Transcriptome Comparisons Shed Light on the Pre-Condition and Potential Barrier for C4 Photosynthesis Evolution in Eudicots

Plant Mol Biol (2016) 91:193–209 DOI 10.1007/s11103-016-0455-x

Transcriptome comparisons shed light on the pre-condition and potential barrier for C4 photosynthesis evolution in eudicots

1 1 1 Yimin Tao • Ming-Ju Amy Lyu • Xin-Guang Zhu

Received: 18 August 2015 / Accepted: 14 February 2016 / Published online: 18 February 2016 Ó Springer Science+Business Media Dordrecht 2016

Abstract C4 photosynthesis evolved independently from Keywords C4 photosynthesis Á C4 evolution Á C3 photosynthesis in more than 60 lineages. Most of the C4 Transcriptome Á Flaveria Á PPT Á Stress lineages are clustered together in the order Poales and the order Caryophyllales while many other angiosperm orders do not have C4 species, suggesting the existence of bio- Introduction logical pre-conditions in the ancestral C3 species that facilitate the evolution of C4 photosynthesis in these lin- C4 photosynthesis has evolved independently from C3 eages. To explore pre-adaptations for C4 photosynthesis photosynthesis in at least 62 lineages, including 36 lineages evolution, we classiﬁed C4 lineages into the C4-poor and in the eudicots, 18 lineages in the grass family and 6 lin- the C4-rich groups based on the percentage of C4 species in eages in the sedge family (Sage et al. 2011). Most of C4 different genera and conducted a comprehensive compar- lineages are clustered together in the order Poales and the ison on the transcriptomic changes between the non-C4 order Caryophyllales while many other angiosperm orders species from the C4-poor and the C4-rich groups. Results do not have any C4 species. This biased distribution of C4 show that species in the C4-rich group showed higher species suggests that few plant groups possess appropriate expression of genes related to oxidoreductase activity, light sets of environmental, metabolic, and anatomical condi- reaction components, terpene synthesis, secondary cell tions that favor the evolution of C4 photosynthesis (Sage synthesis, C4 cycle related genes and genes related to 2001). nucleotide metabolism and senescence. In contrast, C4- Many environmental conditions that promote C4 evo- poor group showed up-regulation of a PEP/Pi translocator, lution have been proposed, such as the decline of atmo- genes related to signaling pathway, stress response, defense spheric CO2 concentration (Christin et al. 2008), high light, response and plant hormone metabolism (ethylene and warm temperature, increasing aridity (Sage 2001), wild ﬁre brassinosteroid). The implications of these transcriptomic (Osborne and Beerling 2006), migration into open canopy differences between the C4-rich and C4-poor groups to C4 system (Osborne and Freckleton 2009) and reduction of evolution are discussed. mean annual precipitation (Edwards and Smith 2010).

Some biological pre-conditions in non-C4 state that pro- Electronic supplementary material The online version of this mote C4 evolution have also been proposed. Some article (doi:10.1007/s11103-016-0455-x) contains supplementary anatomical or biochemical features that were gained in the material, which is available to authorized users. ancestral non-C4 background can facilitate the transition from C photosynthesis to C photosynthesis (Sage 2001). & Xin-Guang Zhu 3 4 [email protected] For example, large bundle sheath cells in the PACMAD clade of grasses could increase the probability of C4 evo- 1 CAS-Key Laboratory for Computational Biology and State lution in this clade (Christin et al. 2013; Grifﬁths et al. Key Laboratory for Hybrid Rice, Partner Institute for 2013). Besides, traits of salt tolerance were found to be Computational Biology, Shanghai Institute for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, present signiﬁcantly more frequently in C4 grass lineages China compared to their C3 counterparts (Bromham and Bennett 123 194 Plant Mol Biol (2016) 91:193–209

2014). Detailed comparative studies in Flaveria genus respectively (Edwards and Ku 1987; McKown et al. 2005). suggested that the re-localization of glycine decarboxylase Clade B and clade A belong to two independent lineages of

(GDC) into the bundle sheath cell and re-fixation of pho- C4 photosynthesis evolution in the genus Flaveria (Sage torespired CO2 (termed C2 photosynthesis) have evolved et al. 2011). Based on the reconstructed phylogenetic tree, before the emergence of C4 metabolic cycle and may also clade A of Flaveria genus contains 4 C3–C4 intermediate have facilitated the subsequent evolution of C4 photosyn- species and 4 C4 species (McKown et al. 2005). Hence the thesis (Sage et al. 2012). Theoretical studies using con- clade A was classified as the C4-rich group (C50 % C4 straint-based model on metabolic networks of the Flaveria abundance). However, Clade B of Flaveria genus contains species further showed that the evolution of C2 photosyn- 8C3–C4 intermediate species and 1 C4-like species but no thesis created an unbalanced nitrogen metabolism between true C4 species (McKown et al. 2005). Though the clade B bundle sheath and mesophyll cells, which could be rescued has no true C4 species identified so far, Flaveria brownii, by the recruitment of some enzymes that were later used as the C4-like species in the clade B, already has a complete components in C4 photosynthesis (Mallmann et al. 2014). C4 photosynthetic pathway, as shown by its gene expres- Recently, John et al. (2014) showed that the ratio of sion profile, gas exchange property and resource use effi- expression abundance between bundle sheath and mesophyll ciency (Monson et al. 1987; Sage et al. 2011). Therefore cells for the genes encoding core C4 enzymes are highly the clade B of Flaveria genus is an independent C4 evo- convergent in different C4 lineages, such as Zea mays, Sor- lutionary emergence (Sage et al. 2011) and was classified ghum bicolor, and Setaria viridis (John et al. 2014). Fur- as a C4-poor group in this study. RNA-seq data of F. thermore, using a deep evolutionary comparison, Aubry et al. anomala and F. ramosissima were retrieved from published (2014) showed that parallel recruitment of homologous dataset with four replicates for each species (Mallmann transcription factors for induction of C4 photosynthesis and et al. 2014). maintenance of cell specific expression patterns in two To discover more general pre-conditions towards C4 independent origins of C4 photosynthesis whose last com- photosynthesis, all eudicot species in the 1KP data set mon ancestor diverged *140 million years ago (Aubry et al. (http://www.oneKP.com/samples/list.php) were screened

2014). Furthermore, though there are metabolic variations of for sampling. All genera with the existence of C4 species C4 photosynthesis, C4 photosynthesis uses similar core were selected for C4 abundance calculation. The total enzymes for the C4 shuttle. All these raise a possibility that number of species was decided based on the estimation of there might be conserved metabolic and regulatory features The Plant List (http://www.theplantlist.org) and C4 species in the ancestors of C4 species, which facilitate the evolution number were collected from literature (Sage et al. 2011) of C4 photosynthesis from C3 ancestors. In this study, we and summarized in the Table 4. Finally, 12 eudicot C3 study this question by comparing transcriptomics profiles of species from 8 genera were identified (Table 5). RNA-seq species in genera with different C4 species abundance, i.e. data of mature leaf tissue for these 12 species (one replicate the ratio of the number of known C4 species to the total for each species) were downloaded from web data source number of species in a genera. By such a comparison, we (http://web.corral.tacc.utexas.edu/OneKP/). assume that (a) extant C3 species in these contrasting groups still reserve regulatory and metabolic features which have Minimal protein coding gene set for Arabidopsis conferred different probabilities of evolving C4 photosyn- thaliana genome thesis; (b) these different regulatory and metabolic features can be reflected in the transcriptomics data. Transcriptomics The 12 species analyzed were widely distributed in eudi- features presented in the C4-rich group but not in the C4-poor cots. Genome information is available for none of them. As group are inferred as potential pre-conditions for C4 photo- a result, the genome of Arabidopsis thaliana (A. thaliana), synthesis evolution. Features of the transcriptomic profile a representative eudicot model plant, was chosen as ref- presented in the C4-poor group but not in the C4-rich group erence for reads mapping. are potential metabolic or regulatory barriers that decrease The genome of A. thaliana experienced two whole the possibility of C4 photosynthesis evolution. genome duplication (WGD) events (WGDa and WGDb) recently (Bowers et al. 2003; Vision et al. 2000), which happened specifically within the cruifer (Brassicaceae) Materials and methods lineage (Tang et al. 2008). Considering the existence of these two WGD events and other lineage-specific gene

Taxon sampling and C4 abundance determination duplications, the minimal genome of A. thaliana was proposed and constructed for gene expression quantiﬁcation

Flaveria anomala and Flaveria ramosissima are C3–C4 across species (Brautigam et al. 2011). In our study, we intermediate species basal to clade B and clade A adopted this minimal protein coding gene set (Brautigam 123 Plant Mol Biol (2016) 91:193–209 195 et al. 2011) and updated it with TAIR10 release (Online expressed in 12 species were identified and used for fol- Resource S4). Compared to the TAIR9, there are 79 new lowing analyses. protein-coding genes added and 42 protein-coding genes removed in the TAIR10 release. For these 79 newly added Identification of the differentially expressed genes genes into TAIR10, homologous search was conducted and significantly changed pathways between against peptide sequences of all protein-coding genes in the C4-rich and C4-poor groups TAIR10 using blastp to identify potential gene duplication events (Camacho et al. 2009). Based on the alignment Procedures for identification of differentially expressed identity, 8 of these 79 protein-coding genes were identified genes and significantly changed pathways between the C4- to be duplicated. Then, we merged the duplicated gene, rich group and the C4-poor group are summarized in Fig. 1. either being segmental duplicated genes and tandem In both two data set, differentially expressed genes were duplicated genes, as one gene and added the gene ID into identified by both DESeq2 (Love et al. 2014) and edgeR the TAIR10 minimal protein-coding gene set to be used in packages (Robinson et al. 2010) with cutoff set as adjusted this study (Online Resource S4). p value \0.1. The overlapping genes identified by these two methods were identified as differentially expressed Gene assembly, annotation and transcript genes. The p values and FDRs estimated by edgeR package quantification are shown in the tables. To characterize significantly changed pathways between

