Promotion of cyclic electron transport around photosystem I during the evolution of NADP–malic enzyme-type C4 photosynthesis in the genus Flaveria

Authors


Summary

  • C4 plants display higher cyclic electron transport activity than C3 plants. This activity is suggested to be important for the production of ATPs required for C4 metabolism.
  • To understand the process by which photosystem I (PSI) cyclic electron transport was promoted during C4 evolution, we conducted comparative analyses of the functionality of PSI cyclic electron transport among members of the genus Flaveria, which contains several C3, C3–C4 intermediate, C4-like and C4 species.
  • The abundance of NDH-H, a subunit of NADH dehydrogenase-like complex, increased markedly in bundle sheath cells with the activity of the C4 cycle. By contrast, PROTON GRADIENT REGULATION5 (PGR5) and PGR5-LIKE1 increased in both mesophyll and bundle sheath cells in C4-like Flaveria palmeri and C4 species. Grana stacks were drastically reduced in bundle sheath chloroplasts of C4-like F. palmeri and C4 species; these species showed a marked increase in PSI cyclic electron transport activity.
  • These results suggest that both the expression of proteins involved in PSI cyclic electron transport and changes in thylakoid structure contribute to the high activity of cyclic electron flow in NADP–malic enzyme-type C4 photosynthesis. We propose that these changes were important for the establishment of C4 photosynthesis from C3–C4 intermediate photosynthesis in Flaveria.

Introduction

C4 photosynthesis requires the coordinated functioning of two cell types: mesophyll (M) cells and bundle sheath (BS) cells. In C3 photosynthesis, CO2 is fixed by ribulose 1,5-bisphosphate carboxylase/oxygenase (RuBisCO). By contrast, in C4 photosynthesis, the initial carboxylation step is catalysed by phosphoenolpyruvate carboxylase (PEPC), forming C4 acids in M cells. The C4 acids are transported to BS cells, where they are decarboxylated to release CO2. The other product of decarboxylation, pyruvate, is returned to M cells, where it is converted to phosphoenolpyruvate (Hatch, 1987; Sage, 1999). As the C4 cycle elevates the intercellular CO2 concentration in BS cells, in which RuBisCO is localized, the oxygenase activity of RuBisCO is decreased. Therefore, C4 plants show a low photorespiration rate, which can be advantageous under conditions of water deficiency and/or heat stress (Long, 1999).

C4 photosynthesis appeared relatively recently in geological time – within the past 35 million yr (Sage et al., 2012). It is considered to be one of the evolutionary convergences of land plants, possibly driven by its survival advantages in the hot, often dry, and nutrient-poor soils of the tropics and subtropics (Sage et al., 2011). Approximately 7500 species of C4 plant classified into 19 families and 62 independent C4 lineages, including monocots and dicots, were recognized in a recent geographical and phylogenetic study (Sage et al., 2011). Of the 62 lineages, 21 contain C3–C4 intermediate species. Studies on C3–C4 intermediate species have suggested that the evolution from C3–C4 photosynthesis progressed stepwise through a C3–C4 intermediate stage (Sage, 2004; Gowik & Westhoff, 2010). The genus Flaveria is one of the youngest C4 lineages, and has been estimated to have evolved in the past 5 million yr (Sage et al., 2012). This genus contains several C3, C4, C3–C4 intermediate and C4-like species; these C3–C4 intermediate and C4-like species are recognized as evolutionary intermediates between C3 and C4 species (Monson & Moore, 1989); therefore, this genus has been widely used to study the evolutionary process of C4 photosynthesis. The C3–C4 intermediate species have a lower compensation point for CO2 than C3 species. This is suggested to be because of the glycine shuttle in the C2 cycle and/or partial operation of the C4 cycle (Ku et al., 1983, 1991; Nakamoto et al., 1983; Rumpho et al., 1984; Monson et al., 1986; Brown & Hattersley, 1989; Hylton et al., 1988). The genus Flaveria also contains C4-like species, such as F. brownii, F. palmeri and F. vaginata. In these species, C4 enzymes are highly expressed, as in C4 species, but RuBisCO is not completely compartmentalized between M and BS cells, resulting in greater O2 sensitivity in CO2 assimilation relative to that in C4 species (Cheng et al., 1988; Moore et al., 1989; Ku et al., 1991; Dai et al., 1996).

