Downloaded from genome. cshlp org on November 3, 2010- Published by Cold Spring Harbor Laboratory Press 感快ERF Coevolution within a transcriptional network by compensatory trans and cis mutations Dwight Kuo, Katherine Licon, Sourav Bandyopadhyay, et al Genome Res. published online October 26, 2010 Access the most recent version at doi: 10. 1101/gr 111765 110 Supplementalhttp:/genome.cshlp.org/content/suppl/2010/09/30/gr.111765.110.dc1.html Material P<P Published online October 26, 2010 in advance of the print journal Open Access Freely available online through the Genome Research Open Access option Email alerting Receive free email alerts when new articles cite this article- sign up in the box at the service top right corner of the article or click here Advance online articles have been peer reviewed and accepted for publication but have not yet appeared in the paper journal(edited, typeset versions may be posted when available prior to final ublication). Advance online articles are citable and establish publication priority; they are indexed y PubMed from initial publication Citations to Advance online articles must include the digital object identifier(DOls)and date of initial publication To subscribe to Genome Research ons http:/genome.cshlp.org/subscripti Copyright 2010 by Cold Spring Harbor Laboratory Press
Access the most recent version at doi:10.1101/gr.111765.110 Genome Res. published online October 26, 2010 Dwight Kuo, Katherine Licon, Sourav Bandyopadhyay, et al. trans and cis mutations Coevolution within a transcriptional network by compensatory Material Supplemental http://genome.cshlp.org/content/suppl/2010/09/30/gr.111765.110.DC1.html P<P Published online October 26, 2010 in advance of the print journal. Open Access Freely available online through the Genome Research Open Access option. service Email alerting top right corner of the article or click here Receive free email alerts when new articles cite this article - sign up in the box at the object identifier (DOIs) and date of initial publication. by PubMed from initial publication. Citations to Advance online articles must include the digital publication). Advance online articles are citable and establish publication priority; they are indexed appeared in the paper journal (edited, typeset versions may be posted when available prior to final Advance online articles have been peer reviewed and accepted for publication but have not yet http://genome.cshlp.org/subscriptions To subscribe to Genome Research go to: Copyright © 2010 by Cold Spring Harbor Laboratory Press Downloaded from genome.cshlp.org on November 3, 2010 - Published by Cold Spring Harbor Laboratory Press
Downloaded from genome. cshlp org on November 3, 2010- Published by Cold Spring Harbor Laboratory Press Research Coevolution within a transcriptional network by compensatory trans and cis mutations Justin Catalana, Timothy Ravasi, Kai Tan, , and Irey ldeker1, g ang> Dwight Kuo, Katherine Licon, Sourav Bandyopadhyay, Ryan Ch ' Departments of Bioengineering and Medicine, University of Califonia, San Diego, La Jolla, California 92093, USA; 'Red Sea Laboratory of Integrative Systems Biology, Division of Chemical and Life Sciences and Engineering, Computational Bioscience Research Center, King Abdullah University of Science and Technology, Thuwal 23955-6900, Kingdom of Saudi Arabia; Departments of Internal Medicine and Biomedical Engineering, University of lowa, lowa City, lowa 52242, USA Transcriptional networks have been shown to evolve very rapidly prompting questions as to how such changes arise and ire tolerated Recent comparisons of transcriptional networks across species have implicated variations in the cis-actin DNA sequences near genes as the main cause of divergence what is less clear is how these changes interact with trans-acting hanges occurring elsewhere in the genetic circuit. Here, we report the discovery of a system of compensatory trans and cis mutations in the yeast AP-l transcriptional network that allows for conserved transcriptional regulation despite continued genetic change We pinpoint a single species, the fungal pathogen Candida glabrata, in which a trans mutation has occurred very recently in a single ap-l family member distinguishing it from its Saccharomyces ortholog. Comparison of chromatin immunoprecipitation profiles between Candida and Saccharomyces shows that despite their different dna-binding domains, the Ap-l orthologs regulate a conserved block of genes. this conservation is enabled by concomitant changes in the cis- regulatory motifs upstream of each gene. Thus, both trans and cis mutations have perturbed the yeast AP-l regulatory ystem in such a way as to compensate for one another this demonstrates an example of "coevolution"between a dna- binding transcription factor and its cis-regulatory site, reminiscent of the coevolution of protein binding partners upplementalmaterialisavailableonlineathttp.www.genome.orgthesequencedatafromthisstudyhavebeen submittedtoheNcblGeneExpressionOmnibus(http://www.ncbi.nlmnihgov/geo/underaccessionno.Gsel5818.] Transcriptional networks are central to understanding both evo- identified transcriptional programs that are dramatically rewired lution and phenotypic diversity among organisms. Of the many over short evolutionary time scales. As with earlier work, many of ways in which transcriptional networks can evolve, much atten- the observed differences in binding and expression have been n has been given to changes in the so-called cis-regulatory re- linked to changes in cis-regulatory regions. For example, Borneman gions of gene promoters (Wray 2007: Wagner and Lynch 2008). et al.