Genome-wide phylogenetic analysis Acyl:CoA synthetase related genes
The adenylate-forming enzyme superfamily that uses a common reaction mechanism includes members from all organisms studied to date, including prokaryotes and eukaryotes (Becker-Andre et al., 1991; Shockey et al., 2003; Schneider et al., 2005). We identified 88 full-length genes encoding adenylate-forming enzymes related to 4CL from genomic databases, using in silico similarity searches based on the amino acid sequences of Arabidopsis 4CL proteins (Ehlting et al., 1999; Hamberger & Hahlbrock, 2004). In this analysis, we focused on three angiosperms with complete genome sequences available (Arabidopsis, poplar and rice), the genomes of maize, Physcomitrella, Chlamydomonas and the genomes of selected other microorganisms (fungi and bacteria) for which complete or substantial genome sequence data is available.
A phylogenetic analysis of over 100 sequences from these various organisms, including bona fide 4CL sequences from Arabidopsis, poplar, and rice, is shown in Fig. 1. This analysis revealed two general groups of adenylate-forming proteins. One large group contained representatives from all organisms analysed, including bacteria, fungi, Chlamydomonas, Physcomitrella and angiosperm plants (large arc in Fig. 1). These probably represent adenylate-forming enzymes with metabolic functions that are conserved in many lineages. As an example of possible functions, one clade in this group contains the Arabidopsis ACN1 gene, which encodes an acetate:CoA ligase that functions as an entry point to the glyoxylate cycle during seed germination (Turner et al., 2005). The clade containing ACN1 is enriched for angiosperm plant sequences, and includes Arabidopsis genes encoding the acyl activating enzymes (AAE) AAE4 and AAE6, which may play housekeeping functions related to fatty acid metabolism (Shockey et al., 2003), as well as two Chlamydomonas genes. A sister clade contains a Saccharomyces cerevisiae gene encoding the FAT2 peroxisomal acyl-CoA synthetase, as well as the Arabidopsis AAE3 gene and several poplar, rice, and Physcomitrella genes, all of unknown function. The ACS protein encoded by the bacterium Streptomyces coelicolor A3(2), ScCCL, has been shown to have high activity against 4-coumaric and cinnamic acids and is thus designated as a cinnamic acid-CoA ligase, ScCLL (Kaneko et al., 2003). This raises the possibility that other bacterial and fungal enzymes in this clade have activities against phenolic substrates. Based on their similarity to Arabidopsis AAE genes, we designated the annotated poplar, rice, and Physcomitrella genes in this large group as AAEL (acyl activating enzyme-like) genes (Table S4).
Figure 1. Phylogenetic tree of 104 proteins related to 4-coumarate:CoA ligase (4CL). 4CL-related (acyl-CoA synthetase (ACS)) sequences corresponding to translated nucleotide sequences from full-length cDNAs and expressed sequence tags (ESTs) were aligned, and an unrooted phylogenetic tree was generated. Nodes with bootstrap values above 70% are indicated by asterisks. Triangles represent the subclades of land plant-specific ACSs most closely related to true 4CL enzymes. The large dashed arc indicates a large group of genes found in all organisms from all lineages investigated (land plants, algae, fungi, protists and prokaryotes); the dashed oval indicates the clade of land plant-specific ACS genes, as discussed in the text. Solid triangle, true 4CL enzymes; stippled triangle, nonperoxisomal ACSs in clade A; hatched triangles, ACS clades B–E as described in the text. A Physcomitrella clade of land-plant ACSs is indicated. Gene name prefixes: At, Arabidopsis thaliana; Os, Oryza sativa; Poptr, Populus trichocarpa; Pp, Physcomitrella patens. Gene names and identifiers are given in Table 1 and the Supporting Information, Table S4. The scale represents 0.1 amino acid changes.
Download figure to PowerPoint
A second group of adenylate-forming proteins (Fig. 1, highlighted by an oval) contained both bona fide 4CL proteins and previously annotated Arabidopsis 4CL-like (Costa et al., 2003; Raes et al., 2003; Shockey et al., 2003; Ehlting et al., 2005) acyl-CoA synthetase (ACS) proteins. Strikingly, this ACS group is apparently land plant-specific. All angiosperm genomes, as well as the genome of the moss Physcomitrella patens, contained genes encoding proteins in this group, while no representatives from other eukaryote lineages were found. It is interesting, however, that Penicillium and Dictyostelium contain genes relatively closely related to this land plant group. Inclusion of full-length gymnosperm ACS gene sequences, when they become available, in this analysis will allow definitive testing of the interpretation that this clade is restricted to land plants.