In C4 plants, both C3 and C4 cycles are functional, thereby increasing the energetic cost of assimilating CO2 relative to that in C3 plants. As a consequence, two extra ATP molecules are required for each CO2 fixed to drive the C4 cycle. In NADP–malic enzyme (ME)-type C4 plants, ATP demand is higher in BS cells than in M cells (Kanai & Edwards, 1999). The extra ATP required for C4 photosynthesis has been suggested to be produced by cyclic electron transport around photosystem I (PSI), which contributes to the generation of a pH gradient across the thylakoid membrane without the net production of NADPH (Kanai & Edwards, 1999). Two cyclic pathways around PSI have been identified in C3 plants: the first involves the chloroplast NADH dehydrogenase-like (NDH) complex, which consists of 28 subunits encoded by the chloroplast and nuclear genomes (Ifuku et al., 2011; Peng et al., 2011); the second is a ferredoxin:plastoquinone oxidoreductase (FQR) pathway that involves PROTON GRADIENT REGULATION5 (PGR5) and PGR5-LIKE1 (PGRL1) proteins. This is considered to be the major cyclic electron transport pathway in C3 plants (Munekage et al., 2002, 2004, 2008; DalCorso et al., 2008). A cell-selective increase in the level of the NDH complex, corresponding to ATP demand depending on cell type, has been shown in NADP-ME-type and NAD-ME-type C4 species (Takabayashi et al., 2005). Although the cell-selective increase in the level of PGR5 is insignificant in Zea mays (Takabayashi et al., 2005), we have shown previously by comparative analysis of C3 and C4 species in the genus Flaveria that the levels of both PGR5 and NDH-H, a subunit of the NDH complex, are higher in C4 species than in C3 species (Munekage et al., 2010). This suggests that both cyclic activities were increased during C4 evolution in the genus Flaveria. In Sorghum bicolor, Z. mays and C4 F. trinervia, which carry out NADP-ME-type C4 photosynthesis, grana-free thylakoid membranes were observed in BS chloroplasts (Laetsch & Price, 1971; Hofer et al., 1992). The reduced grana stacks were reported to be correlated with lower PSII activity, which was associated with fewer subunits of the oxygen-evolving complex (OEC; Sheen et al., 1987; Oswald et al., 1990; Hofer et al., 1992). Interestingly, grana stacks in BS chloroplasts were observed in the C3–C4 intermediates F. floridana and F. linearis and C4-like F. brownii (Holaday et al., 1984). This observation suggests that photosynthetic electron transport in C3–C4 intermediate species probably differs from that in C4 species. As the down-regulation of PSII and the promotion of cyclic electron transport are considered to be related to C4 metabolism, it is important to investigate the C3–C4 intermediate stages of photosynthetic electron transport to understand C4 evolution. Comparative analysis within the genus Flaveria, which contains closely related C3, C3–C4 intermediate, C4-like and C4 species, is useful to clarify the process of C4 evolution.

The aim of this study was to clarify how and when cyclic electron transport around PSI was enhanced with the activity of the C4 cycle during evolution. We investigated the activity of PSI cyclic electron transport, the amounts and cellular location of proteins involved in PSI cyclic electron transport, and the thylakoid membrane structure of BS chloroplasts in members of the genus Flaveria, a model genus of NADP–ME-type C4 evolution.

Materials and Methods

Plant materials and growth conditions

Plants of F. pringlei, F. robusta, F. anomala, F. ramosissima, F. brownii, F. palmeri, F. bidentis and F. trinervia, which are gifts from Dr Peter Westhoff (Heinrich-Heine-University, Germany), were grown in pots in soil for 6–8 wk in a growth chamber (200–300 μmol photons m−2 s−1, 12-h light : 12-h dark photoperiod, 24°C). The fifth or sixth mature leaves were used for the experiments.

Absorbance change at 820 nm

The P700 redox change was measured by monitoring the change in absorbance of the leaves at 820 nm with a dual-wavelength pulse-modulation system, ED-P700DW (Heinz-Walz, Effeltrich, Germany) combined with a PAM101 fluorometer (Heinz-Walz), as described previously by Klughammer & Schreiber (1998). P700 oxidation kinetics were measured under far-red light (720 nm, 17.2 W m−2). The maximum oxidation of P700 was determined using a xenon discharge lamp (50 ms, 1500 W m−2; Heinz-Walz) in the presence of far-red light.

Thermoluminescence analysis

The luminescence emitted by leaf discs was measured with a custom-made apparatus, as described previously (Ducruet, 2003; Havaux et al., 2005). The sample was maintained at 0°C for 30 s in the dark, and then illuminated for 20 s with far-red light (> 715 nm) at this temperature. After switching off the far-red light, the temperature was increased from 0 to 70°C at a rate of 0.5°C s−1. The luminescence emission by the sample was measured during heating by a photomultiplier tube protected by a red filter. The far-red light-induced thermoluminescence band peaking at c. 45°C is the afterglow (AG) band, which reflects heat-induced electron flow from the stroma to the plastoquinone pool (Ducruet et al., 2005; Havaux et al., 2005; Apostol et al., 2006; Peeva et al., 2012). It is usually preceded by a B band peaking at lower temperatures in the range of 18–25°C.

Immunoblot analysis

The extraction of proteins and immunoblot analyses were performed essentially as described by Munekage et al. (2010). Total membrane proteins extracted from mature leaves of each Flaveria plant were separated by 15% sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred onto poly(vinylidene difluoride) (PVDF) membranes (Millipore, Billerica, MA, USA). The separated proteins were stained with anti-NDH-H antibodies (Horvath et al., 2000), anti-PGR5 antibodies (Munekage et al., 2002), anti-Rieske antibodies (Sanda et al., 2011), anti-PGRL1 antibodies raised from recombinant proteins provided by Toru Hisabori, anti-PsbO antibodies provided by the late Akira Watanabe, anti-PsaC antibodies purchased from AgriSera (Vännäs, Sweden), anti-RbcL antibodies (Ogawa et al., 2009) and anti-PEPC antibodies provided by Tsuyoshi Furumoto. Immunocomplexes were detected using the ECL Plus Western blotting system (GE Healthcare Japan, Tokyo, Japan). Chemifluorescence signals were detected and quantified with a Lumino image analyzer (LAS-4000 EPUV; Fujifilm, Co., Tokyo, Japan) and analysed using Multi Gauge ver. 3.2. (Fujifilm Co.).

