Structural and Functional Validation of Pseudomonas Savastanoi Ethylene Forming Enzymes Reveals Flexibility in 2‐Oxoglutarate Binding Mode and Conformation

Abstract

Pseudomonas savastanoi pv. phaseolicola PK2 employs an Fe(II)‐dependent ethylene/succinate‐forming enzyme (PK2 PsEFE) to produce ethylene from 2‐oxoglutarate (2OG). Here we report NMR‐based assays showing that the putative P. savastanoi pv. glycinea PsEFE, which differs from PK2 PsEFE by a single residue, and the P. savastanoi pv.1449B PsEFE, which differs from PK2 PsEFE by 28 residues and a C‐terminal 13‐residue truncation, catalyze ethylene production from 2OG. Like the PK2 PsEFE, they catalyze oxidation of naturally occurring 2OG derivatives to give alcohol and diacid products. Crystallographic analysis demonstrates that the overall fold and active site of 1449B PsEFE is similar to that of PK2 PsEFE. Interestingly, 2OG was observed to adopt an atypical inverse metal ion binding mode in complex with 1449B PsEFE:Mn in which its 2‐oxoacid group is positioned to interact with the guanidinium group of R277, but not the Mn ion, which substitutes for catalytically active Fe(II). Together with reported crystallographic results, this observation indicates that 2OG metal ion binding modes and conformations at the active sites of 2OG oxygenases can vary, possibly in a functionally or disease relevant manner.

Studies with naturally occuring variants of the Pseudomonas savastanoi ethylene‐forming enzyme (PsEFE) inform on the effect of residues on turnover and product ratio. Crystallographic studies imply a potential for metal ion sensing.

1. Introduction

Ethylene is produced by plants, fungi, and bacteria [1, 2, 3, 4, 5, 6, 7]. In seeding plants, 1‐aminocyclopropane‐1‐carboxylate (ACC) oxidase (ACCO) catalyzes production of ethylene from ACC [8]. By contrast, organisms other than seeding plants can produce ethylene using enzymes which do not employ ACC as a substrate [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22]. Nitrogenases catalyze ethylene production from inter alia CO and acetylene [9, 10, 11, 12, 13, 14] and the nitrogenase‐like enzyme methylthio‐alkane reductase (Mar) catalyzes ethylene production from 2‐(methylthio)ethanol [15, 16, 17].

As shown by studies on the Pseudomonas savastanoi pv. phaseolicola PK2 ethylene/succinate‐forming enzyme (PK2 PsEFE), EFEs catalyze the oxidation of 2‐oxoglutarate (2OG) to give ethylene and CO₂ (Figures 1, and S1) [20, 21], in a manner dependent on L‐arginine [23, 24]. PK2 PsEFE also catalyzes the oxidation of 2OG to give succinic acid and CO₂; this reaction is coupled to the C5 hydroxylation of L‐arginine (Figures 1, and S1), to give 5‐hydroxyarginine which fragments to give guanidine and L‐Δ¹‐pyrroline‐5‐carboxylic acid (P5C) [18, 25, 26]. The PK2 PsEFE (co)substrate scope extends to C3‐ and/or C4‐substituted 2OG derivatives which can sustain catalysis to give diacid and alcohol, but not alkene, products, at least in most cases [27, 28, 29].

FIGURE 1.

PsEFE catalysis and structure. (a) Reactions of the P. savastanoi pv. phaseolicola PK2 ethylene/succinate‐forming enzyme (PK2 PsEFE) with 2OG. (b) The DSBH fold (light green) of PK2 PsEFE as observed in a reported PK2 PsEFE:Mn:2OG structure (PDB ID: 5V2X [35]); R326 is in purple (see below). (c) Active site view from a reported PK2 PsEFE:Mn:2OG structure (PDB ID: 5V2X [35]), showing that 2OG can adopt different metal ion binding modes, including an atypical inverted metal ion binding mode (yellow) coordinating to Mn via its C5 carboxylate rather than its 2‐oxoacid group. (d) Active site view of a PK2 PsEFE:Mn:2OG:L‐arginine structure (PDB ID: 5V2Y [35]) showing that 2OG adopts a typical metal ion binding mode in which it coordinates to Mn via its 2‐oxoacid group in the presence of L‐arginine, as typically observed with 2OG oxygenases [26, 32, 33].

