ) and diversity (ha ) of the antioxidant genes in HeLa cells and human primary astrocytes are presented in Figure 2. In HeLa cells (Figure 2a, upper panels), the antioxidant gene activity is significantly altered in all three independent experiments (a, b and c). It shows that HeLa cells submitted to hypoxia have increased the gene expression activity when compared to normoxic controls (Pa=0.01504, FDRa=0.025; Pb=0.02216, FDRb=0.025; Pc=0.0062, FDRc=0.025), and this increase is concentrated in thiol redox functions (Pa=0.0056; FDRa=0.0125; Pb=0.005, FDRb=0.0125; Pc=0.00392, FDRc=0.0125). In order to test whether these results are associated with large-scale and non-specific gene expression changes we also quantified the gene expression activity of several essential pathways (Figure 2a, lower panels). According to KEGG database (19), this panel provides at least one pathway that may characterize the gene expression status of the following cell metabolic functions: i) carbohydrate metabolism, ii) energy metabolism, iii) lipid metabolism, iv) nucleotide metabolism, v) amino acid metabolism, vi) transcription, vii) translation, and viii) folding, sorting and degradation. Accordingly, neither activity nor gene expression diversity of these essential pathways are significantly altered in HeLa cells submitted to hypoxia when compared to normoxic condition (at an FDR < 5%), and in primary astrocytes only one pathway presents gene expression activity significantly increased at an FDR < 5% (n=2) when compared to the control (Figure 2b, lower panels).
4.2. Antioxidant gene network activity in low pH environment
In Figure 3 (upper panels) we present the relative gene expression activity (na ) and diversity (ha ) of the antioxidant GFAGs in SKGT4 cells maintained under low pH condition (pH=6.5). The antioxidant gene sets seem to respond to the low pH environment in a time-dependent manner, showing a progressive increase that is significantly different from the control at 240 min (FDR < 5%) (Figure 3a, upper panels). This result does not seem to be related to global gene expression changes, as indicated by the same analysis on essential pathways (Figure 3a, lower panels). In addition, the gene expression diversity in SKGT4 cells is not significantly altered (at an FDR < 5%) in any evaluated pathway (Figure 3b).
4.3. Gene expression levels of antioxidant genes
In order to compare the gene expression levels among all experiments we normalized the expression values from all microarrays and ordered all genes according to the normalized expression levels in each cell type (HeLa cells, primary astrocytes, and SKGT4 cells). The result of this ranking is depicted in Figure 4 and suggests that the gene expression pattern is similar among the three cell types, that is, genes with high expression levels in HeLa cells also present high expression values in the other two cell types, and vice versa. However, this observation might be biased due to a possible unbalanced distribution of the GFAGs' members (i.e. peroxidases, superoxide dismutases, or thiol redox). Therefore, next we grouped the antioxidant genes by both GFAG membership and gene rank. Such hierarchical subdivision is listed in Table 1 and the relative gene expression activity is depicted in Figure 5. It shows that the response pattern of antioxidant genes is in fact significant different among the three cell types: (i) primary astrocytes do not respond to hypoxic treatment (Figure 5a), (ii) HeLa cells show thiol-redox activation predominantly in two groups (Figure 5b), and (iii) SKGT4 cells present a homogeneous thiol-redox up-regulation (Figure 5c). Taken together, these results indicate that the thiol-redox gene group shows the most responsive antioxidant function when considering hypoxic condition (HeLa cells) and low pH environment (SKGT4 cells).
5. DISCUSSION
Free radicals and other reactive oxygen species are generated in several cellular sites during normal metabolism. One of the major physiological sources of ROS is the mitochondrial electron transport chain (26). The oxidative phosphorylation represents the main pathway in ATP generation in most eukaryotic cells. However, a wide number of cells can survive at low oxygen tension by using the glycolysis pathway to generate ATP. Gene expression is adjusted to oxygen availability by several mechanisms, including regulation of transcription activity by hypoxia-inducible factor α1 (HIF-α1). HIF-α1 regulates the expression of a wide number of genes, including many involved in ROS control (27). In the present work, we evaluated the relative gene expression activity of antioxidant genes in HeLa cells and primary human astrocytes maintained under hypoxic conditions. A network describing the associations among those genes has been proposed by Gelain et al. (7-8), and includes 63 components involved in codifying antioxidants products, grouped into functional categories.