Within the apparent land plant-specific group, 5 clades containing angiosperm ACSs could be further delineated (Fig. 1; clades A–E). The ACS genes encoding proteins in these clades are phylogenetically closely related to bona fide 4CLs, which form a sister clade to clades A–E. As presented in more detail later, each of clades A–E contains at least one sequence representative of each of the four angiosperm plant species analysed (Arabidopsis, poplar, rice and maize), demonstrating that these proteins are evolutionarily conserved in the angiosperm lineage and that a common ancestor of these clades was present before the divergence of monocots and eudicots. The Arabidopsis, poplar, and rice proteins represented in the bona fide 4CL clade have been described and annotated (Ehlting et al., 1999; Hamberger & Hahlbrock, 2004; Tsai et al., 2006; Tuskan et al., 2006; Hamberger et al., 2007). The previously and currently annotated Arabidopsis, poplar, rice and Physcomitrella 4CL and ACS genes are given in Table 1.
Analysis of the Physcomitrella genome revealed four putative 4CL genes encoding proteins that grouped in the bona fide 4CL clade (Table 1; de Azevedo Souza et al., unpublished), as well as five ACS genes falling outside the angiosperm ACS clades B–E. One Physcomitrella ACS sequence grouped into clade A (PpACS6; Table 1). This suggests that 4CL and clade A ACS genes originated early during the evolution of land plants before divergence of tracheophytes (vascular plants) and bryophytes, consistent with a possible role for these enzymes in the biosynthesis of phenylpropanoids and/or the extracellular matrix, a key innovation for adaptation to the land environment (Bowman et al., 2007). Interestingly, Physcomitrella PpACS5 is basal to the ACS clades B–E, suggesting that it could represent an ancestral ACS gene retained in Physcomitrella. The remaining four Physcomitrella ACS sequences formed a distinct clade, suggesting bryophyte-specific evolution and diversification of this ACS gene family, which could encode proteins with functions distinct from those of angiosperm ACSs.
Almost all sequences in clades B, C, D and E, as well as Physcomitrella sequences PpACS1–4, contain a consensus PTS1 peroxisomal target sequence (Table 1, Fig. 2; Reumann et al., 2004) at their C-termini, which suggests they are targeted to this organelle. Localization of ACS proteins to the peroxisome has been experimentally verified for Arabidopsis proteins in clade E (At4g05160, ACS6 and At5g63380, ACS9; Schneider et al., 2005) and the Arabidopsis OPCL1 (At1g20510) protein in clade D (Koo et al., 2006). Interestingly, all fungal ACS-related enzymes identified and shown in Fig. 1 also have peroxisomal target signals. Given that peroxisomes play a major role in fatty acid metabolism in both plants and fungi, and that acyl:CoA ligases are widely used in modification of these molecules, it is possible that ACSs and 4CLs were recruited from fatty acid metabolism early in land plant evolution, to perform their current functions. None of the ACSs in clade A, most closely related to bona fide 4CLs, contained the PTS1 sequence, suggesting that loss of this sequence may have played a role in the acquisition of 4CL and subclade A functions. Furthermore, Physcomitrella PpACS5 located at a position basal to ACS clades B–D also lacks a PTS1 targeting sequence, suggesting a potentially distinct biochemical function for this enzyme, relative to the peroxisomally targeted ACSs.
Figure 2. Phylogenetic relationships of plant-specific acyl-CoA synthetases (ACSs) from three fully sequenced angiosperm genomes. Translated nucleotide sequences corresponding to ACS genes from Arabidopsis, poplar and rice were aligned and an unrooted phylogenetic tree generated. Nodes with bootstrap values above 70% are shown by stars. The 4-coumarate:CoA ligase (4CL) and ACS clades A–E discussed in the text are circled and contain at least one representative of each plant species. Protein names in shaded boxes contain the PTS1 peroxisomal target signal. Bar represents 0.1 amino acid changes.
