Distinctive mitochondrial and chloroplast components contributing to the maintenance of carbon balance during plant growth at elevated CO₂

ABSTRACT

Plant carbon balance depends upon the difference between photosynthetic carbon gain and respiratory carbon loss. In C₃ plants, growth at an elevated atmospheric concentration of CO₂ (ECO₂) stimulates photosynthesis and raises the leaf carbohydrate status, but how respiration responds is less understood. In this study, growth of Nicotiana tabacum at ECO₂ increased the protein amount of the non-energy conserving mitochondrial alternative oxidase (AOX). Growth at ECO₂ increased AOX1a transcript amount, and the transcript amount of a putative sugar-responsive gene encoding a chloroplast glucose-6-phosphate/phosphate translocator (GPT3). We suggest that the elevated amounts of AOX and GPT3 represent distinctive mitochondrial and chloroplast mechanisms to manage an excessive cytosolic pool of sugar phosphates. AOX respiration could consume cytosolic sugar phosphates, without this activity being restricted by rates of ATP turnover. GPT3 could allow accumulating cytosolic glucose-6-phosphate to return to the chloroplast. This could feed starch synthesis or a glucose-6-phosphate shunt in the Calvin cycle. AOX and GPT3 activities could buffer against P_i depletions that might otherwise disrupt mitochondrial and chloroplast electron transport chain activities. AOX and GPT3 activities could also buffer against a down-regulation of photosynthetic capacity by preventing a persistent imbalance between photosynthetic carbon gain and the activity of carbon utilizing sinks.

KEYWORDS:

• Photosynthesis
• respiration
• elevated carbon dioxide
• carbon balance
• alternative oxidase
• glucose-6-phosphate/phosphate translocator
• inorganic phosphate
• source:sink balance

Introduction

Photosynthesis and respiration are two important determinants of plant growth.^1,2 In simplest terms, this is because the amount of carbohydrate available to support growth and maintenance depends upon the difference between photosynthetic carbon gains and respiratory carbon losses. Respiratory carbon losses are significant, usually estimated at about half the photosynthetic carbon gain.³ On the other hand, some aspects of respiratory metabolism may be supportive of carbon gain by benefiting photosynthesis.^4,5 The beneficial effects of respiration on photosynthetic carbon gain are not fully understood, so this remains an area of active research.^6,7

Sources (primarily mature photosynthetic leaves) are net producers and exporters of carbohydrate while sinks (primarily non-photosynthetic tissues) are net importers and consumers of carbohydrate.^8,9 In the light, the immediate carbon gains of photosynthesis are used in the chloroplast for transient starch synthesis, or are transported to the cytosol for either sucrose synthesis or for entry into respiratory and other pathways that support the ongoing maintenance of the source tissue.^8,9 Once synthesized, sucrose is loaded into the phloem and transported to sink tissues, where respiratory and other pathways utilize it for growth, maintenance and storage.^8,9 If leaf sucrose production outpaces sucrose use by sinks, then phloem transport can slow, causing leaf sucrose concentration to rise. Poorly understood feedbacks can then slow the rate of sucrose synthesis, potentially resulting in a buildup of its cytosolic sugar phosphate precursors.^10-13

Plant growth and development require a balance between source and sink activities.^14,15 In part, this balance is accomplished by carbohydrates acting not only as substrates but also as signaling molecules able to influence source and sink activity.^8,13 For example, a persistent buildup of leaf carbohydrate can down-regulate the expression of genes encoding photosynthetic components, reducing photosynthetic capacity.¹⁶

Rates of photosynthetic carbon gain depend upon environmental parameters since photosynthesis involves both a light-dependent electron transport process and a CO₂-dependent carbon assimilation pathway.¹⁷ The chloroplast electron transport chain (cETC) in the thylakoid membrane generates NADPH and ATP for use by the stroma-localized Calvin-Benson (CB) cycle to produce triose phosphates (TP). The CB cycle includes the enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco). While the carboxylase activity of Rubisco supports net TP production, its oxygenase activity supports photorespiration, a pathway that consumes energy but results in a net carbon loss.¹⁸ Stomatal pores on the surface of leaves enable the effective delivery of CO₂ to Rubisco. When water is limiting in the environment, these pores can close to conserve water, disrupting CO₂ supply. Hence, changes in the availability of light, CO₂ and water (amongst other environmental factors) each perturb parts of the photosynthetic process. If left unchecked, such perturbations can result in energy and/or carbon imbalances that can then compromise photosynthesis.^16,19 How plants are able to (mostly) prevent such imbalances is an area of ongoing research.^19-21

