Soybean (Glycine max cv. Primus obtained from Die SAAT Austria) seeds were surface-sterilized with 5% sodium hypochlorite solution and sown into acid-washed pots containing an inert substrate mixture of vermiculite and perlite in a 5/3 (v/v) ratio to exclude contamination. Half of the seeds (N fix) were inoculated with 1:10 diluted commercial Bradyrhizobium japonicum inoculant (Radicin Soja Die SAAT Austria) and received 0.25 mM KNO₃ for only 2 weeks; the other half served as non-symbiotic control (N fed) and was supplied with 10 mM KNO₃. Seedlings were watered every other day with a modified half strength N-free Hoagland nutrient solution, pH 7.2 (Hoagland and Arnon, 1950). After 3 weeks, when nodules were formed and symbiosis was fully established, 0.5 mM tungstate (Na₂WO₄) was added to the nutrient solution. To investigate the effects of tungsten on N-fertilized and symbiotically grown soy plants, a zero tungsten control was realized for both N regimes. The plants were altogether grown in eight biological replicates, four of which were used for proteomic and metabolomic analysis and four for the assessment of physiological parameters and analysis of macro- and micronutrients. The plants were grown under controlled glasshouse conditions with a 12/12 light/dark cycle at 29/21°C and an approximate photosynthetic radiation of 400 µmol m⁻¹s⁻¹ at canopy level. After 2 weeks of tungsten exposure, chlorophyll fluorescence measurements were performed, and the plants were sampled. The roots, nodules, and leaves of four biological replicates were cut off using a stainless-steel razorblade, transferred into precooled 2-mL Eppendorf tubes, immediately put on liquid nitrogen (LN₂) for metabolomic and proteomic analysis, and stored at −80°C. The remaining four replicates were harvested for acid digestion and ICP-OES analysis.

Chlorophyll fluorescence was measured on individual leaves with a mini-pam II/B (Heinz Walz GmbH). The plants were allowed to adapt in the dark for 15 min and then subjected to a rapid light curve (RLC) with increasing photosynthetic active radiation (PAR) intensities (0, 25, 45, 65, 90, 125, 191, 287, 422, 633, 827, 1,157, and 1,507 μmol m⁻² s⁻¹) in increments of 20 s each. The calculation of fluorescence ratio parameters was done with WinControl -3 software (Heinz Walz GmbH) provided by the manufacturers of the pulse-amplitude modulated (PAM) fluorometer.

After harvest, the roots and nodules were washed and ultra-sonicated with CaCl₂ (0.01 M) for 5 min followed by a further 3 min of ultra-sonication with HQ-water. The above-ground biomass was rinsed with deionized water and patted dry. The roots, shoots, leaves, and nodules were separated and dried at 60°C for 1 week and subsequently ground to a homogenous fine powder using a Retsch mill (Retsch MM20). Then, approximately 200 mg of dry, ground plant material was digested in a matrix containing 5 mL HNO₃, 1 mL H₂O₂, and one drop of 1-octanol. For biomass less than 100 mg, the amount of acid and peroxide was reduced by 50%. The digestion was performed using an open digestion unit (Velp Scientifica). The volume of digests was made up to 50 mL/25 mL with HQ-water resulting in ∼6.5% HNO₃. The nutrients and tungsten concentrations in the digests were analyzed using inductively coupled plasma optical emission spectroscopy (ICP-OES Optima 3000 XL Perkin Elmer). The translocation factors (TF) for individual nutrient and tungsten were calculated by dividing the tissue concentrations of shoots by the concentrations in the root.

Frozen plant material was ground to a fine powder on LN₂ using a porcelain mortar. A total of 60 mg (fresh weight) was then used for metabolite and protein extraction for spectrophotometric and mass spectrometric analyses using an adapted sequential extraction procedure by Laware (2015). The extraction procedure for photosynthetic pigments and primary and secondary metabolites is briefly described below. A detailed description of the extraction procedure as well as the protocols for integrative estimation of key plant metabolites can be found in the Supplements (Supplementary S1_1).

