Adaptive modifications in plant sulfur metabolism over evolutionary time

Abstract

Sulfur (S) is an essential element for life on Earth. Plants are able to take up and utilize sulfate (SO₄^2–), the most oxidized inorganic form of S compounds on Earth, through the reductive S assimilatory pathway that couples with photosynthetic energy conversion. Organic S compounds are subsequently synthesized in plants and made accessible to animals, primarily as the amino acid methionine. Thus, plant S metabolism clearly has nutritional importance in the global food chain. S metabolites may be part of redox regulation and drivers of essential metabolic pathways as cofactors and prosthetic groups, such as Fe–S centers, CoA, thiamine, and lipoic acid. The evolution of the S metabolic pathways and enzymes reflects the critical importance of functional innovation and diversifications. Here we review the major evolutionary alterations that took place in S metabolism across different scales and outline research directions that may take advantage of understanding the evolutionary adaptations.

This review delves into major evolutionary alterations that shaped and diversified sulfur metabolism across the domains of life, highlighting the speciation that occurred in different plant taxa.

Introduction

Sulfate (SO₄^2–) is the main source of sulfur (S) utilized by plants. It is taken up from soil into the root system and subsequently distributed to various plant organs and cellular/subcellular compartments that may play specialized roles in plant development and protection. Root uptake of SO₄^2– from soil and internal SO₄^2– distribution largely depend on functions of the membrane-bound sulfate transporters (SULTRs) (Takahashi et al., 2011; Maruyama-Nakashita, 2017). While SO₄^2– can be temporarily stored in the vacuoles, depending on cellular demands for S, it may be mobilized to serve as the substrate for synthesis of S-containing metabolites in the cytosol, plastids, and mitochondria (Fig. 1).

Fig. 1.

Schematic representation of S assimilation in plants. APS, adenosine 5'-phosphosulfate; PAPS, 3'-phosphoadenosine 5'-phosphosulfate; PAP, 3'-phosphoadenosine 5'-phosphate; OAS, O-acetylserine; SULTR, sulfate transporter; ATPS, ATP sulfurylase; APR, APS reductase; APK, APS kinase; SiR, sulfite reductase; SERAT, serine acetyltransferase; OASTL, OAS (thiol)lyase; PAPST, PAPS/PAP transporter; SOT, sulfotransferase; SAL1, 3ʹ,5ʹ-bisphosphate nucleotidase.

Immediately after SO₄^2– import across the plasma membranes or the plastid envelopes, adenylation of SO₄^2– to adenosine 5'-phosphosulfate (APS) takes place in both the plastids and the cytosol through an enzymatic reaction catalyzed by the ATP sulfurylase (ATPS). In higher plants, APS—the activated form of SO₄^2–—resides at the branching point of S metabolism in the plastids where the flux of S can further be partitioned into the reductive and oxidized S assimilation pathways, also described as the primary and the secondary pathways (Kopriva et al., 2012) (Fig. 1). In the reductive S assimilation pathway, the enzymatic reduction of APS to sulfide (S^2–) takes place in a stepwise manner; APS reductase (APR) catalyzes the first step generating sulfite (SO₃^2–) by transferring two electrons to the activated SO₄^2– group of APS, and subsequently in the second step the ferredoxin-dependent sulfite reductase (SiR) provides six electrons to reduce SO₃^2– to S^2– (Saito, 2004; Takahashi et al., 2011). S^2– then becomes the substrate for the O-acetylserine(thiol)lyase (OAS-TL) catalyzing the S^2– incorporation into the amino acid skeleton of O-acetylserine (OAS) to form the S amino acid cysteine (Cys) (Hell and Wirtz, 2011). OAS-TL acts together with serine acetyltransferase (SERAT), the enzyme activating the amino acid Ser to OAS, in the Cys synthase complex (Droux et al., 1998; Wirtz et al., 2004, 2012). While the pathway for SO₄^2– reduction to S^2– localizes exclusively in the plastids, Cys biosynthesis takes place in the plastids as well as in the mitochondria and the cytosol (Fig. 1) (Takahashi et al., 2011; Maruyama-Nakashita, 2017).

Methionine (Met) and glutathione (GSH) are S compounds of vital importance. Met is an essential amino acid for animals and humans as they do not assimilate SO₄^2– into Cys. Translation initiation of every protein requires Met. Apart from its essentiality for protein synthesis, Met serves as an intermediate for synthesis of S-adenosylmethionine, the major donor of methyl groups for numerous biological processes such as DNA and protein methylation and synthesis of various natural products, including polyamines, as well as 1-aminocyclopropane-1-carboxylate (ACC), the substrate for biosynthesis of the plant hormone ethylene (Hesse and Hoefgen, 2003; Hesse et al., 2004; Sauter et al., 2013). GSH is a tripeptide with its thiol group derived from Cys, playing important roles in redox regulation as well as detoxification, conjugation, and transport of various toxic metabolites and heavy metals (Noctor et al., 2012). Subsequent enzymatic oligomerization of the γ-Glu–Cys units of GSH into phytochelatins provides a mitigation mechanism notably against toxic heavy metals (Cobbett and Goldsbrough, 2002). GSH is essential for root development, as the reduction in its content causes a concentration-dependent reduction in root length (Vernoux et al., 2000; Bangash et al., 2019). Cys serves as the reduced S donor for synthesis of both Met and GSH. The Met biosynthesis from Cys occurs in the trans-sulfuration pathway with three distinct enzymatic reactions: the condensation of Cys and O-phosphohomoserine to form cystathionine; the hydrolysis of cystathionine to form homocysteine; and the methylation of homocysteine to produce Met, catalyzed by cystathionine γ-synthase, cystathionine β-lyase, and methionine synthase, respectively (Breitinger et al., 2001; Hesse et al., 2004). On the other hand, as indicated by the chemical formula γ-glutamylcysteinylglycine, GSH is synthesized from its constituting amino acids Glu, Cys, and Gly in a two-step ATP-dependent enzymatic reaction catalyzed by glutamate-cysteine ligase (also known as γ-glutamylcysteine synthetase) and glutathione synthetase (Noctor and Foyer, 1998; Takahashi et al., 2011). Loss of GSH synthesis is, therefore, lethal (Cairns et al., 2006).

