Glutathione levels increase in response to β-carotene and β-apocarotenoid accumulation
We recently reported the conversion of β-apocarotenoids by a set of enzymes so far known as detoxifiers of reactive carbonyl species (RCS), converting aldehydes/ketones into less reactive and potentially toxic alcohols, carboxylic acids or α,β-saturated aldehydes (Fig. 1, Koschmieder et al. 2020). Considering that (non-)enzymatic glutathionylation is an additional, potentially more effective detoxification pathway for RCS (Mano et al. 2019), we investigated the role of (non-)enzymatic glutathionylation in apocarotenoid metabolism.
Cellular levels of GSH are maintained by GSH reductase catalyzing the reduction of oxidized GSH, which forms the dimer glutathione disulfide (GSSG) and thus regenerates GSH. First, cellular levels of reduced and oxidized glutathione (GSH and GSSG, respectively) in roots of two independent PSY-overexpressing, apocarotenoid-accumulating Arabidopsis lines (At12 and At22) were determined. Interestingly, both lines had significantly increased levels of GSH and GSSG by a factor of 1.8 and 2.5, respectively, with the reduced and oxidized form increasing to a practically identical degree (Fig. 2A, B). Thus, the ratio of GSH and GSSG, a measure for the cell redox state, remained unchanged. The analysis of our recently obtained RNA-Seq data (Koschmieder et al. 2020) suggests that increased GSH accumulation is not due to transcriptional induction of the rate-limiting enzyme of GSH biosynthesis, glutamyl-cysteine ligase (At4g23100), or GSH synthase (At5g27380), with their transcript levels remaining unchanged. It must therefore rather be due to more complex regulatory processes in the metabolism of GSH, the supply of its precursor L-cysteine, which has also been considered rate limiting, or the interplay with the ascorbic acid–GSH cycle (Richman and Meister 1975; Hasanuzzaman et al. 2017). In conclusion, GSH increases in response to increased β-carotene and β-apocarotenoid accumulation and might allow sufficient detoxification of reactive compounds formed in subsequent catabolic processes.
Non-enzymatic glutathionylation does not contribute to β-carotene metabolism
Non-enzymatic glutathionylation is known to detoxify many compounds such as apocarotenoid citral, hydrophobic oxylipins and alkenes. This reaction even occurs at physiological pH in phosphate buffer and without catalysts in vitro and corresponding GSH adducts are often detectable in planta (Esterbauer et al. 1975; Davoine et al. 2005, 2006). We therefore investigated non-enzymatic glutathionylation of β-apocarotenoids ranging from C30 to C10 in vitro. High molar excess of GSH is known to strongly favor glutathionylation of reactive electrophiles (Esterbauer et al. 1975). Therefore, assays were conducted with GSH in up to 150-fold molar excess and at near-physiological concentrations of 10 mM for several hours in both phosphate buffer with neutral pH of 7.4 and Tris buffer with an alkaline pH of 9.0. We analyzed for possible reductions of the initial amounts of apocarotenoids as indicator for apocarotenoid–GSH adduct formation. However, the incubations did not reveal a decrease of free apocarotenoid levels at either pH in vitro (Table 1). In contrast, 4-hydroxynonenal, a fatty acid degradation product known to be detoxified by non-enzymatic and enzymatic glutathionylation (Esterbauer et al. 1975; Mano et al. 2019), was readily consumed at pH 7.4 and equimolarity with GSH within only 15 min (Table 1). Accordingly, there was no indication of β-apocarotenoids readily undergoing glutathionylation in vitro in the absence of catalysts.
Investigation on enzymatic apocarotenoid glutathionylation by GSTs
Next, we investigated whether apocarotenoid glutathionylation requires enzymatic catalysis by GSTs. Arabidopsis GSTs of the tau subfamily (GSTU) have recently been suggested to mediate detoxification of RCS by glutathionylation (Yamauchi et al. 2012; Mano et al. 2019), with the substrate specificity of many GSTUs still being unknown. Transcriptome analysis of PSY-overexpressing Arabidopsis roots revealed that ten GSTs were induced upon carotenoid accumulation, mostly by a factor of 3 and included GSTUs and GSTFs of mostly unknown substrate specificity (Supplemental Table S1). We determined the total GST activity in Arabidopsis roots using a colorimetric GST activity assay with the universal GST substrate 1-chloro-2,4-dinitrobenzene (CDNB). We found that the total GST activity is not increased (Fig. 3A).