PK2 PsEFE belongs to the Fe(II)‐dependent 2OG oxygenase structural superfamily which typically employ an HXD/E…H triad of ligands embedded within a distorted double‐stranded β‐helix (DSBH) core fold to bind a single Fe(II) cofactor (Figure 1) [30, 31]. At 2OG oxygenase active sites, 2OG typically adopts an Fe(II) binding mode in which it chelates to the Fe(II) by its C1 carboxylate and C2 ketone groups, with its C5 carboxylate engaging in interactions with residues distal to the Fe(II) binding HXD/E…H triad [26, 32, 33]. 2OG oxygenases normally catalyze 2‐electron substrate oxidations, most commonly hydroxylation or N‐methyl demethylation reactions (via initial hydroxylation), coupled to the oxidative decarboxylation of 2OG to give succinic acid [33]. Note that PK2 PsEFE has a relatively apolar active site, a feature possibly contributing to its unusual reaction to give ethylene via oxidative fragmentation of 2OG (Figure 1) [34, 35].

Reported crystal structures of PK2 PsEFE in complex with 2OG and (as far as is known) catalytically inert Mn(II), substituting for catalytically active Fe(II), indicate that 2OG can adopt different metal ion binding modes at the active site. In one structure, the 2OG is positioned to coordinate Mn via its 2‐oxoacid group and to interact via its C5 carboxylate group with the guanidinium group of R277 via a salt‐bridge (Figure 1) [35], in a manner typically observed with 2OG oxygenases [26, 32, 33]. By contrast, in a structure obtained in the absence of L‐arginine, 2OG was reported to adopt an alternative inverted metal ion binding mode (∼50% occupancy) in which it is positioned to interact with Mn via its C5 carboxylate group and to interact with the guanidinium group of R277 with its 2‐oxoacid group (Figure 1c) [35].

Even closely related P. syringae strains can differ in their ethylene‐producing capacity [36, 37, 38], with the P. syringae and P. savastanoi strains apparently carrying the EFE gene in form of plasmids rather than chromosomal DNA [39]. Pioneering work on PK2 PsEFE variants has revealed that minor sequence variations can ablate their ethylene producing capacity (Table S1), including single residue variations both within and outside the active site [27, 29, 34, 35, 40, 41, 42, 43]. Thus, the ethylene‐producing capacity of enzymes which have been assigned as EFEs based on their structural and/or sequence similarity with PK2 PsEFE should be experimentally validated by investigating the reactivity of the corresponding isolated enzymes [19, 44, 45].

Here we describe spectroscopic studies on catalysis by the predicted EFEs of the P. savastanoi strains pv. glycinea and 1449B (hereafter, glycinea PsEFE and 1449B PsEFE) revealing that they catalyze ethylene production from 2OG, as reported for PK2 PsEFE [20, 21]. The reactivities of the glycinea and 1449B PsEFEs with naturally occurring 2OG derivatives are similar to that of PK2 PsEFE [27, 28], as is the crystallographically observed fold of 1449B PsEFE [34, 35]. The combined results validate the assignment of glycinea PsEFE and 1449B PsEFE as ethylene‐producing enzymes.

2. Results

2.1. Comparison of PsEFE Sequences

We investigated the ethylene‐producing capacity of predicted EFEs from two P. savastanoi strains, that is, P. savastanoi pv. glycinea (glycinea PsEFE; UNIPROT ID: Q7BS32) and pv. 1449B (1449B PsEFE; UNIPROT ID: Q9Z3T0), which differ by 1 residue (R326S; glycinea PsEFE) and which have ∼88% sequence similarity (1449B PsEFE), respectively, compared with the validated ethylene‐producing PK2 PsEFE [20, 21] (UNIPROT ID: P32021; EC:1.13.12.19) (Figure 2), because the ethylene‐producing capacity of Pseudomonas strains differs [36, 37, 38] and because sequence variations can affect the ethylene producing capacity of PK2 PsEFE (Table S1) [27, 34, 35, 40, 41, 42, 43].

FIGURE 2.

Comparison of EFE sequences from P. savastanoi pv. phaseolicola PK2, pv. glycinea, and pv. 1449B. α‐Helices and β‐strands of PK2 PsEFE are numbered based on a PK2 PsEFE structure (PDB ID: 5MOF [34]); β‐turns: TT. Cyan triangles: β‐strands forming the core DSBH fold; green circles: Fe(II)‐binding HXD…H residues; blue circles: polar residues crystallographically observed to directly interact with 2OG; purple pentagon: the R326S variation distinguishes the glycinea PsEFE from the PK2 PsEFE; brown squares: 28 residues distinguishing the 1449B PsEFE from the PK2 PsEFE; orange rectangle: 13 C‐terminal PK2 PsEFE residues (338–350) not present in 1449B PsEFE. Identical residues are in red; residues of similar polarity are in yellow. The figure was generated using ESPript 3.0 [50].