In the oxidized S assimilation pathway, phosphorylation of APS to 3'-phosphoadenosine 5'-phosphosulfate (PAPS) via phospho-transfer from ATP takes place in both the plastids and the cytosol through the function of the APS kinase (APK) (Fig. 1) (Mugford et al., 2009). PAPS is the active form of sulfate used for sulfation of peptides and various specialized metabolites through condensation reactions catalyzed by the sulfotransferases (SOTs) in the cytosol and the Golgi apparatus (Klein and Papenbrock, 2004; Takahashi et al., 2011; Chan et al., 2019). Sulfation is an important modification step for animal proteins and hormones (Gunal et al., 2019), while in plants the best studied sulfated chemical compounds are specialized metabolites, for example glucosinolates, sulfoflavonoids, and peptide hormones such as phytosulfokine (Halkier and Gershenzon, 2006; Bednarek et al., 2009; Chan et al., 2019; Kaufmann and Sauter, 2019). While these modifications of specialized metabolites through sulfation yield a biproduct 3'-phosphoadenosine 5'-phosphate (PAP), subsequent metabolic conversion of this phosphonucleotide completes the pathway and prevents its signaling action (Chan et al., 2019). The chloroplast/mitochondrion dual-localized 3ʹ,5ʹ-bisphosphate nucleotidase SAL1 (Estavillo et al., 2011; Chan et al., 2016) and the PAPS/PAP transporters facilitating PAP import to these organelles (Gigolashvili et al., 2012; Ashykhmina et al., 2019) are key players in this metabolic cycle coupled with the function of the APK enzyme located at the branching point of S metabolism in plants (Fig. 1).

S metabolism is essential for all living organisms, but evolved differently across the evolutionary tree. Not only the arrangement of pathways and enzymatic reactions but also the enzymes themselves catalyzing the same reaction are functionally diversified and can be assembled in entirely different configurations. For example, the trans-sulfuration pathway between the S-amino acids Cys and Met operates in opposite directions in plants and animals (see later) (Giovanelli et al., 1985; Hesse and Hoefgen, 2003; Hesse et al., 2004; Brosnan and Brosnan, 2006). Diverse structural organization of SO₄^2– activation and PAPS biosynthetic enzymes—the ATPS and the APK enzymes found in strikingly different forms in bacteria and eukaryotes (see below)—represents another prominent example of S metabolic diversification which has emerged across kingdoms (Patron et al., 2008). S metabolism impacts life on Earth at different scales given its prevalence and indispensability. The diversity in plant S metabolism extends from a global geological scale to variations among plant taxa and ecotypes. In this review we will discuss plant S metabolism from the evolutionary perspective, highlighting its diversity, and outline open questions in research areas that would benefit from a broader consideration of plant evolution.

Adaptation of S metabolism in early evolution of life

S played a crucial role in shaping the development of life on Earth. Indeed, most of the hypotheses about the origin of life place S at the core of the initial energy sources (Olson, 2021). This is in agreement with evolutionary phylogeny that places organisms conducting dissimilatory S reduction and anoxygenic photosynthesis at the root of the tree of life (Burini et al., 2018). The S reduction probably produced hydrogen sulfide (H₂S) that in turn was utilized as an electron donor in anoxygenic photosynthesis and led to the production of oxidized S compounds, in particular SO₄^2–. Despite the occurrence of the first dissimilatory SO₄^2– reducers originating possibly already 3.5 billion years ago (Canfield and Raiswell, 1999), SO₄^2– concentration in the marine environment increased, reaching ~1 mM around 2.5 billion years ago. This seems to be the onset of oxygenic photosynthesis and further radiation of SO₄^2– reduction (Canfield and Raiswell, 1999). A further increase in SO₄^2– concentration led to changes in Fe deposits and more widespread use of SO₄^2– as a source of S for biochemical reactions (Burini et al., 2018). In addition, the increase in SO₄^2– concentration in seawater has been implicated to have driven changes in abundance of even diverse groups of primary producers in the oceans (Ratti et al., 2011). While in the Paleozoic ocean green algae and cyanobacteria were particularly abundant, with an increase in SO₄^2– in the seawater, the composition of phytoplankton shifted towards Eukaryotic Chl a+c-containing microalgae such as diatoms, coccolithophorids, and dinoflagellates (Ratti et al., 2011).

A remnant of the evolutionary process that has led to oxygenic photosynthesis and SO₄^2– assimilation is the ability of cyanobacteria to switch to anoxygenic photosynthesis in environments where S^2– is present (Fig. 2) (Cohen et al., 1986). Despite the fact that oxygenic photosynthesis evolved in S^2–-rich water (Canfield, 1998), S^2– is toxic to modern cyanobacteria because it inhibits cytochrome c oxidase and inactivates PSII (Cuevasanta et al., 2017). Nevertheless, cyanobacteria are able to grow in the presence of S^2– using several different strategies. Firstly, some cyanobacterial species restrict oxygenic photosynthesis in the presence of S^2–. Secondly, there are some species that are capable of oxidizing S^2– to elemental S by sulfide:quinone reductase (SQR), thus providing electrons to PSI (de Beer et al., 2017). SQR catalyzes the oxidation of S^2– to sulfane S and plays a key role in energy transduction and S^2– detoxification. Thirdly, certain cyanobacteria exhibit various types of D1 proteins, including an S^2–-resistant variant capable of sustaining both oxygenic and anoxygenic photosynthesis concurrently (Mulo et al., 2009). Fourthly, such habitats are advantageous for purely anoxygenic cyanobacteria lacking oxygenic photosynthesis (de Beer et al., 2017), which directly oxidize S^2– to S by SQR and therefore channel the flow of electrons from S^2– to PSI through the quinone pool. SQR is thus directly involved in the anoxygenic photosynthetic activity of cyanobacteria, allowing them to use S^2– as the electron donor (Rahimzadeh Karvansara et al., 2023). Although, in most cyanobacteria, photosynthesis is completely inhibited by sulfide, 45% of the sequenced cyanobacterial genomes in fact contain SQR, indicating the importance of the anoxygenic photosynthetic pathway in fitness under extremely dynamic habitats (Fig. 2) (Xia et al., 2017). Diversity in the photosynthetic metabolic pathways transforming S compounds thus reflects mechanisms that could have played an important role in the shaping of the extant photosynthetic organisms.

Fig. 2.

Effects of H₂S on cyanobacterial photosynthesis. In oxygenic environments, H₂S inhibits PSII and prevents electron transfer and water oxidation (1). Many cyanobacteria possess sulfide:quinone reductase (SQR) able to oxidize H₂S to elemental sulfur (S⁰) and transfer electrons to PSI (2). b6f, cytochrome b₆f complex.