(2007)found that the TF Tecl binds only 20% of the same Such changes include gain, loss, or modification of DNA sequence target genes in comparisons between Saccharomyces cerevisiae and motifs(Cliften et al. 2003; Kelliset aL 2003; Gasch et al. 2004; Stark the closely related Saccharomyces bayanus and saccharomyces et al. 2007) as well as alterations in motif spacing relative to the mikatae, and that this difference is due to gain and loss of canonical start of transcription, or to other motifs (Ihmels et al. 2005; Tanay Tecl cis-regulatory motifs. While some recent studies have asso- et al. 2005). In addition to changes in cis, transcriptional networks ciated genetic variants in TEs with gene expression changes ob- can also evolve through alterations to transcription factor (TF) served in interspecies hybrids(wilson et al. 2008; Wittkopp et al. proteins and other trans-acting factors(Wagner and Lynch 2008). 2008; Tirosh et al. 2009; Bullard et al. 2010; Emerson et al. 2010), in trans, potential mechanisms include mutations to protein struc- Gerke et al. 2009; Sung et al. 2009; Zheng et al. 2010), or in human ture impacting transcriptional activation or DNA-binding do- populations(Kasowski et al. 2010), the picture that emerges is that mains(Wagner and Lynch 2008), modulation of TF expression cis-regulatory regions are incredibly plastic over evolutionary time, Sankaran et al. 2009) or post-translational modifications(Holt while TFs(trans)evolve at a comparatively slower rate(Wray 2007) et al. 2009), or gain and loss of protein-protein interactions amo Given the dramatic changes that appear to be occurring in TFs (Tuch et al. 2008; Lavoie et al. 2010). transcriptional networks, a key question is how such systems re- Recently, a number of genome-scale studies have performed tain essential functions over evolutionary time (Wray 2007).One systematic comparisons of TF-binding patterns(Borneman et al. solution is that changes in cis can occur by replacement of one TI 08: Bra Lavoie et al. 2010: cofactor with another, thereby maintaining regulatory control Schmidt et al. 2010)or mRNA expression profiles across species (Tsong et al. 2006). Alternatively, rather than replacing specific Ihmels et al. 2005; Tanayet al 2005; Hogues et al. 2008; Fieldet al. cofactors, it is conceivable that the DNA-binding domains of the 2009; Wapinski et al. 2010). Almost universally, these studies have TFs that bind these cis-regulatory sequences might be altered in lock-step with changes in cis, similarly to the evolution of protein binding partners(Pazos and Valencia 2008). However, such a mechanism of evolution has yet to be observed Here, we present rticle published or Article and publication date are at a direct example of such"coevolution, " where a specific change to http://www.genome.org/cgi/doi/10.1101/gr.111765.110.Freelyavailable DNA-binding transcription factor and its cis-regulatory site have online through the Genome Research Open Access option. occurred in compensatory fashion. 20: 000-000@ 2010 by Cold Spring Harbor Laboratory Press; ISSN 1088-9051/10
Research Coevolution within a transcriptional network by compensatory trans and cis mutations Dwight Kuo,1 Katherine Licon,1 Sourav Bandyopadhyay,1 Ryan Chuang,1 Colin Luo,1 Justin Catalana,1 Timothy Ravasi,1,2 Kai Tan,3,4 and Trey Ideker1,4 1 Departments of Bioengineering and Medicine, University of California, San Diego, La Jolla, California 92093, USA; 2 Red Sea Laboratory of Integrative Systems Biology, Division of Chemical and Life Sciences and Engineering, Computational Bioscience Research Center, King Abdullah University of Science and Technology, Thuwal 23955-6900, Kingdom of Saudi Arabia; 3 Departments of Internal Medicine and Biomedical Engineering, University of Iowa, Iowa City, Iowa 52242, USA Transcriptional networks have been shown to evolve very rapidly, prompting questions as to how such changes arise and are tolerated. Recent comparisons of transcriptional networks across species have implicated variations in the cis-acting DNA sequences near genes as the main cause of divergence. What is less clear is how these changes interact with trans-acting changes occurring elsewhere in the genetic circuit. Here, we report the discovery of a system of compensatory trans and cis mutations in the yeast AP-1 transcriptional network that allows for conserved transcriptional regulation despite continued genetic change. We pinpoint a single species, the fungal pathogen Candida glabrata, in which a trans mutation has occurred very recently in a single AP-1 family member, distinguishing it from its Saccharomyces ortholog. Comparison of chromatin immunoprecipitation profiles between Candida and Saccharomyces shows that, despite their different DNA-binding domains, the AP-1 orthologs regulate a conserved block of genes. This conservation is enabled by concomitant changes in the cisregulatory motifs upstream of each gene. Thus, both trans and cis mutations have perturbed the yeast AP-1 regulatory system in such a way as to compensate for one another. This demonstrates an example of ‘‘coevolution’’ between a DNAbinding transcription factor and its cis-regulatory site, reminiscent of the coevolution of protein binding partners. [Supplemental material is available online at http://www.genome.org. The sequence data from this study have been submitted to he NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under accession no. GSE15818.] Transcriptional networks are central