For statistical analysis, data was only used if values were present in three or more replicates of at least one treatment. If less than half of the values were missing in one treatment, the missing values were imputed by the k-nearest neighbor algorithm; if more values were missing, the values were imputed with the smallest value detected throughout all samples divided by two. Two samples of one treatment (tungsten N fix root/leaf) were lost during sample processing (metabolite extraction) and thus were not imputed as they were considered systemically missing (n = 3 to 4). The statistical analysis of data from the sequential extraction protocol, starch assay, GC–MS, and ICP-OES as well as the physiological data was performed using Infostat software (InfoStat, RRID : SCR_014310) (Di Rienzo et al., 2012). Data was analyzed by using a linear mixed model and DGC post hoc test. If necessary, heteroscedasticity was corrected by selecting the varIdent model in Infostat software.

For the statistical analysis of the protein data, an analysis of variance (ANOVA) followed by a Tukey HSD post hoc test was performed using R-studio. The LFQ intensities of each treatment were averaged, and the ratios between control and high tungsten treatments were calculated. Proteins were considered significantly changed when the change (ratio) between the compared treatments was higher or at least twofold as well as with a p-value below 0.05. P-value correction was performed using Benjamini–Hochberg to account for false positives due to multiple testing. Adjusted P-values are provided in the summary tables in the Supplement (Supplementary Tables S5, S6).

Additionally, a principal component analysis (PCA) was performed with R-studio to filter for the effect size of individual variables. For the principal component analysis, a new data matrix was created for roots as well as leaves. Each matrix contained all metabolites and significantly changed proteins that were detected in at least three out of four replicates of at least one treatment. The PCA was performed using prcomp-package (principal component analysis, RRID : SCR_014676) using log²-transformed data. Heat maps were created using the R package gplots. Euclidian distance with complete linkage was used for clustering. The PCA loadings for proteins can be found in Supplementary Tables S5, S6. The loadings for metabolites are provided in Supplementary Tables S3, S4. GO enrichment analysis as well as annotation for molecular function and molecular process was performed with the Panther 18.0 analysis tool using the PANTHER overrepresentation test (Released 20231017) and GO Ontology database (DOI: 10.5281/zenodo.8436609, released 2023-10-09) for Glycine max (Ashburner et al., 2000; Thomas et al., 2022; Aleksander et al., 2023). As statistical test, Fisher’s exact was chosen, and false discovery rate (FDR) was calculated for significant enrichments (Supplementary Table S7).

The principal component analysis (PCA) of both leaves and roots shows a clear separation between the different treatments. PC1 to PC2 contribute most to the difference between the treatments and thus were considered for further analysis (Figure 3). In leaves, PC1 accounts for 47.32% and PC2 for 25.24% of the difference between the treatments (Figure 3A). While PC1 represents the difference between controls and tungsten-treated plants, PC2 represents the difference between the N fix control and the nitrogen-fertilized plants. Furthermore, tungsten-stressed N fix plants and controls are separated on PC2.

In roots, PC1 is responsible for 51.17% of the variance and clearly separates the tungsten-stressed plants and controls (Figure 3B). PC2 explains another 14.20% of the variance and shows the difference between the two N regimes. To identify the proteins and metabolites responsible for the separation, a threshold with the largest loadings was set at ±0.065 (Supplementary Figure S2).

Heavy metal ions are known to cause oxidative damage by disrupting cellular processes and enzymatic activity, binding of glutathione and -SH groups of proteins, and displacing nutrients from binding sites as well as partaking in Fenton or Fenton-like reactions (Sharma and Dietz, 2009; Hossain and Komatsu, 2012). This has detrimental effects on cellular redox homeostasis, leads to lipid peroxidation, and disrupts membrane integrity as well as damages proteins and DNA (Michalak, 2006). We have previously found evidence that, similarly to other heavy metals, tungsten interferes with oxidative homeostasis and leads to an increased production of reactive oxygen species (Preiner et al., 2019). Plants combat oxidative stress by keeping the mitochondrial electron transport chain sufficiently oxidized, enhancing the biosynthesis of antioxidative enzymes such as catalase (CAT), superoxide dismutase (SOD), ascorbate peroxidase (APX), peroxidase (PX), and lipoxygenase (LOX), as well as by increasing the production of non-enzymatic antioxidative metabolites such as phenols, flavonoids, and organic acids (Møller, 2001; Gratão et al., 2005).