The R326S variation is located on the surface of the glycinea PsEFE (Figure 1b), relatively far from the active site, and appears to be present in different P. savastanoi and P. syringae strains [35, 37]. To our knowledge, the effect of the R326S variation on catalysis of isolated recombinant PK2 PsEFE has not been investigated (Table S1). 1449B PsEFE lacks the 13 C‐terminal residues of the PK2 and glycinea PsEFEs and differs by 28 additional residues from PK2 PsEFE: the 28 substitutions involve residues both close to and distant from the active site (Figure 2). Note that the 1449B PsEFE sequence is identical to that of the P. syringae pv. pisi GSPB1206 EFE, which is a reported ethylene‐producing P. syringae strain [35, 37]. The sequences of the glycinea, 1449B, and PK2 PsEFEs differ substantially from those of functionally and structurally validated Fe(II)‐dependent ethylene‐producing enzymes of other organisms, including the EFE of Penicillium digitatum [45, 46, 47] as well as Arabidopsis thaliana and Petunia hybrida ACCO1 (Figure S2) [48, 49].

2.2. The glycinea and 1449B PsEFEs Catalyze Ethylene Production from 2OG

Recombinant forms of the isolated glycinea and 1449B PsEFEs were produced in E. coli and purified to high purity (>90% by SDS‐PAGE and mass spectrometric analyses; Supporting Figure 3). To investigate their ethylene‐producing capabilities, we used ¹H NMR assays, employing 3‐(trimethylsilyl)−2,2,3,3‐tetradeuteropropionic acid (TMSP‐d ₄) [51] as an internal standard to quantify substrate conversion; we have previously employed ¹H NMR assays to monitor ethylene formation/2OG depletion by PK2 PsEFE [28, 34]. ¹H NMR analysis (700 MHz) of the reactions of the isolated glycinea and 1449B PsEFEs with 2OG reveals that both the glycinea and 1449B PsEFEs catalyze ethylene production from 2OG in the presence of L‐arginine under the same conditions reported to give ethylene with PK2 PsEFE (Figure 3) [28], confirming their assignment as EFEs. The ¹H NMR assays showed that the glycinea and 1449B PsEFEs also catalyzed oxidation of 2OG to succinic acid in a reaction likely coupled to C5 oxidation of L‐arginine to give P5C, as implied by formation of the characteristic P5C singlet at observed at δ ∼ 7.70 ppm (Figure 3). The glycinea and 1449B PsEFEs did not catalyze efficient 2OG oxidation in the absence of L‐arginine (Figure S4).

FIGURE 3.

The glycinea and 1449B PsEFEs catalyze ethylene production from 2OG. (a) Reactions of PK2, glycinea, and 1449B PsEFE with 2OG. (b) Outcomes of the reaction of 1449B and glycinea PsEFE with 2OG after 90 min, compared to that reported for PK2 PsEFE [28]. Ethylene formation was estimated by subtracting the concentration of succinic acid and 2OG observed after 90 min incubations (obtained from the TMSP‐d ₄ normalized integral) from the starting concentration of 2OG. Note that we did not observe 3‐hydroxypropanoic acid formation by ¹H NMR [40, 44]. (c–e) ¹H NMR spectra (δ ∼0.8 to ∼4.4 ppm and ∼5.4 to ∼8.0 ppm) of a 2OG reaction mixture (90 min) for: (c) 1449B PsEFE or (d) glycinea PsEFE, compared to that reported for: (e) PK2 PsEFE [28]; (f) ¹H NMR spectrum of 2OG in the absence of PsEFE. Conditions: 400 μM 2OG, 500 µM L‐arginine, 500 µM L‐ascorbic acid, 50 µM Fe(II), 800 µM TMSP‐d₄ , and PsEFE (2 µM for PK2 [28]; 5 µM for 1449B; 10 µM for glycinea) in buffer (50 mM sodium phosphate, pH7.4, 10%_v/v D₂O). Note that low levels of dehydroascorbate (∼4.3 and ∼4.2 ppm) [28], succinic acid (∼2.4 ppm), and the hydrated ketone form of 2OG (∼2.0 and ∼2.2 ppm, dashed circles) [52, 53] were observed. The peak at ∼3.75 ppm (labeled: *) corresponds to Tris buffer. Time scales are calibrated to the end of the acquisition of the first ¹H NMR experiment following PsEFE addition to the reaction mixture (t = 0 min), by which time low levels of turnover were manifest. Results are means of independent triplicates (n = 3; mean ± standard deviation, SD).

