Biochemical control systems for small molecule damage in plants

ABSTRACT

As a system, plant metabolism is far from perfect: small molecules (metabolites, cofactors, coenzymes, and inorganic molecules) are frequently damaged by unwanted enzymatic or spontaneous reactions. Here, we discuss the emerging principles in small molecule damage biology. We propose that plants evolved at least three distinct systems to control small molecule damage: (i) repair, which returns a damaged molecule to its original state; (ii) scavenging, which converts reactive molecules to harmless products; and (iii) steering, in which the possible formation of a damaged molecule is suppressed. We illustrate the concept of small molecule damage control in plants by describing specific examples for each of these three categories. We highlight interesting insights that we expect future research will provide on those systems, and we discuss promising strategies to discover new small molecule damage-control systems in plants.

Plant metabolism is not a one-way street

In a simplistic model, central plant metabolism might be seen as a perfect network of well-understood enzymatic reactions. But plant metabolism, just like that of other organisms, not only extends far beyond the limits of central pathways, but is also prone to errors. Small molecules, such as coenzymes, cofactors, metabolic intermediates, and inorganic molecules, are constantly exposed to spontaneous reactions, such as addition, elimination, photolysis, oxidation, or condensation. These reactions frequently lead to the formation of unwanted compounds, especially when plants are challenged by ambient conditions such as temperature extremes, high light intensities, drought, and salinity. Likewise, small molecules may be damaged through the enzymatic action of “promiscuous” enzymes. These enzymes act on non-canonical substrates or catalyze an abnormal reaction on their normal substrate/s. Promiscuous activities of enzymes can be increased by changes in metal cofactors or temperature. Metabolites can also be damaged through the combined action of promiscuous enzymes and spontaneous reactions.

The resulting damaged small molecules are in the best case useless but in most cases harmful to the cell. We propose that at least three types of biochemical systems have evolved in plants to cope with the consequences of damaged small molecules: (i) repair, (ii) scavenging, and (iii) steering (Figure 1).

Figure 1.

Small molecule damage-control systems in plants. The enzymatic action of promiscuous enzymes and spontaneous reactions can lead to damage of small molecules. Three types of biochemical systems evolved in plants to cope with the consequences of damaged small molecules. (i) Repair systems return a damaged metabolite to its original state through one- or multiple-enzymatic steps. (ii) Scavenging systems, which can be enzymatic or non-enzymatic, convert reactive metabolites to harmless products. (iii) Steering systems suppress the possible formation of a damaged metabolite by directed overflow or by changing the supply of reaction substrates.

(i) Repair systems

Repair systems return a damaged small molecule to its original state. These systems of small molecule proofreading consist of one or more enzymatic reactions (Figure 1).¹ The following two examples illustrate single- and multi-enzyme repair mechanisms.

L-2-hydroxyglutarate dehydrogenase, a one-step repair system that corrects the error of mitochondrial malate dehydrogenase

An enzymatic error occurring in mitochondria is the formation of L-2-hydroxyglutarate (L-2HG) by NAD-dependent malate dehydrogenase (mMDH; EC1.1.1.37). The main activity of mMDH is the interconversion of malate to oxaloacetate (OAA), to supply the substrate for the citrate synthase reaction and to provide NAD for the glycine decarboxylase reaction of the photorespiratory pathway depending on cellular needs.² In plants, a side reaction of both mMDH isoforms, mMDH1 and mMDH2, catalyzes the NADH-dependent reduction of 2-ketoglutarate (2-KG) to L-2HG.³ L-2HG is not known to be involved in any further metabolic steps. Instead, it is metabolized back to 2-KG in a one-step reaction by the mitochondrial metabolite repair enzyme L-2HG dehydrogenase (L-2HGDH; EC1.1.99.2).^3,4

This reaction prevents loss of carbon moieties from the tricarboxylic acid cycle and most probably protects cells from the accumulation of L-2HG. In humans, individuals with mutations in L-2HGDH possess elevated levels of L-2HG,^5,6 which contributes to the development of neurologic disorders and the formation and malignant progression of brain tumors.^7-9 Conversely, although A. thaliana loss-of-function mutants in L-2HGDH contain elevated L-2HG levels, they show no obvious phenotype under a wide range of tested growth conditions.³ However, all genomes of embryophytes and green algae possess orthologs of this enzyme,¹⁰ indicating that L-2HGDH makes an essential contribution to plant fitness. Future work will elucidate whether this fitness contribution is small and thus easily overlooked, or whether it is only present under certain conditions in which cellular homeostasis is perturbed at different points simultaneously.