Download figure to PowerPoint
The angiosperm-specific ACS clades were next analysed in more detail, focusing on complete gene families from Arabidopsis, poplar, and rice, for which whole genome sequence information is available. As shown in detail in Fig. 2 and Table 1, all three species contained ACS proteins in each of clades A–E. While the number of ACS genes within each genome was similar (13 in poplar, 12 in rice, and 9 in Arabidopsis; Table 1 and Fig. 2), the number in each clade varied between species. Certain ACS clades were greatly enriched for proteins from a particular species. For example, clade D is an Arabidopsis gene-rich clade, with five Arabidopsis representatives, two from poplar and only one rice member. By contrast, clade E is poplar rich with seven genes, three from rice and only one from Arabidopsis. Clade A, unique in containing ACS genes lacking the PTS1 targeting signal, is the only clade that contained a single representative from each species.
These data clearly show that ACS genes in different clades have undergone differential expansion in each angiosperm lineage, perhaps reflecting different events in the evolution of their genomes and differences in life histories that placed varying selective pressures on the elaboration of biochemical pathways requiring ACS activity. All four Arabidopsis ACS genes in clade D are located in tandem on chromosome 1, suggesting tandem duplication and selection for retention of the duplicated copies. Two of the poplar ACS genes in clade E (PoptrACS11 and PoptrACS12) appear to have arisen by tandem duplication on linkage group XII. However, other members of this and other clades for which the poplar gene models are anchored to linkage groups are physically unlinked and on different linkage groups. This suggests that tandem gene duplication has not played a major role in diversification of the poplar ACS gene family. Many of the poplar ACS genes may rather have been retained after the salicoid whole genome duplication in the poplar lineage, in which chromosome doubling and subsequent rearrangement is thought to have increased the poplar chromosome complement from n = 10 to the current n = 19 (Tuskan et al., 2006). For example, in the poplar-rich clade E, PoptrACS10 and PoptrACS11/12, which are located on duplicated homoeologous linkage groups XII and XV, and PoptrACS6 and PoptrACS7, which are located on duplicated homoeologous linkage groups XIII and X, are likely to have arisen in this manner (Tuskan et al., 2006). Also noteworthy among the poplar clade E ACS proteins is the loss of C-terminal PTS1 peroxisomal targeting sequences in two members (PoptrACS6 and PoptrACS12), suggesting that functional diversification may have taken place at the level of enzyme localization in the poplar lineage after gene duplication. Taken together, these data show conservation of ACS gene number in all three lineages for some clades (A, B and C, with one or two members from each lineage), suggesting possible conservation of function. Conversely, diversification of gene numbers in clades D and E has occurred in a lineage-specific manner, suggesting possible diversification of function in a lineage specific manner.
Developmental expression of poplar and Arabidopsis ACS genes
In order to gain insights into the possible functions of ACS proteins, we examined the developmental gene expression patterns of all Arabidopsis ACS genes, as well as representative poplar genes in clades A-E, by qRT-PCR. Expression profiles in different Arabidopsis and poplar organs and tissues are shown in Fig. 3, relative to the organ or tissue with the highest expression. To evaluate absolute expression levels, we also calculated the expression level of each gene tested relative to the respective Arabidopsis or poplar control gene in the organ or tissue where the gene was most highly expressed (Fig. 3). In general, Arabidopsis and poplar ACS genes from the same clade tended to have similar relative developmental expression patterns, and similar levels of expression. This is especially evident in those clades with single Arabidopsis and poplar ACS representatives. A striking example is clade A, in which AtACS5 expression was strongly flower specific, and PoptrACS13 had a similar pattern of flower-preferred expression. Interestingly, PoptrACS13 expression was specific to male flowers, and a putative AtACS5 orthologue in tobacco shows an anther-preferred expression pattern (Varbanova et al., 2003), suggesting a role for ACS enzymes of this clade in a biochemical pathway important in anther and/or pollen development. Another example is the predominant expression of both Arabidopsis and poplar representatives of clade C in leaves, with lower expression in stem/xylem and flowers. Both the Arabidopsis and poplar representatives were expressed at low levels relative to the respective control genes, suggesting a potential specialized function (for example, expression in restricted cell types). Clade B contains two poplar genes and a single Arabidopsis member. Genes in this clade were expressed in all organs, but both poplar PoptrACS2 and AtACS6 showed highest expression in mature leaves, and lower expression in other organs. The pattern of PoptrACS1 expression differed from that of PoptrACS2, with highest expression in flowers, bark and young leaves, suggesting subfunctionalization in expression patterns, as predicted from genes retained after gene duplication events (Duarte et al., 2006). The Arabidopsis and poplar genes in this clade were expressed at similar, high levels relative to the control genes, but expression of PoptrACS2, the most highly expressed poplar ACS gene, was much higher than that of PoptrACS1, supporting the subfunctionalization of these genes at the level of expression.
