Genome-wide analysis of the chalcone synthase superfamily genes of Physcomitrella patens
Enzymes of the chalcone synthase (CHS) superfamily catalyze the production of a variety of secondary metabolites in bacteria, fungi and plants. Some of these metabolites have played important roles during the early evolution of land plants by providing protection from various environmental assaults including UV irradiation. The genome of the moss, Physcomitrella patens, contains at least 17 putative CHS superfamily genes. Three of these genes (PpCHS2b, PpCHS3 and PpCHS5) exist in multiple copies and all have corresponding ESTs. PpCHS11 and probably also PpCHS9 encode non-CHS enzymes, while PpCHS10 appears to be an ortholog of plant genes encoding anther-specific CHS-like enzymes. It was inferred from the genomic locations of genes comprising it that the moss CHS superfamily expanded through tandem and segmental duplication events. Inferred exon–intron architectures and results from phylogenetic analysis of representative CHS superfamily genes of P. patens and other plants showed that intron gain and loss occurred several times during evolution of this gene superfamily. A high proportion of P. patensCHS genes (7 of 14 genes for which the full sequence is known and probably 3 additional genes) are intronless, prompting speculation that CHS gene duplication via retrotransposition has occurred at least twice in the moss lineage. Analyses of sequence similarities, catalytic motifs and EST data indicated that a surprisingly large number (as many as 13) of the moss CHS superfamily genes probably encode active CHS. EST distribution data and different light responsiveness observed with selected genes provide evidence for their differential regulation. Observed diversity within the moss CHS superfamily and amenability to gene manipulation make Physcomitrella a highly suitable model system for studying expansion and functional diversification of the plant CHS superfamily of genes.
KeywordsPhyscomitrella patens Chalcone synthase superfamily Multigene family Enzyme evolution Gene duplication Retrotransposition Gene regulation cis-acting elements
While flavonoids are ubiquitous in vascular plants, they have been found in only ~40% of liverworts and ~50% of mosses tested (Markham 1988; Iwashina 2000). Nevertheless, it has been proposed that flavonoids played a significant role in the early evolution of land plants, first as chemical messengers and then as UV filters (Stafford 1991). Thus, studies of CHS and related enzymes in non-vascular plants are required to obtain a clearer understanding of evolution of the flavonoid pathway in plants and of evolution of land plants themselves. Although many aspects of gene regulation and evolution of the CHS superfamily of seed plants have been studied, the literature contains little information on the superfamily from non-vascular plants, except that a gene presumably encoding a CHS was shown to be light-regulated in photoautotrophic cells of the liverwort, Marchantia paleacea (Harashima et al. 2004). Recently, a CHS gene (PpCHS) was cloned from the moss, Physcomitrella patens, and PpCHS was shown to exhibit enzymatic characteristics that are very similar to those of spermatophyte CHS enzymes. Thus, PpCHS prefers p-coumaroyl-CoA as the starter substrate, produces mostly pyrone derailment products from sub-optimal substrates (e.g., hexanoyl-CoA), and shows comparable kinetic parameters to those of spermatophyte CHSs (Jiang et al. 2006).
Its informative evolutionary position and unparalleled amenability to gene targeting (Kamisugi et al. 2005, 2006) have made P. patens a popular and powerful model system for studying many plant phenomena. Furthermore, the data from several large-scale studies of the P. patens transcriptome and genome have been published (Nishiyama et al. 2003; Lang et al. 2005; Rensing et al. 2005, 2007; Quatrano et al. 2007). The 480 Mb haploid genome of the plant has been sequenced (Rensing et al. 2008), and the latest annotated version contains approximately 35,938 gene models distributed among 2,106 scaffolds, in turn composed of 19,136 contigs.
Earlier, the P. patens genome was shown to contain as many as 19 putative genes that may encode enzymes of the CHS superfamily (Jiang et al. 2006). The presence of multiple CHS superfamily genes in plants is not uncommon and seems widespread across taxa. Multiple CHS superfamily genes have been found in, to name a few, a pteridophyte Psilotum nudum (Yamazaki et al. 2001), a Pinus species (Fliegmann et al. 1992), and many flowering plants including Petunia (Koes et al. 1989), Ipomoea (Durbin et al. 2000), and Glycine max (Wingender et al. 1989). Many of the moss CHS superfamily genes share greater than 85% sequence identity to PpCHS when their deduced amino acid sequences are compared (Jiang et al. 2006). More interestingly, the P. patens genome contains PpCHS10, a moss homolog of anther-specific CHS-like genes of spermatophytes (Jiang et al. 2008; Ageez et al. 2005), and PpCHS11, which is basal to plant CHS superfamily genes in phylogenetic trees (Jiang et al. 2008). This diversity of P. patensCHS superfamily genes makes the moss a particularly attractive organism for studying evolution and gene regulation of the CHS superfamily. Herein we report our findings on gene architecture, evolution, regulatory elements and responsiveness to light of this gene group in P. patens.
Materials and methods
Genomic sequence analysis
The P. patens genes of the CHS superfamily were identified using the nucleotide and amino acid sequences of PpCHS (Jiang et al. 2006) as BLASTn and BLASTp queries, respectively, against the United States Department of Energy’s Joint Genome Institute’s (JGI) Physcomitrella genome database (http://genome.jgi-psf.org//Phypa1_1/). The GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) database in the JGI site were also term-searched for automatically annotated CHS. Each sequence obtained from these searches was examined for similarity to known CHS genes, scrutinized for an encoded catalytic CHN triad (Fig. 1c) (Ferrer et al. 1999), the GFGPG loop (Suh et al. 2000a) and the conserved Arg residues (Fukuma et al. 2007), and then grouped according to both UTR and coding sequences (CDS). ESTs corresponding to each gene were located in the P. patens EST database (www.cosmoss.org) (Lang et al. 2005). Intron structure was analyzed manually based on amino acid sequence alignments and known EST sequences. Amino acid sequence identity was calculated using the GraphAlign program (Spalding and Lammers 2004).
