Arabidopsis nicotianamine synthases comprise a common core-NAS domain fused to a variable autoinhibitory C terminus

Nicotianamine synthase (NAS) catalyzes the biosynthesis of the low-molecular-mass metal chelator nicotianamine (NA) from the 2-aminobutyrate moieties of three SAM molecules. NA has central roles in metal nutrition and metal homeostasis of flowering plants. The enzymatic function of NAS remains poorly understood. Crystal structures are available for archaeal and bacterial NAS-like proteins that carry out simpler aminobutanoyl transferase reactions. Here, we report amino acids essential for the activity of AtNAS1 based on structural modeling and site-directed mutagenesis. Using a newly developed enzyme-coupled continuous activity assay, we compare differing NAS proteins identified through multiple sequence alignments and phylogenetic analyses. In most NAS of dicotyledonous and monocotyledonous plants (class Ia and Ib), the core-NAS domain is fused to a variable C-terminal domain. Compared to fungal and moss NAS that comprise merely a core-NAS domain (class III), NA biosynthetic activities of the four paralogous Arabidopsis thaliana NAS proteins were far lower. C-terminally trimmed core-AtNAS variants exhibited strongly elevated activities. Of 320 amino acids of AtNAS1, twelve, 287-TRGCMFMPCNCS-298, accounted for the autoinhibitory effect of the C terminus, of which approximately one-third was attributed to N296 within a CNCS motif that is fully conserved in Arabidopsis. No detectable NA biosynthesis was mediated by two representative plant NAS proteins that naturally lack the C-terminal domain, class Ia Arabidopsis halleri NAS5 and Medicago truncatula NAS2 of class II which is found in dicots and diverged early during the evolution of flowering plants. Next, we will address a possible posttranslational release of autoinhibition in class I NAS proteins.

Nicotianamine synthase (NAS) catalyzes the biosynthesis of the low-molecular-mass metal chelator nicotianamine (NA) from the 2-aminobutyrate moieties of three SAM molecules. NA has central roles in metal nutrition and metal homeostasis of flowering plants. The enzymatic function of NAS remains poorly understood. Crystal structures are available for archaeal and bacterial NAS-like proteins that carry out simpler aminobutanoyl transferase reactions. Here, we report amino acids essential for the activity of AtNAS1 based on structural modeling and site-directed mutagenesis. Using a newly developed enzyme-coupled continuous activity assay, we compare differing NAS proteins identified through multiple sequence alignments and phylogenetic analyses. In most NAS of dicotyledonous and monocotyledonous plants (class Ia and Ib), the core-NAS domain is fused to a variable C-terminal domain. Compared to fungal and moss NAS that comprise merely a core-NAS domain (class III), NA biosynthetic activities of the four paralogous Arabidopsis thaliana NAS proteins were far lower. C-terminally trimmed core-AtNAS variants exhibited strongly elevated activities. Of 320 amino acids of AtNAS1, twelve, 287-TRGCMFMPCNCS-298, accounted for the autoinhibitory effect of the C terminus, of which approximately one-third was attributed to N296 within a CNCS motif that is fully conserved in Arabidopsis. No detectable NA biosynthesis was mediated by two representative plant NAS proteins that naturally lack the C-terminal domain, class Ia Arabidopsis halleri NAS5 and Medicago truncatula NAS2 of class II which is found in dicots and diverged early during the evolution of flowering plants. Next, we will address a possible posttranslational release of autoinhibition in class I NAS proteins.
Maintaining adequate uptake, distribution, and storage of essential metals, for example iron (Fe), zinc (Zn), and copper (Cu), is critical for the survival and fitness of all organisms (1,2). Thus, metal homeostasis networks operate by orchestrating a variety of transmembrane metal transport, metal chelation, and metal trafficking processes. In land plants, the nonproteinogenic amino acid nicotianamine (NA) has a central function as a low-molecular-mass chelator molecule which can bind cations of Fe, Zn, Cu, and other metals. Nicotianamine synthase (NAS, EC 2.5.1.43; Figs. 1 and S1) enzymes, first identified in angiosperm plants, catalyze the biosynthesis of one molecule NA from three molecules of SAM in a stepwise fashion (3)(4)(5)(6). In plants, NA can act in the cytosol, vacuole, xylem, and phloem to affect the intracellular sequestration or the intracellular, intercellular, and long-distance partitioning of metals, often in a localized and at least partially metal-specific fashion through an interplay with transmembrane transporters of differing substrate specificities. For instance, NA acts in the transport of iron from the phloem outward in sink organs in young leaves or developing seeds (5,(7)(8)(9)(10)(11). In graminaceous monocotyledons, NA additionally serves as the precursor for the biosynthesis of phytosiderophores, such as 2 0 -deoxymugineic acid, which are secreted into the rhizosphere for scavenging Fe(III) prior to the uptake of Fe(III)-phytosiderophore complexes into root cells in strategy II of plant iron uptake (12). NAS has been receiving increasing attention over the past decade because of its demonstrated potential for iron and zinc biofortification of crops (13)(14)(15)(16)(17).
Although initially thought to be unique to seed plants, NAS proteins and NA production were subsequently identified in the moss Physcomitrium (Physcomitrella) patens within the division of bryophytes and in the filamentous fungus Neurospora crassa (18)(19)(20)(21). Moreover, NAS-like enzymes were described in the bacteria Staphylococcus aureus, Pseudomonas aeruginosa, and Yersinia pestis, as well as in the archaeon Methanothermobacter thermautotrophicus (22)(23)(24)(25). A crystal structure of MtNAS provided seminal insights into its catalytic mechanism (23,26). Plant and fungal NAS enzymes sequentially use the 2-aminobutyrate moieties of three SAM molecules to form one molecule of NA, releasing three molecules of 5 0 -methylthioadenosine (MTA) as a byproduct (12) (Figs. 1 and S1). In bacterial and archaeal NAS-like enzymes, various amino acids can apparently serve as the starter molecule for the reaction, namely glutamate, D-histidine, or L-histidine (Fig. S1). NAS-like enzymes link the α-amino nitrogen of the starter molecule acting as a nucleophile onto the C 4 atom of a 2-aminobutyrate moiety from SAM (Figs. 1 and S1). Subsequently, there can be an additional cycle of extension, either by using another 2-aminobutyrate moiety from SAM similar to plant and fungal NAS or alternatively by using pyruvate or αketoglutarate (Figs. 1 and S1). As a result, NAS-like enzymes catalyze the formation of NA-like compounds thermoNA (tNA), xNA, or yNA (22,27).