In this study, we found further evidence that tungsten exposure leads to oxidative stress and that tungsten stress response involves a significant readjustment of the antioxidative system. The GO enrichment analysis revealed that proteins involved in response to oxidative stress (GO:0006979) as well as oxygen species metabolic processes (GO:0072593) were among the proteins contributing most to the difference between the controls and tungsten-treated plants as well as between the different N regimes (Supplementary Table S7). Additionally, the leaves of N fix plants showed significantly elevated levels of peroxidases, glutathione peroxidase, glutathione reductase, superoxide dismutase (Cu–Zn), and H-type thioredoxin (Supplementary Tables S5, S6) as well as a markedly lower IC50 value (Figure 2). This suggests an enhanced antioxidative capacity of symbiotically grown plants during tungsten exposure. Consistent with this, our data clearly shows that symbiotically grown soy plants show a greater capacity to actively synthesize certain primary and secondary metabolites such as gluconic acid, polyamines, phenolic compounds (flavonoids and tannins elevated but not significant), and proline as well other amino acids to mitigate the deleterious effects of tungsten on plant redox homeostasis (Figures 2, 4). Additionally, our findings indicate that these protective compounds play a role in the chelation and reduction of root-to-shoot trafficking of tungsten, as translocation to leaves was significantly higher in N fed plants compared to that in N fix plants (Table 2).

In leaves, phenolic compounds seem to play a key role in the observed symbiont-specific tungsten stress response (Figure 2). On the one hand, phenolic compounds serve as important non-enzymatic radical scavengers by directly trapping reactive oxygen and nitrogen species and acting as electron donors to peroxidases, mediating the formation of less reactive phenoxyl radicals (Sakihama et al., 2000; Winkel-Shirley, 2002; Maksymiec, 2007; Sytar et al., 2013). On the other hand, flavonoids have been shown to be involved in the chelation and vacuolar sequestration of heavy metals, potentially reducing the toxic effects on cellular metabolism and redox homeostasis. Hale et al. (2001, 2002) showed that anthocyanins are able to chelate not only molybdenum but also tungsten by forming complexes and thus facilitate tungsten accumulation in brassica. Additionally, Xu et al. (2018) found that increased concentrations of flavonoids (quercetin, glycitein, trans-resveratrol, p-coumaric acid, and glycinol) contribute to tolerance to excess molybdenum (100 mg L^-1) in 3-week-old soybean seedlings.

Apart from phenolic compounds, the principal component analysis revealed a possible role of polyamines in tungsten stress response, as they were among the metabolites with the highest loadings (PC1) separating tungsten-treated N fix plants from all the other treatments (Figure 4). Polyamines are low-molecular-weight polycationic amines that are thought to be involved in an array of diverse metabolic processes such as stabilizing membranes and nucleic acids, modulating enzyme activity and scavenging free radicals as well as regulating development, plant growth, protein biosynthesis, and cell division (Walters, 2000; Menéndez et al., 2019; Hidalgo-Castellanos et al., 2022). Although the contribution of polyamines to abiotic stress response in plants is only partially understood, their levels and the activities of enzymes involved in their biosynthesis have been observed to be enhanced in response to various types of stress (Evans and Malmberg, 1989). As comprehensively reviewed by Menéndez et al. (2019), in the context of heavy metal stress, polyamines might play a role in protecting membrane systems and cellular processes from oxidative damage induced by redox-active metal ions. Interestingly, the exogenous application of spermidine to mung bean led to reduced uptake and translocation of Cd as well an increase of antioxidant enzyme activity and levels of non-enzymatic radical scavengers, which effectively prevented oxidative damage and chlorophyll degradation (Nahar et al., 2016; Menéndez et al., 2019). Similarly, the observed increase of putrescine and spermidine concentrations in the leaves of tungsten-stressed N fix plants (Figure 4) might help to reduce oxidative stress and stabilize membrane systems and cellular processes.