To exclude that an only weak, specific apocarotenoid glutathionylation by GSTs activity is suppressed by interfering compounds in the plant extract, we decided to refine the analysis with a protein fraction enriched with GST enzymes. Given the multitude of Arabidopsis GSTs (Dixon and Edwards 2010), we did not aim at heterologous expression and in vitro testing for single GSTs, but decided to non-selectively co-purify endogenous GSTs of the tau and phi subfamilies from Arabidopsis leaves using immobilized GSH affinity chromatography (Edwards and Dixon 2005). The method allowed us to obtain approximately 250 µg of highly pure GSTs from about 100 g of Arabidopsis leaves (Fig. 3B) which showed a high specific activity to glutathionylate the control substrate CDNB with 5.3 nmol CDNB min−1 µg−1 in vitro. However, as indicated by unchanged substrate amounts after several hours of incubation at physiological pH of 7.4 with equimolar amounts of GSH (Table 2), in contrast to the control substrate, apocarotenoids (C30–C10) were not glutahionylated. Currently, we cannot exclude that our assay was not sufficiently sensitive to detect a weak apocarotenoid glutathionylation activity, as only the transcription of seven of the AtGSTs is upregulated in At12 and At22 (Supplemental Table S1). This remains to be explored using recombinant enzymes in future investigations.
Lastly, we investigated whether GSH adducts of apocarotenoids could be detected in At12 and At22 roots by comprehensive metabolite extraction followed by LC–MS analysis. Molecular ions corresponding to putative adducts of the apocarotenoids detected in the At12 and At22 lines (Koschmieder et al. 2020) were searched based on their predicted molecular formula, together with other GSH adducts described previously (Davoine et al. 2005). Besides OPDA-GSH, putative GSH adducts of β-apo-10-carotenoic acid, β-apo-11-carotenoic acid and β-apo-12-carotenol could be detected in all root extracts (Supplemental Figure S1). However, these putative adducts did not accumulate significantly more in the At12 and At22 lines than in the wild type, despite the high levels of apocarotenoid accumulation in the transgenic lines (Koschmieder et al. 2020).
Methylglyoxal and glyoxal as terminal oxidation products of β-carotene
Our experimental data show that GSH and GSSG levels rise in response to β-carotene and β-apocarotenoid accumulation. However, despite increased availability of GSH for potential detoxification processes related to carotenoid metabolism, we could not find any experimental evidence in vitro and in planta supporting non-enzymatic or enzymatic glutathionylation by GSTs for β-apocarotenoids ranging from C10 to C30.
We recently showed that Arabidopsis calli and roots overexpressing PSY and accumulating β-carotene and β-apocarotenoids also accumulate methylglyoxal and glyoxal (Schaub et al. 2018; Koschmieder et al. 2020), two reactive carbonyl species identified also as in vitro degradation products of β-carotene (Benevides et al. 2011; Mogg and Burton 2020). Direct investigations on whether these metabolites also originate from carotenoids in planta are hampered by the difficulties arising from cellular uptake of lipophilic (apo)carotenoids into plastids and/or the multitude of pathways sharing IPP as common precursor. Thus, traditional pulse-chase experiments as conceived to uncover, e.g., the plastid-localized MEP pathway, cannot be performed (Lichtenthaler et al. 1997). If not originating from apocarotenoid oxidation, a possible source of methylglyoxal and even more so of glyoxal is considered to be lipid peroxidation (Paudel et al. 2016). Lipid peroxidation is reported to occur also at high carotenoid levels and high oxygen partial pressure (Yanishlieva et al. 1998; McNulty et al. 2007). However, we previously reported that roots of Arabidopsis lines accumulating β-carotene and β-apocarotenoids do not accumulate lipid peroxidation intermediates or upregulate lipid stress-related genes (Koschmieder et al. 2020). It is therefore reasonable to conclude that accumulating methyglyoxal and glyoxal in At12 and At22 rather originate from continuous carotenoid and apocarotenoid turnover. One known degradation mechanism for the ionone rings of carotenoids is their oxidative cleavage into geronic acid (Burton et al. 2016; Schaub et al. 2017). The remaining linear carbohydrate backbone can theoretically be fully oxidized into methylglyoxal and glyoxal. Importantly, considering a complete degradation of the β-carotene backbone, equimolar amounts of both metabolites are generated.
Involvement of glyoxalase activities
In plants, methylglyoxal and glyoxal detoxification and metabolism strongly depend on cellular levels of reduced GSH and the GSH-dependent glyoxalase (GLX) pathway (Yadav et al. 2005b, a; Upadhyaya et al. 2011; Schmitz et al. 2017). In agreement, we observed increased cellular levels of reduced GSH in At12 and At22 roots, accumulating both methylglyoxal and glyoxal requiring detoxification (Fig. 2). In the glyoxalase pathway, glyoxalase I (GLXI) converts methylglyoxal and GSH to S-D-lactoylglutathione which is converted by glyoxalase II (GLXII) to D-lactate, liberating GSH. Glyoxalase III (GLXIII) enzymes directly convert methylglyoxal to D-lactate in a single step, without using GSH or any other cofactor. Moreover, D-lactate dehydrogenase (D-LDH) is also associated with the glyoxalase system and metabolizes D-lactate into pyruvate which enters into the TCA cycle (Engqvist et al. 2015; Welchen et al. 2016).