To date, the absence of a sensitive and quantitative assay for the analysis of NAS enzyme activity has hampered biochemical studies. Instead, NAS activity was demonstrated upon separation of reaction mixtures either by TLC, for example through autoradiography of [ 14 C]-NA formed from [ 14 C]-SAM (4, 7), or alternatively by HPLC, with photometric or massspectrometric detection of derivatized or underivatized NA (18,28). NAS enzymes are strongly feedback-inhibited by MTA, which is also a spontaneous breakdown product of the labile substrate SAM (7,28). Thus, assays were previously performed in the presence of large amounts of the NAS enzyme and low concentrations of the substrate SAM, therefore requiring sensitive methods to detect the (low amounts of) NA formed.
Here, we report that a continuous enzyme-coupled photometric assay enabled us to quantify the catalytic activities of a number of previously characterized and uncharacterized recombinant NAS proteins in vitro. Compared to activities between 0.2 and 0.8 nkat (mg protein) −1 of fungal and moss NAS, the activity of Arabidopsis thaliana NAS1 (AT5G04950) was less than one-tenth and thus substantially lower. To understand the cause of such strikingly differing activities of NAS proteins from different organisms, we conducted a phylogenetic analysis of NAS homologs predicted from publicly available nucleotide sequence data. We resolved A. thaliana NAS1 (AtNAS1) to AtNAS4 (class Ia), as well as NcNAS (class III), for example, in their expected relative phylogenetic positions. In addition to the previously characterized dicot (class Ia) and monocot (class Ib) NAS proteins, a new group of NAS proteins (class II) is represented in a subset of dicots. Moreover, a fifth class Ia NAS and paralog of NAS1 and NAS2, NAS5, is encoded in the genomes of numerous Brassicaceae species including members of the Arabidopsis genus, but not in the genome of A. thaliana. Amino acid sequence alignments revealed the presence of an elongated C terminus in about 90% of class Ia and class Ib NAS proteins. Amino acid sequences of these C-terminal domains are between 33 and 56 amino acids long and variable overall, but they share conserved amino acids in two regions, which correspond approximately to amino acids 293 to 304 and 309 to 320 of AtNAS1. By contrast, most other NAS and NAS-like proteins, also including NcNAS and PpNAS, consist predominantly of a core-NAS domain. Enzyme activities quantified in vitro of purified recombinant AtNAS mutant variants carrying C-terminal deletions suggested an autoinhibitory role of the elongated C termini of class Ia NAS proteins. Within AtNAS1, which is predicted to comprise 320 amino acids in total, we attributed the autoinhibitory effect to a segment of 12 amino acids at positions 287 to 298 in the elongated C terminus, and the single replacement of N296 by D resulted in an activation to 30% of the maximal activity of a C-terminally truncated AtNAS1 protein.

Establishment of a continuous enzyme-coupled photometric NAS assay
Previously published studies provided qualitative evidence for the enzymatic activities of NAS proteins (18,29). For quantitative comparisons among NAS enzyme activities, we developed an enzyme-coupled photometric NAS activity assay, based on a published method for the quantification of SAMdependent methyltransferase activity (30). SAM-dependent methyltransferases catalyze transmethylation reactions using SAM as the donor of a methyl group and release SAH as a byproduct. Their enzyme activities were quantified in a coupled enzyme assay, in which the byproduct SAH is converted to hypoxanthine by the sequential action of SAH Nucleosidase (SAHN/MtnN) and Adenine Deaminase (AdeD). It was thus possible to monitor the activity of SAM-dependent methyltransferases spectrophotometrically by following the Figure 1. Biochemical function of nicotianamine synthase and general strategy of the coupled spectrophotometric assay to quantify NAS enzyme activity. Nicotianamine synthase (NAS) enzymes catalyze the formation of one molecule of nicotianamine from three molecules of SAM (upper part). The byproduct of this reaction, MTA, can be converted to hypoxanthine through two sequential reactions catalyzed by MtnN and AdeD, respectively (lower part). The formation of hypoxanthine can be monitored spectrophotometrically through a decrease in absorbance at 265 nm. AdeD, Adenine Deaminase; MTA, methylthioadenosine; MtnN, Methylthioadenosine Nucleosidase.
decrease in absorbance at the wavelength of 265 nm, which results from the deamination of adenine to hypoxanthine (30). Since many bacterial SAHN enzymes also accept MTA, the byproduct of NAS, as a substrate, we hypothesized that these coupled reactions could also be employed in an NAS enzyme activity assay (Fig. 1). Thus, we cloned the coding sequences of mtnN (encoding 5 0 -MTA/SAH nucleosidase, MtnN) and adeD from the Escherichia coli laboratory strain XL-1 blue, overexpressed them in the same strain, and purified the enzymes as recombinant His 6 -tagged fusion proteins.
Both recombinantly produced enzymes, MtnN-and AdeD, were active in NAS reaction buffer when assayed individually with their respective substrates, as shown by TLC (Fig. S2, A and B). Using both enzymes together, the continuous photometric monitoring of the two-step conversion of MTA to hypoxanthine was possible ( Figs. 1 and S2C).
Next, we tested if this two-enzyme system can be employed for the quantification of NAS activity using A. thaliana NAS1 (AtNAS1). First, AtNAS1 was coincubated in vitro with MtnN only, and the assay solutions were then analyzed by TLC for the formation of adenine and NA ( Fig. 2A). Importantly, NA and adenine were clearly detectable when active MtnN was added but not in the absence of MtnN. This is, to our knowledge, the first time that in a small-scale in vitro NAS assay, NA formation could be demonstrated simply by ninhydrin staining after TLC, without using radio-labeled SAM or time-consuming purification and concentration procedures of the reaction products. In addition, this result confirms the strong inhibitory effect of the byproduct MTA on NAS enzyme activity, which was described earlier (7,28). Finally, newly formed adenine is a sensitive indicator for NAS activity in this coupled assay (but note that small amounts of adenine seem to be present in the samples as degradation product or contamination of SAM, as visible in lanes 1, 4, and 5 of Fig. 2A).
