Efficient storage of carbon in seeds is crucial to plant fitness and to agricultural productivity. Oil is a major reserve material in most seeds1, and these oils provide the largest source of renewable reduced carbon chains available from nature. However, the conversion of carbohydrate to oil through glycolysis results in the loss of one-third of the carbon as CO2. Here we show that, in developing embryos of Brassica napus L. (oilseed rape), Rubisco (ribulose 1,5-bisphosphate carboxylase/oxygenase) acts without the Calvin cycle2 and in a previously undescribed metabolic context to increase the efficiency of carbon use during the formation of oil. In comparison with glycolysis, the metabolic conversion we describe provides 20% more acetyl-CoA for fatty-acid synthesis and results in 40% less loss of carbon as CO2. Our conclusions are based on measurements of mass balance, enzyme activity and stable isotope labelling, as well as an analysis of elementary flux modes.
During embryogenesis, seeds receive carbon precursors from the mother plant for the synthesis of storage products. In oilseeds such as B. napus, the dominant metabolic flux is the conversion of sugars into triacylglycerols3, resulting in more than 60% of the carbon being stored as oil (about 45% of dry weight). In described pathways for plant oil synthesis, sucrose is converted into pyruvate through glycolysis, which is then transformed into acetyl-CoA, the precursor of fatty-acid biosynthesis4. This conversion of sugars to oil entails the loss of one carbon as CO2 at the pyruvate dehydrogenase (PDH) reaction for each two-carbon acetyl-CoA unit produced for oil synthesis (Fig. 1b). This loss of carbon seems to make oil storage less advantageous for seeds and raises the question why so many plant species use oil as their primary reserve in seeds and whether there is a more efficient way of transforming carbohydrates to oil. Because many seeds are green and have photosystems, even the decreased amount of light that penetrates the fruit wall5 can provide reductant and ATP that may expand the range of pathways available for storage product synthesis.
To measure the efficiency of carbon utilization during oilseed development, B. napus embryos were fed uniformly 14C-labelled carbon sources, and their conversion into oil, protein, carbohydrates and CO2 was determined. Table 1 shows the partitioning of carbon into different biomass fractions and the proportion of carbon liberated as CO2. The measured ratio of carbon stored in oil to carbon liberated as CO2 was close to 3:1, which was substantially higher than expected. The formation of acetyl-CoA from pyruvate by PDH results in a ratio of 2:1, and CO2 production from additional metabolic activities (such as the oxidative pentose-phosphate pathway (OPPP)6 and the tricarboxylic acid (TCA) cycle3) would decrease the ratio further. Thus, considerably less CO2 was produced than expected from the amount of oil formed.
We considered whether the decreased CO2 emission in light was due to its refixation by means of phosphoenolpyruvate carboxylase or pyruvate carboxylase. The product, oxaloacetate, could be converted into amino acids and stored in proteins or possibly secreted from the embryo (for example, in reduced form as malate). However, on the basis of the amino acid composition of B. napus embryos, recovery of CO2 through oxaloacetate into seed protein can account for only about 4% of the CO2 released by PDH (see Supplementary Information I). In addition, no export of malate or other such fixation products from embryos was detected (NMR data not shown). The observation of substantial Rubisco activity in B. napus seeds5,7 indicates that the reassimilation of CO2 by the Calvin cycle might explain the increased efficiency of carbon use. The catalytic capacities of both Rubisco and phosphoribulokinase in developing B. napus seeds are sufficient to potentially fix all the CO2 released by PDH7, and seed CO2 levels8 saturate Rubisco carboxylase activity and prevent its oxygenase activity.
The results of labelling experiments show that CO2 fixation by Rubisco is indeed active but that the Calvin cycle is not functional. Developing embryos were cultured in an atmosphere containing 2% 13CO2—a concentration similar to the CO2 levels that we and others5 measured in the gas-filled spaces, or locules, of developing siliques. Alternatively, alanine in the medium was replaced with [1-13C]alanine or [U-13C3]alanine, which results in the release of 13CO2 internally through the action of alanine aminotransferase and then PDH. In both experiments the distribution of 13C in amino acids and fatty acids shows that only the C1 carbon position of 3-phosphoglyceric acid (PGA) became labelled (Table 2). This is where CO2 fixed by Rubisco is located; fatty acids, which are derived from C2 and C3 of PGA (Fig. 1b), were labelled to extremely low levels (Table 2). This labelling pattern is incompatible with flux through the Calvin cycle, in which the cyclic regeneration of ribulose-1,5-bisphosphate from PGA results in label from CO2 being distributed into all carbon positions of the cycle's intermediates (including PGA)2. We therefore concluded that although Rubisco is active in fixing CO2 by the usual carboxylation reaction, it is not operating as part of the Calvin cycle but in a different context.
