Evolutionary History of the GH3 Family of Acyl Adenylases in Rosids

Plant Mol Biol DOI 10.1007/s11103-011-9776-y

Evolutionary history of the GH3 family of acyl adenylases in rosids

Rachel A. Okrent • Mary C. Wildermuth

Received: 8 October 2010 / Accepted: 10 April 2011 Ó Springer Science+Business Media B.V. 2011

Abstract GH3 amino acid conjugases have been identi- identified in poplar, grape, columbine, maize and rice fied in many plant and bacterial species. The evolution of suggesting descent from a common ancestral chromosome GH3 genes in plant species is explored using the sequenced dating to before the eudicot/monocot split. In addition, the rosids Arabidopsis, papaya, poplar, and grape. Analysis of clade containing PBS3 has undergone a unique expansion the sequenced non-rosid eudicots monkey flower and col- in Arabidopsis, with expression patterns for these genes umbine, the monocots maize and rice, as well as spikemoss consistent with specialized and evolving stress-responsive and moss is included to provide further insight into the functions. origin of GH3 clades. Comparison of co-linear genes in regions surrounding GH3 genes between species helps Keywords GH3 Rosids Phylogeny Synteny Acyl reconstruct the evolutionary history of the family. Com- adenylase Salicylic acid Phytohormone bining analysis of synteny with phylogenetics, gene expression and functional data redefines the Group III GH3 Abbreviations genes, of which AtGH3.12/PBS3, a regulator of stress- Compounds induced salicylic acid metabolism and plant defense, is a BTH 1,2,3-benzothiodiazole-7-carbothioic acid S-methyl member. Contrary to previous reports that restrict PBS3 to ester Arabidopsis and its close relatives, PBS3 syntelogs are IAA Indole-3-acetic acid JA Jasmonic acid Accession numbers: AtGH3.1, At2g14960; AtGH3.2, At4g37390; SA Salicylic acid AtGH3.3, At2g23170; AtGH3.4, At1g59500; AtGH3.5, At4g27260; AtGH3.6, At5g54510; AtGH3.7, At1g23160; AtGH3.8, At5g51470; Genes AtGH3.9, At2g47750; AtGH3.10, At4g03400; AtGH3.11, bZIP Basic-domain leucine-zipper At2g46370; AtGH3.12, At5g13320; AtGH3.13, At5g13350; AtGH3.14, At5g13360; AtGH3.15, At5g13370; AtGH3.16, ERF Ethylene response factor At5g13380; AtGH3.17, At1g28130; AtGH3.18, At1g48670; GDG1 GH3-like defense gene 1 AtGH3.19, At1g48660. GH3 Gretchen Hagen 3 ICS1 Isochorismate synthase 1 Electronic supplementary material The online version of this article (doi:10.1007/s11103-011-9776-y) contains supplementary JAR1 Jasmonic acid resistant 1 material, which is available to authorized users. PBS3 avrPphB susceptible 3 WIN3 HopW1-1-interacting 3 R. A. Okrent M. C. Wildermuth (&) Department of Plant and Microbial Biology, University Organisms of California, 221 Koshland Hall, Berkeley 94720, USA Ac Aquilegia coerulea (columbine) e-mail: [email protected] At Arabidopsis thaliana Present Address: Cp Carica papaya (papaya) R. A. Okrent Mg Mimulus guttatus (monkey flower) Department of Botany and Plant Pathology, Oregon State Os Oryza sativa (rice) University, Corvallis, OR 97331, USA 123 Plant Mol Biol

Pp Physcomitrella patens (moss) Gossypium hirsutum (cotton), all in the Eurosids II subc- Pt Populus trichocarpa (poplar) lade of the rosids superfamily of eudicotyledonous plants Sm Selaginella moellendorffii (spikemoss) (Terol et al. 2006). Substrate specificity of assayed proteins Vv Vitis vinifera (grape) tended to correspond to these phylogenetic relationships Zm Zea mays (maize) (Staswick et al. 2002, 2005). The JAR1 enzyme active on JA was placed in GH3 Group I (comprised of 2 GH3 genes Terms in A. thaliana), with GH3 enzymes active on IAA in Group ML Maximum likelihood II (8 AtGH3 genes), and enzymes including PBS3 active on MP Maximum parsimony neither of these compounds in Group III (9 AtGH3 genes). NJ Neighbor joining We are interested in the evolutionary history of the GH3 PID Percent identity family as a means of gaining insight into the evolved WGD Whole genome duplication functions of these enzymes in plants. As mentioned above, this family of enzymes catalyzes the amino acid conjugation of small molecules, including the hormones IAA and Introduction JA. In doing so, these GH3 proteins alter the activity of the hormone and its extensive impact on plant metabolism and GH3 (Gretchen Hagen 3) genes were originally identified physiology. For example, JAR1 catalyzes the conjugation in Glycine max (soybean) as responsive to the phytohor- of Ile to JA forming JA-Ile, the active form of the hormone, mone auxin (Hagen et al. 1984) and have since been resulting in the degradation of a JA repressor protein and identified in many plant species (Terol et al. 2006). Several the subsequent activation of downstream transcriptional Arabidopsis thaliana genes were identified in genetic responses (Chini et al. 2007). The function of the Group III screens for altered phytohormone-mediated responses to enzymes has remained more elusive, as a substrate for only auxin [e.g., DFL1 (Nakazawa et al. 2001)], or jasmonic one enzyme, PBS3, has been identified (Okrent et al. acid [e.g., JAR1 (Staswick et al. 1992)]. However, the 2009). Though PBS3 does not act directly on the phyto- molecular function of these genes remained unknown until hormone SA, its function is required for full activation of Staswick et al. identified structural similarity between the SA-dependent defense responses (Nobuta et al. 2007; A. thaliana GH3 and the firefly luciferase-like superfamily Jagadeeswaran et al. 2007; Lee et al. 2007). If the Group III of proteins (Staswick et al. 2002). The firefly luciferase- GH3 genes were only present in a small group of related like superfamily, also called the adenylate-forming super- species, it would suggest the encoded enzymes evolved a family, is a diverse group of enzymes that catalyzes the new function. Possibly, this function, e.g., acting on a addition of AMP to carboxyl groups on a wide variety of unique substrate, would be specifically required by these substrates. This family includes nonribosomal peptide species or confer a growth or reproductive benefit. synthetases, 4-coumarate-CoA ligases, acyl-CoA ligases, Herein, we explore the evolutionary history of the GH3 and oxidoreductases (Conti et al. 1996). These enzymes family, focusing particularly on Group III. The recent typically contain three conserved motifs that form a bind- sequencing of multiple plant genomes coupled with new ing pocket for AMP and the substrate (Chang et al. 1997). computational tools such as the CoGe suite of comparative Staswick et al. identified the three conserved motifs in the genomics programs (Lyons and Freeling 2008; Lyons et al. A. thaliana GH3 proteins. Furthermore, in vitro activity 2008) allows us to leverage information from whole gen- assays revealed that one, JAR1 (GH3.11), catalyzes the omes to infer descent from a common ancestral gene by addition of amino acids to the plant hormone jasmonic acid analyzing co-linearity of neighboring genes. We performed (Staswick et al. 2002) and that several others catalyze the our analysis focusing primarily on genome sequence data addition of amino acids to auxin (Staswick et al. 2005). from the rosids Arabidopsis thaliana and Arabidopsis PBS3 (GH3.12), among others, was not active on any of lyrata, order Brassicales; Carica papaya (papaya), order the phytohormone substrates tested (Staswick et al. 2002). Brassicales; Populus trichocarpa (poplar), order Mal- Our subsequent work determined that 4-substituted ben- pighiales; and Vitis vinifera (grape), order Vitales. For zoates serve as substrates of PBS3 (Okrent et al. 2009). comparison, we also used genome data from the asterid Previously published phylogenetic trees constructed Mimulus guttatus (monkey flower); order Lamilaes; the using distance methods divided plant GH3 proteins into basal eudicot Aquilegia coerulea (columbine), order Ran- three major clades, identified as Groups I, II, and III (Felten unculales; the monocot grasses Oryza sativa (rice) and Zea et al. 2009; Staswick et al. 2002; Terol et al. 2006). Groups mays (maize); the lycophyte Selaginella moellendorffii I and II contained genes from many more species than did (spikemoss); and the moss Physcomitrella patens (Fig. 1). Group III, which contained genes from only three species, This syntenic analysis is coupled with investigation of Arabidopsis thaliana, Brassica napus (rapeseed), and expression patterns of the GH3 genes and available 123 Plant Mol Biol

Populus trichocarpa 13 (Rensing et al. 2008)], Populus trichocarpa [JGI, v2 rosids (Tuskan et al. 2006)], Selaginella moellendorfﬁi (JGI v1), Arabidopsis thaliana 19 Vitis vinifera [v1, French-Italian Public Consortium for 9 eudicots Carica papaya 6 Grapevine Genome Characterization (Jaillon et al. 2007)], 3 and Zea mays cultivar B73 [Maize sequence.org v2 (Sch- Vitis vinifera 8 nable et al. 2009)]. The regions of potential synteny com- piled in (Lyons et al. 2008) were used as a starting point, Mimulus guttatus 10 modiﬁed with new genome sequence data, evaluated for

Aquilegia coerulea 6 accuracy, and expanded with additional analyses. The monocots V. vinifera genome, less subject to rearrangements, dupli- Zea mays 13 cations and gene loss than other genomes (Jaillon et al. 2007; Semon and Wolfe 2007), was used as a bridge Oryza sativa 13 between A. thaliana and the other genomes to identify