Plants will experience an elevated atmospheric concentration of CO₂ (ECO₂) in the future. Ongoing research aims to establish how photosynthesis and respiration will respond to this growth condition.^22,23 One important consequence of transferring plants to ECO₂ is that it promotes the carboxylase activity and suppresses the oxygenase activity of Rubisco.¹⁸ In the short term, this can dramatically increase carbon gain. However, this increase tends to be less dramatic in the longer-term due to “photosynthetic acclimation,” where accumulating carbon products feedback on and down-regulate the capacity of the photosynthetic machinery.^16,22 Despite this photosynthetic acclimation, there is consensus that ECO₂ does nonetheless stimulate the long-term photosynthetic carbon gain of C₃ plant species.²² Conversely, there is much less certainty regarding the respiratory response to growth at ECO₂, and hence how respiration will affect overall carbon balance in a future ECO₂ world.^23-25

Plant respiratory pathways [glycolysis, oxidative pentose phosphate (OPP) pathway, tricarboxylic acid (TCA) cycle] process carbohydrates to provide the carbon skeletons, reducing power [NAD(P)H] and ATP required for growth and maintenance.²⁶ As mentioned earlier, the carbon losses associated with respiration are significant. Further, these carbon losses are potentially accentuated in plants by the presence of cytosolic and mitochondrial components that relax the otherwise tight coupling between respiratory carbon oxidation and mitochondrial ATP synthesis.²⁶ A prominent example of this is alternative oxidase (AOX), a component of the mitochondrial electron transport chain (mETC).^27-30 In the mETC, the transfer of electrons from ubiquinol to oxygen can occur using either the cytochrome (cyt) pathway [involving Complex III, cyt c and Complex IV (cyt oxidase)] or AOX. Electron flow through the cyt pathway generates a proton motive force used by Complex V (mitochondrial ATP synthase) to generate ATP. However, electron flow from ubiquinol to AOX does not contribute to this proton motive force, hence not supporting ATP generation.

What is the physiological relevance of AOX respiration? By relaxing the coupling between electron flow and ATP turnover, AOX can reduce the generation of reactive oxygen species by the mETC and can facilitate the provision of carbon skeletons for biosynthesis.^27-29 AOX can also enhance photosynthesis by providing a means to maintain chloroplast energy balance.³⁰ The production ratio of ATP to NADPH by chloroplast linear electron transport (LET) does not match its consumption ratio by the CB cycle. There is a shortfall of ATP relative to NADPH. This shortfall is exaggerated as the ratio of oxygenation to carboxylation by Rubisco increases (i.e. as photorespiration becomes more prevalent), such as during water deficit. Under such conditions, metabolite shuttles³¹ can move “excess” reducing power to the cytosol, for turnover using AOX respiration. This provides a means to match the production ratio of ATP to NADPH in the chloroplast with its consumption ratio, hence enhancing photosynthetic performance.^6,30

Another long-standing hypothesis is that AOX respiration could provide a means to consume “excess” carbohydrate.³² There has been relatively little investigation of this idea, although a few studies have reported higher carbohydrate amounts in AOX knockdown/knockout plants under some conditions^33-35 An important question becomes, when, why and where might it be advantageous for a plant to use AOX respiration to promote carbon loss?