On the basis of the above results we conclude that Rubisco operates as part of a previously undescribed metabolic route between carbohydrate and oil (Fig. 1a). This involves three stages: first, the conversion of hexose phosphates to ribulose-1,5-bisphosphate by the non-oxidative reactions of the OPPP together with phosphoribulokinase; second, the conversion of ribulose-1,5-bisphosphate and CO2 (most of which is produced by PDH3) to PGA by Rubisco; and third, the metabolism of PGA to pyruvate and thence to fatty acids (Fig. 1a). The net carbon stoichiometry of this conversion (for details see Supplementary Information II) is
By contrast, the conversion of the same amount of hexose phosphates through glycolysis is
Thus, Rubisco, together with the non-oxidative enzymes of the pentose phosphate pathway, allows the conversion of carbohydrate into 20% more acetyl-CoA and therefore oil than does glycolysis, with 40% less carbon lost as CO2. This metabolic route is consistent with the mass balance data shown in Table 1, which, given the absence of substantial other CO2 refixing mechanisms, are incompatible with the glycolytic route. Although Rubisco has a central function in the metabolic route that we describe, net fixation of CO2 does not occur because the CO2 fixed by Rubisco is subsequently released by PDH. The increase in carbon economy is therefore achieved only through the combined activity of Rubisco with the non-oxidative reactions of the pentose phosphate pathway.
We used two independent approaches to estimate how large a contribution this metabolic route makes to seed oil synthesis (at a light intensity of 50 µmol m-2 s-1).
First, the carbon balance of the embryo was used. The ratio of fatty-acid synthesis to CO2 release can be formulated as (see Fig. 1b)
where VPDH and VRuB are the fluxes through PDH and Rubisco respectively and VX is the net CO2 production due to the rest of metabolism (primarily the OPPP and the TCA cycle). This can be recast to allow VRuB to be deduced from measurements of R and VX:
We measured R to be 2.9 (Table 1) and determined that VX is positive and small (see refs 3 and 5 and Supplementary Information). Accordingly, VRuB > 0.31VPDH. Because the great majority of PGA is converted to pyruvate and thence to acetyl-CoA9, the total rate of PGA production is very close to VPDH. Because two PGA molecules are produced by the Rubisco carboxylase reaction, the rate of PGA production by Rubisco is 2VRuB. Thus, Rubisco is responsible for more than 62% of the total PGA production (46–75%, considering the range for R (Table 1)).
Second, we supplied [1-13C]Ala and [U-13C3]Ala to developing embryos, which results in the release of 13CO2 internally through the PDH reaction. The flux through Rubisco was estimated from the level of 13C in the C1 position of PGA-derived amino acids that arises from the direct incorporation of 13CO2 by Rubisco. In the absence of flow round the Calvin cycle, the PGA that is produced by glycolysis from triose phosphate will be unlabelled, whereas half the PGA molecules produced by Rubisco are labelled in the C1 position (Fig. 1b). Thus, the labelling level in PGA is a function of the relative contributions of glycolytic flux and the flux through Rubisco, taking into account the enrichments in the precursors triose phosphate and ribulose-1,5-bisphosphate. The relationship (derived from metabolic and isotopic steady-state equations; see Supplementary Information III) is
where VRub and VGAPDH are the fluxes through Rubisco and glyceraldehyde 3-phosphate dehydrogenase, respectively; and FPGA(1) are the fractional 13C enrichments in CO2 within the embryos and in the C1 position of PGA, respectively. We determined FPGA(1) from Phe and Tyr carbon position 1 and we estimated from label in C1 of valine (Table 2, Fig. 1b). On the basis of the data in Table 2 (using FPhe(1) as well as FTyr(1)) and equation (3) we estimate that the flux through Rubisco is responsible for the production of 37–51% of PGA in the developing embryo. This estimate is a minimum because CO2 produced by the OPPP is unlabelled, thus rendering the real value for smaller than estimated.
