Reduction/oxidation-phosphorylation control of DNA binding in the bZIP dimerization network
© Amoutzias et al; licensee BioMed Central Ltd. 2006
Received: 07 March 2006
Accepted: 04 May 2006
Published: 04 May 2006
bZIPs are transcription factors that are found throughout the eukarya from fungi to flowering plants and mammals. They contain highly conserved basic region (BR) and leucine zipper (LZ) domains and often function as environmental sensors. Specifically, bZIPs frequently have a role in mediating the response to oxidative stress, a crucial environmental signal that needs to be transduced to the gene regulatory network.
Based on sequence comparisons and experimental data on a number of important bZIP transcription factors, we predict which bZIPs are under redox control and which are regulated via protein phosphorylation. By integrating genomic, phylogenetic and functional data from the literature, we then propose a link between oxidative stress and the choice of interaction partners for the bZIP proteins.
This integration permits the bZIP dimerization network to be interpreted in functional terms, especially in the context of the role of bZIP proteins in the response to environmental stress. This analysis demonstrates the importance of abiotic factors in shaping regulatory networks.
Control of gene expression at the transcriptional level is vital and several mechanisms exist that may regulate the DNA binding of a transcription factor (TF). These include differential heterodimer formation, methylation of the DNA target site , phosphorylation in the TF DNA-binding domain (DBD) , reduction/oxidation (redox) of the DBD , the concentration of cations (particularly magnesium) in the nuclear environment . A combination of differential heterodimer formation together with the phosphorylation and the redox mechanisms may yield complex behaviours that determine the expression or inhibition of downstream targets. We are particularly interested in the complex behaviour that these 3 mechanisms create in the bZIPs, since these TFs are involved in cell proliferation and apoptosis.
Another mechanism of switching on or off DNA binding is the reduction/oxidation of the cysteine residue in position 19 of the BR of bZIPs (C19) , . Oxidation of C19 blocks DNA binding, and this mechanism has been shown to operate for the AP1 proteins. Several mechanisms have been proposed or predicted for the oxidation of C19, such as reversible formation of sulphenic acid, a disulphide bond , S-glutathiolation , or S-cystenyl cystenylation . Furthermore, C19 can either be protected from oxidation by the MBF1 co-activator , or it can be switched back from its oxidised to its reduced state by the ref-1 protein .
The importance of these two signalling mechanisms (redox versus phosphorylation) is stressed by the conservation of the cysteine or serine at this position. Deppmann et al.  report 55% and 35% occurrence of cysteine and serine, respectively, in position 19 in an alignment of human bZIPs. Here, we provide a more detailed phylogenetic analysis using several vertebrate and invertebrate species and highlight the level of conservation of these amino acids, strongly suggesting functional conservation. In addition to conservation at the sequence level, the same cysteine residue and the redox mechanism have been demonstrated experimentally in both human and Drosophila AP1 proteins , , . For the phosphorylation mechanism, the importance of S19 has been shown experimentally for BATF and C/EBP proteins , , . In addition, mutation of the cysteine or serine in position 19 does not affect heterodimerization properties or DNA-binding recognition , , . This mutation appears to affect the selection of the signal (phosphorylation or oxidative stress) that turns on or off the DNA-binding ability of the dimer. Therefore, the conservation of the amino acids in position 19, and the report of functional conservation across different families, strongly indicates that the same mechanism is preserved from Drosophila to humans across many divergent evolutionary lineages.
In this work, we provide a visualization of the bZIP dimerization network and show the level of conservation of C19 and S19 residues across different phylogenetic lineages. It has yet to be established experimentally, for all bZIP proteins, whether they are under redox or phosphorylation control. However, given that all bZIP DNA-binding regions around the C19 and S19 residues have fundamentally similar properties, and extrapolating from experiments on JUN, FOS, C/EBP and BATF proteins, we predict the bZIP proteins in which the redox and phosphorylation mechanisms are utilised. Based on these predictions, we propose that the control mechanism is linked to the evolutionary history of the bZIP families. Interestingly, certain dimerization types are over-represented, while others are under-represented – suggesting strong preference for particular interaction patterns. Specifically, C19 monomers tend to dimerize with other C19 monomers, probably in order to retain redox control and rarely interact with other monomers. While dimerization with other monomer types can occur, such dimers usually have a repression function, so as to avoid inappropriate gene expression.
Results and discussion
Types of dimers formed by the homo/heterodimerization of the 43 human bZIPs.