This experimental setup allowed us to qualitatively examine the effects of mutations in NAS proteins, which we introduced in order to identify amino acids that are required for the activity of AtNAS1 (Figs. 1 and S1). We thus substituted amino acids corresponding to those suggested to be critical for catalysis in MtNAS (23), i.e., E77 and Y106 of AtNAS1, and amino acids that are conserved among plant NAS proteins but differ between plant NAS and microbial NAS-like proteins, i.e., C69, E73, or the 206-VGMD-209 motif of AtNAS1 (Fig. 3, A  and B). The failure to produce adenine and NA in the assays of the corresponding mutant AtNAS1 proteins E77Q and Y106F suggested that E77 and Y106 are essential for the catalysis of NA formation by AtNAS1 (Figs. 3C; S3, A-C and S4). This observation was in agreement with the proposed role of the corresponding amino acids in the formation of tNA by MtNAS, based on crystal structures (23). In addition, mutations introduced at two sites that are conserved only among plant NAS proteins and predicted to be positioned near the reaction chamber, i.e., the replacement of E73 by Q or the deletion of the 206-VGMD-209 motif, also rendered AtNAS1 inactive, whereas AtNAS1 activity was insensitive to the C69A substitution (Figs. 3C; S3, A-C and S4).
Next, we additionally included AdeD in our assay to test whether this allows to follow the NAS reaction photometrically. In the presence of AtNAS1, MtnN, and AdeD, there was a measurable decrease in light absorbance at an initial rate that was about 3-fold higher than the background in the absence of AtNAS1 and that remained constant over several minutes (Fig. 2, B and C). By comparison, the minor decrease in light absorbance of a reaction mixture containing the enzymes AdeD and MtnN alone may result from the presence of low   (55). Blue boxes mark positions thought to be essential for the overall reaction mechanism of both NAS and NAS-like proteins, red boxes mark positions near the reaction cavity that are conserved among NAS that produce NA. A yellow and a pink box, respectively, mark an LL motif reported as essential for the in vitro activity of OsNAS2 and a YxxΦ motif proposed to be required for the in planta activity of OsNAS2 (46). B, protein model of AtNAS1 (generated by AlphaFold 2), MtNAS (PDB:3FPE), and AhNAS5 (generated by SWISS-MODEL) with highlighted amino acids mentioned in (A) (44). C, AtNAS1 (4.5 μg) and MtnN (10 μg) were coincubated in the presence of SAM (5 mM) in a total volume of 30 μl. The reaction was started by incubating the mixture at 30 C, and it was stopped immediately (0 h) or 2 h after the start of the reaction by flash-freezing. Aliquots of 5 μl per reaction were separated by TLC, and the products were visualized by UV light (adenine) or ninhydrin staining (nicotianamine). Symbols below the images levels of contaminating adenine in the SAM solution (see Fig. 2B). Importantly, there was a linear correlation between the concentration of the NAS protein in the assay and the initial rate quantified according to the change in A 265 over time, indicating against a limitation by insufficient amounts of MtnN and AdeD present in our assay (Fig. 2D).
Proof of concept for a one-pot in vitro synthesis of NA from ATP and methionine Previously described systems for the biosynthetic production of NA used either crude extracts from plants or recombinant NAS proteins from plants or N. crassa in yeast cells (28,(31)(32)(33). We asked whether we could take advantage of the increased activity of AtNAS1 in the presence of MtnN ( Fig. 2A) for the in vitro production of NA. Combining MetK, a bacterial SAM synthetase which synthesizes SAM from ATP and L-methionine, with both AtNAS1 and MtnN should allow the biosynthesis of NA from these precursors. Initially, we cloned the E. coli metK gene, overexpressed it, and purified the enzyme as His 6 -tagged fusion protein for the production of fresh SAM in order to replace the (presumably partially degraded) commercially available SAM in our NAS assays. Subsequently, we combined MetK with MtnN and AtNAS1 for the one-pot synthesis of NA from ATP and methionine (Fig. S5). We approximated that at least 60 nmol (18 μg) NA were produced within 4 h in a volume of 120 μl in the presence of 60 μg MetK, 12 μg MtnN, and 22 μg AtNAS1. In the past, 4.5 nmol (1.3 μg) NA was produced by 350 μg NASHOR1 (a NAS from Hordeum vulgare) using SAM as a substrate, 60 to 750 μg [ 15 N 3 ]-NA was obtained per 120-ml culture of recombinant Schizosaccharomyces pombe cells producing NcNAS, and an engineered strain of Saccharomyces cerevisiae producing AtNAS2 yielded 766 μg NA g −1 wet biomass (28,31,33).

Phylogenetic analysis and sequence comparison
NAS proteins are encoded in the genomes of some archaea, bacteria, fungi, mosses, and all land plants. To understand how the amino acid sequences of different NAS proteins are related among one another, we conducted a phylogenetic analysis. We constructed a phylogenetic tree by Bayesian inference based on the shared core region of 186 NAS and NAS-like proteins, which we defined as the part of the multiple sequence alignment corresponding to the segment from the first to the 275th amino acid of AtNAS1 in order to exclude any N-or C-terminal extensions present in only a subset of proteins (Supporting Data S1).
NAS proteins from dicotyledonous plants grouped in two distinct monophyletic clades with high statistical support (Figs. 4 and S6; Table S1). One of these clades comprised the well-known NAS proteins (termed here class Ia) of A. thaliana and tomato, for example, and the first characterized NAS proteins (from barley, HvNAS) grouped in the sister clade together with NAS proteins of other monocotyledonous plants (class Ib) (5,7,34). The second, distinct clade of NAS proteins from dicotyledonous plants (class II) comprised annotated NAS proteins from the Ranunculaceae family of basal eudicots, the Apiaceae family in the Asterid clade, as well as the Rutaceae, Rosaceae, Fabaceae, Salicaceae, Euphorbiaceae, and Malvaceae from the Rosid clade. Clade credibility values supported only weakly that class II NAS proteins diverged from class I NAS proteins before the origin of extant gymnosperm NAS proteins (class Ic) and NAS proteins of the basal angiosperm Amborella trichopoda (class I/Ic). With stronger support, class II NAS proteins diverged from class I NAS proteins before monocot and dicot class I NAS proteins diverged from one another. NAS from fungi and mosses (class III), archaea (class IV), and bacteria (class V) diverged earlier from all Spermatophyte NAS proteins, consistent with a published phylogeny of a small set of proteins (20) (Fig. 4).