Full-length amino acid sequences were aligned with ClustalX (Thompson et al. 1997). Phylogenetic analysis with the Bayesian inference method was performed using the MrBayes program (v. 3.1.2) (Ronquist and Huelsenbeck 2003). In this study, the “mixed” amino acid model that allows switching between various fixed-rate amino acid models was used. Markov chain Monte Carlo analysis was performed for 500,000 generations with four independent chains. The Markov chain was sampled every 100 generations. At the end of the run, the average standard deviation of split frequencies and the potential scale reduction factors for all the run parameters were used to confirm that the state of convergence was obtained. After all trees sampled during earlier generations were discarded, a consensus tree was constructed based on the remaining trees and displayed using MEGA3.1 (Kumar et al. 2004). The Neighbor-joining (NJ) trees were built with the Poisson correction model as implemented in the MEGA3.1 program. The confidence of the tree topology was assessed by a bootstrap set of 1,000 replicates.
Gene duplication analysis
The tandem and segmental duplication history of CHS superfamily genes in P. patens was inferred by scrutinizing the scaffold sequences comprising each CHS superfamily gene and the flanking regions extending 50 kb in each direction. To investigate whether duplication of CHS genes had been mediated by (retro)transposable elements, the CHS superfamily genes and flanking DNA were scanned for overlapping sequences resembling those of known (retro)transposon families. Detection of gene conversion tracts was attempted using Geneconv (Sawyer 1989) to analyze amino acid sequences aligned in Clustal X and trimmed at the N-terminus where the alignment was poor.
Culture, light treatment and cDNA synthesis
Gametophytes of the pabB4 strain of P. patens were grown axenically and agar-free on solid medium with or without ammonium tartrate at ~21°C under continuous light supplied by Cool White fluorescent tubes (Westinghouse, Regina, SK, Canada) for approximately 3 weeks as described (Ashton et al. 1985). The Petri dishes were covered with clear resin sheets (Roscolux, No. 114; MacPhon Industries, Calgary, AB, Canada) to reduce evaporation from the growth medium. Photon flux at the surface of the medium was 36–70 μmol m−2 s−1. Cultures were then grown under light or in darkness for an additional 3 days, or were exposed to light for 3 days after growing in darkness for 3 days. Tissues were harvested, chilled in liquid nitrogen and stored (−80ºC).
Total mRNA was isolated following the protocol for the Straight A’s mRNA Isolation System (Novagen). cDNA was synthesized using an oligo-dT primer and PowerScript reverse transcriptase (Clontech) according to the manufacturer’s specifications. The cDNA mixture was quantified spectrophotometrically at 260 nm, where A260 of 1.0 was the equivalent of approximately 50 μg/mL.
Gene-specific primers used in semi-quantitative RT-PCR analysis
Primer sequence (5′→3′)
Gene and protein notation
Gene symbols are italicized, e.g., PpCHS1, whereas symbols for the corresponding proteins are non-italicized, e.g., PpCHS1.
Classification of P. patens CHS superfamily genes
CHS superfamily genes from P. patens
GenBank accession no.
JGI protein ID
Gene location Scaffold:base
Sequence identity to PpCHS (%)
Differs at 2 a.a. and 3′-UTR from PpCHS1a
No intron. Differs at 3 a.a. from PpCHS
Differs at 3′-UTR from PpCHS1a
One intron splits Cys72
Differs at 5′-UTR from PpCHS2a
Differs at 5′- and 3′-UTRs from PpCHS2a
Another copy of PpCHS3.1
Another copy of PpCHS3.1
Another copy of PpCHS5.1
One intron splits Cys76
One intron splits Cys77
Two introns split Cys88 and Gly333
Two introns split Cys135 and Arg403
Putative CHS superfamily genes of which full sequences are not available
49 bp 5′-UTR, 847 bp cds
506 bp 5′-UTR, 369 bp cds
950 bp, another copy of PpCHS2b.1
One intron splits Asn2 and Lys3 (?)
Identical to PpCHS2a in 5′-UTR and intron sequences. 16 nt deletion causing frame-shift
Two nonsense mutations, no intron.
The 1,194 bp CDS of PpCHS01 differs from the CDSs of PpCHS1a and PpCHS1c at 11 nucleotide positions, resulting in two amino acid mismatches. PpCHS3 and PpCHS5 are very similar in sequence. Between them, there are six nucleotide and three amino acid mismatches in the CDS. These differences are not due to sequencing errors since there are corresponding ESTs for each gene (see below). There are three copies and two copies of PpCHS3 and PpCHS5, respectively, indicating multiple recent gene duplication events.
PpCHS7 contains an in-frame stop codon (TGA) 366 nt downstream from the start codon, indicating that it might be a pseudogene. However, the stop codon is found at a position where Trp (TGG) is highly conserved in other enzymes of the CHS superfamily (Fig. S1). Thus, it is possible that the observed stop codon reflects a sequencing error and PpCHS7 is a functional gene. There are two ESTs derived from a cDNA clone (pphb3m11) in the cosmoss cDNA database that match the sequence of PpCHS7, although neither EST includes the sequence in question. Another notable aspect of PpCHS7 is that it encodes a Tyr residue in place of the Cys170 that is essential for catalysis and absolutely conserved in the CHS superfamily (Lanz et al. 1991). Hence, the sequence surrounding Cys170 in PpCHS7 is MMCQTGYFGGA, while in other proteins (PpCHS, PpCHS2, etc.) it is MMYQTGCFGGA (Fig. S1). Since the Cys and Tyr codons differ by a single nucleotide, here too we cannot exclude the possibility that a sequencing error has occurred. The high sequence identity of PpCHS7 (90%) and PpCHS suggests that PpCHS7, if functional, is also a CHS. However, the intriguing possibility remains that PpCHS7 is a novel enzyme with different active site architecture.
