Abstract

In metal-scarce environments, some pathogenic bacteria produce opine-type metallophores mainly to face the host's nutritional immunity. This is the case of staphylopine, pseudopaline and yersinopine, identified in Staphylococcus aureus, Pseudomonas aeruginosa and Yersinia pestis, respectively. Depending on the species, these metallophores are synthesized by two (CntLM) or three enzymes (CntKLM), CntM catalyzing the last step of biosynthesis using diverse substrates (pyruvate or α-ketoglutarate), pathway intermediates (xNA or yNA) and cofactors (NADH or NADPH). Here, we explored the substrate specificity of CntM by combining bioinformatic and structural analysis with chemical synthesis and enzymatic studies. We found that NAD(P)H selectivity is mainly due to the amino acid at position 33 (S. aureus numbering) which ensures a preferential binding to NADPH when it is an arginine. Moreover, whereas CntM from P. aeruginosa preferentially uses yNA over xNA, the staphylococcal enzyme is not stereospecific. Most importantly, selectivity toward α-ketoacids is largely governed by a single residue at position 150 of CntM (S. aureus numbering): an aspartate at this position ensures selectivity toward pyruvate, whereas an alanine leads to the consumption of both pyruvate and α-ketoglutarate. Modifying this residue in P. aeruginosa led to a complete reversal of selectivity. Thus, the diversity of opine-type metallophore is governed by the absence/presence of a cntK gene encoding a histidine racemase, and the amino acid residue at position 150 of CntM. These two simple rules predict the production of a fourth metallophore by Paenibacillus mucilaginosus, which was confirmed in vitro and called bacillopaline.

Introduction

In metal-scarce environments, bacteria have to use efficient mechanisms for the uptake of metals required for their growth. This is particularly the case for pathogenic bacteria that have to confront the host's immune system. Indeed, the so-called nutritional immunity induces an additional metal limitation by sequestering iron, zinc or manganese to prevent bacterial growth [14]. To face this metal restriction, bacteria have developed metallophores to recover metals. In this context, nicotianamine-like metallophores have been identified in some bacteria as playing an important role in metal acquisition strategies. Staphylopine, pseudopaline and yersinopine are the three examples currently known and recently identified in Staphylococcus aureus [5], Pseudomonas aeruginosa [6,7] and Yersinia pestis [8], respectively. Depending on the species, the biosynthesis of these nicotianamine-like metallophores occurs in two or three steps. When it is present, such as in S. aureus, CntK, a histidine racemase, transforms l-histidine (l-His) into d-histidine (d-His). CntL, a nicotianamine synthase-like, adds an aminobutyrate moiety coming from S-adenosyl methionine (SAM) on the amino group of its second substrate (l-His in P. aeruginosa and Y. pestis or d-His in S. aureus) to form a pathway intermediate (noted yNA when using l-His, xNA when using d-His). Finally, CntM, an enzyme belonging to the opine dehydrogenase family, condenses the pathway intermediate with an α-ketoacid (pyruvate in the case of S. aureus and Y. pestis or α-ketoglutarate (αKG) in the case of P. aeruginosa) using NAD(P)H to form an opine-type metallophore (Figure 1).

Differences in the staphylopine, pseudopaline, and yersinopine biosynthetic pathways.

Figure 1.
Differences in the staphylopine, pseudopaline, and yersinopine biosynthetic pathways.

Adapted from [5,6,8].

Figure 1.
Differences in the staphylopine, pseudopaline, and yersinopine biosynthetic pathways.

Adapted from [5,6,8].

Opine dehydrogenase catalyzes the NAD(P)H-dependent reductive condensation of the amino group of an amino acid with an α-ketoacid to produce an N-(carboxyalkyl) amino acid, also known as an opine, which exhibits either (l,l) or (d,l) stereochemistry [9]. The variety of amino acids and α-ketoacids that can be used as substrates by opine dehydrogenases results in diverse products. For example, the octopine dehydrogenase catalyzes the production of octopine, lysopine or histopine via reductive condensation of pyruvate and l-arginine, l-lysine and l-histidine, respectively [1013]. Leucinopine, asparaginopine and glutaminopine are other examples of opines produced via reductive condensation of α-ketoglutarate with l-leucine, l-asparagine and l-glutamine, respectively [14,15]. In addition to the diversity of substrates, opine dehydrogenases are also distinguished by their biological roles. Indeed, in some marine invertebrates, opine dehydrogenases participate in the anaerobic metabolism by insuring the last step of anaerobic glycolysis pathway, therefore, participating in the propelling of these animals [16,17]. In plants, diverse opines are found inside crown gall tumors (for example nopaline, agropine, octopine, mannopine or d-l and l-l-succinamopine) that are induced by plant pathogenic bacteria such as Agrobacterium tumefaciens [18,19] In this case, the opines serve as nutrients conferring selective growth advantages to the opine-producing and opine-utilizing microorganisms [2022].

Opine dehydrogenases are also involved in the biosynthesis of nicotianamine-like metallophores in some bacteria and their substrate specificity results in the production of diverse metallophores. The activity of these opine dehydrogenases from S. aureus, P. aeruginosa and Y. pestis (respectively called SaCntM, PaCntM and YpCntM) has been described: SaCntM uses NADPH and pyruvate with xNA to produce staphylopine [5], YpCntM uses pyruvate and NADPH with yNA to produce yersinopine [8], and PaCntM uses NAD(P)H and α-ketoglutarate with yNA to produce pseudopaline [6,7]. Therefore, substrate specificity (and eventually stereospecificity in the case of xNA vs yNA) of CntM leads to the production of diverse opine-type metallophores. These metallophores are involved in metal acquisition strategies, with metal specificity depending on growth media. Staphylopine could transport copper, nickel, cobalt, zinc and iron [5,23,24], and participates in zinc uptake in zinc-scarce environments [23,25]. Pseudopaline is involved in nickel uptake in minimal media, whereas it is responsible for zinc uptake in zinc-scarce environment [6,26]. For both S. aureus and P. aeruginosa, the literature reports a link between the production of opine-type metallophores and bacterial infection. For example, in S. aureus, the staphylopine receptor CntA plays an important role in the optimal function of urease activity and is ultimately linked to its virulence as the deletion of the gene encoding this substrate binding protein leads to a decrease in murine bacteremia and urinary tract infections [24]. In P. aeruginosa, transcriptomic analyses showed that the biosynthetic genes for pseudopaline are overexpressed in burn wound infections in humans [27]. This overexpression would make it possible to bypass the metal limitations put in place during nutritional immunity. Moreover, the pseudopaline exporter CntI plays an important role in the survival and growth of P. aeruginosa in cystic fibrosis airway: deletion of the gene encoding this exporter resulted in an attenuation of this respiratory infection [28]. Similarly, the exporter of staphylopine was found to be important for fitness in abscesses, even before staphylopine discovery [29]. Concerning Y. pestis, no data indicate a link between yersinopine production and virulence at this time and the discovery of this metallophore is so far restricted to in vitro studies [8].

