The dual role of methylglyoxal in plant stress response and regulation of DJ-1 protein

2.1 Non-enzymatic pathways for MG production

Among non-enzymatic pathways of MG production, leakage from the glycolysis pathway contributes to over 90% of MG primary sources. In glycolysis, glucose is converted to fructose-1,6-bisphosphate through a series of reactions. This six-carbon sugar compound is then cleaved into two three-carbon phosphate compounds, glyceraldehyde 3-phosphate (GAP) and its isomer dihydroxyacetone phosphate (DHAP) (Phillips&Thornalley, 1993) (Figure 1a). Due to the instability of these three-carbon phosphates, DHAP and GAP are susceptible to rapid loss of the α-carbonyl proton during the isomerization transition to form the enediolate phosphate intermediate (Richard, 1993). The C3-hydroxyl group in DHAP and its resulting enediolate phosphate 1 is stabilized, requiring internal proton transfer to convert to enediolate phosphate 2. Triose phosphate isomerase (TPI) catalyzes the isomerization reaction by eliminating the β-phosphate group rapidly, leading to the formation of the corresponding ketone and facilitating MG production (Sousa Silva et al., 2013). Therefore, the rate of conversion of GAP to MG is approximately eight times that of DHAP (Phillips & Thornalley 1993). It is worth noting that this process, despite the involvement of TPI, is regarded as non-enzymatic because the MG formation itself does not depend on direct enzyme catalysis. Under certain conditions, such as a strongly alkaline environment, DHAP and GAP can also be directly converted to MG without any enzymes. In humans, only about 0.09–0.4% of the glycolysis flux may be associated with MG production (Martins et al., 2001; Thornalley, 1988). This is mainly because, under normal physiological conditions, the non-enzymatic pathway for MG production exists only as a side reaction in the glycolysis process, and the spontaneous dehydration reaction rate of DHAP and GAP is relatively low. In plants, the glycolysis pathway is connected to photosynthesis, in which the Calvin Cycle reaction produces three-carbon sugars (3-phosphoglycerate (3-PGA)), which are then converted to glucose. The glucose can be used as a substrate for glycolysis to produce MG. MG production during photosynthesis increases with light intensity or CO₂ concentration and correlates positively with O₂ evolution rates (Takagi et al., 2014).

In addition, the process of forming advanced glycation end products (AGEs), a non-enzymatic glycation reaction, is another important non-enzymatic pathway of MG production. It is a slow process first described by the French chemist Louis-Camille Maillard in 1912 as the non-enzymatic browning of food during heating. Interestingly, the glycation reaction that generates MG in plants and other organisms shares similarities with the Maillard reaction in terms of chemical mechanisms. In this process, the aldehyde groups of reducing sugars (e.g., glucose and fructose) readily react with free amino acids or amino acid groups in proteins to produce carbinolamine intermediates, which, upon dehydration, form imine adducts. The imine adducts are converted to MG through spontaneous Amadori rearrangement as well as oxidative cleavage (Davídek et al., 2006) and retro-aldol degradation (Cämmerer et al., 1999; Vistoli et al., 2013) (Figure 1b).

Lipid peroxidation reactions mainly refer to a series of free radical reactions occurring in plant unsaturated fatty acids. This is one of the major damaging changes induced by oxygen radicals and also an important pathway for MG generation (Esterbauer et al., 1982). When leaves are exposed to intense light, double bonds in unsaturated fatty acids are susceptible to ¹O₂ attack to form lipid peroxides (Triantaphylides et al., 2008). In the presence of specific catalysts such as transition metal ions (Fe³⁺) or reducing agents, the lipid peroxidation reaction is enhanced. The intermediates like acetoacetate or acetone spontaneously decompose through the Fenton reaction to generate carbonyls, including MG (Grosch, 1987) (Figure 1c).