Selaginella moellendorffii 20 regions of co-linearity and possible synteny. It should be noted that genes from some of the sequenced plants have Physcomitrella patens 2 not yet been assigned to chromosomes (Phytozome 6.0). Paleohexaploidy Four Arabidopsis GH3 genes were used as seeds to Paleotetraploidy identify GH3 genes in the other eudicot genomes. The CDS Fig. 1 Plant phylogeny showing number of GH3 genes in modern sequences of PBS3 (AtGH3.12, At5g13320), DFL1 species and location of paleohexaploidy and paleotetraploidy events (AtGH3.6, At5g54510), JAR1 (AtGH3.11, At2g46370), and in lineages of interest. Figure adapted from Freeling (2009) with DFL2 (AtGH3.10; At4g03400) were used as BLAST seeds phylogenetic information from the Missouri Botanical Garden’s Angiosperm Phylogeny Project and Phytozome v6 against papaya, grape, poplar, monkey flower, and columbine in CoGe Blast. Similarly, Arabidopsis and rice GH3 functional information to gain insight on potential gene sequences were used to identify homologs in maize, and function. As detailed below, we find that contrary to past moss GH3 sequences were used to identify homologs in reports, Group III GH3 enzymes descended from a com- spikemoss. Additional BLASTN searches were performed mon ancestral chromosome dating to before the eudicot/ with the CDS sequence from each species against the gen- monocot split and are a sister taxa to the Group II IAA- ome of origin to find any other possible matches, and results conjugating enzymes. Furthermore, our analyses find the were checked against genes containing the keyword subsequent expansion of Group III GH3s to be consistent ‘‘GH3’’in Phytozome v6 (http://www.phytozome.net/) from with a role in response to (a)biotic stress. Our identification Department of Energy’s Joint Genome Institute and the of syntelogs (syntenic orthologs) of key Group I, II, and III Center for Integrative Genomics. Sequence length and gene members in agronomically-important species allows one to models were also retrieved from the Phytozome website and prioritize those genes for translational research. Moreover, compared to BLASTX results. The sequences were aligned analysis of conserved genes in syntenic regions may pro- using MUSCLE (Edgar 2004), visualized in JalView and vide insight on ancient and evolving GH3 syntelog func- evaluated for global alignment and presence of AMP-bind- tion(s), as we have explored for PBS3. ing motifs (Chang et al. 1997). As two divergent haplotypes of Selaginella were sequenced, a non-redundant set of sequences was com- Materials and methods posed using a gene set from JGI (http://www.genome.jgi.- psf.org/Selmo1.info.html) and by removing nearly identi- Identification of GH3 genes and syntenic regions cal sequences. Several sequences were found to be flanking GH3 genes incomplete. GSVIV00026990001 (VvGH3.7) was missing motif II and the protein sequence was shorter than expected The CoGe platform for comparison of genome sequences by approximately 200 amino acids based on comparison (http://www.synteny.cnr.berkeley.edu/CoGe/; Lyons and with other GH3 proteins, typically around 600 amino acids Freeling 2008) was used to identify regions of potential long. BLAST searches found peptides and ESTs containing synteny between Arabidopis thaliana (TAIR, v9) and the missing sequence information (ESTs gi 110368758 and Arabidopsis lyrata (JGI, v1) and other sequenced plant gi 110698541 and peptide gi 225454466). The corrected genomes, including Aquilegia coerulea (JGI, v1), Carica sequence length and exon number are shown in Table 1. papaya [v0.4, ASPGB draft genome (Ming et al. 2008)], Two papaya sequences, EVM prediction supercon- Mimulus guttatus (JGI, v1), Oryza sativa [TIGR v5, tig_1065.2 (CpGH3.3) and EVM prediction supercon- (Ouyang et al. 2006)], Physcomitrella patens [JGI v1.1, tig_9.204 (CpGH3.4) were both truncated due to missing 123 Plant Mol Biol sequence data. Some additional base pairs in the C-termi- (PhyML) and maximum parsimony (TNT) tree construc- nus of the p1065.2 gDNA sequence were identified as tion methods were used and MEGA 4.0 (Tamura et al. potentially coding, and added to the CDS sequence. 2007) was employed to visualize and annotate the phylo- However, this correction did not account for all of the genetic trees. Sequences missing AMP binding motifs or missing sequence at the C-terminus. No ESTs were iden- the highly conserved C-terminus were omitted from tified that contain the missing regions for either of the alignments and phylogenetic tree construction. DNA papaya sequences, so they remain incomplete. regions containing sequences related phylogenetically were tested for synteny using the GEvo tool of the CoGe Sequence alignment and phylogenetic tree construction browser, described above.

CDS and peptide sequences were aligned with MUSCLE Analysis of expression data (Edgar 2004) via the online phylogeny.fr platform (http:// www.phylogeny.fr; Dereeper et al. 2008) using default The expression patterns of AtGH3 genes were explored parameters. Distance (BioNJ), maximum likelihood using tools from Genevestigator (Zimmermann et al.

Table 1 GH3 genes in the rosids Populus trichocarpa, Vitis vinifera, and Carica papaya and their relationship to Arabidopsis thaliana genes through common position on an ancestral chromosome Namea Locusb Pos.c Protein Lengthb Exonsb Groupd At syntelog (s)d

PtGH3.1 POPTR_0007s10350 7 597 3 II AtGH3.2, AtGH3.3 PtGH3.2 POPTR_0009S09590 9 690 3 II AtGH3.1 PtGH3.3 POPTR_0001S30560 1 596 3 II AtGH3.1 PtGH3.4 POPTR_0013S14740 13 608 3 II – PtGH3.5 POPTR_0011S13330 11 611 3 II AtGH3.5, AtGH3.6 PtGH3.6 POPTR_0001S43990 1 611 3 II AtGH3.5, AtGH3.6 PtGH3.7 POPTR_0001S12850 1 606 4 III AtGH3.12, AtGH3.17 PtGH3.8 POPTR_0003S15970 3 594 5 III AtGH3.12, AtGH3.17 PtGH3.9 POPTR_0002S20790 2 596 4 III AtGH3.9 PtGH3.10 POPTR_0013S14050 13 595 4 I AtGH3.10 PtGH3.11 POPTR_0019S13450 19 595 4 I AtGH3.10 PtGH3.14 POPTR_0014S09120 14 576 4 I AtGH3.11 PtGH3.15 POPTR_0002S16960 2 576 4 I AtGH3.11 VvGH3.1 GSVIV00007718001 3 598 4 II AtGH3.1 VvGH3.2 GSVIV00019610001 7 600 3 II AtGH3.2, AtGH3.3 VvGH3.3 GSVIV00027472001 19 605 4 II AtGH3.5 AtGH3.6 VvGH3.4 GSVIV00026120001 12 588 5 II – VvGH3.5 GSVIV00027964001 7 596 4 III AtGH3.9 VvGH3.6 GSVIV00026000001 12 592 4 I AtGH3.10 VvGH3.7 GSVIV00026990001e 15 583 4 I AtGH3.11 VvGH3.8 GSVIV00006220001 1 593 5 III AtGH3.12, AtGH3.17 CpGH3.1 EVM supercontig_292.1 292 599 3 II AtGH3.1 CpGH3.2 EVM supercontig_6.74 6 599 3 II AtGH3.2 AtGH3.3 CpGH3.3 EVM supercontig_1065.2 1065 455f 3 II AtGH3.5 AtGH3.6 CpGH3.4 EVM supercontig_9.204 9 491f 3 III AtGH3.9 CpGH3.5 EVM supercontig_34.122 34 607 4 I AtGH3.10 CpGH3.6 EVM supercontig_1483.1 1483 591 4 I AtGH3.11

Syntelogs of PBS3 (AtGH3.12) are in bold a Pt names from Felten et. al (2009), PtGH3.15, Vv, Cp and Mg names ﬁrst described here b Locus names, protein length and exon number from Phytozome v5.0 c Position is the chromosome for Pt and Vv, and supercontig for Cp and Mg d Group and At syntelog from synteny analysis described in Materials and Methods e Gene model in Phytozome v5.0 corrected based on global alignment and EST data, see Materials and Methods f Sequencing error results in missing sequence