Amino acid composition of dimer in position 19
Dominant mechanism that blocks DNA binding of the dimer
Number of human bZIP dimers
Observed and expected C-X and C-C type dimers.
homodimers, intra- and inter-family heterodimers
intra- and inter-family heterodimers
It is conceivable that this under-representation of C-X dimers is an artefact and this could have arisen in two different ways. Firstly, we explored the possibility that C19-bearing proteins tend to homodimerize more than the others. Secondly, we explored the possibility that C19-bearing proteins tend to dimerize more with their closest homologues, that is proteins of the same family that also have a cysteine at position 19. In order to exclude biases created by these two options, we also performed the chi-squared test for heterodimers only, and for heterodimers that are not close paralogues (Table 2). In both cases, the over-representation of C-C and under-representation of C-X dimers is statistically significant.
In the above statistical tests, we used all the paralogues of each family. Nevertheless, most paralogues have similar dimerization patterns. If position 19 is not actually linked to the dimerization pattern, but is only responsible for family-specific DNA core-site recognition, then gene duplication (within a protein family) could have caused an artefactual connection. In order to exclude this possibility, all the paralogues of each family were collapsed into one such that the network was then composed of interacting families. We retained the structure of the network among the families, but we shuffled the amino acids of position 19 across the various families 10,000 times. This model showed clearly that, in less than 2% of the shuffled networks, did we obtain an under-representation of C-X family dimers similar to that observed in the data (11 or fewer families forming C-X dimer types) (see also Additional file 2). Furthermore, we performed the same analysis in other positions of the BR alignment (positions 9, 13, 15, 16, 23) that are also strongly conserved within each family, but we did not observe any under/over-representation of amino acid combinations at a cut-off level of 5% (see also Additional file 2), in contrast to what we observed for position 19. It should also be stressed that the experimental evidence from  shows that mutation of C19 did not affect DNA binding, DNA element recognition, or dimerization of the Zta bZIP protein.
Interestingly, it is apparent that the C-X heterodimer type is not favoured. This can be explained by the dominant nature of phosphorylation over redox control. The presence of only one S19 residue would be sufficient to place the DNA-binding properties of the dimer under phosphorylation control. The loss or decrease of the redox mechanism in the BR is known to increase the transforming activity of the JUN-FOS heterodimer , . Thus, it is presumably important to retain the redox mechanism and avoid heterodimerizing with other types (see Figures 4 &5).
What is the function of the C-type bZIPs that actually dimerize with other types (X-types), thus forming C-X type dimers and exhibiting insufficient redox control? By examining the activation/inhibition activity of the X-type partners in general, it appears that E4BP4 , p21-SNTF  and BATF  have an inhibitory effect when dimerizing with other factors. In the case of E4BP4, this is due to the active repression domain that it possesses. The cases of ATF4, C/EBP-β and C/EBP-γ are more complex because they can exhibit activating or inhibiting effects, depending on post-translational modification , alternative splicing , or the cell type in which they are expressed . Nevertheless, they do have the ability to function as inhibitors. It is reasonable to assume that, for C-type molecules, it is generally acceptable to escape from the redox control, as long as they dimerize with an inhibitor, or if the new heterodimer cannot recognise and bind to promoters of downstream targets that need to be controlled by the redox mechanism – thus preventing uncontrolled activation of downstream targets.
Dimerization is an important mechanism for generating complex behaviour with a limited number of protein "building blocks". Work on other dimerizing TF families, like the bHLH, has revealed a dimerization network with a hub-based structure  that seems to work as a multi-switch , especially in development and the cell cycle. A very different network structure was found for the bZIP proteins, despite the fact that they share a similar crystal structure with the bHLH proteins. Interestingly, there seems to be a pattern in the formation of dimers in the bZIPs (Figures 4 &5). These results indicate that environmental signals (and, particularly, oxidative stress) could have imposed some selective pressure on the dimerizing properties of these proteins. Alternatively, the dimerizing properties of each monomer could have imposed some pressure on the presence of cysteine or serine in position 19 of the BR. The redox mechanism has been implicated in the regulation of DNA binding in other TF proteins: p53, Sp1, NFI, NF-κB, PEBP2/CBF, the nuclear receptor proteins (oestrogen and glucocorticoid receptors) and the bHLH protein, USF (reviewed in ). When cells undergo oxidative stress, the cell cycle is affected and it seems that the redox control of cysteine residues in the DNA-binding domain of various TFs is a simple (but very efficient) mechanism of transducing environmental signals to the transcriptional machinery. In addition, oxidative stress has been implicated in the aetiology of several human diseases, like cancer, ischemia, atherosclerosis, neurodegenerative disorders and ageing (reviewed in ). It will be of great interest to further enhance our understanding of how this mechanism works and affects other dimerizing TF families, like the bHLH and nuclear receptors, determining whether this pattern is global or restricted to the bZIPs.