PpCHS9 is not well represented in the existing EST databases. Of the two ESTs that correspond to this gene, one (PP015028177R) matches the 5′-UTR (from −162 to −746) of PpCHS9 and the other (PP020054304R) covers 229 nt of exon 1 and 116 nt of intron. This is the only EST, among the >500 ESTs analyzed in this study, that contains a portion of intronic sequence. A plausible explanation is that PP020054304R was produced by alternative splicing, although a stop codon exists immediately downstream from the exon–intron splice site. Alternatively, the pre-mRNA may have been incompletely processed. If PpCHS9 is functional, its lower sequence identity with PpCHS (42%) could indicate that it is a non-CHS enzyme, although this has yet to be established.
PpCHS10 displays a substantially higher amino acid sequence similarity (50–60%) to spermatophyte anther-specific CHS-like enzymes (ASCL) than to any other members of the P. patens CHS superfamily (30–40%). ASCLs are specifically expressed in anthers and comprise a monophyletic clade in phylogenetic reconstructions with sequences from various angiosperm and gymnosperm representatives of the CHS superfamily (Ageez et al. 2005; Jiang et al. 2008).
The sequence of PpCHS11 contains four in-frame candidate translation initiation codons (Fig. S2). Based on similarity to the initiation consensus sequence (caA(A/C)aATGGCg) in plants (Lütcke et al. 1987; Joshi et al. 1997) and comparison of predicted amino acid sequences, obtained using the four potential start codons, with those of other P. patens CHS polypeptides, the first (most 5′) and third ATG are the most plausible candidates for the real start codon.
More CHS superfamily genes were recognized although their full-length sequences are not available. A partial sequence (896 bp) of PpCHS is located on scaffold 410, while two additional partial CHS superfamily sequences are found on scaffold 16. One of them, PpCHS13b, is 875 bp long and its sequence matches perfectly the PpCHS sequence except for two nucleotides in the CDS. The other, PpCHS13c, is 1,142 bp long and contains 228 bp of 3′-UTR. A BLASTn search of the cosmoss (www.cosmoss.org) EST database (pp1206_fil) with the PpCHS13c sequence as query yielded six ESTs containing CDSs that are identical to the PpCHS sequence. Thus, PpCHS13c probably has the same CDS as PpCHS while differing from PpCHS in the 3′-UTR. A partial sequence of a second copy of PpCHS2b is located on scaffold 76 (PpCHS2b.2). Its 950 bp-long sequence, including a 258 bp intron, is identical to the sequence of PpCHS2b.1 found on scaffold 463.
No EST sequence corresponding to PpCHS8 was found, suggesting either that PpCHS8 is not expressed under the standard culture conditions employed to obtain tissues for cDNA preparation or that it is a pseudogene. One candidate start codon, if correct, would give an intronless gene encoding a polypeptide of 359 amino acids, ~40 amino acids shorter than other enzymes of the P. patens CHS superfamily. However, upstream from this putative start codon, the well conserved sequence, G(I/V)GTA(V/N), can be discerned in MNKGRSAEGPAVILSIGTAVPPYVHE, derived by virtual translation assuming the presence of a phase-0 intron after the codon for asparagine (N) and thus inserted three nucleotides downstream from the ATG encoding the N-terminal methionine (Fig. S3), suggesting that this is the true start codon. This intron position is unique among known CHS superfamily genes, in which the intron is typically found at a highly conserved position ~180 nucleotides downstream from the start codon, suggesting that the putative intron in PpCHS8, if real, was acquired relatively recently.
Another CHS superfamily gene, PpCHSpg1, is found close to PpCHS2a on the same scaffold. The sequence of PpCHSpg1 is identical to the PpCHS2a sequence in the 5′-UTR and the intron, but differs in the 3′-UTR and more importantly in the CDS. The PpCHSpg1 CDS contains a 15 nt deletion as compared to the PpCHS2a CDS, causing a frame-shift (Fig. S4). A sequence error (deletion of a nucleotide) could be responsible for the frame-shift. However, no EST encompassing the deleted region has been found, giving further credence to the suggestion that PpCHSpg1 is a pseudogene. Also present in the genome is another putative pseudogene (PpCHSpg2) with no matching EST in the databases. Virtual translation revealed that it contains two in-frame stop codons, presumably generated by nonsense mutations that degraded the meaningful CDS of a functional precursor of PpCHSpg2 (Fig. S5).
In the seven P. patens genes, the conserved intron at Cys is a phase-1 intron (T/GT), as is the case in all plant CHS superfamily genes whose gene structures are known. While the second intron in PpCHS10 is also a phase-1 intron (G/GT), the second intron in PpCHS11 is a phase-2 intron (AG/A; Fig. 2b). Conforming to the GT-AG rule, all of the introns begin with the nucleotides ‘GT’ and end with the nucleotides, ‘AG.’ Intron length varies from 107 nt (the second intron of PpCH10) to 325 nt (the first intron of PpCHS11). No other intronic sequence similarities were noticed. Blast searches using the intronic sequences against GenBank and the P. patens genome assembly (v1.1) yielded no significant hits.