Diversification of S metabolism at the kingdom scale

Dissimilatory and assimilatory S metabolism

Organisms with dissimilatory metabolism use inorganic S compounds as electron donors in the redox process to obtain energy, while assimilatory metabolism serves to integrate S into bioorganic compounds. Dissimilatory metabolism can be divided into oxidative and reductive processes. Dissimilatory S-oxidizing organisms are found in the Archaea and the Bacteria domains, and they are distinguished by phototropic and chemolithotrophic aerobic as well as anaerobic life styles (Dahl et al., 2008). The most common substrates are S^2–, elemental S, thiosulfate, and tetrathionate. Further diversification of these dissimilatory S oxidizers are the enzymatic systems involved in these oxidation reactions. The sulfur oxidoreductase (Sox) enzyme system is the main mechanism for oxidation of thiosulfate, but it can also use other substrates and transfers electrons to cytochrome c (Friedrich et al., 2001). Another common enzymatic mechanism is the dissimilatory sulfite reductase (Dsr) system, using various substrates to produce SO₃^2– to be further oxidized to SO₄^2–, usually through the activity of the dissimilatory APR and ATPS, generating ATP (Dahl et al., 2005). The operon structure and compositions of the Dsr system vary across different taxa (Anantharaman et al., 2018). In addition, S^2– can also be oxidized by SQR, as in anoxygenic photosynthesis (Fig. 3). None of these enzymes for oxidative dissimilatory S metabolism, however, has found its way to being integrated into plants. However, sulfide is also oxidized in plants, by the mitochondrial sulfur dioxygenase Ethylmalonic Encephalopathy Protein1 (ETHE1), which is important for detoxification of sulfide in the mitochondria to prevent the inhibition of cytochrome c oxidase (Birke et al., 2012; Krüßel et al., 2014).

Fig. 3.

Dissimilatory sulfur metabolism. In dissimilatory sulfate reduction (in black arrows), SO₄^2– is activated by ATP sulfurylase (ATPS) to APS, reduced to SO₃^2– by dissimilatory APS reductase (APR), and further to S^2– by dissimilatory sulfite reductase (DSR). Dissimilatory sulfite oxidation (in red arrows) uses various sulfur sources (sulfide, thiosulfate, elemental S, polysulfides, etc.) interconverted by different enzymes and complexes, such as sulfur oxidoredutase (SOX), sulfide:quinone oxidoreductase (SQR), and polysulfide reductase (PSR). S^2– is then oxidized to SO₃^2– and SO₄^2– by DSR, APR, and ATPS.

The core dissimilatory SO₄^2– reducing metabolism requires three enzymatic steps identical to the assimilatory SO₄^2– reduction, namely activation of SO₄^2– to APS, reduction of APS to SO₃^2–, and reduction of SO₃^2– to S^2– (Fig. 3). The structure of these enzymes and the reaction mechanisms in dissimilatory SO₄^2– reducers, however, are more similar to those identified in S-oxidizing organisms. This is particularly true for the dissimilatory APR and Dsr. The dissimilatory APR is formed by two subunits, AprA containing FAD cofactors and AprB binding two [4Fe–4S] clusters (Fritz et al., 2000). In contrast, the assimilatory APR or the PAPS reductases are structurally unrelated to AprA and AprB, even though some of them possess an Fe–S center (Patron et al., 2008) (the sulfonucleotide specificities of these assimilatory enzymes are discussed in the section ‘APS/PAPS-dependent SO₄^2– reduction’ below).

The core of the SiR from dissimilatory organisms is also composed of two subunits DsrA and DsrB, while it requires a DsrC protein to bind the S atom transitioning the redox reactions. DsrAB binds four sirohemes and eight [4Fe–4S] clusters as prosthetic groups, although two of the sirohemes might be substituted by sirohydrochlorins, the metal-free form of siroheme (Oliveira et al., 2011). The assimilatory SiRs, on the other hand, exist in at least three different forms. Siroheme is an essential cofactor for SiR and is present in all three assimilatory forms. The monomeric plant SiR is dependent on the function of ferredoxins providing electrons and contains a siroheme and an [4Fe–4S] cluster. In contrast, the bacterial SiR obtains its electrons from NADPH, and it is an oligomer of flavoprotein CysJ catalytic subunits and CysI proteins binding a siroheme and an [4Fe–4S] cluster. SiRs in yeasts and many other fungi are again different, composed of two distinct subunits binding a siroheme, FAD, and FMN (Patron et al., 2008).

ATPS for SO₄^2– activation is also found in different forms, while the core enzyme is less varied than the other two enzymes, except for the bacterial assimilatory ATPS. In addition, unlike the other two enzymes APR and SiR, ATPS is prone to different fusions, mainly with APK. In bacteria, ATPS is formed by two subunits, the CysD ATP pyrophosphatase and the CysN GTPase. Plants, green algae and cyanobacteria, and yeast and fungi possess different forms of a single protein ATPS (Patron et al., 2008). The phylogenetic relationships of these single protein ATPSs are, however, rather unexpected, because plant ATPSs are more similar to the animal counterparts than the green algal ATPS that probably originated from cyanobacteria (Patron et al., 2008).

S metabolism in plants and animals

Major distinctions and interkingdom variations in S metabolism can also be found between the assimilatory SO₄^2– reducers, such as algae and plants, and animals that are unable to reduce SO₄^2– and are dependent on organic S compounds in their nutrition. While plants metabolize SO₄^2– in both the reductive S assimilation pathway to Cys and the oxidized S assimilation pathway to PAPS transferring the SO₄^2– group to bioorganic compounds, animals possess only the latter pathway (Gunal et al., 2019). Thus, the entry of SO₄^2– into metabolism, namely activation of SO₄^2– by adenylation, is conserved between plants and animals as are the functions of the two enzymes, ATPS and APK. However, unlike plants, animals do not possess the enzymes APR and SiR and are thus unable to reduce SO₄^2–. Interestingly, despite the evolutionary distance and different arrangement of SO₄^2– assimilation, ATPS and APK from plants and animals are structurally conserved, as demonstrated by a greater extent of similarities of the plant-derived ATPSs found with the homologous enzymes from animals than with those of green algal and cyanobacterial origins (Patron et al., 2008).