In an effort to understand the diversity of bacterial opine-type metallophores, we studied the substrate specificity of CntM in S. aureus and P. aeruginosa by combining bioinformatic and structural analyses with chemical synthesis and enzymatic studies. We found that a single amino acid residue is responsible for the preferential binding of NADPH, and a single residue is involved in the specificity toward pyruvate or α-ketoglutarate. This finding led us to establish simple rules involved in opine-type metallophore diversity and prompted us to search into available genomes for a bacteria possessing a cntK homolog together with a cntM gene predicted to use α-ketoglutarate. This exploration ultimately led to the discovery of a new opine-type metallophore called bacillopaline in Paenibacillus mucilaginosus.

Materials and methods

Bioinformatic analyses

SaCntM (sav2468) protein sequence was analyzed by searching for homologs using Psi-BLAST search [30] through the NCBI databases (National Center for Biotechnology Information, Bethesda, Maryland, U.S.A.) and the Pfam database (European Bioinformatics Institute, Hinxton, England, U.K.). Sequence alignment was done using the Muscle program [31] with defaults criteria from Jalview (version 2.10.3) [32]. Residues were colored following the Clustal coloring scheme with intensity modified by a conservation color increment of 30%. Gene synteny was inspected using the MaGe MicroScope web interface [33] added to sequence alignment analysis.

Cloning, expression and purification of proteins

The gene encoding the SaCntM protein was cloned in pET-SUMO and pET-101 [5,34] and the one encoding the PaCntM protein was cloned in pET-TEV [6]. Similarly, the genes encoding the PmCntL and PmCntM proteins were cloned in pET-TEV from genomic DNA. The primers used for these constructions are listed in Supplementary Table S1. After co-transformation with plasmid pRARE (encoding for rare codon in Escherichia coli), E. coli BL21 strains were aerobically cultivated with horizontal shaking in LB media supplemented with appropriate antibiotics (kanamycin at 50 µg ml−1 for pET-SUMO and pET-TEV, ampicillin at 50 µg ml−1 for pET-101 and chloramphenicol at 25 µg ml−1 for pRARE). These strains were grown in diverse conditions (37°C or 16°C, with or without induction of protein expression by the addition of 0.1 mM IPTG when the OD of the culture was ∼0.6; Supplementary Table S2). After overnight growth, cells were recovered by centrifugation at 5000g for 20 min at 4°C. Cells were resuspended in buffer A and disrupted using a Constant cell disruption system operating at 1.9 Kbar. Cell debris was removed by centrifugation at 8000g for 20 min. The supernatants were centrifuged at 100,000g for 45 min at 4°C to remove cell wall debris and membrane proteins. The resulting soluble fraction was purified by batch using a nickel-charged affinity resin (Ni-NTA Agarose resin, ThermoFisher Scientific). The proteins were eluted stepwise with imidazole (15 mM wash, 250 mM or 500 mM elution). Collected fractions were transferred into imidazole-free buffer B (see Supplementary Table S2 for details on the buffers used and Supplementary Figure S8 for the monitoring of protein purifications).

Site-directed mutagenesis

Site-directed mutagenesis were performed according to the standard protocol from the QuickChange II Site-Directed Mutagenesis kit (Agilent Technologies). The only difference was that E. coli DH10β Competent Cells were used instead of XL1-Blue Supercompetent Cells for transformations. Selection of colonies was done after spreading on LB plate supplemented with appropriate antibiotics. Plasmid pET-SUMO containing the gene encoding the SaCntM protein and plasmid pET-TEV containing the gene encoding the PaCntM protein were used as mutagenesis templates for the D150A and A153D substitutions, respectively. For the R33H substitution from SaCntM, mutagenesis was performed from plasmid pET-101 containing the gene encoding the SaCntM protein as a template. Primer pairs were designed for single substitutions and were then synthesized by Eurofins Genomics (Supplementary Table S1).

Chemical synthesis of xNA and yNA

The chemical synthesis of xNA has recently been described [34]. The same strategy was used for the chemical synthesis of yNA, although l-His-OMe was used instead of d-His-OMe as the starting material. Purified intermediates and their characterization are described in the Supplementary Materials.

CntM activity assay

Enzymatic reactions of CntM were performed at 28°C in microplates and with purified proteins. They were carried out in a reaction volume of 100 µl in buffer C (50 mM BisTrisPropane, 100 mM NaCl, pH = 7 for SaCntM and PmCntM or pH = 8 for PaCntM) containing 2 or 5 µg of enzyme, 0.2 mM of NADH or NADPH (Sigma–Aldrich), 0.2 mM of xNA or yNA (chemically synthesized), and 1 mM of pyruvate or α-ketoglutarate (Sigma–Aldrich), unless otherwise stated. The absorbance at 340 nm was measured using a microplate reader (Infinite 200 Pro; Tecan) to follow the oxidation of NADPH illustrating the progress of the reaction. Activities were calculated from the initial rate and the amount of enzyme used. Kinetic parameters were estimated according to the Michaelis–Menten kinetic without (1) or with the substrate inhibition model (2) using SigmaPlot. These values were used to plot a fit on the experimental data. (Vm: Maximum velocity; KM: Michaelis constant; [S]: substrate concentration; Ki: Inhibition constant). 
formula
(1)
 
formula
(2)

Fluorescence resonance energy transfer (FRET) studies

Fluorescence studies were performed using a Cary Eclipse spectrophotometer (Agilent). The FRET experiment was done using a protein concentration of 5 μM and an excitation wavelength at 280 nm (tryptophan excitation). The emission at 340 nm was transferred to the NAD(P)H and the signal was recorded between 400 and 500 nm. Five spectra were averaged in order to increase the signal to noise ratio.