In general terms, the respiration rate of plant tissues could depend upon either the availability of respiratory substrate (principally carbohydrates), the capacity of respiratory components (principally enzymes) to process that substrate, or the rate at which various respiratory products [carbon skeletons, NAD(P)H, ATP] are being consumed for growth and maintenance.³⁶ Of these possibilities, the rate of ATP turnover is often regarded as having a dominant control (so-called adenylate control) over respiration rates, providing a rigorous means to match energy generation with cellular energy needs. Nonetheless, the presence of components such as AOX provides a means to relax adenylate control,³⁷ meaning that other factors, such as substrate availability, could become more influential in controlling respiration rates. Some studies have observed a close positive relationship between leaf carbohydrate amounts and respiration rate.^38-41 However, whether increasing respiration rates in these cases are due to a preferential increase in AOX respiration remains mostly unknown.

Results and discussion

Growth at ECO₂ is accompanied by an increase in leaf carbohydrate amount.²² To examine the influence of growth CO₂ concentration on the mETC, immunoblots were used to quantify the leaf protein amounts of cyt oxidase subunit II (COXII) and AOX over time in Nicotiana tabacum (tobacco) plants grown since germination at ambient atmospheric concentrations of CO₂ (ACO₂, 400 ppm) or ECO₂ (1000 ppm). Overall, COXII protein amount appeared slightly higher in the ECO₂- than ACO₂-grown plants, but in both cases, the amount was highest at the earlier time points and then progressively declined over time (Figure 1b). AOX showed a similar pattern over time, but only in the ACO₂-grown plants (Figure 1a). A different pattern occurred in plants growing at ECO₂. Here, the AOX protein amount increased significantly over time. Hence, while AOX amount was similar between ACO₂- and ECO₂-grown plants at the first time point, at later time points, AOX amount was much higher in the ECO₂- than ACO₂-grown plants. These results illustrate a fundamental difference in how the mETC responded over time to growth at ACO₂ compared to ECO₂. The results suggest a gradual shift toward more leaf AOX respiration in plants grown at ECO₂, compared to plants grown at ACO₂.

Figure 1. The leaf protein amount of AOX (a) and COXII (b) in Nicotiana tabacum germinated and grown at ACO₂ (400 ppm) or ECO₂ (1000 ppm). In each growth condition, leaf 5 was sampled at six different days after potting. These sampling times were chosen so that leaf 5 was of comparable size in the two growth conditions at each sampling time. For example, leaf 5 of ACO₂-grown plants at Day 11 was comparable in size to leaf 5 of ECO₂-grown plants at Day 9. All protein amounts were normalized to the protein amount in the ACO₂-grown plant at Day 11, which was set to 1. Data are the mean ± SEM of three independent experiments (n = 3). For each independent experiment and treatment, protein was isolated from leaf tissue pooled from three replicate plants. Protein was analyzed by Western blot using antibodies AS04054 (AOX1/2) and AS04053A (COXII) (Agrisera, Vännäs, Sweden). Data were analyzed by one-way ANOVA, followed by a Tukey multiple comparison test, comparing all pairs within a growth condition (ACO₂ or ECO₂). Data bars not sharing a common letter are significantly different from one another (P < .05).

In some respects, the apparent shift toward more AOX respiration at ECO₂ relative to ACO₂ is surprising. The chloroplast should be less reliant upon the mETC as a means to consume “excess” chloroplast reductant at ECO₂ since less photorespiration means that LET will generate less of a shortfall of ATP to NADPH in the chloroplast.³⁰ Further, since ECO₂ stimulates TP production, there should be a high cytosolic ATP demand for sucrose synthesis and export. Hence, respiration might be expected to shift toward the cyt pathway to satisfy ATP demand.⁴

Could it be that the role of tobacco AOX respiration at ECO₂ is to consume “excess” leaf carbohydrate? To consider this possibility, an important first question becomes, is there actually a “physiological excess” of carbohydrate in the ECO₂-grown plants? To begin addressing this question, we examined a model “sugar-responsive” gene that encodes a chloroplast transporter whose expression in leaf appears to relate closely with carbohydrate status. As background, the chloroplast inner envelope membrane contains a family of plastid phosphate translocators that includes the triose phosphate/phosphate translocator (TPP), the phosphoenolpyruvate/phosphate translocator, the glucose-6-phosphate/phosphate translocator (GPT) and the xylulose-5-phosphate/phosphate translocator.^42-44 In almost all dicots, the GPT subfamily includes two distinct groups termed GPT1 and GPT2. The Arabidopsis thaliana genome contains one GPT1 and one GPT2 gene.⁴² GPT1 is expressed in heterotrophic tissues, where the plastid import of Glc-6-P, in exchange for P_i, supports starch synthesis and OPP pathway activity.⁴⁵ GPT1 is essential for gametophyte, embryo and seed development, so gpt1 mutants are lethal.^46,47 Alternatively, gpt2 mutants have no visible phenotype.⁴⁶