For the Flaveria dataset, gene expression of F. anomala the C4-poor group and the C4-rich group, gene ontology and F. ramosissima were retrieved from published dataset (GO) enrichment analysis was conducted. In detail, sig- (Mallmann et al. 2014). For the 1KP dataset, gene nificantly enriched GO terms in those genes up-regulated in expressions were quantified after de novo assembly and the C4-poor group or the C4-rich group as compared to all annotation. In detail, RNA-seq raw reads of these 12 spe- detected genes were predicted by conducting Fishers’ exact cies were downloaded and de-novo assembled individually test implemented in agriGO (Du et al. 2010). The p value using Trinity (version trinityrnaseq_r20140717) (Grabherr was adjusted by Benjamini and Hochberg method (Ben- et al. 2011). The N50 of assembled contigs ranged from jamini and Hochberg 1995). FDR cutoff was set to be 0.05. 728 nt to 1231 nt in different species (Online Resource S3 For the 1KP dataset, considering the limited number of Table 1). The assembled contigs were further mapped differentially expressed genes detected, we used a different against TAIR10 protein database (Lamesch et al. 2012) for method to analyze the significantly changed pathways annotation applying blastx search implemented in blast? besides GO enrichment. In detail, log2 transformed suite (ncbi-blast - 2.2.28?) (Camacho et al. 2009) with a expression fold change between the C4-rich and the C4-poor cutoff e value \1e - 5. About 48.3–73.6 % of assembled group was first calculated in DESeq2 (Love et al. 2014) and contigs can be annotated by blasting against the TAIR10 edgeR packages (Robinson et al. 2010). In both softwares, protein database (Online Resource S3 Table 2). In different the fold change was calculated as the regressed expression species, 66.87–88.70 % of all detected genes had single fold change based on a generalized linear model (GLM) transcript (Online Resource S3 Table 3). For those genes (Love et al. 2014; Robinson et al. 2010). The glm considers with multiple transcripts, only 1.26–4.5 % of genes com- the gene-specific expression variation in the differentially posed multiple transcripts that are not consistently mapped expressed genes’ calling. The calculated fold changes in to the same protein of A. thaliana (Online Resource S3 these two software showed a high level of correlation Table 3). In this case, the gene was annotated as the protein (Pearson correlation = 0.945) (Online Resource S3 Fig. 1). that its longest transcript mapped to. After gene mergence Therefore, in this study, the log2 transformed expression based on our constructed A. thaliana minimal protein- fold changes calculated in edgeR package (Robinson et al. coding gene set, about 12,000 genes were detected in 12 2010) were chosen to identify significantly changed path- species. (Online Resource S3 Table 2). ways between the C4-poor group and the C4-rich group. After de novo assembly and mapping, the raw read Specifically, log2 transformed expression fold changes of count and expression level of each assembled contig were genes in a certain pathway or gene family were compared calculated with the RSEM software (Li and Dewey 2011). against the expression fold changes of the background 7096 Then the raw read counts and expression levels were added genes detected in all the studied species by conducting up and attributed to TAIR10 gene IDs based on the Wilcoxon rank-sum test, which is implemented in the annotation of contigs. Furthermore, raw read counts and MapMan software (3.5.1R2) (Thimm et al. 2004). The FPKM of duplicated TAIR10 genes were added up based p value is adjusted by Benjamini and Hochberg method on the minimal protein coding gene set constructed above (Benjamini and Hochberg 1995). FDR cutoff was set to (Online Resource S4). Finally, 6790 genes that co- be 0.1. 123 196 Plant Mol Biol (2016) 91:193–209

Fig. 1 The procedure for the identification of differentially expressed p value\0.1. GO enrichment analyses were conducted in agriGO (Du genes and significantly changed pathways in two datasets. Differential et al. 2010). Significantly changed pathways were predicted by expressed genes were estimated by both DESeq2 (Love et al. 2014) conducting Wilcoxon rank-sum test implemented in the MapMan and edgeR packages (Robinson et al. 2010) with cutoff set as adjusted software (3.5.1R2) (Thimm et al. 2004)

Results enriched in the C4-poor group were diverse, which inclu- ded 8 terms related to peptidase activity, 2 terms related to Comparison of gene expression proﬁle in Flaveria phosphatase, 2 terms related to nucleotide transporter genus activity, one term related to 1-aminocyclopropane-1-synthase, one relate to oxygen binding, etc. 4 out of the 8 GO

Firstly, detailed comparisons were conducted in Flaveria terms in biological process enriched in the C4-poor group genus, which composes two closely related C4 lineages were related to the alkene/ethylene metabolic process, (clade A and clade B) with different C4 abundance (Ed- including cellular alkene metabolic process, ethylene wards and Ku 1987; McKown et al. 2005; Sage et al. metabolic process, ethylene biosynthetic process and 2011). In particular, F. anomala and F. ramosissima, the alkene biosynthetic process. 2 out of the 8 GO terms in

C3–C4 intermediate species from clade B (C4-poor) and biological process were related to the steroid metabolic clade A (C4-rich) respectively, were compared for their process, including phytosteroid metabolic process and differences in transcriptomic levels. brassinosteroid metabolic process. 2096 protein-coding genes were differentially expressed We further examined the significantly differentially between F. anomala (the C4-poor group) and F. ramosis- expressed photosynthesis related genes (Table 2). In detail, sima (the C4-rich group) (Online Resource S1). 1061 genes the photosynthesis related genes were filtered out from all were significantly up-regulated in F. ramosissima (the C4- significantly differentially expressed genes based on their rich group) (Online Resource S1). GO enrichment analysis MapMan functional annotations (MapMan bin code starts was conducted on these differentially expressed genes with 1). (Table 1). For the GO terms in cellular component, genes As a result, there were 22 photosynthesis related genes up-regulated in F. ramosissima (C4-rich group) were sig- significantly up-regulated in F. ramosissima (the C4-rich nificantly enriched in chloroplast part, thylakoid mem- group) and 5 photosynthesis related genes significantly up- brane, thylakoid part, etc. (Table 1). For the GO terms in regulated in F. anomala (the C4-poor group). In the C4-rich molecular function, genes up-regulated in the C4-rich group, 19 out of these 22 genes were related to the light group were significantly enriched in oxygen binding, reaction in photosynthesis, including genes encoding light transferase activity, oxi-reductase, and arsenate reductase harvesting complex, subunits in photosystem I (PSI) and activity etc. subunits in photosystem II (PSII) (Table 2), which is

However, genes up-regulated in F. anomala (C4-poor consistent with the enriched GO terms presented in the C4- group) were signiﬁcantly enriched in 16 GO terms in rich group (Table 1). In detail, seven genes were related to molecular function and 8 GO terms in biological process photosystems I and cyclic electron transfer; nine genes (Table 1). The GO terms related to the molecular function were related to photosystem II; three genes were related to

123 Plant Mol Biol (2016) 91:193–209 197

Table 1 Enriched gene ontology terms of signiﬁcantly up-regulated genes in Flaveria anomala (the C4-poor group) and Flaveria ramosissima (the C4-rich group) respectively Group GO subtype GO ID Description p value Adjusted p value

C4-poor Molecular function GO:0003993 Acid phosphatase activity 1.50E-07 1.30E-04 GO:0016847 1-Aminocyclopropane-1-carboxylate synthase activity 9.60E-06 4.10E-03 GO:0016846 Carbon–sulfur lyase activity 1.60E-05 4.60E-03 GO:0004197 Cysteine-type endopeptidase activity 5.10E-05 1.10E-02 GO:0019825 Oxygen binding 9.90E-05 1.70E-02 GO:0070008 Serine-type exopeptidase activity 3.30E-04 3.10E-02 GO:0017171 Serine hydrolase activity 4.00E-04 3.10E-02 GO:0070011 Peptidase activity, acting on L-amino acid peptides 3.90E-04 3.10E-02 GO:0004185 Serine-type carboxypeptidase activity 3.30E-04 3.10E-02 GO:0008236 Serine-type peptidase activity 4.00E-04 3.10E-02 GO:0015932 Nucleobase, nucleoside, nucleotide and nucleic acid 2.80E-04 3.10E-02 transmembrane transporter activity GO:0004180 Carboxypeptidase activity 5.30E-04 3.50E-02 GO:0005337 Nucleoside transmembrane transporter activity 5.40E-04 3.50E-02 GO:0004722 Protein serine/threonine phosphatase activity 8.00E-04 4.30E-02 GO:0008238 Exopeptidase activity 8.10E-04 4.30E-02 GO:0008233 Peptidase activity 7.30E-04 4.30E-02 Biological process GO:0006952 Defense response 2.10E-05 3.90E-02 GO:0043449 Cellular alkene metabolic process 1.30E-04 4.90E-02 GO:0009692 Ethylene metabolic process 8.50E-05 4.90E-02 GO:0009693 Ethylene biosynthetic process 8.50E-05 4.90E-02 GO:0043450 Alkene biosynthetic process 1.30E-04 4.90E-02 GO:0016128 Phytosteroid metabolic process 2.10E-04 5.70E-02 GO:0016131 Brassinosteroid metabolic process 2.10E-04 5.70E-02 GO:0006555 Methionine metabolic process 3.30E-04 7.70E-02