According to Figure 2a, hypoxic treatment significantly increases the relative expression of the antioxidant genes in HeLa cells and this up-regulation is not followed by a general increase in transcriptional activity. Surprisingly, even glycolysis genes (which could be expected to be upregulated under hypoxic conditions) are not altered in HeLa cells. This is in contrast with primary astrocytes, which shows significant changes in glycolysis genes, but no significant alterations in the antioxidant GFAGs (Figure 2b). This result is consistent with the original paper that produced the transcriptomes from HeLa cells and astrocytes (20). According to Mense & cols (20), primary astrocytes present significant increase in the induction of genes involved in glycolysis, glucose transport, and lactate transport when compared to HeLa cells. The authors suggest that this gene expression pattern might contribute to the higher resistance of astrocytes to hypoxia, indicating that the oxygen-dependent gene expression regulation is cell type specific. It is possible that the unchanged status of the antioxidant gene groups in astrocytes is related to its ability to deal with anaerobic metabolism. In contrast, HeLa cells represent a highly proliferative tumor cell type that requires additional energy supply, obtained mainly through oxidative phosphorylation. Therefore, conditions of low O2 tension might cause severe cellular stress in this cell type when compared to human astrocytes, which might be better adapted to hypoxic conditions. In some extent, it is possible that such cellular stress may modulate the thiol-redox apparatus expression, as showed in Figure 2a.
Cells have mechanisms that regulate the intracellular pH (pHi), such as the Na+/H+ antiport. However, an intense glycolytic activity may lead to acidosis if an excessive H+ production overcomes the cellular pHi regulating systems. Thus, a common hypoxic consequence is cellular acidosis, which may trigger a selective gene expression response (21). To determine the relative expression of the antioxidant genes submitted to acidosis, we accessed microarray data from esophageal adenocarcinoma SKGT4 cells maintained under low pH conditions. As shown in Figure 3a, gene expression levels of thiol-redox members are increased in these cells. It shows that the antioxidant gene network (mainly thiol-redox genes) in SKGT4 cells is responsive to low pH environment, suggesting that cellular acidosis can be one of the conditions able to up regulate the expression of thiol-redox genes. Additionally, the gene expression diversity in SKGT4 cells is not altered in any experimental condition, indicating that the observed gene expression changes are due to a coordinated response (not caused by isolated gene expression peaks), which reinforces the idea that antioxidant genes work as a system.
The contrast among the relative gene expression patterns from HeLa, astrocytes and SKGT4 cells rises two interesting observations: (i) the most expressed genes are consistently overlapped among the three cell types and (ii) the functional subdivision of the gene rank shows that the thiol-redox response to hypoxia and acidosis does not seems to be merely related to the gene expression levels. As a summary of this last observation, in Figure 6 we provide a network representation of the antioxidant response to low pH treatment in the SKTG4 cell line, which showed the most consistent thiol-redox response to acidosis. In this sense, some key points might be useful to contemplate these results. Firstly, there are a number of thiol-redox genes codifying peptidic but non-enzymatic antioxidant products, such as metallothioneins. Contrasting to SODs and peroxidases, thiol-redox components are not controlled by an equivalent post-translational regulatory process (such as enzymatic activity). In this condition transcriptional regulation may acquire an essential role for thiol-redox functioning. Secondly, the interconnectivity between multiple thiol/disulfide couples on redox environments are maintained at stable but non-equilibrium steady states in which the abundance of antioxidant-active components (i.e. thioredoxins and glutaredoxins) can modulate the redox network function (28). Removing a thiol-redox component will necessarily perturb the overall thiol-redox homeostasis of the cell, which is a typical trait of biological networks (29). As the thiol-redox network presents more components and is more interconnected than the other antioxidant functions (7-8), it is likely that such genes are collectively more responsive to redox environments. In addition, thiol-containing molecules represent the most important redox buffers within cells. Taking into account that alterations in pH status affect the cellular redox state (30), the increased expression of thiol-redox genes described here is consistent with a physiological response to maintain the redox state homeostasis in cells submitted to acidosis.
6. CONCLUSIONS
In this work we provide evidences that the human antioxidant gene network described by Gelain et al (8) can respond selectively to oxidative stress stimulus. We anticipate that this network-based model will be particularly useful not only to describe stress-responsive expression profiles but also to detect other aspects of transcriptome functioning possibly not detected in the typical single-gene analysis.
7. ACKNOWLEDGMENTS
We acknowledge the Brazilians funds CNPq, CAPES, and PROPESQ/UFRGS.