Our phylogenetic analysis further suggested that the amino acid sequences of the NAS proteins from the mosses Physcomitrium patens and Ceratodon purpureus are more closely related to fungal than to plant NAS proteins (Figs. 4 and S6). This was further supported by the fact that both fungal and moss NAS genes contain an intron at a conserved position that also conserves the phase of the intron in relation to the codons of the coding sequence (Fig. S7), whereas all NAS genes from seed plants are intron-free. The genomes of the mosses Marchantia polymorpha, Sphagnum fallax, and Sphagnum magellanicum, for example, do not contain any NAS genes. The NCBI protein database contains 321 entries for fungal NAS proteins from species in various phyla including Ascomycota, Basidiomycota, Mucoromycota, Zoopagomycota, and Chytridiomycota. These observations support that the NAS gene of P. patens and C. purpureus is of fungal origin and probably arose through a horizontal gene transfer. Yet, we cannot exclude horizontal gene transfer in the opposite direction from moss to fungus.
Multiple sequence alignment revealed that all class Ib and many of the class Ia NAS proteins have an extended C terminus of approximately 45 aa (Figs. 5 and S8, termed here long NAS, about 320 aa long in total), in contrast to almost all other proteins in class Ic and classes II to V (termed here short NAS, length of about 280 aa) (Fig. 4). The basal angiosperm A. trichopoda possesses both a short NAS protein which grouped among the gymnosperm sequences (AmNAS2) and a long NAS protein positioned basally to class Ia and Ib (AmNAS1). Our analysis suggested that there were several independent secondary losses of the extended C terminus in class Ia, as exemplified by the NAS isoforms MetrNAS4 and AhNAS5 (Figs. 4 and 5).

Autoinhibitory effect of the elongated C terminus in purified recombinant plant NAS enzymes
We quantified the activities of a set of short and long NAS proteins using the coupled photometric assay in vitro.
indicate whether (+) or not (−) NA was detectable by MS in the reaction mix at the end point. Full images are provided in Fig. S4. AtNAS1, Arabidopsis thaliana NAS1; MtnN, Methylthioadenosine Nucleosidase; NA, nicotianamine.
Although we had initially used AtNAS1, a long NAS, to establish the enzyme activity assay, its activity was only very low (0.02 ± 0.01 nkat (mg protein) −1 in two independent preparations). By contrast, the activities of NcNAS (0.81 ± 0.01 and 0.51 ± 0.05 nkat (mg protein) −1 in two independent preparations) were the highest, followed by PpNAS (0.37 ± 0.03 and 0.21 ± 0.01 nkat (mg protein) −1 ), both of them being short NAS proteins (Fig. 6). To our knowledge, the activity of the NAS enzyme from the moss P. patens had not been demonstrated earlier. The biosynthesis of NA was confirmed through MS (see Fig. S3, C and D). These data suggested the possibility of an influence of the C-terminal domain on the activity of NAS proteins.
As a representative of a new group of NAS enzymes in the Brassicaceae and an example of a short NAS of secondary origin in class Ia, we tested AhNAS5 activity in vitro. However, its activity, as well as any formed NA, were below our detection limits (see Fig. S3, C and D). Given the conserved sequence changes and the wide distribution of NAS5 homologs in the Brassicaceae, a neo-functionalization for a different,  (Table S1) obtained by Bayesian inference. Short NAS (generally about 280 amino acids in length) end after the core-NAS domain (protein names given in light gray fonts). Long NAS (generally about 320 amino acids in length) comprise additional amino acids at the C terminus in almost all cases (protein names given in black fonts). Colored backgrounds of short protein names reflect taxonomic groups. Colors of lines at branch positions reflect clade credibility values between 0.52 (red) and 1 (green). NAS, nicotianamine synthase. yet unidentified, function is possible (Fig. S9). It should be mentioned that the plant model organism A. thaliana has lost most of its NAS5 gene, but a remaining segment of it encoding a partial protein homologous to the 57 C-terminal amino acids of AhNAS5 is still present in the genome (AGI code AT4G26483). The encoded protein is predicted to have a length of 84 aa and is unlikely to have any NAS or NAS-related activity (Fig. S9C). Furthermore, we tested the in vitro activity of MetrNAS2, a short NAS of class II. Similar to AhNAS5, both MetrNAS2 activity as quantified in our photometric assay and the levels of formed NA as analyzed by LC-MS remained below our detection limits (Fig. S3D).
In order to address a possible influence of the elongated C terminus on the enzyme activity of plant long NAS proteins, we generated C-terminally truncated variants lacking this domain and containing only the core-NAS domain for each of the four NAS homologs of A. thaliana and quantified their enzyme activities (Fig. 7, A and B). As for AtNAS1, the enzyme activities of the full-length AtNAS2, AtNAS3, and AtNAS4 were low or even around the limit of detection. C-terminal truncation caused a strong, 30-and 20-fold increase in the activities of AtNAS1 and AtNAS2 in vitro, respectively. The truncated variants of AtNAS3 and AtNAS4 also showed strongly increased activities that remained below 25% of those of C-terminally truncated AtNAS1 and AtNAS2 (Fig. 7B).
Next, we analyzed the effects of progressive C-terminal deletions on the activity of AtNAS1 (Fig. 7A, lower part, and 7 C). Removal of the C-terminal 13 or 22 amino acids of AtNAS1 did not result in any increase in its enzyme activity. Deletion of 31 amino acids resulted in 42% of the maximal activity that was observed upon deletion of either 34 or 42 amino acids from the C terminus. Compared to the maximal activity, AtNAS1 activity decreased again when it was Cterminally truncated by more than 42 amino acids. C-terminal truncation by 46 residues removed a lysine residue that is highly conserved among plant NAS proteins, and this eliminated enzyme activity (see Figs. 5, 7C and S8). Thus, we attributed the repression of AtNAS1 activity to 12 amino acids, residues 287 to 298 of the C terminus. The corresponding amino acid sequence, TRGCMFMPCNCS, is partly conserved between AtNAS1 and AtNAS2 to AtNAS4, with full conservation of the final four amino acids, CNCS. Of the inhibitory  (Table S1). NAS, nicotianamine synthase. Figure 6. Comparison of enzyme activities of NAS from different species. Enzyme activities were quantified in the coupled spectrophotometric assay (see Fig. 2, B and D). Bars show mean ± SD (n = 3 technical replicates) from two independent protein purifications (black and gray, respectively). Different characters denote statistically significant differences between means (p < 0.01; ANOVA with post hoc Tukey's HSD test). At, Arabidopsis thaliana; NAS, nicotianamine synthase; Nc, Neurospora crassa; Pp, Physcomitrium patens. effect, about 58% appeared to reside in the first three, nonconserved amino acids 287-TRG-289 of the 12-amino-acid autoinhibitory region.