The NJ phylogram suggests progressive evolution with substantial late (moss lineage-specific) expansion of the CHS superfamily (Fig. 3a). PpCHS9 and the putative pseudogene PpCHS8 appear to have diverged relatively early in that order while PpCHS6 diverged later prior to the relatively recent evolution of the gene clusters. One cluster contains PpCHS4, PpCHS3 and PpCH5 and the other cluster is comprised of PpCHS2, PpCHS, PpCHS1, PpCHS01 and PpCHS7.
Next, the evolutionary relationship of the P. patens CHS superfamily and other plant enzymes was studied. Selected for phylogenetic analysis were the P. patensCHS superfamily genes, all known genes encoding ASCL enzymes, plus representative CHS genes of the same species from which ASCLs were taken or, if not available, CHSs from closely related species. Excluding PpCHS11, shown in a previous study (Jiang et al. 2008) to be sister to the plant CHS superfamily, the plant genes are distributed in two sister groups in a Bayesian-inferred phylogram (Fig. 3b), one of them comprised exclusively of ASCLs. As expected based on sequence similarity, PpCHS10 is included in the ASCL gene cluster (Fig. 3b). In the other sister group, PpCHS9 and PpCHS8 are basal and sister, respectively to the rest of the genes. The remaining P. patens genes form an exclusive sister group to the CHS genes from other plant taxa.
Currently, the P. patens genome sequences are assembled in contigs and scaffolds. Half of the assembly is contained in 111 scaffolds, each of which is at least 1.3 Mb in length. Of the 24 CHS superfamily genes, pseudogenes and partial gene sequences in the P. patens genome, 11 gene sequences (four pairs and one triplet) are clustered on five scaffolds. Additionally, pairs of scaffolds contain similar clusters of one or more CHS superfamily genes along with genes from other gene families.
PpCHS3.1 and PpCHS3.2 are found in head-to-head orientation with their start codons separated by ~8.8 kb, linked to a sequence similar to a lysyl-tRNA synthetase gene from Arabidopsis ~14.6 kb downstream of PpCHS3.2 (Fig. 4d). Similarly, PpCHS5.1 and PpCHS5.2 are linked head-to-head with ~18.9 kb between their start codons and in turn are linked to a second lysyl-tRNA synthetase-like sequence ~19.6 kb downstream of PpCHS5.1. PpCHS2a and PpCHS2b.1, each linked tail-to-tail to a gene encoding a chlorophyll a/b binding protein (CAB), form tandem gene pairs in a cluster including the related pseudogene, PpCHSpg1 spanning ~54.3 kb on scaffold 463 (Fig. 4e). PpCHS2c is located downstream of a third CAB gene in head-to-tail orientation on scaffold 1, with ~18.9 kb separating the start codon of PpCHS2c from the stop codon of the CAB gene.
Sequences related to (retro)transposable elements were found in the regions flanking CHS genes but did not overlap the CHS genes. Geneconv did not detect any evidence of gene conversion between CHS genes.
ESTs corresponding to PpCHS4 are found only in the cDNA library (ppgs) derived from green sporophytes (e.g., GenBank accession no. DC957761, etc.), but not in the cosmoss database, indicating that this gene is exclusively expressed in moss sporophytes. There are more than 2500 ESTs corresponding to PpCHS11 probably due to cloning artifacts, and the expression pattern of PpCHS11 could not be studied.
The CHS superfamily in P. patens
The complete complement of genes in P. patens belonging to the CHS superfamily comprises 17 genes, for 14 of which the complete CDS is known, plus 3 putative pseudogenes and multiple gene fragments. Three of these genes exist in multiple copies (PpCHS2b, PpCHS3 and PpCHS5) and all are expressed under laboratory conditions. In most cases what enzymes are encoded by these genes is unknown. Only one of the genes, PpCHS, has been demonstrated to encode a bona fide CHS (Jiang et al. 2006). Lower sequence identity (<42%) of PpCHS9, PpCHS10 and PpCHS11 to PpCHS suggests that they encode non-CHS enzymes, although this has yet to be demonstrated. However, consistent with this contention, PpCHS11 was shown to produce long-chain (C12–C18) alkylpyrones and related compounds in vitro (unpublished data) and in this respect is similar to a pentaketide resorcylic acid synthase from the fungus N. crassa (Funa et al. 2007). Moreover, in a previous study (Jiang et al. 2008), phylogenetic reconstruction demonstrated that PpCHS11 is positioned in a sister group separate from the rest of the CHS superfamily and it was proposed that PpCHS11 is an extant enzyme that may resemble more closely their most recent common ancestor than do other members of the plant CHS superfamily.
Among the P. patens genes, PpCHS10 is unique in being orthologous to seed plant ASCLs (Fig. 3b), raising the exciting possibility that its function is similar to theirs. Spermatophyte ASCLs, vital for male fertility, are expressed in the tapetum during the early stages of microspore/pollen development and probably play a role in the biosynthesis of exine (outer wall of pollen grains) (Atanassov et al. 1998; Wu et al. 2008). A recent study showed that two Arabidopsis ASCLs (AthCHSL1 and AthCHSL2 in Fig. 3b) sequentially condense a long-chain (up to C20) acyl-CoA with malonyl-CoA molecules to produce alkylpyrones in vitro (Mizuuchi et al. 2008). Our experiments indicated that PpCHS10 also functions in vitro as a long-chain alkylpyrone synthase (unpublished data). PpCHS10 is not expressed in moss gametophytes but ESTs corresponding to PpCHS10 are found in a P. patens sporophytic cDNA library derived from sporophytes harvested at different stages (thus containing developing spores) (S. Rensing, U. of Freiburg, personal communication). A case can be made that moss spores and pollen are homologous structures since both result from meiotic divisions, are protected by a multilayered coat containing sporopollenin (Domínguez et al. 1999) and germinate giving rise initially to a single filamentous apical cell or pollen tube which grows one-dimensionally. Therefore, PpCHS10 is potentially a valuable tool for studying the roles of ASCLs in spore/pollen formation and seed plant male fertility.