Despite the structural similarities, the redox regulation of the APK enzyme appears to have evolved specifically in land plants, with conserved Cys residues playing pivotal roles in switching the enzyme activity on and off as they change the redox states and hence the protein conformation (Ravilious et al., 2012). In plants, the mechanism to control the APK enzyme activity may well be paralleled with redox regulations of the chloroplast-localizing APR isoenzyme (Bick et al., 2000) and the SAL1 3ʹ,5ʹ-bisphosphate nucleotidase (Chan et al., 2016). Under oxidative stress, the inhibition of the redox-sensitive APK enzyme may increase the flux of APS reduction and subsequent reductive S assimilation that leads to Cys biosynthesis by taking advantage of the oxidatively activated form of the APR isoenzyme (Bick et al., 2000), but limit the S metabolic flux toward PAPS synthesis that couples with the activity of the PAPS/PAP exchanger, sulfation of specialized metabolites in the cytosol and the Golgi, and dephosphorylation of PAP through the oxidation-sensitive SAL1 enzyme in the chloroplast (Gigolashvili et al., 2012; Chan et al., 2016, 2019). The proposed regulatory mechanism thus facilitates switching the flux of S at the branching point toward the reductive S assimilation pathway to meet the demand for synthesis of reduced GSH when plants are exposed to oxidative stress. The conserved Cys residues for the redox control of the APK enzyme appear to have been established after the emergence of land plant species, in conjunction with the bifurcating S assimilation pathway (Fig. 1) (Ravilious et al., 2012). It remains to be investigated how these redox control mechanisms in the oxidized and reductive S assimilation pathways are coordinated with the postulated function of PAP as a retrograde signal of oxidative stress from the plastids to the nucleus (Estavillo et al., 2011; Chan et al., 2016, 2019).

The retention of the oxidized S assimilation pathway—SO₄^2– activation to APS and subsequent synthesis of PAPS—indicates an expansion of biological sulfation in animals. Animals produce a wide variety of sulfated products, including hormones, carbohydrates, and proteins, through the aid of the gene family of SOT with broader functional diversities than in plants (Coughtrie, 2016; Gunal et al., 2019). In fact, the common occurrence of sulfated extracellular carbohydrates in humans, such as chondroitin sulfate or heparan sulfate, is not observed in plants, where no sulfated carbohydrates from cell walls have been isolated (Gunal et al., 2019). Both plants and animals modify Tyr in the polypeptide chain by sulfation; however, in plants, this modification is present only in small peptides, and the sulfation of proteins as a mechanism for post-translational regulation is most probably confined to animals as it has not yet been described in plants (Kaufmann and Sauter, 2019; Stewart and Ronald, 2022). The sulfated small peptides play various roles in control of plant development and growth, such as those represented by function of the phytosulfokines, the Casparian strip integrity factors CIF1 and CIF2, root meristem growth factor RGF, or TWISTED SEED1 (Matsuzaki et al., 2010; Doblas et al., 2017; Nakayama et al., 2017; Doll et al., 2020).

Another major difference between plants and animals is found in pathways that serve for synthesis of S-containing amino acids (Fig. 4). Plants synthesize Cys directly from S^2– and OAS, and use Cys for synthesis of homocysteine and Met through the trans-sulfuration and methylation pathways (Giovanelli et al., 1985; Hesse and Hoefgen, 2003; Hesse et al., 2004). The thiol group of Cys reacts with O-phosphohomoserine, the activated form of homoserine, to form cystathionine in the γ-replacement reaction catalyzed by cystathionine γ-synthase. Cystathionine then splits into homocysteine, pyruvate, and ammonium by cystathionine β-lyase. The thiol group of homocysteine is then methylated by methionine synthase to form Met. In contrast, the Cys synthase complex is not present in animals. They instead form Cys from Met through the transmethylation and reverse trans-sulfuration pathway (Brosnan and Brosnan, 2006). The pathway involves activation of Met to S-adenosylmethionine, transfer of the methyl group to methyl acceptor substrates, and hydrolysis of the resultant S-adenosylhomocysteine to homocysteine. The thiol group of homocysteine then joins with serine to form cystathionine in the β-replacement reaction catalyzed by cystathionine β-synthase. The cystathionine subsequently becomes the substrate for cystathionine γ-lyase breaking it down into Cys, α-ketobutyrate, and ammonium (Brosnan and Brosnan, 2006). Thus, plants and animals direct the flow of metabolism between the two S-containing amino acids in an opposite manner (Fig. 4).

Fig. 4.

Methionine–cysteine interconversions in plants and animals. Met synthesis in plants (upper panel) and trans-sulfuration pathway in animals (lower panel). CGS, cystathionine γ-synthase, CBL, cystathionine β-lyase; MS, methionine synthase; MAT, methionine adenosyltransferase; MT, methyltransferase; SAHH, S-adenosylhomocysteine hydrolase; CBS, cystathionine β-synthase; CSE, cystathionine γ-lyase; Acc, acceptor; CH₃-Acc, methylated acceptor; THF, tetrahydrofolate.

Despite these distinctions of the S metabolic pathways, protein persulfidation on exposure to H₂S appears consequential to regulation of enzymes in both plants and animals, and persulfidation is common to all kingdoms of life (Ogata et al., 2023). The discovery of the signaling function of S^2– and its prevalent roles in human and animal diseases led to establishing similar concepts in plants. Recent studies in fact provide implications that S^2– can act as a stress protectant and control a number of important physiological processes. Sulfide triggers stomatal closure and inhibits autophagy induction by ATG4 inactivation (Laureano-Marín et al., 2020; Scuffi et al., 2014; Aroca et al., 2020). However, the greatest difference from the animal system is that, in plants, S^2– is an intermediate of the S assimilation pathway determining the flux of Cys biosynthesis (Calderwood and Kopriva, 2014). The full understanding of the proposed multifaceted roles of S^2– in plants is still lacking.

APS/PAPS-dependent SO₄^2– reduction

The reduction of activated SO₄^2– is another step of the S assimilation pathway that underwent diversification in different organisms. The major difference is the nature of the substrate sulfonucleotide, APS or PAPS, as the activated form of SO₄^2– being reduced, and the corresponding enzyme APS reductase or PAPS reductase, respectively, catalyzing the reactions (Kopriva and Koprivova, 2004; Carroll et al., 2005) (Fig. 5). The mechanism of SO₄^2– reduction and the identity of the substrate and the enzyme have been controversially discussed especially in the plant community (Kopriva and Koprivova, 2004).

