While many of the phenotypic differences between human and chimpanzee may result from changes in gene regulation, only a handful of functionally important regulatory differences are currently known. As a first step towards identifying transcriptional pathways that have been remodeled in the human lineage, we focused on a transcription factor, FOXO1a, which we had previously found to be up-regulated in the human liver compared to that of three other primate species. We concentrated on this gene because of its known role in the regulation of metabolism and in longevity.
Using a combination of expression profiling following siRNA knockdown and chromatin immunoprecipitation in a human liver cell line, we identified eight novel direct transcriptional targets of FOXO1a. This set includes the gene for thioredoxin-interacting protein (TXNIP), the expression of which is directly repressed by FOXO1a. The thioredoxin-interacting protein is known to inhibit the reducing activity of thioredoxin (TRX), thereby hindering the cellular response to oxidative stress and affecting life span.
Our results provide an explanation for the repeated observations that differences in the regulation of FOXO transcription factors affect longevity. Moreover, we found that TXNIP is down-regulated in human compared to chimpanzee, consistent with the up-regulation of its direct repressor FOXO1a in humans, and with differences in longevity between the two species.
Citation: de Candia P, Blekhman R, Chabot AE, Oshlack A, Gilad Y (2008) A Combination of Genomic Approaches Reveals the Role of FOXO1a in Regulating an Oxidative Stress Response Pathway. PLoS ONE 3(2): e1670. https://doi.org/10.1371/journal.pone.0001670
Academic Editor: Justin O. Borevitz, University of Chicago, United States of America
Received: January 16, 2008; Accepted: January 30, 2008; Published: February 27, 2008
Copyright: © 2008 de Candia et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was supported by NIH grant GM077959.
Competing interests: The authors have declared that no competing interests exist.
In addition to substitutions at the protein level, changes in gene regulation are likely to underlie many phenotypes of interest, including human-specific adaptations and diseases –. But while many human-specific adaptations in gene copy number and protein sequence have been documented, only a few differences in gene regulation between humans and other apes are known –.
In order to identify human-specific changes in regulatory pathways, we focused on a transcription factor, the Forkhead box O1A transcription factor (FOXO1a), which we had previously found to be significantly up-regulated in human livers compared to that of three non-human primates . We concentrated on this gene because of its pivotal role in the regulation of metabolism and in longevity (reviewed by ), a phenotype that differs markedly between humans and other primates .
The FOXO transcription factors are key targets of the insulin/IGF signaling pathway (reviewed by ). Humans and mice have four functional FOXO genes (−1, 3, 4, and 6), while flies (dFOXO) and worms (DAF-16) have one . Changes in the regulation of FOXO transcription factors affect the median and maximum life span in C. elegans ,  and D. melanogaster  and, in rodents, the inhibition of the insuling/IGF-1 signaling pathway in mice ,  and rats  results in increased longevity. It has further been shown that inhibition of FOXO transcription factors in worms, flies, and mammalian cellular systems results in differences in expression of a large number of genes, and in particular, leads to decreased expression of enzymes that protect against or repair oxidative damage and, as a result, to higher sensitivity to oxidative stress –. Since oxidative stress is thought to be an important determinant of the rate of aging (reviewed by ), at least one mechanism by which changes in the regulation of FOXO affect life span may be through the regulation of genes involved in protection from reactive oxygen species (ROS) , , .
These functions of FOXO in the insulin signaling pathway and the response to ROS, and its role in promoting longevity, appear to be evolutionarily conserved: When the expression level of FOXO is perturbed, the corresponding changes in gene expression patterns as well as the resulting phenotypes are similar across distantly related species (reviewed by ). However, while dosage manipulations of FOXO result in expression level changes at a large number of genes, to date, only a few have been shown to be directly regulated by FOXO transcription factors , . In particular, although FOXO has been shown to regulate the expression of several genes involved in ROS detoxification , , the direct transcriptional targets through which FOXO mediates the cellular response to oxidative stress and life span remained elusive.