More complex expression patterns were observed in clades where expansion of gene family members in either Arabidopsis or poplar has occurred. In clade D, the duplicated and highly similar poplar genes PoptrACS4 and PoptrACS5 appeared to have similar expression patterns across a range of organs (Fig. 3), and shared similar, high expression levels. However, the transcribed portions of the two poplar genes were so similar that cross-detection cannot be excluded. By contrast, for the five representatives of the Arabidopsis members of clade D, distinct and complementary expression patterns were observed throughout the majority of organs tested, and the genes were expressed at remarkably different levels (0.51 relative to the control gene level for AtASC2, to 1482 relative to the control gene for OPCL1), which suggests expression subfunctionalization of these genes. The demonstration that the Arabidopsis OPCL1 gene in this clade encodes an OPDA-CoA ligase (Koo et al., 2006) is consistent with the very high developmental expression of OPCL1 in flowers, where JA plays a role in anther development (Ito et al., 2007). Given their phylogenetic relationships to OPCL1, other genes in this clade could share this activity, especially since some of them are wound and methyl jasmonate (MJ) inducible (Koo et al., 2006 and later), and since the opcl1 mutant retains the ability to make JA both developmentally and in response to wounding (Koo et al., 2006). The distinct developmental expression patterns and expression levels of the Arabidopsis clade D ACS genes could therefore suggest specialization of at least some of the genes for developmental biosynthesis of JA in different organs.
The only Arabidopsis representative in clade E, AtACS9, was most highly expressed (at a high absolute level) in seedlings, followed by flowers. The three poplar homologues most closely related to AtACS9 (PoptrACS10, PoptrACS11 and PoptrACS12) displayed expression patterns largely complementary to each other, covering leaves, roots and male flowers, although all had high expression in roots. Like AtACC9, expression levels of the poplar genes were generally high, although they varied over 10-fold between each other. Thus, expansion of this family of ACS genes in poplar appears to have been accompanied by subfunctionalization for developmental expression.
Stress-induced expression of poplar and Arabidopsis ACS genes
The stress-responsiveness of the set of ACS genes, except the flower-specific genes in clade A (which are not expressed in vegetative organs in Arabidopsis), was tested at various times after mechanical wounding in both Arabidopsis and poplar. The ACS promoter-GUS fusions were constructed, using genomic sequences upstream of the ATG start codon, between 1.5 kb and 2 kb in length. Transgenic lines were generated, and promoter activity following wounding assayed in representative lines. Wound-induced transcription of Arabidopsis and poplar genes was also tested by qRT-PCR. In addition, expression of the poplar ACS genes was tested by qRT-PCR after each of the following treatments: herbivory by the forest tent caterpillar (Malacosoma disstria; FTC), simulated herbivory (SH; wounding plus Malacosoma disstria regurgitant) and exposure to MJ.
Wound-induced expression data are shown in Fig. 4, and the responses of poplar genes are shown in Fig. 5. Arabidopsis 4CL2, known to be wound inducible (Ehlting et al., 1999), was used as a positive control for Arabidopsis wound treatments, and was up-regulated over 5 fold at 4 h after wounding (data not shown). The endogenous AtACS6 gene and AtACS6 promoter-GUS transgene were not wound inducible (Fig. 4), and PoptrACS2 in the same clade (B) was unresponsive both to wounding and other stress treatments (Figs 4 and 5). However, PoptrACS1, also in clade B, was upregulated 1.6-fold after 4 h wounding, and was strongly and transiently upregulated by SH and FTC treatments (Fig. 5). In a separate microarray expression profiling experiment, AtACS6 expression was not activated by diamondback moth herbivory (J. Ehlting and J. Bohlmann, unpublished). These data suggest that AtACS6 and PoptrACS2 do not function in wound or herbivory stress-related biochemical pathways, but that PoptrACS1 may have gained this function after duplication of the gene in the poplar lineage, and that the biochemical pathway using the product of the enzyme encoded by PoptrACS1 may play a role in defence against herbivory.