The glycinea and 1449B PsEFEs were not observed to catalyze formation of 2OG‐derived products other than succinic acid and ethylene (Figure 3), in accord with our reported ¹H NMR observations on the reactivity of PK2 PsEFE with 2OG [28, 34]. This observation contrasts with the reported ability of PK2 PsEFE and related EFEs to catalyze formation of low levels of 3‐hydroxypropionic acid, possibly because of different assay conditions and/or because ¹H NMR‐based assays are less sensitive than LCMS/infrared‐based assays previously used to detect 3‐hydroxypropionic acid [40, 44, 45].

Specific activities were calculated for the PK2, 1449B, and glycinea PsEFEs by analyzing 2OG depletion ∼10 min post incubation with PsEFE. The results reveal that the specific activities decrease in the order: PK2 PsEFE > 1449B PsEFE > glycinea PsEFE, assuming that the purified PsEFEs were fully active (Table 1). Although ethylene formation was not directly quantified because ethylene may evaporate from the reaction mixture, it was estimated by subtracting the concentration of succinic acid and 2OG observed at a given time (obtained from the TMSP‐d ₄ normalized integral) from the starting concentration of 2OG. Time‐course studies measuring 2OG depletion and succinic acid formation imply that ethylene was continuously formed from 2OG during the reaction with glycinea and 1449B PsEFE (Figure S5‐S6). The ethylene:succinic acid ratio was estimated to be ∼3:2 for the 1449B PsEFE, but was ∼1:1 for both the glycinea and PK2 PsEFE after 90 min incubations (Figure 3b).

TABLE 1.

Specific activities and product ratios of the PK2, 1449B, and glycinea PsEFEs with 2OG and 2OG derivatives.

2OG derivative	Specific activities, a nmol·min −1 ·mg −1			Succinic acid:ethylene ratios (for 2OG) b or diacid:alcohol ratios c
	PK2 PsEFE	1449B PsEFE	glycinea PsEFE	PK2 PsEFE	1449B PsEFE	glycinea PsEFE
2OGb	357 ± 26	136 ± 4	17 ± 6	1.0 ± 0.1 : 1	0.7 ± 0.1 : 1	0.8 ± 0.1 : 1
2OAc	52 ± 1	23 ± 14	2 ± 1	1.2 ± 0.1 : 1	1.3 ± 0.1 : 1	1.2 ± 0.1 : 1
4‐Methyl‐2OG (3)c	72 ± 4	46 ± 7	3 ± 1	1.9 ± 0.3 : 1	1.3 ± 0.3 : 1	1.3 ± 0.2 : 1
3‐Methyl‐2OG (4)c	62 ± 1	36 ± 3	8 ± 1	1.2 ± 0.1 : 1	1.4 ± 0.1 : 1	0.9 ± 0.1 : 1
4‐Hydroxy‐2OG (5)c	69 ± 2	10 ± 4	3 ± 2	2.2 ± 0.3 : 1	2.6 ± 0.9 : 1	1.4 ± 0.4 : 1

^a Specific activities (nmol·min⁻¹·mg⁻¹) based on 2OG/2OG derivative depletion after ∼10 min incubations with PsEFE.

^b Succinic acid:ethylene ratios ∼90 min post incubation with PsEFE.

^c Diacid:alcohol ratios ∼ 90 min post incubation with PsEFE. Assay conditions: 400 µM 2OG or 2OG derivative, 500 µM L‐arginine, 500 µM L‐ascorbic acid, 50 µM Fe(II), 800 µM TMSP‐d₄ , and PsEFE (using: 2 µM PK2, 5 µM 1449B, and 10 µM glycinea PsEFE with 2OG; or using: 10 µM PK2 or 1449B and 30 µM glycinea PsEFE with 2OG derivatives) in buffer (50 mM sodium phosphate, pH 7.4, 10% _v/v D₂O). Results are means of independent triplicates (n = 3; mean ± SD).