Recently, an extensive phylogenetic analysis of plastid phosphate translocators reported that, within the asterid clade of dicots, which includes the family Solanaceae, GPT2 is replaced by another GPT group that was designated GPT3.⁴⁴ N. tabacum is a member of the Solanaceae but was not included in this prior analysis. Hence, we examined the relationship between N. tabacum GPT protein sequences and those of other dicots, including the well-characterized Arabidopsis GPT1 and GPT2 proteins (Figure 2). This analysis showed that N. tabacum has a GPT1 transporter, like all other dicots. The analysis also showed the presence of a second transporter that, similar to other Solanaceae, grouped separately from GPT2. Based on these results and the prior analysis⁴⁴ we conclude that N. tabacum, like other asterids, contains both the GPT1 and GPT3 groups of GPT transporters (Figure 2). In the text below, we use the terms GPT2 and GPT3 interchangeably, depending upon the dicot species in question.

Figure 2. A phylogenetic tree of dicot GPT proteins, constructed using the maximum-likelihood method within MEGA-X software.⁴⁸ Protein sequences were aligned using the MUSCLE algorithm and the LG+F substitution model was used to determine branch lengths. The numbers shown on the branches are bootstrap values (performed 1000 repeats). Rooting of the tree used a monocot (Oryza sativa) GPT1 protein sequence.

Typically, there is a poor expression of GPT’s in photosynthetic tissue, correlating with an inability of chloroplast envelopes to transport Glu-6-P. However, studies indicate that GPT2 expression can be strongly induced in leaf tissue by signal(s) relating to carbohydrate status (see details and references^12,23,49-69 in Table 1). In Arabidopsis, treatment with exogenous sugars (e.g. sucrose, glucose) rapidly increases GPT2 (but not GPT1) transcript amount.^53,57,62,63 Mutants displaying high leaf soluble sugar amounts such as starch metabolism mutants or sugar transport mutants also show increased GPT2 expression,^52,53,59,60 as do plants with an increased sugar status owing to a shift to higher irradiance or growth at ECO₂.^12,23,50 Transcription factors involved in this sugar-induced gene expression are being elucidated, but the specific sugar metabolite(s) and/or other factors required for this signaling remains mostly unresolved.^12,63

Table 1. A summary of factors altering leaf GPT2 (Arabidopsis thaliana, Glycine max) or GPT3 (Nicotiana tabacum) gene expression. See text for further details.