Activity assay followed by TLC

The assay consisted in incubating the purified enzymes at a final concentration of 2.5 μM and using carboxyl-[14C]-labeled SAM (2.5 μM), NADPH or NADH (30 μM), l- or d-histidine (10 μM) and α-ketoacid (pyruvate or α-ketoglutarate; 1 mM). The total volume was 100 µl in buffer D (50 mM of Hepes, 1 mM of DTT, 1 mM of EDTA, pH = 9). The mixtures were incubated for 30 min at 28°C. The reactions were stopped by adding ethanol to a final concentration of 50% (v/v) and the products were then separated by a thin-layer chromatography. An aliquot of 10 µl of the reaction mixtures were spotted on HPTLC (High-Performance TLC) Silica Gel 60 Glass Plates (Merck KGaA), and the plates were developed with a phenol:n-butanol:formate:water (12 : 3 : 2 : 3 v/v) solvent system. These separation parameters were used in the initial biochemical characterization of plant nicotianamine synthase [35] and of staphylopine and pseudopaline [5,6]. HPTLC plates were dried and exposed to a [14C]-sensitive imaging plate for one day. Imaging plates were then scanned on a Typhoon FLA 7000 phosphorimager (GE Healthcare).

Cell culture conditions of Paenibacillus mucilaginosus

Paenibacillus mucilaginosus, sub sp. 1480D (from DSMZ collection) was grown by inoculating stock bacteria into 20 ml of sterile medium (TSB 1/10) at 30°C for 48 h. Genomic DNA was extracted using the protocols and pretreatments for Gram-positive bacteria from DNeasy Blood and Tissue kit (Qiagen).

Results and discussion

Specificity toward NADH or NADPH

The structure of CntM has been solved in a binary/tertiary complex with NADPH or NADPH and xNA, revealing the residues that are involved in the complex formation and building the active site [8,34] (pdb code 6C4T and 6H3D, respectively). Focusing on the NADPH binding site of SaCntM, we observed that the phosphate group of NADPH is sandwiched by the side chains of two positively charged residues (R33 and K39) that are rather conserved in the CntM family (Figure 2A). However, a sequence alignment of nine CntM homologs shows that R33 residue in SaCntM is conserved in YpCntM, but replaced by a histidine in PaCntM (Figure 2B). As it is known that SaCntM and YpCntM specifically use NADPH [5,8], while PaCntM can use NADH or NADPH [68], we hypothesized that the nature of the amino acid residue at this position could determine the NADH/NADPH selectivity by CntM.

CntM structure, sequence conservation around NADPH and operon diversity highlight protein selectivity.

Figure 2.
CntM structure, sequence conservation around NADPH and operon diversity highlight protein selectivity.

(A) Details of the NADPH-binding site on SaCntM with side chains of residues located up to 4 Å around the NADPH represented in stick. The protein is colored from N-terminus (blue) to C-terminus (red). The pdb code used for this figure corresponds to 6H3D and shows two alternative positions for the adenine moiety of NADPH. Dashed lines indicate hydrogen bonds and salt bridges. The nicotinamide moiety is where the reductive condensation of xNA with pyruvate occurs and is involved in a putative hydrogen bond with D150 side chain. (B) Sequence alignment of nine CntM protein sequences from bacteria. Threshold for the Clustal coloring scheme corresponds to 30% sequence conservation as defined in Jalview. Arrows point to residues involved in NADPH/NADH selectivity and pyruvate/α-ketoglutarate selectivity. A weblogo representation of the entire UniProt family available from Pfam (318 entries) is drawn below the alignment, showing the conservation of position equivalent to R33 (R: 90%, H: 6% and K: 4%) and D150 (D: 72%, A: 25% and T: 3%) (C) Genomic organization of the biosynthetic genes of four different cnt operons in bacteria.

Figure 2.
CntM structure, sequence conservation around NADPH and operon diversity highlight protein selectivity.

(A) Details of the NADPH-binding site on SaCntM with side chains of residues located up to 4 Å around the NADPH represented in stick. The protein is colored from N-terminus (blue) to C-terminus (red). The pdb code used for this figure corresponds to 6H3D and shows two alternative positions for the adenine moiety of NADPH. Dashed lines indicate hydrogen bonds and salt bridges. The nicotinamide moiety is where the reductive condensation of xNA with pyruvate occurs and is involved in a putative hydrogen bond with D150 side chain. (B) Sequence alignment of nine CntM protein sequences from bacteria. Threshold for the Clustal coloring scheme corresponds to 30% sequence conservation as defined in Jalview. Arrows point to residues involved in NADPH/NADH selectivity and pyruvate/α-ketoglutarate selectivity. A weblogo representation of the entire UniProt family available from Pfam (318 entries) is drawn below the alignment, showing the conservation of position equivalent to R33 (R: 90%, H: 6% and K: 4%) and D150 (D: 72%, A: 25% and T: 3%) (C) Genomic organization of the biosynthetic genes of four different cnt operons in bacteria.

We, therefore, sought to determine the role of this residue in NAD(P)H selectivity by replacing this arginine by a histidine in a SaCntM:R33H variant. Because the histidine could either be neutral or positively charged at pH above or below its pKa, the binding of NAD(P)H was followed at two pHs (6.0 and 8.5). We found that the WT enzyme, whatever the pH, still preferentially binds NADPH over NADH (Figure 3). On the contrary, the R33H mutant behaves as the WT at pH 6.0, whereas it loses its preferential binding property at a higher pH. This shows that a histidine at this position could serve as a selective residue, stabilizing NADPH at acidic pH when the imidazole ring of histidine is positively charged, and favoring NADH binding when histidine is neutral at basic pH, overall explaining the difference of selectivity in the literature [68]. In other words, at acidic pH the histidine side chain of the R33H mutant is charged and behaves like the WT, whereas at basic pH this histidine is neutral, it could not form a salt bridge with the phosphate of NADPH and therefore could not discriminate between NADPH and NADH. Interestingly, we noted that Fictibacillus arsenicus possesses a histidine residue at the conserved position equivalent to K39 in S. aureus, suggesting the same possibility of preferential NADH binding at basic pH.

Titration of NADPH (black circles) and NADH (white circles) binding to SaCntM (5 µM) at pH = 6.0 or pH = 8.5 followed by fluorescence energy transfer between tryptophan excitation (280 nm) and NAD(P)H emission (450 nm).