2.3. The glycinea and 1449B PsEFEs Can Accept 2OG Derivatives as (co)substrates

The reactivities of the glycinea and 1449B PsEFEs with four naturally occurring 2OG derivatives, that is, 2‐oxoadipate (2OA, 2) [54, 55], 4‐methyl‐2OG (3) [56, 57, 58], 3‐methyl‐2OG (4) [59], and 4‐hydroxy‐2OG (5) [60], were investigated to compare the reaction outcomes with those reported for PK2 PsEFE [28]. ¹H NMR analyses (700 MHz) reveal that the glycinea and 1449B PsEFEs catalyzed formation of a diacid and an alcohol product from all the four tested 2OG derivatives (Figure 4 and S5‐S10), as also observed for PK2 PsEFE [27, 28]. The estimated diacid:alcohol ratios of the reactions of the 1449B and glycinea PsEFEs with 2OG derivatives were similar within experimental error and did not differ substantially from those reported for PK2 PsEFE (Table 1) [28]. As observed with 2OG (Figure 3), the specific activities of the three PsEFEs with 2OG derivatives decreased in the order: PK2 PsEFE > 1449B PsEFE > glycinea PsEFE (Table 1). Note that the concentration of glycinea PsEFE in the ¹H NMR assays with the 2OG derivatives was 3‐fold that of PK2 and 1449B PsEFE to enable robust product formation.

FIGURE 4.

The glycinea and 1449B PsEFEs catalyze oxidation of 2OG derivatives. Outcomes of the reactions of the glycinea and 1449B PsEFEs with: (a) 2‐oxoadipate (2OA, 2), (b) 4‐methyl‐2OG (3), (c) 3‐methyl‐2OG (4), and (d) 4‐hydroxy‐2OG (5), compared to those reported for PK2 PsEFE [28]. Conditions: 400 µM 2OG derivative, 500  µM L‐arginine, 500  µM L‐ascorbic acid, 50  µM Fe(II), 800 µM TMSP‐d₄ , and PsEFE (PK2 [28] and 1449B PsEFE: 10  µM; glycinea PsEFE: 30 µM) in buffer (50 mM sodium phosphate, pH 7.4, 10% v/v D₂O). Results are means of independent triplicates (n = 3; mean ± SD). Note that the 2OG derivatives 3, 4, and 5 were used as racemic mixtures; there is potential for stereoselectivity in their reactions, including that one enantiomer may be a PsEFE substrate and the other an inhibitor and/or that one enantiomer favors a particular PsEFE reaction pathway.

2.4. Crystallography

1449B PsEFE was crystallized in the presence of 2OG and Mn(II), the latter substituting for catalytically active Fe(II). 1449B PsEFE crystallized in the P43 space group (1.65 Å resolution; Figure 5 and Table S2), whereas PK2 PsEFE was crystallized in different space groups [34, 35, 41, 61]. The 1449B PsEFE:Mn:2OG complex crystal structure was solved by molecular replacement using a reported PK2 PsEFE structure as a search model (PDB ID: 5MOF [34]). Superimposition of the 1449B PsEFE:Mn:2OG structure and a reported PK2 PsEFE:Mn:2OG structure (PDB ID: 5MOF [34]) reveal a similar overall fold (RMSD: 0.415 Å) (Figure S11) and active site geometry, in accord with the ability of 1449B PsEFE to catalyze conversion of 2OG to ethylene (Figure 3).

FIGURE 5.

2OG adopts an inverted binding mode in complex with 1449B PsEFE:Mn. (a) Locations of 1449B PsEFE residues in the 1449B PsEFE:Mn:2OG crystal structure (PDB ID: 9SCF) that differ from PK2 PsEFE (in red; see Figure 2); the corresponding PK2 PsEFE residues are in black parentheses; the 1449B PsEFE DSBH is in cyan. (b) Representative Polder omit electron density map contoured to 3σ around 2OG and R277 in the 1449B PsEFE:Mn:2OG structure. 2OG coordinates to Mn and interacts with the R171 guanidinium group with its C5 carboxylate; it interacts with the R277 guanidinium group via its C1 carboxylate and C2 ketone. (c) Active site view of a PK2 PsEFE:Mn:Cl:2OG structure (PDB ID: 5MOF [34]). (d) Superimposition of active site views from the 1449B PsEFE:Mn:2OG (PDB ID: 9SCF) and PK2 PsEFE:Mn:2OG:Cl (PDB ID: 5MOF [34]) structures reveal different 2OG metal ion binding modes. (e) Active site view of a reported PK2 PsEFE:Mn:2OG structure (PDB ID: 5V2X [35]). (f and g) Superimposition of views from the 1449B PsEFE:Mn:2OG (PDB ID: 9SCF) and PK2 PsEFE:Mn:2OG (PDB ID: 5V2X [35]) structures. Note that 2OG adopts two binding modes in the reported PK2 PsEFE:Mn:2OG structure, with Binding Mode A shown in: (f) and Binding Mode B shown in: (g) [35]. Binding Modes A and B differ from that observed in the 1449B PsEFE:Mn:2OG structure. (h) Superimposition of active site views from the 1449B PsEFE:Mn:2OG (PDB ID: 9SCF) and PK2 PsEFE:Mn:2OG:L‐arginine (PK2 PsEFE: orange; Mn: violet; carbon‐backbone of 2OG: turquoise; carbon‐backbone of L‐arginine: salmon; PDB ID: 5V2Y [35]) structures imply that L‐arginine binding affects the 2OG conformation.