Separation of M and BS fractions

M and BS fractions were separated from mature leaves of F. bidentis according to the method of Hofer et al. (1992) with some modifications. Mature leaves were cut into 2–3-mm2 pieces and homogenized in a Waring laboratory blender at 8400 rpm for 10 s in buffer containing 0.4 M sorbitol, 1 mM MnCl2, 10 mM NaCl, 0.8 mM KH2PO4, 4 mM l-cysteine, 2 mM Na-ascorbate, 44 mM Mes-KOH (pH 6.1) and one tablet of Complete Protease Inhibitor Cocktail Tablets (Roche Diagnostics Japan, Tokyo, Japan) for 50 ml of buffer. The suspension was filtered through a nylon mesh with a mean pore size of 37 μm, and the filtrate was used as the M fraction. The residue was transferred to the same buffer without protease inhibitor, and homogenized in a Waring laboratory blender for 1 min at 15 600 rpm. This procedure was repeated until most of the adhering M cells were released from the BS strands, which was confirmed by observation under a light microscope. The purified BS strands were used as the BS fraction.

In situ immunohistochemistry

Small pieces of mature leaves were fixed in fixation buffer (4% paraformaldehyde, 0.1 mM CaCl2, 50 mM Pipes-NaOH (pH 7.2), 50% ethanol) overnight at 4°C. The samples were dehydrated with an ethanol series and subsequently impregnated with t-butyl alcohol and finally embedded in Paraplast® X-TRA (Sigma-Aldrich, St. Louis, MO, USA). Embedded sections were sliced into 10-μm sections and placed onto slide glass. Deparaffinized sections were stained with toluidine blue for structural observations. For immunolabelling, deparaffinized sections were incubated in blocking solution consisting of Tris-Buffered Saline and Tween 20 (TBST) with 10% bovine serum albumin, and then incubated with preimmunization serum or primary antibody. After washing in TBST, sections were stained with the secondary antibody, anti-rabbit-IgG-fluorescein isothiocyanate (FITC) conjugate (Sigma-Aldrich), and then washed in TBST. FITC fluorescence was detected under a confocal laser scanning microscope (Model FV-1000; Olympus, Tokyo, Japan).

Electron microscopy

The chloroplast structure was studied as described previously (Watanabe et al., 1998). The length of thylakoid membranes was measured by Image analysis software (WinROOF; Mitani Co., Tokyo, Japan; see Supporting Information Fig. S4).

Chlorophyll fluorescence analysis

Chlorophyll fluorescence was measured using a MINI-PAM fluorometer (Heinz-Walz). The transient increase in chlorophyll fluorescence after actinic light illumination (53 μmol photons m−2 s−1) for 4 min was detected in measuring light (0.1 μmol photons m−2 s−1), as described by Shikanai et al. (1998).

Results

Activities of PSI cyclic electron transport in C3, C3–C4 intermediate, C4-like and C4 Flaveria species

PSI cyclic electron transport activity was investigated in vivo in two C3, two C3–C4 intermediate, two C4-like and two C4 species in the genus Flaveria. We measured the P700 oxidation kinetics in response to far-red illumination by monitoring changes in leaf absorbance at 820 nm (Fig. 1a). The activity of cyclic electron transport around PSI is known to slow this process (Asada et al., 1993; Joliot & Joliot, 2006; Okegawa et al., 2007). Plants were dark adapted for 2 h, and then illuminated with far-red light. In C3 F. pringlei and F. robusta, P700 was rapidly oxidized under far-red illumination. In both of these species, it took c. 2 s to reach the maximum oxidation level under far-red light. By contrast, P700 displayed a lag phase and was slowly oxidized under far-red illumination in C4 F. bidentis and C4 F. trinervia (t = 17–19 s, where t is the time required to reach the maximum P700 oxidation level under far-red light). C3–C4 intermediate and C4-like species showed a variety of P700 oxidation rates, which ranged between those of C3 and C4 species of Flaveria. In C3–C4 intermediate F. anomala, P700 was rapidly oxidized, as in C3 species, but showed a short lag phase followed by slow oxidation kinetics (t  = 6 s). In C3–C4 intermediate F. ramosissima, the lag phase of P700 oxidation started earlier and lasted for longer than that in F. anomala (t  = 10 s). In C4-like F. brownii, the P700 oxidation kinetics were similar to those in F. ramosissima (t  = 10 s), whereas they were much slower in C4-like F. palmeri (t  = 16 s), as in C4 species. To investigate whether these slow oxidation kinetics were caused by cyclic electron transport activity, leaf discs prepared from dark-adapted plants were infiltrated with either water or methyl viologen (MV), which inhibits PSI cyclic electron transport activity by diverting electrons from the reducing side of PSI to O2. Infiltration with water slowed slightly the first phase of P700 oxidation kinetics, whereas infiltration with MV strongly enhanced the rate of P700 oxidation and resulted in a complete loss of the second slow oxidation phase in C3–C4 intermediate, C4-like and C4 species (t  = c. 3 s in all species investigated with MV infiltration; Figs 1b, S1). These results showed that the activity of cyclic electron transport around PSI was greatly enhanced in C4-like F. palmeri and C4 species, and was moderately enhanced (to a level similar to the mid-range level of C4 species) in C3–C4 intermediate F. ramosissima and C4-like F. brownii.

Figure 1.