C4-rich Cellular component GO:0044434 Chloroplast part 1.30E-05 5.60E-03 GO:0044435 Plastid part 1.00E-04 2.20E-02 GO:0034357 Photosynthetic membrane 2.80E-04 3.80E-02 GO:0009579 Thylakoid 3.60E-04 3.80E-02 GO:0009507 Chloroplast 1.00E-03 7.80E-02 GO:0044436 Thylakoid part 1.10E-03 7.80E-02 GO:0009706 Chloroplast inner membrane 1.50E-03 7.90E-02 GO:0009536 Plastid 1.30E-03 7.90E-02 GO:0009535 Chloroplast thylakoid membrane 2.20E-03 9.50E-02 GO:0055035 Plastid thylakoid membrane 2.20E-03 9.50E-02 Molecular function GO:0019825 Oxygen binding 1.90E-08 1.90E-05 GO:0016765 Transferase activity, transferring alkyl or aryl (other 1.70E-04 8.60E-02 than methyl) groups GO:0030611 Arsenate reductase activity 3.40E-04 9.20E-02 GO:0016628 Oxidoreductase activity, acting on the CH–CH group 3.70E-04 9.20E-02 of donors, NAD or NADP as acceptor p value is adjusted by Benjamini and Hochberg method (Benjamini and Hochberg 1995). FDR cutoff is set to be 0.05 oxygen evolution. Besides these light reaction related group (Table 2). In the C4-poor group, 5 photosynthesis genes, one gene encoding the fructose-1,6-bisphosphatase related genes showed signiﬁcantly higher expression as

(FBPase) in the Calvin cycle and two genes encoding the compared to the C4-rich group (Table 2). In detail, genes hydroxypyruvate reductase in the photorespiration pathway encoding the PsbZ (ATCG00300) and the PsbTc also showed signiﬁcantly higher expression in the C4-rich (ATCG00690), two subunits in the photosystem II, showed 123 198 Plant Mol Biol (2016) 91:193–209

Table 2 Signiﬁcantly differentially expressed photosynthesis related genes between Flaveria anomala (the C4-poor group) and Flaveria ramosissima (the C4-rich group) Gene ID MapMan pathway bin TAIR description log2FC p value Adjusted (C4rich/ p value C4poor)

Photosynthesis related genes signiﬁcantly up-regulated in the C4-poor group at4g26530 PS.calvin cycle.aldolase Fructose-bisphosphate aldolase, putative -0.84 1.16E-02 8.56E-02 atcg00150 PS.lightreaction.ATP synthase Encodes a subunit of ATPase complex CF0 -0.95 8.12E-03 6.67E-02 at5g07950 PS.lightreaction.other electron carrier Unknown protein -1.60 1.45E-05 5.20E-04 (ox/red).ferredoxin atcg00300 PS.lightreaction.photosystem II. PSII Encodes PsbZ, a subunit of photosystem II -2.40 2.16E-14 1.07E-11 polypeptide subunits atcg00690 PS.lightreaction.photosystem II. PSII Encodes photosystem II 5 kD protein subunit -1.03 4.29E-03 4.23E-02 polypeptide subunits PSII-T

Photosynthesis related genes significantly up-regulated in the C4-rich group atcg00600 PS.lightreaction.cytochrome b6/f Cytochrome b6-f complex, subunit V 1.32 1.93E-08 2.16E-06 atcg00590 PS.lightreaction.cytochrome b6/f Hypothetical protein 1.37 1.99E-08 2.19E-06 at3g16250 PS.lightreaction.other electron carrier NDF4, NDH-dependent cyclic electron flow 1 0.46 1.21E-02 8.80E-02 (ox/red).ferredoxin at4g32590 PS.lightreaction.other electron carrier Ferredoxin-related 0.47 7.61E-03 6.39E-02 (ox/red).ferredoxin at5g28450 PS.lightreaction.photosystem I. LHC-I Chlorophyll A–B binding protein 1.06 1.89E-06 9.54E-05 at2g20260 PS.lightreaction.photosystem I. PSI PSAE-2, photosystem I subunit E-2 0.49 1.11E-02 8.33E-02 polypeptide subunits at1g03130 PS.lightreaction.photosystem I. PSI PSAD-2, photosystem I subunit D-2 0.52 9.55E-03 7.51E-02 polypeptide subunits at1g31330 PS.lightreaction.photosystem I. PSI PSAF, photosystem I subunit F 0.51 5.72E-03 5.19E-02 polypeptide subunits at1g30380 PS.lightreaction.photosystem I. PSI PSAK, photosystem I subunit K 0.57 5.00E-03 4.71E-02 polypeptide subunits at5g64040 PS.lightreaction.photosystem I. PSI PSAN, calmodulin binding 0.62 3.75E-03 3.86E-02 polypeptide subunits at3g08940 PS.lightreaction.photosystem II. LHC-II LHCB4.2,light harvesting complex PSII 0.54 5.94E-03 5.34E-02 at1g29910 PS.lightreaction.photosystem II. LHC-II CAB3, chlorophyll A/B binding protein 3 0.58 9.26E-03 7.34E-02 at1g76570 PS.lightreaction.photosystem II. LHC-II Chlorophyll A–B binding family protein 0.66 4.20E-03 4.17E-02 at5g54270 PS.lightreaction.photosystem II. LHC-II LHCB3, light-harvesting chlorophyll 0.85 5.31E-04 9.09E-03 B-binding protein 3 at1g15820 PS.lightreaction.photosystem II. LHC-II LHCB6, light harvesting complex PSII 0.87 2.24E-04 4.67E-03 subunit 6 at2g06520 PS.lightreaction.photosystem II. PSII PSBX, PSII subunit X 0.65 5.56E-03 5.10E-02 polypeptide subunits at1g76450 PS.lightreaction.photosystem II. PSII Oxygen-evolving complex-related 0.69 3.16E-04 6.12E-03 polypeptide subunits at3g01440 PS.lightreaction.photosystem II. PSII Oxygen evolving enhancer 3 (PsbQ) family 0.89 1.91E-05 6.49E-04 polypeptide subunits protein at4g37230 PS.lightreaction.photosystem II. PSII Oxygen-evolving enhancer protein, putative 1.16 1.72E-06 8.85E-05 polypeptide subunits at1g79870 PS.photorespiration.hydroxypyruvate reductase Oxidoreductase family protein 0.42 7.86E-03 6.53E-02 at2g45630 PS.photorespiration.hydroxypyruvate reductase Oxidoreductase family protein 1.30 4.38E-08 3.93E-06 at5g64380 PS.calvin cycle.FBPase Fructose-1,6-bisphosphatase family protein 1.14 5.06E-04 8.80E-03 The photosynthesis related genes were filtered out from all significantly differentially expressed genes based on their MapMan functional annotations (MapMan bin code starts with 1) p value is adjusted by Benjamini and Hochberg method (Benjamini and Hochberg 1995). FDR cutoff is set to be 0.1

123 Plant Mol Biol (2016) 91:193–209 199

significantly higher expression in the C4-poor group. Genes the C4-rich group, which is much less than the 1061 encoding the FBP aldolase and the CF0 component of the identified genes in the Flaveria data set. This is potentially ATP synthase also showed higher expression in the C4- due to the diverse evolutionary background in this 1KP poor group. dataset, i.e. 12 C3 eudicot species from 8 genera. Besides these significantly differentially expressed pho- We analyzed those significantly up-regulated genes in tosynthesis related genes, the expression profile of major C4- the C4-rich group from the 1KP dataset (Table 6). First, a cycle related genes were also analyzed. As summarized in number of enzymes related to terpene metabolism were up-

Table 3, genes encoding phosphoenolpyruvate carboxylase regulated in the C4-rich group. These include chorismate (PEPC), phosphoenolpyruvate carboxylase kinase (PEPC- synthase (AT1G48850), shikimate kinase 1 (AT2G21940), kinase), NADP-dependent malic enzyme (NADP-ME), and mevalonate diphosphate decarboxylase (AT2G38700). NAD-dependent malic enzyme (NAD-ME), and aspartate The shikimate kinase 1 can function to increase carbon flux aminotransferase (Asp-AT) showed significant higher to specific metabolite pool in response to environmental expression in F. ramosissima (the C4-rich group) (Table 3), stress such as heat stress or during senescence stage or in indicating the recruitment of partial C4-cycle in F. ramo- embryo development (Fucile et al. 2008). The mevalonate sissima (the C4-rich group). Many of these genes were diphosphate decarboxylase (AT2G38700) is one of the related to the amino acid metabolism, i.e. either generation of final steps in the mevalonate pathway which provides carbon skeleton for amino acid synthesis (PEPC, PPCK, precursor for synthesis of triterpenes, sesquiterpenes, NADP-ME, pyruvate kinase), or the direct ammonia residue phytosterols, ubiquinone, vitamin D et al. (Simkin et al. transfer reaction (ASP). 2011). Consistent with this up-regulation of enzymes Based on these transcriptome features identified in the involved in the terpene synthesis, a number of proteins