Identifying the direct transcriptional targets of FOXO1a
As a first step of our analysis of FOXO1a regulatory pathways, we validated the original microarray observation of FOXO1a mRNA expression differences between humans and other primates by using quantitative RT-PCR on human and chimpanzee liver RNA samples (Figure S1). We also confirmed that the expression of FOXO1a at the protein level is elevated in the human liver compared to that of chimpanzee (Figure S1). Available genomic sequences (http://genome.ucsc.edu/) indicate that the human and chimpanzee FOXO1a proteins only differ at one residue (at position 62), which is not within the forkhead box DNA binding domain or any known protein-protein interaction domain, and is not known to be a target of any regulatory post-translational modification. This observation suggests that the human and chimpanzee FOXO1a orthologs have similar biochemical properties - including DNA binding - and that their regulation at the protein level (e.g., their localization) may be similar. Thus, the observed difference in FOXO1a gene expression levels between human and chimpanzee likely results in differences in the regulation of FOXO1a transcriptional targets between the two species .
To identify direct FOXO1a transcriptional targets in the human liver, we used a combination of approaches. First, we examined changes in gene expression levels following a knockdown of FOXO1a in human liver cell lines by using siRNA transfection (see Materials and Methods). The knockdown of FOXO1a resulted in a significant (FDR<0.05) change in the expression of 490 genes (Figure 1 and Table S1). Only a subset of the 490 differentially expressed genes are likely to be direct targets of FOXO1a, since many gene expression changes likely result from regulatory network perturbations (e.g., the genes may be regulated by the direct targets of FOXO1a, or by genes that are farther downstream in the cascade. In addition, the knockdown of a transcription factor may affect the cellular environment in ways that may trigger larger changes in the gene expression profiles, not directly related to the regulatory effects of the perturbed transcription factor).
A. FOXO1a Western blots are shown for one of the three siRNA biological replicates, indicating that the level of the FOXO1a protein is dramatically reduced. B. Zoom into a picture of a cDNA microarray co-hybridization of RNA from one biological replicate of cells treated with FOXO1a siRNA (Cy3 - green) and RNA from untreated cells (Cy5 - red). The circle marks the cDNA probe for FOXO1a. As can be seen, FOXO1a mRNA levels are reduced following the knockdown. We note that this microarray result was validated by using quantitative RT-PCR. C. A volcano plot for results of the comparison of gene expression profiles following FOXO1a knockdown to the control siRNA treatment. The eight confirmed direct transcriptional target of FOXO1a are indicated by arrows. In the plot, all P-values smaller than 10−9 are plotted as P = 10−10 (P-values ranged from 1 to 10−43).
To hone in on the subset of direct targets, we then searched for the known binding motif of FOXO1a in the putative promoters of the 490 differentially expressed genes. Our analysis was limited to the ∼1 kb segments upstream of known transcription start sites (see Materials and Methods for details), and hence was far from exhaustive. Nonetheless, 21 genes whose expression levels were significantly elevated or reduced by the knockdown were found to contain FOXO1a binding motif in their promoters. These 21 genes are likely direct transcriptional targets of FOXO1a (Table 1).
One concern is that computational searches for transcription factor binding sites are known to have a high rate of false positives . We therefore validated the in silico analysis using Chromatin ImmunoPrecipitation (ChIP) with a FOXO1a antibody, followed by PCR amplification of the 21 promoter regions predicted to contain a FOXO1a binding site (see Materials and Methods). The promoters for eight (38%) of the 21 genes were found to be enriched in PCR amplifications following the ChIP with FOXO1a antibody, compared to the control experiment (Figure 2). In summary, by intersecting the results of expression profiling following a FOXO1a knockdown, computational analyses and PCR amplification of ChIP enriched promoter regions, we identified eight novel direct transcriptional targets of FOXO1a.
Gel electrophoresis pictures of PCR amplifications following FOXO1a ChIP, for the putative promoters of the eight direct transcriptional targets of FOXO1a.