In clade C, AtACS7 expression was downregulated by wounding to less than half the level of the unwounded control, and a similar result was obtained for the single poplar homologue in this group, PoptrACS3. Interestingly, PoptrACS3 was strongly and transiently upregulated by SH but not other stresses. These results suggest that enzymes in this clade are not likely to play roles in biochemical pathways related to herbivory and similar stresses, although the strong response of PoptrACS3 to SH treatment warrants further investigation.
Arabidopsis and poplar genes in clade D were particularly responsive to wounding stress. Histochemical assays of transgenic promoter-GUS lines showed that the Arabidopsis ACS2, ACS3 and OPCL1 promoters were strongly responsive to wounding (Fig. 4a). Accumulation of ACS2 and ACS3 mRNA was also strongly upregulated by wounding, with ACS3 levels 14-fold above the untreated the control within 1 h of wounding (Fig. 4b). While wound-induced accumulation of OPCL1 and AtACS8 mRNA was not obvious under the conditions used here, Koo et al. (2006) previously showed that expression of OPCL1 is strongly, and ACS8 weakly, wound inducible. OPCL1 has been shown to encode a wound and MJ-inducible OPDA:CoA ligase involved in JA biosynthesis (Koo et al., 2006). Our data are consistent with potential roles of other Arabidopsis ACS genes in this clade in JA biosynthesis, in addition to OPCL1, perhaps as members of an OPCL gene family, as proposed by Koo et al. (2006). In support of this, AtACS8, as well as OPCL, is activated by diamondback moth herbivory based on microarray expression profiling (J. Ehlting and J. Bohlmann, unpublished data). Furthermore, analysis of public microarray gene expression data using the Bio-Array Resource, (http://www.bar.utoronto.ca; Toufighi et al., 2005) showed that expression of OPCL1, AtASC2 and AtASC8 is induced by MJ treatment in 7-d-old seedlings. AtASC3 is not represented in the microarray probe sets used in these experiments, but was strongly wound inducible in our experiments (Fig. 4). Unlike the other Arabidopsis members of Clade D, there is no evidence that AtACS1 is wound or MJ inducible (Fig. 2), and it lacks apparent activity against JA precursors in vitro (Kienow et al., 2008) suggesting that it is less likely to be involved in JA biosynthesis. However, further analyses will be required to determine if Arabidopsis Clade D proteins other than AtOPCL1 have in vivo activities against OPDA, or preferentially use other substrates.
The expression of the poplar genes in this clade, PoptrACS4 and PoptrACS5, was induced up to fivefold 4 h after wounding, and remained high 24 h after treatment (Fig. 4b). PoptrACS4 and PoptrACS5 were also strongly upregulated by herbivory, SH and MJ, with the last treatment leading to a transient 20-fold increase in expression by 2 h, with mRNA returning to control levels by 24 h (Fig. 5). These two highly similar poplar genes are most similar to Arabidopsis OPCL1 based on phylogenetic reconstruction (Fig. 2), and like OPCL1 are highly expressed in a developmental context (Fig. 3d). These data are consistent with PoptrACS4 and PoptrACS5 encoding poplar OPDA:CoA ligases, a function that OsASC4 (Os03g04000), the only rice gene in clade D (Fig. 2), could share. Interestingly, if this is the case, the two duplicated and highly similar poplar OPCL genes, and the single rice gene, contrast sharply in number to the expanded clade D family of potential OPCLs in Arabidopsis, suggesting that the regulation of JA biosynthesis in Arabidopsis may be more complex than in these other two angiosperm lineages.
Expression of the single Arabidopsis clade E gene, AtACS9 was downregulated after 1 h in response to wounding, and expression stayed below the control levels up to 24 h. Of the three poplar homologues most related to AtACS9, PoptrACS12 showed the most similar expression pattern, whereas PoptrACS10 and PoptrACS11 showed little or no change in expression in response to wounding. The poplar genes in clade E had similar responses to FTC, SH, and MJ treatments (PoptrACS12 was downregulated; PoptrACS10 and PoptrACS11 showed only minor fluctuations). The enzyme encoded by AtACL9 has been shown to be a fatty acyl:CoA synthetase, with activity in vitro with fatty acids and OPDA, a precursor in JA biosynthesis (Schneider et al., 2005; Kienow et al., 2008). It is localized to the peroxisome and has been suggested to be a potential OPDA:CoA ligase (Schneider et al., 2005; Kienow et al., 2008). However, the lack of wound, herbivory and MJ activation of AtACL9 and of the most closely related poplar ACS genes, as well as the phylogenetic distance between this gene and clade D containing the Arabidopsis OPCL1 gene, do not support a role for these enzymes in the stress induced synthesis of JA, and the in vivo substrate of AtACL9 remains to be clarified.