8. REFERENCES
1. H. Liu, R. Colavitti, I. I. Rovira, T. Finkel: Redox-Dependent Transcriptional Regulation. Circ Res, 97(10):967-974 (2005).
http://dx.doi.org/10.1161/01.RES.0000188210.72062.10
2. S.G. Rhee: Cell Signaling: H2O2, a necessary evil for cell signaling. Science, 312(5782):1882-1883 (2006).
http://dx.doi.org/10.1126/science.1130481
3. E. A. Veal, A. M. Day, B. A. Morgan: Hydrogen peroxide sensing and signaling. Mol Cell, 26(1):1-14 (2007)
http://dx.doi.org/10.1016/j.molcel.2007.03.016
4. A. L. Jackson, L. A. Loeb: The contribution of endogenous sources of DNA damage to the multiple mutations in cancer. Mutat Res-Fund Mol M, 477(1-2):7-21 (2001)
http://dx.doi.org/10.1016/S0027-5107(01)00091-4
5. F. Dal Pizzol, F. Klamt, M. S. Benfato, E. A. Bernard, J. C. F. Moreira: Retinol supplementation induces oxidative stress and modulates antioxidant enzyme activities in rat Sertoli cells. Free Radical Res, 34(4):395-404 (2001)
http://dx.doi.org/10.1080/10715760100300331
6. M. R. de Oliveira, R. B. Silvestrin, T. M. Souza, J. C. M. Moreira: Therapeutic vitamin A doses increase the levels of markers of oxidative insult in Substantia nigra and decrease locomotory and exploratory activity in rats after acute and chronic supplementation. Neurochem Res, 33(3):378-383 (2008)
http://dx.doi.org/10.1007/s11064-007-9438-2
7. R. J. S. Dalmolin, D. P. Gelain, M. A. A. Castro: Human Antioxidant Genes. http://www.ufrgs.br/icbs/hag/index.htm 2008.
(doi not found)
8. D. P. Gelain, R. J. S. Dalmolin, V. L. Belau, J. C. F. Moreira, F. Klamt, M. A. A. Castro: A systematic review of human antioxidant genes. Front Biosci, 144:457-463 (2009)
http://dx.doi.org/10.2741/3541
9. D. Bonatto: A systems biology analysis of protein-protein interactions between yeast superoxide dismutases and DNA repair pathways. Free Radical Bio Med, 43(4):557-567(2007)
http://dx.doi.org/10.1016/j.freeradbiomed.2007.05.013
10. S. Casagrande, V. Bonetto, M. Fratelli, E. Gianazza, I. Eberini, T. Massignan, M. Salmona, G. Chang, A. Holmgren, P. Ghezzi: Glutathionylation of human thioredoxin: A possible crosstalk between the glutathione and thioredoxin systems. P Natl Acad Sci Usa, 99(15):9745-9749 (2002)
http://dx.doi.org/10.1073/pnas.152168599
11. Finkel T. Oxidant signals and oxidative stress. Curr Opin Cell Biol, 15(2):247-254 (2003)
http://dx.doi.org/10.1016/S0955-0674(03)00002-4
12. J. Haendeler, J. Hoffmann, V. Tischler, B. C. Berk, A. M. Zeiher, S. Dimmeler: Redox regulatory and anti-apoptotic functions of thioredoxin depend on S-nitrosylation at cysteine 69. Nat Cell Biol, 4(10):743-749 (2002)
http://dx.doi.org/10.1038/ncb851
13. M. A. A. Castro, J. C. M. Mombach, R. M. C. de Almeida, J. C. F. Moreira: Impaired expression of NER gene network in sporadic solid tumors. Nucleic Acids Res, 35(6):1859-1867 (2007)
http://dx.doi.org/10.1093/nar/gkm061
14. M. M. Xiong, J. Y. Zhao, H. Xiong: Network-based regulatory pathways analysis. Bioinformatics, 20(13):2056-2066 (2004)
http://dx.doi.org/10.1093/bioinformatics/bth201
15. H. V. Westerhoff, B. O. Palsson: The evolution of molecular biology into systems biology. Nat Biotech, 22(10):1249-1252 2004
http://dx.doi.org/10.1038/nbt1020
16. I. Borodina, J. Nielsen: From genomes to in silico cells via metabolic networks. Curr Opin Biotech, 16(3):350-355 (2005)
http://dx.doi.org/10.1016/j.copbio.2005.04.008
17. H. M. Wain, M. J. Lush, F. Ducluzeau, V. K. Khodiyar, S. Povey: Genew: the Human Gene Nomenclature Database, 2004 updates. Nucleic Acids Res, 32:D255-D257 (2004)