Based on previous findings, glyoxalase GLXI activity is rate-limiting for the entire glyoxalase pathway and directly proportional to the GSH concentration available to the enzyme (Lages et al. 2012; Rabbani and Thornalley 2014; Nigro et al. 2017). In view of this, our findings suggest that increased GSH levels in At12 and At22 roots probably increase glyoxalase pathway activity and flux, allowing for increased detoxification of apocarotenoid-derived methyglyoxal and glyoxal (Fig. 1).
Additionally, we investigated upregulation of the glyoxalase GLX pathway and GLX-like at the transcript and protein level. Interestingly, the expression of 2 of the 11 known GLXI and GLXI-like genes in Arabidopsis (Schmitz et al. 2017, 2018) were upregulated in At12 and At22; among the strongest GLXI-like;7 (At1g80160) by factor 29 and 10, respectively, and GLXI-like;4 (At1g15380) by factor 4 and 3, respectively (Table 3). Notably, however, GLXI-like;4 and GLXI-like;7 are considered non-functional glyoxalases, although a final confirmation with recombinant proteins is still required (Schmitz et al. 2018). Genes annotated as GLXII were not found to be differentially regulated in PSY-overexpressing roots. Interestingly, the expression of GLXIII (At3g02720) encoding an enzyme with highest in vitro activity on methylglyoxal and glyoxal among all GLXIII enzymes (Kwon et al. 2013) was not upregulated, while the expression of GLXIII (At3g54600), encoding an enzyme of low in vitro activity on methylglyoxal, was upregulated by factor 2 and 4 in A12 and At22, respectively (Table 3).
Therefore, transcriptome analysis in AtPSY-expressing roots revealed no evidence for transcriptional induction of the canonical glyoxalase GLX pathway, rather pointing toward induction of GLXIII for one-step conversion of the cytotoxic metabolites. In line with this, we found that overall activity of GLXI, the rate-limiting enzyme of the pathway in plants and in mammals (Rabbani and Thornalley 2014), remained unchanged in A12 and At22 roots (Fig. 4). We propose that in addition to GLXIII-mediated methylglyoxal detoxification, increased availability of both substrates—GSH and apocarotenoid-derived methyglyoxal and glyoxal in A12 and At22 roots—is likely sufficient to increase flux through the glyoxalase system without the need for induction of the rate-limiting enzyme GLXI at the transcript or enzyme level.
Metabolomic analysis indicates methylglyoxal and glyoxal detoxification
Given the indications for an enhanced detoxification of apocarotenoid-derived methylglyoxal and glyoxal, we decided to perform a metabolomic analysis in roots of hydroponically grown plants. Relative amounts of a set of 71 metabolites were quantified in wild-type and transgenic lines by GC–MS after derivatization (Lisec et al. 2006) and differential metabolite accumulation among lines was tested (Supplementary Figure S2). Among the metabolites analyzed, we focused on those which were significantly different from wild-type roots grown in parallel and responded consensually in both carotenoid-accumulating lines, thus either increasing or decreasing metabolite levels compared with wild-type roots (Fig. 5).
Interestingly, we found evidence for the detoxification of both apocarotenoid-derived breakdown products glyoxal and methylglyoxal. Firstly, D-lactate levels increased by twofold in both carotenoid-accumulating lines compared to wild-type levels, suggesting its origin from apocarotenoid-derived methylglyoxal. Although mitochondrial D-LDH was not induced transcriptionally in neither At12 nor At22, we suggest that the capacity is sufficient to convert D-lactate into pyruvate, also considering that other metabolic pathways to detoxify D-lactate are not known in plants (Maurino and Engqvist 2015; Welchen et al. 2016). Levels of pyruvate were similarly unchanged, however, especially assuming an increased demand for plastid-derived IPP for carotenoid biosynthesis in PSY-overexpressing roots. We consider that methylglyoxal-derived pyruvate might readily be used for IPP biosynthesis via the MEP pathway, thus explaining its constant amounts like in wild-type roots.