2.2 Enzymatic pathways for MG production

During the catabolism of glucose, amino acids, and fatty acids, the production of MG in bacteria and animals involves both enzymatic and non-enzymatic pathways. The enzymatic pathways have not been reported in plants. Methylglyoxal synthase (MGS) is the only enzyme known to catalyze the direct conversion of DHAP to MG during glycolysis (Cooper, 1984; Cooper & Anderson, 1970; Hopper & Cooper, 1972). MGS is widely found in microorganisms (e.g., E. coli and Pseudomonas aeruginosa), and the enzyme has been isolated from goat liver (Cooper & Anderson, 1970). Additionally, there are several other enzymes in microorganisms and animals that are involved in the production of MG. For example, during the catabolism of glycine and threonine to form aminoacetone intermediates, semicarbazide-sensitive amine oxidase (SSAO) catalyzes the production of MG from these intermediates (Lyles & Chalmers, 1992). Acetoacetate formed in the process of fatty acid catabolism can be oxidized by myeloperoxidases (MPO) to generate MG (Panasenko et al., 2022). Similarly, cytochrome P450 (CYP450) oxidizes acetone to MG using nicotinamide adenine dinucleotide phosphate (NADPH) as a cofactor (Koop & Casazza, 1985). During triacylglycerol hydrolysis in lipid metabolism, glycerol kinase also converts the resulting glycerol phosphate intermediates into MG (Jung et al., 2011) (Figure 1d). Clearly, the production of MG is an inevitable and multifaceted process. However, the equivalent enzymatic pathways have not yet been discovered in plants, presenting a great opportunity for investigating the mechanisms of MG production in the plant kingdom. The aforementioned discoveries made in microorganisms and animals can certainly inform plant research.

4.1 MG in low concentration acts as a signalling molecule under salt stress

Signalling molecules are typically small molecules that respond rapidly to environmental changes (de Bari et al., 2020) and can be quickly produced inside the cell and trigger downstream signalling pathways. MG may be regarded as a signalling molecule potentially involved in a variety of plant physiological processes, including stomatal movement, plant growth and development, and stress response (Hossain et al., 2016).

Plants have evolved different mechanisms to sense and respond to environmental stresses, such as regulating gas exchange and water loss through stomatal pores on leaf surfaces (Jia&Zhang, 2008). Light-induced stomatal opening is mediated by K⁺ accumulation in guard cells, whereas stomatal closure is associated with an increase in cytosolic calcium concentration (Ca²⁺_cyt). Ca²⁺_cyt in guard cells oscillates in response to environmental stresses (Pei et al., 2000; Young et al., 2006). Hoque et al. found that MG inhibits photoinduced stomatal opening and induces stomatal closure but does not affect the guard cell viability. It decreases K⁺ influx by directly inhibiting inward-rectifying potassium channels (K_in) in the guard cells, thereby affecting plant water transpiration and gas exchange (Hoque et al., 2012a; Hoque et al., 2012b). Additionally, MG-induced stomatal closure involves extracellular peroxidase-mediated oxidative damage and (Ca²⁺_cyt) oscillations. This process is completely inhibited by a peroxidase inhibitor salicylhydroxamic acid (SHAM) (Samuilov & Kiselevsky, 2016), but is not affected by the respiratory burst NADPH oxidase mutants atrbohD and atrbohF (Dietrich et al., 2001). Thus, RBOH-dependent ROS production does not appear to be involved in MG signalling.

Exogenously applied MG modulates other plant signalling molecules, such as phytohormones gibberellins and abscisic acid (ABA), which play key roles in the regulation of plant growth and development and response to environmental stresses. For instance, the addition of exogenous MG significantly inhibits seed germination and root elongation. This effect is closely related to the ABA signaling pathway, especially through the upregulation of stress-responsive genes, like RD29B and RAB18. These genes play pivotal roles in salt stress response, suggesting that MG exerts a regulatory role in this process by modulating the ABA-dependent signalling pathway (Hoque et al. 2012a). Li et al. found that exogenous MG can cross the plasma membrane into plant cells and activate GLO I and GLO II in the shoots and roots of wheat seedlings under salt stress (Li et al., 2017). This activation helps to mitigate the increase of MG levels and the accumulation of GSH. Moderate accumulation of MG activates specific signalling pathways, inducing the expression of antioxidant enzymes such as superoxide dismutase (SOD), catalase (CAT), ascorbate peroxidase (APX), and glutathione reductase (GR). This process also enhances the levels of ascorbate (ASA), GSH, and osmolytes such as proline, glycine, betaine, and soluble sugars, thereby alleviating the increase in ^•O₂⁻, H₂O₂, and MDA under oxidative stress. Roy et al. (2004) proposed that 0.5 mM MG can completely replace cytokinin to effectively promote the differentiation of healing tissues into plantlets (Roy et al., 2004). MG can also self-regulate its levels in cells through a feedback mechanism of activating TPI. In the isomerization reaction, the balance shifts towards the formation of GAP, thereby reducing the levels of MG precursor DHAP (Sharma et al., 2012). GAP, in turn, supplements the energy reserves via glycolysis, thereby helping plants to survive adverse conditions. Exogenously added MG may act as a stress signal, which has a positive effect on seed germination, growth and development, alleviation of oxidative stress, and even acquisition of stress resilience.