Specificity toward xNA or yNA

In vivo, CntM uses the product of CntL i.e. xNA or yNA depending on the species. However, using enzymatically produced xNA or yNA, McFarlane et al. [8] suggested that SaCntM could use both diastereoisomers. Activities from S. aureus (SaCntM) and from P. aeruginosa (PaCntM) were therefore compared for their ability to use chemically synthesized xNA and yNA as substrates (Figure 4). The chemical synthesis of xNA was recently reported [34] and we were able to synthesize yNA by following the same approaches (see the experimental procedures and the supplementary materials). Reactions were then performed in vitro using purified proteins and a concentration range of xNA and yNA with a saturating concentration of other substrates: 0.2 mM of NADPH and 1 mM of pyruvate (when evaluating SaCntM) or α-ketoglutarate (when evaluating PaCntM). Overall, we found that the activity of CntM from S. aureus and P. aeruginosa toward yNA and xNA was different. In the case of SaCntM, although xNA is used in vivo to produce staphylopine, its activity is higher when using yNA in vitro. Indeed, the kcat for the reaction with yNA is ∼2-fold higher than the one with xNA (3.18 s−1 and 1.67 s−1, respectively). However, the KM for the reaction with yNA is ∼2-fold higher than the one with xNA (47 µM and 23 µM, respectively), leading to a catalytic efficiency (kcat/KM) of the same order of magnitude for the two substrates (Table 1). In the case of PaCntM, the enzyme uses yNA to produce pseudopaline in vivo. In agreement with this, we found that the reaction with yNA is more efficient than the one with xNA, the reaction with yNA exhibiting a catalytic efficiency ∼10-fold higher than with xNA (29 618 M−1 s−1 and 2918 M−1 s−1, respectively).

Activity profile of SaCntM and PaCntM using variable concentrations of xNA (black circle) and yNA (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and 1 mM of pyruvate when evaluating SaCntM, or 0.2 mM of NADH and 1 mM of α-ketoglutarate when evaluating PaCntM.

Figure 4.
Activity profile of SaCntM and PaCntM using variable concentrations of xNA (black circle) and yNA (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and 1 mM of pyruvate when evaluating SaCntM, or 0.2 mM of NADH and 1 mM of α-ketoglutarate when evaluating PaCntM.

The data points are means of three replicates with standard deviations. The fits are made using the Michaelis–Menten model considering (continuous line) or not (dashed line) a substrate inhibition.

Figure 4.
Activity profile of SaCntM and PaCntM using variable concentrations of xNA (black circle) and yNA (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and 1 mM of pyruvate when evaluating SaCntM, or 0.2 mM of NADH and 1 mM of α-ketoglutarate when evaluating PaCntM.

The data points are means of three replicates with standard deviations. The fits are made using the Michaelis–Menten model considering (continuous line) or not (dashed line) a substrate inhibition.

Table 1
Kinetic parameters of SaCntM and PaCntM activities established for a concentration range of xNA and yNA with fixed concentrations of other substrates: 0.2 mM of NADPH and 1 mM of pyruvate when evaluating SaCntM, and 0.2 mM of NADH and 1 mM of α-ketoglutarate when evaluating PaCntM
Protein xNA vs yNA KM (µM) Ki (mM) kcat (s−1kcat/KM (M−1 s−1
SaCntM yNA 47 ± 9 2.3 ± 0.8 3.18 ± 0.25 67 584 
SaCntM xNA 23 ± 3 NA 1.67 ± 0.04 72 711 
PaCntM yNA 21 ± 7 2.4 ± 1.0 0.62 ± 0.06 29 618 
PaCntM xNA 54 ± 20 NA 0.16 ± 0.02 2918 
Protein xNA vs yNA KM (µM) Ki (mM) kcat (s−1kcat/KM (M−1 s−1
SaCntM yNA 47 ± 9 2.3 ± 0.8 3.18 ± 0.25 67 584 
SaCntM xNA 23 ± 3 NA 1.67 ± 0.04 72 711 
PaCntM yNA 21 ± 7 2.4 ± 1.0 0.62 ± 0.06 29 618 
PaCntM xNA 54 ± 20 NA 0.16 ± 0.02 2918 

The data and the standard errors associated with were generated by SigmaPlot according to the Michaelis–Menten model with or without substrate inhibition. (NA: Not Applicable; Vm: Maximum velocity; KM: Michaelis constant; Ki: Inhibition constant; kcat: Catalytic constant (or turnover number); kcat/KM: Catalytic efficiency).

In the past decades, several opine dehydrogenases have been studied and their biosynthetic reactions generally exhibit a substrate stereospecificity toward the amino group with most enzymes using the l-stereoisomer as substrate [9]. For example, substrate stereospecificity has been outlined for the octopine dehydrogenase from Pecten maximus [12] and the structure of this enzyme shows a negatively charged cavity acting as a ‘charge ruler', which favors l-arginine binding. Here, we found that even if SaCntM uses xNA in vivo, it is also capable of using yNA in vitro. This trend confirms the one outlined by McFarlane et al. [8] using enzymes from different species. Under their experimental conditions, PaCntM exhibited a kcat of 0.016 s−1 in the presence of xNA, which was biosynthesized with SaCntL. Accordingly, we found that PaCntM preferentially used yNA whether in vitro or in vivo. All these data, therefore, suggest that there is a substrate stereospecificity in the case of PaCntM, which is not found in SaCntM. We further noted that using yNA but not xNA, a drop in enzyme activity is visible at high substrate concentrations, which suggests a mechanism of substrate inhibition (Figure 4). Indeed, the fits made with a substrate inhibition model added to the Michaelis–Menten kinetic better cover the experimental data (Figure 4; plain lines). However, this substrate inhibition is not visible when using xNA as substrate.

Specificity toward pyruvate or α-ketoglutarate

To find amino acid residues involved in the pyruvate/α-ketoglutarate selectivity by CntM, we searched for residues located in the vicinity of the active site (i.e. the nicotinamide moiety) and conserved in species known to use pyruvate (S. aureus and Y. pestis) but differing in species known to use α-ketoglutarate (P. aeruginosa). This pointed to the aspartic acid residue at position 150 in SaCntM, which is also present in YpCntM but replaced by an alanine in PaCntM (corresponding to residue 153 in this enzyme). Because this residue is either an aspartic acid or an alanine, we hypothesized that the nature of the amino acid residue at this position would determine the pyruvate/α-ketoglutarate selectivity by CntM. This postulate was also proposed by McFarlane et al. [8]. To test the role of this amino acid in pyruvate/α-ketoglutarate selectivity, we replaced the aspartic acid by an alanine in SaCntM (D150A variant) and did the opposite mutation in PaCntM (A153D). The activities of these proteins were then compared for their ability to use pyruvate and α-ketoglutarate as substrates (Figure 5 and Supplementary Figure S8 for the monitoring of protein purifications). The reactions were performed using varying pyruvate and α-ketoglutarate concentrations with a fixed concentration of other substrates: 0.2 mM of NAD(P)H and 0.2 mM of xNA (when using SaCntM) or yNA (PaCntM). As a control, we verified that the mutation did not affect the binding of the NADPH on SaCntM (Supplementary Figure S9).

Activity profile of SaCntM (WT and D150A mutant) and PaCntM (WT and A153D mutant) using variable concentrations of pyruvate (black circle) and α-ketoglutarate (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and xNA when evaluating SaCntM, and 0.2 mM of NADH and yNA when evaluating PaCntM.