123 Plant Mol Biol

2004), NascArrays (Craigon et al. 2004), the eFP brower Results (Toufighi et al. 2005; Winter et al. 2007)(http://www.bar. utoronto.ca/efp/cgi-bin/efpWeb.cgi) and analysis of the Identification of GH3 genes relevant literature. Experiments in which GH3 genes were expressed in response to abiotic or biotic stress or hormone Potential orthologs of the AtGH3 genes were identified by treatments were identified in Genevestigator and the eFP BLASTN search using CoGe Blast with AtGH3, OsGH3, browser, and data downloaded. Most experiments analyzed and PpGH3 genes as seeds, aligned using MUSCLE (Edgar were from the AtGenExpress series (Goda et al. 2008). The 2004), visualized in JalView and evaluated for global NascArrays experiment reference numbers are shown in alignment and presence of AMP-binding motifs and the the corresponding data tables in the ‘‘Results’’ section. highly conserved C-terminal domain (see Materials and Values greater than 2-fold increase relative to control Methods). For the rosids, in addition to the 19 Arabidopsis experiments were reported and significance tested using thaliana GH3 genes, 6 GH3 genes were identified in Carica student’s t tests at a = 0.05 for experiments with three papaya,13inPopulus trichocarpa [one newly described replicates. Experiments with two replicates were manually gene plus 12 described in (Felten et al. 2009)], and 8 GH3 examined for reproducibility of experimental and control genes were identified in Vitis vinifera (Fig. 1, Table 1). samples. PtGH3 and VvGH3 expression patterns were For comparison, we also identified GH3 genes in non- analyzed using the Plant Expression Database [Plexdb, rosid eudicots Mimulus guttatus (6 GH3 genes) and Aqui- http://www.plexdb.org (Wise et al. 2007)]. legia coerulea (10 genes), as well as in the monocot grasses Zea mays (13 GH3 genes) and Oryza sativa [13 genes; previously described in (Jain et al. 2006)], and in Selagi- Analysis of transcription factor binding motifs nella moellendorffii (20 GH3 genes) and Physcomitrella patens [2 genes; previously described in (Bierfreund et al. Potential transcription factor binding sites 1 kb upstream of 2004) and (Ludwig-Muller et al. 2009)] (Fig. 1, Online AtGH3 transcriptional start sites were identified using the Resource 1). It should be noted that there may be additional Arabidopsis cis-regulatory element database (AtcisDB, M. guttatus and C. papaya GH3 genes, as the sequencing http://www.arabidopsis.med.ohio-state.edu/AtcisDB/; and annotation of those genomes are still incomplete. Molina and Grotewold 2005) of the Arabidopsis Gene Regulatory Information Server from Ohio State University. Loss and gain of GH3 genes in rosids Several of the tandemly duplicated genes have fewer than 1 kB between the next gene: AtGH3.13 (100 bp), Analysis of synteny is complicated by gene duplication, AtGH3.15 (288 bp), AtGH3.16 (711 bp) and AtGH3.18 either due to whole genome duplication (WGD) or local (989 bp). Promoter motifs associated with transcription duplication, gene loss or insertion (Lyons et al. 2008). Plant factor families of interest were tallied and summarized. genomes are heavily duplicated, with WGD events quite These plant transcription factors include WRKYs, basic- common in the evolutionary history of many plant lineages. domain leucine-zippers (bZIPs), MYBs, and dehydration Following episodes of genome duplication, selective gene response element binding proteins (DREBs) known to loss, called fractionation, is typically observed (Freeling mediate plant response to biotic stress (Singh et al. 2002). 2009). As shown in Fig. 1, the rosids all show evidence of Furthermore, ethylene responsive element binding factors a pre-rosid hexaploidy event with subsequent fractionation. (ERFs), auxin response factors (ARFs) and MYC2 tran- Though the exact classification of V. vinifera remains scription factors (Dombrecht et al. 2007) were included to uncertain, sequence comparisons of chloroplast genomes reflect responses mediated by the hormones ethylene, suggest that V. vinifera is a basal rosid (Jaillon et al. 2007; auxin, and jasmonate, respectively. Cis-acting regulatory Jansen et al. 2006) as depicted. Although A. thaliana has elements are defined as follows: WRKY transcription the smallest genome of any sequenced plant, two rounds of factors recognize the W-box: ttgact/c; bZIPs recognize WGD have occurred since its divergence from the primary actcat (ATB2/AtbZip53) and acacttg (DPBF1&2) motifs; rosid lineage (Blanc et al. 2003). One round of WGD has the MYB and MYB-like transcription factors bind MYB also occurred in the P. trichocarpa lineage. This can lead to (aaccaaac, taactaac) and aaatct (MYB-related CCA1) as many as four copies of A. thaliana and two of motifs; DREBs recognize tgccgacaa, gaccgacct, and aacc- P. trichocarpa for each V. vinifera or C. papaya gene. gacca motifs; ERFs bind the GCC-box (gccgcc), ARF1 Though Fig. 1 does not show a WGD event as part of binds tgtctc; and MYC2 binds the G-box (cacgtg), T/G-box M. guttatus lineage, several members of the Mimulus genus (cacgtt), and cacatg. have been shown to have evidence of polyploidy (Wu et al.

123 Plant Mol Biol

2007). Completion of the M. guttatus annotation will allow C. papaya did not share a region of synteny with PBS3; future resolution of this issue. however, this needs to be reexamined once the final com- As noted on Fig. 1, our analysis indicates that the pleted and annotated genome is available. Regions of common ancestor prior to the pre-rosid hexaploidy WGD M. guttatus showed evidence of synteny with the region event had 3 GH3 genes, leading to nine after the hexa- surrounding PBS3 (see Table 3), however, none of these ploidy WGD event, one of which was lost in the lineage M. guttatus regions contain a GH3 sequence. It remains prior to the divergence of grape. V. vinifera has retained all possible, however, that a GH3 sequence resides in the 8 GH3 genes, but the other rosids analyzed have lost missing region of M. guttatus scaffold 39 (open boxes in multiple genes (Table 2). Uniquely, A. thaliana lost 21 Table 3). As shown in Online Resource 2, Aquilegia genes but also gained 7 GH3 genes, due to whole genome coerulea does contain a syntenic region (mapped to scaf- duplication, fractionation, local duplication and transposi- fold 49) and GH3 gene, AcGH3.6. Similarly, a region of tion. Of particular interest, the genes gained are all in Oryza sativa chromosome 11 containing OsGH3.13 Group IIIA defined below. (Os11g32520) (Terol et al. 2006) and a region of Zea mays containing ZmGH3.13 demonstrate co-linearity of genes Identification of GH3 syntenic sets with the region containing Arabidopsis PBS3 (Online Resource 2). No syntenic region is detectable between Using the GEvo tool of the CoGe genome comparison S. moellendorffii or P. patens. browser (Lyons and Freeling 2008), eight GH3 syntenic A detailed comparison of the syntenic regions flanking sets were identified, incorporating ten genomic regions of PBS3 shows that the five AtGH3 genes in the Arabidopsis Arabidopsis thaliana that flank GH3 genes. These include chromosome 5 region correspond to a single gene in each several areas of A. thaliana chromosomes 1, 2, 4, and 5 and of the other genomes. This suggests that these additional account for 14 of the 19 A. thaliana GH3 (AtGH3) genes. Arabidopsis GH3 genes result from local duplication. Two The gene names, locus, chromosome, contig, or scaffold additional genes (At5g13330 and At5g13340, annotated as number, predicted protein length, number of exons, and an ERF/AP2 transcription factor family member and of Arabidopsis syntelog are shown in Table 1 for the rosids unknown function, respectively) are located between PBS3 and Online Resource 1 for the others. and the other four AtGH3 genes. One of these, the gene As an example, Fig. 2 shows part of the Arabidopsis encoding an ERF/AP2 transcription factor, also has syn- PBS3 (AtGH3.12/At5g13320) syntenic set including telogs in many of the species examined. In total, in addition approximately 40 kB of Arabidopsis chromosome 5 with to the GH3 genes, seven Arabidopsis genes in the chro- 100 kB of V. vinifera chromosome 1, 90 kB P. trichocarpa mosome 5 region and three in the chromosome 1 region chromosome 1, and 60 kB of P. trichocarpa chromosome display synteny to genetic regions of other species. These 3. The Arabidopsis chromosome 5 region contains PBS3 genes and their annotations are summarized in a slightly (AtGH3.12, At5g13320) and four other AtGH3 (AtGH3.13- abbreviated version for the eudicots in Table 3 and more 16) genes. Though not shown in Fig. 2,70kBofArabid- fully for all species examined in Online Resource 2. opsis chromosome 1 containing AtGH3.17 is also syntenic Five A. thaliana, one P. trichocarpa (PtGH3.4), one (see Table 3 and Online Resource 2). Of the rosids, only A. coerulea (AcGH3.1) and two O. sativa GH3 genes (OsGH3.7 and OsGH3.12) are not members of syntenic sets (Online Resources 1 and 3); there is no detectable co-linearity between surrounding genes and GH3 genes from other Table 2 Loss and gain of GH3 genes in each syntenic group since species. In addition, it is not possible to detect co-linearity divergence from common ancestor between chromosomal regions surrounding the S. mo- Syntenic group At Cp Pt Vv ellendorfii and P. patens genes and the other species studied (data not shown). Four of the five Arabidopsis genes not in a IA -3NC-1NC syntenic set (AtGH3.7, AtGH3.8, AtGH3.18, AtGH3.19) are IB -3NCNCNC in Group IIIA (defined below), which contains PBS3, and IIA1 -2NC-1NC can be explained by local duplication and gene insertion IIA2 -2NCNCNC (Fig. 3). AtGH3 genes, AtGH3.18 (At1g48670) and IIB1 -2NCNCNC AtGH3.19 (At1g48660), as well as a severely truncated GH3 IIB2 -4 -1 -1NC gene (At1g48690), likely arose from insertion and duplica- IIIA 22/17 21 21NCtion (Fig. 3) and there are several retrotransposons nearby IIIB -3 -1NCNC(e.g. At1g48680). AtGH3.7 (At1g23160), most similar in Total 19 5 12 8 protein sequence to PBS3, is present in A. lyrata but not Group in bold contains PBS3. NC no change papaya, suggesting that it was inserted into an ancestor of 123 Plant Mol Biol

PBS3 AtGH3.13 AtGH3.15 At AtGH3.12 AtGH3.14 AtGH3.16

Pt 1

Pt 2

Fig. 2 Synteny between the region of Arabidopsis chromosome 5 areas of similarity from BLASTz between genomic regions are shown surrounding PBS3 and grape and poplar chromosomes. A screenshot as colored blocks above or below the gene models. Each pairwise of the BLASTz output from the CoGe browser is shown. Each large comparison is shown in a different color. Brown and pink lines are horizontal bar represents one genomic region, with the dashed line drawn between similar regions of At and the other species. Lines dividing the top (50 on left) and bottom (50 on right) strand. The connecting other pairwise comparisons were omitted for clarity. The genome origin is indicated on the right (At Arabidopsis thaliana,Vv analysis can be regenerated at http://www.genomevolution.org/r/1d9i. Vitis vinifera,PtPopulus trichocarpa). The colored arrows represent The 70 kB syntenic region of Arabidopsis chromosome 1 containing gene models: green are CDS, blue are RNA, and gray are introns. The AtGH3.17 is not shown here but is included in Table 3