Since sequence identity of non-CHS enzymes to CHS from the same plant is typically <70%, the enzymes of the P. patens CHS superfamily that show sequence identity of ~90% to PpCHS are probably also CHSs (Table 2). PpCHS6, which is 86% identical to PpCHS at the amino acid level, is a sister to the remaining moss CHS genes and thus may be most directly related to the gene encoding the ancestral enzyme of the P. patens CHS group (Fig. 3a). Ancestral enzymes of the plant CHS superfamily may have employed simple acyl-CoA esters as substrates before the metabolic pathway furnishing phenylpropanoid-CoA esters (e.g., p-coumaroyl-CoA) evolved (Jiang et al. 2008). In this context, PpCHS6, if not a CHS, may be a non-CHS such as phlorisovalerophenone synthase (VPS, Paniego et al. 1999) that differs from CHS only in the choice of starter-CoA and accepts isovaleryl-CoA (Fig. 1b).
The presence of a CHS multigene family is common in seed plants. Ipomoea, Petunia, and soybean species have all been shown to have at least 6–10 CHS genes. However, the fact that the moss has as many as 17 CHS genes and gene copies (assuming that those genes with ~90% amino acid sequence identity code for CHS) is unexpected and puzzling. Chalcones, the products of CHS, are considered minor products in seed plants. CHS functions together with other downstream enzymes in the flavonoid pathway to produce secondary metabolites with diverse functions such as UV protection, antimicrobial defense and flower pigmentation (Fig. 1a). Therefore, multiple CHS genes in spermatophytes are under differential temporal and spatial regulation to serve the various branches of the pathway. In contrast, mosses do not have such a well-developed flavonoid pathway and produce fewer flavonoids (Markham 1988). Flavones, biflavones, aurones, isoflavones, and 3-deoxyanthocyanins have been found in various moss species (Basile et al. 2003; Brinkmeier et al. 1999; Geiger and Markham 1992). However, to the best of our knowledge, no phytochemical study has been carried out on the flavonoid profile of P. patens.
The P. patens genome contains two genes for chalcone isomerase (CHI) and five genes for flavanone 3-hydroxylase (F3H). CHI converts chalcone to flavanone in the second step of flavonoid pathway, and F3H converts flavanone to dihydroflavonol in the third step. Flavone synthase that also utilizes flavanone to produce flavone seems to be absent in P. patens (Fig. 1a). Putative CHS genes outnumber these downstream genes and, thus, maintenance of gene stoichiometry does not explain the retention of the observed number of CHS genes in P. patens. It is possible that, in the moss, chalcones and/or their downstream flavonoid products (albeit of simpler profile) play various physiological roles that are performed by more sophisticated metabolites in seed plants. In this case, the multiple CHS genes are probably differentially regulated in response to varying metabolic needs in different parts of the plant at different developmental stages. This argument is supported by the EST distribution data showing divergent expression patterns of the CHS genes (Fig. 5). Some groups of genes share the same CDS but have dissimilar UTRs (e.g., PpCHS2a, PpCHS2b and PpCHS2c), suggesting that these genes are differently regulated. The 5′-UTR of PpCHS3.1 diverges from those of PpCHS3.2 and PpCHS3.3 beyond 409 nt from the translation initiation site, although the three genes are classified as gene copies in this study. Thus, PpCHS3.1 may also be subjected to different regulation from PpCHS3.2 and PpCHS3.3. At least one example of different expression of such closely related genes was found in this study (see below).
Most plant CHS superfamily genes contain a single phase-1 intron in the codon for the conserved Cys (T/GT or T/GC) in the consensus sequence of (K/Q)R(M/I)C(D/E)KS. The P. patens CHS gene family exhibits an unusual collection of genes in terms of exon–intron architectures in that there are intronless genes (PpCHS, PpCHS01, PpCHS1, PpCHS3, PpCHS4, PpCHS5, PpCHS7, and PpCHS13) and genes with two introns (PpCHS10 and PpCHS11) as well as genes with one intron (PpCHS2, PpCHS6, and PpCHS9). While no other intronless CHS superfamily gene has been discovered from other plants, a few genes with more than two introns have been reported. Benzalacetone synthase (PcPKS2) from Polygonum cuspidatum is encoded by a three intron gene (Ma et al. 2009). The genes coding for Antirrhinum majus CHS (AmCHS, Sommer and Saedler 1986) as well as Arabidopsis ASCL (AthASCL2) contain two introns each (Fig. 2). All of these multi-intron genes share the first intron at the conserved Cys. Interestingly, fungal CHS superfamily genes differ from their plant counterparts in intron structure. Most fungal genes also have two introns. For example, the Aspergillus oryzae csyA gene (Seshime et al. 2005) has two intron sites at Ala59 (GC/T) and Gly274 (G/GA) (Fig. 2a). While the second intron splice site is conserved among the fungal genes, the first intron site is conserved in its position only (data not shown). Although the size of the first and second exons in the plant and fungal genes are similar, there is no apparent relatedness between plant and fungal intron splice sites. Taken together, these findings suggest that the acquisition of the first intron at the Cys in plant genes occurred at an early stage of plant evolution before the divergence of bryophytes and other lineages of plants, independently of the fungal lineage.
