Metabolite profiles reveal interspecific variation in operation of the Calvin-Benson cycle in both C4 and C3 plants

Abstract

Low atmospheric CO2 in recent geological time led to the evolution of carbon-concentrating mechanisms (CCMs) such as C4 photosynthesis in >65 terrestrial plant lineages. We know little about the impact of low CO2 on the Calvin-Benson cycle (CBC) in C3 species that did not evolve CCMs, representing >90% of terrestrial plant species. Metabolite profiling provides a top-down strategy to investigate the operational balance in a pathway. We profiled CBC intermediates in a panel of C4 (Zea mays, Setaria viridis, Flaveria bidentis, and F. trinervia) and C3 species (Oryza sativa, Triticium aestivum, Arabidopsis thaliana, Nicotiana tabacum, and Manihot esculenta). Principal component analysis revealed differences between C4 and C3 species that were driven by many metabolites, including lower ribulose 1,5-bisphosphate in C4 species. Strikingly, there was also considerable variation between C3 species. This was partly due to different chlorophyll and protein contents, but mainly to differences in relative levels of metabolites. Correlation analysis indicated that one contributory factor was the balance between fructose-1,6-bisphosphatase, sedoheptulose-1,7-bisphosphatase, phosphoribulokinase, and Rubisco. Our results point to the CBC having experienced different evolutionary trajectories in C3 species since the ancestors of modern plant lineages diverged. They underline the need to understand CBC operation in a wide range of species.

Keywords: C3; C4; Calvin–Benson cycle; interspecies variation; metabolite profiles; photosynthesis.

Fig. 1.

CBC metabolite and 2PG profiles in different species. Growth and harvest conditions can be found in Supplementary Table S1. Note that 2PG amounts are multiplied by 10 for better visibility. The results are shown as mean (nmol g FW⁻¹) ±SD. The original data are provided in Supplementary Dataset S1.

Fig. 2.

CO₂ assimilation rate in Z. mays and A. thaliana, and CBC metabolite profiles in Z. mays and A. thaliana at different short-term irradiances. Zea mays and A. thaliana were grown at 550 µmol m⁻² s⁻¹ and 120 µmol m⁻² s⁻¹ irradiance, respectively. CO₂ assimilation rate in (A) Z. mays (n=10) and (B) A. thaliana (n=9). The results are shown as mean (µmol CO₂ m⁻² s⁻¹ and µmol CO₂ g FW⁻¹ h⁻¹, for Z. mays and A. thaliana, respectively) ±SD. Arrows indicate the irradiances at which leaves were sampled for metabolite analysis. (C) Zea mays was harvested at growth irradiance (Zm, medium irradiance) or after being subjected for 4 h to 160 µmol m⁻² s⁻¹ (ZmL, low irradiance) from the beginning of the light period. Arabidopsis thaliana was harvested at growth irradiance (At, medium irradiance) or subjected for 15 min to 80 µmol m⁻² s⁻¹ or 280 µmol m⁻² s⁻¹ (AtL, low and AtH, high irradiances, respectively). Quenching of metabolism and harvest of leaf tissue were performed at least 4 h after the beginning of the light period. 2PG amounts are multiplied by 10 for better visibility. Asterisks indicate graphs already presented in Fig. 1. The results are shown as mean (nmol g FW⁻¹) ±SD. (D) Correlation analysis. The metabolite data shown in (B) and (C) were used to perform Pearson’s correlation analysis between data sets from the same species at different irradiances, and correlations between different species. Before performing the correlation analysis, each data set was normalized by calculating the amount of carbon in a given metabolite, and dividing it by the total carbon in all metabolites in that data set. This was done to avoid secondary correlation due to any interspecies differences in leaf composition. The results are given as r and the higher correlations are indicated in bold. All correlations were positive. The original data are presented in Supplementary Dataset S1.

Fig. 3.

Total chlorophyll (A) and protein (B) content in different species. Measurements were performed in Z. mays at low and medium irradiance (ZmL and Zm, respectively), S. viridis (Sv), F. bidentis (Fb), F. trinervia (Ft), O. sativa (Os), T. aestivum (Ta), and A. thaliana at low, medium, and high irradiance (AtL, At, and AtH, respectively), N. tabacum (Nt) and M. esculenta (Me). Growth and harvest conditions can be found in Supplementary Table S1. The results are shown as mean (mg g FW⁻¹) ±SD. The original data are presented in Supplementary Dataset S1.

Fig. 4.