A. thaliana before the divergence from A. lyrata but after neighbor joining (BioNJ) algorithms. The tree found to best papaya (data not shown). The region surrounding AtGH3.8 correspond to the syntenic relationships identified above (At5g51470) contains many stress response genes, including was constructed using a multiple sequence alignment PBS2, also known as RAR1, which was identified in the same curated using Gblocks with relaxed settings and PhyML. mutant screen for altered disease resistance as PBS3 (War- This phylogenetic tree comprised of the eudicot GH3 ren et al. 1999) (data not shown, can be recapitulated sequences is shown in Fig. 4, with the complete phyloge- http://www.genomevolution.org/r/37d). Regions of V. vinif- netic tree including eudicot, monocot, moss and spikemoss era chromosome 16 and P. trichocarpa chromosomes 12 sequences provided as Online Resource 3. and 15 display co-linear genes with the region around The eight sets of GH3 syntenic genes are distributed AtGH3.8 (At5g51470), although no GH3 gene is present, between Groups I, II, and III. The proteins from all species suggesting that AtGH3.8 was inserted at this site. Though other than S. moellendorffii and P. patens separate into AtGH3.4 (At1g59500) is grouped with the IIA GH3 proteins these three Groups in the phylogenetic trees, with each (below), synteny was ambiguous as only one other gene in syntenic set containing one V. vinifera protein. Each of the the surrounding region matched genes from other species. two Group I sets contains one VvGH3, AtGH3 (AtGH3.11/ JAR1 in Set IA and AtGH3.10/DFL2 in Set IB), CpGH3, Classification of GH3 proteins into groups based MgGH3, and AcGH3 protein and two poplar proteins, with on synteny and phylogenetic relationships Set IA also including monocot grass syntelogs. Group II contains four VvGH3 proteins, each of which Corrected GH3 protein sequences were aligned using corresponds to a syntenic subset of sequences (IIA1, IIA2, MUSCLE (Edgar 2004), and curated by removing gaps IIB1, and IIB2). Within IIB, IIB1 contains eight proteins manually or using Gblocks (Castresana 2000) the phylog- including AtGH3.5/WES1 and AtGH3.6/DFL1, other rosid eny.fr server (Dereeper et al. 2008) as described in proteins, as well as proteins from Mimulus and Aguilegia, ‘‘Materials and methods’’. Alignments were constructed while IIB2 is comprised only of VvGH3.4 and PtGH3.4. and clustered into phylogenetic trees using maximum Within subgroup IIA, the sequences are more evenly dis- parsimony (TNT), maximum likelihood (PhyML), and tributed between IIA1 and IIA2 syntentic sets, with

123 123

Table 3 Comparison of syntenic regions including PBS3 (AtGH3.12, At5g13320) At locus 1 At locus 2 Vv locus 1 Pt locus 1 Pt locus 2 Mg locus 1 Mg locus 2 Mg locus 3 Annotation Ch 5 Ch 1 Ch 1 Ch 1 Ch 3 scaffold 39 scaffold 7 scaffold 54

At5g13250 At1g28080 None None None None None Unknown, plastid At5G13320 At1g28130 GSVIV00006220001 POPTR_0001s12850 POPTR_0003s15970 None None PBS3, GH3 family protein (AtGH3.12) (AtGH3.17) (VvGH3.8) (PtGH3.7) (PtGH3.8) At5g13330 At1g28160 GSVIV00006201001 POPTR_0001s12820 POPTR_0003s15940 None None ERF/AP2 transcription factor At5g13340 None None None None None None Unknown At5g13350 None None None None None None GH3 family protein At5g13360 None None None None None None GH3 family protein At5g13370 None None None None None None GH3 family protein At5g13380 None None None None None None GH3 family protein At5g13390 None GSVIV00006219001 POPTR_0001s12860 POPTR_0003s15980 mgf012815m None None NEF1, chloroplast At5g13400 None GSVIV00006214001 POPTR_0001s12890 None mgf016173m None None Proton-dependent oligopeptide transport, plastid None None GSVIV00006211001 POPTR_0001s12910 POPTR_0003s16010 mgf08773m None None Cytochrome P450 At5g13410 None GSVIV00006210001 POPTR_0001s12920 None mgf015787m None None Peptidyl-prolyl cis–trans isomerase, chloroplast At5g13420 None GSVIV00006209001 POPTR_0001s12930 POPTR_0003s16030 mgf012100m None None Transaldolase, chloroplast None None GSVIV00006208001 None None None mgf004556m mgf014939m Unknown None None GSVIV00006206001 None None None mgf015946m mgf014306m, Unknown mgf000651m, mgf008051m At5g13430, None GSVIV00006205001 POPTR_0001s12960 POPTR_0003s16060 mgf004959m None None Ubiquinol-cytochrome C At5g13440 reductase mitochondrion None At1G28140 GSVIV00006202001 None None mgf003850m None None Unknown, chloroplast ln o Biol Mol Plant Plant Mol Biol

Syntenic set: (At2g47750) IIIA IIIB

(At1g28130) (At5g13360) (At5g13370) (At5g13380) local duplication (At5g13350) (At1g48670) (At1g48660) local duplication (At5g51470) insertion

(At1g23160) insertion

Fig. 3 Maximum likelihood phylogenetic tree of Group III rosid the same symbol represent a set with evidence of synteny. Sequences GH3 proteins, with evidence for expansion in Arabidopsis Group IIIA are from the rosid species At Arabidopsis thaliana,CpCarica detailed. The tree was constructed using PhyML on multiple sequence papaya,PtPopulus trichocarpa, and Vv Vitis vinifera. PBS3 is alignments from MUSCLE curated using Gblocks with reliability of highlighted internal branch length tested using the aLRT method. Branches with

Fig. 4 Maximum likelihood phylogenetic tree of eudicot III GH3 proteins. The tree was constructed using PhyML on A multiple sequence alignments from MUSCLE curated using Gblocks with reliability of B internal branch length tested A using the aLRT method. Branches with the same symbol represent a set with evidence of synteny. Sequences are from the species Aq Aquilegia coerulea, At Arabidopsis thaliana, Cp Carica papaya,MgMimulus guttatus,PtPopulus I trichocarpa, and Vv Vitis vinifera. PBS3 is highlighted B

Syntenic sets: IA IB B IIA1 IIA2 II IIB1 IIB2 A IIIA IIIB

AtGH3.1 and AtGH3.3 in IIA1 and AtGH3.2/YDK1 and Resource 3). If MgGH3.3 is misassigned and should be AtGH3.3 in IIA1 and AtGH3.1 in IIA2. As the evolu- placed in IIA1 instead of IIA2, then a duplication of the tionary distance of the species increases, it is increasingly Group II sequences after the divergence of the rosids would difﬁcult to distinguish subsets of IIA and IIB. Though we explain the presence of non-rosid species in only one did assign subsets of IIA and IIB for Mimulus and subset of IIA (i.e. IIA1) and IIB (i.e. IIB1). Aguilegia, there is less certainty about this assignment than Group III contains two VvGH3 proteins, corresponding for the rosids. For the monocots, we could not parse out to two syntenic sets, both of which contain proteins from these subsets and differentiate only IIA from IIB (Online species other than rosids. Set IIIA, contains the Arabidopsis