In vivo analysis of photosystem I (PSI) cyclic electron flow activity in C3, C3–C4, C4-like and C4 Flaveria species. (a) P700 oxidation kinetics in response to far-red light illumination in attached leaves of C3 F. pringlei, C3 F. robusta, C3–C4 F. anomala, C3–C4 F. ramosissima, C4-like F. brownii, C4-like F. palmeri, C4 F. bidentis and C4 F. trinervia. Data were normalized to the maximum P700 oxidation level, which was determined under a saturating xenon flash. Kinetics shown are the average from three individual plants. (b) P700 oxidation kinetics in leaf discs of F. bidentis infiltrated with distilled water (dH2O), 50 μmol methyl viologen (MV), 50 μmol dichlorophenyl dimethylurea (DCMU) and 100 μmol dibromomethylisopropyl benzoquinone (DBMIB) before measurement. Kinetics shown are the average from three individual plants. (c) Thermoluminescence signals in C3 F. pringlei, C3–C4 F. ramosissima, C4-like F. palmeri and C4 F. bidentis. The band peaking at 45°C in F. pringlei is the afterglow (AG) band; the band in the range 18–23°C is the B band.

In C4 F. bidentis, infiltration with dibromomethylisopropyl benzoquinone (DBMIB), which inhibits electron transport from plastoquinol to cytochrome b6 by binding at the QO site, also enhanced the rate of P700 oxidation, as did infiltration with MV (Fig. 1b). By contrast, infiltration with dichlorophenyl dimethylurea (DCMU), which inhibits electron transport from QB to plastoquinone, enhanced the rate of P700 oxidation during the first phase, but not during the second slow oxidation phase (t = 16 s). This result suggested that cyclic electron transport operated via the donation of electrons from the stromal reducing pool, but its activity was enhanced by donation of electrons from PSII in C4 F. bidentis.

The PSI cyclic electron transport activities in C3–C4 intermediate, C4-like and C4 species were also evaluated by measuring their AG thermoluminescence signals after far-red illumination (Fig. 1c). Leaves of C3 F. pringlei displayed a typical AG band peaking at c. 45°C, similar to the signals measured previously in other C3 plants, such as Arabidopsis thaliana or Nicotiana tabacum (Havaux et al., 2005). The B band peaking at c. 20°C is caused by charge recombinations within PSII between S2/3 and math formula. These charge recombinations are generated by the weak excitation of PSII by far-red light. Further warming induces a flow of electrons from the stroma to silent S2/3 QB, which leads to the luminescence-emitting S2/3 math formula states, thereby generating the AG band (Miranda & Ducruet, 1995; Ducruet, 2003). It has been shown that this heat-induced electron flow occurs via the cyclic electron pathways (Havaux et al., 2005). In F. bidentis, the AG band was markedly decreased, appearing as a shoulder in a strongly stimulated B band (Fig. 1c). Previous studies have shown that a shift in the AG band towards lower temperatures, which can ultimately result in fusion of the AG band with the B band, corresponds to the stimulation of cyclic electron transport (Ducruet et al., 2005; Apostol et al., 2006; Lintala et al., 2009; Peeva et al., 2012). When cyclic pathways are fully activated, electrons flow freely from the stroma to QB without the need for thermal activation. The thermoluminescence signal from F. palmeri was similar to that measured in F. bidentis, whereas that of F. ramosissima was intermediate. Thus, the thermoluminescence measurements indicated a marked stimulation of the PSI cyclic electron pathway in C4 F. bidentis and C4-like F. palmeri, in agreement with the measurements of P700 oxidation kinetics.

Relative abundances of PGR5, PGRL1 and NDH-H in Flaveria C3, C3–C4 intermediate, C4-like and C4 species

Next, we investigated whether the activity of PSI cyclic electron transport was enhanced by the greater expression of proteins involved in cyclic electron transport. We determined the relative abundances of PGR5, PGRL1 and NDH-H (a subunit of the chloroplastic NDH complex) by immunoblot analysis, and quantified the subunits of major complexes in the photosynthetic electron transport chain in each Flaveria species (Fig. 2). The abundance of the Rieske protein (a subunit of the cytochrome b6f complex) in total membrane proteins was similar among all of the studied species. The abundance of PsbO (a subunit of the PSII complex) was slightly lower in C3–C4 intermediate, C4-like and C4 species (Fig. 2a) than in C3 species. The abundance of PsaC (a subunit of the PSI complex) was slightly higher in C4-like F. palmeri and C4 F. bidentis than in the other species. The abundances of PGR5 and NDH-H were increased in C4 F. bidentis (Fig. 2a). These findings were consistent with the results of our previous study (Munekage et al., 2010). Quantitative analysis showed that the abundance of PGR5 was three times higher and that of NDH-H was > 10 times higher in C4 F. bidentis than in C3 species (Fig. 2b). Consistent with the greater abundance of PGR5, that of PGRL1 was three times higher in C4 F. bidentis than in C3 species. In C3–C4 intermediate species and C4-like F. brownii, the abundances of both PGR5 and PGRL1 were similar to those in C3 species. By contrast, the abundances of PGR5 and PGRL1 were approximately four times higher in C4-like F. palmeri than in C3 species. In C3–C4 intermediate and C4-like species, the level of NDH-H was increased with increases in expression and activity of C4 enzymes, including PEPC, pyruvate, phosphate dikinase (PPDK) and NADP-ME (Cheng et al., 1988; Moore et al., 1988, 1989; Ku et al., 1991; see Fig. S2). The relative abundance of NDH-H in C3–C4 F. ramosissima, C4-like F. brownii and C4-like F. palmeri was two times, seven times and > 10 times higher, respectively, than that in C3 species (Fig. 2b). Consistent with the relative abundance of NDH-H, NDH activity, estimated from the amplitude of the transient increase in chlorophyll fluorescence after actinic light illumination, was 1.3-fold higher in C3–C4 F. ramosissima and 2.5-fold higher in C4-like F. brownii, C4-like F. palmeri and C4 F. bidentis compared with its activity in C3–C4 F. anomala and C3 species (Fig. S3).