http://dx.doi.org/10.1093/nar/gkh072
18. S. D, Hooper, P. Bork: Medusa: a simple tool for interaction graph analysis. Bioinformatics, 21(24):4432-4433 (2005)
http://dx.doi.org/10.1093/bioinformatics/bti696
19. M. Kanehisa, S. Goto, M. Hattori, K. F. Aoki-Kinoshita, M. Itoh, S. Kawashima, T. Katayama, M. Araki, M. Hirakawa: From genomics to chemical genomics: new developments in KEGG. Nucleic Acids Res, 34:D354-D357 (2006)
http://dx.doi.org/10.1093/nar/gkj102
20. S. M. Mense, A. Sengupta, M. Zhou, C. Lan, G. Bentsman, D. J. Volsky, L. Zhang: Gene expression profiling reveals the profound upregulation of hypoxia-responsive genes in primary human astrocytes. Physiol Genomics, 25(3):435-449 (2006)
http://dx.doi.org/10.1152/physiolgenomics.00315.2005
21. S. P. Duggan, W. M. Gallagher, E. J. P. Fox, M. M. Abdel-Latif, J. V. Reynolds, D. Kelleher: Low pH induces co-ordinate regulation of gene expression in oesophageal cells. Carcinogenesis, 27(2):319-327 (2006)
http://dx.doi.org/10.1093/carcin/bgi211
22. M. A. A. Castro, T. T. G. Onsten, R. M. C. de Almeida, J. C. F. Moreira: Profiling cytogenetic diversity with entropy-based karyotypic analysis. J Theor Biol, 234(4):487-495 (2005)
http://dx.doi.org/10.1016/j.jtbi.2004.12.006
23. M. A. A. Castro, T. G. H. Onsten, J. C. F. Moreira, R. M. C. de Almeida: Chromosome aberrations in solid tumors have a stochastic nature. Mutat Res-Fund Mol M, 600(1-2):150-164 (2006)
http://dx.doi.org/10.1016/j.mrfmmm.2006.04.003
24. Y. Benjamini, Y. Hochberg: Controlling the false discovery rate - A practical and powerful approach to multiple testing. J Roy Stat Soc B Met, 57(1):289-300 (1995)
(doi not found)
25. M. A. A. Castro, J. L. Rybarczyk-Filho, R. J. S. Dalmolin, M. Sinigaglia, J. C. F. Moreira, J. C. M. Mombach, R. M. C. de Almeida: ViaComplex: software for landscape analysis of gene expression networks in genomic context. Bioinformatics, 25(11):1468-1469 (2009)
http://dx.doi.org/10.1093/bioinformatics/btp246
26. I. Ohsawa, M. Ishikawa, K. Takahashi, M. Watanabe, K. Nishimaki, K.Yamagata, K. Katsura, Y. Katayama, S. Asoh, S. Ohta: Hydrogen acts as a therapeutic antioxidant by selectively reducing cytotoxic oxygen radicals. Nat Med; 13(6):688-694 (2007)
http://dx.doi.org/10.1038/nm1577
27. F. J. Giordano: Oxygen, oxidative stress, hypoxia, and heart failure. J Clin Invest, 115(3):500-508 (2005)
http://dx.doi.org/10.1172/JCI24408
28. M. Kemp, Y. M. Go, D. P. Jones: Nonequilibrium thermodynamics of thiol/disulfide redox systems: A perspective on redox systems biology. Free Radical Bio Med, 44(6):921-937 (2008)
http://dx.doi.org/10.1016/j.freeradbiomed.2007.11.008
29. M.B. Toledano, C. Kumar, N. Le Moan, D. Spector, F. Tacnet: The system biology of thiol redox system in Escherichia coli and yeast: Differential functions in oxidative stress, iron metabolism and DNA synthesis. FEBS Lett, 581(19):3598-607 (2007).
http://dx.doi.org/10.1016/j.febslet.2007.07.002
30. F. Q. Schafer, G. R. Buettner: Redox environment of the cell as viewed through the redox state of the glutathione disulfide/glutathione couple. Free Radical Bio Med, 30(11):1191-1212 (2001)
http://dx.doi.org/10.1016/S0891-5849(01)00480-4
Key Words: Human Antioxidant Genes, HAG Database, Gene Network, System Biology, Reactive Oxygen Species, Free radicals.
Send correspondence to: Rodrigo Juliani Siqueira Dalmolin, Universidade Federal do Rio Grande do Sul, Rua Ramiro Barcelos 2600-anexo, Porto Alegre, RS, Brazil, Tel: 55-51-33085578, Fax:55-51-33085540, E-mail:rodrigo.dalmolin@ufrgs.br