We next considered whether any other metabolic routes are possible for the conversion of hexose to fatty acids that could account for our observations. To do this we used elementary flux mode analysis10,11 applied to the enzymes of glycolysis, the OPPP, the Calvin cycle and fatty-acid synthesis (see Supplementary Information IV). This analysis showed that only the group of elementary flux modes that include the bypass of glycolysis by Rubisco can explain the observed increase in carbon conversion efficiency and labelling results. Three additional sets of flux modes were identified that describe the conventional glycolytic route, the Calvin cycle and the conversion of hexose phosphates to pentose phosphates through the oxidative decarboxylation of hexose with the subsequent formation of PGA through phosphoribulokinase and Rubisco. Linear combinations of the different flux modes showed that increased carbon efficiency is achieved only by increasing flux through Rubisco but comes at the cost of increasing requirement for external (photosynthetic) NADPH. The analysis also shows that with only a small fraction (less than 15%) of the NADPH or ATP required for the Calvin cycle, Rubisco in seeds can carry most of the metabolic flux to oil (see Supplementary Information IV). The flux distributions that are actually available to a developing seed will depend on the amount of light available for generating ATP and NADPH by photosynthetic electron transport. Indeed, as shown in Fig. 2, the ratio of carbon stored in oil to carbon liberated as CO2 increases linearly with light intensity between 10 and 100 µmol m-2 s-1.
In addition to demonstrating the operation of the new pathway in B. napus and the increased carbon efficiency that it confers, we also assessed the efficiency of carbon metabolism in developing non-green seeds. For two sunflower varieties developing in culture, the oil:CO2 ratio ranged from 1.2 to 1.6 (n = 3 for each variety). This is similar to the ratio expected for the conventional glycolytic pathway to oil (taking into account some CO2 release from respiration and the OPPP) and is in keeping with our prediction that the increased efficiency conferred by the new pathway requires light and photosystems to provide cofactors.
For green seeds, the presence of glycolysis, the OPPP, Rubisco and photosystems provides alternative metabolic routes that allow adaptations to different environments and thereby maximize the use of both the carbon provided by the mother plant and the light available to the embryo. The survival advantage to plants of the more efficient metabolic flux to carbon storage that we describe might explain why seeds of many species are green and contain substantial Rubisco activity during development despite the absence of sufficient light for the operation of the Calvin cycle.
Isotopically labelled substrates
[U-14C]glucose, [U-14C]sucrose, [U-14C]glutamine and [U-14C]alanine were purchased from Amersham Biosciences. NaH14CO3, NaH13CO3, [1-13C]alanine and [U-13C3]alanine were purchased from Sigma-Aldrich.
Embryo culture system
Embryos of Brassica napus L (cv. Reston) collected early in the oil accumulation stage were cultured for 3 days in 5 ml liquid medium as described previously6,9. Carbon sources were sucrose (80 mM), glucose (40 mM), glutamine (35 mM) and alanine (10 mM) at concentrations mimicking the endosperm liquid in which embryos develop in planta.
Measurement of the CO2 balance for growing embryos
All organic carbon sources were uniformly labelled with 14C by adding [U-14C]sucrose, [U-14C]glucose, [U-14C]glutamine and [U-14C]alanine to the medium, resulting in a specific radioactivity of 1.42 mCi per mol of carbon. Culture flasks were sealed with sleeve stoppers containing two inlets. A 2% 14CO2 atmosphere was created inside the flasks by placing a glass vial into the flask containing 18.6 mg NaH14CO3, and injecting HCl directly into the vial to release the 14CO2.
Determination of 14CO2 efflux
After culture, the flasks were flushed for 2 h with nitrogen at 45 ml min-1 with exhaust gas bubbled through a 250-ml wash bottle containing 140 ml of 1 M KOH. The efficiency of the CO2 trap for 14CO2 recovery was on average 99.6% (s.d. = 3.3%). An aliquot (2.5 ml) of the trapping solution was counted by liquid scintillation.
Separation of 14C-labelled compounds
To extract lipids, 14C-labelled embryos were homogenized with a glass microgrinder at 4 °C in 1 ml iso-octane:isopropanol (2:1, v/v). The samples were centrifuged for 5 min at 5,000g and the supernatants were pipetted into glass test tubes (lipid fraction). Lipid extraction was repeated three times. To recover proteins, the pellet was extracted twice with 1 ml of 0.01 M sodium phosphate saline buffer (pH 7.4) containing 1 mM EDTA, 10 mM 2-mercaptoethanol, 0.02% sodium azide and 0.0125% (w/v) SDS. After extraction of protein, a fibre/polysaccharide fraction was recovered by hydrolysis of the pellet in 0.5 ml of 67% aqueous sulphuric acid. 14C in each fraction was determined by liquid scintillation counting.