Factor	Species	Change in Leaf GPT2 (or GPT3) Transcript Amount	Reference
Light	A. thaliana	Increased (up to 10-fold) within 5 h following a shift to higher irradiance. However, the increase did not occur in leaves pretreated with DCMU directly before the irradiance shift.	^Citation49
		Increased (34-fold) at 1 d following a shift from 100 to 400 µmol photons m^−2 s^−1. Most upregulated transcript on the gene microarray.	^Citation50
		Increased transiently (several 100-fold) within 2 h following a shift from 120 to 500 µmol photons m^−2 s^−1. However, the increase did not occur when light shift accompanied by a shift to lower CO2 (from 400 to 20 ppm).	^Citation12
CO2	A. thaliana	Increased in ECO2 (550 ppm, free-air CO2 enrichment experiment) compared to ambient CO2 in two of the ecotypes tested (Cvi-0 and WS). However, transcript goes down in the third ecotype tested (Col-0).	^Citation51
	G. max	Increased (132–146%) in ECO2 (550 ppm, free-air CO2 enrichment experiment) compared to ambient CO2. Second most upregulated transcript on the gene microarray.	^Citation23
	N. tabacum	Increased following short-term shift to or long-term growth at 1000 ppm CO2, compared to 400 ppm CO2.	Current study
Carbohydrate	A. thaliana	Increased (37-fold) in a pho3 mutant that accumulates sucrose and other carbohydrates in leaves due to a defective SUC2 gene, encoding a sucrose transporter. On the other hand, the transcript amount of GPT1 and other plastid phosphate translocators decreased.	^Citation52
		Increased (100’s-fold) at 6 h following the addition of 50 mM sucrose to dark-grown seedlings. Also responded, but less so, to glucose additions. Increased in a starchless pgm mutant, but only toward the end of the photoperiod, when sugars were higher than in the WT.	^Citation53
		Increased at 6 h following the addition of 3% glucose to sugar-depleted seedlings.	^Citation54
		Increased in response to glucose treatments, as well as during developmental leaf senescence, when glucose amounts increase.	^Citation55
		Increased following floating of mature leaves on sucrose for 24 h in darkness. Less increase in an impaired sucrose induction 1 mutant defective in sugar-responsive gene expression.	^Citation56
		Increased (8.8-fold) following floating of leaf sections on 100 mM sucrose for 16 h in dim light. Most upregulated transcript on the gene microarray.	^Citation57
		Increased within 30 min of addition of 15 mM sucrose to carbon-starved seedlings in liquid culture.	^Citation58
		Increased in the sweetie mutant that accumulates sucrose, glucose, fructose and starch due to a defect in carbohydrate utilization.	^Citation59
		Increased in several different starch biosynthesis mutants, correlating with increased amounts of soluble sugars in the mutants during the light period.	^Citation60
		Increased in seedlings grown in the presence of 50 mM glucose, fructose or sucrose.	^Citation61
		Increased (4-fold) in seedlings transferred to sucrose-containing medium for 3 h. Most upregulated gene in the RNA-seq experiment.	^Citation62
		Increased (20-fold within 1 h) in carbon-starved seedlings transferred to medium containing 15 mM glucose. Response dependent on WRKY18 and WRKY53 transcription factors. GPT1 and other plastid phosphate translocators did not respond to the glucose treatment.	^Citation63
Phosphate	A. thaliana	Increased in hydroponic-grown plants in response to P withdrawal.	^Citation64
		Increased (37-fold) in P-starved compared to P-sufficient plants. The increase significantly reversed within 3 h of P re-supply.	^Citation65
		Increased (3.8-fold) in P-starved compared to P-sufficient plants.	^Citation66
Abiotic stress	A. thaliana	Increased by combination of heat + drought (3.3 log2-fold) or drought alone (3.8 log2-fold). Both treatments resulted in accumulation of sucrose and other sugars.	^Citation67
Nitro-oxidative stress	N. tabacum	Increased within 3 h following an increase in nitric oxide (due to infiltration of leaf with nitric oxide donor) or H2O2 (due to transfer of catalase-deficient plant to high light).	^Citation68
Other	A. thaliana	Decreased (to 23% of WT amount) in a mutant with an inactivated chloroplast ATP synthase, resulting in a high accumulation of protons in the lumen.	^Citation69

We examined the effects of growth CO₂ concentration on N. tabacum GPT3 transcript amounts, in comparison with GPT1 and the major AOX gene family member, AOX1a. Both GPT3 and AOX1a transcript amounts were higher in plants germinated and grown long term at ECO₂ compared to ACO₂ (Figure 3). This long-term difference in transcript amount between the two growth conditions was also evident in short-term (24 h) transfer experiments, and regardless of the direction of transfer. That is, ACO₂-grown plants transferred to ECO₂ displayed an increase in GPT3 and AOX1a transcript 24 h after transfer, while ECO₂-grown plants transferred to ACO₂ displayed a decrease in GPT3 and AOX1a transcript 24 h after transfer (Figure 3). On the other hand, GPT1 transcript showed little change in abundance following either of these short-term transfers, and a slight decrease in abundance in plants grown long term at ECO₂ compared to ACO₂ (Figure 3). Overall, these results show that the two GPT isoforms responded differently to growth CO₂ concentration, and that GPT3 and AOX1a transcript amounts responded similarly to growth CO₂ concentration, both increasing at ECO₂.