Figure 5.
Activity profile of SaCntM (WT and D150A mutant) and PaCntM (WT and A153D mutant) using variable concentrations of pyruvate (black circle) and α-ketoglutarate (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and xNA when evaluating SaCntM, and 0.2 mM of NADH and yNA when evaluating PaCntM.

The data points are means of three replicates with standard deviations. The fits are made using the Michaelis–Menten model.

Figure 5.
Activity profile of SaCntM (WT and D150A mutant) and PaCntM (WT and A153D mutant) using variable concentrations of pyruvate (black circle) and α-ketoglutarate (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and xNA when evaluating SaCntM, and 0.2 mM of NADH and yNA when evaluating PaCntM.

The data points are means of three replicates with standard deviations. The fits are made using the Michaelis–Menten model.

With regard to SaCntM, we confirmed that the WT was only able to use pyruvate, with a kcat of 1.0 s−1 and a KM of 51 µM, resulting in a catalytic efficiency of 19 592 M−1 s−1 (Figure 5 and Table 2). On the contrary, the D150A variant of SaCntM could use both pyruvate and α-ketoglutarate as substrates. Indeed, although the maximum activity is not reached within the concentration range tested, the kinetic parameters calculated for both pyruvate and α-ketoglutarate are in the same order of magnitude. This single substitution, therefore, led to a decreased activity when using pyruvate but most of all, significantly led to an increased activity when using α-ketoglutarate. Even if we take a lower limit for the kcat of 1.65 s−1 for α-ketoglutarate, this would correspond to a more than 40-fold increased as compared with the WT SaCntM. We then investigated whether the opposite mutation in PaCntM would trigger the opposite effect on substrate selectivity. Here again, we confirmed that the WT enzyme could only use α-ketoglutarate with a kcat of 0.27 s−1 and a KM of 133 µM. Strikingly, we found that the A153D substitution in PaCntM led to a complete reversal of selectivity, with the variant only being able to use pyruvate and unable to use α-ketoglutarate anymore. This mutation is indeed accompanied by a more than 20-fold increase in kcat for pyruvate and ∼30-fold decrease for α-ketoglutarate, i.e. a complete switch in substrate specificity. Consequently, the substrate specificity of CntM toward pyruvate or α-ketoglutarate is mainly governed by this single amino acid (position 150 in SaCntM or 153 in PaCntM): an aspartate ensures the selection of pyruvate, whereas an alanine leads to α-ketoglutarate specificity. To our knowledge, this is the first example showing that substrate specificity might be tuned in opine/opaline dehydrogenases family. There are, however, some examples of redesigned substrate specificity in dehydrogenases such as the production of a highly active malate dehydrogenase starting from lactate dehydrogenase [36]. In this case, the Gln102Arg mutation of Bacillus stearothermophilus lactate dehydrogenase led to a shift in kcat/KM with malate so that it is equal that of native lactate dehydrogenase for its natural substrate. Examples of site-directed mutagenesis studies in the opine/opaline family were centered on the catalytic residues and all led to decreased activities toward substrates and none explored the putative α-ketoacid specificity [12,37]. Finally, there is also an example of opine-binding proteins in which a single residue is a key to the recognition of pyruvate vs αKG opines (i.e. octopines vs nopaline) [38].

Table 2
Kinetic parameters of SaCntM (WT and D150A mutant) and PaCntM (WT or A153D mutant) activities established for a concentration range of pyruvate and αKG with fixed concentrations of other substrates: 0.2 mM of NADPH and xNA when evaluating SaCntM, and 0.2 mM of NADH and yNA when evaluating PaCntM
Protein Pyruvate vs αKG KM (µM) kcat (s−1kcat/KM(M−1 s−1
SaCntM pyruvate 51 ± 6 1.00 ± 0.02 19 592 
SaCntM αKG ND 0.04 ± 0.01 ND 
SaCntM:D150A pyruvate 681 ± 115* 1.23 ± 0.08* 1807* 
SaCntM:D150A αKG 2 040 ± 431* 1.65 ± 0.22* 809* 
PaCntM pyruvate ND 0.02 ± 0.01 N.D. 
PaCntM αKG 133 ± 39 0.27 ± 0.02 2058 
PaCntM:A153D pyruvate 14 986 ± 2964* 0.46 ± 0.07* 31 
PaCntM:A153D αKG ND 0.01 ± 0.01 ND 
Protein Pyruvate vs αKG KM (µM) kcat (s−1kcat/KM(M−1 s−1
SaCntM pyruvate 51 ± 6 1.00 ± 0.02 19 592 
SaCntM αKG ND 0.04 ± 0.01 ND 
SaCntM:D150A pyruvate 681 ± 115* 1.23 ± 0.08* 1807* 
SaCntM:D150A αKG 2 040 ± 431* 1.65 ± 0.22* 809* 
PaCntM pyruvate ND 0.02 ± 0.01 N.D. 
PaCntM αKG 133 ± 39 0.27 ± 0.02 2058 
PaCntM:A153D pyruvate 14 986 ± 2964* 0.46 ± 0.07* 31 
PaCntM:A153D αKG ND 0.01 ± 0.01 ND 

The data and the standard errors associated with were generated by SigmaPlot according to the Michaelis–Menten model. (ND: Not Determined).

*

Because the maximum enzyme activity is not reached, these values are not well defined and must be taken with caution.

Identification of a novel nicotianamine-like metallophore

Having established the molecular determinant for the α-ketoacid selectivity of CntM, we propose two simple rules governing the production of nicotianamine-like metallophores: (1) the presence or absence of a cntK homolog, which may or may not lead to the production of d-His from l-His, respectively (then used by CntL to produce xNA or yNA, respectively), and (2) the presence of an aspartate or an alanine at position 150 (S. aureus numbering), which results in pyruvate or α-ketoglutarate incorporation, respectively. Applying these rules, we searched for a species capable of producing the missing metallophore variant, which would use xNA and incorporate α-ketoglutarate, i.e. a species possessing a cntK homolog and an alanine at position responsible for α-ketoglutarate selectivity. Digging into available genomes in silico, we identified Paenibacillus mucilaginosus as a good candidate. Indeed, this species carries an A153 in CntM (as P. aeruginosa) and possesses a cntK in its cnt operon (as S. aureus) (Figure 2B,C).