Unlike the first intron that is conserved in the plant multi-intron genes, the second and third introns are not shared by all. However, their positions do not seem to be random but instead are situated within the codons for conserved amino acid residues (Fig. 2b, c). PpCHS10 and PcPKS2 share the same intron splice site within the codon for a highly conserved Gly, two amino acids downstream from the catalytic His (Jez and Noel 2000; Suh et al. 2000b). The second introns of AmCHS and PcPKS2 occur at the codon for another conserved Gly adjacent to the catalytic Cys (Ferrer et al. 1999), although the AmCHS intron splice site (phase-0) occurs one nucleotide 5′ from the PcPKS2 intron splice site (phase-1). The second PpCHS11 intron splits the codon for an Arg residue, which has also shown to be highly conserved (Fukuma et al. 2007). The second AthASCL2 intron is within the codon for an amino acid residue (Glu in AthASCL2) that is typically either Glu (GAA or GAG), Asp (GAT or GAC), Asn (AAT or AAC), or Gly (GGN) in plant CHS enzymes. The number of multi-intron genes is small and it may be premature to draw any conclusions regarding the intron evolution; nevertheless, two alternative explanations are proposed. (a) The five multi-intron plant CHS superfamily genes may have gained additional introns relatively recently and independently. In this case, our data provide some support for the controversial proto-splice site hypothesis (Dipp and Newman 1989; Long and Rosenberg 2000), which postulates that intron gain occurs in the consensus exon/intron 5′ splice junction of (C/A)AG|R, where ‘|’represents the intron insertion site. According to the hypothesis, such sequences may originate and persist in the absence of introns because of coding constraints and may serve as “hot spots” for intron gain (Dipp and Newman 1989). A few plant gene families including grass catalases have been presented as supporting evidence for intron gain at proto-splice sites (Frugoli et al. 1998 and references therein). As shown in Fig. 2b, the exonic nucleotide sequences flanking the second and third introns of the multi-intron CHS superfamily genes agree well with the consensus sequence of the proposed proto-splice site. (b) An alternative proposal contends that the ancestral plant CHS gene had two or more introns, of which all but the intron at the conserved Cys have been lost in most lineages. Sequential intron loss has been observed in a few plant gene families such as terpene synthase genes (Trapp and Croteau 2001) and legumin genes (Häger et al. 1996). In these gene families, more evolved genes lack one or more introns that are present in primitive genes, which differs from the situation observed in the CHS superfamily genes. PpCHS11, the gene proposed to most closely resemble a common ancestor of plant CHS superfamily genes (Jiang et al. 2008), has a unique second intron. Furthermore, PpCHS10 and AthASCL2, both of which belong to the ASCL clade (Fig. 3b), do not share the same second intron. The fact that the insertion sites of AmCHS and PcPKS2 differ by one nucleotide also provides an argument against the ancient origin of the second intron of these genes. Distinguishing which alternative is correct requires discovery and evaluation of more multi-intron genes in this superfamily.
Phylogenetic analysis, gene structure and sequence analysis and evolution of the CHS superfamily
The importance of gene duplication in evolution is well documented (Zhang 2003). Many genes encoding enzymes of plant secondary metabolism have been recruited following gene duplication. Gene families that have evolved via this mechanism include methyltransferases and terpene synthases (Ober 2005 and references therein).
Careful consideration and comparison of the topologies of trees shown in Fig. 3, as well as of others published elsewhere (Jiang et al. 2008), allow the following inferences concerning the contributions of gene duplication to evolution of the plant CHS superfamily. (1) Gene duplications giving rise to the progenitor of PpCHS11 and the ancestor of all other CHS superfamily genes and giving rise to the progenitor of all ASCLs including PpCHS10 and the ancestor of the remaining CHS superfamily genes preceded the separation of lineages leading, respectively to mosses and tracheophytes that occurred ~400 mya (Bateman et al. 1998). (2) Tracheophyte orthologs of PpCHS11 were lost during evolution of the vascular plant lineage. (3) The genome of P. patens contains an ortholog, PpCHS10, of seed plant ASCLs, all of which were derived from the same ancestral gene. (4) Further expansion of the plant ASCL and CHS families has occurred after the separation of moss and tracheophyte lineages.
Tree topologies, gene architectural similarities and low Ks values indicate that some genes have evolved recently. Thus, intron loss and subsequent duplication have generated 10 intronless genes and their copies in P. patens (PpCHS, PpCHS01, 1a, 1c, 3, 4, 5, 7, 13b, and 13c). This is in marked contrast to other plants in which no intronless CHS gene has been discovered to date. Those sequences that have been acquired relatively recently, showing ~90% amino acid sequence identity to PpCHS, form two separate clusters in the NJ tree (Fig. 3a). PpCHS4, PpCHS3 and PpCHS5 form one cluster and the rest form the other. It is plausible that the observed clustering represents two independent evolutionary lineages. Whether evolution of these lineages has been accompanied by functional divergence requires investigation. Thus, at least one of the enzymes belonging to the PpCHS4 cluster needs to be characterized. A parsimonious mechanism that plausibly accounts for the evolution of both clusters of intronless moss CHS genes involved two sequential retrotranspositions. One such event gave rise to an intronless progenitor (indicated by a filled circle in Fig. 3a) of the PpCHS4, PpCHS3, PpCHS5 gene cluster, while the other gave rise to another intronless progenitor (also indicated by a filled circle in Fig. 3a) of the PpCHS1, PpCHS, PpCHS01 cluster. Subsequent tandem and segmental duplications of these progenitor genes and their derivatives generated the two intronless clusters (see below for further discussion).