Figure 3. The Nicotiana tabacum leaf transcript amount for genes encoding AOX (gray bars), GPT1 (black bars) and GPT3 (open bars). Plants were germinated at ACO₂ (400 ppm) or ECO₂ (1000 ppm) for two weeks and then potted and grown for an additional 15 days under their original growth condition, prior to either transcript analysis or transfer to an alternate growth condition. The left-most set of data (ACO₂ to ECO₂) shows the transcript amount 24 h after transfer of ACO₂-grown plants to ECO₂, compared to the amount measured in control ACO₂-grown plants. The center set of data (ECO₂) shows the transcript amount in ECO₂-grown plants, compared to the amount measured in control ACO₂-grown plants. The right-most set of data (ECO₂ to ACO₂) shows the transcript amount 24 h after transfer of ECO₂-grown plants to ACO₂, compared to the amount measured in control ECO₂-grown plants. All transcript amounts were normalized to that of a reference gene (EF1) and are relative to the control amount (log2 fold-change). Data are the mean ± SEM of three independent experiments (n = 3). For each independent experiment and treatment, RNA was isolated from leaf tissue (leaf 5) pooled from three replicate plants. RNA was analyzed by qPCR, using the primers listed in Table S1.

As discussed earlier, tobacco AOX respiration acts to optimize photosynthesis during periods of water deficit by promoting chloroplast energy (ATP/NADPH) balance. Therefore, we examined the response of GPT3 to an increasing severity of water deficit. Well-watered tobacco plants maintained a leaf relative water content (RWC) of approximately 90% (Figure 4a). This RWC was maintained until approximately Day 2 after withholding water, after which time it gradually declined to approximately 31% by Day 8 (Figure 4a). The gradual decline in RWC following Day 2 was accompanied by a gradual increase in AOX protein, to approximately 10-fold its initial amount by Day 8 (Figure 4c). On the other hand, GPT3 transcript amount did not increase in response to water deficit. In fact, on most days over the water stress period, its amount was slightly lower than in the well-watered plant (Figure 4b).

Figure 4. The relative water content (a), GPT3 transcript amount (b) and AOX protein amount (c) of Nicotiana tabacum (leaf 5) in response to water deficit. Plants were grown at ACO₂ and were watered daily for 21 days. Plants were then given a final watering (Day 0 on the graph), after which time water was withheld. The GPT3 transcript amounts were normalized to that of a reference gene (GPT1) and are relative to the amount in the well-watered (Day 0) plants (log2 fold-change). GPT1 was used as the reference gene (rather than EF1, as used in Figure 3) since GPT1 amounts remained stable in response to water deficit, while EF1 amounts did not. The AOX protein amounts were normalized to the protein amount in the well-watered (Day 0) plants, which was set to 1. Data are the mean ± SEM of three independent experiments (n = 3). For each independent experiment and treatment, leaf tissue pooled from three replicate plants was used for RNA and protein isolation, as well as determination of relative water content. Protein and RNA were analyzed by Western blot and qPCR, respectively, as described in Figures 1 and 3.

During water deficit, AOX was necessary to maintain chloroplast energy balance,⁶ but not to maintain leaf carbon balance since starch, sucrose, hexose, and hexose phosphate pools were similar between WT, AOX knockdown, and AOX overexpression plants during water deficit.^35,70,71 The current results suggest that GPT3 has no role during water deficit since its expression is not responsive to this growth condition (Figure 4). On the other hand, both AOX and GPT3 appear to have a role during growth at ECO₂, as both displayed enhanced expression (Figures 1 and 3). Interestingly, we previously showed that leaf carbohydrates (particularly Glu-6-P, sucrose, and starch) were elevated in AOX knockdowns, relative to WT, during growth at ECO₂, hinting at a role for AOX in maintaining carbon balance.³⁵ The results with GPT3, a gene putatively responsive to leaf carbohydrate status, are consistent with this interpretation.