MG and its derivatives (e.g., AGEs) can affect cell signalling pathways through various mechanisms, inducing post-translational modifications (PTMs) (Donnellan et al., 2022; Galligan et al., 2018; Lu & Holmgren, 2014). In one study, Zheng et al. identified a mechanism by which MG regulated gene expression through methylglyoxalylation of histones under salt stress (Fu et al., 2021). Histones are prone to PTMs because of their high proportion of lysine and arginine residues. When histones undergo PTMs, they can affect gene expression by altering chromatin structure and recruiting various specific binding proteins. Thus, histone modifications play a key role in a variety of biological processes, such as cell differentiation, tissue development, cancer induction, and environmental response. Salt stress can lead to the accumulation of MG in plants, which can induce histone PTMs and improve the accessibility of plant chromatin, thus promoting the expression of salt stress-responsive genes, alleviating the damage to plants, especially in cells lacking GLO I. Modern proteomics, biochemistry and genetics tools will greatly facilitate the characterization of specific PTMs mediated by MG and its signalling role in plant cells.

4.2 High concentrations of MG are toxic to plants under salt stress

Not all PTMs of proteins induced by MG are beneficial. A typical example is non-enzymatic glycation (the Maillard reaction) in cells, leading to non-functional proteins that should be eliminated by proteolysis (Verzijl et al., 2000). Under salt stress, plants accumulate large amounts of MG (Kaur et al., 2016) that react easily with nitrogen, oxygen or sulfur atoms containing lone electron pairs, such as the amino or thiol groups in proteins (Figure 3), ultimately forming AGEs. Therefore, high concentrations of MG (e.g., > 75 μM) are toxic to plant cells. Glycated residues mainly accumulate on extracellular proteins, often making them inactive due to covalent modifications to the catalytic or binding sites and structural changes, resulting in irreversible damage (M Ciulla et al., 2011). Furthermore, glycation most notably targets the guanidino group of arginine residues, and condensation of MG with its Nη atom yields the corresponding carbinolamine amine (Henle et al., 1994), which is then cyclized to form a diol intermediate and dehydrated twice to produce hydroimidazolone derivatives (MG-H1, MG-H2, MG-H3) (Vistoli et al., 2013) (Figure 3. 3.1–3.3), which are in a dynamic equilibrium state because they can be ring-opened to form N⁷-carboxyethylarginine (CEA) (Figure 3. 3.4). This can, in turn, be recyclized, while the addition of MG continues to form argpyrimidine (Figure 3. 3.5) or N_δ-(4-carboxy-4,6-dimethyl-5,6-dihydroxy-1,4,5,6-tetrahydropyrimidine-2-yl)-L-ornithine (THP) (Figure 3. 3.6), with MG-H1 and argpyrimidine being the major glycation adducts. The reaction between MG and lysine residues produces N_ε-(carboxyethyl) lysine (Figure 3. 3.7) and a cross-link between two lysines, 6–1-(5S)-5-ammonio-6-oxido-6-oxohexyl-4-methyl-imidazolium-3-yl-L-norleucine (MOLD, Figure 3. 3.8). Moreover, after MGO has reacted with lysine, it could further react with arginine residues to form a stable carbonyl cross-link, 2-ammonio-6-({2-[(4-ammonio-5-oxido-5-oxopentyl)amino]-4-methyl-4,5-dihydro-1H-imidazol-5-ylidene}amino) hexanoate (Lederer & Klaiber, 1999) (MODIC, Figure 3. 3.9). In addition, MG can also react with cysteine residues to produce reversible hemiacetal adducts (Figure 3. 3.10). These MG-derived AGEs are collectively referred to as MAGEs. They are thought to accumulate slowly throughout the life span, making their concentration represent a good biomarker of plant senescence, and may serve as a label for proteolytic degradation of dysfunctional proteins (Havé et al., 2015).