123 Plant Mol Biol proteins PBS3 (AtGH3.12) and AtGH3.13-16 on chromo- susceptible to pathogens such as Botrytis cinerea that some 5 and AtGH3.17 on chromosome 1. PBS3 syntelogs activate ET/JA-dependent defenses (Ferrari et al. 2003). in other species include VvGH3.8, PtGH3.7 and PtGH3.8, AtGH3.10/DFL2 is in a distinct syntenic set from JAR1 AcGH3.6, OsGH3.13 and ZmGH3.13 (Online Resources 2 and does not appear to function in JA signaling and and 3). Set IIIB, contains AtGH3.9 and a variety of other response (Fig. 5). Instead, it mediates red light-specific eudicot proteins. AtGH3.17 of Set IIIA and AtGH3.9 of Set hypocotyl elongation with its expression controlled by light IIIB had previously been classified with Group II proteins (Takase et al. 2003). Interestingly, jar1 mutants also (Staswick et al. 2005). exhibit far red light insensitivity (Hsieh et al. 2000). Nei- ther Group I Arabidopsis gene contains cis-acting regula- Analysis of GH3 functional data tory elements bound by ARFs, ERFs, or dehydration response element binding proteins (DREBs) in their 1 kb Examination of expression data and transcription factor promoters (Online Resource 4). However, they do contain binding motifs in gene promoters complements biochemi- other biotic-stress associated cis-acting regulatory elements cal and phenotypic data and can help provide insight into including those bound by MYC2 transcription factors. Of gene function. For example, the first GH3 genes were particular interest, the promoter of JAR1 contains 4 MYC2 identified as inducible by auxin (Hagen et al. 1984), years binding sites. Proteolysis of the JAZ family of JA repres- before their function was known. Subsequent analysis sors is mediated by the JA-Ile conjugate whose formation revealed GH3 enzymes that catalyze the conjugation of is catalyzed by JAR1. The proteolysis of the JAZ repressor amino acids to the auxin IAA regulating IAA activity and then allows for downstream JA-associated gene expression function (Staswick et al. 2005). Most of the available through MYC2 transcription factors (Dombrecht et al. functional data is for the Arabidopsis GH3 genes/proteins, 2007). the focus of our functional analysis. An analysis of publicly available expression data Arabidopsis GH3s induced by IAA are in Group II showed the Arabidopsis Group III GH3 genes tend to have (Fig. 5). Though all Group II AtGH3 genes are induced by higher expression then those in Group I and II, excluding IAA, only two of the six analyzed contain a cis-acting AtGH3.13 and 16, whose expression has not been observed regulatory element bound by auxin responsive transcription (Figs. 5 and 6; Online Resource 5). Two of the seven tested factors (ARFs). However, on a percentage basis Group II and expressed Group III A. thaliana GH3 genes are promoters are more likely to contain at least one cis-acting induced by phytohormone treatment (Fig. 5) with five of regulatory element bound by an ARF at 33%, compared the seven induced by pathogen or abiotic stress (Fig. 6). with 0 and 20% for Groups I and III, respectively (Online Group III GH3s tend to contain cis-acting regulatory ele- Resource 4). It should be noted that AtGH3.17 and ments associated with (a)biotic stress response in their AtGH3.9, which we reclassify as Group III above, are not promoters (Online Resource 4). It is interesting that motifs induced by IAA (Fig. 5.) We find the Group II GH3s are bound by DREBs are only present in the promoters of the also induced by (a)biotic stress including pathogens Group III genes AtGH3.12/PBS3, AtGH3.7 (the most (Fig. 6) with the exception of the lowly expressed similar A. thaliana protein to PBS3), and AtGH3.15.In AtGH3.1. In addition to Group II promoters being enriched addition, PBS3 and AtGH3.7 are induced by osmotic stress in cis-acting regulatory elements bound by ARFs, they are (Fig. 6); AtGH3.15 does not have a gene-specific probe on also enriched for regulatory elements bound by ethylene the ATH1 array. In terms of enzymatic activity, AtGH3.9 responsive binding factors (ERFs) known to regulate and 17, now classified into distinct syntenic Group III sets response to (a)biotic stress (Gutterson and Reuber 2004). are active on IAA, though their expression is not induced Group II Arabidopsis GH3 proteins are also active on IAA, by it [Fig. 5 (Khan and Stone 2007; Staswick et al. 2005)]. with mutants in these genes resulting in an IAA phenotype AtGH3.12 (PBS3), a syntelog of AtGH3.17, is not active when tested (Fig. 5). Of note, AtGH3.5/WES1 is active on on IAA, but on 4-substituted benzoates (Okrent et al. both IAA and salicylic acid (2-hydroxybenzoate) (Staswick 2009). Its expression is induced by SA [Fig. 5 and (Jaga- et al. 2002; Park et al. 2007b; Zhang et al. 2007). deeswaran et al. 2007)], with mutants exhibiting compro- In contrast to the Arabidopsis Group II genes, which are mised SA accumulation and pathogen resistance (Okrent all induced by IAA and encode proteins active on IAA, the et al. 2009; Lee et al. 2007; Jagadeeswaran et al. 2007). two Group I genes AtGH3.11/JAR1 and AtGH3.10/DFL2 Substrates for other Arabidopsis Group III members remain have minimal unifying functional data. Surprisingly, unknown. though JAR1 is active on JA (Staswick and Tiryaki 2004), In addition to the functional data discussed above for JAR1 is not induced by JA, other hormones, or in response Arabidopsis GH3 genes, data for P. trichocarpa, V. vinif- to a diverse set of pathogens (Figs. 5 and 6). jar1 mutants era, and O. sativa provide further evidence for selected do however exhibit JA-associated phenotypes and are more biotic stress induction of Group II and III GH3 members. 123 Plant Mol Biol

Induced Activity Phenotype

JA IAA SA JA IAA B JA IAA SA

Fig. 5 Maximum Likelihood phylogenetic tree of the Arabidopsis for activity data are (Staswick and Tiryaki 2004; Okrent et al. 2009). GH3 proteins showing induction by phytohormones, enzyme activity, Shaded boxes in the ‘‘Phenotype’’ column indicate that plants with and mutant phenotype. Symbols in branches are as in Fig. 4.JAis mutations in the genes have altered signaling through the phytohor- jasmonic acid, IAA is indole-3-acetic acid, SA is salicylic acid, B is mone indicated, with black shaded boxes moderate/strong and gray as benzoates. In the ‘‘Induced’’ column, gray shaded boxes correspond weak phenotype. References for phenotypic data are as follows: to C2-fold increase in expression compared to control treatment, AtGH3.1, (Staswick et al. 2005); AtGH3.2 (YDK1) (Staswick et al. black shaded boxes correspond to C10-fold expression compared to 2005: Takase et al. 2004); AtGH3.5 (WES1) (Staswick et al. 2005; control treatment. Boxes with dashed line were not tested or did not Park et al. 2007b; Zhang et al. 2007); AtGH3.6 (DFL1) (Nakazawa have gene speciﬁc data on array. PBS3 is highlighted. Microarray et al. 2001); AtGH3.9 (Khan and Stone 2007); AtGH3.11 (JAR1) data from NASCArrays set 174 (JA), 175 (IAA), 192 (SA) and 392 (Staswick et al. 1992); AtGH3.12 (PBS3) (Nobuta et al. 2007; (SA-analogue BTH). Shaded boxes in the ‘‘Activity’’ column indicate Jagadeeswaran et al. 2007; Lee et al. 2007); AtGH3.17 (Staswick that the enzyme is active on the corresponding substrate. References et al. 2005; Khan and Stone 2007)

For example, in poplar (Populus tremula 9 Populus alba) 2009). Furthermore, the IIA genes PtGH3.1 and PtGH3.2 tree roots colonized by the ectomycorrhizal fungus (EMF) were induced in EMF-colonized tree roots (Felten et al. Laccaria bicolor, the Group III PBS3 syntelogs PtGH3.7, 2009) and VvGH3.2 (probeset: 1610880_s_at) expression and Pt.GH3.8 were induced (Felten et al. 2009). In con- was elevated in response to infection with Bois Noir phy- trast, PtGH3.7 (probeset PtpAffx.140928.1.A1_at) and toplasma (Albertazzi et al. 2009). PtGH3.8 (probeset PtpAffx.210014.1.S1_at) were not signiﬁcantly elevated in response to infection with the Melampsora rust fungus (Plant Expression Database). No Discussion probesets were identiﬁed for the grape Group III members the PBS3 syntelog VvGH3.8 or the AtGH3.9 syntelog GH3 phylogeny VvGH3.5; however, the PBS3 syntelog from rice, OsGH3.13, is active on IAA, highly upregulated during A careful analysis of sequence data reconciled with syn- drought conditions and to a lesser extent, treatment with tenic analysis indicates that GH3 phylogeny is not as clear- the phytohormones IAA, SA and ABA, and confers cut as previously reported. Previous analyses of the GH3 enhanced drought tolerance when overexpressed in rice gene family relied on global sequence similarities (Stas- (Zhang et al. 2009). wick et al. 2005; Felten et al. 2009; Terol et al. 2006), For Group II, OsGH3.8 (in IIA) is upregulated following which are not necessarily indicative of true evolutionary infection with the bacterial pathogen Xanthomonas oryzae descent. The increasing number of sequenced genomes and pv oryzae and overexpresion of OsGH3.8 in rice is corre- new comparative genomic tools makes it possible to use lated with increased levels of IAA-Asp conjugates and synteny to evaluate phylogenetic trees. For example, Jun enhanced resistance to Xanthomonas (Ding et al. 2008). et al. recently evaluated the use of local synteny for iden- Similarly, constitutive expression of OsGH3.1 (in IIB) tifying orthologous genes in mammals, and found it quite alters auxin homeostasis and enhances resistance to the effective, particularly in cases of local duplication and fungal pathogen Magnaporthe grisea (Domingo et al. transposition (Jun et al. 2009). We used the syntenic data to 123 Plant Mol Biol

Induced Phenotype nonhost vir avr Abiotic stress P. infestans Pst Pst Psp Biotic B. cinera

Fig. 6 Maximum likelihood phylogenetic tree of Arabidopsis GH3 phaseicola (120), Botrytis cinerea (167), Phytopthora infestans (123), proteins and expression in response to biotic and abiotic stress. seedlings treated with cold/osmotic/salt/drought (138–141). For the Symbols in branches correspond to syntenic set as in Fig. 4. For the ‘‘Phenotype’’ column, shaded boxes indicate phenotypes have been ‘‘Induced’’ column, black shaded boxes correspond to C2-fold observed in mutants in response to biotic stress. Assessment of abiotic expression compared to control treatment and gray shaded boxes stress response has been minimal so it has not been included here. correspond to C10-fold increase in expression compared to control References are as follows: AtGH3.5 (WES1; Park et al. 2007a, b; treatment at a = 0.05. Boxes with dashed line were not tested or did Zhang et al. 2007); AtGH3.6 (DFL1; Zhang et al. 2007); AtGH3.11 not have gene-specific data on array. PBS3 is highlighted. Microarray (JAR1; Ferrari et al. 2003); AtGH3.12 (PBS3, GDG1, WIN3; Nobuta data is from adult leaves unless otherwise noted treated as indicated et al. 2007; Jagadeeswaran et al. 2007; Lee et al. 2007) from the following NASCArrays datasets: virulent, avirulent Pseu- domonas syringae pv. tomato and nonhost Pseudomonas syringae pv. guide the choice of phylogenetic methods in order to best literature over the ‘‘best’’ method for phylogenetic analysis, understand how PBS3 is related to the other GH3 genes. often defined as how much a method can tolerate violations Phylogenetic trees were constructed using neighbor-joining of its assumptions (Huelsenbeck and Hillis 1993). For (NJ), maximum parsimony (MP), and ML methods with example, there is a contentious debate over the relative various alignment curation methods. merits of ML particularly when different positions in a The ML method is a discrete data method that begins sequence evolve over different rates (Steel 2005). Many with a model of rates of evolutionary change and alters the experts suggest constructing phylogenetic trees using model until in fits the observed data (Mount 2004). This is multiple methods and evaluating them carefully (Hall contrasted with distance methods such as NJ for which the 2005; Mount 2004; Thornton and Kolaczkowski 2005). percentage of aligned positions that differ between two When available, addition of data from syntenic analysis sequences is computed pair-wise for all sequences in an can help determine the choice of phylogenetic method. We alignment and the values arranged so that sequences with found the tree constructed via ML best fit the data from lower difference scores are closer together on the tree syntenic analysis and was supported by the available (Mount 2004). Distances methods tend to be favored by functional data. molecular biologists as they are straightforward and com- As shown in the eudicot GH3 ML phylogenetic tree in putationally efficient. The drawback of distance methods is Fig. 4, there are two clades with high statistical support. One that they can be misleading when using an incorrect evo- of these is Group I, which in turn forms two subclades: one lutionary model (Huelsenbeck and Hillis 1993). While ML with JAR1 and its syntelogs (set IA) and one with DFL2 and methods are traditionally computationally intensive, new its syntelogs (set IB). Set IA predates the moncot/eudicot split algorithms, such as that employed by PhyML (Guindon whereas set IB contains only eudicots (Online Resource 3). and Gascuel 2003) reduce calculation time sufficiently to As discussed in Results, JAR1 is active on JA and mediates allow for routine use. There is considerable argument in the JA-dependent developmental and (a)biotic stress responses,