Since FOXO1a expression levels are elevated in humans compared to chimpanzees, we hypothesized that a subset of the eight direct transcriptional targets of FOXO1a would be differentially expressed between the species. Specifically, based on the expression profiling following FOXO1a knockdown (Table S1 and Figure 2), we predicted that the expression levels of the genes ABCB1, EiFG4, RARS, and TAF11 would be elevated in humans while the genes B4GALT3, PRDX4, PRG4, and TXNIP would show reduced expression levels in humans compared to chimpanzees. To test this hypothesis, we measured the expression levels of these eight genes in RNA samples from the livers of six human and six chimpanzee individuals, using quantitative RT-PCRs (see Materials and Methods). As can be seen in figure 3, four of the eight genes were found not to be significantly differentially expressed between humans and chimpanzees (at the 5% level), while for one gene, PRDX4, the difference in expression between the species was not consistent with our prediction. A likely explanation is that compensatory mutations in humans offset the effect of elevated FOXO1a levels on the expression of these five genes, as have been observed previously in fruit flies ,  and inferred from a comparison of human and mouse regulatory sequences . We conclude that changes in FOXO1a expression levels cannot explain the observed inter-species gene expression profiles for these five genes.
Mean fold differences (y-axis) and standard errors for six biological replicates (different individuals) are given for either the human (dark bars) or chimpanzee (clear bars) liver RNA samples. For each gene (x-axis), results were standardized based on the species with the lower expression level (set to 1). Stars indicate gene that are differentially expressed between human and chimpanzee at either P<0.1 (*) or P<0.05 (**) (see text).
In contrast, our predictions were met for three genes: We found a significant inter-species difference in gene expression for the genes TAF11 (one-tailed P<0.001) and TXNIP (P = 0.02), and a marginally significant, consistent difference (P = 0.09) for RARS (Figure 3). For these three genes, it is likely that elevated level of FOXO1a gene expression in humans compared to chimpanzees resulted in an inter-species difference in transcript levels.
FOXO1a regulates the oxidative stress response pathways
Of the three genes, relatively little is known about the function of TAF11 (TATA binding protein-associated factor 11) and RARS (arginyl-tRNA synthetase). In contrast, TXNIP (Thioredoxin interacting protein, also termed vitamin D3 up-regulated protein 1 – VDUP1, and thioredoxin-binding protein 2 – TBP2) has been studied extensively. In particular, TXNIP has been shown to inhibit the reducing activity of thioredoxin (TRX) through direct protein-protein interaction –. Because TRX plays a critical role in regulating the cellular response to oxidative stress –, the presence of high levels of its inhibitor, TXNIP, increases the vulnerability of the cell to ROS . Thus, our results point to a direct link between changes in the regulation of FOXO1a and the cellular response to oxidative stress.
The role of TRX in the ROS detoxification pathway is well understood –, and the protein-protein interaction between TXNIP and TRX has been clearly demonstrated –. We wanted to provide similarly strong evidence that binding of FOXO1a to the promoter of TXNIP indeed affects TXNIP expression level in humans. To do so, we used site directed mutagenesis to mutate the FOXO1a binding site in the promoter of TXNIP. We then examined the difference in TXNIP promoter activity with and without the binding site for FOXO1a, by using reporter gene assays in human liver cell lines (see Material and Methods). As can be seen in figure 4, TXNIP promoter activity is significantly (P<10−5) elevated when the binding site for FOXO1a is mutated, consistent with our observation that FOXO1a is a direct repressor of TXNIP.
By using a combination of genomic approaches, we found that TXNIP is a direct regulatory target of FOXO1a. The effect of changes in the regulation of TXNIP and TRX on the response to oxidative stress and life span was found to be conserved across worms and mice , , and more speculatively, in flies  and pigs . Although increased response to oxidative stress may not be the only mechanism through which changes in FOXO regulation affect life span , our findings provide an explanation for the repeated observations that elevated levels of FOXO transcription factors enhance the response to ROS and increase longevity in a number of model organisms. As a direct transcriptional repressor of TXNIP, elevated expression levels of FOXO result in lower levels of TXNIP, which in turn results in increased TRX-reducing activity , , improved cellular response to oxidative stress , and ultimately increased life span , .