Nucleic acids can also be irreversibly modified by DNA glycation. Under physiological conditions, the most reactive nucleotide is deoxyguanosine (dG). In vivo, MG reacts with dG to generate nucleotide AGEs, mainly the imidazolopurinone derivative dG-MG (Thornalley et al., 2010) (Figure 3. 3.11), which is self-cyclized from the respective aminocarbinol (Figure 3. 3.12). In addition, stable adducts of N₂-(1-carboxyethyl)-deoxyguanosine (CEdG) (Figure 3. 3.13) have also been reported (Papoulis et al., 1995; Thornalley, 2003), but their relative content was less than that of dG-MG. In cultured HL60 cells, the DNA glycation products were reported to be at the same level or even significantly more abundant than the major DNA oxidation product 8-oxoguanine (per 10⁶ nucleotides: 8-oxo-dG, 1.48 ± 0.39; dG-MG, 18.2 ± 6.5 and CEdG, 1.33 ± 1.12), suggesting that MG glycosylation as DNA damage may be as important as oxidation (Thornalley et al., 2010). There is evidence that DNA glycosylation leads to a loss of genomic integrity. High concentrations of MG produce interstrand DNA cross-links (Rahman et al., 1990) and strand breaks (Pischetsrieder et al., 1999), and increase mutation frequency (Cajelli et al., 1987; Migliore et al., 1990), the majority of the mutations being G:C > C:G and G:C > T:A transversions (Murata-Kamiya et al., 1997, 2000).

6.1 The structure and function of the DJ-1 superfamily

The DJ-1 superfamily is widely distributed in nature, with the core structure and specific active sites conserved in humans, bacteria, and plants. Except for the extended and terminal regions, the tertiary structures of the members of the DJ-1 superfamily are similar (Görner et al., 2007), and all of them share a core α/β sandwich fold (Bonifati et al., 2003), also known as the α/β hydrolase fold. This structure is highly conserved, with a motif called the “ribonuclease elbow”, which carries an invariant cysteine residue, Cys106, in the human protein (Holmquist, 2000). This residue plays a central role in protease/peptidase activity by acting as a nucleophile (Lee et al., 2003), and its oxidation is critical for the cytoprotective activity of DJ-1 (Canet-Avilés et al., 2004; Neumann et al., 2004). In the vicinity of Cys106 in the active site, a glutamate residue Glu18 (or Asp18 in some homologs) is responsible for lowering the pK_a value of Cys106, facilitating its oxidation (Witt et al., 2008).

Despite sharing similar tertiary structures, DJ-1 superfamily members exhibit diverse biochemical activities. It has been shown that human DJ-1 has chaperone activity (Lee et al., 2003), whereas Hsp31 from E. coli has both peptidase activity and chaperone activity (Fu et al., 2023; Mujacic & Baneyx, 2007). Conversely, YhbO from E. coli and DR1199 from Deinococcus radiodurans lack both chaperone and peptidase activities, although they are closely related to human DJ-1 and E. coli Hsp31(Abdallah et al., 2006; Abdallah et al., 2007). The reason for this functional diversity is that they have different interfacial regions, which can form diverse dimerization/oligomerization patterns. Members of the DJ-1 superfamily can homo-oligomerize in at least four different binding orientations through interfacial regions I–IV, which directly affects the oligomerization state and function of the protein. As a result, DJ-1 proteins can be classified into four major types. The dimerization form generated from interface I is called the DJ-1 type, and that formed through interface II is called the YhbO type. Regions III and IV play unique roles in the dimerization pattern within the DJ-1 superfamily; in Hsp31 proteins, this N-terminal extension contains about 45 amino acid residues, a structural feature crucial for the formation of its dimerization contacts. Moreover, Hsp31 possesses an insert near regions I and II, which provides an additional structural platform for dimer formation and protein function. Together, these elements form a unique dimerization surface responsible for mediating specific protein–protein interactions. The form of dimer produced by region IV is referred to as the YDR type. The dimer interface of region IV is completely different from the other types in the DJ-1 superfamily. In the yeast YDR533Cp protein, this region consists of β7, β8 and β9 strands.