To check the validity of this hypothesis, genes encoding the CntL and CntM from P. mucilaginosus (PmCntL and PmCntM, respectively) were amplified from genomic DNA, cloned in expression vectors, purified and used to determine the substrates they consumed in vitro (Figure 6). Enzyme activities were first assayed using TLC separation, using carboxyl-[14C]-labeled SAM substrate. In this assay, [14C]-labeled SAM shows a characteristic profile with one strong band and two other bands of much lower intensity (Figure 6A). The incubation of [14C]-labeled SAM with PmCntL led to another prominent band in the presence of d-His, which was not found using l-His. This novel band migrated below the SAM band and therefore corresponds to the xNA intermediate. This indicates that PmCntL uses d-His and not l-His, and further validates the link between the presence of a cntK gene and the use of d-His by CntL. We then tested the substrate specificity of PmCntM by co-incubating both PmCntL and PmCntM together with diverse substrates: pyruvate or α-ketoglutarate with NADH or NADPH. When using NADPH and d-His with either pyruvate or α-ketoglutarate, we neither detected the SAM nor xNA pattern, but found a novel band migrating just above the SAM in the presence of pyruvate (which corresponds to staphylopine) and just below the SAM in the presence of α-ketoglutarate. Surprisingly, we thus found that PmCntM could use both pyruvate and α-ketoglutarate in this TLC assay. However, TLC experiments were run using a defined incubation time (30 min), which might hinder differences in enzyme efficiency. We, therefore, determined the enzymatic parameters of PmCntM using purified protein and a concentration range of pyruvate and α-ketoglutarate, with a fixed concentration of other substrates: 0.2 mM of NADPH and 0.2 mM of xNA (Figure 6B). As a result, we found that the catalytic efficiency is 10-fold better for α-ketoglutarate than for pyruvate (10 367 M−1 s−1 and 1016 M−1 s−1, respectively; Table 3). These data, therefore, suggest that our hypothesis is valid: CntL and CntM from P. mucilaginosus produce an additional variant of opine-type metallophore with d-His, NADPH and α-ketoglutarate. This metallophore has been called bacillopaline as it belongs to the opaline family (Figure 6C). Moreover, we found that, in addition to using both pyruvate and α-ketoglutarate, CntM from P. mucilaginosus is also able to use both xNA and yNA (Supplementary Figure S10, Supplementary Table S3).

Activity of CntL and CntM from Paenibacillus mucilaginosus.

Figure 6.
Activity of CntL and CntM from Paenibacillus mucilaginosus.

(A) TLC separation of reaction products incubating carboxyl-[14C]-labeled with a purified enzyme (PmCntL and PmCntM) and various substrates (l-His (L) or d-His (D), pyruvate (Pyr) or α-ketoglutarate (αKG), NADH or NADPH). (B) Activity profile of PmCntM using variable concentrations of pyruvate (black circle) and α-ketoglutarate (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and 0.2 mM of xNA. The data points are means of three replicates with standard deviations. The fits are made using the Michaelis–Menten model. (C) Biosynthetic pathway for the assembly of bacillopaline from d-His, SAM and α-ketoglutarate.

Figure 6.
Activity of CntL and CntM from Paenibacillus mucilaginosus.

(A) TLC separation of reaction products incubating carboxyl-[14C]-labeled with a purified enzyme (PmCntL and PmCntM) and various substrates (l-His (L) or d-His (D), pyruvate (Pyr) or α-ketoglutarate (αKG), NADH or NADPH). (B) Activity profile of PmCntM using variable concentrations of pyruvate (black circle) and α-ketoglutarate (white circle) with fixed concentrations of other substrates: 0.2 mM of NADPH and 0.2 mM of xNA. The data points are means of three replicates with standard deviations. The fits are made using the Michaelis–Menten model. (C) Biosynthetic pathway for the assembly of bacillopaline from d-His, SAM and α-ketoglutarate.

Table 3
Kinetic parameters of PmCntM activities established for a concentration range of pyruvate and αKG with fixed concentrations of other substrates: 0.2 mM of NADPH and 0.2 mM of xNA
Protein Pyruvate vs αKG KM (µM) kcat (s−1kcat/KM (M−1 s−1
PmCntM pyruvate 180 ± 39 0.18 ± 0.01 1016 
PmCntM αKG 42 ± 5 0.44 ± 0.01 10 367 
Protein Pyruvate vs αKG KM (µM) kcat (s−1kcat/KM (M−1 s−1
PmCntM pyruvate 180 ± 39 0.18 ± 0.01 1016 
PmCntM αKG 42 ± 5 0.44 ± 0.01 10 367 

The data and the standard errors associated with were generated by SigmaPlot according to the Michaelis–Menten model.

Contrary to human pathogens like S. aureus, P. aeruginosa or Y. pestis in which opine-type metallophores were discovered [58], P. mucilaginosus is a soil bacterium. The production of opine-type metallophores was shown to be regulated by zinc through the zur (zinc uptake repressor) repressor [6,25], which is likely the case in P. mucilaginosus, with a putative zur box upstream the cnt operon. A possible hypothesis would be that bacillopaline production could be induced in calcareous soils, where zinc bioavailability has been shown to be low [39]. In the past decade, P. mucilaginosus have been studied for its capacity in wastewater treatment, but also as biofertilizer and plant growth promoting rhizobacteria [4042]. Indeed, in addition to producing bioflocculants or plant hormones, this species is able to solubilize phosphates, to fix nitrogen and to produce ammonia. Moreover, it is able to produce siderophores, which contributes to plant growth by indirectly preventing the growth of plant pathogens [41,43]. Similarly, we could hypothesize that bacillopaline could contribute in the same way to maintain plant micronutrients homeostasis.

In conclusion, studying the substrate specificity of the enzyme catalyzing the last step of nicotianamine-like metallophore biosynthesis allowed us to determine simple rules governing their production. First, the presence or absence of a cntK gene leads to the use of respectively d-His or l-His by CntL resulting in the incorporation of respectively xNA or yNA by CntM. Second, the presence of an aspartate or an alanine at position 150 on CntM (S. aureus numbering) results in pyruvate or α-ketoglutarate incorporation, respectively. Thanks to these simple rules, it is now possible to predict the nature of the nicotianamine-like metallophore produced by all bacteria possessing a cnt operon in their genome.

Abbreviations

     
  • KG

    ketoglutarate

  •  
  • SAM

    S-adenosyl methionine

Author Contributions

C.L. and P.A. designed the experiments. C.L., C.B., C.H. and L.O. carried out the experiments. C.L., C.B. and P.A. analyzed the data. G.C, R.F synthesized xNA and yNA under the supervision of F.C. C.L. and P.A. wrote the manuscript with contributions from F.C.