Our observations that the FOXO1a expression level is elevated in human livers compared to chimpanzee livers and consistently, that TXNIP expression levels are lower, provide one of very few well documented examples of differences in regulatory pathways between the species and raise an intriguing hypothesis, namely that the cellular response to ROS is increased in humans compared to chimpanzee. While currently we cannot exclude the possibility that compensatory changes offset the effect of FOXO1a and TXNIP, we note that the binding site for FOXO1a in the promoter of TXNIP is highly conserved across species, and in particular, is identical in humans and chimpanzees (Figure S2). Thus, our findings are consistent with the hypothesis that increased resistance to ROS contributes to greater life span in humans , and is particularly intriguing given the well-documented difference in life expectancy and maximum life span between humans and chimpanzees , . Functional studies of oxidative stress response are needed in order to directly test this hypothesis.
Materials and Methods
We performed quantitative RT-PCR in order to: (i) Validate the original microarray observation of differences in FOXO1a gene expression between humans and chimpanzees, (ii) confirm the FOXO1a knockdown in HepG2/C3A liver cells (see below), and (iii) test for inter-species differences expression of the eight FOXO1a direct transcriptional targets (see below). Total RNA was extracted from liver cell lines using the RNA Mini kit (Qiagen, Valencia, CA), and from human and chimpanzee liver tissue samples using Trizol (Invitrogen, Carlsbad, CA). In all cases, we synthesized first-strand cDNA using a poly-T oligonucleotide and the Superscript enzyme (Invitrogen, Carlsbad, CA). The first strand cDNA was then used as template for quantitative RT-PCR with the JumpStart Taq ReadyMix kit (Sigma-Aldrich, St. Louis, MO). For all reactions, PCR primers and probes were designed in sequences that are identical between human and chimpanzee (based on their available genomic sequence (http://genome.ucsc.edu/)). In each reaction, the final concentrations of the primers and the probe were 200 nM and 100 nM, respectively. The cycling conditions were as follows: initial denaturation at 94°C for 2 min, following by 40 cycles of denaturation at 94°C for 15 sec and annealing/extension at 60°C for 1 min. β−Actin was used as control for gene expression analyses. Inter-species differences in gene expression were evaluated using a t-test.
We performed Western blots in order to: (i) Confirm that FOXO1a protein expression level is elevated in human compared to chimpanzees, and (ii) confirm the FOXO1a knockdown in HepG2/C3A liver cells (see below). In both cases (cell lines or tissue samples), proteins were extracted in RIPA buffer (Tris-HCl: 50 mM, pH 7.4 NP-40: 1%; Na-deoxycholate: 0.25%; NaCl: 150 mM; EDTA: 1 mM) and proteases inhibitors (PMSF: 1 mM, Aprotinin, leupeptin, pepstatin: 1 µg/ml each). The protein extracts subjected to electrophoresis using a MiniGel apparatus and then transferred onto the Immuno-Blot PVDF Membrane (Biorad Laboratories, Hercules, CA). The antibody against FOXO1a was purchased from Cell Signaling Technology (Danvers, MA) and visualized with the ECL plus Western Blotting Detection System (Amersham Biosciences, Piscataway, NJ).
siRNA knockdown and microarray hybridizations
In order to knockdown FOXO1a in HepG2/C3A human liver cells, we transfected the cells with two different siRNAs (Ambion, Austin, TX), which target different region of the gene. As a control, we transected the cells with Ambion Silencer® Negative Control siRNA. Each transfection was performed in three biological replicates. Total RNA was extracted from each biological replicate, as well as from untreated cells, using the RNA Mini kit (Qiagen, Valencia, CA). First strand cDNA was synthesized using a T7-poly-T oligo and the superscript enzyme (Invitrogen, Carisbad, CA). Second strand cDNA was synthesized using DNA Pol I enzyme (Invitrogen, Carisbad, CA). The double strand cDNA was subjected to linear amplification using MEGAscript (Ambion, Austin, TX), and RNA was purified using the RNeasy kit (Qiagen, Valencia, CA). For each microarray hybridization, 4 µg of amplified RNA were used for amino-allele labeling (BD Bioscience, Palo Alto, CA) with either Cy3 (for the specific or control siRNA treatment) or Cy5 (for the untreated cells) dyes (Amersham, Buckinghamshire, UK). Labeled samples were co-hybridized to the multi-species cDNA array described previously ,  according to a reference design where the RNA from untreated cells serve as the reference and using two technical replicates for each biological sample (for a total of 12 hybridizations). Hybridization and washes were carried out as described in reference (1).