Fig. 5.

Diversity of enzymes catalyzing APS reduction in S assimilation. The schematic representation of enzymes reducing activated SO₄^2– in bacteria and cyanobacteria, yeast and fungi, basal plants, flowering plants, and eukaryotic microalgae, namely the diatom Thalassiosira pseudonana (top) and the dinoflagellate Heterocapsa triquetra (bottom). Orange-colored APR enzymes mark the presence of the Fe–S clusters. Trx, thioredoxin or thioredoxin-like domain.

Since the reductive S assimilation pathway was first resolved in unicellular model organisms, Escherichai coli and yeast, both of which use PAPS and PAPS reductase, and because of the fact that the PAPS reductase activity was also present in photosynthesizing organisms, such as the cyanobacterium Synechococcus PCC 6301 (Peck, 1961; Wagner et al., 1978), the PAPS-dependent pathway was long considered universally responsible for the reduction of SO₄^2– and believed to have also been adopted by plants. However, a later discovery of the APS-dependent activity in green algae and plants with SO₃^2– bound to a carrier as a reaction product led to the identification of the corresponding enzyme named APS sulfotransferase (Schmidt, 1972). The bacterial and fungal PAPS reductase utilizes thioredoxin as an electron donor. In contrast, the plant APS sulfotransferase was shown to be independent of thioredoxin. In addition, APS sulfotransferase was regulated by a number of environmental conditions and metabolites, and was shown to play an important role in control of S assimilation in plants and algae (Brunold, 1990). When the corresponding gene for this enzyme was first identified in plants, the APS sulfotransferase was found to be a protein with two domains, one similar to bacterial PAPS reductases and the other a thioredoxin-like domain (Gutierrez-Marcos et al., 1996; Setya et al., 1996). The latter domain has a function of glutaredoxin, reflected in the ability of GSH to serve as the reductant for the enzyme (Bick et al., 1998). Subsequently, the enzyme was shown to carry an Fe–S center as a cofactor, and therefore, it was renamed as APS reductase (APR) (Kopriva et al., 2001) to be an enzyme specific to eukaryotes. However, a number of bacterial taxa demonstrated activities to utilize APS in SO₄^2– reduction as well, but in this case using thioredoxin as an electron donor (Abola et al., 1999; Bick et al., 2000). Consequently, the presence of the Fe–S center was postulated to be the key factor determining the APS or PAPS specificity of the enzyme (Kopriva et al., 2002). However, this view changed when an APS-dependent form without the thioredoxin domain and the Fe–S cluster, APR-B, was found in the moss Physcomitrium patens (Kopriva et al., 2007; Stevenson et al., 2013). The APS and PAPS reductases have a common reaction mechanism, transferring the activated SO₄^2– group from the sulfonucleotide to a conserved Cys residue of the protein, forming a stable intermediate, which is then reductively released as free SO₃^2– by thioredoxin, glutaredoxin, or the C-terminal domain of the plant enzyme (Weber et al., 2000; Carroll et al., 2005).

The diversity of the sulfonucleotide reductases (Fig. 5) has been further extended by identification of genes in eukaryotic microalgae with APR-B fused to thioredoxin, or with APR being fused to ATPS (Patron et al., 2008). However, it seems that among the eukaryotes the division of these reductases is very clear, as algae and plants exclusively use APS-dependent enzymes for SO₄^2– reduction, while yeast and fungi use PAPS reductase. Among prokaryotes, the boundary is not that simple: the APS-dependent pathway seems to be more widespread among different taxa, while the PAPS-dependent pathway is confined to γ-proteobacteria and cyanobacteria, in both of which however we may still find APS-reducing taxa (Patron et al., 2008). The PAPS-dependent reduction might have evolved from APR, possibly through an adaptation to Fe- and/or S-poor conditions. Limited Fe availability might have led to the replacement of the Fe–S-requiring mechanism with a costly alternative that would instead require one additional ATP for SO₄^2– reduction. Supporting this hypothesis, the dissimilatory SO₄^2–-reducing bacteria use APS reductase with Fe–S clusters.

Speciation of S metabolism at the species scale

S-containing specialized metabolites

The prime example of species-specific diversity of S metabolism are S-containing specialized metabolites. Plants produce various S-containing natural products—some highly specialized only in a few species, while some are more common (Fig. 6). S in these S-containing specialized metabolites can be present in oxidized or reduced form. The oxidized S modifications in various metabolites are products of biological sulfations catalyzed by a variety of SOTs (Hirschmann et al., 2014). These compounds have important and varied functions in different aspects of plant physiology and environmental interactions, and are thus often variable not only on the species level but also within ecotypes of the same species (Kliebenstein et al., 2001a). The full extent of S-containing plant metabolites is not fully described yet; even in the model plant species Arabidopsis thaliana, a large number of detected S-containing metabolites could not be structurally identified (Gläser et al., 2014).

Fig. 6.

Examples of S-containing specialized metabolites in cyanobacteria and plants. Sulfur atoms are marked green when they originate from reduced S assimilation or orange when they result from oxidized metabolism using sulfation.

Glucosinolates

Glucosinolates (GLSs) are the best characterized group of S-containing specialized metabolites specific to a small number of taxa as they are only found in the Brassicaceae (Halkier and Gershenzon, 2006). GLSs are amino acid-derived compounds with important roles in defense and S homeostasis in plants possessing S in both oxidized and reduced forms (Fig. 6). The ~170 GLS structures are divided into three classes depending on the amino acids they originate from; that is, the aliphatic GLS from amino acids Met, Ala, Leu, Ile, and Val (e.g. glucoraphanin, Fig. 6); the indolic GLSs from Trp (e.g. glucobrassicin); and the aromatic GLSs from Phe (e.g. gluconasturtiin) (Fahey et al., 2001; Sonderby et al., 2010). While the core structure of GLSs with an N-sulfated imine linked through a thioglucoside bond with glucose is identical, the modifications of the amino acid-derived side chain give the structural variety (Sonderby et al., 2010). One level of such variations occurs through chain elongation of Met before they enter the core pathway for GLS synthesis, while the others involve decorative modifications, such as hydroxylation of the amino acid-derived side chains and methylation of the hydroxyl group.