Discussion
A major problem in the biochemical characterization of NAS enzymes had been the low enzyme activity so far, which was probably caused by a strong inhibitory effect of the byproduct MTA on the reaction catalyzed by the NAS enzyme (7,28). To alleviate this, researchers had used only low concentrations of the substrate SAM (20 μM) in NAS activity assays, which, in turn, required sensitive methods to detect the small amounts of NA formed. MTA may already be present as contaminating or breakdown product in commercially available SAM preparations, further interfering with the sensitivity of conventional assays for NAS activity. Here we report that the inclusion of E. coli MtnN in the NAS assay, which irreversibly hydrolyzes MTA to adenine and methylthioribose, substantially increased the amount of NA formed by AtNAS1 in an end point assay ( Fig. 2A). We attribute this effect to the degradation of newly formed MTA by MtnN, which counteracts a gradual accumulation of MTA over time in the reaction mix and thus prevents the inhibition of NAS by MTA. Consequently, the combination of NAS with MtnN allows a simple, semiquantitative NAS assay, in which many samples can be analyzed in parallel for the formation of NA and adenine after separation of the reaction products by TLC (Figs. 3C and S4). By further adding AdeD, which converts adenine to hypoxanthine, the NAS reaction can be continuously monitored by spectrophotometry (Fig. 2, B-D).
It should be noted that the activities of poorly active NAS enzymes, for example, the full-length AtNAS1, remain close to the detection limit of this spectrophotometric assay. In the future, it might be possible to further increase the sensitivity of the assay by including a xanthine oxidase, which acts on hypoxanthine to release hydrogen peroxide for monitoring by colorimetric or fluorometric detection methods using Figure 7. Inhibitory effect of the variable C-terminal domain on the activity of Arabidopsis NAS proteins. A, alignment of the C termini of AtNAS1, AtNAS2, AtNAS3, and AtNAS4 (upper part) and stepwise C-terminally truncated AtNAS1 variants (bottom part). Numbers following ΔC indicate the number of amino acids deleted from the C terminus of AtNAS1. Amino acids fully conserved/similar by comparison to the short NAS PpNAS and NcNAS (or one of them) are highlighted in black/gray (see Fig. 5). B, activity of full-length and C-terminally truncated variants for the four paralogous NAS proteins of Arabidopsis thaliana. The C-terminal end of each truncated variant is marked by a small gray triangle (A, upper part). C, effect of progressive C-terminal deletions on AtNAS1 activity. Bars show mean ± SD (n = 3 technical replicates). Asterisks (B) denote statistically significant differences between means (***p < 0.001, **p < 0.01; Student's t test, two-tailed, equal variance). Different characters (C) denote statistically significant differences between means (p < 0.01; ANOVA with post hoc Tukey's HSD). AtNAS1, Arabidopsis thaliana NAS1; HP, holoprotein; NAS, nicotianamine synthase. additional reagents. This principle is used in commercial kits for quantifying the activities of SAM-dependent methyltransferases (35).
The strong inhibitory effect of MTA on NAS enzymes in vitro raises the question of whether this is of physiological relevance in planta. MTA is not only a byproduct of NA biosynthesis but also of ethylene and polyamine biosynthesis. Plants possess MtnN orthologs which are part in the methionine salvage pathway. Two MTN genes reside in the genome of A. thaliana, but coexpression of MTN and NAS genes has not been reported, which may suggest that the basal MTN activity is sufficient to prevent the accumulation of an excess of MTA in the cytosol. A coordinated upregulation of NAS and MTN genes under iron deficiency was described in rice and wheat (36)(37)(38). NA biosynthetic rates might be higher in these plants than in A. thaliana because strategy II plants use NA also as a precursor for the biosynthesis of phytosiderophores.
A sequence alignment and a phylogenetic analysis including 186 NAS amino acid sequences from a variety of species revealed the existence of two different types of NAS in land plants (Fig. 4, Supporting Data S1). Long NAS proteins, found almost exclusively in class Ia and Ib, differ from short NAS by an additional variable domain at the C terminus which consists of approximately 45 amino acids (Figs. 5 and S8). Monocot plants possess only long NAS (class Ib), whereas both long and short NAS are found in dicot plants (class Ia and II) (Figs. 4 and  S6). Interestingly, class II that exclusively contains short NAS proteins from dicots appears to have arisen before the divergence of the monocot and dicot lineages. Note that long NAS of class Ia and b exhibit some shared C-terminal sequence features (see Figs. 5 and S8; Supporting Data S1), which were absent in the single class II long NAS protein, Ricinus communis NAS3. Indeed, the global multiple sequence alignment clarified that this protein carries an N-terminal instead of a C-terminal extension (Supporting Data S1). Our phylogenetic analysis indicated that the earliest NAS were short NAS proteins and that NAS proteins containing an extended C terminus likely arose early during Angiosperm evolution. This was supported by the fact that only short NAS proteins are found in archaea, bacteria, fungi, and moss. Moreover, our phylogenetic analysis suggests that class II NAS arose early in the Spermatophyte or Angiosperm lineage and were secondarily lost at least in the monocots and several clades of the dicots. Alternatively, class II may have arisen at an early stage during the evolution of the dicot lineage through horizontal gene transfer from an unknown eukaryotic organism (Fig. S6). This remains to be examined in more detail.