Figure 2.

Relative abundances of PROTON GRADIENT REGULATION5 (PGR5), PGR5-LIKE1 (PGRL1), NDH-H (a subunit of the NDH complex), PsaC (photosystem I, PSI), Rieske (Cyt b6f) and PsbO (photosystem II, PSII) in C3, C3–C4, C4-like and C4 Flaveria species. (a) Immunoblot analysis using polyclonal antibodies against PGR5, PGRL1, NDH-H, PsaC, Rieske and PsbO performed on total membrane proteins extracted from leaves of C3 F. pringlei, C3 F. robusta, C3–C4 F. anomala, C3–C4 F. ramosissima, C4-like F. brownii, C4-like F. palmeri, and C4 F. bidentis. Lanes were loaded with 20 μg of protein to detect PGR5 and Rieske proteins, 10 μg of protein to detect PsaC, 5 μg of protein to detect PGRL1 and PsbO, and a dilution series of F. bidentis as indicated. Typical images of immunoblots for each protein are shown. (b) Relative abundances of PGR5, PGRL1, NDH-H, PsaC and Rieske in Flaveria species quantified from chemifluorescence signal intensities in immunoblot analyses by a luminescent image analyser, LAS-4000. Bars are means ± SD of three independent plants. Values for F. bidentis were set to 1.0.

Distribution of PGR5, PGRL1 and NDH-H proteins in M and BS cells

In our previous study, we observed strong immunolabelling signals from both PGR5 and NDH-H in BS chloroplasts in C4 species (Munekage et al., 2010). To quantitatively compare the expression levels of PGR5, NDH-H and PGRL1 between M and BS cells, we fractionated M and BS cells from C4 F. bidentis and performed immunoblot analyses on their proteins. The purity of M cell fractions was estimated by PEPC, and that of BS cells was estimated by RbcL (the large subunit of RuBisCO), because PEPC specifically localizes in M cells and RbcL in BS cells of C4 plants (Moore et al., 1989). The M and BS cell fractions were diluted in a series from 1 to 1 : 16 (Fig. 3a). Contamination of BS cells in the M cell fraction was estimated as < 20%, and contamination of M cells in the BS cell fraction was estimated as < 2%. We performed immunoblotting analyses of PGR5, PGRL1, NDH-H and the Rieske protein in the M and BS fractions (Fig. 3b) and calculated the relative abundances of each protein between M and BS cells taking into account the contamination ratio of each fraction. The relative amount of the Rieske protein was the same in M and BS cells in C4 F. bidentis. The amount of NDH-H was three-fold higher in BS cells than in M cells in C4 F. bidentis, whereas the amounts of PGR5 and PGRL1 were similar in M and BS cells. These results showed that the NDH complex was more abundant in BS cells than in M cells, whereas PGR5 and PGRL1 were present at similar levels in M and BS cells in C4 F. bidentis. Compared with their respective abundances in the total membrane proteins of C3 F. pringlei, the amounts of NDH-H, PGRL1 and PGR5 were all greater in both M and BS cells in C4 F. bidentis.

Figure 3.

Distribution of PROTON GRADIENT REGULATION5 (PGR5), PGR5-LIKE1 (PGRL1), NDH-H (a subunit of the NDH complex) and Rieske proteins between mesophyll (M) cells and bundle sheath (BS) cells in C4 Flaveria bidentis. (a) Estimation of the purity of the M cell fraction (MCF) and BS cell fraction (BSCF). Phosphoenolpyruvate carboxylase (PEPC) and RbcL (the large subunit of RuBisCO) were used as specific markers for M cells and BS cells, respectively. Soluble proteins extracted from MCF (MC-S) and BSCF (BSC-S), and total soluble proteins (TS), were loaded onto an acrylamide gel (5 μg per lane) and immunolabelled using antiserum against PEPC and RbcL. From the dilution series of MC-S and BSC-S (1 to 1 : 16), the contamination of BS cells in MCF was estimated as < 20% and the contamination of M cells in BSCF was estimated as < 2%. (b) Immunoblot analysis of PGR5, PGRL1, NDH-H and Rieske proteins in M membrane proteins extracted from MCF (MC-M) and BS membrane proteins extracted from BSCF (BSC-M), and total membrane proteins (TM). Lanes were loaded with 20 μg of protein to detect PGR5, NDH-H and Rieske, 5 μg of protein to detect PGRL1, and a dilution series of membrane proteins of each fraction.