Phylogenetic analyses revealed that all plant L-2HGDH – except for additional homologs in the basal plants Selaginella moellendorffii and the moss Physcomitrella patens – cluster together and are distantly related to animal and bacterial L-2HGDH sequences, which form a separated clade.¹⁰ Selaginella and Physcomitrella possess a second copy of L-2HGDH that belongs to the subfamily found in animals and bacteria.¹⁰ This indicates that the ancestor of photosynthetic eukaryotes possessed two homologs of L-2HGDH, one of which was subsequently lost in green algae and in angiosperms. The maintenance of L-2HGDH copies from both subfamilies in Selaginella and Physcomitrella provides an interesting starting point for the analysis of the putative involvement of this enzyme family in metabolite repair in these basal plants.

The oxidative photosynthetic carbon cycle, a multiple-step repair system that evolved to cope with the product of the oxygenase activity of Rubisco

All oxygenic photosynthetic organisms use Rubisco (Ribulose-1,5-bisphosphate carboxylase oxygenase) in the first reaction of the Calvin-Benson cycle to carboxylate ribulose 1,5-bisphosphate (RuBP), its canonical substrate, and produce 3-phosphoglycerate (3-PGA). The action of this enzyme is inefficient, as it possesses a low catalytic rate and rather low CO₂ affinity.¹¹ In a side reaction, Rubisco catalyzes the oxygenation of RuBP, yielding one molecule of 3-PGA and one molecule of 2-phosphoglycolate (2-PG). 2-PG is a dead-end metabolic product. It is of high toxicity, as it inhibits chloroplastic triose phosphate isomerase and phosphofructokinase.^12-14

During early plant evolution, when the atmosphere was essentially free of O₂ and the levels of CO₂ were high (1500–3000 ppm;¹⁵), the production of 2-PG was negligible.¹⁶ Due to this, there was originally no selective pressure against Rubisco’s potential promiscuity.¹⁵ Drastic changes in the composition of atmospheric gases occurred after the onset of oxygenic photosynthesis: CO₂ was continuously fixed and not released back into the atmosphere, while massive amounts of O₂ were produced.¹⁷ As a selective response to Rubisco’s increased oxygenase activity, the oxidative photosynthetic carbon cycle – also known as the photorespiratory pathway – evolved as a multiple-step repair system to convert 2-PG back to 3-PGA. Through the reactions of the Calvin-Benson cycle, 3-PGA is converted back to RuBP, recovering almost 75% of the carbon diverted by the oxygenase activity of Rubisco.^18,19 Once established, this multi-step repair cycle became intimately interconnected with primary metabolism. Plants use the photorespiratory pathway or some of its reactions for the synthesis of glycine and serine during the day,^20,21 as well as for the avoidance of photoinhibition via the regeneration of acceptor molecules for the light reactions.^22,23 Today, atmospheric O₂ is roughly 500-fold more abundant than CO₂, with a correspondingly high flux through the photorespiratory pathway; the rate of CO₂ release from glycine decarboxylation through the photorespiratory pathway can reach five times the rate of normal tricarboxylic acid cycle activity.²⁰

The photorespiratory pathway evolved as a multi-step repair system in all organisms carrying out oxygenic photosynthesis, although the individual biochemical steps and subcellular localization may differ. The best-studied photorespiratory pathway is that of higher plants, which involves 16 reactions distributed across chloroplast, peroxisome, mitochondrion, and the cytosol.^18,19 Future work should be directed towards elucidating the exact biochemical steps of this multiple-step repair system and their subcellular localizations in algae of all lineages, especially in those that originated through secondary endosymbiosis (e.g., diatoms), which seem to have evolved particular characteristics.²⁴

(ii) Scavenging systems

Scavenging systems convert reactive molecules produced during metabolism into more harmless ones. These systems include non-enzymatic and enzymatic reactions as parts of single- or multiple-step mechanisms (Figure 1). The following examples illustrate the most important scavenging mechanisms of plant cells.