Analysis of electron density maps for the 1449B PsEFE:Mn:2OG structure reveals both 2OG and Mn bound at the active site with 2OG being modeled in a single conformation with an inverted metal ion binding mode compared to that typically observed for 2OG [26, 32, 33, 34, 35]. In this structure, the 2‐oxoacid group of 2OG is positioned to interact with the guanidinium group of R277 via its C1 carboxylate and C2 ketone groups (2.5 and 3.0 Å, respectively) rather than with the metal ion, as typically observed [26, 32, 33]. Mn is complexed by one of the two 2OG C5 carboxylate oxygen atoms (2.2 Å), with the other 2OG C5 carboxylate oxygen atom being positioned to interact with the guanidinium group of R171 (2.8 Å) (Figure 5b). No evidence for the typical 2OG binding mode was apparent, but the presence of low levels of this cannot be excluded. As anticipated, Mn is coordinated by the side chain imidazole rings of H189 and H268 (2.2 Å), the side chain carboxylate of D191 (2.2 Å) [30], and the water molecules W1 and W2 (1.6 and 2.0 Å, respectively) (Figure 5b).

The two Mn‐bound water molecules W1 and W2 are part of a water network which extends across the active site (Figure 6a). The water network links the 2OG C1 and C5 carboxylate groups, the D191 side chain carboxylate, and the Y192 and Y318 side chain alcohols (Y192 is involved in binding of the L‐arginine substrate, as observed in the PK2 PsEFE:Mn:2OG:L‐arginine (PDB ID: 5V2Y [35]) structure) with Q37, T241, V246, Q200, and T248, with some of latter residues being distal to the 1449B PsEFE active site. The positions of water molecules within, and proximal to, the 1449B PsEFE active site is similar, but not identical to those observed in reported PK2 PsEFE structures (Figure 6d) [34, 35]. Note that in the PK2 PsEFE:Mn:Cl:2OG structure (PDB ID: 5MOF [34]) a chloride ion was modeled in between R277 and the 2OG C5 carboxylate (∼30% occupancy), binding of which is stabilized by interactions with water molecules (Figure 5c‐d) [34]. PsEFE turnover assays are typically performed in the absence of high chloride concentrations [19, 23, 24, 27, 28, 34, 35, 40, 41, 44, 45, 61], though chloride ions from the enzyme preparation are present and do not ablate catalytic activity [28].

FIGURE 6.

An extensive water network stabilizes 2OG binding in PsEFE:Mn:2OG complex as observed in crystal structures in the absence of L‐arginine. (a) Three water molecules are within hydrogen bonding distance (<3.2 Å) of 2OG in the 1449B PsEFE:Mn:2OG structure, apparently stabilizing 2OG binding. (b and c) Close up views of the waters in proximity of: (b) the 2OG C1 carboxylate and (c) the 2OG C5 carboxylate in the 1449B PsEFE:Mn:2OG structure. (d) A water network similar to that at the 1449B PsEFE active site was observed in a PK2 PsEFE:Mn:2OG crystal structure (PDB ID: 5V2X [35]). (e) The PK2 PsEFE:Mn:2OG:L‐arginine structure (PDB ID: 5V2Y [35]) reveals that the presence of L‐arginine at the PsEFE active site alters the water network.