123 Plant Mol Biol though it is not induced by JA nor (a)biotic stresses tested preference has routinely been assessed only for (Figs. 5 and 6). Our analysis identifies JAR1 syntelogs in 2-hydroxybenzoate (SA). Indeed, though PBS3 is not agronomically important species including grape, poplar, active on SA, it is active on a series of other related rice, and maize as potential targets for enhanced disease benzoates with a strong preference for 4-substituted ben- resistance. By contrast, DFL2 is not active on JA or involved zoates such as 4-HBA (Okrent et al. 2009). Additional in JA-associated responses (Figs. 5 and 6). Instead it mediates functional analyses of the Group III (and Group II) GH3 red light-specific hypocotyl elongation with its expression members including high throughput assays [described in controlled by light (Takase et al. 2003, 2004)anditssyntelogs (Okrent et al. 2009)] to assess GH3 activity on a variety present only in the eudicots. of substrates combined with a comprehensive ancestral The second major clade contains the rest of the analysis as more genomes are sequenced should eventu- sequences, which divides into two subclades, referred to as ally allow for a fuller understanding of the ancestral Group II and Group III. Group II contains two major sets function(s) and evolution of these proteins. Importantly, (IIA and IIB) which precede the moncot/eudicot split. As the early emergence of GH3 proteins active on benzoates discussed in the Results, Group II members where tested is suggested by the ability of Lemna paucicostata (an are induced by the growth phytohormone IAA, active on early diverging monocot) to produce benzoyl-Asp (Suzuki IAA, with mutants exhibiting auxin-associated phenotypes et al. 1988) and the detection of 4-HBA-Glu and p-cou- (e.g. Figs. 5 and 6). In addition, the promoters of the maroyl-Glu in the hornwort Anthoceros agrestis (Tren- Arabidopsis GH3 Group II member genes are enriched in nheuser et al. 1994). ARF- and ERF-binding motifs (Online Resource 4) consistent with their induction by IAA and for some, by Function of PBS3 and its syntelogs pathogens (Figs. 5 and 6). Group II GH3 proteins in other eudicot and moncot species are also induced by pathogen Contrary to previous reports that restrict PBS3 to Ara- and exhibit both auxin and altered susceptibility pheno- bidopsis and its close relatives, we identify PBS3 synte- types (see ‘‘Results’’). logs in poplar, grape, columbine, maize and rice Group III sequences consist of set IIIA (PBS3 and its suggesting descent from a common ancestral chromosome syntelogs, including AtGH3.17 which is active on IAA) and dating to before the eudicot/monocot split. Furthermore, set IIIB (AtGH3.9 which is active on IAA and its syntelogs). our analysis of co-linear genes found a syntenic relation- As discussed earlier, AtGH3.9 and AtGH3.17 and their ship between the Arabidopsis PBS3/AtGH3.12 gene and orthologs had been previously classified as Group II proteins AtGH3.17 that was obscured by sequence similarity alone. though they did group separately from the other Group II The PBS3 syntelogs for which expression data exist are proteins (Staswick et al. 2005; Terol et al. 2006; Felten et al. induced by biotic interactions and/or abiotic stress, with 2009; Liu et al. 2005). Similar to Group II GH3 genes, Group the exception of some of the recently acquired Arabidopis III genes are often induced by (a)biotic stress and altered genes (see ‘‘Results’’). As mentioned earlier, AtPBS3 is expression can result in (a)biotic stress phenotypes [i.e. At- active on 4-substituted benzoates while AtGH3.17 and PBS3 (Nobuta et al. 2007; Lee et al. 2007; Jagadeeswaran OsGH3.13 are active on IAA. However, these latter et al. 2007) and OsGH3.13 (Zhang et al. 2009)]. enzymes were not tested for activity on 4-HBA. Loss of The most parsimonious explanation for the presence of PBS3 function alters benzoate metabolism, with a sub- IAA- conjugating enzymes in both Group II and III is that stantial impact on total SA accumulation and disease the ancestral gene encoded an enzyme that used IAA as a resistance (Nobuta et al. 2007; Jagadeeswaran et al. 2007; substrate. Indeed, both rice Group II and III members Lee et al. 2007). Enhanced expression of OsGH3.13/TLD1 have been found to be active on IAA (Zhang et al. 2009; alters plant architecture, IAA homeostasis, and enhances Ding et al. 2008; Domingo et al. 2009). However, as drought tolerance (Zhang et al. 2009). However, a TLD1 described in the Results, genes in both Group II and III loss of function mutant displayed no obvious growth or can be induced by SA and are active on benzoates [e.g. drought phenotypes, presumably due to the compensatory Group II AtGH3.5 is active on SA (2-HBA) and Group III action of other GH3s (Zhang et al. 2009). Similarly, an PBS3/AtGH3.12 is active on 4-HBA]. IAA-amino acid Arabidopsis gh3.17 mutant exhibited only very minor conjugates and their function in regulating auxin auxin-associated phenotypes; it was not tested for altered homeostasis has been a subject of long-standing investi- (a)biotic stress resistance (Staswick et al. 2005). Since gation [reviewed in (Woodward and Bartel 2005)]. pbs3 mutants exhibit substantial defects in total SA However, despite the fact that a variety of benzoate and accumulation and disease resistance; benzoate metabolism cinnamate amino acid conjugates have been detected in is likely to be the primary target of PBS3 and perhaps also plants (e.g. Suzuki et al. 1988; Bourne et al. 1991; of its syntelogs. Cross-talk between SA, auxin, and the Trennheuser et al. 1994), GH3 protein substrate drought-induced phytohormone ABA [e.g. (Park et al. 123 Plant Mol Biol

2007b; Zhang et al. 2009)], might then explain the phe- susceptibility and Arabidopsis plants with altered PBS3 notypes observed when OsGH3.13/TLD1 is overexpressed. expression have not been tested for drought tolerance. In any An examination of the predicted or known function of case, it is clear that PBS3 and its syntelogs can mediate conserved genes in the PBS3 syntenic regions (Online (a)biotic stress independent of flowering, with this role Resource 2) may provide additional insight into the ancient supported by the (a)biotic stress-induction and, where tested, and perhaps conserved function of the PBS3 syntelogs. function of its conserved neighbors NEF1 and RAP2.6L. Though benzoates are produced in the plastid, there is no evidence that PBS3 is plastid-localized. However, many of Expansion of the clade containing PBS3 the genes in the PBS3 syntenic region are known or pre- in Arabidoposis dicted to be in the plastid (Online Resource 2). NEF1 (At5g13390) syntelogs are present in most species exam- Of the rosids analyzed, A. thaliana was unique in that it not ined including eudicots and monocots. NEF1 is plastid- only lost GH3 genes but also gained seven GH3 genes (Table localized and involved in exine formation of pollen 2). Intriguingly, the genes gained in Arabidopsis all are in (Ariizumi et al. 2004). The sculptured wall exine consists Group IIIA which contains PBS3. AtGH3.7 and AtGH3.8 of phenols and fatty acid derivatives and plays an important were inserted into their location in the genome, as evidenced role in protecting the pollen from various (a)biotic stresses by corresponding regions in papaya, grape, and poplar but no (in Ariizumi et al. 2004). Because NEF1 is expressed in GH3 gene (data not shown). AtGH3.13-16 (syntelogs of flowers, and benzoates including SA are known to impact PBS3) were locally duplicated, likely from the ancestral the induction of flowering (Martinez et al. 2004), pollina- gene of PBS3 as the proteins encoded by the four duplicated tion strategy (e.g. to act as specific pollinator attractants) genes are more similar to each other than they are to PBS3 and defense (e.g. Dudareva and Pichersky 2000), we (data not shown). The genes AtGH3.18 and AtGH3.19 rep- examined whether PBS3 and the other Arabidopsis resent a tandem duplication following insertion. All of these genes in the PBS3 syntenic regions are expressed in rearrangements and insertions suggest a rapidly evolving flowers. With the exception of the duplicated At5g13430 group of genes. Researchers have shown that genes that gene At5g13440, all Arabidopsis genes in the syntenic respond to (a)biotic stress are overrepresented in arrays of region surrounding PBS3 are expressed in flowers (Online tandemly duplicated genes (Hanada et al. 2008; Rizzon et al. Resource 2). Indeed, using a PBS3 promoter::GUS fusion, 2006) and tend to be retained following local duplication Jagadeeswaran et al. (2007) found PBS3 to be expressed at events (reviewed in Freeling 2009). In addition, the same multiple stages of flower development and in specific floral genes that tend to be locally duplicated also tend to transpose organs. Benzoate metabolism in flowers is integrated with within chromosomal regions with high recombination rates. the functions of other phytohormones and can be modu- Much of the work studying the divergence in expression of lated by herbivory (e.g. see Kessler et al. 2010). Therefore, duplicated genes has been done in yeast. However, expres- a possible conserved function of PBS3 syntelogs may be to sion patterns of tandemly and segmentally duplicated genes modulate stress-induced benzoate metabolism in flowers in A. thaliana have been found to be similar, with a weak but and by so doing provide a reproductive benefit. significant correlation between expression pattern and pro- In addition to its expression in flowers, NEF1 is strongly moter similarity (Haberer et al. 2004). The expression of the induced by osmotic stress, heat, and cold in seedlings (abi- tandemly duplicated sets of A. thaliana genes in the GH3 otic stress, At-TAX series). Both PBS3 and OsGH3.13 are family, PBS3/AtGH3.12 and AtGH3.13-16; and AtGH3. 18 also induced by drought/salt stress in addition to pathogens and AtGH3.19, do not retain correlated expression (data not and SA, as is the nearby ERF transcription factor At5g13330/ shown). The expression of the genes in different types of RAP2.6L (At-TAX dataset; Jagadeeswaran et al. 2007; tissues suggests possible specialization (Online Resource 5). Zhang et al. 2009; Krishnaswamy et al. 2011). Similar to In addition, they differ in frequency of putative transcription overexpression of the PBS3 syntelog OsGH3.13 in rice factor binding motifs (Online Resource 4), suggesting evi- (Zhang et al. 2009), overexpression of RAP2.6L in Arabid- dence for promoter evolution. In many cases, these newer opsis results in reduced stature, enhanced salt and drought AtGH3 genes are induced in response to (a)biotic stress tolerance (Krishnaswamy et al. 2011). By contrast, rap2.6L (Fig. 6) suggesting an evolving role for these genes in Arabidopsis mutants exhibit enhanced expression of SA- response to stress. The rare effector HopW1-1 from Pseu- dependent defense genes (e.g. PR1) and enhanced resistance domonas syringae pv. maculicola has been shown to bind to to Pseudomonas syringae (Sun et al. 2010), while pbs3 the AtPBS3 protein (Lee et al. 2008) indicating unique tar- mutants are compromised in SA-dependent gene expression geting of PBS3. This further illustrates the importance of the and resistance to P. syringae (e.g. Nobuta et al. 2007). PBS3 protein in disease resistance/susceptibility and sug- Unfortunately, rice plants with altered OsGH3.13 expression gests host counter-evolution could play a role in the have not been tested for altered pathogen resistance/ expansion of this clade in Arabidopsis. 123 Plant Mol Biol

