The nitrogen cost of photosynthesis (2024)

,

Article history

Abstract

Introduction

Just over a century has passed since the discovery of the Haber–Bosch method to reduce atmospheric dinitrogen and produce ammonia, which paved the way for large-scale production of nitrogenous fertilizer. There is a close correlation between the production of nitrogenous fertilizer and the production of the three key cereals that dominate the human diet (wheat, rice and maize; http://www.fao.org/faostat). Crop production reflects photosynthesis integrated over the life of the crop. The process of photosynthesis requires a system that is comprised of many proteins, and which accounts for the majority of nitrogen in any plant. It is this large nitrogen requirement to construct a photosynthetic system that results in the need for nitrogenous fertilizer by highly productive crops.

The photosynthetic rate and other leaf attributes have been measured for an extensive number of species. By combining two attributes, nitrogen content and the leaf dry mass, both expressed per unit leaf area, it is possible to predict the photosynthetic capacity. This has proved a useful way of parameterizing photosynthesis over the large areas of natural ecosystem that is necessary for global models (Rogers et al., 2017a). There are differences between species in the relationship between photosynthesis and leaf nitrogen content (Kattge et al., 2011). These reflect underlying differences in the allocation of nitrogen between proteins and the properties of those proteins, or they are a consequence of anatomical differences. Nitrogen and photosynthesis are central to each of these interrelated topics (Box 1), which are considered in this review.

Leaf nitrogen budget

It is timely to revisit the nitrogen budget of a leaf. Firstly, X-ray crystallography of protein complexes reveals atomic resolution, providing accurate pigment-to-protein stoichiometries. Secondly, a vast number of proteins and their relative abundance can now be determined using mass spectrometry (MS).

Dividing nitrogen between different pools can take several directions. At a cellular level, soluble and membrane fractions can be separated from a cell-wall pool. Alternatively, nitrogen can be partitioned between different organelles. These two approaches rely on different methodologies and generally no approach accounts for all of the nitrogen. Consequently, melding together these disparate pieces of information requires adjustments to reach an average total. This average may not apply to a particular leaf due to effects of age, environment, and species, but it provides a useful common starting point for C₃ species.

Using MS, thousands of proteins and their relative abundance in a range of organisms have been measured. The PaxDb resource (Wang et al., 2015) provides estimates of protein abundance derived from spectral counts across many experiments and tissue types. The Arabidopsis thaliana database comprises 46 datasets, covering 76% of the expected proteome. More than 90% of protein is accounted for by the 1000 most-abundant proteins. However, protein quantification by MS has an inherent bias, over-representing more-abundant proteins when low-abundance proteins fall below the instrument detection limits. Identification of proteins by MS can also be biased due to a range of factors affecting peptide detection, such as peptide solubility, enzymatic digestion efficacy, and differing ion efficiencies (reviewed in Lundgren et al., 2010).

Consequently, the PaxDb values cannot be taken at face value (Li et al., 2017). For example, the abundance of Rubisco large subunits outnumbers that of the small subunit by more than 8-fold. One would expect that the amounts of these two subunits should be similar as the mature Rubisco enzyme contains eight large and eight small subunits. Rubisco represents about 40% of soluble protein (Eckardt et al., 1997), or 20% of leaf nitrogen (Evans and Seemann, 1984), which equates to about 119000 ppm for each subunit (see Supplementary Data S1 at JXB online). Because Rubisco is such an abundant protein, this potentially introduces a significant bias unless it is corrected (Li et al., 2017). Further, the stoichiometry in PaxDb of proteins within and between complexes does not necessarily match expectations, perhaps reflecting the fact that not all proteins are quantitatively captured during tissue preparation and subsequent measurement. However, the data available from MS allows a deeper understanding of nitrogen distribution between proteins than previous techniques have afforded. Moving forward, new data-independent acquisition proteomic techniques, such as SWATH MS (Law and Lim, 2013) will allow greater accuracy and a much finer resolution of leaf nitrogen allocation between proteins within leaves.