The integration of genomic, phylogenetic and functional data reveals a preference in the interaction partners of bZIP proteins that is linked to oxidative stress. Specifically, bZIP proteins whose DNA binding is controlled by redox tend to dimerize, with a frequency more than that expected by chance, with other bZIP proteins that are also controlled by redox. These results demonstrate that abiotic factors may play a major role in shaping regulatory networks. While the dimerization networks of bHLH proteins and nuclear receptors are hub-based, that of the bZIP proteins is not. Nevertheless we have demonstrated that this network is not random. It follows a logic which strongly links its structure with the network's functional role in environmental sensing.
Protein-protein interaction data for all the human bZIPs were taken from . In that study, the coiled-coils of 43 out of the 51 human bZIPs that we had identified were checked for the presence of an interaction with any of the other bZIP coiled-coils. Each protein was used both as a surface-bait and a probe. Therefore, a given heterodimerization is represented by two different symmetrical (across the diagonal) points in an interaction matrix. We considered an interaction as valid if its Z-score was greater than 2.5 in both directions, where the Z-score is a measure of the signal-to-noise ratio .
bZIP sequences were obtained by genome-wide scanning using custom-made HMMs . The training of the HMM models was based on protein sequences annotated as bZIPs in the TRANSFAC database (version 4) . Four vertebrate (Homo sapiens, Gallus gallus, Takifugu rubripes, Danio rerio), four invertebrate (Ciona intestinalis, Drosophila melanogaster, Apis melifera, Anopheles gambiae) and six fungal (Schizosaccharomyces pombe, Yarrowia lipolytica, Debaryomyces hansenii, Kluyveromyces lactis, Candida glabrata, Saccharomyces cerevisiae) genomes were scanned for bZIP sequences. In addition, cnidarian bZIP sequences were retrieved, using a keyword search, from the NCBI protein database. The inclusion of a sequence as a bZIP 'hit' required both the presence of an LZ and a typical DNA-binding region, as defined by  and the 2ZIP program . This strict criterion was imposed because LZ domains may appear by chance, due to abundance of the leucine residue and the short length of the domain .
Multiple sequence alignments were performed for family members with T-COFFEE  (Notredame et al., 2000) and among different families using CLUSTALW  (Thompson et al., 1994). The alignment was based on the BR and LZ domain. Phylogenetic analysis (neighbour-joining) of the BR was performed by the PHYLIP package , using the PROTDIST and NEIGHBOUR programs, using the JTT model of amino acid replacement. The neighbour-joining tree was visualised with TreeEdit .
Classification of bZIPs in protein families
We classified the 51 human bZIPs into 19 families, based on the neighbour-joining phylogenetic analysis of the BR-LZ domain, combined with the distribution of orthologues and the domain architecture of the whole sequence (G. D. Amoutzias, PhD Thesis, The University of Manchester, 2005). Specifically, the designations of our analysis were based on the presence of invertebrate orthologues and distinct domain architectures for each family. The new designations are: (1) the split of the OASIS family into OASIS and OASIS-B, (2) the split of CNC family into NFE2 and BACH, and (3) the split of the C/EBP family into C/EBP and C/EBP-γ.
We thank Elgar Pichler (AstraZeneca) and Marc Robinson-Rechavi (UNIL) for useful comments and discussions. GDA received a CASE studentship from the EPSRC and AstraZeneca, and was also supported by an EPSRC platform grant (GR/R80810/01) to SGO. Work on protein interactions in DLR's and SGO's groups is supported by the BBSRC and DTI/Beacon. GDA gratefully acknowledges support from Dimitris and Vasiliki Amoutzias.