Growth at ECO₂ stimulates the availability of TPs for starch and sucrose synthesis, and hence is expected to challenge the plants' ability to maintain source:sink balance. Under such conditions, sucrose export could slow, resulting in sucrose accumulation and then a slowing of sucrose synthesis.¹³ Hence, our current hypothesis is that AOX and GPT3 represent mitochondrial and chloroplast mechanisms, respectively, to adjust leaf carbohydrate metabolism when there is too high source relative to sink activity. Below we discuss how AOX and GPT3 might be functioning is this regard.

When sugar phosphates such as Glc-6-P accumulate in the cytosol, rather than being used for sucrose synthesis and export, cytosolic P_i will decline. This can then slow the exchange of cytosolic P_i for stromal TPs by the TPT. As 3-phosphoglycerate (3-PGA) and TPs then accumulate in the stroma, stromal P_i will also decline, limiting the chloroplast ATP synthase activity supporting photosynthesis.^17,21,72,73 A high stromal 3-PGA/P_i ratio should then enhance the rate of starch synthesis by further activating ADP-glucose pyrophosphorylase.⁷⁴ However, the rate of starch synthesis can also become limited by the activity of chloroplast phosphoglucoisomerase, which is inhibited by 3-PGA, and which supplies the Glc-6-P necessary for starch synthesis.⁷⁵ In such conditions, induction of GPT3 would allow accumulating cytosolic Glc-6-P to reenter the stroma and fuel starch synthesis. The additional starch synthesis would then free up P_i from the carbon pool, without this process being limited by phosphoglucoisomerase activity.⁷⁶

Another possibility is that Glc-6-P returning to the stroma via GPT3 could enter the oxidative branch of the chloroplast OPP pathway. In the light, this Glc-6-P shunt is hypothesized to feed Glc-6-P carbon back into the CB cycle as ribulose-5-phosphate.^76–78 The shunt is a futile cycle, releasing CO₂ in the light, but might be advantageous at ECO₂. First, it would reduce the rate of net carbon gain under conditions when carbon excess is becoming problematic. Second, it would consume chloroplast ATP under conditions where a lack of photorespiration has otherwise reduced the chloroplast demand for ATP relative to NADPH. Hence, shunt activity could contribute to maintaining leaf carbon balance and/or chloroplast energy balance at ECO₂.

In the scenarios described above, one could hypothesize that a cellular P_i indicator (perhaps low cytosolic or stromal P_i) may act as a signal to induce GPT2/GPT3 expression. There is some evidence in the literature consistent with this idea. One example involves an Arabidopsis mutant with an inactivated chloroplast ATP synthase.⁶⁹ This mutant displays a high accumulation of protons in the thylakoid lumen in the light since the proton gradient is not being dissipated for ATP synthesis. Hence, one can speculate that this mutant also maintains a higher than usual stromal P_i concentration. Interestingly, this mutant showed reduced expression of GPT2,⁶⁹ consistent with the hypothesis that GPT2 expression is responsive to P_i. Another example involves several independent studies showing that nutritional P deficiency increases Arabidopsis GPT2 expression and that this elevated expression is rapidly reversed upon P resupply.^64-66 Nutritional P deficiency may make metabolism more susceptible to P_i limitations, despite the buffering effect of a vacuolar store of P_i.⁷⁹ These studies hint that, besides being a sugar-responsive gene, some aspect of tissue P status influences GPT2 expression. In turn, GPT2 transport activity will influence the partitioning of both carbon and P between the chloroplast and cytosol.

Another means to release P from an accumulating pool of leaf sugar phosphates (besides synthesis of sucrose or starch) would be to process these intermediates in respiratory pathways. In particular, when phosphoenolpyruvate (PEP) is acted upon by PEP carboxylase or pyruvate kinase (PK), the P in PEP is either directly released as P_i (in the case of PEP + HCO₃⁻ conversion to oxaloacetate + P_i by PEP carboxylase) or released into the adenine nucleotide pool (in the case of PEP + ADP conversion to pyruvate + ATP by PK). There is evidence that these enzymes may be activated in leaf by high carbohydrate status. PEP carboxylase is activated by Glc-6-P and by a serine phosphorylation that is promoted by sucrose and trehalose-6-P via an unknown mechanism.^80,81 Phosphorylation of cytosolic PK, potentially by an SnRK1 kinase, can inhibit its activity and promote its degradation.⁸¹ Hence, inhibition of SnRK1 kinases by trehalose-6-P and Glu-6-P¹³ could promote PK activity. In both tobacco⁸² and potato,⁸³ source leaf PK activity has been shown to increase over the later half of the light period, when leaf sugar status is expected to be highest. In turn, activation of PEPC and PK will stimulate glycolysis by lowering PEP, a potent inhibitor of upstream glycolytic metabolism.^26,81