The detection of several peaks of retrotransposon activity in the P. patens genome during the last 10 my (Rensing et al. 2008) lends support to this hypothesis. However, there is no consistent association of intronless CHS genes and retrotransposon sequences in the genome and there was no sign of reverse transcribed polyA sequences downstream from their stop codons. It should be noted that a Bayesian-inferred phylogram (not shown), similarly built using the P. patens genes, had similar overall tree topology and branch lengths to those in the NJ tree (Fig. 3a). One difference was that the intronless PpCHS4 was sister to the other seven genes comprising the two recently evolved clusters, requiring a third retrotransposition to be invoked.
Phylogenetic reconstruction, linkage analysis and comparison of gene architectures indicate both tandem and segmental duplications have contributed to expansion of the P. patens CHS superfamily. PpCHS8 and PpCHS9, although distinctly dissimilar in sequence, are linked in tandem on scaffold 25 in tail-to-tail orientation with only 1.4 kb separating their stop codons, suggesting they were produced by an ancient gene duplication (Fig. 4c). By contrast, a recent tandem duplication probably accounts for PpCHS13b and PpCHS13c, although the possibility that PpCHS13b and PpCHS13c are portions of a single gene erroneously separated by misassembly cannot be dismissed (Fig. 4b).
Three very similar CHS genes located on different scaffolds, PpCHS, PpCHS1a and PpCHS01, are linked to PAL genes that, taken pair-wise, are 97.1–98.5% identical at the nucleotide level supporting the occurrence of two relatively recent segmental duplication events (Fig. 4a). Segmental duplication of PpCHS3 and a putative lysyl-tRNA synthetase-like gene that produced PpCHS5 and a second putative lysyl-tRNA synthetase-like gene or vice versa is another recent event since PpCHS3 and PpCHS5 are 99.5% identical at the nucleotide level and the lysyl-tRNA synthetase-like genes are 97.7% identical at the nucleotide level. Even more recently, both CHS genes duplicated in tandem, since all three copies of PpCHS3 exhibit identical coding sequences as do both copies of PpCHS5. Similarly, three genes (PpCHS2a, PpCHS2b.1 and PpCHS2c) are products of tandem and segmental duplication events that probably occurred very recently since even the intronic sequences are identical. Furthermore, each possible pair of the three CAB genes linked to PpCHS2 differ by only 1 or 2 nucleotides out of the 1,220 nt that comprise the CDS and the single intron.
Nonfunctionalization does not appear to be the fate of the majority of the recently duplicated genes. One exception is PpCHSpg1, which is highly similar to the PpCHS2 genes (2a, 2b.1, 2b.2 and 2c). Thus, only one of five copies of PpCHS2 has degenerated into a pseudogene. The existence of several copies of each of PpCHS1, PpCHS2, PpCHS3, PpCHS5 and PpCHS13 in which the coding sequences of the copies are identical, suggests that neofunctionalization of gene copies has not occurred. However, it is notable that the copies of PpCHS1 and PpCHS2 differ in their UTRs, indicating that the copies may be differentially regulated and that subfunctionalization may have occurred. Subfunctionalization may be a transitional step preceding neofunctionalization (Ober 2005) and it is possible that neofunctionalization may be the eventual fate of some copies of these genes. Retention of redundant genes may be required for production of sufficient protein product of some genes. Dosage balance of genes encoding proteins that form complexes has also been suggested as a selective advantage for retention of redundant gene copies. However, gene copies that have been retained for a long period of time would be expected to contain synonymous substitutions and, possibly, non-synonymous substitutions in codons corresponding to non-critical amino acid residues. It is possible, therefore, that redundancy of CHS superfamily genes in P. patens is, at least in some cases, a temporary condition following a recent burst of duplications.
Linkage of PpCHS1a, PpCHS01, and PpCHS to PAL genes (Fig. 4a) could indicate that these genes, which act in the same pathway (Fig. 1a), are co-regulated. However, although a flavonol reductase/cinnamoyl-CoA reductase gene was found approximately 47 kb downstream from PpCHS4, no other flavonoid pathway gene was found within 50 kb of a CHS gene, indicating that physical proximity to other genes in the flavonoid pathway is not a general feature of CHS genes in P. patens. The significance, if any, of the linkage of CHS genes with CAB and LysRS genes in the moss genome is unknown.
Overrepresentation of metabolic genes in the moss genome has been noted previously and attributed to preferential retention of their paralogs after an ancient genome duplication that occurred between 30 and 60 mya (Rensing et al. 2007). This study adds CHS superfamily genes to the list of metabolic genes that are overrepresented. This may have provided mosses with the variability and plasticity of secondary metabolism that would have allowed them to adapt more readily to environmental changes.
Gene expression and light responsiveness
Since flavonoids (UV protection), isoflavonoids (Rhizobium nodulation), anthocyanins (flower color), and antimicrobial phytoalexins are all derived from polyketides synthesized by the members of the CHS superfamily, elucidating the properties of these enzymes and discovering how the genes encoding them are regulated have been important within the applied sciences of agriculture and horticulture. Thus, it is not surprising that the CHS promoter is one of the best studied plant promoters. Spatio-temporal patterns of synthesis vary substantially among the flavonoids and many cis-acting elements have been identified that play roles in the regulation of the seed plant CHS genes. For examples, the G-box (with the consensus sequence of CACGTG) is involved in UV-light induced expression of CHS genes in spermatophytes (Staiger et al. 1989). The H-box (ACCTAC) is known to play a role in feed-forward activation of the CHS promoter by p-coumarate, an upstream intermediate in the phenylpropanoid-pathway (Fig. 1a; Loake et al. 1992). The W-box (CTGACC/T) is a cis-acting element involved in pathogen- or elicitor-induction of CHS and other genes involved in plant defense (Trognitz et al. 2002). Some CHS promoters contain multiple cis-elements. Arabidopsis thalianaCHS (M20308) and Pueraria lobataCHS (D63855) genes each contain a G-box and an H-box in their promoters, while the Phaseolus vulgarisCHS15 (X59469) promoter has a G-box and two H-box elements. Several R2R3-Myb and bHLH type trans-acting factors have been found to bind to these cis-acting elements and regulate CHS gene expression (Jin and Martin 1999; Qian et al. 2007; Cominelli et al. 2008). Thus, differential interactions among cis-acting elements and trans-acting factors lead to the complex expression patterns of the CHS genes documented in seed plants (Hartmann et al. 2005).