GLSs accumulated in plant organs remain inactive until exposed to and activated by myrosinases, which hydrolyze the thioglucoside bond, causing the aglycone to rearrange into the volatile isothiocyanate, nitrile, or epithionitrile (Halkier and Gershenzon, 2006). The chemical identity of the volatile products depends on the presence and nature of specifier proteins (Kuchernig et al., 2012). Traditionally, because of these volatile cocktails, GLSs were primarily seen as chemical defense compounds against insects and other herbivores, but also as attractants for specialized insects or as the basis for tritropic interactions (Ratzka et al., 2002). However, GLSs also serve an important role in plant immunity triggered by fungal and bacterial pathogens, and recently they were also shown to affect the plant microbiome (Bednarek et al., 2009; Clay et al., 2009; Fan et al., 2011; Jacoby et al., 2021). In particular, the indolic GLSs play an essential role in plant interactions with both pathogenic and beneficial fungi. Some GLSs provide clear health benefit to humans, while on the other hand, high GLS content in feed is detrimental for animal weight gains (Traka and Mithen, 2009). Correspondingly, a high GLS broccoli variety has been developed with increased content of glucoraphanin, the precursor of sulforaphane with anti-carcinogenic properties, while modern oilseed rape varieties have low GLS content instead enabling the use of the seed press cakes for animal feed (Mithen et al., 2003). Beside defense, herbivory, and microbe interactions, GLSs as S-rich compounds have been considered to represent the S storage in plants; however, the S recycling pathway from GLSs to primary metabolism has been shown only recently (Sugiyama et al., 2021).

GLS profiles differ not only across the plant species but also within the ecotypes of a single species. Such a difference has been instrumental in dissection of the regulation of their synthesis and identification of both regulatory and biosynthetic genes. The qualitative and quantitative variation in GLS accumulation in Arabidopsis, particularly of the aliphatic class of GLSs, is generally explainable by four genetic loci. GS-Elong, containing the methylthioalkylmalate synthase (MAM) genes, controls chain elongation of the aliphatic GLSs as they are responsible for reactions cyclically adding a carbon to Met and subsequently to the chain-elongated forms (Kroymann et al., 2001). In Arabidopsis and Brassica, two MAM genes control synthesis of the aliphatic GLSs with either three or four carbon molecules (Abrahams et al., 2020). However, long-chain aliphatic GLSs are also produced in Arabidopsis with C7 and C8, and in other Brassicaceae species with up to C10 and C11 side chains (Fahey et al., 2001). Another major quantitative trait locus (QTL) is caused by variation in 2-oxoacid-dependent dioxygenase AOP2 that converts methylsulfinyl GLSs to alkenyl GLSs (Kliebenstein et al., 2001b). Interestingly, AOP2 affects not only qualitative but also quantitative accumulation of GLSs (Chan et al., 2011). Variation of a neighboring gene AOP3 is responsible for production or absence of 3-hydroxypropyl GLS (Kliebenstein et al., 2001a). Interestingly, due to the duplications and rearrangements of the AOP3 locus in A. thaliana, 3-hydroxypropyl GLS accumulates in the seeds of all accessions but is much more limited in the leaves (Chan et al., 2011). The fourth QTL, GS-OH, is responsible for the ability to hydroxylate alkenyl GLS, for which the causal gene has been identified to encode 2-oxoacid-dependent dioxygenase unrelated to the AOP2/AOP3 enzymes (Hansen et al., 2008). The allelic configuration at these four loci can predict profiles of the aliphatic GLSs in Arabidopsis and partly in Brassicas, but other polymorphisms are responsible for further variation in other species, such as the BCMA locus in Boechera stricta, which controls production of the branched-chain aliphatic GLSs (Prasad et al., 2012). Interestingly, much less is known about the genes underlying variation in the indolic GLSs. Despite several QTLs being identified, only one gene, the CYP81F2 converting indol-3-yl-methylglucosinolate (I3M) to 4-hydroxy-indol-3-yl-methylglucosinolate (4OH-I3M), has been cloned as the causal gene for the Indole Glucosinolate Modifier1 (IGM1) locus (Pfalz et al., 2009).

Sulfoflavonoids and cysteine sulfoxides

Another large group of sulfated metabolites beside GLSs are the sulfoflavonoids (Fig. 6). Sulfoflavonoids have been identified in ~250 species in different forms and with different numbers of SO₄^2– residues (Barron et al., 1988; Teles et al., 2018). Flavonol-specific SOTs have been described in A. thaliana (Gidda and Varin, 2006; Hashiguchi et al., 2013) and in several species of the genus Flaveria (Varin et al., 1992; Marsolais and Varin, 1998). Since different Flaveria species, however, produce different combinations of sulfoflavonoids, these metabolites were considered as possible markers differentiating the type of photosynthesis (e.g. C₃ versus C₄; Hannoufa et al., 1994). However, a more recent study revealed that there is no obvious correlation between the profiles of sulfoflavonoids and the photosynthetic types (Kleinenkuhnen et al., 2019). While flavonoids are known to represent numerous functions in plants, such as pigmentation, photoprotection, developmental regulation, and plant–microbe interactions, the biological function(s) of sulfated flavonoids is mostly unknown (Teles et al., 2018; Kleinenkuhnen et al., 2019). The sulfoflavonoids, such as quercetin 3-sulfate, can form a substantial S pool in plant leaves, as determined in Flaveria pringlei, and may thus serve as S storage compounds like GLSs (Sugiyama et al., 2021). However, the mechanism of the remobilization of the sulfate group is unclear since a sulfatase enzyme seems to be lacking in plants. Alternatively, sulfoflavonoids could represent an overflow mechanism to maintain S homeostasis during excessive SO₄^2– supply. Whether the distribution of sulfoflavonoids in different taxa follows any ecological and/or evolutionary pattern, however, remains to be elucidated. It has to be noted that sulfoflavonoids are also detectable in human tissues, produced by sulfation of flavonoids during Phase II of xenobiotic metabolism (Pai et al., 2001).