We observed a striking difference in activity between a long NAS, namely full-length AtNAS1, and its C-terminally truncated variant comprising only the core-NAS domain, AtNAS1 ΔC42, which exhibited a much higher in vitro enzyme activity (Fig. 7). We obtained similar results for AtNAS2, AtNAS3, and AtNAS4. In accordance with this, enzyme activities of the short NAS proteins PpNAS and NcNAS were also much higher than that of AtNAS1 (Fig. 6). The NAS substrate SAM is an important methyl donor in a variety of cellular biochemical reactions, such as the methylation of DNA and RNA, and SAM is also a substrate for the biosynthesis of polyamines and of the phytohormone ethylene (39,40). Consequently, the consumption of SAM by NAS might require tight regulation. Although the sequences of the C termini are variable, there are short regions containing more conserved amino acids across long NAS proteins of class I. These are the regions corresponding to aa 293 to 304 of AtNAS1 comprising 294-PCNCS-298, with N, K, or R at the position corresponding to N296 that is important for autoinhibition, as well as the region corresponding to aa 309 to 320 of AtNAS1 comprising acidic residues 312-(M/I/L) IEE-315 (Figs. 5, 7 and S8). It is thus possible that the C terminus of other long NAS in class Ia and Ib also has an autoinhibitory role and that posttranslational mechanisms are required to activate these NAS proteins in planta. The C termini contain amino acids that can act in the binding of metal ions, such as cysteine, histidine, glutamate, or methionine. It may be speculated that these NAS proteins are activated upon the binding of metal cations to, or upon the loss of bound metal cations from, the C-terminal extensions. In planta, NAS activation could operate through metal-or protein-binding-dependent or metal-or proteinbinding-independent alterations in three-dimensional structure, posttranslational modification, or proteolytic processing (41,42). The in planta biosynthesis of NAS protein in an enzymatically inactive form and its controlled posttranslational activation would allow a tight control of NAS activity and thus of SAM consumption and MTA production by NAS. Future work will address these possibilities. In addition, it will be important to address whether our findings on AtNAS proteins are indeed representative of long NAS proteins of class I in general.
In the bacterial NAS-like protein S. aureus CntL that comprises merely a core-NAS domain, Luo et al. (43) identified a linker region between its N-and C-terminal domain which is capable of conformational alternation between the open and closed states of SaCntL. The authors suggest that this linker region could have pivotal roles in allowing substrate entry, substrate recognition, and catalysis (43). The linker region corresponds to the amino acids between H96 and P125 of AtNAS1 (Fig. S10A). A comparison of a crystal structure of SaCntL with the structure of AtNAS1 as predicted by Alpha-Fold 2 suggests that the putative linker region is within reach of the C-terminal extension of AtNAS1 ( Fig. S10B; (44)). We observed an increase in enzyme activity of the N296D variant of AtNAS1 (Fig. 7C). Since asparagine can act differently from aspartate in the formation of hydrogen bonds, it is possible, for example, that N296 binds to the linker region of AtNAS1 via hydrogen bonding. As a consequence, the movement of the linker region might be impaired, thus interfering with the catalytic cycle of AtNAS1. Alternatively, hydrogen bonding of N296 might occur to a different region of the NAS protein surface. Further research is required to identify the mechanism of N296-dependent autoinhibition and of how additional C-terminal amino acids contribute to autoinhibition of AtNAS1.
As a representative of Brassicaceae NAS5 proteins, AhNAS5 was inactive in the biosynthesis of NA from SAM despite the absence of an elongated C terminus. Based on an amino acid sequence comparison of AhNAS5 with several NAS and NASlike proteins, there are replacements of conserved amino acids, for example, AtNAS1 C69 to R and E73 to V in the corresponding positions of AhNAS5 (Fig. 3A). Here we observed that E73Q, but not C69A, eliminates AtNAS1 activity (Figs. 3C, S3 and S4). A protein model of AhNAS5 superimposed onto the known protein structure of MtNAS positions these amino acids near the reaction chamber indicating the possibility that these mutations render AhNAS5 inactive (Fig. 3B). Further studies on AhNAS5 could test the activity of AhNAS5 R69C and V73E mutant variants or even address a possible neo-functionalization of AhNAS5 in the biosynthesis of an NA-related molecule.
To our knowledge, the only functionally characterized class II NAS (without a C-terminal extension) is Medicago truncatula NAS2, which is required for symbiotic nitrogen fixation (45). A nas2 loss-of-function Tnt1 insertion mutant had no growth defect under nonsymbiotic conditions. Importantly, overall NA content was not significantly different from the WT in either shoots, roots, or nodules of the mutant. It is thus possible that MetrNAS2 catalyzes the biosynthesis of an NArelated molecule in vivo or possesses a different biological role.
Site-directed mutagenesis of AtNAS1 indicated that amino acids E77 and Y106, which are conserved by comparison to MtNAS, are necessary for AtNAS1 activity. Similarly, we observed that E73 and 206-VGMD-209, which are conserved among plant NAS proteins and predicted to localize near the reaction chamber, are required for AtNAS1 activity. Earlier work had described a dileucine motif at positions 112/113 of OsNAS2 as essential for in vitro activity (46) (see Fig. 3A). These amino acids correspond to positions 116/117 of AtNAS1, which were not examined in this study. Mutation of 107-YVNL-110 of OsNAS2 to AVDL was reported not to affect the in vitro activity, but to interfere with the in planta activity, of OsNAS2 (46) (see Fig. 3). The mutated motif includes the position corresponding to AtNAS1 Y106, which is critical for in vitro activity according to the results presented here. Although these results are overall similar, the differences between our findings on AtNAS1 and earlier findings on OsNAS2 will require further study.
In conclusion, we report here a continuous assay suitable for quantifying NAS enzyme activities. We identify an extended C terminus present in class Ia dicot and class Ib monocot NAS proteins, but not in other classes of NAS proteins. We show that this extended C terminus has an autoinhibitory function in all class Ia NAS proteins of A. thaliana in vitro. By contrast, activities of fungal and moss NAS proteins that lack the C-terminal extension are much higher. We show that class II NAS proteins are present in a number of dicots in addition to class Ia NAS proteins. Class II diverged from class I NAS proteins before the divergence of class I into monocot (Ib) and dicot (Ia) NAS proteins, and class II NAS proteins lack a C-terminal extension. Our results identify that amino acid residues are essential for NAS enzyme activity. The methods and the classification of NAS proteins provided here will facilitate their future functional characterization. The autoinhibitory role of the C terminus of several class I NAS proteins from plants identified here in vitro warrants future studies of the possibility of posttranslational regulation of NAS activity in planta.

Protein modeling
The predicted protein model of AtNAS1 by AlphaFold 2 was identified by searching for the Uniprot number of AtNAS1 (Q9FF79) (44). The predicted model of AhNAS5 was calculated by SWISS-MODEL (47) with default settings, since no model was available in AlphaFold 2. All protein models were visualized and further modified in PyMOL (https://pymol. org/2/, Schrödinger LLC, Version 2.1.1).