Analysis of microarray hybridizations
The 12 cDNA arrays were scanned using a GenePix Axon scanner and data were extracted with GenePix 6 (Molecular Devices, Sunnyvale CA), resulting in Cy5 and Cy3 foreground and background intensities (using the morph background estimation procedure). Subsequent analysis was performed using the R computing environment (http://www.r-project.org). Background corrected Cy5 and Cy3 intensities were produced using the ‘normexp’ method with an offset of 50, implemented in the limma software package , and within-array lowess normalization was performed using all probes.
The microarray that we used includes orthologous probes from humans, and three other closely related primates . We have previously shown that differential expression between samples from the same species can be estimated using probes from a closely related species . Hence, we were able to combine data from all probes on the array (i.e., including those for non-human species). The expression log-ratios of the Cy5 to Cy3 intensity (M) for each gene were analyzed using the following linear mixed model where we have suppressed the individual gene labels:(0.1)
Here μt is the fixed effect for the treatment t (either FOXO1a or control siRNA treatment) and the term πp is the fixed effect for the probe where p = h,c,o or r (for human, chimpanzee, orangutan or rhesus macaque) is the probe species. αtri is a random effect for technical replicate i within each biological replicate r, which is assumed to be uncorrelated with mean zero and variance σ2α. εtrip is the residual error term with variance σ2, assumed to be uncorrelated with mean zero. The random effect for technical replicates was handled by pooling the variances across replicates using the method of reference . Tests of significance were conducted using empirical Bayes moderated t-tests which ensure stable inference even with small sample size . Differentially expressed genes were identified at a false discovery rate  of 5%.
Identifying FOXO1a binding sites
Our goal was to identify FOXO1a binding sites in the promoters of the 490 genes whose expression levels were significantly different following FOXO1a knockdown (see above). To do so, we first used the database of transcription start sites (DBTSS, http://dbtss.hgc.jp/) to identify an empirically validated transcription start site (TSS) for 287 of the 490 differentially expressed genes. We then defined putative promoters as the sequences ranging from 1000 bp upstream to 200 bp downstream of the TSS. To search for the signature of the FOXO1a binding site in these 287 putative promoters, we used MATCH , together with FOXO1a positional weight matrices from the TRANSFAC database (http://www.gene-regulation.com/). We identified FOXO1a binding sites in the putative promoters of 21 genes (table 1). For three of these genes (SEPP1, B4GALT, and CREG) we found two putative FOXO1a binding sites in the promoter. In subsequent analysis of these promoters (see below), we only tested one site for each promoter - the one with highest similarity to the FOXO1a consensus binding site (assessed by p-values output by MATCH ).
In order to validate the computational prediction of FOXO1a binding sites, we used Chromatin Immunoprecipitation (ChIP), following the EZ ChIP protocol (Upstate Millipore, Billerica, MA). Human liver HepG2/C3A cells were crosslinked in 1% formaldehyde, then lysed at a concentration of 107 cells/ml in 1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1. Subsequently, DNA was sheared by sonication in a Bioruptor (Diagenode) to a range of 300–1000 bp. We used 106 cell equivalents of lysate for one immunoprecipitation and incubated over night at 4°C with 2 µg of either the antibody against FOXO1a or with the rabbit IgG as negative control (both from Santa Cruz Biotechnology, Santa Cruz, CA). After precipitation, the chromatin was first de-crosslinked and then purified by using the PCR product purification kit (Qiagen, Valencia, CA). Enrichment of specific promoter regions was evaluated by PCR amplification using 1/50 of the immunoprecipitated chromatin as template, with the GoTaq Flexi DNA polymerase (Promega, Madison, WI), for 35 cycles in a DNA Engine Peltier Thermal Cycler (Biorad Laboratories).