Thylakoid nitrogen costs

Within the chloroplast, protein complexes in the thylakoid membranes are involved with light-capture, photosynthetic electron transport from water to NADP, and ATP synthesis. The relative abundance of these protein complexes varies in response to growth irradiance, which also changes the electron transport capacity per unit of chlorophyll. It is convenient to divide thylakoid nitrogen between two pools: light-capture and bioenergetics. The reaction centres of both photosystem II and I capture light and perform electron transport, but under unstressed conditions neither determine the electron transport capacity. Consequently, it is appropriate to place them in the pool associated with light-capture, together with the light-harvesting chlorophyll a/b complexes (LHCs). The distribution of chlorophyll between these complexes can be used to estimate the nitrogen associated with each, if the chlorophyll-to-protein stoichiometry is known (Table 1). The majority of chlorophyll is associated with the LHCs (56%), each of which binds 14 chlorophyll molecules (Liu et al., 2004). Photosystem I with its four associated LHCs accounts for 30% of leaf chlorophyll in complexes that bind 156 chlorophyll molecules (Scheller et al., 2001; Caspy and Nelson, 2018). Photosystem II with the antenna proteins CP26 and CP29 bind 63 chlorophyll molecules (Wei et al., 2016) and account for the remaining 14% of chlorophyll. Putting these three fractions together results in an average nitrogen cost for light-capture of 37.3 mol N mol⁻¹ Chl (Table 1).

The second thylakoid nitrogen pool, bioenergetics, is associated with photosynthetic electron transport and ATP synthesis. The relative abundances of the cytochrome b6f and ATP synthase complexes covary depending on the growth irradiance, and are directly correlated with the electron transport capacity (Evans, 1987; Yamori et al., 2011). Consequently, the cytochrome f content provides a way to link photosynthetic performance to the nitrogen cost of these complexes. As quantitative measures of ATP synthase were lacking when the thylakoid nitrogen budget was first assembled, a ratio of cyt f:ferredoxin-NADP⁺ reductase (FNR):ATP synthase of 1:1:1.2 was assumed, which resulted in a nitrogen cost for bioenergetics of 8.85 mol N mmol^–1 cyt f (Evans and Seemann, 1989). Now, with the availability of PaxDb (Wang et al., 2015; Li et al., 2017), we have reassessed this assumption (Supplementary Data S2) and obtained a ratio of cyt f:FNR:ATP synthase of 1:0.85:1.35, which leads to a revised cost for bioenergetics of 10.86 mol N mmol⁻¹ cyt f. The actual ratio assumed for ATP synthase makes a significant impact on the total nitrogen cost of bioenergetics as it represents about 80% of this pool.

The nitrogen cost of thylakoids with respect to their electron transport capacity can be represented graphically. In Box 2, cytochrome f content per unit chlorophyll, which is directly proportional to the electron transport capacity per unit chlorophyll, varies along the x-axis; the total thylakoid nitrogen cost per unit chlorophyll is presented on the y-axis. The symbols represent actual measurements taken from spinach and pea leaves that were grown under different irradiances, as well as several C₄ species where mesophyll and bundle sheath cells were analysed separately (Evans, 1987; Terashima and Evans, 1988; Evans and Seemann, 1989; Ghannoum et al., 2005). The rectangular area labelled ‘light capture’ represents the average nitrogen cost associated with LHCs and the two photosystem complexes (37.3 mol N mol⁻¹ Chl). For simplicity, the minor variation in chlorophyll distribution between pigment protein complexes has been ignored here (Leong and Anderson, 1984). The triangular area labelled ‘bioenergetics’ represents the increasing cost of nitrogen as the electron transport capacity increases per unit chlorophyll. Two upper bounds are shown depending on the nitrogen cost assumed for bioenergetics (8.85 and 10.86 mol N mmol⁻¹ cyt f being the original and revised estimates, respectively). On average for a leaf growing in sunlight, there are about 55 mol N mol⁻¹ Chl in chloroplast thylakoid membranes.

