Introduction

Global warming in recent years has caused extreme weather events such as severe droughts and heat waves, which affect ecosystems and vegetation and reduce crop production. Under drought, plants close their stomata. This regulation is important to prevent water loss in plants but also limits the entry of CO2. The resulting decrease in intracellular CO2 enhances the oxygenase activity of ribulose-1,5-bisphosphate carboxylase-oxygenase (RuBisCO) in chloroplasts. High temperature also enhances RuBisCO oxygenase activity by decreasing the RuBisCO specificity to CO2 relative to O2 in addition to decreasing the ratio of dissolved O2 to dissolved CO2 in the chloroplast (Jordan and Ogren 1984; Long 1991). At the current atmospheric CO2 concentration of approximately 400 ppm, photorespiration occurs at a rate of 25% of photosynthesis at 30–35 °C and over 40% of photosynthesis at 35–40 °C (Sage et al. 2012). C4 plants are able to suppress photorespiration by concentrating CO2 at RuBisCO sites; therefore, they have a great advantage for survival in hot and semiarid environments compared with C3 plants. Metabolic pathways have been extensively studied, and attempts have been made to introduce the C4 cycle into C3 plants. However, the engineering of C4 photosynthesis is still a work in progress, and the entire system that coordinates C4 photosynthesis needs to be clarified. (Ermakova et al. 2021; Lin et al. 2020; Taniguchi et al. 2008). Meanwhile, phylogenetic studies of various genera, phenotypic comparisons, and a recent comprehensive transcriptome have demonstrated that C4 evolution occurred gradually through C3–C4 intermediate stages (Lauterbach et al. 2017; Lyu et al. 2015; Mallmann et al. 2014; Williams et al. 2013). C4 evolution has occurred more than 66 times in angiosperms, indicating that there may be a common system in C3 plants that leads to C4 evolution (Sage et al. 2012). The genus Flaveria in the family Asteraceae evolved relatively recently three hundred million years ago and contains a variety of species, including C3, C3–C4 intermediate, C4-like and C4 species, making it a good model genus for studying the evolution of C4 photosynthesis (Christin et al. 2011; Ku et al. 1991; Powell 1978). Biochemical and biophysical evidence of Flaveria is accumulating, and we recently published draft genome sequences of C3 Flaveria robusta, C3–C4 intermediate Flaveria floridana, C4-like Flaveria brownii and C4 Flaveria bidentis (Taniguchi et al. 2021). By tracing the genomic changes associated with changes in photosynthesis, we hope to clarify the molecular mechanisms underlying C4 evolution and to provide missing information for engineering C4 photosynthesis in C3 plants. In this review, we present the characteristics of C3, C3–C4 intermediate, C4-like, and C4 Flaveria species and propose a scheme for C4 evolution derived from studies of Flaveria.

During the 3 million years of C4 evolution in the genus Flaveria, genome size, not the number of encoded genes, has changed enormously

Twenty-three species are recognized in the genus Flaveria, and most of these species were reported to be diploid (n = 18), except for F. pringlei (n = 36), which is an allopolyploid of F. pringlei and F. angustifolia (Lyu et al. 2015; McKown et al. 2005; Powell 1978). Figure 1 shows the phylogenetic tree of 20 species in the genus Flaveria and the genome sizes of selected species among them (McKown et al. 2005; Taniguchi et al. 2021). Phylogenetic studies showed that C3–C4 intermediate species appeared between 3.6 and 3.1 million years ago, except for F. sonorensis, which appeared 2.8 million years ago (Christin et al. 2011; McKown et al. 2005). The transition from C3–C4 intermediates to C4-like traits has occurred twice in the genus Flaveria and is estimated to have occurred between 1.8 and 1.3 million years ago in clade A and after 0.4 million years ago in clade B (Christin et al. 2011; McKown et al. 2005). The transition from the C4-like trait to the C4 trait occurred only in clade A, which is estimated to have occurred after 1 million years ago. They are mostly distributed in tropical and subtropical regions, such as Mexico, the Gulf Coast of the United States, and the West Indies (McKown et al. 2005). Geographic studies have suggested that the transition from C3 to C3–C4 intermediates, C4-like, and C4 photosynthesis in the genus Flaveria may have been triggered by high temperatures, frequent droughts, and increased salinity, as suggested by the evolution of many other C4 species (Edwards et al. 2010; McKown et al. 2005; Sage et al. 2012). Interestingly, although the genome sizes of these Flaveria species vary widely, indicating that their genomes have changed significantly during evolution (Fig. 1), the number of protein-coding genes predicted by mapping mRNAs to whole-genome draft data was not affected (Taniguchi et al. 2021). The genome size of the basal Group C3 F. robusta is the smallest at 0.49 Gb, the genome size of C4 F. bidentis in clade A is twice that at 1.01 Gb, and the genome size of C3–C4 intermediate F. floridana and C4-like F. brownii in clade B is 1.39 Gb and 1.58 Gb, respectively, but all four species contain approximately 40,000 protein-coding genes (Taniguchi et al. 2021). These species have almost the same number of gene families of C4 cycle enzymes, and the expression of one of the genes was upregulated and acquired a cell-specific expression pattern during C4 evolution (Taniguchi et al. 2021). These results suggest that basal group Flaveria species have the potential to evolve to C4 and/or are already in the preliminary stages of C4 evolution, and changes in cis-elements and/or trans-factors that control gene expression may have led to actual C4 evolution.