Scavenging of reactive oxygen species to prevent oxidative damage and create cellular messengers

Reactive oxygen species (ROS) are small inorganic molecules such as singlet oxygen (¹O₂), the superoxide anion radical (O₂^−.), the hydroxyl radical (OH·), and hydrogen peroxide (H₂O₂).²⁵ They are continuously produced as an unavoidable consequence of aerobic metabolism and in addition, abiotic and biotic stresses enhance their generation.^26,27 In plant cells, chloroplasts, mitochondria and peroxisomes are major sources of ROS as they possess high rate of electron flow and oxidizing metabolic activity.^28-32

Increased ROS levels may cause oxidative stress and damage to important biological molecules such as lipid peroxidation, protein oxidation, and DNA structural modifications.³³ Cellular ROS levels are determined by the rates of their production and elimination via a high number of scavenging systems.²⁵ The highly toxic nature of ROS explain the evolution of complex enzymatic and non-enzymatic ROS scavenging systems in plants.³⁴ These systems are interlinked through shared metabolites and reducing equivalents and can be found in different subcellular compartments.

ROS enzymatic scavenging systems include superoxide dismutase, which removes O₂^−. by catalyzing its dismutation, resulting in H₂O₂ and O₂; catalase, which removes H₂O₂ generated in peroxisomes by its dismutation into H₂O and O₂; ascorbate peroxidase, which is involved in H₂O₂ scavenging in the in water-water and glutathione-ascorbate cycles utilizing ascorbic acid as electron donor; glutathione reductase, which catalyzes the NADPH dependent reduction of disulphide bond of GSSG and thus maintains the GSH pools; NADH dependent monodehydroascorbate reductase, which is an enzymatic component of the glutathione-ascorbate cycle that reduces monodehydroascorbate to ascorbate; dehydroascorbate reductase, which regulates ascorbic acid cellular redox state by regenerating its oxidized state; glutathione peroxidase, which uses glutathione to reduce H₂O₂ and organic and lipid hydroperoxide; and glutathione-S-transferase, which catalyzes the conjugation of electrophilic xenobiotic substrates with glutathione.³⁴

To the ROS non-enzymatic scavenging systems belongs ascorbic acid (vitamin C), which donates electrons in a number of enzymatic and non-enzymatic reactions. Ascorbic acid provides protection to membranes by directly scavenging O₂^−. and OH· and by regenerating α-tocopherol from its radical, it supports dissipation of excess excitation energy acting as cofactor of violaxantin de-epoxidase, and participates in the glutathione-ascorbate cycle. Other important non-enzymatic scavenging systems are glutathione, which is a pivotal component of the glutathione-ascorbate cycle, carotenoids and phenolic compounds (such as flavonoids), which neutralize radicals through their capacity of oxidizing them and serving as more stable, less-reactive radicals, and α-tocopherol (vitamin E), which is major antioxidant in biomembranes by preventing chain propagation in lipid autooxidation.³⁴

The activity of all mentioned scavenging systems is important not only to eliminate toxic ROS, but also to maintain them at non-harmful levels and thus allow them to act as cellular messengers to regulate plant development and stress responses.³⁵ Amongst different ROS, H₂O₂ is an ideal signalling molecule due to its relative stability and diffusibility.³⁶ Chloroplasts and peroxisomes are cellular compartments with capacity to generate and release H₂O₂ as signal molecules. H₂O₂ produced in the chloroplasts was shown to modulate the control of gene transcription and secondary-signaling messengers while H₂O₂ produced from peroxisomes most likely induces stress tolerance or protective responses.³¹ Further work will be needed to functionally dissect the components of the signaling network and to characterize specific cellular and whole organismal responses, such as DNA damage/repair and the responses of plants to a/biotic challenges.

Reactive carbonyl species are maintained at non-harmful cellular levels through the glyoxalase system

Reactive carbonyl species (RCS) are small electrophilic mono- and di-carbonyl molecules that are unavoidably produced in normal cellular metabolism as well as stress conditions.³⁷ RCS are highly reactive toward cellular macromolecules, resulting in their irreversible modification and the formation of adducts.^38-40 RCS can be generated through non-enzymatic processes such as lipid peroxidation, amino acid oxidation, glycation, autoxidation of glucose or triose phosphates and through disrupted enzymatic reaction.