While changes in phenol and polyamine levels were primarily found in leaf tissue (Figure 4), we observed significant alterations in the abundance of certain enzymes involved in the glutathione–ascorbate cycle (glutathione reductase, glutathione peroxidase, and glutathione S-transferase) and of certain antioxidative enzymes (peroxidases and catalases) in the roots of N fix plants (Supplementary Table S6). This could be the result of an adaptive response to bacterial symbiosis, as reactive oxygen species and the plant’s antioxidative defense system play a crucial role in nodule formation as well as the establishment and functioning of symbiosis (Stambulska et al., 2018). As nodules are thought to mitigate stress by serving as an important sink for heavy metals, an effective antioxidative system is necessary to protect nitrogenase from oxidative damage (Stambulska et al., 2018). In addition to this enzyme-mediated antioxidative response, gluconic acid was significantly increased in the roots of tungsten-stressed plants of both N regimes (Figure 4). However, this accumulation was again more prominent in the roots of symbiotically grown plants (50-fold increase) compared to the N fed treatment (Figure 4; Supplementary Table S4). The inoculation with gluconic acid-producing bacteria has been shown to enhance plant growth and restore root elongation of cadmium (Cd²⁺)-stressed plants (Kavita et al., 2008). Furthermore, it improves the bioavailability and uptake of phosphate and certain trace elements while alleviating the symptoms of heavy metal toxicity by forming stable complexes (Stella and Halimi, 2015; Mandal et al., 2016; Wood et al., 2016). Research by Xu et al. (2018) found more than 107-fold increase in gluconic acid concentrations in the roots of soybean plants (non-symbiotic) exposed to excess molybdenum. They hypothesize that, together with other organic acids, gluconic acid reduces molybdenum toxicity by chelating molybdenum and thus hindering root-to-shoot trafficking of the transition element. Interestingly, gluconic acid has been found to form stable complexes not only with molybdenum (VI) but also tungsten (VI) in aqueous solutions (Llopis et al., 1986; Ramos et al., 1997).

The idea that tungsten uptake and translocation within the plant is affected by protective compounds such as polyamines and gluconic acid is supported by our results. Tungsten-stressed N fix plants, which showed the highest fold changes of these metabolites, exhibited a significantly lower root-to-shoot translocation of tungsten (Table 2). Additionally, our proteomic data revealed significant alterations of cell wall metabolism in the roots to facilitate tungsten sequestration and mitigate the deleterious effects of heavy metal stress. Two proteins with some of the highest loadings of PC1, Berberine bridge enzyme-like proteins 8/28 (I1MG80 and I1K2W8), are interesting in this respect (Figure 5, Table 3B). These enzymes belong to the FAD-linked oxidase superfamily and are involved in the inactivation of oligogalacturonides, which are cell wall fragments that serve as elicitors of defense response and seem to regulate plant growth (auxin antagonist) and cell wall formation/degradation (Ferrari et al., 2013). Oligogalacturonide (OG) accumulation as well as external application has been found to result in reduced growth, disturbed development, necrosis of tissue, and, eventually, plant death (Chen et al., 2022). Berberine bridge enzyme-like proteins have been found to oxidize OGs and thus protect cell walls from degradation and maintain plant growth, as the H₂O₂ formed during OG oxidation can be used by peroxidases to reinforce cell walls (Chen et al., 2022; Scortica et al., 2023). Interestingly, peroxidases were also among the proteins with the highest loadings in PC1 (I1J7M4 and I1JKZ9) (Figure 5; Table 3B). Our results indicate that the immobilization of tungsten in the root system, together with the observed changes in the antioxidative system as well as the enhanced production of protective compounds (polyamines, gluconic acid, and phenols), potentially resulted in reduced oxidative damage in N fix plants (Figure 6).