Funding

We thank the Agence Nationale de la Recherche [grant ANR-14-CE09-0007-02] for initial support and the association Vaincre la Mucoviscidose [VLM, grant RFI20160501495] for financial support.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Capdevila
,
D.A.
,
Wang
,
J.
and
Giedroc
,
D.P.
(
2016
)
Bacterial strategies to maintain zinc metallostasis at the host-pathogen interface
.
J. Biol. Chem.
291
,
20858
20868
2
Hood
,
M.I.
and
Skaar
,
E.P.
(
2012
)
Nutritional immunity: transition metals at the pathogen–host interface
.
Nat. Rev. Microbiol.
10
,
525
537
3
Weinberg
,
E.D.
(
1975
)
Nutritional immunity. Host's attempt to withold iron from microbial invaders
.
JAMA
231
,
39
41
4
Zygiel
,
E.M.
and
Nolan
,
E.M.
(
2018
)
Transition metal sequestration by the host-defense protein calprotectin
.
Annu. Rev. Biochem.
87
,
621
643
5
Ghssein
,
G.
,
Brutesco
,
C.
,
Ouerdane
,
L.
,
Fojcik
,
C.
,
Izaute
,
A.
,
Wang
,
S.
, et al.  (
2016
)
Biosynthesis of a broad-spectrum nicotianamine-like metallophore in Staphylococcus aureus
.
Science
352
,
1105
1109
6
Lhospice
,
S.
,
Gomez
,
N.O.
,
Ouerdane
,
L.
,
Brutesco
,
C.
,
Ghssein
,
G.
,
Hajjar
,
C.
, et al.  (
2017
)
Pseudomonas aeruginosa zinc uptake in chelating environment is primarily mediated by the metallophore pseudopaline
.
Sci. Rep.
7
,
17132
7
McFarlane
,
J.S.
and
Lamb
,
A.L.
(
2017
)
Biosynthesis of an opine metallophore by Pseudomonas aeruginosa
.
Biochemistry
56
,
5967
5971
8
McFarlane
,
J.S.
,
Davis
,
C.L.
and
Lamb
,
A.L.
(
2018
)
Staphylopine, pseudopaline, and yersinopine dehydrogenases: A structural and kinetic analysis of a new functional class of opine dehydrogenase
.
J. Biol. Chem.
293
,
8009
8019
9
Thompson
,
J.
and
Donkersloot
,
J.A.
(
1992
)
N-(Carboxyalkyl)amino acids: occurrence, synthesis, and functions
.
Annu. Rev. Biochem.
61
,
517
557
10
Bates
,
H.A.
,
Kaushal
,
A.
,
Deng
,
P.N.
and
Sciaky
,
D.
(
1984
)
Structure and synthesis of histopine, a histidine derivative produced by crown gall tumors
.
Biochemistry
23
,
3287
3290
11
Biemann
,
K.
,
Lioret
,
C.
,
Asselineau
,
J.
,
Lederer
,
E.
and
Polonsky
,
J.
(
1960
)
On the structure of lysopine, a new amino acid isolated from crown gall tissue
.
Biochim. Biophys. Acta
40
,
369
370
12
Smits
,
S.H.J.
,
Mueller
,
A.
,
Schmitt
,
L.
and
Grieshaber
,
M.K.
(
2008
)
A structural basis for substrate selectivity and stereoselectivity in octopine dehydrogenase from Pecten maximus
.
J. Mol. Biol.
381
,
200
211
13
Van Thoai
,
N.
,
Huc
,
C.
,
Pho
,
D.B.
and
Olomucki
,
A.
(
1969
)
Octopine déshydrogénase : purification et propriétés catalytiques
.
Biochim. Biophys. Acta, Enzymol.
191
,
46
57
14
Chang
,
C.-C.
and
Chen
,
C.-M.
(
1983
)
Evidence for the presence of N2-(1,3-dicarboxypropyl)-L-amino acids in crown-gall tumors induced by Agrobacterium tumefaciens strains 181 and EU6
.
FEBS Lett.
162
,
432
435
15
Chang
,
C.C.
,
Chen
,
C.M.
,
Adams
,
B.R.
and
Trost
,
B.M.
(
1983
)
Leucinopine, a characteristic compound of some crown-gall tumors
.
Proc. Natl Acad. Sci. U.S.A.
80
,
3573
3576
16
Gäde
,
G.
,
Weeda
,
E.
and
Gabbott
,
P.A.
(
1978
)
Changes in the level of octopine during the escape responses of the scallop, Pecten maximus (L.)
.
J. Comp. Physiol. B
124
,
121
127
17
Harcet
,
M.
,
Perina
,
D.
and
Pleše
,
B.
(
2013
)
Opine dehydrogenases in marine invertebrates
.
Biochem. Genet.
51
,
666
676
18
Moore
,
L.W.
,
Chilton
,
W.S.
and
Canfield
,
M.L.
(
1997
)
Diversity of opines and opine-catabolizing bacteria isolated from naturally occurring crown gall tumors
.
Appl. Environ. Microbiol.
63
,
201
207
PMID:
[PubMed]
19
Dessaux
,
Y.
and
Faure
,
D.
(
2018
) Niche Construction and Exploitation by Agrobacterium: How to Survive and Face Competition in Soil and Plant Habitats. In
Agrobacterium Biology: From Basic Science to Biotechnology
(
Gelvin
,
S.B.
, ed.), pp.
55
86
,
Springer International Publishing
,
Cham
20
Tremblay
,
G.
,
Gagliardo
,
R.
,
Chilton
,
W.S.
and
Dion
,
P.
(
1987
)
Diversity among opine-utilizing bacteria: identification of coryneform isolates
.
Appl. Environ. Microbiol.
53
,
1519
1524
PMID:
[PubMed]
21
Lang
,
J.
,
Vigouroux
,
A.
,
Planamente
,
S.
,
Sahili
,
A.E.
,
Blin
,
P.
,
Aumont-Nicaise
,
M.
et al.  (
2014
)
Agrobacterium uses a unique ligand-binding mode for trapping opines and acquiring a competitive advantage in the niche construction on plant host
.
PLoS Pathog.
10
,
e1004444
22
Montoya
,
A.L.
,
Chilton
,
M.D.
,
Gordon
,
M.P.
,
Sciaky
,
D.
and
Nester
,
E.W.
(
1977
)
Octopine and nopaline metabolism in Agrobacterium tumefaciens and crown gall tumor cells: role of plasmid genes
.
J. Bacteriol.
129
,
101
107
PMID:
[PubMed]
23
Grim
,
K.P.
,
San Francisco
,
B.
,
Radin
,
J.N.
,
Brazel