Acknowledgments We would like to thank Dr. Eric Lyons for his Domingo C, Andres F, Tharreau D, Iglesias DJ, Talon M (2009) assistance with the CoGe browser, Dr. Divya Chandran for careful Constitutive expression of OsGH3.1 reduces auxin content and reading of the manuscript, and the William Carroll Smith Graduate enhances defense response and resistance to a fungal pathogen in Research Fellowship in Plant Pathology (to R.A.O) and UC Berkeley rice. Mol Plant-Micro Interact 22:201–210 awards (to M.C.W.) for financial support. Some of the genome sequence Dudareva N, Pichersky E (2000) Biochemical and molecular genetic data described here was analyzed prior to publication by the sequencing aspects of floral scents. Plant Physiol 122:627–634 projects. Of these, the Aquilegia coerulea, Mimulus guttatus, and Edgar RC (2004) Muscle: multiple sequence alignment with high Selaginella moellendorffii data were produced by the US Department of accuracy and high throughput. Nucleic Acids Res 32: Energy Joint Genome Institute. Carica papaya data were produced by 1792–1797 the ASGPB Hawaii Papaya Genome Project (http://www. Felten J, Kohler A, Morin E, Bhalerao RP, Palme K et al (2009) The asgpb.mhpcc.hawaii.edu/papaya/). Zea mays data were produced by ectomycorrhizal fungus Laccaria bicolor stimulates lateral root the Genome Sequencing Center at Washington University School of formation in poplar and Arabidopsis through auxin transport and Medicine in St. Louis and can be obtained from http://www. signaling. Plant Physiol 151:1991–2005 maizesequence.org/. Ferrari S, Plotnikova JM, De Lorenzo G, Ausubel FM (2003) Arabidopsis local resistance to Botrytis cinerea involves salicylic acid and camalexin and requires EDS4 and PAD2, but not SID2, EDS5, or PAD4. Plant J 35:193–205 Freeling M (2009) Bias in plant gene content following different sorts References of duplication: Tandem, whole-genome, segmental, or by transposition. Annu Rev Plant Biol 60:433–453 Albertazzi G, Milc J, Caffagni A, Francia E, Roncaglia E et al (2009) Goda H, Sasaki E, Akiyama K, Maruyama-Nakashita A, Nakabayashi Gene expression in grapevine cultivars in response to Bois noir K et al (2008) The AtGenExpress hormone and chemical phytoplasma infection. Plant Sci 176:792–804 treatment data set: experimental design, data evaluation, model Ariizumi T, Hatakeyama K, Hinata K, Inatsugi R, Nishida I, Sata data analysis and data access. Plant J 55:526–542 S, Kato T, Tabata S, Toriyama K (2004) Disruption of the Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm novel plant protein NEF1 affects lipid accumulation in the to estimate large phylogenies by maximum likelihood. Syst Biol plastids of the tapetum and exine formation of pollen, 52:696–704 resulting in male sterility in Arabidopsis thaliana. Plant J Gutterson N, Reuber TL (2004) Regulation of disease resistance 39:170–181 pathways by AP2/ERF transcription factors. Curr Opin Plant Bierfreund NM, Tintelnot S, Reski R, Decker EL (2004) Loss of GH3 Biol 7:465–471 function does not affect phytochrome-mediated development in Haberer G, Hindemitt T, Meyers BC, Mayer KFX (2004) Transcrip- a moss, Physcomitrella patens. J Plant Physiol 161:823–835 tional similarities, dissimilarities, and conservation of cis- Blanc G, Hokamp K, Wolfe KH (2003) A recent polyploidy elements in duplicated genes of Arabidopsis. Plant Physiol superimposed on older large-scale duplications in the Arabid- 136:3009–3022 opsis genome. Genome Res 13:137–144 Hagen G, Kleinschmidt A, Guilfoyle T (1984) Auxin-regulated gene Bourne DJ, Barrow KD, Milborrow BV (1991) Salicyloylaspartate as expression in intact soybean Glycine max cultivar Wayne an endogenous component of the leaves of Phaseolus vulgaris. hypocotyl and excised hypocotyl sections. Planta 162:147–153 Phytochemistry 30:4041–4044 Hall BG (2005) Comparison of the accuracies of several phylogenetic Castresana J (2000) Selection of conserved blocks from multiple methods using protein and DNA sequences. Mol Biol Evol alignments for their use in phylogenetic analysis. Mol Biol Evol 22:792–802 17:540–552 Hanada K, Zou C, Lehti-Shiu MD, Shinozaki K, Shiu SH (2008) Chang K-H, Xiang H, Dunaway-Mariano D (1997) Acyl-adenylate Importance of lineage-specific expansion of plant tandem motif of the acyl-adenylate/thioester-forming enzyme superfam- duplicates in the adaptive response to environmental stimuli. ily: a site directed mutagenesis study with the Pseudomonas sp. Plant Physiol 148:993–1003 Strain CBS3 4-chlorobenzoate:Coenzyme A ligase. Biochemis- Hsieh H-L, Okamoto H, Wang M, Ang L-H, Matsui M et al (2000) try 36:15650–15659 FIN219, an auxin-regulated gene, defines a link between Chini A, Fonseca S, Fernandez G, Adie B, Chico JM et al (2007) The phytochrome A and the downstream regulator COP1 in light JAZ family of repressors is the missing link in jasmonate control of Arabidopsis development. Genes Dev 14:1958–1970 signalling. Nature 448:666 Huelsenbeck JP, Hillis DM (1993) Success of phylogenetic methods Conti E, Franks NP, Brick P (1996) Crystal structure of firefly in the four-taxon case. Syst Biol 42:247–264 luciferase throws light on a superfamily of adenylate-forming Jagadeeswaran G, Raina S, Acharya BR, Maqbool SB, Mosher SL et al enzymes. Structure 4:287–298 (2007) Arabidopsis GH3-like Defense Gene 1 is required for Craigon DJ, James N, Okyere J, Higgins J, Jotham J et al (2004) accumulation of salicylic acid, activation of defense responses and Nascarrays: a repository for microarray data generated by NASC’s resistance to Pseudomonas syringae. Plant J 51:234–246 transcriptomics service. Nucleic Acids Res 32:D575–D577 Jaillon O, Aury JM, Noel B, Policriti A, Clepet C et al (2007) The Dereeper A, Guignon V, Blanc G, Audic S, Buffet S et al (2008) grapevine genome sequence suggests ancestral hexaploidization Phylogeny.Fr: robust phylogenetic analysis for the non-special- in major angiosperm phyla. Nature 449:463–467 ist. Nucleic Acids Res 36:W465–W469 Jain M, Kaur N, Tyagi AK, Khurana JP (2006) The auxin-responsive Ding X, Cao Y, Huang L, Zhao J, Xu C et al (2008) Activation of the GH3 gene family in rice (oryza sativa). Funct Integr Genomics indole-3-acetic acid-amido synthetase GH3–8 suppresses expan- 6:36–46 sin expression and promotes salicylate- and jasmonate-indepen- Jansen R, Kaittanis C, Saski C, Lee S-B, Tomkins J et al (2006) dent basal immunity in rice. Plant Cell 20:228–240 Phylogenetic analyses of Vitis (Vitaceae) based on complete Dombrecht B, Xue GP, Sprague SJ, Kirkegaard JA, Ross JJ et al chloroplast genome sequences: effects of taxon sampling and (2007) MYC2 differentially modulates diverse jasmonate- phylogenetic methods on resolving relationships among rosids. dependent functions in Arabidopsis. Plant Cell 19:2225–2245 BMC Evol Biol 6:32