Heavy metal stress interferes with plant metabolic pathways on various levels. One hallmark of heavy metal exposure is the synthesis and accumulation of various N-containing metabolites such as small peptides, amino acids, and polyamines with the potential to mitigate the heavy metal toxicity (Sharma and Dietz, 2006). Among various N-containing protective metabolites that showed significantly increased concentrations in the tungsten-stressed N fix plants, proline had the highest loadings (PC1) and highest fold change (100-fold in leaves and 500-fold in roots) (Figure 4). The accumulation of proline is well documented in response to various types of biotic and abiotic stresses including pathogen attack, salinity, drought, flooding, and heavy metal stress (Spormann et al., 2023). Similarly, a variety of roles such as antioxidative, osmoregulatory, or signaling function have been attributed to proline during stress (Spormann et al., 2023). High constitutive proline contents found in taxonomically distant metal-tolerant plant species suggests a function of proline in metal stress tolerance (Schat et al., 1997; Sharma and Dietz, 2006; Szepesi and Szőllősi, 2018). It is thought to play a role in stabilizing cellular homeostasis, providing reducing equivalents for mitochondrial respiration and protecting enzymatic function during stress as well as binding and chelating heavy metals (Hare and Cress, 1997; Sharma and Dietz, 2006; Szabados and Savouré, 2010; Szepesi and Szőllősi, 2018). Proline accumulation during stress is thought to be driven by proteolysis and increased synthesis as well as reduced catabolism (Szabados and Savouré, 2010; Spormann et al., 2023). We not only found a pronounced increase in proline levels but we also observed significantly higher levels of P5CS, a central enzyme of proline biosynthesis, in N fix roots (Supplementary Table S6). Together with higher levels of proline precursors such as glutamate and ornithine (Supplementary Table S4B), this indicates an enhanced proline biosynthesis in symbiotically grown soy plants. Although the proline levels were also increased in N fed plants, the magnitude of the observed increase of proline concentrations in N fix plants (more than 100-fold) points to a specific symbiosis-induced response (Figure 4; Supplementary Tables S3B, S4B). Interestingly, Selim et al. (2021) found that inoculation with actinobacteria ameliorates the adverse effects of tungsten nanoparticles on legumes (alfalfa, pea, and soybean) by the accumulation of proline, soluble sugars, and polyamines.

To ensure a sufficient N supply for the biosynthesis of proteins and protective N compounds such as proline and polyamines, N assimilation and subsequent metabolic pathways play a critical role in heavy metal tolerance and stress response (Sharma and Dietz, 2006). In the context of tungsten stress, this seems especially interesting as it has been shown that a major component of tungsten toxicity in plants is the inhibition of nitrogen metabolic processes by substitution of the molybdenum ion bound to the enzymatic co-factor of nitrate reductase (Harper and Nicholas, 1978; Deng et al., 1989). The picture seems not to be as clear for the Fe–Mo cofactor containing bacterial nitrogenase utilized by symbiotic rhizobia. Some studies have shown that bacterial nitrogenase remains functional upon tungsten exposure (Kletzin and Adams, 1996; Schwarz et al., 2007; Oburger et al., 2018; Preiner et al., 2019).

In this experiment, we therefore exposed plants to tungsten only after the establishment of rhizobium symbiosis, when nodules were fully developed. Although tungsten clearly affected plant growth, photosynthesis, and biomass (Table 1), we found no effect on nodule number (Table 1). Together with our previous findings (Oburger et al., 2018; Preiner et al., 2019) as well as the increased concentrations of nitrogen compounds such as ureides (allantoin and allantoic acid) and amino acids, this can be seen as further evidence for maintained nitrogen fixation activity during tungsten exposure (Figure 4; Supplementary Tables S3, S4). As discussed above, we also found a systemic metabolic readjustment clearly connected to the symbiosis with N₂-fixing bacteria, which enabled the plants to synthesize a variety of protective metabolites such as amino acids and polyamines as well as phenolic compounds (Figures 2, 4).