For labelling with stable isotope, embryos were grown for 3 days in a liquid medium as described above in a 2% CO2 (v/v) atmosphere. CO2 or alanine were replaced with 13CO2, [1-13C]alanine or [U-13C3]alanine, respectively. After growth, lipids and proteins were extracted, and fatty-acid methyl ester and amino acids were obtained as described previously9. Amino acids were analysed as N,O-t-butyldimethylsilyl derivatives and fatty acids as methyl esters by gas chromatograph/mass spectrometer as described earlier9. Mass spectra of Phe, Tyr, Val and oleic acid showed significant enrichment of 13C only in the m1 peaks of each of the fragments observed. The apparent absence of molecules 13C-labelled at multiple positions (m2, m3, and so on) allowed the calculation of the fractional 13C enrichments in C1 of Phe, Tyr and Val as follows: FPhe(1) = m1,Phe(1–9) - m1,Phe(2–9); FTyr(1) = m1,Tyr(1–9) - m1,Tyr(2–9); FVal(1) = m1,Val(1–5) - m1,Val(2–5); FPGA(2–3) = FVal(2–5) = m1,Val(2–5); FPGA(2–5) = Foleic acid(1–18) = m1,oleic acid(1–18) (fractional enrichment is shown as Fmetabolite(carbon numbers), and mass peak mX corrected for natural isotope abundance as mX,metabolite(carbon numbers)). Only in the case of feeding [U-13C3]Ala did the mass spectra of Val and oleic acid reveal molecules to be labelled with 13C at multiple positions (significant abundance of m2, m3, and so son). In this case FVal(1) was calculated by subtracting the average 13C enrichment for fragments Val1–5 and Val2–5 from each other.
Levin, D. A. The oil content of seeds: An ecological perspective. Am. Nat. 108, 193–206 (1974)
Bassham, J. A. et al. The path of carbon in photosynthesis. XXI. The cyclic regeneration of carbon dioxide acceptor. J. Am. Chem. Soc. 67, 1760–1770 (1954)
Schwender, J., Ohlrogge, J. & Shachar-Hill, Y. Understanding flux in plant metabolic networks. Curr. Opin. Plant Biol. 7, 309–317 (2004)
Neuhaus, H. E. & Emes, M. J. Non-photosynthetic metabolism in plastids. Annu. Rev. Plant Physiol. Plant Mol. Biol. 51, 111–140 (2000)
King, S. P., Badger, M. R. & Furbank, R. T. CO2 refixation characteristics of developing canola seeds and silique wall. Aust. J. Plant Phys. 25, 377–386 (1998)
Schwender, J., Ohlrogge, J. B. & Shachar-Hill, Y. A flux model of glycolysis and the oxidative pentose phosphate pathway in developing Brassica napus embryos. J. Biol. Chem. 278, 29442–29453 (2003)
Ruuska, S. A., Schwender, J. & Ohlrogge, J. B. The capacity of green oilseeds to utilize photosynthesis to drive biosynthetic processes. Plant Physiol. 136, 2700–2709 (2004)
Goffman, F. D., Ruckle, M., Ohlrogge, J. B. & Shachar-Hill, Y. Carbon dioxide concentrations are very high in developing oil seeds. Plant Physiol. Biochem. 42, 703–708 (2004)
Schwender, J. & Ohlrogge, J. Probing in vivo metabolism by stable isotope labeling of storage lipids and proteins in developing Brassica napus embryos. Plant Physiol. 130, 347–361 (2002)
Schuster, S., Fell, D. & Dandekar, T. A general definition of metabolic pathways useful for systematic organization and analysis of complex metabolic networks. Nature Biotechnol. 18, 326–332 (2000)
Schuster, S., Dandekar, T. & Fell, D. A. Detection of elementary flux modes in biochemical networks: a promising tool for pathway analysis and metabolic engineering. Trends Biotechnol. 17, 53–60 (1999)
This work was supported by grants from the Department of Energy, the National Science Foundation and the USDA. Acknowledgement is also made to the Michigan Agricultural Experiment Station for its support of this research.
The authors declare that they have no competing financial interests.
Contains an analysis of how much carbon dioxide can be assimilated and stored in biomass by carboxylation reactions that produce oxaloacetate. Alternative metabolic routes for converting hexose into acetyl-CoA are described and analysed as elementary flux modes. The labelling experiments using 13CO2 or 13C-labelled alanine are discussed with additional detail. (DOC 354 kb)
About this article
Cite this article
Schwender, J., Goffman, F., Ohlrogge, J. et al. Rubisco without the Calvin cycle improves the carbon efficiency of developing green seeds. Nature 432, 779–782 (2004). https://doi.org/10.1038/nature03145
Thermomagnesium as a fertilizer for soybean: carbohydrate metabolism, silicon–magnesium fertilizer, and grain yield
Journal of Plant Nutrition (2021)
Plant Physiology (2021)
Integrative Modeling of Gene Expression and Metabolic Networks of Arabidopsis Embryos for Identification of Seed Oil Causal Genes
Frontiers in Plant Science (2021)
New insight into the rapid growth of the Mikania micrantha stem based on DIA proteomic and RNA-Seq analysis
Journal of Proteomics (2021)
Microbial Biotechnology (2021)