C4-rich group, the metabolic pre-adaptations towards C4 involved in chlorophyll synthesis showed higher expres- photosynthesis were inferred as the up-regulation of sion in the C4-rich group, i.e. AT5G56260. oxidoreductase activity, light reaction components on The C4-rich group in the 1KP dataset showed higher chloroplast membrane and C4 cycle related genes, which expression of genes related to the cell wall metabolism. For is consistent with current understanding of C4 evolution example, a coumarate 3-hydroxylase (C3H), a P450-depen- in Flaveria genus (Nakamura et al. 2013;Sageetal. dent monooxygenase (AT2G40890) involved in secondary 2013). cell wall synthesis (Taylor-Teeples et al. 2015), showed higher expression. A pyridoxamine/pyridoxine-50-phos- Analyses of differentially expressed genes among 8 phate oxidase (AT2G04690), which is expressed in cell wall genera with different C4 abundance in the 1KP data and affiliated with flavin (Ahmed et al. 2015; Irshad et al. set 2008), had higher expression as well. The C4-rich group showed increased expression of AtLigB (AT4G15093), Then, we investigated the transcriptomic differences which is an aromatic ring-opening dioxygenase enzyme and among non-C4 species from more genera that have differ- involved in synthesis of Arabidopyrones (Aps). Aps are ent C4 species’ abundance. Based on the C4 abundance compounds enriched in stem (Weng et al. 2012). The higher calculation, we identified 6 genera as the C4-poor group expression of genes related to the cell wall metabolism might and 2 genera as the C4-rich group (Table 4). Transcrip- be related to the increased cell wall thickness in the bundle tomics data of 12 C3 species from these 8 distinct eudicot sheath cell wall during the C4 evolution. genera were compared to identify the pre-conditions Some proteins related to the primary metabolism also towards C4 photosynthesis (Table 5). showed higher expression in the C4-rich group as compared Raw reads of these 12 RNA-seq data sets were down- to the C4-poor group. A putative 2-hydroxyacid dehydro- loaded and de novo assembled into contigs with N50 genase (AT2G45630), a candidate gene involved in the ranging from 728 nt to 1231 nt (Online Resource S3 photorespiratory pathway, and a cytosolic pyrophosphory- Table 1). In different species, about half of assembled lase, which influences seed composition (Meyer et al. contigs could be annotated by blasting against the TAIR10 2012), showed higher expression in C4-rich group. In protein database (Online Resource S3 Table 2). Finally, addition, an uncharacterized but potential plastid solute 6790 genes that co-expressed in 12 species were identified transporter (AT2G02590) was identified (Tyra et al. 2007). and used for the following analyses. In addition to these, the non-symbiotic hemoglobin In the case of 1KP data set, only 46 genes were statis- (AT3G10520), which can scavenge NO to form nitrite and tically significantly differentially expressed between the influence the synthesis and transport of auxin (Elhiti et al.

C4-poor and the C4-rich groups, 10 of which were novel 2013), was also shown to have higher expression in C4-rich proteins with unknown function in the TAIR annotations group. Furthermore, tetrapyrrole (Corrin/Porphyrin) (Fig. 2; Table 6). 34 out of 46 genes were up-regulated in methylases, related to the synthesis of heme group of 123 200 Plant Mol Biol (2016) 91:193–209

Table 3 Summary of major C4 cycle related genes between Flaveria anomala (the C4-poor group) and Flaveria ramosissima (the C4-rich group) Enzyme Locus Name log2fold change p value Adjusted p value (C4-rich/C4-poor) beta-CA AT3G01500 beta-CA1 -0.12 8.94E – 01 9.76E – 01 PEPC AT1G53310 PEPC1 0.68 1.60E-04 3.54E-03 AT2G42600 PEPC2 0.86 7.77E-06 3.09E-04 AT3G14940 PEPC3 N.A. – – AT1G68750 PEPC4 0.65 2.47E-04 5.05E-03 PEPC-kinase AT3G04530 PPCK2 0.60 8.77E-03 7.08E-02 AT1G08650 PPCK1 -0.42 3.29E-01 6.46E-01 MDH AT3G47520 pNAD-MDH -0.33 6.15E-01 8.44E-01 AT5G43330 cNAD-MDH N.A. – – AT5G58330 NADP-MDH -0.56 1.57E-01 4.37E-01 NADP-ME AT2G19900 NADP-ME1 0.37 2.77E-02 1.53E-01 AT5G11670 NADP-ME2 0.17 1.72E-01 4.62E-01 AT5G25880 NADP-ME3 N.A. – – AT1G79750 NADP-ME4 0.48 8.15E-03 6.69E-02 NAD-ME AT2G13560 NAD-ME1 1.17 4.75E-05 1.34E-03 AT4G00570 NAD-ME2 -0.47 1.75E-01 4.65E-01 PCK AT4G37870 PEPCK1 0.51 1.97E-02 1.22E-01 AT5G65690 PEPCK2 0.46 8.26E-02 3.01E-01 PPDK AT4G15530 PPDK 0.25 1.28E-01 3.90E-01 PPDK-RP AT4G21210 PPDK-RP1 -0.04 6.79E-01 8.75E-01 AT3G01200 PPDK-RP2 -0.01 5.91E-01 8.31E-01 Ala-AT AT1G72330 AlAAT2 N.A. – – AT1G17290 ALAAT1 0.16 1.35E-01 4.02E-01 Asp-AT AT2G30970 ASP1 0.10 2.13E-01 5.15E-01 AT5G19550 ASP2 0.77 5.04E-05 1.40E-03 AT5G11520 ASP3 1.00 8.09E-07 4.80E--05 AT1G62800 ASP4 0.35 2.23E-02 1.33E-01 AT4G31990 ASP5 -0.13 8.35E-01 9.54E-01 PPT AT5G33320 PPT1 -0.14 9.06E-01 9.81E-01 AT3g01550 PPT2 -0.21 – – Dit AT5G12860 Dit1 -0.24 8.30E-01 9.52E-01 AT5G64290 Dit2.1 N.A. – –

Genes encoding PEPC, PEPC-kinase, NADP-ME, NAD-ME and Asp-AT were predicted to be up-regulated in C4-rich group signiﬁcantly, which are indicated in bold black. Transcriptome of Flaveria anomala and Flaveria ramosissima were retrieved from published data (Mallmann et al. 2014) p value is adjusted by Benjamini and Hochberg method (Benjamini and Hochberg 1995). FDR cutoff is set to be 0.1 PEPC phosphoenolpyruvate carboxylase, PEPC-kinase phosphoenolpyruvate carboxylase kinase, MDH malate dehydrogenase, NADP-ME NADP-dependent malic enzyme, NAD-ME NAD-dependent malic enzyme, PCK phosphoenolpyruvate carboxykinase, PPDK pyruvate orthophosphate dikinase, PPDK-RP pyruvate orthophosphate dikinase regulatory protein, Ala-AT alanine aminotransferase, Asp-AT aspartate aminotransferase, PPT PEP/Pi translocator, Dit dicarboxylate transporter

hemoglobin, showed higher expression in the C4-rich group (AT1G65420) was identiﬁed (Jung and Niyogi 2010). In as well. Considering that all these genes can either directly addition, AT2G22650, a FAD-dependent oxireductase or indirectly related to the amino acid metabolism, it is protein with no functional characterization, showed higher likely that the C3 species in the C4-rich group might have expression in the C4-rich group. A purple acid phosphatase, specialized amino acid metabolism. a member of the metallo-phosphoesterase family

The C4-rich group in the 1KP data set showed strong (AT5G57140), also showed higher expression. perturbation of redox metabolism. A chloroplast-localized We found higher expression of genes related to

YCF20-like gene involved in photochemical quenching nucleotide metabolism in the C4-rich group. A gene

123 Plant Mol Biol (2016) 91:193–209 201

Table 4 The percentage of C 4 Genus C species number Total species number C abundance group species in 8 genera used in this 4 4 study Aerva 414C4-poor

Alternanthera 15 87 C4-poor

Cleome \3 206 C4-poor

Mollugo 218C4-poor

Sesuvium 111C4-poor

Flaveria 523C4-poor

Heliotropium 120 156 C4-rich

Atriplex 200–300 258 C4-rich

6 of these 8 genera are labeled as C4-poor group while the other 2 genera are labeled as C4-rich group. The information of C4 species number was retrieved from Sage et al. (2011). The number of total species for each genus was retrieved from http://www.theplantlist.org

Table 5 The C3 species and sample tissues used in this study from involved in DNA polymerase activity, POLD2 the 1KP data set http://1KP-project.com/blast.html (at2g42120), showed increased expression in the C4-rich Group 1KP ID Species Tissue Type group. An RNA methyltransferase (NSUN5; AT5G26180), which speciﬁcally methylates nuclear large subunit 25S C -poor VKOE Aerva lanata Mature leaf C 4 3 rRNA, showed higher expresion in C4-rich group (Burgess KTQI Alternanthera brasiliana Mature leaf C 3 et al. 2015). C4-rich group showed increased expression LUNL Alternanthera sessilis Mature leaf C3 of dUTP-Pyrophosphatase like 1 (DUT1, AT3G46970) HURS Mollugo pentaphylla Mature leaf C3 (Dubois et al. 2011). Finally, the nudix hydrolyse OPZX Sesuvium verrucosum Mature leaf C3 (AT5G19460), which hydrolyzes ribonucleoside and UPZX Cleome viscosa Mature leaf C3 deoxyribonucleoside triphosphates, nucleotide sugars etc., KJBH Flaveria cronquistii Mature leaf C3 also showed higher expression. These increased expression RZSW Flaveria pringlei Mature leaf C3 of genes related to nucleotide metabolism might be related

C4-rich RGMN Heliotropium calcicola Mature leaf C3 to the stronger demand for nucleotide maintenance and

NIGS Heliotropium karwinskyi Mature leaf C3 repair.