Next, we analysed the two cell types, M and BS cells, to determine which cell type contained higher levels of PGR5 and NDH-H in C4-like species. We performed in situ immunolabelling using specific antibodies against PGR5 and NDH-H in leaves of C4-like F. brownii, C4-like F. palmeri and C4 F. bidentis (Fig. 4). Transverse sections of leaves stained with toluidine blue O showed numerous chloroplasts in both M and BS cells in all of the studied species (Fig. 4a–c). Transverse sections prepared from the same leaf samples were labelled with either preimmune serum or immune serum (Fig. 4d–i for NDH-H, j–o for PGR5) and then labelled with secondary antibodies conjugated to FITC. We obtained overlaid images of the FITC fluorescence in green and bright field by confocal microscopy (Fig. 4g–i for NDH-H, m–o for PGR5). The background labelling with preimmune serum was very low in all cases (Fig. S4; Munekage et al., 2010). In the section immunolabelled for NDH-H, there was a strong FITC signal in BS chloroplasts and a faint FITC signal in M chloroplasts in C4 F. bidentis (Fig. 4d,g). In the section immunolabelled for PGR5, there was an FITC signal in both M and BS chloroplasts, but it was stronger in BS chloroplasts than in M chloroplasts in C4 F. bidentis (Fig. 4j,m), as reported in our previous study (Munekage et al., 2010). In C4-like F. palmeri, there was a strong NDH-H immunolabelling signal in BS chloroplasts and a weaker signal in M chloroplasts (Fig. 4e,h), whereas the intensity of the PGR5 immunolabelling signal was equivalent between M and BS chloroplasts (Fig. 4k,n). In C4-like F. brownii, the NDH-H immunolabelling signal was also observed in BS chloroplasts (Fig. 4f,i), whereas the PGR5 signal was very weak, indicating lower expression of PGR5 in F. brownii than in C4-like F. palmeri (Figs 2b, 4l,o). In C3 F. pringlei, the PGR5 and NDH-H immunolabelling signals were very weak (data not shown). These results showed that the abundance of the NDH complex was exclusively increased in BS chloroplasts in F. palmeri and F. brownii, whereas that of PGR5 was increased in both M and BS chloroplasts in C4-like F. palmeri.

Figure 4.

In situ immunolocalization of PROTON GRADIENT REGULATION5 (PGR5) and NDH-H (a subunit of the NDH complex) in leaf tissue of C4-like Flaveria brownii, C4-like F. palmeri and C4 F. bidentis. Transverse sections of lamina were stained with toluidine blue O for anatomical observations (a–c), or with anti-NDH-H or anti-PGR5 sera. The localization of PGR5 (d–i) and NDH-H (j–o) was visualized by the green fluorescence of fluorescein isothiocyanate (FITC)-labelled antibody. FITC fluorescence in green (d–f, j–l) and overlaid images of FITC fluorescence in green and bright field (g–i, m–o) were visualized by confocal microscopy. Bars, 50 μm.

Ultrastructure of BS chloroplasts in Flaveria C3, C3–C4 intermediate, C4-like and C4 species

To evaluate the development of grana stacks of BS chloroplasts in C3, C3–C4 intermediate, C4-like and C4 species, we investigated the ultrastructure of BS and M chloroplasts in the studied species by transmission electron microscopy (Fig. 5 for BS chloroplasts, data not shown for M chloroplasts). We measured the total length of grana and stroma thylakoid membranes of M and BS chloroplasts in each Flaveria species (Table 1, Fig. S5), and then calculated the grana index (%; total length of grana/total length of thylakoid membranes × 100) in M and BS chloroplasts (Fig. 6a). The grana index of M chloroplasts was similar among C3, C3–C4 intermediate, C4-like and C4 species (> 50%). It has been reported that C3 plants also have BS cells containing a small number of chloroplasts (Brown & Hattersley, 1989; Kinsman & Pyke, 1998; Muhaidat et al., 2011). We investigated the chloroplasts of BS cells adjacent to the vascular bundle in C3 F. pringlei and C3 F. robusta. The M cells contained many chloroplasts, whereas the transverse sections of BS cells showed only two or three chloroplasts per cell in both F. pringlei and F. robusta. These BS chloroplasts had some developed grana structures (Fig. 5a,b). In C4 species, grana stacks were dramatically reduced in BS chloroplasts (Fig. 5g,h). The grana index of BS chloroplasts was 15% in F. bidentis and 19% in F. trinervia (Fig. 6a); this was consistent with the results of a previous report showing that the PSII complex and PSII activity are not depleted completely in these species (Hofer et al., 1992). In two C3–C4 species, F. anomala and F. ramosissima, the BS chloroplasts contained well-developed grana thylakoids (Fig. 5c,d). The grana index of BS chloroplasts was 57% in F. anomala and 63% in F. ramosissima (Fig. 6a). In these species, there were many mitochondria located near chloroplasts. Well-developed grana thylakoids were also observed in BS chloroplasts of F. brownii (Fig. 5e), in which the grana index was 50% (Fig. 6a). However, in C4-like F. palmeri, grana stacks were absent from the central area of BS chloroplasts, but present in the peripheral region of the chloroplast envelope (Fig. 5f). The grana index of BS chloroplasts in C4-like F. palmeri was 16%, similar to that in C4 species (Fig. 6a). However, the number of thylakoid membrane stacks per grana was greater in C4-like F. palmeri than in C4 species (Fig. 6b). In C4 species, 65–75% of grana consisted of three to four layers of thylakoid membranes, and < 1% of grana had > 10 layers of thylakoid membranes. However, 35% of the grana in C4-like F. palmeri had three to four layers of thylakoid membranes, and 15% had > 10 layers.

Table 1. Total lengths of grana thylakoid membrane and stroma thylakoid membrane in palisade mesophyll (PM) chloroplasts and bundle sheath (BS) chloroplasts in Flaveria C3, C3–C4 intermediate, C4-like and C4 species
 SpeciesCell typeGrana thylakoid length (nm)Stroma thylakoid length (nm)Total thylakoid length (nm)
  1. Values for the lengths of grana thylakoid, stroma thylakoid and total thylakoid are averages (= 15; five chloroplasts per plant, three individual plants).