Fig. 1
figure 1

Phylogenetic tree of 20 species in the genus Flaveria and genome sizes of selected species among them. Relative branch points in the phylogenetic tree and clade classification of species were based on McKown et al. (2005). Genome size of Flaveria species was based on Taniguchi et al. (2021)

Full size image

Alteration of leaf anatomy supporting a two-celled metabolic cycle during the transition from C3 to C4 photosynthesis

To compare plant phenotypes and gene expression, plants were grown in a growth chamber with a light intensity of 200–300 μmol photons m−2 s−1 at 24 °C in our study (Munekage et al. 2010; Nakamura et al. 2013; Taniguchi et al. 2021). Under this condition, Flaveria C3 and C4 species have typical CO2 concentration points (Γ) and O2 inhibition of photosynthesis compared with other C3 and C4 plants, respectively, and C3–C4 intermediate and C4-like species have values between C3 and C4 species (Table 1) (Jordan and Ogren 1984; Ku et al. 1991). Since flowering in the genus Flaveria is induced under short-day conditions, for morphological analysis, plants were grown under long-day conditions to prevent early flowering (Fig. 2). C3 F. pringlei, C3 F. robusta, C3–C4 F. floridana, C4 F. bidentis and C4 F. trinervia have broad leaves (Fig. 2A, C, E, M and O), whereas C3–C4 F. ramosissima, C4-like F. brownii and C4-like F. palmeri have narrow leaves (Fig. 2G, I and K). The size and shape of leaves are genetically or conditionally determined as a result of adaptation to the environment. For example, plants form smaller leaf surface areas to minimize water loss in dry environments, but leaf size and shape are not related to the type of photosynthesis in the genus Flaveria (Fig. 2). On the other hand, leaf anatomy is closely related to the type of photosynthesis (Fig. 2).

Table 1 Characterization of C3, C3–C4 intermediate, C4-like and C4 species in the genus Flaveria
Full size table
Fig. 2
figure 2

Visible phenotype and leaf cross sections of F. pringlei (A, B), F. robusta (C, D) classified as a C3 species, F. floridana (E, F) and F. ramosissima (G, H) classified as C3–C4 intermediate species, F. brownii (I, J) and F. palmeri (K, L) classified as C4-like species, and F. bidentis (M, N) and F. trinervia (O, P) classified as C4 species grown for 8 weeks in a growth chamber at a light intensity of 200–300 μmol photons m−2 s−1 with a 16 h light–8 h dark photoperiod at 24 °C. Leaf cross sections were prepared using fully expanded leaves as described in Nakamura et al. (2013). Scale bars indicate 5 cm for the picture of the plants (A, C, E, G, I, K, M, O) and 100 μm for the leaf cross section (B, D, F, H, J, L, N, P). Asterisks indicate examples of bundle sheath cells. Arrowhead indicate examples of chloroplasts in the centripetal position of bundle sheath cells