Not all S-containing specialized metabolites are sulfated. The flavor of the Allium vegetables, for example, is formed by a mix of organic sulfides, disulfides, and other volatiles containing reduced S (Block, 1985; Marcinkowska and Jelen, 2022). Similar to GLSs, these flavor compounds are formed after enzymatic degradation of the non-volatile precursor sulfoxides (Block, 1985). These precursor sulfoxides accumulate in the cytosol and, upon cellular damage, are metabolized by alliinase, a vacuole-sequestered C-S lyase, to allicin and sulfenic acids. These primary products are then rapidly converted into a variety of volatile sulfides, disulfides, and trisulfides, or more complex cyclic dithiins and sulfines (Block, 1992). Four precursor sulfoxides have been detected in varying amounts in different Allium species and accessions: S-allyl cysteine sulfoxide (alliin), S-methyl cysteine sulfoxide (methiin), S-propyl cysteine sulfoxide (propiin), and S-trans-prop-1-enyl cysteine sulfoxide (isoalliin), converted to a much greater variation of organic S compounds in garlic oils or extracts (Block, 1992; Jones et al., 2004). These variations in accumulation of the S compounds detected among the onion or garlic accessions are associated with their varying flavors, and antimicrobial and health properties. However, the list of sulfoxide precursors and active volatile compounds appears far from complete, and possible involvement of additional protein factors in control of altering the chemical compositions of Allium S volatiles, which may be paralleled with the specifier proteins for GLS degradation (Eisenschmidt-Bonn et al., 2019), remains to be investigated (Block, 1992; Jones et al., 2004; Avgeri et al., 2020).

Specialized S metabolites of cyanobacteria

Cyanobacteria also produce a range of specialized metabolites with divergent structures, such as peptides, polyketides (PKs), peptide/PK hybrids, alkanes, terpenes, etc., that are often limited to a few strains (Nunnery et al., 2010; Kehr et al., 2011). Many cyanobacterial specialized metabolites have biological activities as cyanotoxins, while some may also serve as potential therapeutics, such as the cryptophycins (Nunnery et al., 2010). Prominent examples of S-containing specialized metabolites in cyanobacteria are thiohistidines, such as ovothiol (Fig. 6) and ergothioneine, found in several cyanobacterial strains such as Microcystis aeruginosa and Cylindrospermopsis raciborskii (Liao and Seebeck, 2017; Brancaccio et al., 2021). Although their exact physiological role is not known, thiohistidines seem to protect cyanobacteria against reactive oxygen species (Liao and Seebeck, 2017; Brancaccio et al., 2021). Sulfated specialized metabolites might also be present in cyanobacterial strains. In fact, SOT homologs have been found in several operons of metabolic genes in cyanobacteria. One example of a cyanobacterial sulfated metabolite is the cyanotoxin cylindrospermopsin produced by Cylindrospermopsis raciborskii, Umezakia natans, Aphanizomenon ovalisporum, and Raphidiopsis curvata (Yang et al., 2021) (Fig. 6). Cylindrospermopsin is composed of a guanidine modified by a SO₄^2– group and an additional uracil side chain (Ohtani et al., 1992), and it functions most probably as an allelopathic metabolite to inhibit the growth of competing organisms (Teneva et al., 2023). Thus, also in cyanobacteria, different lineages utilize S for synthesis of metabolites with specialized functions, adding to the large variation of S biochemistry in the photosynthetic organisms.

Localization of S assimilation in bundle sheath versus mesophyll cells

A different kind of metabolic diversification of S assimilation has been observed between the plant lineages concerning the spatial distribution of the responsible enzymes. The cell type-specific localization of enzymes has become a prominent topic in plant physiology since the discovery of C₄ photosynthesis, as this photosynthetic mechanism is dependent on separation of primary CO₂ assimilation by the phosphoenolpyruvate carboxylase in the mesophyll cells and the decarboxylation of the resulting C₄ acids and recapture of the CO₂ by the Rubisco enzyme expressed in the bundle sheath cells (Hatch and Slack, 1966; Slack et al., 1969). However, not only is the pathway for the CO₂ assimilation differentially distributed between the mesophyll and bundle sheath cells of C₄ plants, but so are the reductive assimilation pathways for two major mineral nutrients, NO₃^– and SO₄^2–, confined to these specific cell types (Mellor and Tregunna, 1971; Moore and Black, 1979; Gerwick et al., 1980; Passera and Ghisi, 1982). In contrast to the NO₃^– reduction pathway found specifically in the mesophyll, the SO₄^2– reduction pathway is localized to the bundle sheath cells of a number of C₄ species (Gerwick et al., 1980). While this cell type-specific localization seems to be conserved, at least in a few independent C₄ lineages, its significance is still unknown (Weckopp and Kopriva, 2014). In attempts to clarify the importance of spatial separation of the SO₄^2– assimilation pathway for the evolution of C₄ photosynthesis, C₃–C₄ intermediate species of the model genus Flaveria were employed for a comparative analysis, from which the separation was found to be confined to the monocot C₄ species (Koprivova et al., 2001). This has been confirmed further by RNA-seq analyses of another dicot C₄ species, Cleome gynandra, which showed no differential expression of genes for the SO₄^2– assimilation pathway between the mesophyll and bundle sheath cells (Kulahoglu et al., 2014).

Surprisingly, however, a translatome analysis of a C₃ species Arabidopsis revealed a strong enrichment of pathway enzymes for SO₄^2– assimilation as well as GLS biosynthesis in bundle sheath cells (Aubry et al., 2014). The bundle sheath cell-specific expression of the GLS biosynthetic genes seems to be controlled by the MYB transcription factors MYB28 and MYB29, both characterized previously as controlling aliphatic GLS biosynthesis, as well as the basic helix–loop–helix (bHLH) factors MYC2, MYC3, and MYC4 that are essential for this pathway and interacting with MYB28, MYB29, and other MYB transcription factors otherwise responsible for indolic GLS biosynthesis (Dickinson et al., 2020). Interestingly, this trait appears to be conserved in C₃ plants, because the monocot rice also showed the same spatial gene expression pattern for SO₄^2– assimilation (Hua et al., 2021). However, the bundle sheath localization of SO₄^2– assimilation seems to not always be conserved ubiquitously among the C₃ monocots, as demonstrated by activities for corresponding enzymes found in both cell types in wheat (Schmutz and Brunold, 1984). Thus, the variation in the cell type-specific localization of S assimilation does exist in plants, differentiating the C₄ species with good evidence to support the exclusive presence of the entire pathway in the bundle sheath cells of the C₄ monocots while possibly there is no such spatial separation in the C₄ dicots. In contrast, for the C₃ species, not enough evidence beyond the two model plant species exists and the jury is thus still out on whether the observed bundle sheath localization found in Arabidopsis and rice can be generalized.