The clade E gene PoptrACS12 has distinct expression patterns relative to the most closely related genes in poplar, PoptrACS10 and PoptrACS11. It shows downregulated expression in response to wounding and related stresses (Figs 4 and 5), and shows a contrasting developmental expression pattern, with highest expression in roots, phloem and bark (Fig. 3e). Interestingly, the PoptrACS12 protein lacks the PTS1 peroxisomal targeting sequence (Fig. 2) suggesting, that while PoptrACS10 and PoptrACS11 may have enzymatic functions similar to the AtACS9 protein as peroxisomally localized fatty acyl:CoA synthetases, PoptrACS12 may have acquired a new function specific to the poplar lineage, as a nonperoxisomal CoA ligase.
Our data show that the 4CL-like ACS enzymes are a conserved, land plant-specific group of adenylate-forming enzymes most closely related to true 4CL enzymes among a larger group of plant and microbial acyl activating and acyl:CoA synthetase enzymes. The fact that ACS representatives, as well as true 4CL enzymes, are found in the moss Physcomitrella suggests that both groups of enzymes evolved early in the transition of plants to the terrestrial environment and were present in the most recent common ancestor of bryophytes and vascular plants. The ACS enzymes are largely predicted to be peroxisomally located and are related to peroxisomal enzymes found in fungi and prokaryotes. This suggests that the progenitors of land plant 4CL and the related ACS enzymes may have originally performed peroxisomal functions and that bona fide 4CL enzymes and clade A ACS enzymes evolved from such progenitor enzymes by loss of the PTS1 peroxisomal targeting sequence and acquisition of new functions in other cellular compartments. It appears that loss of peroxisomal targeting sequences can occur relatively easily in this group of enzymes, since two apparent independent examples in the duplicated poplar ACS genes (PoptrACS6 and PoptrACS12, both in clade E), and the Physcomitrella PpACS5 gene were found. The nonperoxisomal poplar proteins have presumably acquired new functions in the poplar lineage, subsequent to the gene duplication events.
The in vivo biological functions of the angiosperm ACS enzymes are for the most part still unknown, although such functions may be conserved within the conserved angiosperm clades A–E, and in many cases they have expression patterns that suggest functions in developmental and/or stress-related biochemical pathways not related to phenylpropanoid metabolism. In addition to Arabidopsis OPCL1 and its poplar homologues, other stress-induced ACS enzymes in clade D may have acyl-activating functions associated with chain-shortening reactions in JA biosynthesis. AtACS6 (At4g05160; clade B) and AtACS9 (At5g63380; clade E) have activity against a variety of acyl substrates including OPDA or its derivatives and long-chain fatty acids (Schneider et al., 2005; Kienow et al., 2008), although in vivo substrates have not been identified. Thus, many or all of the ACS proteins may accept a diversity of different fatty acid or other acyl substrates. Additional data from surveys of potential substrates (Schneider et al., 2005; Kienow et al., 2008), 4CL and ACS structural information, and in vivo functional assays (for example using reverse genetic approaches) will be necessary to determine functions of Arabidopsis ACL enzymes.
We targeted Arabidopsis ACS genes in two of the clades, A and E, with single Arabidopsis representatives (AtACS5 and AtACS9, respectively, Fig. 2) for reverse genetic analysis using T-DNA insertion null-mutant lines (data not shown). While no metabolic or morphological phenotypes were evident for a homozygous acs9 mutant line (data not shown), the acs5 mutant showed a striking male sterility phenotype (data not shown; C. de Azevedo Souza, C. Douglas et al., unpublished). This is consistent with roles for AtACS5 and the poplar homologue PoptrACS13 in anther development predicted from their developmental expression patterns (Fig. 3a). Further comparative biochemical studies on enzymes encoded by the corresponding poplar, rice, and Physcomitrella ACS genes will help to shed light on the conservation of pathways utilizing ACS enzymes in land plant lineages.