Full size image

Common anatomical features that have changed in association with the evolution to C4 photosynthesis include (i) a decrease in intervascular distance correlated with a decrease in the number of mesophyll cells located between bundle sheath cells, (ii) an enlargement of the volume of bundle sheath cells relative to the volume of mesophyll cells, (iii) an increase in the number of mitochondria and chloroplasts oriented centripetally or centrifugally within bundle sheath cells, and (iv) radial patterning of a single mesophyll cell layer surrounding bundle sheath cells (Lundgren et al. 2014; Sage et al. 2012). These features are important for the two-celled metabolism of C4 photosynthesis. The enlargement of bundle sheath cells and the increase in the number of organelles in these cells are necessary to support the metabolic activities of bundle sheath cells. The reduced intervascular distance and the radial patterning of a single mesophyll cell layer that surrounds the bundle sheath cells also contribute to the rapid exchange of metabolites between mesophyll and bundle sheath cells. These C4-type features, often collectively referred to as Kranz anatomy, must be regulated by different molecular mechanisms, and it is likely that they were obtained in a stepwise fashion through intermediate stages.

(i) In the genus Flaveria, intervascular distance correlates well with the degree of C4 evolution: the C3 basal species have longer intervascular distances and four or more mesophyll cells between vascular bundles, whereas the C3–C4 intermediate species have reduced intervascular distances, and the C4-like and C4 species have more pronounced reductions in intervascular distances and approximately two mesophyll cells between vascular bundles (Fig. 2) (McKown and Dengler 2007). This trait is attributed to the enhanced development of minor veins (McKown and Dengler 2009). Auxin signaling appears to be an important factor controlling vein density because vein differentiation is induced by auxin maxima that are directed by PIN-FORMED auxin efflux carriers and their regulators (Linh et al. 2018; Sedelnikova et al. 2018). Although it was shown that auxin biosynthesis and the expression of related genes are higher in C4 species (Huang et al. 2017), the mechanisms underlying the initiation of minor vein and C4-type vein patterning have not been elucidated.

(ii) In terms of parameters for bundle sheath cell enlargement, the volume of the bundle sheath cells do not differ between C3 and C4 Flaveria species, but the relative proportion of mesophyll to bundle sheath tissue area in mature leaves is well correlated with the type of photosynthesis (McKown and Dengler 2007). The C3–C4 intermediate species have a ratio of mesophyll to bundle sheath tissue area between C3 and C4 species, while the C4-like species have a ratio of mesophyll to bundle sheath tissue area close to C4 species (McKown and Dengler 2007). This is thought to be due to the early cessation of proliferation and elongation of mesophyll cells during leaf expansion, resulting in a decrease in the number and volume of mesophyll cells (McKown and Dengler 2009).

(iii) The number of mitochondria and chloroplasts in the bundle sheath cells is significantly higher in C3–C4 intermediate species, and this phenotype is closely related to the development of a photorespiration-dependent CO2 concentration system (Sage et al. 2014; Voznesenskaya et al. 2017; Yorimitsu et al. 2019). C3 photosynthesis occurs within the mesophyll cell, and the bundle sheath cells do not contribute much to starch synthesis, but in C3–C4 intermediate species, photorespiration-dependent CO2 concentration using mesophyll and bundle sheath cells (called C2 photosynthesis) occurs, and starch synthesis occurs in both cells (Bauwe 2011). In the C3–C4 intermediate species, most of the activity of glycine decarboxylase (GDC) is lost in mitochondria of mesophyll cells; therefore, glycine is transported to bundle sheath cells and converted to serine by GDC in mitochondria of bundle sheath cells (Fig. 3A). A large proportion of chloroplasts were found to accumulate in the centripetal position of bundle sheath cells together with mitochondria in Flaveria C3–C4 intermediate species (Fig. 2F and H) (Sage et al. 2013). These two-celled C2 cycle and organelle arrangements contribute significantly to the recapture of CO2 released from photorespiration by elevating the CO2 level at a site of RuBisCO in bundle sheath chloroplasts. Since C3 F. pringlei and C3 F. robusta were found to have more chloroplasts and mitochondria in bundle sheath cells than C3 F. cronquistii and other closely related C3 species, and GDC expression was higher in bundle sheath cells than in mesophyll cells, these species have been classified as proto-Kranz C3 species (Sage et al. 2013, 2014). The C4-like and C4 species have enlarged chloroplasts located at the centripetal position in the bundle sheath cells (Fig. 2J, L, N and P), but the number of chloroplasts in these species has not changed or has decreased except for F. brownii, which has a large number of small chloroplasts in bundle sheath cells (Araus et al. 1990; Brown and Hattersley 1989; Sage et al. 2014). Maize GOLDEN2-like transcription factors were shown to promote the development of bundle sheath chloroplasts in rice (Wang et al. 2017). In rice, two genes encoding GOLDEN2-like, OsGLK1 and OsGLK2, were shown to be redundantly involved in chloroplast development (Rossini et al. 2001; Wang et al. 2013). On the other hand, in maize, ZmG2, which is related to OsGLK2, was preferentially expressed in bundle sheath cells, and was suggested to function in chloroplast development in these cells (Hall et al. 1998; Rossini et al. 2001). However, the mechanism that determines chloroplast enlargement, increase, and positioning remains to be elucidated.