Reporter gene assays
We designed PCR primers to amplify a product from ∼100 bp downstream of the putative TSS of TXNIP to ∼900 bp upstream of it. We ligated the PCR products into the Luciferase reporter gene vector pGL4.14 (Promega), and cloned them in JM109 competent cells. We then used the Quikchange II site-directed mutagenesis kit (Stratagene) to introduce individual nucleotide changes to the promoter, which removed the binding site for FOXO1a while maintaining the exact length of the construct. Specifically, we mutated the FOXO1a binding site ‘AAACA’ into ‘TAAGA’ – a sequence that is not known to be an exact motif of any transcription factor based on the TRANSFC database, which currently (August 2007) includes 443 human binding motifs (the binding motifs of transcription factors ZF5, CTF, and NF1 are similar to the mutated sequence, but none of these transcription factors is expressed in the liver based on the Novartis gene expression atlas (http://expression.gnf.org/cgi-bin/index.cgi)).
We used touch-down PCR to amplify and then sequence (using an ABI3730 automated sequencer) the insert from individual colonies in order to confirm that no Taq-generated errors were incorporated in either the original or mutated promoters. Once the sequence of the insert from individual colonies was confirmed, we proceeded by extracting the plasmid and using it in transfections of human liver HEP cells by using Lipofectamine 2000 (Invitrogen) with 200 ng of each plasmid. The HEP cells were also transfected with 20 ng of the Renilla vector pGL4.73 (Promega). The co-transfection allows us to normalize across experiments for transfection efficiency. Luciferase and Renilla activity were measured 24 hours after transfection, using Dual-glo Luciferase kit (Promega) in a Veritas 96-well plate luminometer (Turner Biosystems).
Reporter gene study design and analysis
The Luciferase activity of each construct was measured using five replicates (independent transfections). In addition, we measured Luciferase activity for an empty (i.e., with no promoter) pGL4.14 vector, in five replicates, in order to estimate background Luciferase transcription levels. For each replicate, we normalized Luciferase by Renilla luminescence values in order to control for transfection efficiency. We then standardized the normalized luminescence values by the background activity (of the empty vector). We used a t-test to test for difference in activity between the original and the mutated promoter. The choice of a t-test is appropriate as we can not reject the hypothesis that the data is normally distributed (using Shapiro-Wilk test for normality). Unfortunately, since chimpanzee liver cell lines are not available, we could not perform the reciprocal experiment.
Many thanks to Kay Macleod, Coleen Murphy, George Perry, Molly Przeworski, Gordon Smyth, Marc Tatar and Terry Unterman for comments on the manuscript and/or helpful discussions.
Conceived and designed the experiments: YG Pd. Performed the experiments: AC Pd. Analyzed the data: RB AC AO Pd. Wrote the paper: YG RB AO Pd.
- 1. Britten RJ, Davidson EH (1971) Repetitive and non-repetitive DNA sequences and a speculation on the origins of evolutionary novelty. Q Rev Biol 46: 111–138.
- 2. King MC, Wilson AC (1975) Evolution at two levels in humans and chimpanzees. Science 188: 107–116.
- 3. Carroll SB (2003) Genetics and the making of Homo sapiens. Nature 422: 849–857.
- 4. Jin W, Riley RM, Wolfinger RD, White KP, Passador-Gurgel G, et al. (2001) The contributions of sex, genotype and age to transcriptional variance in Drosophila melanogaster. Nat Genet 29: 389–395.
- 5. Abzhanov A, Protas M, Grant BR, Grant PR, Tabin CJ (2004) Bmp4 and morphological variation of beaks in Darwin's finches. Science 305: 1462–1465.
- 6. Shapiro MD, Marks ME, Peichel CL, Blackman BK, Nereng KS, et al. (2004) Genetic and developmental basis of evolutionary pelvic reduction in threespine sticklebacks. Nature 428: 717–723.
- 7. Iftikhar R, Kladney RD, Havlioglu N, Schmitt-Graff A, Gusmirovic I, et al. (2004) Disease- and cell-specific expression of GP73 in human liver disease. Am J Gastroenterol 99: 1087–1095.
- 8. Taron M, Rosell R, Felip E, Mendez P, Souglakos J, et al. (2004) BRCA1 mRNA expression levels as an indicator of chemoresistance in lung cancer. Hum Mol Genet 13: 2443–2449.