Glyoxal (GO) and methylglyoxal (MGO) are the most common RCS found in plant tissues. Sources of GO in plant metabolism are assumed to be lipid peroxidation and the fragmentation of glycated proteins, whereas MGO will be mainly generated via non-enzymatic mechanisms and also through the combination of the enzymatic action of triose phosphate isomerase with a non-enzymatic decomposition of the reaction intermediate.^41-43 Triose phosphates are prone to the loss of the α-carbonyl protons resulting in an enediolate phosphate intermediate. Subsequently, MGO is formed by a spontaneous β-elimination of the phosphate group. During glycolysis, the canonical action of triose phosphate isomerase is the catalysis of the isomerisation of glyceraldehyde-3-phosphate and dihydroxyacetone phosphate forming the enediolate phosphate intermediate. This intermediate can escape the catalytic site and decompose to MGO and inorganic phosphat.^41,42

To avoid cellular damage the levels of RCS should be maintained in a low certain range. In recent times it is often quoted that RCS might also act as cellular signaling messengers in plants,⁴⁴ although definitive evidence for this is lacking. The main pathway to metabolize GO and MGO is the glyoxalase system, which consists of the enzymes S-D-lactoylglutathione lyase (glyoxalase I, GLXI) and S-2-hydroxyacylglutathione hydrolase (glyoxalase II, GLXII) acting in tandem.⁴⁵ Through this pathway, GO is converted into glycolate and MGO into D-lactate.²⁴ Glycolate can be further metabolized by peroxisomal glycolate oxidases,⁴ while D-lactate is converted to pyruvate through mitochondrial D-lactate dehydrogenase.^46,47

In plants MGO formation is closely connected to the fluxes through glycolysis and the Calvin-Benson cycle. Accordingly, at least in the model plant A. thaliana, the glyoxalase system is functional in the cytosol and chloroplasts and the expression of one out of three GLXI isoforms is regulated by the cellular sugar status.²⁴ It is assumed that stress increasing the cellular sugar status, such as osmotic stress,⁴⁸ or the possible autoxidation of glucose under heat stress will also influence the expression of the GLX system. In A. thaliana, specific GLXI isoforms change their specificity to GO or MGO depending on the metal cofactor used. Moreover, the functional analysis of loss-of-function mutants in the GLXI paralogs indicated that elimination of toxic GO and MGO during germination and seedling establishment depends only on the activity of one specific GLXI cytosolic isoform.²⁴ These facts rise the question whether individual GLXI isoforms are able to act on other RCS than GO and MGO and whether they are recruited at different developmental stages and cellular metabolic conditions. We propose that the glyoxalase system possess high plasticity to respond to the actual necessities of the cells. Future work will indicate if this system is able to metabolize other RCS found in plant tissues and which components of the system act in concert under specific cellular conditions induced by changes in metabolic homeostasis.

(iii) Steering systems

Steering systems evolved to by-pass and thus prevent the formation of damaged small molecules. These systems include mechanisms through which small molecules are diverted out of the canonical metabolic pathways in which they are involved (Figure 1). Through the mechanism of directed overflow, control enzymes prevent the accumulation of molecules – that in excesses can be prone to be damaged – by converting them into less harmful ones.^49,50 Another type of mechanism reduces the unwanted side reaction of an enzyme by changing the supply of the substrates for that reaction.

Few examples of steering systems in plants are currently found in the literature, as this concept is very new. We anticipate that many more small molecule steering systems in plants will be discovered and characterized in the near future. The next examples illustrate steering mechanisms in plants.

A family of metal-dependent phosphatases are involved in a directed overflow mechanism

Coenzyme A (CoA) precursors such as phosphopantothenate and phosphopantetheine can accumulate under certain conditions and be converted into harmful oxidized forms (S-sulfonate, sulfonate, or other forms).⁵¹ Members of a family of metal-dependent phosphatases (DUF89 proteins) were recently found to dephosphorylate those molecules thus limiting the pool size of the regularly occurring phosphometabolites and consequently decreasing the potential of harmful build-ups.⁵¹ In addition, these control enzymes have also the capacity to act on the already damaged molecules to convert them into harmless ones.

The C₄ photosynthetic pathway, a mechanism that evolved to reduce the production of toxic 2-phosphoglycolate by the oxygenase activity of Rubisco

The C₄ photosynthetic pathway is a CO₂-concentrating pump that evolved in Angiosperms at least 66 times independently (C₄ plants) as a metabolite steering mechanism in order to stimulate carboxylation and inhibit oxygenation of RuBP by Rubisco (see above).⁵²

C₄ photosynthesis metabolically concentrates CO₂ from the intercellular air spaces into a cellular compartment where Rubisco is localized. The biochemical steps involved in the C₄ photosynthetic pathway are spatially separated. In most cases, this separation occurs at the cellular level: the first steps take place in the mesophyll cells, where CO₂ is initially fixed into C₄ acids (malate or aspartate). The C₄ acids diffuse to the bundle-sheath cells, where they undergo decarboxylation. In this way, CO₂ is liberated for the assimilation through Rubisco, which is exclusively located in the chloroplasts of bundle-sheath cell. Through this CO₂-concentrating pump C₄ plants achieve CO₂ concentrations around 10 times higher (1,500 μmol mol^–1 air) in bundle-sheath cells than those in mesophyll cells.⁵²