Natural variation in genes for SO₄^2– assimilation in control of S homeostasis

Natural accessions of a plant species can differ not only by distinct profiles of specialized metabolites, such as GLSs, but also by SO₄^2– accumulation. The variation in SO₄^2– accumulation has been explored extensively to dissect genetic control mechanisms associated with SO₄^2– assimilation and general S homeostasis in Arabidopsis. Indeed, a >5-fold difference in SO₄^2– accumulation was observed between the recombinant inbred lines (RILs) of the Arabidopsis accessions, Bay-0 and Shahdara (Loudet et al., 2007), and the subsequent QTL analysis resulted in identifying two key genes controlling a large part of the variation. The causal genes associated with these QTLs encoded the major isoforms of the S assimilation pathway enzymes, APR2 and ATPS1 (Loudet et al., 2007; Koprivova et al., 2013). Interestingly, the variation in SO₄^2– accumulation was governed by two different mechanisms modifying the activity and the quantity of the corresponding enzymes in Arabidopsis accessions, Bay-0 and Shahdara. The first variation identified was associated with the Shahdara haplotype of APR2 with a non-synonymous single nucleotide polymorphism (SNP) that changes Ala399 into Glu and thus inhibits the binding of the enzyme to the electron donor GSH (Loudet et al., 2007). On the other hand, the second variation was associated with a deletion of an enhancer element in the first intron of the ATPS1 gene that led to a reduction in the ATPS1 transcript levels in Bay-0 relative to Shahdara, although the primary amino acid sequences of ATPS1 in Bay-0 and Shahdara were identical (Koprivova et al., 2013). These weak alleles of APR2 and ATPS1 lowered the enzymatic activity and the presence of the corresponding proteins, respectively, and thus reduced the S metabolic flux through the S assimilation pathway and ultimately led to the accumulation of the entry substrate SO₄^2– (Koprivova et al., 2013; Chao et al., 2014). Interestingly, the less active form of APR2 from Shahdara had only a limited effect on steady-state levels of Cys and GSH accumulations, unless challenged by a stress, such as an exposure to selenate (Loudet et al., 2007; Grant et al., 2011).

An independent screen for genes causing higher accumulation of S and SO₄^2– identified another weak APR2 allele that replaces Gly266 with Arg, inactivating the enzyme activity in the Hod accession of Arabidopsis (Chao et al., 2014). In a reverse approach, several accessions with various APR2 haplotypes were tested for SO₄^2– accumulation, revealing a third natural variation with the Phe265 to Ser replacement in Lov-5 and three other Swedish accessions (Chao et al., 2014). Also, for ATPS1, an allele with a non-synonymous SNP replacing Gly342 with Asp was found to inactivate the enzyme in the Naes-2 accession (Herrmann et al., 2014). While these inactivating SNPs were either unique or rare among the Arabidopsis accessions, the SNPs associated with dysfunction of the transcriptional enhancer in the first intron of ATPS1 including the Bay-0 allele were more frequently observed in the population (Koprivova et al., 2013). These findings implicate conserved influence and prevalence of natural variations in APR2 and ATPS1 over SO₄^2– and total S accumulations. However, current datasets suggest other mechanisms to be possibly involved in S homeostasis, since even within the four accessions with the APR2 Phe265–Ser haplotype, a large variation in both S and SO₄^2– content was still observed (Chao et al., 2014). Indeed, recent detailed analysis of S homeostasis in Arabidopsis accessions revealed a tight connection of foliar S levels with SO₄^2– uptake, but also a correlation with GSH and GLS synthesis (de Jager et al., 2023).

Natural variations in the mitochondrial OAS-TL—also known as OASC—suggest an influence of Cys biosynthesis on SO₄^2– accumulation. These OASC variations were initially demonstrated using associative transcriptomics, a variant approach of a genome-wide association study (GWAS) with Brassica napus (Koprivova et al., 2014), while two different haplotypes were later identified from the Arabidopsis population (Koprivova et al., 2022). The SO₄^2– accumulation profiles associated with the SNPs identified from B. napus and Arabidopsis suggested that they could alter the function of the OASC differently. In B. napus the causal SNP for the Gln355–Arg replacement led to an increase in SO₄^2– accumulation, while in A. thaliana the Lys81–Arg polymorphism was causative of a decrease in SO₄^2– accumulation (Koprivova et al., 2022). Unlike APR2 and ATPS1, however, the mechanism by which OASC affects SO₄^2– accumulation is still not well established. OAS-TL is an essential catalytic component of the Cys synthase complex whose enzymatic function is strictly controlled by the presence or absence of the substrates OAS and sulfide, modulating the flux of Cys biosynthesis. Findings of the arsenite-tolerant astol1 mutant of rice and the causal mutation of the chloroplast-localizing OAS-TL stabilizing the assembly of the Cys synthase complex suggest a mechanism associated with an organelle-localized OAS-TL altering SO₄^2– accumulation in plants (Sun et al., 2021). The mitochondrial pathway is the major route for synthesis of OAS (Watanabe et al., 2008; Wirtz et al., 2012). How naturally occurring polymorphisms of the mitochondrial OAS-TL influence the activity and composition of the Cys synthase complex and how would they modulate the mitochondrial OAS levels that may take control of the postulated S-sensing function of the enzyme complex await further investigation.

Conclusion

The biochemical properties of S make it pre-destined to be involved in a great variety of reactions that are indispensable. Changes in S availability in different chemical forms and redox states shaped life on Earth, such as the expansion of groups of microorganisms using S compounds as their energy source or alteration of the phytoplankton composition in the oceans. The extant organisms display functional innovations and extensive adaptive evolution that took place for the same S metabolic processes. Metabolic diversification is found across all scales, from diverse and complementary features of S-containing specialized metabolites in different accessions of the same plant species to a reverse flow of S between Cys and Met in plants and animals. While our comprehension of plant S metabolism and its integration with general S homeostasis has increased immensely alongside growing traction toward genetic and genomic research, we know very little about how and under what evolutionary drivers the ancestral species pivoted to establish the great variety of the S metabolic pathways and enzyme functions. Delving into the evolutionary history, however, holds the promise of unlocking the potential for optimizing these metabolic pathways for the future, paving the way for advancements in enhancing crop productivity and resilience.

Conflict of interest

The authors have no conflicts to declare.

Funding

The work of SK and PRK is supported by the Deutsche Forschungsgemeinschaft (DFG) under Germany’s Excellence Strategy—EXC 2048/1—project 390686111. SK is additionally funded by a DFG grant no. 426501900. HT acknowledges funding support from AgBioResearch USDA-NIFA (Hatch Project 1018575).