Fig. 3
figure 3

Schematic representation of the metabolic pathways of the C3–C4 intermediate stage, C4-like stage and C4 photosynthesis. In the C3–C4 intermediate stage, noncell-specific expression of C4 enzymes is induced in the background of C2 photosynthesis, where glycine produced by photorespiration in mesophyll cells is shuttled to BS cells and decarboxylated by glycine decarboxylase localized in mitochondria of bundle sheath cells (A). In the C4-like stage, the two-celled C4 cycle is strongly promoted by cell-specific regulation of C4 enzyme expression, and C2 photosynthesis activity is reduced by suppression of RuBisCO expression in mesophyll cells (B). C4 photosynthesis is established by completely restricting RuBisCO expression in BS cells (C). Chloroplasts and mitochondria are represented in green and orange, respectively. M mesophyll cell, BS bundle sheath cell, C3 C3 cycle, C4 cycle, C2 C2 photosynthesis, RuBisCO ribulose 1,5-bisphosphate carboxylase/oxygenase, PEPC phosphoenolpyruvate carboxylase, PPDK pyruvate orthophosphate dikinase, ME NADP-malic enzyme, and GDC glycine decarboxylase

Full size image

(iv) Radial patterning, in which a single mesophyll cell layer surrounds a bundle sheath cell layer, was observed not only in monocotyledonous C4 species but also in eudicotyledonous C4 species (Edwards and Voznesenskaya 2011). This patterning is partly related to the number of cell layers in immature ground tissues and the loss of cell division during leaf development. In the genus Flaveria, basal C3 species have eight layers of ground tissue cells in developing leaves (McKown and Dengler 2007). Of eight cell layers, cells in the first and second layers at the adaxial side of the subepidermis differentiate into palisade cells. C3–C4 intermediate F. floridana and C4-like F. brownii in clade B both have six layers of immature ground tissue cells in developing leaves (McKown and Dengler 2007), but in mature leaves, F. floridana has two layers of palisade cells, whereas F. brownii has mostly one layer of mesophyll cells, and the extra mesophyll cells not adjacent to the bundle sheath cells contain few chloroplasts (Fig. 2F and J) (Araus et al. 1990; Cheng et al. 1988). Furthermore, large intercellular spaces were observed between epidermal cells in adaxial or abaxial mesophyll cells in F. brownii (Fig. 2J). This implies that early cessation of proliferation and development of mesophyll cells occur during leaf expansion in C4-like F. brownii. Clade A species, including C3–C4 intermediate F. ramosissima, C4-like F. palmeri, C4 F. bidentis and C4 F. trinervia, basically exhibit five layers of ground tissue cells in developing leaf tissue, and they all form a single layer of palisade mesophyll cells (Fig. 2H, L, N and P) (McKown and Dengler 2007), suggesting that the number of ground tissue layers is genetically determined in clade A. On the other hand, in C4-like and C4 species in clade A, the mesophyll cells are arranged around the bundle sheath cells, and most of them are adjacent to the bundle sheath cells (Fig. 2L, N and P).

Gradual replacement of the two-celled C2 cycle with the two-celled C4 cycle during C4 evolution

The C3–C4 intermediate species exhibit a lower CO2 compensation point than the C3 species due to their photorespiration-dependent CO2 concentration mechanism (C2 photosynthesis), but their O2 inhibition rate is high (40–44%), close to that of the C3 species (Table 1) (Ku et al. 1991). The lower O2 inhibition rates found in the C4-like species (26% in F. brownii and 7% in F. palmeri) correlated with the amount of RuBisCO expression remaining in the mesophyll cell (Table 1) (Ku et al. 1991; Taniguchi et al. 2021).