to understanding both evolution and phenotypic diversity among organisms. Of the many ways in which transcriptional networks can evolve, much attention has been given to changes in the so-called cis-regulatory regions of gene promoters (Wray 2007; Wagner and Lynch 2008). Such changes include gain, loss, or modification of DNA sequence motifs (Cliften et al. 2003; Kellis et al. 2003; Gasch et al. 2004; Stark et al. 2007) as well as alterations in motif spacing relative to the start of transcription, or to other motifs (Ihmels et al. 2005; Tanay et al. 2005). In addition to changes in cis, transcriptional networks can also evolve through alterations to transcription factor (TF) proteins and other trans-acting factors (Wagner and Lynch 2008). Although there have been fewer reports of evolutionary changes in trans, potential mechanisms include mutations to protein structure impacting transcriptional activation or DNA-binding domains (Wagner and Lynch 2008), modulation of TF expression (Sankaran et al. 2009) or post-translational modifications (Holt et al. 2009), or gain and loss of protein–protein interactions among TFs (Tuch et al. 2008; Lavoie et al. 2010). Recently, a number of genome-scale studies have performed systematic comparisons of TF-binding patterns (Borneman et al. 2007; Tuch et al. 2008; Bradley et al. 2010; Lavoie et al. 2010; Schmidt et al. 2010) or mRNA expression profiles across species (Ihmels et al. 2005; Tanay et al. 2005; Hogues et al. 2008; Field et al. 2009; Wapinski et al. 2010). Almost universally, these studies have identified transcriptional programs that are dramatically rewired over short evolutionary time scales. As with earlier work, many of the observed differences in binding and expression have been linked to changes in cis-regulatory regions. For example, Borneman et al. (2007) found that the TF Tec1 binds only 20% of the same target genes in comparisons between Saccharomyces cerevisiae and the closely related Saccharomyces bayanus and Saccharomyces mikatae, and that this difference is due to gain and loss of canonical Tec1 cis-regulatory motifs. While some recent studies have associated genetic variants in TFs with gene expression changes observed in interspecies hybrids (Wilson et al. 2008; Wittkopp et al. 2008; Tirosh et al. 2009; Bullard et al. 2010; Emerson et al. 2010), in outbred crosses (Brem and Kruglyak 2005; Landry et al. 2005; Gerke et al. 2009; Sung et al. 2009; Zheng et al. 2010), or in human populations (Kasowski et al. 2010), the picture that emerges is that cis-regulatory regions are incredibly plastic over evolutionary time, while TFs (trans) evolve at a comparatively slower rate (Wray 2007). Given the dramatic changes that appear to be occurring in transcriptional networks, a key question is how such systems retain essential functions over evolutionary time (Wray 2007). One solution is that changes in cis can occur by replacement of one TF cofactor with another, thereby maintaining regulatory control (Tsong et al. 2006). Alternatively, rather than replacing specific cofactors, it is conceivable that the DNA-binding domains of the TFs that bind these cis-regulatory sequences might be altered in lock-step with changes in cis, similarly to the evolution of proteinbinding partners (Pazos and Valencia 2008). However, such a mechanism of evolution has yet to be observed. Here, we present a direct example of such ‘‘coevolution,’’ where a specific change to a DNA-binding transcription factor and its cis-regulatory site have occurred in compensatory fashion. 4 Corresponding authors. E-mail tideker@ucsd.edu. E-mail kai-tan@uiowa.edu. Article published online before print. Article and publication date are at http://www.genome.org/cgi/doi/10.1101/gr.111765.110. Freely available online through the Genome Research Open Access option. 20:000–000 ! 2010 by Cold Spring Harbor Laboratory Press; ISSN 1088-9051/10; www.genome.org Genome Research 1 www.genome.org Downloaded from genome.cshlp.org on November 3, 2010 - Published by Cold Spring Harbor Laboratory Press
ownloaded from genome. cshlp org on November 3, 2010- Published by Cold Spring Harbor Laboratory Press Kuo et al As a model of transcriptional network evolution, we exam- tin immunoprecipitation with microarray hybridization( ChIP- ined the yeast AP-1 (yAP-1)family, which, with a total of eight chip) data in Sc(Harbison et al. 2004; Tan et al. 2008) have de- (Fernandes et al. 1997; Rodrigues-Pousada et al. 2010). Like other Yap8) recognize YRE-O, whereas two family members(Yap4 and paralogous families, AP-1 factors have been born through the pro- Yap6)recognize YRE-A We examined the binding of the remaining cess of gene duplication, which gives rise to multiple copies that Sc AP-1 member Yap3 by ChIP-chip and determined it preferre are free from selective pressure and may functionally diverge from YRE-A sites in both complete media and stress conditions (Sup- their duplicates by sub- or neofunctionalization(Hittinger and plemental Fig. 1). This preference for YRE-o or YRE-A-binding Carroll 2007). AP-1 also provides a classic example of the basic sites in Sc AP-ls correlates precisely with the presence of arginine leucine zipper(bZIP) motif, which is widely conserved across or lysine at residue 12(Fig. 1A) eukaryotes(Tan et al. 2008; Rodrigues-Pousada et al. 2010).In Interestingly, residue 12 is part of an alpha-helical surface that humans, AP-1 TEs have been heavily studied due to their crucial forms multiple contacts to DNA (basic region residues 7-15)(Fig. role in cell proliferation, death, and