AKTA Atriplex hortensis Mature leaf C3 In addition, many genes earlier shown to be associated

EPVF Atriplex prostrata Mature leaf C3 with senescence were up-regulated in the C4-rich group. These genes include AT2G03350, AT4G29400,

Fig. 2 Distribution of 46 differentially expressed genes in the background of log2 transformed fold changes of total genes in the 1KP data set

123 202 Plant Mol Biol (2016) 91:193–209

Table 6 Signiﬁcantly differentially expressed genes between the C4-poor and the C4-rich groups in the comparison of 1KP data set Locus Name TAIR description log2FC p value Adjusted (C4rich/ p value C4poor)

Genes signiﬁcantly up-regulated in the C4-poor group AT1G07870 Protein kinase superfamily protein -1.50 3.73E-04 6.50E-02 AT1G53035 Unknown -2.28 4.52E-04 7.16E-02 AT1G53040 Unknown -1.54 5.43E-05 1.52E-02 AT1G66950 ABCG39 Encodes a plasma membrane-localized ABC transporter. -0.90 6.74E-04 8.78E-02 AT3G01550 ATPPT2 Phosphoenolpyruvate (pep)/phosphate translocator 2 -3.00 2.23E-05 1.03E-02 AT3G17100 AIF3 DNA binding transcription factors -1.70 6.72E-04 8.78E-02 AT3G22890 APS1 Encodes ATP sulfurylase, the ﬁrst enzyme in the sulfate assimilation -3.41 2.89E-05 1.11E-02 AT4G26270 PFK3 Phosphofructokinase 3 -1.21 3.97E-04 6.76E-02 AT4G34500 Protein kinase superfamily protein -1.60 3.03E-04 5.72E-02 AT4G37310 CYP81H1 CYTOCHROME P450, FAMILY 81, SUBFAMILY H, -1.17 4.18E-04 6.81E-02 POLYPEPTIDE 1 AT5G43745 Unknown -1.02 4.70E-06 3.80E-03 AT5G52660 Homeodomain-like superfamily protein -2.31 4.72E-04 7.23E-02

Genes signiﬁcantly up-regulated in the C4-rich group AT1G03687 Unknown 1.55 5.83E-08 4.07E-04 AT1G26640 Amino acid kinase family protein 0.78 4.20E-04 6.81E-02 AT1G45110 Tetrapyrrole (Corrin/Porphyrin) Methylases 1.48 5.81E-04 7.95E-02 AT1G47710 ATSERPIN1 Serine protease inhibitor (SERPIN) family protein 3.44 5.27E-06 3.80E-03 AT1G48850 EMB1144 Embryo defective 1144, chorismate synthase activity 1.92 9.53E-05 2.38E-02 AT1G54310 S-adenosyl-L-methionine-dependent methyltransferases superfamily 0.69 6.28E-06 3.98E-03 protein AT1G65420 NPQ7 Chloroplast localized YCF20-like gene involved in nonphotochemical 1.64 5.03E-05 1.46E-02 quenching AT1G76150 ATECH2 Encodes a monofunctional enoyl-CoA hydratase 2 2.08 6.79E-04 8.78E-02 AT2G02590 Unknown 1.44 4.82E-04 7.23E-02 AT2G03350 Unknown 2.09 5.59E-04 7.80E-02 AT2G04690 Pyridoxamine 50-phosphate oxidase family protein 3.32 2.03E-07 7.08E-04 AT2G04900 Unknown 1.97 3.18E-05 1.11E-02 AT2G21940 ATSK1 Shikimate kinase 1.64 7.45E-04 9.28E-02 AT2G22650 FAD-dependent oxidoreductase family protein 0.57 5.58E-04 7.80E-02

AT2G36930 Zinc ﬁnger (C2H2 type) family protein 1.90 5.99E-05 1.61E-02 AT2G38700 ATMVD1 Encodes mevalonate diphosphate decarboxylase 0.20 2.55E-04 5.23E-02

AT2G40890 CYP98A3 Encodes coumarate 3-hydroxylase (C3H), a P450-dependent 1.61 1.12E-04 2.52E-02 monooxygenase AT2G42120 POLD2 DNA POLYMERASE DELTA SMALL SUBUNIT 0.66 1.35E-05 6.75E-03 AT2G42760 Unknown 1.56 3.97E-05 1.26E-02 AT2G45630 D-isomer speciﬁc 2-hydroxyacid dehydrogenase family protein 1.96 2.69E-06 3.13E-03 AT3G08600 Unknown 1.41 8.01E-05 2.07E-02 AT3G10520 AHB2 Encodes a class 2 non-symbiotic hemoglobin 3.21 5.42E-06 3.80E-03 AT3G46940 DUT1 DUTP-PYROPHOSPHATASE-LIKE 1 2.63 2.77E-05 1.11E-02 AT3G53620 ATPPA4 Encodes a soluble protein with inorganic pyrophosphatase activity 1.94 1.04E-04 2.42E-02 AT4G08790 Nitrilase/cyanide hydratase and apolipoprotein N-acyltransferase family 0.38 5.02E-04 7.29E-02 protein AT4G15093 ATLIGB Catalytic LigB subunit of aromatic ring-opening dioxygenase family 2.93 2.39E-04 5.05E-02 AT4G29400 Unknown 1.51 7.00E-04 8.88E-02 AT4G35750 SEC14 cytosolic factor family protein 3.16 1.01E-05 5.89E-03 AT4G38020 tRNA/rRNA methyltransferase (SpoU) family protein 0.45 4.87E-04 7.23E-02

123 Plant Mol Biol (2016) 91:193–209 203

Table 6 continued Locus Name TAIR description log2FC p value Adjusted (C4rich/ p value C4poor)

AT5G19460 ATNUDT20 Nudix hydrolase homolog 20 2.10 1.38E-06 2.34E-03 AT5G26180 S-adenosyl-L-methionine-dependent methyltransferases superfamily 1.07 2.36E-05 1.03E-02 protein AT5G51830 pfkB-like carbohydrate kinase family protein 2.07 3.33E-05 1.11E-02 AT5G56260 Ribonuclease E inhibitor RraA/Dimethylmenaquinone 3.09 1.03E-04 2.42E-02 methyltransferase AT5G57140 ATPAP28 Purple acid phosphatase 28 (PAP28) 0.75 3.52E-04 6.45E-02 p value is adjusted by Benjamini and Hochberg method (Benjamini and Hochberg 1995). FDR cutoff is set to be 0.1

AT4G35750, At1G47710, and AT1G76150. Each of these pathway/gene family were compared against the expression genes have been shown earlier to be up-regulated in fold changes of whole expressed gene set as background to senescence plant or tissue (Ascencio-Ibanez et al. 2008; detect signiﬁcantly changed pathways by conducting Wil- Gepstein et al. 2003; Goepfert et al. 2006; Lampl et al. coxon rank-sum test (Table 7). This pathway-level compar- 2013) or under abiotic stress (Ciftci-Yilmaz et al. 2007; ison shows that protein synthesis and gene family of Nudix

Kreps et al. 2002). Again, these might suggest that C3 hydrolases were significantly up-regulated in the C4-rich plants in the C4-rich group might experience more severe group (Table 7). In the C4-poor group, however, pathways of stress as compared to C3 plants in the C4-poor group. signaling, hormone metabolism, stress response and protein Out of the 12 genes showing higher expression in the C4- degradation were significantly up-regulated (Table 7). poor group, two genes related to plant primary metabolism were identified, i.e. the phosphofructokinase (PFK), an enzyme involved in glycolysis, and phosphoenolpyruvate/ Discussion phosphate transporter (PPT2). C4-poor group showed higher expression of a number of genes associated with oxidative This study aims to identify difference in transcriptomic stress, i.e., a putative cytochrome P450 protein (AT4G37310) profile between non-C4 plants in groups with different (Stanley Kim et al. 2005), AT1G53035 (Charron et al. 2008) abundances of C4 species. We identified major transcrip- and AtPDR11 (AT1G66950). Two transcription factors, i.e. a tomic features in the C4-rich group, which are potential basic helix-loop-helix protein modulating brassinosteroid metabolic or regulatory pre-conditions for C4 photosynthesis (BR) signaling (AT3G17100; AIF3) (Wang et al. 2009)anda evolution. These features include up-regulation of genes homeodomain-like superfamily (AT5G52660) involved in related to oxidoreductase activity, terpene synthesis, light dehydration response (Ding et al. 2013), showed increased reaction components, C4 cycle related genes and genes expression in the C4-poor group. An enzyme involved in related to nucleotide metabolism and senescence (Tables 1, sulfur metabolism, i.e. the ATP sulfurylase, the first enzyme in 6, 7). In addition, we identified major transcriptomic features the sulfate assimilation (AT3G22890), showed higher presented in the C4-poor group, which are potential meta- expression in the C4-poor group (Bohrer et al. 2015). Two bolic barriers that decrease the possibility of C4 photosyn- protein kinases (AT1G07870 and AT4G34500) were identi- thesis evolution. The identified genes that showed increased

ﬁed to have higher expression in the C4-poor group, however, expression in the C4 poor group include the up-regulation of the function of them is unknown. a PEP/Pi translocator, genes involved in signaling pathway, stress response, defense response and plant hormone meta- Analyses of the signiﬁcantly changed pathways bolism (ethylene and brassinosteroid) (Tables 1, 6, 7). In this in genera with different C4 abundance in the 1KP session, we discuss the implications of the major differences data set between the C4-rich and the C4-poor groups to C4 evolution.