The first step of C4 evolution was the gain of C2 photosynthesis that occurred in the transition from the C3 to C3–C4 intermediate stage, which can be explained by a single event, the localization of GDC activity to the mitochondria of bundle sheath cells, as described above. C2 photosynthesis is important under high-temperature conditions when the activity of RuBisCO oxygenase is enhanced and was suggested to have bridged the evolution of C3 to C4 photosynthesis (Bauwe 2011; Sage et al. 2018). GDC is composed of four proteins, P-, L-, T-, and H-protein, which catalyze the conversion of glycine to serine through a multistep enzymatic system. In the multistep reaction, the P-protein functions as the actual decarboxylation unit. Suppression of GDC P-protein in mitochondria of mesophyll cells was found in C3–C4 intermediate species in a number of genera, including Steinchisma, Moricandia, Mollugo, Flaveria and Heliotropium (Bauwe 2011; Sage et al. 2014). Studies in the genus Flaveria have provided an example of how bundle sheath cell-specific expression of P-protein in GDCs was acquired (Schulze et al. 2013, 2016). In Flaveria, the P-protein of GDC is encoded by three genes, GLDPA, GLDPB and GLDPC. While GLDPA and GLDPB were shown to be involved in photorespiration, GLDPC was suggested to be involved in the maintenance of basal C1 metabolism (Schulze et al. 2013). Promoter analysis showed that GLDPA was expressed only in bundle sheath cells in C3 and C4 Flaveria species, whereas GLDPB was expressed both in mesophyll and bundle sheath cells in C3 species and became a pseudogene in C4 Flaveria species (Schulze et al. 2013; Wiludda et al. 2012). In the C3–C4 intermediate species, the expression of GLDPA was upregulated, while that of GLDPB was downregulated, allowing C2 photosynthesis to function at a high activity (Schulze et al. 2013). The bundle sheath cell-specific expression of the GLDP gene can be achieved by altering the cis-regulatory elements of the promoter region during evolution, as shown in Arabidopsis thaliana, where the deletion of a cis-regulatory module required for mesophyll cell-specific expression, called the “M-box,” resulted in bundle sheath cell-specific expression of the GLDP genes (Adwy et al. 2015).

The next step was how the C4 cycle was developed in the C3–C4 intermediate stage toward the establishment of C4 photosynthesis. The genus Flaveria contains a number of C3–C4 intermediate species classified as type II C2 species that express C4 cycle enzymes to some extent, as shown in Table 1, but these C4 cycle enzymes do not contribute to the concentration of CO2 (Ku et al. 1991; Monson et al. 1988). It has been suggested that these C4 cycle enzymes are upregulated to rebalance nitrogen metabolism under C2 photosynthesis (Mallmann et al. 2014; Schulze et al. 2016). Since the glycine decarboxylation reaction releases toxic ammonia that should be taken up by the bundle sheath chloroplasts, it creates an imbalance in nitrogen metabolism between mesophyll and bundle sheath chloroplasts. By using a computer simulation model, a malate/alanine shuttling between mesophyll and bundle sheath cells was predicted to be coupled with a glycine/serine shuttling (Mallmann et al. 2014). Operation of this metabolic pathway was supported by upregulation of alanine aminotransferase together with NADP-malic enzyme (NADP-ME) in C3–C4 intermediate species (Mallmann et al. 2014). However, C4 cycle enzymes were expressed in both mesophyll and bundle sheath cells in C3–C4 intermediate species (Moore 1988; Taniguchi et al. 2021), which indicates that the produced C4 compounds could be metabolized within a cell without transfer to adjacent cells (Fig. 3). While equal distribution of the C4 enzyme between mesophyll and bundle sheath cells was observed in F. floridana, weak cell selective distribution of the C4 cycle enzyme was observed in F. ramosissima (Table 1) (Taniguchi et al. 2021). The selective cell distribution became stronger in C4-like F. brownii (Table 1) (Taniguchi et al. 2021). In this species, the level of PEPC expression does not reach the level of that in C4 species (0.3 times that of C4 species, Table 1), but functional operation of the C4 cycle that contributes to CO2 concentration in bundle sheath cells was reported (Cheng et al. 1988; Monson et al. 1988). This evidence shows that cell-specific expression was gradually gained during the C3–C4 intermediate stage and elevated flux of the C4 cycle between mesophyll and bundle sheath cells. The following scenarios were possible: (1) in the early C3–C4 intermediate stage, multiple metabolic pathways, including 2-oxoglutarate/glutamate, pyruvate/alanine and malate/aspartate shuttles, may have been used to balance nitrogen metabolism under C2 photosynthesis, as predicted by computer simulation (Mallmann et al. 2014). C4 cycle enzymes and N- and C-balancing enzymes, such as alanine-aminotransferase and aspartate-aminotransferase, were upregulated in response to metabolic imbalance but in a noncell selective manner; therefore, the flux of the two-celled C4 cycle must have been very low (Fig. 3A). (2) Cell-selective expression of C4 cycle enzymes was gradually acquired to correct the metabolic imbalance more efficiently, and the resulting increase in the flux of the two-celled C4 cycle may have functioned to concentrate CO2. Subsequent suppression of RuBisCO expression in mesophyll cells may have reduced RuBisCO oxygenase activity and replaced the C2 cycle with the C4 cycle at a C4-like stage (Fig. 3B). Suppression of RuBisCO expression in mesophyll cells was not observed in F. ramosissima but in F. brownii, indicating that it was induced after the establishment of a high flux of the two-celled C4 cycle. (3) Finally, C4 photosynthesis was established by complete suppression of RuBisCO expression in mesophyll cells (Fig. 3C).