Therefore, DJ-1 superfamily members exhibit diverse functional capabilities, in part through variation of the stoichiometry of the complexes they form (Pei et al., 2000). DJ-1 proteins usually form dimers or hexamers in vivo, and this oligomerization state is critical for their stability and biochemical activity (Kim et al., 2005; Wei et al., 2007). It has been shown that dimerization of DJ-1 helps to maintain its active structure, thus ensuring the proper execution of its function. For example, the dimerized form of certain DJ-1 proteins is required for their antioxidant response and molecular chaperone functions. Altered oligomerization states (e.g., abnormal aggregation) have been associated with the development of neurodegenerative diseases, underscoring the importance of the correct DJ-1 oligomerization for the maintenance of its normal biological function. Except for interface III, which contains a long N-terminal extension, patches I, II, and IV are all contained within the core structure of the DJ-1 superfamily, suggesting the possibility of higher-order oligomerization through the combination of several interface types. It has been reported that dimerization contacts can form hexamers in YhbO-type proteins and present them in the solution (Jia & Zhang, 2008). Only hexamers exhibit protein hydrolysis activity, and rotational contacts between the two interfaces may be responsible for the assembly of functional hexamers.

6.2 The research focus on DJ-1 protein and MG metabolism

Recent studies on the function of DJ-1 proteins have focused on their ability to detoxify MG and revert MAGEs. The mechanism of action of DJ-1 on MG is complex, and it has triggered a wide range of discussions and controversies. DJ-1 was initially reported to act as a unique GLO III-like activity capable of removing MG and related compounds in a GSH-independent manner (Lee et al., 2012). This could hinder the glycation of important macromolecules by MG while sparing GSH for other reactions in response to oxidative stress. Additionally, there is evidence to support that DJ-1 has only glyoxalase activity, and some studies have shown that human DJ-1 indirectly reduces glycation damage mainly by consuming free MG but does not directly act on glycation adducts (Andreeva et al., 2019; Pfaff et al., 2017). Recombinant DJ-1 converts MG or the related α-dicarbonyl probes into the corresponding α-hydroxy acid products to prevent stable glycation (Coukos et al., 2023). This protective effect of DJ-1 does not require physical interaction with the target protein, providing direct evidence for a GSH-independent glyoxalase rather than a deglycase mechanism.

Subsequent studies raised controversy over the function of the DJ-1 protein. It was reported that DJ-1 removes MG adducts from arginine, lysine, and cysteine side chains and releases deglycated proteins and lactate, thereby directly repairing glycated amino acid residues (Richarme et al., 2015). In addition, Richarme et al. found that DJ-1 and its E. coli homologs can directly deglycate MG-derived nucleotide adducts in the same way(Richarme et al., 2017). Matsuda et al. also found that recombinant DJ-1 was able to protect CoA and β-alanine, its precursor in the CoA synthesis pathway, from MG. L-lactic acid was formed as a byproduct instead of D-lactic acid produced by traditional glyoxalase, supporting the direct deglycation activity of DJ-1 (Matsuda et al., 2017). Based on these findings, human DJ-1 is currently classified as a “deglycase” in several major databases such as UniProt and NCBI (Figure 4). X-ray crystallographic studies of DJ-1 with its inhibitors (Tashiro et al., 2018) map the active site pocket near the protein surface, a structural layout that would allow DJ-1 to exclusively recognize and interact with the glycation site. This specific binding mode is suitable for the selective removal of specific glycation modifications without interference from other parts of the substrate. Therefore, DJ-1 can efficiently recognize and remove specific glycation modifications caused by MG without broadly interacting with large portions of a substrate macromolecule. This structural property is the key functional basis of DJ-1 as a putative deglycase, enabling it to target modified proteins or other biomolecules for precise repair.

These studies collectively demonstrate that DJ-1 can protect against glycation of proteins and nucleic acids. However, most DNA repair pathways and enzymes are typically substrate-specific, and the finding that DJ-1 can repair a wide range of structurally altered substrates, including carbonylated nucleotides, DNA/RNA, amino acids, and even proteins, is striking. It is also noteworthy that glyoxalase directly converts free reactive carbonyl groups like MG via a chemical reaction rather than acting on already formed glycation modifications. In contrast, deglycases usually remove glycation modifications directly from proteins or other biomolecules, thereby reversing them to their unmodified state, such as proteins modified by MG. Therefore, the mechanism of DJ-1 function in repairing MG by mainly targeting glycated adducts or free small molecule aldehydes, has become an important point of debate. In other words, is DJ-1 a GLO III or a deglycase?