Principal component analyses of the CBC metabolite profiles in all tested species. The analyses were performed on the metabolite data set excluding 2PG and SBP (2PG was omitted to avoid systematic bias between C₃ and C₄ species due the differing rates of RuBP oxygenation, and SBP was omitted because in some species part of the pool may not be involved in the CBC; see text for details). Metabolite amounts were normalized on (A) FW, (B) total chlorophyll content, or (C) protein content, or (D) were transformed into a dimensionless data set. For dimensionless data set determination, in a given sample, the level of each metabolite was first transformed to C equivalent values by multiplying the amount (nmol g FW⁻¹) by the number of C atoms in the metabolite. The C equivalent amounts of all CBC intermediates plus 2PG were then summed. In the last step, the C equivalent value of a given metabolite was divided by the summed C equivalent values. The transformed values and calculation steps are provided in Supplementary Dataset 1. This transformation generates a dimensionless data set (provided in Supplementary Dataset S1) in which each metabolite receives a value equal to its fractional contribution to all the C in CBC metabolites plus 2PG. As this data set is dimensionless, there is no systematic bias due to differences in leaf composition. The distribution of C₄ species (green) and C₃ species (black) is shown on PC1 and PC2 (Z. mays, Zm and ZmL; S. viridis, Sv; F. bidentis, Fb; F. trinervia, Ft; O. sativa, Os; T. aestivum, Ta; A. thaliana, AtL, At, and AtH; N. tabacum, Nt; M. esculenta, Me). The loadings of CBC intermediates in PC1 and PC2 are shown in red. Principal component analyses with the full metabolite data set and with all metabolites except either 2PG or SBP are shown in Supplementary Fig. S4 (amounts normalized on FW), Supplementary Fig. S5 (amounts normalized on total chlorophyll content), Supplementary Fig. S6 (amounts normalized on protein content), and Supplementary Fig. S7 (dimensionless). The original data are presented in Supplementary Dataset S1.

Fig. 5.

Principal component analysis of the CBC metabolite contents in C₃ species only. The analyses were performed on the metabolite data set excluding SBP. Metabolite data were normalized on (A) total chlorophyll content and (B) total protein, and (C) using a dimensionless data set (see legend of Fig. 4). The distribution of C₃ species is shown on PC1 and PC2 (O. sativa, Os; T. aestivum, Ta; A. thaliana, AtL, At, and AtH; N. tabacum, Nt; M. esculenta, Me). The loadings of CBC intermediates in PC1 and PC2 are shown in red. PC analyses with the full metabolite data set and with all metabolites except either 2PG or SBP are shown in Supplementary Fig. S8 (amounts normalized on total chlorophyll content), Supplementary Fig. S9 (amounts normalized on total protein), and Supplementary Fig. S10 (dimensionless). The original data are presented in Supplementary Dataset S1.

Fig. 6.

Coefficient of variance (CV) of CBC metabolites between species. The CV (SD/mean×100) is a standardized quantity describing the dispersion of a population distribution (Simpson and Roe, 1939). The analysis was performed on a dimensionless data set that was generated as described for Fig. 4D. The transformed data (presented in Supplementary Dataset S1) were used to calculate the bootstrapped CV for each metabolite (30 bootstrap iterations). The 95% confidence interval was estimated using the basic bootstrap method. Statistically significant differences between metabolites are indicated by letters (ANOVA on the bootstrap results followed by the Tukey’s HSD post-test). (A) All species, (B) only C₄ species, and (C) only C₃ species.

Fig. 7.

Correlation between levels of CBC metabolites. The correlations were performed with all individual samples from a dimensionless data set, generated as described for Fig. 4D. The transformed data were used to calculate the Pearson’s correlation matrix on every pair of metabolites. Correlation values are given in the figure panels and indicated by a heat map. The adjacent dendrograms show clusters defined using the complete linkage method (Sørensen, 1948). Non-significant correlations (P≥0.05; two-sided Student’s t-test) are set as zero. Metabolite pairs that are linked by irreversible reactions are indicated by a black box. (A) All species, (B) only C₄ species, and (C) only C₃ species. An alternative display is provided in Supplementary Fig. S11, with the same fixed order of metabolites in each panel, corresponding to the reaction sequence in the CBC. The same heat map scale is used for (A–C). (D) Schematic representation of interspecies variance in the ratio of substrate abundance:product abundance for different CBC enzymes. Enzymes that catalyse irreversible reactions are highlighted in bold. For each enzyme reaction, the substrate and product that were compared are indicated in the list below the display. This display is schematic because some metabolites were not measured (erythrose 4-phosphate, E4P; and glyceraldehyde 3-phosphate, GAP) or were not separated (Ru5P and Xu5P). For reactions using GAP, it is assumed that GAP and dihydroxyacetone phosphate (DHAP) are in equilibrium. For transketolase (TK), two reactions were separated (termed TK^a and TK^b). For TK^a, the reactant E4P was missing, and only the relationships between triose-P and F6P and Ru5P+Xu5P are shown. For Tk^b, the plot focuses on the relationship between S7P and R5P or Ru5P+Xu5P. The display shows the alternative pairs of metabolites compared, with the upper and lower symbols in the display corresponding to the upper and lower pair in the list. A similar display mode is used for the carboxylation and oxygenation reactions of Rubisco. The correlation coefficients are taken from (B) and (C), using the same heat map scale. Results are shown separated for correlations between the four C₄ species (squares) and the five C₃ species (circles). The analysis is not shown for the combined C₄ plus C₃ species set because, in this case, some relationships are driven by differences between C₄ and C₃ species. Additional abbreviations: fructose 1,6-bisphosphate aldolase (FBP ald), phosphoglycerate kinase (PGK), ribose 5-phosphate isomerase (R5P isom), sedoheptulose 1,7-bisphosphate aldolase (SBP ald).