The inverted 2OG binding mode observed in the 1449B PsEFE:Mn:2OG structure is similar, but not identical, to the inverted 2OG binding mode observed in a reported PK2 PsEFE:Mn:2OG structure (Binding Mode B; Figure 5g) (PDB ID: 5V2X [35]). Notably, in our 1449B PsEFE:Mn:2OG structure and in the reported Binding Mode B of the reported PK2 PsEFE:Mn:2OG structure, 2OG adopts a C2/C5 anti‐periplanar conformation (Figure 5b,g), whereas it adopts a C2/C5 gauche conformation in the reported Binding Mode A of the PK2 PsEFE:Mn:2OG structure (Figure 5f). The PK2 PsEFE:Mn:Cl:2OG (PDB ID: 5MOF [34]) and PK2 PsEFE:Mn:2OG:L‐arginine (PDB ID: 5V2Y [35]) structures manifest a substantially different 2OG binding mode, in which 2OG coordinates to Mn via its 2‐oxoacid group, as typically observed in 2OG oxygenase crystal structures in complex with 2OG (Figure 5h) [26, 32, 33]. The 2OG binding mode in the 1449B PsEFE:Mn:2OG structure also differs from that observed in a reported Penicillium digitatum EFE structure in complex with Mn and 2OG (PDB ID: 9EIS [45]), in which only the 2OG C1 carboxylate group, but not the 2OG C2 ketone group, was modeled in a position to interact with the guanidinium group of R336, the R277‐equivalent residue in the Penicillium digitatum EFE [45].

3. Discussion

Analysis of specific activities indicates that, at least under the tested conditions, the glycinea PsEFE and the 1449B PsEFE are less efficient at catalyzing 2OG depletion and ethylene production compared to PK2 PsEFE (Table 1). This observation is in accord with reported studies showing that the R326S PK2 PsEFE variation reduced levels of ethylene production in cell extracts of different Pseudomonas species (Table S1) [37]. The glycinea PsEFE S326 residue is located on the surface of the protein and is relatively close to the C‐terminal region, a region which has been shown to affect 2OG oxygenases catalysis, at least in some cases, in a manner involving conformational changes [62, 63, 64, 65, 66].

Notably, although the R326S PK2 PsEFE variation is conserved amongst several Pseudomonas pathovars [35, 37], work with intact cells implies that these Pseudomonas pathovars produce variable levels of ethylene, despite bearing PsEFEs with identical sequences [67]. Results with cell extracts suggested that the 1449B PsEFE is less efficient in producing ethylene than the glycinea PsEFE [37]; by contrast, our work with isolated enzymes indicates that the 1449B PsEFE is more efficient in catalyzing ethylene production than the glycinea PsEFE (Figure 3; Table 1). Thus, the combined results suggest that ethylene production in Pseudomonas pathovars does not only depend on the PsEFE sequence but also on other factors.

The glycinea PsEFE and PK2 PsEFE produced ethylene and succinic acid in approximately equal ratios under the tested conditions (Figure 3). By contrast, the 1449B PsEFE apparently favored ethylene over succinic acid production (Figure 3), possibly reflecting the lower sequence similarity of 1449B and PK2 PsEFE (Figure 2). Although functional and structural studies on 2OG oxygenases have revealed that the C‐terminus of some 2OG oxygenases is proximal to the active site and can affect catalysis [62, 63, 64, 65, 66], our work supports reported results showing that the 13 C‐terminal PK2 PsEFE residues are not required for ethylene production and L‐arginine hydroxylation (Figure 3) [35]. The combined results thus suggest that one or more of the 28 other residues distinguishing the 1449B and PK2 PsEFEs affect the ethylene:succinic acid ratio, which is reported to be ∼1:1 for the PK2 PsEFE variant lacking the 13 C‐terminal residues (as for wildtype PK2 PsEFE) [28, 35], but which is ∼3:2 for 1449B PsEFE (Figure 3). As many of the 28 residues distinguishing the 1449B and PK2 PsEFEs are not at the active site (Figure 5), they may influence second coordination sphere interactions, possibly involving the active site water network (Figure 6) [68, 69].

The observation that the PsEFE sequence affects the ethylene:succinic acid ratio is precedented by work on PsEFE‐related enzymes, including the Penicillium digitatum EFE which manifests ∼55% sequence similarity with PK2 PsEFE (Figure S2) and which also favors ethylene over succinic acid production [44, 45]. Thus, 1449B PsEFE‐related EFEs might exist or might be evolved, which exclusively catalyze ethylene production from 2OG, but not C5 hydroxylation of L‐arginine. Such EFEs are attractive from biocatalysis perspectives to enable sustainable ethylene production from renewable resources [70, 71].