differentiation(Shaulian and 1B; Fujii et al. 2000). Previously, this residue was predicted as Karin 2002). In yeast, yAP-1-mediated transcriptional networks a likely determinant of DNA half-site spacing preference in Gcn4 carry out overlapping, but distinct biological responses to stress another bZIP family TF(Kim and Struhl 1995). Although in vitro (Tan et al. 2008; Rodrigues-Pousada et al. 2010). In contrast to the testing of Gcn4 mutants was not able to confirm this prediction despread divergence in tF binding that has been demonstrated(Kim and Struhl 1995), it has become apparent that such variations reviously(Borneman et al. 2007; Tuch et al. 2008; Lavoie et al. in half-site recognition are best distinguished in vivo(Suckow and 2010), we show that coupled trans and cis mutations enable con- Hollenberg 1998; Berger et al. 2008; Maerkl and Quake 2009) servation of a subset of genes targeted by yAP-1. These result provide an example of compensatory coevolution of a trans and Residue 12 point mutations cause rewiring of AP-1 regulatory system. transcriptional interactio To further examine the regulatory role of residue 12, we mutated Results this residue in Yapl, a representative YRE-O-binding factor, and a trans mutation is associated with Ap-1 DNA-binding motif specificity A B To identify trans mutations that could be Sc yAP-1 Family ★ NRAAQRAF associated with AP-l-binding preference Yap8(20-40)LRA 9 TTACTAA KNN we pertormed an amino acid sequence AATGATT of all eight AP-l-like TEs in S. cerevisiae cYap2(47-67 (Sc). This alignment and its associated phylogenetic tree(Fig. 1A)were searched CATT Yap Half-site to identify the key polymorphic amino ogen Bond acids whose pattern ScYap1 vs. ScYap1 R79K Binding ScYap4 vS ScYap4 K252R Binding divergence best explain the phylogeny Number of Bound Promoters D Number of Bound Promote (Methods, Evolutionary Trace Ana Such residues have been shown to fre. ■ ScRap4 bound quently play important evolutionary roles YRE-O YRE-O ■ ScRap4.K252 R Bound (Innis et al. 2000). Using this approach, we identified residue 12 of the dna- RE-A a ScYap1. R79K Bound YRE-A binding domain basic region as the most important evolutionarily divergent posi- E ScYap1 vs. ScYap1 R79K mRNA Expression F ScYap4 vs ScYap4 K252R mRNA Expression tion across the yAP-1 family (i. e, the one that was most highly correlated with the Fraction of Promoters with DNA-Binding Motifs Fraction of Promoters with DNA-Binding Motifs phylogeny; Fig 1A) Residue 12 was also predictive of Yap1, R79K AP-1 family DNA-binding motif prefer Background p=24×101 ence(Fig. 1A)(MacIsaac et al. 2006; Tar Background p=1.8x103 et al. 2008). AP-1 family members bind DNA as homo- or heterodimers, where Figure 1. A single residue determines yAP-1 DNA-binding motif specificity.(A) Alignment and each constituent monomer recognize the consensus sequence TTAC (Suckow ar) is predictive of preference for overlapping (YRE-O)or adjacent(YRE-A)DNA-bindi (left) YRE.O et al.1999; Fujii et al. 2000). These"half- 2004). Positions affecting Gcn4 half-site spacing preference(Kim and Struhl 1995)are shown(gray sites"are positioned in either adjacent or stars ). (B) Recognition of the yAP-1 half-site(Fuji et al. 2000).Residue 12(red star)is in close proxin overlapping fashion( Fig. 1A), which we to residues conferring AP-1 sequence specificity. (C, D)ScYap1 R79K and ScYap4 K252R mutants have refer to as yAP-1 response element adja- altered half-site spacing preference as evidenced cent (YRE-A)or yAP-1 response element with either YRE-O or YRE-A sites as assessed by Fishers exact test.(E, F) ScYapl R79 and ScYap4 K252 mRNA expression changes among with YRE-o and YRE-A overlapping (YRE.O), respectively. Pre- sites among the top 50 most differentially expressed genes. P-values denote the significance of YRE-A vious analyses of genome-wide chroma- and YRE-o motifs among gene promoters compared with the genomic background. 2 Genome Research
As a model of transcriptional network evolution, we examined the yeast AP-1 (yAP-1) family, which, with a total of eight members, is one of the largest paralogous TF families in S. cerevisiae (Fernandes et al. 1997; Rodrigues-Pousada et al. 2010). Like other paralogous families, AP-1 factors have been born through the process of gene duplication, which gives rise to multiple copies that are free from selective pressure and may functionally diverge from their duplicates by sub- or neofunctionalization (Hittinger and Carroll 2007). AP-1 also provides a classic example of the basic leucine zipper (bZIP) motif, which is widely conserved across eukaryotes (Tan et al. 2008; Rodrigues-Pousada et al. 2010). In humans, AP-1 TFs have been heavily studied due to their crucial role in cell proliferation, death, and differentiation (Shaulian and Karin 2002). In yeast, yAP-1-mediated transcriptional networks carry out overlapping, but distinct biological responses to stress (Tan et al. 2008; Rodrigues-Pousada et al. 2010). In contrast to the widespread divergence in TF binding that has been demonstrated previously (Borneman et al. 2007; Tuch et al. 2008; Lavoie et al. 2010), we show that coupled trans and cis mutations enable conservation of a subset of genes targeted by yAP-1. These results provide an example of compensatory coevolution of a trans and cis regulatory system. Results A trans mutation is associated with AP-1 DNA-binding motif specificity To identify trans mutations that could be associated with AP-1-binding preference, we performed an amino acid sequence alignment of the DNA-binding domains of all eight