Optimization of the photochemical reaction and energy supply by reduction of photosystem II (PSII) activity and upregulation of cyclic electron transport

The electron transport system in chloroplasts was modified to optimize the energy supply during C4 evolution. Since in the linear electron transport from water to NADPH, the number of protons transferred with an electron transfer is fixed, the ratio of ATP to NADPH production was estimated to be 9/7 (Allen 2003). C4 photosynthesis requires more ATP to drive the C4 cycle so that the ratio of ATP/NADPH demand in chloroplasts increases with the development of the C4 cycle during C4 evolution. In NAD-malic enzyme (NAD-ME)-type C4 photosynthesis, ATP/NADPH demand increased in mesophyll cells, whereas in NADP-ME-type C4 photosynthesis, it increased in bundle sheath cells because reducing power was shuttled as malate from mesophyll to bundle sheath cells (Kanai and Edwards 1999). A part of the C3 cycle from phosphorylation of 3-PGA and subsequent reduction to production of triose phosphate is known to occur in mesophyll cells in C4 species (Fig. 3C) (Kanai and Edwards 1999). This metabolic pathway is important in allocating energy requirements in mesophyll chloroplasts but is not likely to be able to compensate for imbalanced ATP/NADPH demand between mesophyll and bundle sheath cells (Kanai and Edwards 1999; Munekage and Taniguchi 2016).

Cyclic electron transport (CET) around photosystem I can generate proton motive force driving ATP synthesis without the production of NADPH by recycling electrons from ferredoxin to plastoquinone (Munekage 2016; Yamori and Shikanai 2016). There are two pathways of cyclic electron transport: the PGR5-PGRL1-dependent pathway and the NDH complex-dependent pathway (DalCorso et al. 2008; Munekage et al. 2002; Peltier et al. 2016). The abundances of NDH subunits were higher corresponding to the elevation of ATP demand in bundle sheath chloroplasts in NADP-ME type C4 species or in mesophyll chloroplasts in NAD-ME type C4 species (Kubicki et al. 1994; Majeran et al. 2008; Takabayashi et al. 2005), indicating that NDH-dependent pathways were used to supply the ATP required for C4 photosynthesis. Interestingly, the NDH subunit was elevated in the C3–C4 intermediate F. ramosissima and C4-like F. brownii, correlating with enhanced CET activity inferred from P700 oxidation kinetics (Nakamura et al. 2013) (Table 1). In C3 photosynthesis, ATP/NADPH demand was estimated to be 1.55 when photorespiration/photosynthesis occurred at 1/4 (Osmond 1981). In C3–C4 intermediate species, if glycine/serine shuttling was taken into account, ATP/NADPH demand was only slightly increased to 1.57 in mesophyll cells by a phosphorylation step of glycerate to produce 3-PGA and regeneration steps of RuBP from 3-PGAs that were also produced by RuBisCO oxygenation; whereas NH3 uptake by glutamate synthase (GS) and glutamine-oxoglutarate aminotransferase (GOGAT), which consumed one molecule of ATP and 2 electrons counted as one molecule of NADPH, decreased the ATP/NADPH demand in bundle sheath cells, indicating that glycine/serine shuttling did not influence the ATP/NADPH demand in chloroplasts. However, if the C4 cycle was coupled with the glycine/serine shuttle, the total ATP/NADPH demand was increased to 1.65. If reducing power was shuttled as malate from mesophyll to bundle sheaths where it was decarboxylated by NADP-ME, it elevated ATP/NADPH demand in bundle sheath chloroplasts. CET activity was only slightly elevated in C3–C4 intermediate F. floridana but was substantially elevated in C3–C4 intermediate F. ramosissima and C4-like F. brownii, corresponding to the phenotypes where cell-selective distribution of the C4 enzyme was observed (Table 1) (Nakamura et al. 2013). These results suggest that the acquisition of cell-selective expression of the C4 enzyme increased the flux of the C4 cycle, consequently increasing the demand for ATP/NADPH and that CET activity, especially NDH-dependent CET activity, was upregulated to fine-tune the ATP supply in the C3–C4 intermediate stage.