The organic acids produced by PEP carboxylase and PK (oxaloacetate and pyruvate, respectively) are further processed by the TCA cycle, where the activity of several enzymes is tightly controlled by the end-product (NADH) inhibition.²⁶ However, this biochemical control is relaxed in tobacco if AOX is active since it reduces the respiratory ATP yield associated with NADH turnover.³⁷ In tobacco leaves, pyruvate amounts are two-fold higher in light than dark⁸² and pyruvate is a potent activator of tobacco AOX.³⁷ Hence, AOX (like GPT2) could provide a means to manage an accumulating cytosolic pool of sugar phosphates. In fact, both tobacco AOX knockdowns³⁵ and an Arabidopsis GPT2 knockout⁸⁴ showed increased leaf amounts of Glu-6-P at high photosynthetic rates. Similar to GPT2, studies have shown that AOX increases during nutritional P deficiency, providing a means to maintain respiration when low pool sizes of P_i and/or ADP can otherwise limit respiratory carbon flow and oxidative phosphorylation.^85,86

Under a wide range of growth irradiance and water status conditions, AOX protein amount in tobacco leaf correlates strongly and positively with chloroplast excitation pressure, a chlorophyll fluorescence parameter used to estimate the redox state of the plastoquinone pool within the cETC.⁸⁷ This suggests that a signal closely associated with chloroplast redox state can control AOX amount in response to changes in irradiance and water status. This would ensure that a rise in the reduction state of the cETC could be stabilized by an increased amount of AOX respiration acting as an electron sink. However, growth at ECO₂ lowers excitation pressure,³⁵ yet also results in an increase in AOX amount (Figure 1). This suggests that another signal is responsible for controlling AOX amount in response to changes in growth CO₂ concentration. We hypothesize that both AOX and GPT3 gene expression may be responding to a carbohydrate and/or P_i signal at ECO₂. In general, carbohydrate and P signaling are thought to be closely linked.^52,57 Reactive oxygen and/or reactive nitrogen signals have been shown to increase both AOX1a⁸⁸ and GPT3⁶⁸ transcript amounts in tobacco. Hence, these also represent potential signals in the response of these genes to growth at ECO₂.

We speculate that GPT3 and AOX act in a daily recurring manner to manage leaf sugar phosphate pools during growth at ECO₂. If the stimulation of source activity by ECO₂ generates an imbalance between source and sink activity, then leaf sugar phosphates would be expected to rise over the course of the day, particularly as the transient starch storage pool begins to saturate.⁸⁹ Left unchecked, this could result in the threshold carbohydrate signals able to induce photosynthetic acclimation. However, GPT3 and AOX activities would provide means to manage a late-in-day carbon imbalance, hence avoiding the photosynthetic acclimation that would then compromise the longer-term potential for carbon gain. In other words, a daily late-in-day management or even loss of carbon might be more than made up for if it avoids a robust long-term down-regulation of photosynthetic capacity. Our results also suggest that, at ECO₂, AOX respiration becomes progressively more important over time, as plant size increases (Figure 1). This may indicate that the source:sink imbalance is amplified as the plant develops and acquires more source leaves, while the overall relative growth rate is likely declining. Hence, future studies examining the influence of AOX and GPT3 on growth, carbon balance, and photosynthetic acclimation at ECO₂ will need to consider both developmental and time-of-day effects.

Disclosure of Potential Conflicts of Interest

The authors report no conflict of interest.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Additional information

Funding

This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada to G.C.V. (grant no. RGPIN-2019-04362)