AP-1-like TFs in S. cerevisiae (Sc). This alignment and its associated phylogenetic tree (Fig. 1A) were searched to identify the key polymorphic amino acids whose patterns of conservation and divergence best explain the phylogeny (Methods, Evolutionary Trace Analysis). Such residues have been shown to frequently play important evolutionary roles (Innis et al. 2000). Using this approach, we identified residue 12 of the DNAbinding domain basic region as the most important evolutionarily divergent position across the yAP-1 family (i.e., the one that was most highly correlated with the phylogeny; Fig. 1A). Residue 12 was also predictive of AP-1 family DNA-binding motif preference (Fig. 1A) (MacIsaac et al. 2006; Tan et al. 2008). AP-1 family members bind DNA as homo- or heterodimers, where each constituent monomer recognizes the consensus sequence TTAC (Suckow et al. 1999; Fujii et al. 2000). These ‘‘halfsites’’ are positioned in either adjacent or overlapping fashion (Fig. 1A), which we refer to as yAP-1 response element adjacent (YRE-A) or yAP-1 response element overlapping (YRE-O), respectively. Previous analyses of genome-wide chromatin immunoprecipitation with microarray hybridization (ChIPchip) data in Sc (Harbison et al. 2004; Tan et al. 2008) have determined that five AP-1 family members (Yap1, Yap2, Yap5, Yap7, Yap8) recognize YRE-O, whereas two family members (Yap4 and Yap6) recognize YRE-A. We examined the binding of the remaining Sc AP-1 member Yap3 by ChIP-chip and determined it preferred YRE-A sites in both complete media and stress conditions (Supplemental Fig. 1). This preference for YRE-O or YRE-A-binding sites in Sc AP-1s correlates precisely with the presence of arginine or lysine at residue 12 (Fig. 1A). Interestingly, residue 12 is part of an alpha-helical surface that forms multiple contacts to DNA (basic region residues 7–15) (Fig. 1B; Fujii et al. 2000). Previously, this residue was predicted as a likely determinant of DNA half-site spacing preference in Gcn4, another bZIP family TF (Kim and Struhl 1995). Although in vitro testing of Gcn4 mutants was not able to confirm this prediction (Kim and Struhl 1995), it has become apparent that such variations in half-site recognition are best distinguished in vivo (Suckow and Hollenberg 1998; Berger et al. 2008; Maerkl and Quake 2009). Residue 12 point mutations cause rewiring of AP-1 transcriptional interactions To further examine the regulatory role of residue 12, we mutated this residue in Yap1, a representative YRE-O-binding factor, and Figure 1. A single residue determines yAP-1 DNA-binding motif specificity. (A) Alignment and phylogeny of AP-1 DNA-binding domain basic regions (residues 6 to 20 are shown). Residue 12 (red star) is predictive of preference for overlapping (YRE-O) or adjacent (YRE-A) DNA-binding motifs (left). Note that Yap8 possesses an Asp at residue 12 and binds a 2-bp overlapping YRE-O (Harbison et al. 2004). Positions affecting Gcn4 half-site spacing preference (Kim and Struhl 1995) are shown (gray stars). (B) Recognition of the yAP-1 half-site (Fujii et al. 2000). Residue 12 (red star) is in close proximity to residues conferring AP-1 sequence specificity. (C,D) ScYap1.R79K and ScYap4.K252R mutants have altered half-site spacing preference as evidenced by ChIP-chip (Methods). P-values refer to differences in binding to genes with either YRE-O or YRE-A sites as assessed by Fisher’s exact test. (E,F ) ScYap1.R79K and ScYap4.K252R mutations cause mRNA expression changes among genes with YRE-O and YRE-A sites among the top 50 most differentially expressed genes. P-values denote the significance of YRE-A and YRE-O motifs among gene promoters compared with the genomic background. 2 Genome Research www.genome.org Kuo e t al. Downloaded from genome.cshlp.org on November 3, 2010 - Published by Cold Spring Harbor Laboratory Press
ownloaded from genome. cshlp org on November 3, 2010- Published by Cold Spring Harbor Laboratory Press Coevolution within a transcriptional network Yap 4(also known as Cin5), a representative YRE-A-binding fac We used ChIP-chip to determine whether this Lys 12 sub- tor. This process involved generating mutants Yapl. R79K and stitution had a functional effect on CgApl binding (Methods). To Yap4. K252R, changing arginine to lysine in Yapl and lysine to facilitate this assay, we tagged CgApl with the TAP epitope and arginine in Yap4 (Methods). Next, Yapl. R79K binding and designed a custom microarray tiling the C genome(Methods). As Yap4. K252R binding were assayed in vivo using ChIP-chip(Meth- a control on both the TAP construct and the array design, we used ods). Comparison of the top 50 promoters bound by Yap1. R79K ChIP-qPCR to successfully validate a panel of five randomly cho- with the top 50 promoters bound by wild-type Yapl (as de- sen Cg gene promoters that were determined to be bound by termined in Tan et al. 2008)showed that mutation of Yapl CgApl in the ChIP-chip experiment(Supplemental Fig. 5) nificantly altered its preference for YRE-O and YRE-A sites(Fig. 1c; We found that CgApl bound the promoters of a total of 114 Fisher's exact test P=0.0002). Comparison of promoters bound by genes, 90 of which had known orthologs in Sc(Methods). Com- mutant and wild-type Yap 4 also showed the predicted shift in parison of these data with ChIP profiles for each of the AP-1 factors binding preference(Fig. 1D; Fisher's exact test P=0.037). These in Sc grown under the same treatment conditions(as determined results were not dependent on the number of promoters examined in Tan et al. 