In C4 species in the genus Flaveria, ATP/NADPH demand was estimated to be 1.9 and 5 in mesophyll and bundle sheath cells, respectively, where leakage of CO2 from the bundle sheath to mesophyll cells is neglected (Munekage and Taniguchi 2016). Corresponding to the elevated ATP/NADPH demand, CET activity was further upregulated in C4 species (Table 1). In these species, not only malate but also aspartate is transported to the bundle sheath cells, where it is converted back to oxaloacetate and then reduced to malate, which is decarboxylated by NADP-ME; therefore, the PSII activity of bundle sheath chloroplasts remains up to 20% of that of mesophyll chloroplasts to produce NADPH via linear electron transport (Hofer et al. 1992; Meister et al. 1996). The grana index correlated well with PSII activity and was relatively higher in bundle sheath chloroplasts in Flaveria C4 species (15–19%) than in those in Zea mays and Sorghum bicolor, which have little PSII activity (Table 1) (Andersen et al. 1972; Nakamura et al. 2013; Woo et al. 1970). Notably, bundle sheath chloroplasts in C4-like F. brownii showed a high grana index (50%) similar to those observed in mesophyll chloroplasts (Holaday et al. 1984; Nakamura et al. 2013). Because C4-like F. palmeri and C4 F. bidentis showed much slower P700 oxidation kinetics than C4-like F. brownii (Table 1), the nonstacked thylakoid membrane structure and the suppression of PSII may contribute to the elevation of CET activity. These results also suggest that the optimization of electron transport by suppression of PSII was induced at a late stage of C4 evolution.

Conclusions

C4 evolution proceeded through various C3–C4 intermediate stages, where a photorespiration-dependent CO2 enrichment system (C2 photosynthesis) was first acquired, which may have led to the acquisition of two-celled C4 cycles. The genus Flaveria is one of the most useful models to study C4 evolution since it contains a large number of C3–C4 intermediate and C4-like species that are closely related to C4 species. The intermediate features between C3 and C4 displayed by these species indicate that most key C4 traits, including localization of GDCs in mitochondria of bundle sheath cells, upregulation and cell-selective regulation of C4 cycle-related genes, suppression of RuBisCO in mesophyll cells, upregulation of cyclic electron transport activity and suppression of PSII activity in bundle sheath chloroplasts, were all gradually acquired during C4 evolution. The fact that these traits were acquired at different times suggests that a change that is dominant for survival triggers the next dominant change by natural selection, optimizing the system by modifying the balance of metabolism, gene expression, and energy supply.

Currently, the direct introduction of the C4 cycle into C3 plants is being attempted as a way to engineer C4 photosynthesis. However, in these transformants, it was reported that bundle sheath chloroplasts were not well developed, resulting in insufficient expression of bundle sheath chloroplast proteins and that the C4 cycle did not function well and there was an imbalance in metabolism and reducing power (Ermakova et al. 2021; Lin et al. 2020; Taniguchi et al. 2008). Since C4 photosynthesis is a well-optimized and sophisticated system, it is necessary to modify the support system simultaneously with the introduction of the C4 cycle. The evolutionary process of C4 photosynthesis shows a trait change toward C4 based on genomic changes. Using technologies such as genome editing to introduce genome modifications that mimic the evolution of C4 photosynthesis may enable the engineering of C4 photosynthesis in the future.