- Iguchi-Ariga SM, Schaffner W: CpG methylation of the cAMP-responsive enhancer/promoter sequence TGACGTCA abolishes specific factor binding as well as transcriptional activation. Genes Dev. 1989, 3: 612-9.PubMedView ArticleGoogle Scholar
- Mahoney CW, Shuman J, McKnight SL, Chen HC, Huang KP: Phosphorylation of CCAAT-enhancer binding protein by protein kinase C attenuates site-selective DNA binding. J Biol Chem. 1992, 267: 19396-403.PubMedGoogle Scholar
- Abate C, Patel L, Rauscher FJ, Curran T: Redox regulation of fos and jun DNA-binding activity in vitro. Science. 1990, 249: 1157-61.PubMedView ArticleGoogle Scholar
- Moll JR, Acharya A, Gal J, Mir AA, Vinson C: Magnesium is required for specific DNA binding of the CREB B-ZIP domain. Nucleic Acids Res. 2002, 30: 1240-6. 10.1093/nar/30.5.1240.PubMedPubMed CentralView ArticleGoogle Scholar
- Fujii Y, Shimizu T, Toda T, Yanagida M, Hakoshima T: Structural basis for the diversity of DNA recognition by bZIP transcription factors. Nat Struct Biol. 2000, 7: 889-93. 10.1038/82822.PubMedView ArticleGoogle Scholar
- Miller M, Shuman JD, Sebastian T, Dauter Z, Johnson PF: Structural basis for DNA recognition by the basic region leucine zipper transcription factor CCAAT/enhancer-binding protein alpha. J Biol Chem. 2003, 278: 15178-84. 10.1074/jbc.M300417200.PubMedView ArticleGoogle Scholar
- Deppmann CD, Thornton TM, Utama FE, Taparowsky EJ: Phosphorylation of BATF regulates DNA binding: a novel mechanism for AP-1 (activator protein-1) regulation. Biochem J. 2003, 374: 423-31. 10.1042/BJ20030455.PubMedPubMed CentralView ArticleGoogle Scholar
- Xanthoudakis S, Miao G, Wang F, Pan YC, Curran T: Redox activation of Fos-Jun DNA binding activity is mediated by a DNA repair enzyme. EMBO J. 1992, 11: 3323-35.PubMedPubMed CentralGoogle Scholar
- Marshall HE, Merchant K, Stamler JS: Nitrosation and oxidation in the regulation of gene expression. FASEB J. 2000, 14: 1889-900. 10.1096/fj.00.011rev.PubMedView ArticleGoogle Scholar
- Klatt P, Molina EP, De Lacoba MG, Padilla CA, Martinez-Galesteo E, Barcena JA, Lamas S: Redox regulation of c-Jun DNA binding by reversible S-glutathiolation. FASEB J. 1999, 13: 1481-90.PubMedGoogle Scholar
- Jindra M, Gaziova I, Uhlirova M, Okabe M, Hiromi Y, Hirose S: Coactivator MBF1 preserves the redox-dependent AP-1 activity during oxidative stress in Drosophila. EMBO J. 2004, 23: 3538-47. 10.1038/sj.emboj.7600356.PubMedPubMed CentralView ArticleGoogle Scholar
- Okuno H, Akahori A, Sato H, Xanthoudakis S, Curran T, Iba H: Escape from redox regulation enhances the transforming activity of Fos. Oncogene. 1993, 8: 695-701.PubMedGoogle Scholar
- Morgan IM, Havarstein LS, Wong WY, Luu P, Vogt PK: Efficient induction of fibrosarcomas by v-jun requires mutations in the DNA binding region and the transactivation domain. Oncogene. 1994, 9: 2793-7.PubMedGoogle Scholar
- Trautwein C, van der Geer P, Karin M, Hunter T, Chojkier M: Protein kinase A and C site-specific phosphorylations of LAP (NF-IL6) modulate its binding affinity to DNA recognition elements. J Clin Invest. 1994, 93: 2554-61.PubMedPubMed CentralView ArticleGoogle Scholar
- Fernandes L, Rodrigues-Pousada C, Struhl K: Yap, a novel family of eight bZIP proteins in Saccharomyces cerevisiae with distinct biological functions. Mol Cell Biol. 1997, 17: 6982-93.PubMedPubMed CentralView ArticleGoogle Scholar
- Barabasi AL, Oltvai ZN: Network biology: understanding the cell's functional organization. Nat Rev Genet. 2004, 5: 101-113. 10.1038/nrg1272.PubMedView ArticleGoogle Scholar
- Amoutzias GD, Robertson DL, Oliver SG, Bornberg-Bauer E: Convergent evolution of gene networks by single-gene duplications in higher eukaryotes. EMBO Rep. 2004, 5: 274-9. 10.1038/sj.embor.7400096.PubMedPubMed CentralView ArticleGoogle Scholar