C4 F. trinervia PM57 88846 459104 347
BS31 883136 053167 936
F. bidentis PM75 32657 962133 288
BS23 382135 296158 678
C4-like F. palmeri PM121 63793 115214 752
BS67 151347 522414 673
F. brownii PM31 46625 37856 844
BS46 09945 92892 027
C3–C4 F. ramosissima PM64 29232 15296 444
BS90 25951 961142 220
F. anomala PM75 41430 955106 369
BS58 32443 780102 104
C3 F. robusta PM48 92626 66375 589
F. pringlei PM40 10721 60861 715
Figure 5.

Ultrastructure of bundle sheath (BS) chloroplasts in C3, C3–C4, C4-like and C4 Flaveria species. Transmission electron micrographs of BS chloroplasts in C3 F. pringlei (a), C3 F. robusta (b), C3–C4 F. anomala (c), C3–C4 F. ramosissima (d), C4-like F. brownii (e), C4-like F. palmeri (f), C4 F. bidentis (g) and C4 F. trinervia (h). In C3–C4 F. anomala and C3–C4 F. ramosissima, mitochondria were located near chloroplasts (c, d). Bars, 2.5 μm.

Figure 6.

Grana index and number of thylakoids per granum of mesophyll (M) chloroplasts and bundle sheath (BS) chloroplasts in Flaveria C3, C3–C4 intermediate, C4-like and C4 species. (a) Grana index in M chloroplasts and BS chloroplasts in C3 F. pringlei, C3 F. robusta, C3–C4 F. anomala, C3–C4 F. ramosissima, C4-like F. brownii, C4-like F. palmeri, C4 F. bidentis and C4 F. trinervia. The grana index (%) represents the total length of grana/total length of thylakoid membrane × 100 in M chloroplasts (grey bar) and BS chloroplasts (white bar). Bars are means ± SD (= 15). (b) Relative distribution of thylakoid membranes per granum in a chloroplast of BS cells in C4 F. trinervia, C4 F. bidentis and C4-like F. palmeri. Bars are means ± SD (= 5).

Discussion

Promotion of cyclic electron transport with development of the C4 cycle

In our comparative analyses of C3, C3–C4 intermediate, C4-like and C4 Flaveria species, we observed a strong correlation between the promotion of cyclic electron transport and C4 cycle activity reported in the literature. In C3–C4 intermediate F. anomala, the operation of a functional C4 cycle is not obvious (Rumpho et al., 1984; Monson et al., 1986). Therefore, its lower compensation point for CO2 than that of C3 species is thought to be achieved via glycine shuttling in the C2 cycle (Monson et al., 1986). This is supported by the fact that there were abundant mitochondria near chloroplasts in BS cells, as often observed in C3–C4 intermediate plants (Fig. 5c; Brown et al., 1983; Brown & Hattersley, 1989). In F. anomala, P700 oxidation was not significantly slowed, and there were no detectable increases in the levels of NDH-H, PGRL1 and PGR5 (Figs 1a, 2b; Munekage et al., 2010), indicating that PSI cyclic electron transport was not promoted in this species. By contrast, in C3–C4 intermediate F. ramosissima, in which functional integration between the C4 cycle and the C3 cycle has been suggested (Monson et al., 1986; Moore et al., 1988), PSI cyclic electron transport activity estimated by both P700 oxidation and AG thermoluminescence increased to a level approximating the middle level of the range in C4 species (Fig. 1a,c). Increased activity of cyclic electron transport was also observed in C4-like F. brownii (Fig. 1a), in which the operation of the C4 cycle is more evident (Cheng et al., 1988; Ku et al., 1991). Although the time required to reach the maximum P700 oxidation level was similar in F. ramosissima and F. brownii, the lag phase began earlier in F. brownii than in F. ramosissima (Fig. 1a), indicating that PSI cyclic electron transport activity was higher in F. brownii than in F. ramosissima. This is probably because of the increased amount of the NDH complex, as NDH-H was increased two-fold in C3–C4 intermediate F. ramosissima and seven-fold in C4-like F. brownii, compared with that in C3 species. This increase in NDH-H corresponded to increased NDH activity, as estimated by chlorophyll fluorescence (Fig. S3), whereas neither the expression of PGR5 or PGRL1 or the thylakoid structure showed changes in these species. These results showed that the increase in cyclic electron transport activity involving the NDH complex was dependent on the activity of the C4 cycle; further regulatory mechanisms are needed to explain the expression of PGR5 and PGRL1 and the structural changes in thylakoid membranes induced in F. palmeri and C4 species. It is tempting to speculate that the increase in the amount of the NDH complex was induced by the accumulation of NADPH, resulting from the higher requirement for ATP than for NADPH during operation of the C4 cycle. In the NADP–ME-type C4 cycle, NADPH is released during the decarboxylation step by NADP–ME, leading to the accumulation of NADPH in BS chloroplasts. This could have occurred in F. brownii because NADP–ME is highly expressed in BS chloroplasts (Cheng et al., 1988), and we observed BS cell-selective accumulation of NDH-H by immunolabelling. However, in the glycine shuttle of the C2 cycle, there is no obvious change in the ATP/NADPH requirement in chloroplasts compared with that in C3 photosynthesis. If the flux of the C2 cycle is increased in C3–C4 intermediate species, the required ATP/NADPH ratio would be increased slightly in M chloroplasts during the step in which glycerate is phosphorylated by glycerate 3-kinase to form 3-phosphoglyceric acid. In this case, the imbalance of ATP and NADPH might not be sufficiently large to increase the level of the NDH complex, especially in F. anomala.