Although the debate is ongoing, this complexity highlights the potential versatility of DJ-1 in the cell through both direct repair of glycation adducts and an indirect effect on the glycation status by scavenging free MG. Further studies are needed to clarify these mechanisms and to explore the specific role of DJ-1 in different organisms, cells and their responses to stress and disease.

6.3 DJ-1 family in plants

Recently, it has been reported in plants that the DJ-1/PfpI superfamily in A. thaliana comprises six genes that share similarities with the recently discovered bacterial and animal glyoxalases, each possessing two tandem DJ-1/PARK7 domains in A. thaliana while only one DJ-1/PARK7 domain in bacterial and animal (Kwon et al., 2013). Among them, each homolog differs in function. In these Arabidopsis DJ-1 homologs, there are two complete DJ-1/PARK7 polypeptide chains, which all contain functionally conserved DJ-1/PARK7 amino acid residues, including specific residues associated with Parkinson's disease (PD) in humans. Comparative analysis of amino acid sequences has revealed the occurrence of a gene duplication event to create two tandem DJ-1/PARK7 domains in A. thaliana. During the process of evolution, such gene duplication events provide the opportunity for the two DJ-1/PARK7 domain copies to diverge functionally: one copy may retain its original function, while the other may evolve to acquire entirely new functions. AtDJ-1A is an oxidative response protein that improves plant tolerance to abiotic stress by activating SOD (Xu et al., 2010). AtDJ-1B is a bifunctional protein that is sensitive to H₂O₂ and has a glyoxalase enzyme (Lewandowska et al., 2019). AtDJ-1C is localized to chloroplasts and is essential for plant growth and development but lacks the conserved Cys residue, resulting in no glyoxalase activity (Lin et al., 2011). AtDJ-1D contains a hemi-sulfide acetal, which is typical of GLO III (Choi et al., 2014b), and is also a stable deglycase (Prasad et al., 2022). AtDJ-1D is widely expressed in seeds, roots, leaves, and pollen (Schmid et al., 2005). It was hypothesized that AtDJ-1D may be a major detoxification enzyme in A. thaliana (Kwon et al., 2013). Since AtDJ-1E and AtDJ-1F do not have catalytic residues in their C-terminal domains, they exhibit very low enzyme activities, so their exact functions are unknown. In addition to Arabidopsis, the DJ-1 protein in date palm has also been functionally validated for the detoxification of MG. Overexpression of PdDJ-1 in Arabidopsis may negatively affect MG homeostasis, inhibiting the photosynthesis machinery and plant functions (Jana et al., 2021).

It is worth noting that DJ-1 proteins may be involved in plant stress responses by affecting the stability, localization, or activity of Nonexpressor of Pathogenesis-Related 1 (NPR1) and dehydration-responsive element binding proteins (DREBs). NPR1 is a key regulator in plant defence responses, activating the expression of a series of defence genes mainly by regulating the signalling pathway mediated by salicylic acid (Zavaliev et al., 2020). DREBs play a central role in plant responses to abiotic stresses such as drought, low temperature, and high salinity, enhancing stress tolerance by specifically binding to dehydration-responsive elements (DREs) in target genes to regulate their expression levels (Lata & Prasad, 2011). This is similar to the mechanism by which DJ-1 in animal cells responds to oxidative stress by regulating nuclear factor erythroid 2-related factor 2 (Nrf2) activity (Zhao et al., 2021). DJ-1 may enhance plant tolerance to abiotic stress by regulating intranuclear translocation and/or stability of NPR1 or regulating the activity of DREBs, thereby modulating plant disease resistance responses. Such regulation may involve direct binding of DJ-1 to these transcription factors or PTMs, such as redox, phosphorylation or protein proteolysis. With the fast advancement of proteomics technologies (Couto-Rodríguez et al., 2023; Liu et al., 2022; Perron et al., 2022), addressing these overarching hypotheses will greatly advance the field of plant biology, and specifically the progress in plant DJ-1 and MG research.