Nitrogen distribution within the cell

To establish the relative distribution of nitrogen between the cellular organelles, it is necessary to juggle different sources of information as none provide the complete picture. An average distribution for mature leaves of C₃ plants is chloroplast 75%, mitochondria 5%, peroxisomes 2.5%, cytosol 7.5%, and cell walls 10% (Makino and Osmond, 1991; Wang et al., 2015; Li et al., 2017; Onoda et al., 2017). Alternatively, the nitrogen distribution can be grouped by function and this can be superimposed onto the organellar structure, as shown in Box 3. The relative size of each pool related to photosynthesis has been scaled to represent the fraction of leaf nitrogen associated with it, in total accounting for 54% of the total. In the case of the photorespiratory cycle, this occurs across three organelles. Within chloroplasts, about 16% of the nitrogen is associated with other proteins and molecules not directly associated with photosynthesis and protein synthesis. For the remainder of the cell, another 13% is left in the ‘other’ category, which includes the nucleus, cytosol, and non-photorespiratory mitochondrial processes.

Scaling to the ecosystem

Given the diversity of plant species and ecosystems, it is a challenge to represent them through generalizations. Leaf dry mass and nitrogen contents per unit area have been determined for samples collected in the field for many species, and for those leaves that also had photosynthetic attributes measured in the field, relationships have emerged. Linear relationships between photosynthetic capacity and leaf nitrogen content per unit area exist for different plant types (Kattge et al., 2009) although, perhaps surprisingly, nitrogen-fixing legumes overlap with non-leguminous dicotyledonous crop species (Adams et al., 2018). Since there are many more measurements of leaf nitrogen than photosynthesis on field-grown material, these relationships are widely embedded into ecosystem and global models. However, given the variability in the slope relating photosynthetic capacity to leaf nitrogen content per unit area between plant types, ground-truthing is still required, for example in arctic biomes (Rogers et al., 2017b). Field gas-exchange measurements can establish the relationship between Rubisco capacity and leaf nitrogen content, although this may not reflect the actual allocation of nitrogen in Rubisco (Bahar et al., 2017). Improvements in remote sensing capability are increasing our ability to estimate plant characteristics from reflectance spectra (Martin et al., 2018). Whether it is possible to use hyperspectral reflectance to derive estimates of Rubisco capacity directly (Serbin et al., 2015; Yendrek et al., 2017; Silva-Perez et al., 2018) or indirectly by first predicting nitrogen content (Dechant et al., 2017) is currently an active area of research.

An analysis of multiple publications has revealed four features associated with increasing leaf mass per unit area between species (Onoda et al., 2017). Firstly, there is an apparent decrease in nitrogen allocated to Rubisco. Secondly, there is a decrease in mesophyll conductance per unit of mesophyll cell surface exposed to intercellular airspace. Thirdly, the draw-down in CO₂ partial pressure between intercellular airspaces and the sites of carboxylation inside chloroplasts during photosynthesis increases with increasing leaf mass per unit area (LMA). Fourthly, there is an increase in the fraction of leaf nitrogen associated with the cell wall. The combination of these features reduces photosynthetic capacity per unit of leaf N in species with greater LMA. Given that LMA is associated with leaf lifespan, rather than achieving an instantaneous high photosynthetic rate per unit leaf nitrogen, species with high LMA may instead achieve greater lifetime photosynthetic return from a given investment of nitrogen into a leaf.

Fertilizer—photosynthesis—food

In the forty years 1962–2002, the combined global production of wheat, rice, and maize increased from 682 to 1752 Mt year⁻¹ and nitrogen fertilizer production increased from 13.6 to 88.2 Mt year⁻¹ (http://www.fao.org/faostat/en/#data). There was a close linear relationship between the two, with 13.8 tonnes of grain produced per tonne of nitrogen fertilizer. Assuming an average grain nitrogen content for wheat, rice, and maize of 1.9% (Uribelarrea et al., 2008; Jaksomsak et al., 2017; Rapp et al., 2018), harvested grain accounts for one quarter of global nitrogen fertilizer. This is remarkable given that the fertilizer is not only applied to these three crops, that the harvested grain represents only part of the nitrogen in the crop at maturity, that there are losses of nitrogen from leaching, erosion, and denitrification, and that there is some residual nitrogen left in the soil. However, the environmental costs associated with nitrogen escape are a growing cause for concern and there are pressing demands for improving the efficiency in the use of nitrogen applied in agriculture in order to reduce environmental damage, economic cost, and atmospheric greenhouse gas consequences both during the production of fertilizer and NO_x emissions from fields.