- Cinquin O, Demongeot J: High-dimensional switches and the modelling of cellular differentiation. J Theor Biol. 2005, 233: 391-411. 10.1016/j.jtbi.2004.10.027.PubMedView ArticleGoogle Scholar
- Newman JR, Keating AE: Comprehensive identification of human bZIP interactions with coiled-coil arrays. Science. 2003, 300: 2097-101. 10.1126/science.1084648.PubMedView ArticleGoogle Scholar
- Schelcher C, Valencia S, Delecluse HJ, Hicks M, Sinclair AJ: Mutation of a single amino acid residue in the basic region of the Epstein-Barr virus (EBV) lytic cycle switch protein Zta (BZLF1) prevents reactivation of EBV from latency. J Virol. 2005, 79: 13822-8. 10.1128/JVI.79.21.13822-13828.2005.PubMedPubMed CentralView ArticleGoogle Scholar
- Cowell IG, Hurst HC: Transcriptional repression by the human bZIP factor E4BP4: definition of a minimal repression domain. Nucleic Acids Res. 1994, 22: 59-65.PubMedPubMed CentralView ArticleGoogle Scholar
- Bower KE, Zeller RW, Wachsman W, Martinez T, McGuire KL: Correlation of transcriptional repression by p21(SNFT) with changes in DNA.NF-AT complex interactions. J Biol Chem. 2002, 277: 34967-77. 10.1074/jbc.M205048200.PubMedView ArticleGoogle Scholar
- Hai T, Hartman MG: The molecular biology and nomenclature of the activating transcription factor/cAMP responsive element binding family of transcription factors: activating transcription factor proteins and homeostasis. Gene. 2001, 273: 1-11. 10.1016/S0378-1119(01)00551-0.PubMedView ArticleGoogle Scholar
- Martinez-Jimenez CP, Gomez-Lechon MJ, Castell JV, Jover R: Transcriptional regulation of the human hepatic CYP3A4: Identification of a new distal enhancer region responsive to CCAAT/enhancer binding protein beta isoforms (LAP and LIP). Mol Pharmacol. 2005, 6: 2088-101. 10.1124/mol.104.008169.View ArticleGoogle Scholar
- Parkin SE, Baer M, Copeland TD, Schwartz RC, Johnson PF: Regulation of CCAAT/enhancer-binding protein (C/EBP) activator proteins by heterodimerization with C/EBPgamma (Ig/EBP). J Biol Chem. 2002, 277: 23563-72. 10.1074/jbc.M202184200.PubMedView ArticleGoogle Scholar
- Morel Y, Barouki R: Repression of gene expression by oxidative stress. Biochem J. 1999, 342 (Pt 3): 481-96. 10.1042/0264-6021:3420481.PubMedPubMed CentralView ArticleGoogle Scholar
- Toone WM, Morgan BA, Jones N: Redox control of AP-1-like factors in yeast and beyond. Oncogene. 2001, 20: 2336-2346. 10.1038/sj.onc.1204384.PubMedView ArticleGoogle Scholar
- Eddy SR: Hidden Markov models. Curr Opin Struct Biol. 1996, 6: 361-365. 10.1016/S0959-440X(96)80056-X.PubMedView ArticleGoogle Scholar
- Matys V, Fricke E, Geffers R, Gossling E, Haubrock M, Hehl R, Hornischer K, Karas D, Kel AE, Kel-Margoulis OV, Kloos DU, Land S, Lewicki-Potapov B, Michael H, Munch R, Reuter I, Rotert S, Saxel H, Scheer M, Thiele S, Wingender E: TRANSFAC: transcriptional regulation, from patterns to profiles. Nucleic Acids Res. 2003, 31: 374-378. 10.1093/nar/gkg108.PubMedPubMed CentralView ArticleGoogle Scholar
- Vinson C, Myakishev M, Acharya A, Mir AA, Moll JR, Bonovich M: Classification of human B-ZIP proteins based on dimerization properties. Mol Cell Biol. 2002, 22: 6321-35. 10.1128/MCB.22.18.6321-6335.2002.PubMedPubMed CentralView ArticleGoogle Scholar
- Bornberg-Bauer E, Rivals E, Vingron M: Computational approaches to identify leucine zippers. Nucleic Acids Res. 1998, 26: 2740-2746. 10.1093/nar/26.11.2740.PubMedPubMed CentralView ArticleGoogle Scholar
- Brendel V, Karlin S: Too many leucine zippers?. Nature. 1989, 341: 574-5. 10.1038/341574a0.PubMedView ArticleGoogle Scholar
- Notredame C, Higgins DG, Heringa J: T-Coffee: A novel method for fast and accurate multiple sequence alignment. J Mol Biol. 2000, 302: 205-17. 10.1006/jmbi.2000.4042.PubMedView ArticleGoogle Scholar
- Thompson JD, Higgins DG, Gibson TJ: CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 1994, 22: 4673-80.PubMedPubMed CentralView ArticleGoogle Scholar