Multiple sequence alignments and phylogenetic analyses
The amino acid sequence of NAS1 of A. thaliana (AT5G04950) was used in a blastp search against all proteins from annotated genomes in Phytozome (https://phytozomenext.jgi.doe.gov/, version 12.1.5) (48). In total, 164 full-length NAS sequences from 52 plant species were retrieved using an Expect value (E) cut-off of 10 −90 . Likewise, four NAS proteins were identified in the Ginkgo Database (https://ginkgo. zju.edu.cn/) (49) and seven NAS proteins in the PLAZA 5.0 database (https://bioinformatics.psb.ugent.be/plaza/) (50). Further amino acid sequences of NAS and NAS-like proteins from bacteria, archaea, and fungi were included based on published data (18,20,23,(51)(52)(53). While NAS of Magnaporthe oryzae (XP_003719353.1) was studied in a published phylogenetic analysis (20), NAS of Diaporthe ampelina (KKY38707.1) was newly identified here as described above and included to obtain a more robust tree. A complete list of all NAS and NAS-like proteins used in our analysis can be found in the Table S1, with additional information.
The multiple sequence alignment for the phylogenetic tree shown in Figure 4 was conducted in Mega11 (54) using ClustalW with standard settings. The full alignment consisting of 603 sites (including gaps) was subsequently trimmed at the N-and C-terminal ends to remove highly divergent and gaprich regions. The final alignment comprised 412 sites, corresponding to the amino acids 1 to 275 of AtNAS1 (Supporting Data S1), and displayed using Multiple Align Show (55). Phylogenetic analysis was performed using the program MrBayes (version 3.2.1) (56) with a mixed amino acid rate matrix in two independent Markov Chain Monte Carlo analyses for 2 million generations each. The burn-in was set to 25%, and a 50% majority rule tree was obtained. The tree was visualized in iTOL (version 5) (57).
NAS5 sequences of A. thaliana, Arabidopsis halleri, A. lyrata, Boechera stricta, Brassica rapa, Capsella rubella, C. grandiflora, and Eutrema salsugineum were identified in phytozome by blastp using the amino acid sequences of AtNAS1 and AhNAS5 as queries against the genome of each species. The multiple sequence alignment was conducted in ClustalW with standard settings. Phylogenetic analysis was conducted, and the tree visualized, using the program MegaX (https://www.megasoftware.net/) based on the full alignment. The phylogenetic tree was calculated using the maximum likelihood method and JTT matrix-based model with 500 bootstrap replicates (58).

Constructs for recombinant protein production
The ORFs encoding SAM Synthase (MetK, EC 2.5.1.6), MtnN (EC 3.2.2.9), and AdeD (EC 3.5.4.2) were amplified by PCR from genomic DNA (gDNA) of the E. coli strain XL1-Blue, omitting the stop codon (Tables S2-S4). Amplification of metk and mtnN were carried out with DreamTaq Polymerase (Thermo Fisher Scientific) according to manufacturer's instructions. Amplification of adeD was carried out using Phusion Polymerase (Thermo Fisher Scientific) according to manufacturer's instructions. For T/A cloning in to pGEM-T Easy vector (Table S2), the adeD PCR product was additionally gel-purified (NucleoSpin Gel and PCR Clean-Up Kit, Macherey-Nagel), and 100 ng of purified DNA was subsequently incubated with 0.2 M dATP and 5 U DreamTaq Polymerase (Thermo Fisher Scientific) in DreamTaq buffer (10 μl total volume) according to manufacturer's instructions. All ORFs were cloned into the pGEM-T Easy vector according to manufacturer's instructions. Subsequently, sequences were subcloned into pET-21b (+) ( Table S2) in front of the C-terminal His 6 -tag using the NdeI and XhoI restriction sites, and the construct was verified by Sanger sequencing.
The genomic sequence encoding M. thermautotrophicus NAS (MTH675) was amplified by PCR from a pET101 vector (Table S2) provided by the lab of Pascal Arnoux (CEA), using Phusion Polymerase (Thermo Fisher Scientific) according to manufacturer's instructions (Tables S3 and S4). The amplicon was cloned directly into pET-21b (+) as described above using the NdeI and NotI restriction sites.
Point mutations were introduced into AtNAS1 according to the QuikChange protocol from Stratagene modified as described in Zheng et al. 2004 (59). The previously generated pET-21b-AtNAS1 construct was used as a template. The mutations corresponding to variant proteins AtNAS1 C69A, AtNAS1 E73Q, and AtNAS1 ΔVGMD were introduced using KAPA Hifi-Polymerase (PeqLab Biotechnologie) according to manufacturer's instructions (Tables S3 and S4). The mutations corresponding to variant proteins AtNAS1 E77Q, AtNAS1 Y106F, and AtNAS1 N296D were introduced using a Phusion Polymerase (Thermo Fisher Scientific) according to manufacturer's instructions (Tables S3 and S4). The ORFs encoding C-terminally truncated variants of AtNAS1, AtNAS2, AtNAS3, and AtNAS4 were amplified by PCR using Phusion Polymerase (Thermo Fisher Scientific), the same forward primer and different reverse primers according to manufacturer's instructions (Tables S3 and S4). The amplicons encoding AtNAS1, all C-terminally truncated variants of AtNAS1, AtNAS4, and AtNAS4 ΔC47, were cloned directly into pET-21b (+) using the NdeI and NotI restriction sites as described above. Amplicons encoding AtNAS2 ΔC43 and AtNAS3 ΔC42 and their respective full-length ORFs were cloned into pET-21b (+) using the NheI and NotI restriction sites. The ORF of AhNAS5 was amplified from gDNA of A. halleri (Lan 3.1) via PCR using Phusion Polymerase (Thermo Fisher Scientific) according to manufacturer's instructions (Tables S3 and S4). The amplicon was subcloned first into pGEM-T Easy and subsequently into its destination vector pET-21b (+) as described above using the NdeI and NotI restriction sites (Table S2).