The sustained or even increased concentrations of allantoin and allantoic acid as well as the amino acids glutamine (gln) and glutamic acid (glu) in the roots and leaves of tungsten-stressed N fix plants (Figure 4) suggest that sufficient nitrogen (N) for the production of amino acids and protective compounds is provided by either N₂ fixation or a significant reduction of ureide catabolism (Ohyama et al., 2017). In ureidic legumes such as soybeans, ammonia from N₂ fixation is rapidly assimilated into glutamine and subsequently to glutamate by glutamate synthetase (GOGAT); the fixed nitrogen is then used for de novo synthesis of purine nucleotides, which then is metabolized via uric acid to allantoin and allantoic acid, which represent the main forms of transport of fixed nitrogen from root nodules to leaf tissue (Ohyama et al., 2017; Ono et al., 2021). Indeed we found key enzymes involved in ureide biosynthesis (glutamate dehydrogenase, glutamate synthetase, glutamine synthetase, uricase, hydroxyisourate hydrolase, and allantoinase) as well as ureide catabolism (urease and ureidoglycolate amidohydrolase) to be significantly more highly abundant in the roots of tungsten-treated N fix plants compared to the control (Supplementary Table S6). Additionally, no significant reduction of ureide catabolic enzymes was observed in the leaves of N fix plants, where the highest levels of free amino acids and ureides were measured. Together with the reduction of several TCA cycle organic acids in leaves (Supplementary Tables S3, S4), this suggests that the ureides were transported from the root nodules to the sink organs where they were used for amino acid biosynthesis. The notion that the observed metabolic changes are a result of maintained N₂ fixation rather than ureide catabolism is also supported by our previous results, which showed that tungsten does not directly affect N₂ fixation rate (activity per nodule) but rather hampers nodule formation and development (Oburger et al., 2018; Preiner et al., 2019).

The increase in asparagine (asn), glutamic acid (glu), and glutamine (gln) concentrations (Supplementary Tables S3, S4) in N fix plants also indicates enhanced protein turnover and recycling of amino acids as well as hints at the function of these amino acids as transient N storage (Carter and Tegeder, 2016; Ohyama et al., 2017). This is supported by the fact that despite significantly higher levels of proteolytic enzymes as well as protease inhibitors in leaves (Table 3), the overall protein concentrations remained comparable in tungsten-stressed N fix plants (Table 1). Together with the increase in abundance of numerous proteins, especially of proteins with stress-responsive, antioxidative, and enzymatic function in N fix plants upon stress (Supplementary Tables S5, S6), this indicates that the significant amino acid accumulation is a result of an overall metabolic readjustment of N fix plants upon tungsten stress.

In N fed plants, conversely, the lower levels of nitrogen compounds (allantoin, allantoic acid, and asparagine) (Supplementary Tables S3, S4) as well as the depletion of enzymes involved in nitrate assimilation (nitrite reductase) (Supplementary Table S6) in the tungsten treatment show the direct inhibitory effect of tungsten on nitrate reduction. This is furthermore supported by previous research where the direct inhibitory action of tungsten on nitrate reductase activity was shown (Nicholas et al., 1976; Deng et al., 1989; Oburger et al., 2018; Preiner et al., 2019).

Taken together, our data supports the idea that the increase in amino acids and protective N-metabolites is fueled by ureide de novo biosynthesis and consequently N₂ fixation by associated rhizobia rather than by a reduction of ureide catabolism or enhanced proteolytic activity. Nitrogen-fertilized plants (N fed) conversely did not possess the capacity to respond to tungsten-induced oxidative stress in the same way that the assimilation of nitrogen via nitrate reduction was inhibited by tungsten (Figure 6).