Consistent with the central role of plant flavonoids in UV protection, CHS is highly regulated by light (Wingender et al. 1989). The P. patensCHS multigene family provides an opportunity to study the regulation of CHS genes in a non-seed plant, about which little is known. We have studied in particular the expression of representative genes of the P. patensCHS superfamily in response to light. In an experiment using a proven light responsive POR gene as a positive control, PpCHS01 and PpCHS2c exhibited light responsiveness (Fig. 6). By scrutinizing upstream DNA sequences, we found that both PpCHS01 and PpCHS2c have the G-box core sequence (ACGT) at −411 and −471 bp upstream from their ATG codons, respectively. Since the same sequence is also found in the 5′-UTRs of PpCHS1a, PpCHS2b.1, PpCHS10 and PpCHS11, the functional significance of the ACGT sequence in the light responsiveness of PpCHS01 and PpCHS2c is unclear. Further study should provide insight into the evolution of the roles of cis-acting elements in the regulation of CHS genes. It is still notable that PpCHS2a and PpCHS2c showed different responses to light, since their 5′-UTR sequences are identical up to −249 bp and must had been diverged very recently. This provides an example of differentially expressed duplicated genes (subfunctionalization).
Other notable putative cis-elements are also found in P. patens CHS genes. A W-box-like sequence (TTGACC) is found in the 5′-UTRs of PpCHS3.1 (at −935 bp from the start codon), PpCHS3.2 and PpCHS3.3 (−610), PpCHS5.1 (−1,000 and −930), PpCHS6 (−341), PpCHS9 (−670), and PpCHS10 (−661). Some of these genes were not responsive to light and we speculate that some of them may be involved in plant defense. Many genes with anther-specific expression possess, in their promoter regions, a unique sequence called an anther-box (Höfig et al. 2003). The 5′-UTR of PpCHS10 contains an 18 bp sequence (TAGAGAATGCTTGAAATC at −1149) that appears to be homologous to the anther box. This provides further support for the contention that PpCHS10 is the moss ortholog of seed plant ASCLs.
In conclusion, the combination of amenability to genetic manipulation and its diversity of CHS superfamily genes makes Physcomitrella an exciting and unique model system for further studies of CHS superfamily gene functions, regulation and evolution.
This research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the University of Regina. E. I. B. and C. C. C. are recipients of NSERC postgraduate scholarships (PGS-D and CGS-M, respectively).
- Dipp NJ, Newman AJ (1989) Evidence that introns arose at proto-splice site. EMBO J 8:2015–2021Google Scholar
- Fliegmann J, Schröder G, Schanz S, Britsch L, Schröder J (1992) Molecular analysis of chalcone and dihydropinosylvin synthase from Scots pine (Pinus sylvestris), and differential regulation of these and related enzyme activities in stressed plants. Plant Mol Biol 18:489–503CrossRefPubMedGoogle Scholar
- Gross F, Luniak N, Perlova O, Gaitatzis N, Jenke-Kodama H, Gerth K, Gottschalk D, Dittmann E, Muller R (2006) Bacterial type III polyketide synthases: phylogenetic analysis and potential for the production of novel secondary metabolites by heterologous expression in pseudomonads. Arch Microbiol 185:28–38CrossRefPubMedGoogle Scholar
- Hartmann U, Sagasser M, Mehrtens F, Stracke R, Weisshaar B (2005) Differential combinatorial interactions of cis-acting elements recognized by R2R3-MYB, BZIP, and BHLH factors control light-responsive and tissue-specific activation of phenylpropanoid biosynthesis genes. Plant Mol Biol 57:155–171CrossRefPubMedGoogle Scholar
- Loake GJ, Faktor O, Lamb CJ, Dixon RA (1992) Combination of H-box [CCTACC(N)7CT] and G-box (CACGTG) cis elements is necessary for feed-forward stimulation of a chalcone synthase promoter by the phenylpropanoid-pathway intermediate p-coumaric acid. Proc Natl Acad Sci USA 89:9230–9234CrossRefPubMedGoogle Scholar
- Markham KR (1988) Distribution of flavonoids in the lower plants and its evolutionary significance. In: Harborne JB (ed) The flavonoids. Chapman and Hall, London, pp 427–468Google Scholar
- Nishiyama T, Fujita T, Shin-I T, Seki M, Nishide H, Uchiyama I, Kamiya A, Carninci P, Hayashizaki Y, Shinozaki K, Kohara Y, Hasebe M (2003) Comparative genomics of Physcomitrella patens gametophytic transcriptome and Arabidopsis thaliana: implication for land plant evolution. Proc Natl Acad Sci USA 100:8007–8012CrossRefPubMedGoogle Scholar
- Trognitz F, Manosalva P, Gysin R, Niñio-Liu D, Simon R, del Herrera MR, Trognitz B, Ghislain M, Nelson R (2002) Plant defense genes associated with quantitative resistance to potato late blight in Solanum phureja × dihaploid S. tuberosum hybrids. Mol Plant Microbe Interact 15:587–597CrossRefPubMedGoogle Scholar