Plants need to balance carbon gain with the synthesis of organic nitrogen compounds. As a consequence of the oxygenation reaction catalysed by Rubisco, the photorespiratory pathway recycles two molecules of phosphoglycolate to produce one phosphoglycerate (PGA). At the same time, one molecule of ammonia is released in mitochondria and is refixed by Gln synthetase and glutamine oxoglutarate aminotransferase (GS/GOGAT). The widely used Farquhar, von Caemmerer, and Berry biochemical model of C₃ photosynthesis (Farquhar et al., 1980) assumes complete recycling, although this may not always be the case (Abadie et al., 2017; Bloom and Lancaster, 2018; Busch et al., 2018). At 25 °C and current atmospheric CO₂ concentrations, approximately six carbon atoms are fixed per ammonia recycled (see Supplementary Data S4). By comparison, new biomass requires 33 carbon to be fixed for each new nitrogen, assuming the plant contains 2% N, 40% C, and respires 30% of daily carbon fixed during the production of this new biomass. Incorporation of ammonia during photorespiration or de novo incorporation in leaves uses the same GS/GOGAT enzymatic pathway. Therefore, for plants converting inorganic nitrogen into organic compounds in their leaves, 85% of the GS/GOGAT flux is dealing with photorespiration on average. At any given instant, this proportion would change as it is affected by temperature, irradiance, and CO₂ concentration. One consequence of rising atmospheric CO₂ concentrations is that the C:N balance of plant tissue is changing. Elevated CO₂ reduces photorespiration and, with the exception of legumes that can fix atmospheric nitrogen symbiotically, plants grown under elevated atmospheric CO₂ have lower nitrogen concentrations (Feng et al., 2015). This translates into lower grain protein concentrations, which may have dietary implications in the future (Myers et al., 2014; Zhu et al., 2018).

Engineering photosynthesis to improve crop yield

Our detailed knowledge of photosynthesis has led to the identification of many proteins that can be targeted to increase carbon gain. A selection of targets that have been identified are presented in Box 4. In some cases, initial proof of concept has been obtained with transformed model plants (Kromdijk et al., 2016; Driever et al., 2017; Lopez-Calcagno et al., 2018; Salesse-Smith et al., 2018). Field trials with crop plants are underway and their outcome is eagerly awaited. Given the central importance of Rubisco in determining the rates of CO₂ assimilation and photorespiration, and because it accounts for so much of leaf nitrogen, much attention is focussed on ways to improve its performance. Approaches fall into two categories: firstly, those where the catalytic properties of Rubisco are altered, for example from C₄ species or other organisms (Orr et al., 2016), and secondly, those where the CO₂ partial pressure around Rubisco is increased, for example CO₂-concentrating mechanisms such as carboxysomes (Hanson et al., 2016; Rae et al., 2017; Long et al., 2018), greater mesophyll conductance (Groszmann et al., 2017), or photorespiratory bypass (Peterhansel and Maurino, 2011). While some variation in kinetic properties of Rubisco between wheat relatives has been identified (Prins et al., 2016), detailed crop modelling is needed to assess the impact and cost/benefit from engineering an alternative form into elite wheat. While there are several crop models available (Song et al., 2017; Yin and Struik, 2017; Wu et al., 2018), it is a complex task to deal with plant functions that are not necessarily well represented or fully parameterized. The perennial debate about whether plant growth and yield are determined by source photosynthesis or sink demand continues. In the case of rice, increasing sink capacity has led to a dramatic increase in yield (Ashikari et al., 2005). The current focus on improving photosynthesis is because the gains in harvest index (grain yield/above-ground biomass) associated with the introduction of dwarfing genes have been largely maximized, but maintaining or increasing both sink strength and harvest index is also crucial.