Carbohydrate metabolism is highly sensitive to environmental stress. Heavy metals are known to affect carbohydrate metabolism by disrupting photosynthesis, source-to-sink transport, and enzyme activities as well as respiration and thus have a detrimental effect on plant growth, seed development, and productivity (Samarakoon and Rauser, 1979; Viehweger, 2014).

Although no changes in photosynthetic pigment contents were observed (Supplementary Table S1A), our results clearly show that tungsten impairs the photosynthetic activity and efficiency of PSII (Supplementary Figure S1; Supplementary Table S1B). Additionally, it led to a reduction of proteins of the photosynthetic apparatus in both N regimes (Supplementary Table S5). Interestingly, symbiotically grown tungsten-stressed plants exhibited a distinct metabolic response compared to their N fed counterparts (Figure 6). Despite a reduced photosynthetic activity, we found significantly higher levels of soluble carbohydrates (glucose, fructose, galactose, and raffinose) and a notable reduction of starch contents in the roots and leaves of tungsten-stressed N fix plants (Figure 4; Supplementary Tables S3A, S4A). In N fed plants, this increase was less pronounced, and the starch concentrations were not affected or even slightly elevated (Supplementary Tables S3, S4). Our proteomic analysis revealed that in the leaves of tungsten-stressed N fix plants, granule-bound starch synthase (A1YZE0) was markedly reduced and among the proteins with the highest negative loadings in PC1 (Figure 4). Furthermore, alpha- and beta-amylases and the levels of sucrose synthase were significantly increased in the leaves and roots of tungsten-stressed N fix plants (Supplementary Table S5). Interestingly, a reduction of sucrose synthase activity has been shown to be responsible for the inhibition of nitrogen fixation during drought in Medicago truncatula (Larrainzar et al., 2007). The instance that we found a significantly increased abundance of sucrose synthase in tungsten-stressed N fix plants indicates that the carbohydrate allocation found might also serve to “feed” bacterial symbionts and again points to a maintenance of N₂ fixation. The reduced starch content in tungsten-stressed N fix plants partially contradicts our previous results (Preiner et al., 2019), where we observed increased starch concentrations in whole shoot tissue of both N regimes. However, it is important to note that, in this study, only leaf tissue was used for starch content measurement. Overall, our findings suggest that the suppression of starch synthesis, together with enhanced sucrose synthesis, resulted in the accumulation of soluble carbohydrates in N fix plants. Starch depletion, paired with an increase of soluble carbohydrates during the day, when usually transitory starch concentrations increase due to photosynthetic activity, has been observed in plants growing under various adverse conditions (Thalmann and Santelia, 2017). When photosynthetic activity is reduced by environmental constraints, starch is degraded in order to ensure a supply of soluble carbohydrate to maintain cellular respiration and water potential and to protect metabolic processes from oxidative stress (Thalmann et al., 2016). Additionally, certain sugars such as sucrose might also indirectly enhance the antioxidative capacity of plants by acting as secondary messengers that influence the photosynthetic activity as well as the expression of radical scavengers such as anthocyanins (Van den Ende and El-Esawe, 2014).

Apart from their central role as energy source and osmolyte as well as structural component of plant cells, soluble carbohydrates also serve as the carbon backbone of secondary metabolites and protective compounds such as phenols, polyamines, and amino acids (Jeandet et al., 2022). The evident symbiont-specific investment in the synthesis of amino acids and protective compounds thus depends not only on a constant supply of N but also on the significant incorporation of carbon. Our results clearly show that N fix plants underwent a significant metabolic readjustment to provide carbon sources for nitrogen fixation as well as for the biosynthesis of secondary metabolites and protective compounds (Figure 6). We believe that the increased investment in stress alleviation, together with reduced photosynthesis rates, most likely resulted in the mobilization of carbon storage pools in N fix plants in an effort to maintain cellular respiration and osmotic balance. A similar interplay of transitory starch, soluble carbohydrates, and protective compounds such as proline has been reported in drought-stressed Arabidopsis (Zanella et al., 2016).