- 9. Rockman MV, Hahn MW, Soranzo N, Zimprich F, Goldstein DB, et al. (2005) Ancient and recent positive selection transformed opioid cis-regulation in humans. PLoS Biol 3: e387.
- 10. Huby T, Dachet C, Lawn RM, Wickings J, Chapman MJ, et al. (2001) Functional analysis of the chimpanzee and human apo(a) promoter sequences: identification of sequence variations responsible for elevated transcriptional activity in chimpanzee. J Biol Chem 276: 22209–22214.
- 11. Rockman MV, Hahn MW, Soranzo N, Goldstein DB, Wray GA (2003) Positive selection on a human-specific transcription factor binding site regulating IL4 expression. Curr Biol 13: 2118–2123.
- 12. Gilad Y, Oshlack A, Smyth GK, Speed TP, White KP (2006) Expression profiling in primates reveals a rapid evolution of human transcription factors. Nature 440: 242–245.
- 13. Tran H, Brunet A, Griffith EC, Greenberg ME (2003) The many forks in FOXO's road. Sci STKE 2003: RE5.
- 14. Harvey PH, Clutton-Brock TH (1985) life history variation in primates. Evolution 39: 559–581.
- 15. Barthel A, Schmoll D, Unterman TG (2005) FoxO proteins in insulin action and metabolism. Trends Endocrinol Metab 16: 183–189.
- 16. Lin K, Dorman JB, Rodan A, Kenyon C (1997) daf-16: An HNF-3/forkhead family member that can function to double the life-span of Caenorhabditis elegans. Science 278: 1319–1322.
- 17. Ogg S, Paradis S, Gottlieb S, Patterson GI, Lee L, et al. (1997) The Fork head transcription factor DAF-16 transduces insulin-like metabolic and longevity signals in C. elegans. Nature 389: 994–999.
- 18. Giannakou ME, Goss M, Junger MA, Hafen E, Leevers SJ, et al. (2004) Long-lived Drosophila with overexpressed dFOXO in adult fat body. Science 305: 361.
- 19. Bluher M, Kahn BB, Kahn CR (2003) Extended longevity in mice lacking the insulin receptor in adipose tissue. Science 299: 572–574.
- 20. Holzenberger M, Dupont J, Ducos B, Leneuve P, Geloen A, et al. (2003) IGF-1 receptor regulates lifespan and resistance to oxidative stress in mice. Nature 421: 182–187.
- 21. Shimokawa I, Higami Y, Utsuyama M, Tuchiya T, Komatsu T, et al. (2002) Life span extension by reduction in growth hormone-insulin-like growth factor-1 axis in a transgenic rat model. Am J Pathol 160: 2259–2265.
- 22. Honda Y, Honda S (1999) The daf-2 gene network for longevity regulates oxidative stress resistance and Mn-superoxide dismutase gene expression in Caenorhabditis elegans. Faseb J 13: 1385–1393.
- 23. Junger MA, Rintelen F, Stocker H, Wasserman JD, Vegh M, et al. (2003) The Drosophila forkhead transcription factor FOXO mediates the reduction in cell number associated with reduced insulin signaling. J Biol 2: 20.
- 24. Kops GJ, Dansen TB, Polderman PE, Saarloos I, Wirtz KW, et al. (2002) Forkhead transcription factor FOXO3a protects quiescent cells from oxidative stress. Nature 419: 316–321.
- 25. Kregel KC, Zhang HJ (2007) An integrated view of oxidative stress in aging: basic mechanisms, functional effects, and pathological considerations. Am J Physiol Regul Integr Comp Physiol 292: R18–36.
- 26. Morris BJ (2005) A forkhead in the road to longevity: the molecular basis of lifespan becomes clearer. J Hypertens 23: 1285–1309.
- 27. Murphy CT (2006) The search for DAF-16/FOXO transcriptional targets: Approaches and discoveries. Experimental Gerontology doi:10.1016/j.exger.2006.06.040.
- 28. Oh SW, Mukhopadhyay A, Dixit BL, Raha T, Green MR, et al. (2006) Identification of direct DAF-16 targets controlling longevity, metabolism and diapause by chromatin immunoprecipitation. Nat Genet 38: 251–257.