There was a marked increase in cyclic electron transport activity in C4-like F. palmeri, C4 F. bidentis and C4 F. trinervia. In these species, NDH-H was increased > 10-fold and PGR5 and PGRL1 were increased three- to four-fold compared with their respective abundances in C3 species. In contrast with NDH-H, which accumulated exclusively in BS cells, the distribution of PGR5 was similar in M and BS cells (Figs 2, 4). These results suggested that the expression of PGR5 and the NDH complex was controlled differently in C4-like F. palmeri and C4 species. PGR5 and PGRL1 have been reported to be important factors for FQR activity (Munekage et al., 2002, 2004, 2008; DalCorso et al., 2008). The overexpression of PGR5 in Arabidopsis thaliana resulted in increased activity of cyclic electron transport (Okegawa et al., 2007). Moreover, PGRL1 has been reported to accept electrons from ferredoxin in a PGR5-dependent manner (Hertle et al., 2013). Therefore, the increased abundances of PGR5 and PGRL1 in C4-like F. palmeri and C4 Flaveria are considered to contribute to increased cyclic electron transport activity.

Effect of decreased grana thylakoid structure on the promotion of cyclic electron transport activity

In C4-like F. palmeri, C4 F. bidentis and C4 F. trinervia, there were markedly fewer grana stacks in BS chloroplasts, compared with those in C4-like F. brownii and C3–C4 intermediate species (Figs 5, 6a). The reduced grana stacks have been reported to be correlated with lower PSII activity (Sheen et al., 1987; Oswald et al., 1990). Corresponding to thylakoid structure, the activities of PSII and linear electron transport were reported to be decreased in BS cells in F. palmeri, F. bidentis and F. trinervia, but not in F. brownii (Hofer et al., 1992; Ketchner & Sayre, 1992). Conversely, the total length of stroma lamellae per chloroplast was dramatically increased in BS chloroplasts in F. palmeri, F. bidentis and F. trinervia compared with that in M chloroplasts or BS chloroplasts of other species (Table 1). Although PGR5 and the NDH complex are located in stroma lamellae (Rumeau et al., 2005; Munekage et al., 2010), the abundances of these proteins were not correlated with the ratio of the length of grana thylakoids to the length of stroma lamellae in Flaveria species, indicating that the accumulation of PGR5 and the NDH complex is independent of thylakoid structure. Interestingly, in F. palmeri, grana stacks were absent from the central area of BS chloroplasts, but were located in the end-membrane regions (Fig. 4f). The number of thylakoid membranes per granum was as large in F. palmeri as in C3–C4 intermediate species (Fig. 6b). However, we observed extremely long stroma lamellae between grana in F. palmeri, resulting in a low grana index (Fig. 6a). As the PSI cyclic electron transport pathway shares electron carriers with the linear electron transport pathway from plastoquinone to ferredoxin, structural changes in the thylakoid membrane would be important to avoid competition between these two pathways. Considering the result that the cyclic electron transport activity of F. palmeri was as high as that in C4 species, compartmentalization of grana stacking and stroma lamella regions is feasible to avoid competition between cyclic electron transport operating in the stroma lamellae and linear electron transport operating in the grana.

A phylogenetic study showed that there are two evolutionarily distinct lineages, Clades A and B, in the genus Flaveria. Flaveria ramosissima, F. palmeri, F. bidentis and F. trinervia are classified in Clade A, and F. anomala and F. brownii are classified in Clade B (McKown et al., 2005). Among the species in Clade B, F. brownii has the photosynthesis phenotype closest to C4 photosynthesis (with C4-like features). There are no species with true C4 photosynthesis in Clade B. Considering the evidence that both clades include species with increased NDH-H, but only Clade A species show structural changes in BS chloroplasts, the genetic background inducing structural changes in thylakoid membranes probably differs from that inducing the expression of the NDH complex. As NADPH is produced by decarboxylation of malate in BS cells in NADP–ME-type C4 photosynthesis, suppression of PSII activity, as well as an increase in cyclic electron transport activity, is favourable for NADP–ME-type C4 photosynthesis. This genetic alteration may have been important in the establishment of true NADP–ME-type C4 photosynthesis from the C3–C4 intermediate stage during evolution.

In conclusion, our data provide evidence that cyclic electron transport is promoted by C4 cycle activity in C3–C4 intermediate, C4-like and C4 Flaveria species. The increase in NDH-H in BS cells is closely associated with C4 cycle activity, whereas an increase in PGR5 in both M and BS cells and a decrease in grana stacks in BS chloroplasts occur in C4-like F. palmeri and C4 species classified into Clade A. Together with the observation that PSI cyclic electron transport is markedly increased in C4-like F. palmeri and C4 species, our results show that both the expression of the NDH complex and of PGR5 and PGRL1, and structural changes in thylakoid membranes, contribute to promote PSI cyclic electron transport activity in NADP–ME-type C4 photosynthesis.

Acknowledgements

We thank Peter Westhoff for his gift of seeds of Flaveria species. Antibodies against PEPC, PGRL1, PsbO and NDH-H were kindly provided by Tsuyoshi Furumoto, Toru Hisabori, the late Akira Watanabe and Dominique Rumeau, respectively. We thank Shio Inoue, Ayumi Yonemitsu and Rina Nagai for technical assistance. This work was partly supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (Grant no. 22770039) and a Funding Program for Next Generation World-Leading Researchers from the Japan Society for the Promotion of Science (Grant no. GS019).

Ancillary