2008)showed significant overlap between the targets of CgApl and ScYapl(17 genes, P< 10) Overlap with other Sc ext, to assess the functional implications of changes in AP-1 factors was less substantial(Fig 3A). This pattern of overlap yAP-1 binding, we generated genome-wide mRNA expression pro- was reinforced by sequence analysis, in which phylogenetic clus- files for each mutant in comparison to the unmutated parental tering of AP-1 DNA-binding domains places CgApl definitively strain (Methods). Both mutations, Yapl.R79K and Yap4 K252R, with ScYapl and not with other Sc AP-1 sequences(Fig. 2C; altered the expression of genes whose promoters were highly Methods) enriched for AP-1-binding sites (YRE-O and YRE-A, Fig. lE, F; Sup. We were therefore faced with the following conundrum: On plemental Fig 3). These genes were also enriched for Yapl. R79K the one hand, the CgApl sequence diverges from Yapl orthologs at and Yap4 K252R binding(P<10), respectively residue 12, suggesting a shift in dna binding On the other hand, he CgApl-binding profile is quite specifically conserved with that An apparent paradox: Candida AP-l diverges at residue 12, of Yapl, calling into question the importance of residue 12 for but its targets are conserved Based on our observation that residue 12 affects binding of AP-1 paralogs in S. cerevisiae, we next asked whether changes in this Cg apl prefers yre-a rather than YRe-O sites residue could lead to divergent binding of AP-1 orthologs across To investigate this apparent contradiction, we next turned to the pecies. We searched the yeast phylogeny (Wapinski et al. 2007) for gene promoters targeted by CgApl in the ChIP assay. Promoters AP-l orthologs that were anomalous in their use of Arg 12 or Lys targeted by Cg Apl showed a clear preference for YRE-A sites over 12, suggesting lineage-specific mutation(Supplemental Fig. 4). YREO sites(49 vs. four promoters, respectively ) This preference mong TFs orthologous to Sc YAPl, we found that the Candida significantly differs from ScYapl, which prefers YRE-O over YRE-A glabrata( Cg)ortholog CgAPI diverges from other yeasts(Fig. A-C)( Fisher's exact test P= 3.5 X 10-; Fig. 3B, 21 vs. 12 promoter due to the presence of lysine at residue 12, in contrast to other respectively). This preference could not be attributed to threshole yeasts in its clade that possess an arginine. This CgApl amino acid effects on binding-site calls, as direct comparison of motif scores substitution was confirmed by sequencing of genomic DNA from confirmed a preference for YRE-A over YRE-O sites( Mann-Whitney two independent Cg isolates, 2001HTU and NCCLS84( Fig 2B) U test, P=0.0072). This preference was also observed via de novo nd even among the Cg orthologs of all B Species Position Sequence We further analyzed this cis-regula- Soto MMaMlodow tory preference by examining the ortho- logs of genes targeted by both CgApl and Sbayanus ScYapl across 20 sequenced yeast ge- nomes(Wapinski et al. 2007). C glabrata c Phylogeny of Se and Cg yAP-1 DNA Binding Domains stood out clearly as the only species with enrichment for YRE-A sites( Fig. 3D). In contrast. the YRE-o site was enriched in all neighboring species in the yeast other sensu stricto species (S paradoxus S. mikatae, and S. bayanus) as well as the more diverged Saccharomyces castellii, romyces walti, Kluyveromyces lactis Ashbya gosspyii, and Candida tropicalis ocfasporus These results indicate that upstream DNA-binding motifs of CgApl targets Fs (A)CgApl possesses a lysine at residue 12(CgAp1 46) have evolved from YRE-o to YRE-A(Fig. as rtv eals tateterepr and scrap COPhustr. nter al branch paint unders refer to the accompanied by concordant changes in osterior probability a measure of confidence( Drummond and Rambaut 2007) secondary cis-regulatory DNA motifs
Yap4 (also known as Cin5), a representative YRE-A-binding factor. This process involved generating mutants Yap1.R79K and Yap4.K252R, changing arginine to lysine in Yap1 and lysine to arginine in Yap4 (Methods). Next, Yap1.R79K binding and Yap4.K252R binding were assayed in vivo using ChIP-chip (Methods). Comparison of the top 50 promoters bound by Yap1.R79K with the top 50 promoters bound by wild-type Yap1 (as determined in Tan et al. 2008) showed that mutation of Yap1 significantly altered its preference for YRE-O and YRE-A sites (Fig. 1C; Fisher’s exact test P = 0.0002). Comparison of promoters bound by mutant and wild-type Yap4 also showed the predicted shift in binding preference (Fig. 1D; Fisher’s exact test P = 0.037). These results were not dependent on the number of promoters examined (Supplemental Fig. 2). Next, to assess the functional implications of changes in yAP-1 binding, we generated genome-wide mRNA expression profiles for each mutant in comparison to the unmutated parental strain (Methods). Both mutations, Yap1.R79K and Yap4.K252R, altered the expression of genes whose promoters were highly enriched for AP-1-binding sites (YRE-O and YRE-A, Fig. 1E,F; Supplemental Fig. 3). These genes were also enriched for Yap1.R79K and Yap4.K252R binding (P < 10!5 ), respectively. An apparent paradox: Candida AP-1 diverges at residue 12, but its targets are conserved Based on our observation that residue 12 affects binding of AP-1 paralogs in S. cerevisiae, we next asked whether changes in this residue could lead to divergent binding of AP-1 orthologs across species. We searched the yeast phylogeny (Wapinski et al. 2007) for AP-1 orthologs that were anomalous in their use of Arg 12 or Lys 12, suggesting lineage-specific mutation (Supplemental Fig. 