The limited number of differentially expressed genes detected General transcriptomic features associated cannot be analyzed by GO enrichment analysis. To solve this with the C4-rich group issue, we designed a new method to analyze the transcriptome proﬁles and identify differentially expressed pathways. In C4 photosynthesis was proposed to evolve through a step- particular, expression fold changes of genes in a certain wise manner, i.e. each of the individual features were

123 204 Plant Mol Biol (2016) 91:193–209

Table 7 Signiﬁcantly changed pathways between the C4-poor and the C4-rich groups in the comparison of 1KP dataset Category Pathway #Genes attributed in p value Adjusted p value

C4-poor Signalling 339 1.99E-04 2.74E-02 Signalling.receptor kinases 87 1.93E-07 1.15E-04 Hormone metabolism 96 5.58E-04 6.52E-02 Hormone metabolism.brassinosteroid.synthesis-degradation 11 6.74E-04 6.71E-02 Stress 174 1.99E-04 3.39E-02 Protein.degradation.ubiquitin.E3 187 6.00E-04 6.52E-02 Protein.degradation.ubiquitin.E3.RING 103 1.45E-04 2.90E-02

C4-rich Protein.synthesis 248 4.20E-06 1.67E-03 Protein.synthesis.ribosomal protein 132 1.36E-08 1.63E-05 Protein.synthesis.ribosomal protein.prokaryotic 69 6.98E-05 2.09E-02 Protein.synthesis.ribosomal protein.eukaryotic 55 2.94E-04 4.39E-02 Nucleotide metabolism.salvage.NUDIX hydrolases 7 5.03E-04 6.52E-02 Each MapMan pathway bin was tested by Wilcoxon rank sum test followed by multi-hypothesis correction (Benjamini and Hochberg method). The signiﬁcantly changed pathways are classiﬁed into two categories, one showing higher expression in the C4-poor group and another showing higher expression in the C4-rich group. FDR cutoff is set to be 0.1

gained independently and potentially sequentially even expressed in the chloroplast of rosette leaves (Ogawa et al. though different C4 lineages might have taken different 2005, 2008). Nudix hydrolases play an important role in sequences of feature acquisition (Sage 2004; Williams maintaining cellular homeostasis by hydrolyzing oxidized et al. 2013). The two datasets analyzed in this study were nucleotides, which could decrease the probability of errors used to study two major transitions during the C4 evolution, in DNA replication or RNA transcription (Kraszewska i.e. the transition from C3 photosynthesis to C4 photosyn- 2008). Overexpression of AtNUDX2 can enhance the tol- thesis (the 1KP data set) and the transition from C3–C4 erance to oxidative stress in A. thaliana (Ogawa et al. intermediate state to the C4 photosynthesis (the Flaveria 2009). data set). We first discuss the major features identified from Furthermore, we observed increased expression of genes the 1KP data set. The C4-rich group in the 1KP data set related to nucleotide metabolism, including methyltrans- showed higher expression of genes related to terpene ferase (NSUN5; AT5G26180) and dUTP-Pyrophosphatase metabolism and also secondary cell wall synthesis. like 1 (AT3G46970). dUTP-Pyrophosphatase like 1 is Increasing terpene-related product can help plants cope needed for hydrolysis of dUTP to dUMP and critical for with stresses, such as heat (Fucile et al. 2008) while protection of NDA from uracil incorporation (Dubois et al. increasing secondary cell wall synthesis can potentially be 2011). The increased expression of these enzymes related related to resistance or tolerance to drought (Griffiths et al. to nucleotide metabolism, together with the high expres-

2013). Both stresses are closely related to the emergence of sion of AtNUDX2, in the C4-rich group suggests that the C4 photosynthesis (Sage 2004). The C4-rich group also C3 ancestors of the C4 plants have higher demand for DNA showed higher expression of genes related to redox regu- protection in the initial stage of C4 evolution, which is lation, e.g. YCF20, AT2G22650, which suggests that the consistent with the environment conditions which pro-

C3 ancestors of C4 photosynthesis might have perturbed moted the evolution of C4 photosynthesis. The emergence cellular redox metabolism. of C4 photosynthesis has been shown to be highly corre- At the pathway level, gene family of Nudix hydrolases lated with reduction of annual rainfall (Edwards and Smith showed signiﬁcant up-regulation in the C4-rich group 2010), movement into open habitats (Osborne and Freck- (Table 7). The Nudix hydrolases are a group of enzymes leton 2009), and reduction of atmospheric CO2 levels that catalyze the hydrolytic breakdown of nucleoside (Christin et al. 2008). All these factors can potentially lead diphosphates linked to some other moiety (NU-dix) to over-reduction of electron transfer chain and hence (Bessman et al. 1996). In A. thaliana (A. thaliana), there generation of reactive oxygen species, which can cause are 24 genes encoding Nudix hydrolases, which are widely oxidative stress or damage or senescence. Consistent with expressed in different cellular compartments, such as this reasoning, in the C4-rich group, many genes related to chloroplast, mitochondira and cytosol (Ogawa et al. 2005, senescence indeed showed higher expression levels as

2008). The ATNUDT20 was reported to be speciﬁcally compared to the C4-poor group (Table 6).

123 Plant Mol Biol (2016) 91:193–209 205

In the comparison between F. anomala (the C4-poor effectively (Gowda et al. 2011; Lynch 2013; Uga et al. group) and F. ramosissima (the C4-rich group), 19 genes 2013). Alternations in the leaf anatomical features, such as related to the light reaction showed significant higher increased vein density, have been postulated to confer expression in the C4-rich group as compared to the C4-poor plants with different benefits, including increased tolerance group. These include genes encoding light harvesting to high temperature and drought (Griffiths et al. 2013; Sack complex, subunits in PSI and PSII (Table 2). These results and Scoffoni 2013; Sage et al. 2012; Tholen et al. 2012). suggest that the capacity of photosynthetic electron flow in So far, no systematic study on difference in stress tol-

F. ramosissima (the C4-rich group) might be critical for the erance between C3 species with different probability of establishment of C4 metabolic cycle from the C3–C4 evolving C4 photosynthesis is available. However, C4 intermediate species. Considering the increased possibility plants have been shown to be highly sensitive to water of oxidative stress during the initial stage of C4 evolution, a stress (Ghannoum 2009) and C4 plants do lost their higher linear electron transport capacity could be a strategy advantage in photosynthetic efficiency comparing to their to use the excess light energy. Furthermore, C4 photosyn- C3 counterparts under drought conditions (Ripley et al. thesis requires two extra ATP for fixing one CO2. The extra 2007). All these evidences suggest a possibility that demand of ATP in C4 plants can be fulfilled by increasing ‘weaker C3 ancestors’ (less adapted to stresses) evolved C4 cyclic electron flow around PSI (Kanai and Edwards 1999). photosynthesis, while species capable of coping with these

Earlier comparative studies in Flaveria genus showed that stress conditions (drought, high temperature or low CO2 major cyclic electron ﬂow components (PGR5 and NDH- etc.) managed to survive through these stresses with

H) increase remarkably in the C4-like and the C4 Flaveria diverse stress response mechanisms and hence maintained species but not in the C3 and the C3–C4 intermediate the C3 photosynthesis. Systematic study is needed to test Flaveria species (Nakamura et al. 2013). Analysis from this hypothesis. this study further shows an higher expression of genes involved in the cyclic electron transfer, such as components The gene encoding a PEP/Pi translocator of photosystems I, cytochrome b6f, NDF4 etc. in Flaveria is up-regulated in the C4-poor group ramossisima in C4-rich clade A, as compared to F. anomala in the C4-poor clade B (Table 2). Taken together, the To complete a C4 cycle, PEP needs to be transported out of up-regulation of both linear and cyclic electron transfer the chloroplast in mesophyll cells through a PEP/Pi related genes observed might be a pre-condition for C4 translocator (Weber and von Caemmerer 2010). Several photosynthesis evolution. transcriptomic analyses have conﬁrmed the up-regulation

of PEP/Pi translocator genes in different C4 species General transcriptomic features observed (Brautigam et al. 2011; Gowik et al. 2011). In our com- in the C4-poor group parison among 8 eudicot genera, a gene encoding a PEP/Pi translocator (orthologous to ATPPT2, AT3G01550 in A.

Compared to the C4-rich group, genes related to stress thaliana) showed a significantly higher expression in the responses showed higher expression in the C4-poor group. C4-poor group but not the C4-rich group (Fig. 3; Table 6), These include genes involved in biotic stress response and which suggests that the up-regulation of PEP/Pi translo- defense response, together with those genes involved in cator in the C4-poor group might have been a barrier during hormone metabolism (Tables 1, 7). These observations the evolution of C4 photosynthesis. These results suggest a suggest that species in the C4-poor group have more different metabolic circuit on PEP (phosphoenolpyruvate) mechanisms to cope with stresses; in another word, the C4 utilization between the C4-poor and the C4-rich groups. In photosynthesis might have been just one of many potential C3 photosynthesis, PEP is mostly produced in cytosol mechanisms plants use to survive through or cope with (through glycolysis, for example) and transported into those stress conditions which promoted the evolution of C4 chloroplasts for shikimate pathway (Herrmann and Weaver photosynthesis, such as drought, low CO2 etc. This is in 1999). The up-regulation of a phosphoenolpyruvate phos- line with the fact that the total number of C4 species in the phate translocator (PEP/Pi translocator; AT3G01550, world, estimated to be around 7500 (Sage et al. 2011), is ATPPT2) and a phosphofructokinase (AT4G26270) in the far less than the total number of C3 species (Sage et al. C4-poor group are in line with such a possibility. 2011), even though theoretically the photosynthetic energy In A. thaliana, there are two distinct PEP/Pi transloca- conversion efficiency of C4 plants is much higher than C3 tors, ATPPT2 (AT3G01550) and ARAPPT (AT5G33320; plants in a broad range of environmental conditions (Zhu namely the ATPPT1 in literatures), which show different et al. 2008). The potential mechanisms that C3 species use expression profiles (Knappe et al. 2003). In maize, the to cope with different stresses are diverse. For example, homolog of the ARAPPT (AT5G33320) is highly expres- plants with deep roots can cope with drought more sed in leaf (Brautigam et al. 2008) and presents preferential 123 206 Plant Mol Biol (2016) 91:193–209

Fig. 3 Expression of two PEP/ Pi transporters (PPT) in 12 C3 species in the 1KP data set. These two PPTs are homologs of ATPPT2 and ARAPPT in Arabidopsis thaliana

Fig. 4 The up-regulation of ATPPT2 might be an obstacle for C4 evolution in the C4-poor group. a The proposed major ﬂuxes in C3 species of the C4- poor group where the PEP is produced in cytosol and largely imported into Chloroplasts for the shikimate pathway; b Less PEP is imported into chloroplasts in C3 species of the C4-rich group

accumulation in mesophyll cells compared to bundle the C4-poor group may have increased glycolysis for the sheath cells at the protein level (Majeran and van Wijk generation of PEP in the cytosol (Table 6). The up-regulation