The establishment of the C₄ metabolism not only involved the foundation of a new biochemical pathway but also a broad range of cellular and physiological adaptations such as an increase of bundle-sheath cell volume, cell-type-specific expression patterns and specific kinetic and regulatory properties of many enzymes and transporters.^53-55 The origins of C₄ species would have been driven by a drop in atmospheric CO₂ levels that favoured these species over C₃ ones because of their greater photosynthetic efficiency at low CO₂ levels, particularly when combined with high temperatures and drought stress.¹⁵ Due to this, it is desirable to transfer this mechanism into C₃ plants lacking this steering mechanism.⁵⁶ However, genetic engineering towards C₄ metabolism is limited by incomplete understanding of the system, which is a complex trait.⁵⁷ Taking into account the importance for improved modeling of C₄ regulatory networks aimed at redesigning and engineering the C₄ pathway, there is still a need to understand the action and interaction of the major C₄ enzymatic activities at a molecular scale and to discover and experimentally verify the importance of regulation of this steering system.

Discovering new small molecule damage-control systems

Identifying novel small molecule damage control systems is a challenge for several reasons. There may be no information on the identity of the small molecules themselves, which enzymes produce them, and which downstream enzymes act on them. Several computational approaches have been used to predict products of promiscuous enzyme activities or novel salvage pathways.^58-60 A fundamental limitation of these algorithms is that enzymes have often not been experimentally tested with a large set of chemically diverse substrates, resulting in scarce availability of training data.⁶¹ High-throughput methods and approaches in biochemistry will be needed to generate such data by profiling a large number of known enzymes for their activity with alternative substrates.⁶² Novel approaches leveraging machine learning can efficiently reduce the screening burden of such efforts by predicting which substrates are most informative.⁶¹

Even if a novel promiscuous enzymatic reaction has been identified as causal in the production of a particular metabolite in one organism, it is unclear whether this will be true for orthologs. Many existing computational pipelines predict functional annotation for genes that share sequence similarity to previously characterized ones. However, systematic analyses show that the sequence identity must be at least 50–60% to transfer primary enzyme function with relatively high accuracy.^63-66 It is still an open question to what extent enzyme substrate promiscuity is conserved in orthologs, and therefore whether it can be predicted bioinformatically using sequence identity. The conservation of L-2HGDH genes across plants indicates that the promiscuous function of mitochondrial MDH is widespread¹⁰ and indicates that bioinformatic prediction may indeed be possible.

Several approaches make use of metabolomics for identifying new enzyme functions and metabolic pathways.^67,68 This could prove a powerful way of identifying enzymes acting on enzyme side products even when those side products are not yet known, as well as on unwanted compounds that are formed non-enzymatically. A notable application of this approach identified 241 potential novel enzymes by testing all 1,275 functionally uncharacterized genes from the Escherichia coli genome. This was achieved using supplemented metabolite extracts, recombinant enzymes, and non-targeted meta-bolomics.⁶⁹ The strength of this approach is that it relies on metabolite extracts as a source of substrates, so that it is applicable even when the substrates have not yet been identified. The main disadvantage of this approach is that up to 98% of identified mass peaks cannot be annotated,⁷⁰ and determining the substrate identity can therefore be cumbersome.

However, computational approaches that predict metabolites resulting from enzyme promiscuity show great promise in aiding this identification.⁷¹ Such methods predict metabolites that are likely to occur based on known metabolites and common biochemical reactions. The predicted metabolites might, in turn, inform the generation of new hypotheses and form starting points for the search for enzymes that catalyze their conversion in repair, scavenging, or steering pathways. One way to identify enzymes potentially involved in such pathways would be to find genomically encoded homologs of enzymes known to consume the predicted metabolites, utilizing the power of comparative genomics.⁷² There are still close to 7,000 uncharacterized enzymes and transporters in Arabidopsis;⁷³ we expect that a significant number of these belong to important undiscovered biochemical systems to control small molecule damage.

Funding Statement

This work was supported by the Deutsche Forschungsgemeinschaft under grants FOR 1186, MA2379/11–2, and EXC 1028 to VGM.