4). Among TFs orthologous to Sc YAP1, we found that the Candida glabrata (Cg) ortholog CgAP1 diverges from other yeasts (Fig. A–C) due to the presence of lysine at residue 12, in contrast to other yeasts in its clade that possess an arginine. This CgAp1 amino acid substitution was confirmed by sequencing of genomic DNA from two independent Cg isolates, 2001HTU and NCCLS84 (Fig. 2B). We used ChIP-chip to determine whether this Lys 12 substitution had a functional effect on CgAp1 binding (Methods). To facilitate this assay, we tagged CgAp1 with the TAP epitope and designed a custom microarray tiling the Cg genome (Methods). As a control on both the TAP construct and the array design, we used ChIP-qPCR to successfully validate a panel of five randomly chosen Cg gene promoters that were determined to be bound by CgAp1 in the ChIP-chip experiment (Supplemental Fig. 5). We found that CgAp1 bound the promoters of a total of 114 genes, 90 of which had known orthologs in Sc (Methods). Comparison of these data with ChIP profiles for each of the AP-1 factors in Sc grown under the same treatment conditions (as determined in Tan et al. 2008) showed significant overlap between the targets of CgAp1 and ScYap1 (17 genes, P < 10!17). Overlap with other Sc AP-1 factors was less substantial (Fig. 3A). This pattern of overlap was reinforced by sequence analysis, in which phylogenetic clustering of AP-1 DNA-binding domains places CgAp1 definitively with ScYap1 and not with other Sc AP-1 sequences (Fig. 2C; Methods). We were therefore faced with the following conundrum: On the one hand, the CgAp1 sequence diverges from Yap1 orthologs at residue 12, suggesting a shift in DNA binding. On the other hand, the CgAp1-binding profile is quite specifically conserved with that of Yap1, calling into question the importance of residue 12 for sequence recognition. CgAp1 prefers YRE-A rather than YRE-O sites To investigate this apparent contradiction, we next turned to the gene promoters targeted by CgAp1 in the ChIP assay. Promoters targeted by CgAp1 showed a clear preference for YRE-A sites over YRE-O sites (49 vs. four promoters, respectively). This preference significantly differs from ScYap1, which prefers YRE-O over YRE-A (Fisher’s exact test P = 3.5 3 10!8 ; Fig. 3B, 21 vs. 12 promoters, respectively). This preference could not be attributed to threshold effects on binding-site calls, as direct comparison of motif scores confirmed a preference for YRE-A over YRE-O sites (Mann-Whitney U test, P = 0.0072). This preference was also observed via de novo motif search in these promoters (Fig. 3C) and even among the Cg orthologs of all ScYap1 targets (Q = 0.05). We further analyzed this cis-regulatory preference by examining the orthologs of genes targeted by both CgAp1 and ScYap1 across 20 sequenced yeast genomes (Wapinski et al. 2007). C. glabrata stood out clearly as the only species with enrichment for YRE-A sites (Fig. 3D). In contrast, the YRE-O site was enriched in all neighboring species in the yeast phylogeny, including S. cerevisiae and other sensu stricto species (S. paradoxus, S. mikatae, and S. bayanus) as well as the more diverged Saccharomyces castellii, Kluyveromyces waltii, Kluyveromyces lactis, Ashbya gosspyii, and Candida tropicalis. These results indicate that upstream DNA-binding motifs of CgAp1 targets have evolved from YRE-O to YRE-A (Fig. 3E). Such a switch may have also been accompanied by concordant changes in secondary cis-regulatory DNA motifs Figure 2. Evolution of the yAP-1 TFs. (A) CgAp1 possesses a lysine at residue 12 (CgAp1.46), while most other species possess an arginine. (B) Sequencing of CgAP1 in two unrelated isolates shows complete identity to the Cg reference genome. (C ) Phylogenetic clustering of all Sc and Cg AP-1 DNAbinding domains reveals that CgAp1 and ScYap1 co-cluster. Internal branch point numbers refer to the Bayesian posterior probability, a measure of confidence (Drummond and Rambaut 2007). Coevolu tion wi thin a transcrip tional ne twork Genome Research 3 www.genome.org Downloaded from genome.cshlp.org on November 3, 2010 - Published by Cold Spring Harbor Laboratory Press
ownloaded from genome. cshlp org on November 3, 2010- Published by Cold Spring Harbor Laboratory Press K al D with CgAp1 Overlap Significant 4001 005901 39 ScRap p=002 ScYap3I5 p=0.02 DNA-Binding Motif Preference nidulans CgAp Ancestral Yap1 ScYap2 SeRap ScRap? IrlcTAA TTCH →TSAT ScRap YRE-A Site TTACT TTAC T CATT ScRape has been rewired. (A)F by each yAP.] t site is enriched(star)among in other yeas Compensator i both trans and ) but not Cg(Q-1.0). The YRE-A site is enriched among these targets in Cg(star)but no maintain AP-1 binding (Supplemental material, Yap Cis-Regulatory Motifs Are Coincident Examination of the protein sequences of all AP-1 family with Those of Rtg3 and Aftl; Supplemental Fig. 6) and possible members across 20 available yeast genomes(Wapinski et al. 2007) functional divergence(Supplemental material, Divergence and suggests that mutations in residue 12 have occurred frequently Conservation of Yapl Function). The most plausible explanation is during AP-1 family evolution( Supplemental Fig 4). Interestingly, that these motifs have coevolved with a Lys 12 mutation in CgApl, we found that all yeasts possess at least one AP-l TF with Arg 12 with the result that this transcriptional system has retained reg.