Production and purification of recombinant proteins
We expressed metK, mtnN, and adeD in the E. coli BL21-CodonPlus (DE3)-RIL strain (Table S2). The main bacterial culture for recombinant protein production was inoculated from a 30-ml overnight culture and was grown in 1-l erlenmeyer flasks containing 300 ml of 2× YT media supplemented with 100 μg ml −1 ampicillin and 30 μg ml −1 chloramphenicol at 37 C and 220 rpm for 3 h until the cell culture reached an A 600 of 0.6 to 0.8. The expression was then induced by adding IPTG to a final concentration of 1 mM, and the cells were further grown at 37 C and 220 rpm for 3 h. For harvest, cells were pelleted by centrifugation in a Sorvall SLA-1500 fixed rotor at 4 C and 15,180g for 10 min. The supernatant was discarded, and the pellets were flash-frozen in liquid N 2 and stored at −80 C. On the day of protein purification, 30 ml of lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 1 mg ml −1 lysozyme, pH 7.5) was added to the pellet while thawing on ice for approximately 30 min. Cell lysis was done by ultrasonication (processor UP50H with tip MS3 from Hielscher Ultrasonics) on ice four times for 1 min at an amplitude of 100. The supernatant was applied to a 2-ml column of Ni-NTAagarose at a flow rate of 1 ml min −1 . The column was washed with 25-ml wash buffer (50 mM NaH 2 PO 4 pH 7.5, 300 mM NaCl, 40 mM imidazole, pH 8) and the protein eluted in 2.5-ml elution buffer (50 mM NaH 2 PO 4 pH 7.5, 300 mM NaCl, 250 mM imidazole, pH 8), all with a flow rate of 1 ml min −1 . The elution was applied to a PD-10 column, which was previously equilibrated to the respective storage buffer (MetK; 100 mM Tris-HCl, 1 mM EDTA, pH 8; MtnN: 50 mM potassium phosphate, pH 7; AdeD: 50 mM Tris/HCl, 1 mM DTT, 250 mM NaCl, pH 8). The flowthrough was discarded, and the protein solution was eluted in 3.5 ml storage buffer, aliquoted, and stored at −80 C.
The protocol used for the production of all recombinant NAS and NAS-like proteins was conducted as described above for MetK, MtnN, and AdeD, with the exception that the expression was induced with 0.1 mM IPTG at 30 C for 4 h. In contrast to the purification of MetK, MtnN, and AdeD, NAS and NAS-like proteins were purified via gravity flow. On the day of protein purification, 40 ml of lysis buffer (50 mM NaH 2 PO 4 , 500 mM NaCl, 50 mM imidazole, 0.1% (v/v) Tween 20, pH 7.5, 1 tablet cOmplete, EDTA-free Protease Inhibitor Cocktail (Roche)) was added to the combined pellets from a total of 600 ml expression culture while thawing on ice for approximately 30 min. Cell lysis was achieved by ultrasonication (processor UP50H with tip MS3 from Hielscher Ultrasonics) on ice six times for 30 s at an amplitude of 100. The supernatant was mixed with 1 ml Ni-NTA-agarose slurry (Qiagen) in an overhead rotator (10 rpm) at 4 C for 1 h. The suspension was applied to a column (5-ml polypropylene column, Qiagen) with subsequent washing (5 column volumes, 50 mM NaH 2 PO 4 , 500 mM NaCl, 60 mM imidazole, 0.1 (v/v) Tween 20, pH 7.5) and elution in 2.5 ml elution buffer (50 mM NaH 2 PO 4 , 500 mM, 250 mM imidazole, pH 7.5). The eluate was applied to a PD-10 column, previously equilibrated with 1 mM Tris/HCl, pH 8, and was eluted in storage buffer (1 mM Tris/HCl, pH 8), aliquoted, and stored at −80 C.

TLC for enzyme activity testing
To assess MtnN activity, 0.1 mg ml −1 MtnN and 2.5 mM MTA were mixed in NAS assay buffer (50 mM Tris/HCl, pH 8.7). AdeD activity was assessed by mixing 0.5 mg ml −1 AdeD and 3 mM adenine in NAS assay buffer. Activities of AtNAS1 mutants were assessed by mixing 0.15 mg ml −1 NAS, 0.1 mg ml −1 MtnN, 0.1 mg ml −1 AdeD, and 5 mM SAM in NAS assay buffer. All reactions were carried out in a total volume of 30 μl at 30 C and were stopped by flash freezing in liquid N 2 . Five microliters of each reaction mixture were loaded onto a TLC plate (TLC Silica gel 60 F 254 , Merck). The mobile phase was seven vol-parts of 1-propanol mixed with eight vol-parts of H 2 O. After running for approximately 1 h in a closed TLC chamber at RT, the TLC plate was dried using a blow-dryer and documented under UV light. Additionally, the plate was sprayed with a ninhydrin solution (0.2% w/v in EtOH) using a spray bottle and incubated at 100 C for 5 min to visualize NA.

Spectrophotometry
Spectrophotometric quantifications of NAS enzyme activities were carried out in a 96-well plate (UV-STAR MICRO-PLATE, Greiner) at 37 C in a Syntex multimode reader with monitoring of light absorbance at 265 nm. Each measurement was done simultaneously in three wells (technical replicates) with a final volume of 100 μl per well. Each reaction included 1 mg ml −1 MtnN, 1 mg ml −1 AdeD, 50 mM Tris-HCl, pH 8.7, and 0.15 mg ml −1 NAS unless otherwise mentioned. The mixture was preincubated at 37 C for 3 min, and the enzymatic reaction started by adding 0.125 mM SAM and mixing by pipetting 10% of the total reaction volume up and down five times. The initial rate of change in light absorbance over time was estimated with the setting "Maximize Slope Magnitude" in ICEKAT (https://icekat.herokuapp.com/icekat), and the enzyme activity was calculated in nkat mg −1 protein using Lambert-Beer's law and an extinction coefficient for adenine (λ = 265 nm) of 6700 M −1 cm −1 (30). All NAS proteins examined in this study and their calculated molar masses are listed in Table S6.
Reaction mixtures containing NAS were set up as described above in "TLC for enzyme activity testing." After the reactions, one volume of 100% EtOH was added to the reaction mixture, followed by incubation at −80 C for 2 h, to precipitate any NAS in the sample. Samples were then centrifuged at 15,115g and 4 C for 30 min, the supernatant was transferred into a new reaction tube, and dried in a rotating vacuum concentrator (Genevac Quattro miVac concentrator, Schlee GmbH). The pellet was dissolved in 120-μl ultrapure water containing 0.1% (v/v) formic acid (LC-MS grade) and stored at −20 C. Samples were injected directly into the LC-HRMS system. MS spectra were analyzed for compounds eluting at retention times between 0.31 and 0.59 min (Fig. S11). NA standard was purchased from Toronto Research Chemicals.

Statistics
Student's t tests were conducted in Microsoft Excel. ANOVA and Tukey's HSD were conducted at https://astatsa. com/OneWay_Anova_with_TukeyHSD/.

Data availability
All data are contained in this manuscript and associated Supporting Information.