2009), indicating the C4 version PEP/Pi translocator is of ATPPT2 might have been an obstacle for C4 evolution in recruited from the gene homolog of ARAPPT the C4-poor group, which prevents the establishment of C4 (AT5G33320) but not ATPPT2 (AT3G01550). In our metabolic cycle (Fig. 4). study, two homologs of PEP/Pi translocator were both Furthermore, the up-regulation of ATPPT2 could detected and quantiﬁed in all 12 species (Fig. 3). The non- increase the shikimate pathway, which is responsible for

C4 version PEP/Pi translocator (homolog to ATPPT2, producing chorismate in plants. Chorismate is the precursor AT3G01550), but not the C4 version PPT (homolog to for aromatic amino acids (Phe, Tyr and Trp) (Herrmann ARAPPT, AT5G33320), showed signiﬁcant up-regulation 1995; Herrmann and Weaver 1999). A variety of secondary in the C4-poor group. metabolites are synthesized from these three aromatic amino In the background of C3 photosynthesis, PEP is only pro- acids, such as phenylpropanoids, alkaloids and hormones duced in cytosol and transported into chloroplast for the (Maeda and Dudareva 2012), which help plants cope with shikimate pathway (Fischer et al. 1997; Flugge et al. 2011). stress conditions. For example, phenylpropanoid compounds

The up-regulation of ATPPT2 in the C4-poor group would are responsive to various biotic and abiotic stresses, such as import PEP from cytosol into chloroplasts in mesophyll cells, high light/UV, wounding, low temperature, low nitrogen and which is opposite to the direction of PEP transporting in a C4 pathogen attack (Dixon and Paiva 1995). This is also con- cycle. Besides, the up-regulation of phosphofructokinase, the sistent with the up-regulation of stress response related

ﬁrst committed step of glycolysis, suggests that ancestors of genes in the C4-poor group (Tables 1, 7).

123 Plant Mol Biol (2016) 91:193–209 207

Conclusions manuscript, and generated the ﬁgures. All authors read and approved the ﬁnal manuscript.

In this study, we aim to identify potential metabolic or Compliance with ethical standards regulatory pre-conditions for C4 photosynthesis evolution by comparing transcriptome between extant non-C4 spe- Conflict of interest The authors declare no competing interests. cies from plant lineages with different C4 abundance. Our analyses show that species in the C4-rich group have References higher expression of genes related to oxidoreductase activity, secondary cell wall synthesis, nucleotide meta- Ahmed FH et al (2015) Sequence-structure-function classification of a bolism, terpene metabolism, light reaction components, C4 catalytically diverse oxidoreductase superfamily in mycobacte- cycle related genes, Nudix hydrolases, 2-hydroxyacid ria. J Mol Biol 427:3554–3571 dehydrogenase, and cytosolic pyrophosphorylase etc. In Ascencio-Ibanez JT, Sozzani R, Lee TJ, Chu TM, Wolfinger RD, Cella R, Hanley-Bowdoin L (2008) Global analysis of Ara- addition, metabolic features presented in the C4-poor bidopsis gene expression uncovers a complex array of changes group but not in the C4-rich group were identified, which impacting pathogen response and cell cycle during geminivirus include the up-regulation of a PEP/Pi translocator, genes infection. Plant Physiol 148:436–454 related to signaling pathway, stress response, defense Aubry S, Kelly S, Kumpers BMC, Smith-Unna RD, Hibberd JM (2014) Deep evolutionary comparison of gene expression response and plant hormone metabolism (ethylene and identifies parallel recruitment of trans-factors in two independent brassinosteroid). Based on these observations, we propose origins of C4 photosynthesis. PLoS Genet 10(6):e1004365 that C4 photosynthesis might have evolved from those Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: weak C ancestors, which lack other efficient mechanisms a practical and powerful approach to multiple testing. J R Stat 3 Soc B Met 57:289–300 to cope with stress conditions that lead to C4 evolution. A Bessman MJ, Frick DN, O’Handley SF (1996) The MutT proteins or potential mechanistic linkage between the high expression ‘‘nudix’’ hydrolases, a family of versatile, widely distributed, of PEP/Pi translocator and decreased probability of C4 ‘‘housecleaning’’ enzymes. J Biol Chem 271:25059–25062 evolution is also proposed. One caveat in this study is that Bohrer AS, Yoshimoto N, Sekiguchi A, Rykulski N, Saito K, Takahashi H (2015) Alternative translational initiation of ATP due to the large phylogenetic distances among the studied sulfurylase underlying dual localization of sulfate assimilation species or genera and un-replicated sampling strategy used pathways in plastids and cytosol in Arabidopsis thaliana. Front in the 1KP dataset used here, the number of identified Plant Sci 5:750 differences between the C -poor and C -rich group might Bowers JE, Chapman BA, Rong JK, Paterson AH (2003) Unravelling 4 4 angiosperm genome evolution by phylogenetic analysis of represent the minimum. Analyses with species with chromosomal duplication events. Nature 422:433–438 smaller phylogenetic distance and with more replicates are Brautigam A, Hoffmann-Benning S, Weber APM (2008) Compara- needed to further confirm pre-conditions and barriers for tive proteomics of chloroplast envelopes from C3 and C4 plants C evolution in different C lineages. Further experiments reveals specific adaptations of the plastid envelope to C4 4 4 photosynthesis and candidate proteins required for maintaining are also needed to study the physiological or functional C4 metabolite fluxes. Plant Physiol 148:568–579 consequences of these observed transcriptomic changes in Brautigam A et al (2011) An mRNA blueprint for C4 photosynthesis extant C3 and C4 plants and correspondingly their rele- derived from comparative transcriptomics of closely related C3 vance to C engineering. Another caveat is that this study and C4 species. Plant Physiol 155:142–156 4 Bromham L, Bennett T (2014) Salt tolerance evolves more frequently focuses on comparing transcriptomics and does not study in C4 grass lineages. J Evol Biol 27:653–659 other regulatory mechanisms, such as translational control, Burgess AL, David R, Searle IR (2015) Conservation of tRNA and post-translational modifications etc., which might also play rRNA 5-methylcytosine in the kingdom Plantae. BMC Plant Biol an important role during the evolution of C 15:199 4 Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer photosynthesis. K, Madden TL (2009) BLAST plus: architecture and applications. BMC Bioinform 10:421 Acknowledgments The authors thank anonymous reviewers for Charron JBF, Ouellet F, Houde M, Sarhan F (2008) The plant constructive comments. The National Basic Research and Develop- apolipoprotein D ortholog protects Arabidopsis against oxidative ment Program of the Ministry of Science and Technology of China stress. BMC Plant Biol 8:86 (#2015CB150104), the Bill and Melinda Gates Foundation Christin PA, Besnard G, Samaritani E, Duvall MR, Hodkinson TR, (#OPP1014417) and the international collaboration program of the Savolainen V, Salamin N (2008) Oligocene CO2 decline National Science Foundation of China, Ministry of Science and promoted C4 photosynthesis in grasses. Curr Biol 18:37–43 Technology of China (# 2011DFA31070) supported this study. We Christin PA et al (2013) Anatomical enablers and the evolution of C4 thank Dr. Gane Wong, who organized the 1KP project on which this photosynthesis in grasses. Proc Natl Acad Sci USA study was based. 110:1381–1386 Ciftci-Yilmaz S et al (2007) The EAR-motif of the Cys2/His2-type Authors’ contribution XGZ conceived the study, wrote and zinc finger protein Zat7 plays a key role in the defense response revised the manuscript. ML and YT conducted all analyses, wrote the of Arabidopsis to salinity stress. J Biol Chem 282:9260–9268