In addition to D-lactate which is the conversion product of methylglyoxal, we found glyoxylate which is likely derived from glyoxal to be increased by twofold in apocarotenoid-accumulating lines. The formation of glyoxylate from glyoxal involves the intermediate glycolate which was unchanged in apocarotenoid-accumulating roots and requires the activities of glycolate oxidase (GOX; converting glycolate to glyoxylate) and the glycolate transporter PLGG1 (At1g32080; Fig. 6). These two proteins are involved in photorespiration in photosynthetically active tissues (Nunes-Nesi et al. 2014; Maurino and Engqvist 2015; Modde et al. 2016). However, photorespiratory enzymes are assumed to fulfill other metabolic requirements in heterotrophic tissues (Engqvist et al. 2015; Schmitz et al. 2020), such as roots which are investigated in this work. As a confirmation, we found transcriptional upregulation of some genes encoding enzymes involved in photorespiration, although transcript levels in roots are very low compared to green tissues (Table 3). Remarkably, the plastidal glycolate/glycerate translocator 1 PLGG1, which exports glycolate from plastids into peroxisomes (Pick et al. 2013), was induced 11-fold. Moreover, the peroxisomal enzymes glycolate oxidase GOX1 (At3g14420), converting glycolate into glyoxylate, and serine:glutamate aminotransferase SGAT (At2g13360) were induced four- and eightfold, respectively (Table 3). Genes encoding proteins involved in photorespiration, namely GOX2 (At3g14415), GDC-T (At1g11860), GDC-P (At4g33010), SHM1 (At4g37930), HPR1 (At1g68010) and GLYK (At1g80380), were not considerably induced. These findings agree with recent reports on alternative functions of enzymes linked with photorespiration in heterotrophic tissues (Nunes-Nesi et al. 2014; Engqvist et al. 2015; Schmitz et al. 2020). In summary, genes coding for enzymes involved in the steps involving glycolate/glyoxylate metabolism were considerably upregulated. This might represent an alternative biological function of those activities, here in non-photosynthetic tissues, allowing for increased metabolism of apocarotenoid-derived glyoxal ultimately allowing for carbon recycling via primary pathways (Fig. 6).
Similarly, the level of malate which is the product of glyoxylate metabolization catalyzed by malate synthase remained uchanged, while the transcript levels of malate synthase were strongly increased 13-fold in At12 and At22. Likewise, malate dehydrogenase, which converts malate into oxaloacetate, was induced by threefold. Interestingly, and in support of the involvement of malate entry of glyoxal-derived carbon, transcripts of all other enzymes of the tricarboxylic acid cycle (TCA) remained unchanged. Oxaloacetate might further be converted into phosphoenolpyruvate, which is supported by increased expression levels of the enzymes phosphoenolpyruvate carboxykinase and phosphoenolpyruvate carboxykinase 2 (three- and sixfold increases in both lines). Finally, phosphoenolpyruvate can be converted into pyruvate by pyruvate kinase; however, the expression levels of the corresponding enzyme remained unchanged.
Other metabolites induced in apocarotenoid-accumulating roots
Remarkably, apart from the metabolites of primary carbon metabolic pathways, the levels of a number of amino acids were found to be changed in response to apocarotenoid accumulation. Serine and glycine, substrate and product of the induced photorespiratory SGAT enzyme, respectively, were unchanged, while other amino acids like β-alanine, asparagine, glutamine, lysin, phenylalanine and threonine increased up to twofold in abundance. Interestingly, the same amino acids were increased in fruits of tomato plants with increased carotenoid levels upon overexpression of the fruit-specific phytoene synthase (Fraser et al. 2007). Although tomatoes accumulate mainly linear lycopene and not cyclic β-carotene like the Arabidopsis roots investigated in this work, oxidative breakdown of the common carotenoid backbone most likely occurs with similar end products methylglyoxal and glyoxal. This might explain similar responses of primary pathway metabolites including amino acids. Possibly, a number of amino acids can be formed by interconversions mediated by amino transferases. The similarities between lycopene and β-carotene-induced alterations in amino acid abundances suggest similar balancing metabolic responses toward altered primary carbon metabolism which requires further investigations.
Finally, several intermediates of the glycolytic pathway were present in higher abundance, such as fructose-6-P and glucose-6-P increased up to threefold in apocarotenoid-accumulating roots, while glucose levels were slightly reduced. In agreement with this, several enzymes of the pathway were induced: fructose-1,6-bisphosphatase (twofold), aldolase (threefold), glycerinaldehyde-3-phosphate dehydrogenases (fourfold and twofold, respectively) and phosphoglycerate mutases (up to sevenfold; Table 3). The reduced levels of glucose in carotenoid-accumulating roots and the induction of enzymes, indicative of glycolysis (Giegé et al. 2003), suggests that these processes are required to mobilize carbon from glucose to fuel the high carotenoid biosynthesis activity introduced by the transgenic expression of the rate-limiting enzyme PSY.