- 29. Seidman JG, Seidman C (2002) Transcription factor haploinsufficiency: when half a loaf is not enough. J Clin Invest 109: 451–454.
- 30. Vavouri T, Elgar G (2005) Prediction of cis-regulatory elements using binding site matrices–the successes, the failures and the reasons for both. Curr Opin Genet Dev 15: 395–402.
- 31. Ludwig MZ, Bergman C, Patel NH, Kreitman M (2000) Evidence for stabilizing selection in a eukaryotic enhancer element. Nature 403: 564–567.
- 32. Ludwig MZ, Palsson A, Alekseeva E, Bergman CM, Nathan J, et al. (2005) Functional evolution of a cis-regulatory module. PLoS Biol 3: e93.
- 33. Dermitzakis ET, Clark AG (2002) Evolution of transcription factor binding sites in Mammalian gene regulatory regions: conservation and turnover. Mol Biol Evol 19: 1114–1121.
- 34. Patwari P, Higgins LJ, Chutkow WA, Yoshioka J, Lee RT (2006) The interaction of thioredoxin with Txnip. Evidence for formation of a mixed disulfide by disulfide exchange. J Biol Chem 281: 21884–21891.
- 35. Junn E, Han SH, Im JY, Yang Y, Cho EW, et al. (2000) Vitamin D3 up-regulated protein 1 mediates oxidative stress via suppressing the thioredoxin function. J Immunol 164: 6287–6295.
- 36. Yoshida T, Kondo N, Oka S, Ahsan MK, Hara T, et al. (2006) Thioredoxin-binding protein-2 (TBP-2): its potential roles in the aging process. Biofactors 27: 47–51.
- 37. Jee C, Vanoaica L, Lee J, Park BJ, Ahnn J (2005) Thioredoxin is related to life span regulation and oxidative stress response in Caenorhabditis elegans. Genes Cells 10: 1203–1210.
- 38. Mitsui A, Hamuro J, Nakamura H, Kondo N, Hirabayashi Y, et al. (2002) Overexpression of human thioredoxin in transgenic mice controls oxidative stress and life span. Antioxid Redox Signal 4: 693–696.
- 39. Yoshida T, Oka S, Masutani H, Nakamura H, Yodoi J (2003) The role of thioredoxin in the aging process: involvement of oxidative stress. Antioxid Redox Signal 5: 563–570.
- 40. Yu M, Geiger B, Deeb N, Rothschild MF (2007) Investigation of TXNIP (thioredoxin-interacting protein) and TRX (thioredoxin) genes for growth-related traits in pigs. Mamm Genome.
- 41. Cutler RG (2005) Oxidative stress profiling: part I. Its potential importance in the optimization of human health. Ann N Y Acad Sci 1055: 93–135.
- 42. Hill K, Boesch C, Goodall J, Pusey A, Williams J, et al. (2001) Mortality rates among wild chimpanzees. J Hum Evol 40: 437–450.
- 43. Gilad Y, Rifkin SA, Bertone P, Gerstein M, White KP (2005) Multi-species microarrays reveal the effect of sequence divergence on gene expression profiles. Genome Res 15: 674–680.
- 44. Smyth GK (2005) Limma: linear models for microarray data. In: Gentleman R, Carey V, Dudoit S, Irizarry R, Huber W, editors. Bioinformatics and Computational Biology Solutions using R and Bioconductor. New York: Springer. pp. 397–420.
- 45. Oshlack A, Chabot AE, Smyth GK, Gilad Y (2007) Using DNA microarrays to study gene expression in closely related species. Bioinformatics.
- 46. Smyth GK (2004) Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 3: Epub 2004 Feb 2012.
- 47. Benjamini Y, Hochberg Y (1995) Controlling the False Discovery Rate: a Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society B 57: 289–300.
- 48. Kel AE, Gossling E, Reuter I, Cheremushkin E, Kel-Margoulis OV, et al. (2003) MATCH: A tool for searching transcription factor binding sites in DNA sequences. Nucleic Acids Res 31: 3576–3579.