123 208 Plant Mol Biol (2016) 91:193–209

Ding Y, Liu N, Virlouvet L, Riethoven JJ, Fromm M, Avramova Z Jung HS, Niyogi KK (2010) Mutations in Arabidopsis YCF20-like (2013) Four distinct types of dehydration stress memory genes in genes affect thermal dissipation of excess absorbed light energy. Arabidopsis thaliana. BMC Plant Biol 13:229 Planta 231:923–937 Dixon RA, Paiva NL (1995) Stress-induced phenylpropanoid Kanai R, Edwards GE (1999) The biochemistry of C4 photosynthesis. metabolism. Plant Cell 7:1085–1097 In: Sage RF, Monson RK (eds) C4 plant biology. Academic Du Z, Zhou X, Ling Y, Zhang ZH, Su Z (2010) agriGO: a GO Press, San Diego, pp 49–87 analysis toolkit for the agricultural community. Nucleic Acids Knappe S, Lottgert T, Schneider A, Voll L, Flugge UI, Fischer K Res 38:W64–W70 (2003) Characterization of two functional phosphoenolpyruvate/ Dubois E et al (2011) Homologous recombination is stimulated by a phosphate translocator (PPT) genes in Arabidopsis AtPPT1 may decrease in dUTPase in Arabidopsis. PLoS One 6:e18658 be involved in the provision of signals for correct mesophyll Edwards GE, Ku MS (1987) Biochemistry of C3–C4 intermediates. development. Plant J 36:411–420 Biochem Plants 10:275–325 Kraszewska E (2008) The plant nudix hydrolase family. Acta Edwards EJ, Smith SA (2010) Phylogenetic analyses reveal the Biochim Pol 55:663–671 shady history of C4 grasses. Proc Natl Acad Sci USA Kreps JA, Wu YJ, Chang HS, Zhu T, Wang X, Harper JF (2002) 107:2532–2537 Transcriptome changes for Arabidopsis in response to salt, Elhiti M, Hebelstrup KH, Wang AM, Li CL, Cui YH, Hill RD, osmotic, and cold stress. Plant Physiol 130:2129–2141 Stasolla C (2013) Function of type-2 Arabidopsis hemoglobin in Lamesch P et al (2012) The Arabidopsis information resource the auxin-mediated formation of embryogenic cells during (TAIR): improved gene annotation and new tools. Nucleic Acids morphogenesis. Plant J 74:946–958 Res 40:D1202–D1210 Fischer K, Kammerer B, Gutensohn M, Arbinger B, Weber A, Lampl N, Alkan N, Davydov O, Fluhr R (2013) Set-point control of Hausler RE, Flugge UI (1997) A new class of plastidic RD21 protease activity by AtSerpin1 controls cell death in phosphate translocators: a putative link between primary and Arabidopsis. Plant J 74:498–510 secondary metabolism by the phosphoenolpyruvate/phosphate Li B, Dewey CN (2011) RSEM: accurate transcript quantification antiporter. Plant Cell 9:453–462 from RNA-seq data with or without a reference genome. BMC Flugge UI, Hausler RE, Ludewig F, Gierth M (2011) The role of Bioinform 12:323 transporters in supplying energy to plant plastids. J Exp Bot Love MI, Huber W, Anders S (2014) Moderated estimation of fold 62:2381–2392 change and dispersion for RNA-seq data with DESeq2. Genome Fucile G, Falconer S, Christendat D (2008) Evolutionary diversifi- Biol 15:550 cation of plant shikimate kinase gene duplicates. PLoS Genet Lynch JP (2013) Steep, cheap and deep: an ideotype to optimize 4:e1000292 water and N acquisition by maize root systems. Ann Bot Gepstein S et al (2003) Large-scale identification of leaf senescence- 112:347–357 associated genes. Plant J 36:629–642 Maeda H, Dudareva N (2012) The shikimate pathway and aromatic Ghannoum O (2009) C4 photosynthesis and water stress. Ann Bot amino acid biosynthesis in plants. Annu Rev Plant Biol 103:635–644 63:73–105 Goepfert S, Hiltunen JK, Poirier Y (2006) Identification and Majeran W, van Wijk KJ (2009) Cell-type-specific differentiation of functional characterization of a monofunctional peroxisomal chloroplasts in C4 plants. Trends Plant Sci 14:100–109 enoyl-CoA hydratase 2 that participates in the degradation of Mallmann J, Heckmann D, Brautigam A, Lercher MJ, Weber AP, even cis-unsaturated fatty acids in Arabidopsis thaliana. J Biol Westhoff P, Gowik U (2014) The role of photorespiration during Chem 281:35894–35903 the evolution of C4 photosynthesis in the genus Flaveria. Elife Gowda VRP, Henry A, Yamauchi A, Shashidhar HE, Serraj R (2011) 3:e02478 Root biology and genetic improvement for drought avoidance in McKown AD, Moncalvo JM, Dengler NG (2005) Phylogeny of rice. Field Crop Res 122:1–13 Flaveria (Asteraceae) and inference of C4 photosynthesis Gowik U, Brautigam A, Weber KL, Weber APM, Westhoff P (2011) evolution. Am J Bot 92:1911–1928 Evolution of C4 photosynthesis in the genus Flaveria: how many Meyer K, Stecca KL, Ewell-Hicks K, Allen SM, Everard JD (2012) and which genes does it yake to make C4? Plant Cell Oil and protein accumulation in developing seeds is influenced 23:2087–2105 by the expression of a cytosolic pyrophosphatase in Arabidopsis. Grabherr MG et al (2011) Full-length transcriptome assembly from Plant Physiol 159:1221–1234 RNA-seq data without a reference genome. Nat Biotechnol Monson RK, Schuster WS, Ku MSB (1987) Photosynthesis in 29:644–652 Flaveria brownii Powell, A.M.: a C4-like C3–C4 intermediate. Griffiths H, Weller G, Toy LFM, Dennis RJ (2013) You’re so vein: Plant Physiol 85:1063–1067 bundle sheath physiology, phylogeny and evolution in C3 and C4 Nakamura N, Iwano M, Havaux M, Yokota A, Munekage YN (2013) plants. Plant Cell Environ 36:249–261 Promotion of cyclic electron transport around photosystem I Herrmann KM (1995) The shikimate pathway as an entry to aromatic during the evolution of NADP-malic enzyme-type C4 photosyn- secondary metabolism. Plant Physiol 107:7–12 thesis in the genus Flaveria. New Phytol 199:832–842 Herrmann KM, Weaver LM (1999) The shikimate pathway. Annu Ogawa T, Ueda Y, Yoshimura K, Shigeoka S (2005) Comprehensive Rev Plant Physiol 50:473–503 analysis of cytosolic nudix hydrolases in Arabidopsis thaliana. Irshad M, Canut H, Borderies G, Pont-Lezica R, Jamet E (2008) A J Biol Chem 280:25277–25283 new picture of cell wall protein dynamics in elongating cells of Ogawa T, Yoshimura K, Miyake H, Ishikawa K, Ito D, Tanabe N, Arabidopsis thaliana: confirmed actors and newcomers. BMC Shigeoka S (2008) Molecular characterization of organelle-type Plant Biol 8:94 nudix hydrolases in Arabidopsis. Plant Physiol 148:1412–1424 John CR, Smith-Unna RD, Woodfield H, Covshoff S, Hibberd JM Ogawa T, Ishikawa K, Harada K, Fukusaki E, Yoshimura K, (2014) Evolutionary convergence of cell-specific gene expres- Shigeoka S (2009) Overexpression of an ADP-ribose pyrophos- sion in independent lineages of C4 grasses. Plant Physiol phatase, AtNUDX2, confers enhanced tolerance to oxidative 165:62–75 stress in Arabidopsis plants. Plant J 57:289–301

123 Plant Mol Biol (2016) 91:193–209 209

Osborne CP, Beerling DJ (2006) Nature’s green revolution: the Taylor-Teeples M et al (2015) An Arabidopsis gene regulatory remarkable evolutionary rise of C4 plants. Philos Trans R Soc network for secondary cell wall synthesis. Nature 517:571-U307 Lond B Biol Sci 361:173–194 Thimm O et al (2004) MAPMAN: a user-driven tool to display Osborne CP, Freckleton RP (2009) Ecological selection pressures for genomics data sets onto diagrams of metabolic pathways and C4 photosynthesis in the grasses. Proc R Soc Lond B Biol other biological processes. Plant J 37:914–939 276:1753–1760 Tholen D, Boom C, Zhu XG (2012) Opinion: prospects for improving Ripley BS, Gilbert ME, Ibrahim DG, Osborne CP (2007) Drought photosynthesis by altering leaf anatomy. Plant Sci 197:92–101 constraints on C4 photosynthesis: stomatal and metabolic Tyra HM, Linka M, Weber AP, Bhattacharya D (2007) Host origin of limitations in C3 and C4 subspecies of Alloteropsis semialata. plastid solute transporters in the first photosynthetic eukaryotes. J Exp Bot 58:1351–1363 Genome Biol 8:R212 Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: a Biocon- Uga Y et al (2013) Control of root system architecture by DEEPER ductor package for differential expression analysis of digital ROOTING 1 increases rice yield under drought conditions. Nat gene expression data. Bioinformatics 26:139–140 Genet 45:1097–1102 Sack L, Scoffoni C (2013) Leaf venation: structure, function, Vision TJ, Brown DG, Tanksley SD (2000) The origins of genomic development, evolution, ecology and applications in the past, duplications in Arabidopsis. Science 290:2114–2117 present and future. New Phytol 198:983–1000 Wang H, Zhu YY, Fujioka S, Asami T, Li JY, Li JM (2009) Sage RF (2001) Environmental and evolutionary preconditions for the Regulation of Arabidopsis brassinosteroid signaling by atypical origin and diversification of the C4 photosynthetic syndrome. basic helix-loop-helix proteins. Plant Cell 21:3781–3791 Plant Biol 3:202–213 Weber APM, von Caemmerer S (2010) Plastid transport and Sage RF (2004) The evolution of C4 photosynthesis. New Phytol metabolism of C3 and C4 plants: comparative analysis and 161(2):341–370 possible biotechnological exploitation. Curr Opin Plant Biol Sage RF, Christin PA, Edwards EJ (2011) The C4 plant lineages of 13:257–265 planet Earth. J Exp Bot 62:3155–3169 Weng JK, Li Y, Mo HP, Chapple C (2012) Assembly of an Sage RF, Sage TL, Kocacinar F (2012) Photorespiration and the evolutionarily new pathway for alpha-pyrone biosynthesis in evolution of C4 photosynthesis. Annu Rev Plant Biol 63:19–47 Arabidopsis. Science 337:960–964 Sage TL et al (2013) Initial events during the evolution of C4 Williams BP, Johnston IG, Covshoff S, Hibberd JM (2013) Pheno- photosynthesis in C3 species of Flaveria. Plant Physiol typic landscape inference reveals multiple evolutionary paths to 163:1266–1276 C4 photosynthesis. eLife 2:e00961 Simkin AJ et al (2011) Peroxisomal localisation of the final steps of Zhu X-G, Long SP, Ort DR (2008) What is the maximum efficiency the mevalonic acid pathway in planta. Planta 234:903–914 with which photosynthesis can convert solar energy into Stanley Kim H et al (2005) Transcriptional divergence of the biomass? Curr Opin Biotechnol 19:153–159 duplicated oxidative stress-responsive genes in the Arabidopsis genome. Plant J 41:212–220 Tang HB, Bowers JE, Wang XY, Ming R, Alam M, Paterson AH (2008) Synteny and collinearity in plant genomes. Science 320:486–488

123