If a plant could be engineered to fix more carbon per unit of nitrogen associated with photosynthesis, then unless de novo incorporation of nitrogen was also enhanced, there would be a lowering of the nitrogen concentration of the plant and most likely also the protein content of the grain. An increase in carbon gain per unit photosynthetic N could free up nitrogen for investment in new tissues elsewhere and increase growth. This is observed when plants are grown under elevated atmospheric CO₂ (Ainsworth and Long, 2005). However, unless additional organic nitrogen is incorporated into other tissues, the conversion of that increased growth into greater yield would result in lower grain protein. If the additional organic nitrogen incorporated elsewhere in the plant could not provide any improvement above that gained from greater photosynthesis per unit of photosynthetic N, what is the benefit from raising photosynthetic rate per unit N?

A second concern is that for cereal crops, nitrogen is remobilized from leaves and stems during grain filling. At maturity, the grain can account for 80–90% of above-ground nitrogen (Barraclough et al., 2010; Gaju et al., 2014). For a crop yielding 10 t ha^–1 with a 2.5% N concentration in the grain, this represents 250 kg N ha⁻¹. To contain this within a crop canopy with a leaf area index of 7 (Shearman et al., 2005), the leaf nitrogen content would need to be 3.6 g m⁻². This is close to the maximum leaf nitrogen content that is observed (Silva-Perez et al., 2018). If increasing photosynthesis per unit N resulted in lower nitrogen contents per unit leaf area, then a greater fraction of this remobilizable nitrogen would need to be present in the sheath and stem fractions. In the case of wheat, the ear can also make a substantial photosynthetic contribution to the grain (Maydup et al., 2012; Zhou et al., 2016). While these tissues can contribute to canopy photosynthesis, the relative efficiency of leaf and stem needs to be investigated in order to assess the consequences. The point is, that to increase yield while maintaining grain protein concentration requires increasing both photosynthetic carbon gain and de novo nitrogen incorporation. In addition, the crop canopy has to be capable of holding the vast majority of that nitrogen in its leaves to enable its relocation into developing grain. An alternative is to continue de novo nitrogen incorporation during grain filling, which requires continued root growth, nitrogen uptake (perhaps associated with a late application of fertilizer), and incorporation into protein while leaves are senescing.

Future work

Given that Rubisco constitutes the largest fraction of nitrogen in leaves of C₃ plants, it justifiably attracts great attention. In the absence of complete kinetic information to describe the performance of Rubisco from different species, the default has frequently been to assume kinetic values of tobacco Rubisco (Bernacchi et al., 2002). However, the kinetic properties of Rubisco from diverse species need to be determined. Some of the variation between species in the apparent Rubisco activity per unit leaf N might be associated with variation in kinetic properties, but other factors could also be involved, such as different allocation of nitrogen towards Rubisco and different activation states. With improved quantification of relative protein abundance, the extent to which variation in nitrogen allocation to pigment protein complexes is associated with Rubisco performance will be revealed. The limited number of species for which thylakoid nitrogen cost has been quantified should be expanded. In particular, the nitrogen allocated to ATP synthase needs attention, given its apparent significant cost.

Supplementary data

Supplementary data are available at JXB online.

Data S1. Rescaling PaxDb to account for Rubisco abundance.

Data S2. Nitrogen cost of bioenergetics.

Data S3. Nitrogen distribution within the cell.

Data S4. Nitrogen fixed per carbon assimilated.

Acknowledgements

This work was supported by the Australian Research Council Centre of Excellence for Translational Photosynthesis (CE140100015) and the Grains Research Development Corporation (ANU00025). Thanks to Harvey Millar and Nic Taylor from UWA for proteomics information, and Christine Raines for encouragement.

References

© The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. All rights reserved. For permissions, please email: journals.permissions@oup.com

Issue Section:

Expert View

Comments

0 Comments

I agree to the terms and conditions. You must accept the terms and conditions.

Email alerts

Citing articles via

More from Oxford Academic