123 Plant Mol Biol

Jun J, Mandoiu I, Nelson C (2009) Identification of mammalian Rizzon C, Ponger L, Gaut BS (2006) Striking similarities in the orthologs using local synteny. BMC Genomics 10:630 genomic distribution of tandemly arrayed genes in Arabidopsis Kessler D, Diezel C, Baldwin IT (2010) Changing pollinators as a and rice. PLoS Comput Biol 2:e115 means of escaping herbivores. Curr Biol 20:237–242 Schnable PS, Ware D, Fulton RS, Stein JC, Wei F et al (2009) The Khan S, Stone JM (2007) Arabidopsis thaliana GH3.9 influences B73 maize genome: complexity, diversity, and dynamics. primary root growth. Planta 226:21–34 Science 326:1112–1115 Krishnaswamy S, Verma S, Rahman MH, Kav NNV (2011) Semon M, Wolfe KH (2007) Consequences of genome duplication. Functional characterization of four APETALA2-family genes Curr Opin Genet Dev 17:505–512 (RAP2.6, RAP2.6L, DREB19 and DREB26) in Arabidopsis. Singh KB, Foley RC, Onate-Sanchez L (2002) Transcription factors Plant Mol Biol 75:107–127 in plant defense and stress responses. Curr Opin Plant Biol Lee MW, Lu H, Jung HW, Greenberg JT (2007) A key role for the 5:430–436 Arabidopsis WIN3 protein in disease resistance triggered by Staswick PE, Tiryaki I (2004) The oxylipin signal jasmonic acid is Pseudomonas syringae that secrete AvrRpt2. MPMI 20:1192–1200 activated by an enzyme that conjugates it to isoleucine in Lee MW, Jelenska J, Greenberg JT (2008) Arabidopsis proteins Arabidopsis. Plant Cell 16:2117–2127 important for modulating defense responses to Pseudomonas Staswick PE, Su W, Howell SH (1992) Methyl jasmonate inhibition syringae that secrete HopW1–1. Plant J 54:452–465 of root growth and induction of a leaf protein are decreased in an Liu Kl, Kang B-C, Jiang H, Moore SL, Li H et al (2005) A GH3-like Arabidopsis thaliana mutant. Proc Natl Acad Sci USA gene, CcGH3, isolated from Capsicum chinense l. Fruit is 89:6837–6840 regulated by auxin and ethylene. Plant Mol Biol 58:447–464 Staswick PE, Tiryaki I, Rowe ML (2002) Jasmonate response locus Ludwig-Muller J, Julke S, Bierfreund NM, Decker EL, Reski R JAR1 and several related Arabidopsis genes encode enzymes of (2009) Moss (Physcomitrella patens) GH3 proteins act in auxin the firefly luciferase superfamily that show activity on jasmonic, homeostasis. New Phytol 181:323–338 salicylic, and indole-3-acetic acids in an assay for adenylation. Lyons E, Freeling M (2008) How to usefully compare homologous plant Plant Cell 14:1405–1415 genes and chromosomes as DNA sequences. Plant J 53:661–673 Staswick PE, Serban B, Rowe M, Tiryaki I, Maldonado MT et al Lyons E, Pedersen B, Kane J, Alam M, Ming R et al (2008) Finding (2005) Characterization of an Arabidopsis enzyme family that and comparing syntenic regions among Arabidopsis and the conjugates amino acids to indole-3-acetic acid. Plant Cell outgroups papaya, poplar, and grape: CoGe with rosids. Plant 17:616–627 Physiol 148:1772–1781 Steel M (2005) Should phylogenetic models be trying to ‘‘Fit an Martinez C, Pons E, Prats G, Leon J (2004) Salicylic acid regulates elephant’’? Trends Genet 21:307–309 flowering time and links defence responses and reproductive Sun F, Liu P, Xu J, Dong H (2010) Mutation in RAP2.6L, a development. Plant J 37:209–217 transactivator of the ERF transcription factor family enhances Ming R, Hou S, Feng Y, Yu Q, Dionne-Laporte A et al (2008) The resistance to Pseudomonas syringae. Physiol Mol Plant Pathol draft genome of the transgenic tropical fruit tree papaya (Carica 74:295–301 papaya linnaeus). Nature 452:991–996 Suzuki Y, Yanaguchi I, Murofushi N, Takahasi N (1988) Biological Molina C, Grotewold E (2005) Genome wide analysis of Arabidopsis conversion of benzoic acid in Lemna paucicostata 151 and its core promoters. BMC Genomics 6:25 relation to flower induction. Plant Cell Physiol 29:439–444 Mount DW (2004) Bioinformatics: sequence and genome analysis. Takase T, Nakazawa M, Ishikawa A, Manabe K, Matsui M (2003) Cold Spring Harbor Laboratory Press, Cold Spring Harbor DFL2, a new member of the Arabidopsis GH3 family is involved Nakazawa M, Yabe N, Ichikawa T, Yamamoto YY, Yoshizumi T in red light-specific hypocotyl elongation. Plant Cell Physiol et al (2001) DFL1, an auxin-responsive GH3 gene homologue, 44:1071–1080 negatively regulates shoot cell elongation and lateral root Takase T, Nakazawa M, Ishikawa A, Kawashima M, Ichikawa T, formation, and positively regulates the light response of Takahashi N, Shimada H, Manabe K, Matsui M (2004) ydk1-D, hypocotyl length. Plant J 25:213–221 an auxin-responsive GH3 mutant that is involved in hypocotyl Nobuta K, Okrent RA, Stoutemyer M, Rodibaugh N, Kempema L and root elongation. Plant J 37:471–483 et al (2007) The GH3 acyl adenylase family member PBS3 Tamura K, Dudley J, Nei M, Kumar S (2007) MEGA4: molecular regulates salicylic acid-dependent defense responses in Arabid- evolutionary genetics analysis (MEGA) software version 4.0. opsis. Plant Physiol 144:1144–1156 Mol Biol Evol 24:1596–1599 Okrent RA, Brooks MD, Wildermuth MC (2009) Arabidopsis Terol J, Domingo C, Talon M (2006) The GH3 family in plants: GH3.12 (PBS3) conjugates amino acids to 4-substituted genome wide analysis in rice and evolutionary history based on benzoates and is inhibited by salicylate. J Biol Chem 284: EST analysis. Gene 371:279–290 9742–9754 Thornton JW, Kolaczkowski B (2005) No magic pill for phylogenetic Ouyang S, Zhu W, Hamilton J, Lin H, Campbell M et al (2006) The error. Trends Genet 21:310–311 TIGR rice genome annotation resource: improvements and new Toufighi K, Brady SM, Austin R, Ly E, Provart NJ (2005) The botany features. Nucleic Acids Res 35:D883–D887 array resource: E-northerns, expression angling, and promoter Park J-E, Seo PJ, Lee A-K, Jung J-H, Kim Y-S et al (2007a) An analyses. Plant J 43:153–163 Arabidopsis GH3 gene, encoding an auxin-conjugating enzyme, Trennheuser F, Burkhard G, Becker H (1994) Anthocerodiazonin, an mediates phytochrome B-regulated light signals in hypocotyl alkaloid from Antheroceros agretis. Phytochemistry 37:899–903 growth. Plant Cell Physiol 48:1236–1241 Tuskan GA, DiFazio S, Jansson S, Bohlmann J, Grigoriev I et al Park J-E, Park J-Y, Kim Y-S, Staswick PE, Jeon J, Yun J, Kim (2006) The genome of black cottonwood, populus trichocarpa S-Y, Kim J, Lee Y-H, Park C-M (2007b) GH3-mediated (torr. & gray). Science 313:1596–1604 auxin homeostasis links growth regulation with stress Warren RF, Merritt PM, Holub E, Innes RW (1999) Identification of adaptation response in Arabidopsis. J Biol Chem 282: three putative signal transduction genes involved in R gene- 10036–10046 specified disease resistance in Arabidopsis. Genetics 152:401–412 Rensing SA, Lang D, Zimmer AD, Terry A, Salamov A et al (2008) Winter D, Vinegar B, Nahal H, Ammar R, Wilson GV et al (2007) An The Physcomitrella genome reveals evolutionary insights into ‘‘electronic fluorescent pictograph’’ browser for exploring and the conquest of land by plants. Science 319:64–69 analyzing large-scale biological data sets. PLoS One 2:e718

123 Plant Mol Biol

Wise RP, Caldo RA, Hong L, Shen L, Cannon E et al (2007) during Arabidopsis-Pseudomonas syringae interaction. Plant Barleybase/plexdb: a uniﬁed expression proﬁling database for Physiol 145:450–464 plants and plant pathogens. In: Edwards D (ed) Plant bioinfor- Zhang S-W, Li C-H, Cao J, Zhang Y-C, Zhang S-Q et al (2009) matics—methods and protocols. Humana Press, Totowa, Altered architecture and enhanced drought tolerance in rice via pp 347–363 the down-regulation of indole-3-acetic acid by tld1/osgh3.13 Woodward AW, Bartel B (2005) Auxin: regulation, action, and activation. Plant Physiol 151:1889–1901 interaction. Ann Bot 95:707–735 Zimmermann P, Hirsch-Hoffmann M, Hennig L, Gruissem W (2004) Wu CA, Lowry DB, Cooley AM, Wright KM, Lee YW et al (2007) GENEVESTIGATOR. Arabidopsis microarray database and Mimulus is an emerging model system for the integration of analysis toolbox. Plant Physiol 136:2621–2632 ecological and genomic studies. Heredity 100:220–230 Zhang Z, Li Q, Li Z, Staswick PE, Wang M et al (2007) Dual regulation role of GH3.5 in salicylic acid and auxin signaling

123