,
E.B.
,
Kelliher
,
J.L.
,
Párraga Solórzano
,
P.K.
et al.  (
2017
)
The metallophore staphylopine enables Staphylococcus aureus to compete with the host for zinc and overcome nutritional immunity
.
mBio
8
,
e01281-17
24
Remy
,
L.
,
Carrière
,
M.
,
Derré-Bobillot
,
A.
,
Martini
,
C.
,
Sanguinetti
,
M.
and
Borezée-Durant
,
E.
(
2013
)
The Staphylococcus aureus Opp1 ABC transporter imports nickel and cobalt in zinc-depleted conditions and contributes to virulence: nickel and cobalt uptake in Staphylococcus aureus
.
Mol. Microbiol.
87
,
730
743
25
Fojcik
,
C.
,
Arnoux
,
P.
,
Ouerdane
,
L.
,
Aigle
,
M.
,
Alfonsi
,
L.
and
Borezée-Durant
,
E.
(
2018
)
Independent and cooperative regulation of staphylopine biosynthesis and trafficking by Fur and Zur: regulation of S. aureus cnt operon by Fur and Zur
.
Mol. Microbiol.
108
,
159
177
26
Mastropasqua
,
M.C.
,
D'Orazio
,
M.
,
Cerasi
,
M.
,
Pacello
,
F.
,
Gismondi
,
A.
,
Canini
,
A.
, et al.  (
2017
)
Growth of Pseudomonas aeruginosa in zinc poor environments is promoted by a nicotianamine-related metallophore: metallophore-mediated zinc uptake in Pseudomonas aeruginosa
.
Mol. Microbiol.
106
,
543
561
27
Bielecki
,
P.
,
Puchałka
,
J.
,
Wos-Oxley
,
M.L.
,
Loessner
,
H.
,
Glik
,
J.
,
Kawecki
,
M.
et al.  (
2011
)
In-vivo expression profiling of Pseudomonas aeruginosa infections reveals niche-specific and strain-independent transcriptional programs
.
PLoS ONE
6
,
e24235
28
Gi
,
M.
,
Lee
,
K.-M.
,
Kim
,
S.C.
,
Yoon
,
J.-H.
,
Yoon
,
S.S.
and
Choi
,
J.Y.
(
2015
)
A novel siderophore system is essential for the growth of Pseudomonas aeruginosa in airway mucus
.
Sci. Rep.
5
,
14644
29
Ding
,
Y.
,
Fu
,
Y.
,
Lee
,
J.C.
and
Hooper
,
D.C.
(
2012
)
Staphylococcus aureus NorD, a putative efflux pump coregulated with the Opp1 oligopeptide permease, contributes selectively to fitness in vivo
.
J. Bacteriol.
194
,
6586
6593
30
Altschul
,
S.F.
,
Madden
,
T.L.
,
Schäffer
,
A.A.
,
Zhang
,
J.
,
Zhang
,
Z.
,
Miller
,
W.
et al.  (
1997
)
Gapped BLAST and PSI-BLAST: a new generation of protein database search programs
.
Nucleic Acids Res.
25
,
3389
3402
31
Edgar
,
R.C.
(
2004
)
MUSCLE: multiple sequence alignment with high accuracy and high throughput
.
Nucleic Acids Res.
32
,
1792
1797
32
Clamp
,
M.
,
Cuff
,
J.
,
Searle
,
S.M.
and
Barton
,
G.J.
(
2004
)
The Jalview Java alignment editor
.
Bioinformatics
20
,
426
427
33
Vallenet
,
D.
,
Belda
,
E.
,
Calteau
,
A.
,
Cruveiller
,
S.
,
Engelen
,
S.
,
Lajus
,
A.
, et al.  (
2013
)
Microscope—an integrated microbial resource for the curation and comparative analysis of genomic and metabolic data
.
Nucleic Acids Res.
41
,
D636
D647
34
Hajjar
,
C.
,
Fanelli
,
R.
,
Laffont
,
C.
,
Brutesco
,
C.
,
Cullia
,
G.
,
Tribout
,
M.
, et al.  (
2019
)
Control by metals of staphylopine dehydrogenase activity during metallophore biosynthesis
.
J. Am. Chem. Soc.
141
,
5555
5562
35
Higuchi
,
K.
,
Kanazawa
,
K.
,
Nishizawa
,
N.-K.
and
Mori
,
S.
(
1996
)
The role of nicotianamine synthase in response to Fe nutrition status in Gramineae
.
Plant Soil
178
,
171
177
36
Wilks
,
H.M.
,
Hart
,
K.W.
,
Feeney
,
R.
,
Dunn
,
C.R.
,
Muirhead
,
H.
,
Chia
,
W.N.
et al.  (
1988
)
A specific, highly active malate dehydrogenase by redesign of a lactate dehydrogenase framework
.
Science
242
,
1541
1544
37
Müller
,
A.
,
Janßen
,
F.
and
Grieshaber
,
M.K.
(
2007
)
Putative reaction mechanism of heterologously expressed octopine dehydrogenase from the great scallop, Pecten maximus (L)
.
FEBS J.
274
,
6329
6339
38
Vigouroux
,
A.
,
El Sahili
,
A.
,
Lang
,
J.
,
Aumont-Nicaise
,
M.
,
Dessaux
,
Y.
,
Faure
,
D.
et al.  (
2017
)
Structural basis for high specificity of octopine binding in the plant pathogen Agrobacterium tumefaciens
.
Sci. Rep.
7
,
18033
39
Alloway
,
B.J.
(
2009
)
Soil factors associated with zinc deficiency in crops and humans
.
Environ. Geochem. Health
31
,
537
548
40
Deng
,
S.
,
Bai
,
R.
,
Hu
,
X.
and
Luo
,
Q.
(
2003
)
Characteristics of a bioflocculant produced by Bacillus mucilaginosus and its use in starch wastewater treatment
.
Appl. Microbiol. Biotechnol.
60
,
588
593
41
Goswami
,
D.
,
Parmar
,
S.
,
Vaghela
,
H.
,
Dhandhukia
,
P.
and
Thakker
,
J.N.
(
2015
)
Describing Paenibacillus mucilaginosus strain N3 as an efficient plant growth promoting rhizobacteria (PGPR)
.
Cogent Food Agric.
1
,
1000714
42
Liu
,
S.
,
Tang
,
W.
,
Yang
,
F.
,
Meng
,
J.
,
Chen
,
W.
and
Li
,
X.
(
2017
)
Influence of biochar application on potassium-solubilizing Bacillus mucilaginosus as potential biofertilizer
.
Prep. Biochem. Biotechnol.
47
,
32
37
43
Haas
,
D.
and
Défago
,
G.
(
2005
)
Biological control of soil-borne pathogens by fluorescent pseudomonads
.
Nat. Rev. Microbiol.
3
,
307
319