(Fig 4). In contrast, several yeasts lack AP-1 TFs with Lys 12, and ulatory control of the same set of target genes over evolutionary these species are the most evolutionarily diverged from Sc. These time results suggest that the common yeast AP-1 ancestor encoded ar- ginine and that the emergence of Tfs using lysine is a more recent n evolutionary innovation(Supplemental material, AP-1 Family Ancestry) Which mutation came first: the cis or trans? It is possible to envi- Within the Candida clade, several species(. tropicalis, C. sion two equally plausible scenarios(Fig. 3E): (1)An initial muta. albicans, C. parapsilosis, and Lodderomyces elongospors) have AP-1 tion in the Yapl tf provided selective pressure for subsequent families based exclusively on Arg 12, while others(C lusitaniae, cis-regulatory changes in Yapl targeted genes;(2)a change from Debaryomyces hanseni, and C guilliermondii (Fig 4)represent both YRE-O to YRE-A-binding site in key Yapl target(s) provided selec- Arg 12 and lys 12 across the AP-1 family. This suggests two equally tive pressure for a mutation in the Yapl TE. In either scenario, plausible scenarios for the emergence of lys 12 in yAP-1 TFs: (1) mutations in trans and cis may have been facilitated by other AP-1 Lysine emerged following the divergence of Yarrowia lipolyt family members. The large size and interconnectivity of the AP-1 from other hemi-ascomycetes, followed by a lineage-specific loss family may serve as a buffer for accumulation of cis and trans within the Candida clade. (2) Lysine emerged following the split of mutations, allowing for highly plastic evolution of the AP-1 reg. the Candida clade from the rest of the hemi-ascomycetes and ulatory network. In support of this hypothesis, several yAP-Is have emerged again within the Candida clade. In either scenario, been shown to bind each other along with common target genes a switch from arginine(coded by AGA or AGG)to lysine( coded by which might compensate for some loss in regulation by paralogs AAA or AAG)could be acc shed by a simple single base-pair (Tan et al. 2008) mutation Genome Research
(Supplemental material, Yap Cis-Regulatory Motifs Are Coincident with Those of Rtg3 and Aft1; Supplemental Fig. 6) and possible functional divergence (Supplemental material, Divergence and Conservation of Yap1 Function). The most plausible explanation is that these motifs have coevolved with a Lys 12 mutation in CgAp1, with the result that this transcriptional system has retained regulatory control of the same set of target genes over evolutionary time. Discussion Which mutation came first: the cis or trans? It is possible to envision two equally plausible scenarios (Fig. 3E): (1) An initial mutation in the Yap1 TF provided selective pressure for subsequent cis-regulatory changes in Yap1 targeted genes; (2) a change from YRE-O to YRE-A-binding site in key Yap1 target(s) provided selective pressure for a mutation in the Yap1 TF. In either scenario, mutations in trans and cis may have been facilitated by other AP-1 family members. The large size and interconnectivity of the AP-1 family may serve as a buffer for accumulation of cis and trans mutations, allowing for highly plastic evolution of the AP-1 regulatory network. In support of this hypothesis, several yAP-1s have been shown to bind each other along with common target genes which might compensate for some loss in regulation by paralogs (Tan et al. 2008). Examination of the protein sequences of all AP-1 family members across 20 available yeast genomes (Wapinski et al. 2007) suggests that mutations in residue 12 have occurred frequently during AP-1 family evolution (Supplemental Fig. 4). Interestingly, we found that all yeasts possess at least one AP-1 TF with Arg 12 (Fig. 4). In contrast, several yeasts lack AP-1 TFs with Lys 12, and these species are the most evolutionarily diverged from Sc. These results suggest that the common yeast AP-1 ancestor encoded arginine and that the emergence of TFs using lysine is a more recent evolutionary innovation (Supplemental material, AP-1 Family Ancestry). Within the Candida clade, several species (C. tropicalis, C. albicans, C. parapsilosis, and Lodderomyces elongosporus) have AP-1 families based exclusively on Arg 12, while others (C. lusitaniae, Debaryomyces hansenii, and C. guilliermondii) (Fig. 4) represent both Arg 12 and Lys 12 across the AP-1 family. This suggests two equally plausible scenarios for the emergence of Lys 12 in yAP-1 TFs: (1) Lysine emerged following the divergence of Yarrowia lipolytica from other hemi-ascomycetes, followed by a lineage-specific loss within the Candida clade. (2) Lysine emerged following the split of the Candida clade from the rest of the hemi-ascomycetes and emerged again within the Candida clade. In either scenario, a switch from arginine (coded by AGA or AGG) to lysine (coded by AAA or AAG) could be accomplished by a simple single base-pair mutation. Figure 3. The CgAp1 transcriptional network has been rewired. (A) For the promoters targeted by each yAP-1 transcription factor in Sc, the overlap with CgAp1 targets is shown (of 90 CgAp1 targets total). (B) CgAp1 prefers YRE-A-binding sites compared with ScYap1 (Fisher’s exact test). (C ) The CgAp1 DNA-binding motif (green) clusters with YRE-A rather than YRE-O motifs. (D) The YRE-O site is enriched (star) among common ScYap1 and CgAp1 targets in other yeasts (hypergeometric test, Q < 0.05), but not Cg (Q ; 1.0). The YRE-A site is enriched among these targets in Cg (star) but not other yeasts. (E ) Compensatory mutations in both trans and cis maintain AP-1 binding. Kuo e t al. 4 Genome Research www.genome.org Downloaded from genome.cshlp.org on November 3, 2010 - Published by Cold Spring Harbor Laboratory Press