The 1449B PsEFE:Mn:2OG structure reveals an atypical inverted 2OG metal ion binding mode in which the 2OG 2‐oxoacid group is positioned tointeract with the guanidinium group of R277 rather than with the Mn ion, which is instead complexed by the 2OG C5 carboxylate (Figure 5). This observation is precedented by similar, but not identical 2OG binding modes in reported PK2 PsEFE:Mn:2OG and Penicillium digitatum EFE:Mn:2OG structures (Figure 5) [35, 45]. It is unclear whether this atypical binding mode of 2OG in complex with PsEFE is relevant in cells, including because the crystal structures reported to date suggest that 2OG only adopts these inverted modes in complex with PK2 PsEFE and catalytically inert Mn(II), but not Fe(II), and in the absence of L‐arginine. At least in crystals, the presence of L‐arginine and/or substituting Mn(II) by Fe(II) results in a 2OG Fe(II)‐binding mode typically observed in 2OG oxygenases, that is, chelation of the active site metal with the 2OG 2‐oxoacid motif (Figure 5h) [35, 41].

The atypical 2OG metal ion binding modes observed with PK2 PsEFE, 1449B PsEFE, and Penicillium digitatum EFE might nonetheless be of in vivo relevance [35, 45], though likely not for productive catalysis given that 2OG adopts an Fe(II)‐binding mode typically observed for 2OG oxygenases in the presence of Fe(II) [35, 41]. A possible functional relevance could relate to metal sensing as some Fe(II)‐dependent oxygenases are proposed to act as non‐Fe transition metal ion sensors as a result of Fe(II) displacement and consequent activity loss [72, 73, 74, 75, 76, 77]. It is possible that catalytically inactive 2OG binding modes at the PsEFE active site may be involved in metal ion sensing and/or in enabling different reactions; in the latter case, it is notable that structurally distinct dioxygenases are catalytically active using either Fe(II) or other divalent transition metal ions [78, 79]. In this regard it is of interest that mutations in a Mn(II) transporter protein lead to accumulation of Mn(II) in cells, causing a genetic disease the symptoms of which include increased erythropoietin and, consequently, polycythemia [80]. These symptomes are proposed to result from competition between Mn(II) with catalytically active Fe(II) for binding at the at the active site of the 2OG dependent hypoxia inducible factor‐α prolyl hydroxylases (PHDs) which negatively regulate hypoxia inducible factor promoted transcription of the erythropoietin gene [81]. An analogous PHD inhibition mechanism is proposed for the mechanism of action of (now discontinued) anemia treatment by CoCl₂ and for inhibition of other 2OG oxygenases by catalytically inactive transition metal ions [74, 75, 76]. Although many structures of 2OG oxygenases have been obtained with transition metal ions other than Fe(II) with 2OG complexed in a non‐inverted mode, the results presented here and elsewhere, support the possibility that atypical 2OG binding modes may play a functionally relevant role in 2OG oxygenase inhibition by transition metal ions in cells.

2OG not only adopts atypical metal ion binding modes in complex with PsEFE and Mn, but also atypical (i.e., non anti‐periplanar) conformations, i.e., with its C2/C5 groups being in a gauche arrangement (Figure 5f) [35]. Such 2OG conformations have also been observed in complexes of 2OG oxygenases other than PsEFE. For example, 2OG was observed to adopt a C2/C5 gauche conformation with the 2OG C5 carboxylate group complexing Mn via a water molecule in crystal structures with disease‐related active site variants of human aspartate/asparagine‐β‐hydroxylase (AspH), whereas it adopted a C2/C5 antiperiplanar conformation in complex with wildtype AspH [82, 83, 84]. The oncometabolite D‐2‐hydoxyglutarate (2HG) has been observed to adopt a C2/C5 gauche conformation in complex with the JmjC N ^ε ‐methyllysine demethylase 5B (KDM5B) and Mn, that is, with the 2HG C5 carboxylate group being positioned to interact with the hydroxy group of 2HG [85]. Notably, oxaloacetate has been observed in an inverted metal ion binding mode in complex with KDM5B and Mn [85]. Although oxaloacetate lacks one of the two 2OG methylene groups, the combined observations indicate that 2OG binding modes and conformations may be flexible at 2OG oxygenase active site, possibly in a manner affecting catalysis.

Given that the glycinea, 1449B, and PK2 PsEFEs can employ naturally occurring 2OG derivatives as cosubstrates to sustain L‐arginine hydroxylation and alcohol formation, but not ethylene production (Figure 4) [27, 28], it is possible that the 2OG derivative substitution pattern affects the binding mode and/or conformation of the 2OG derivatives at the PsEFE active site, as observed in the case of AspH [82], potentially in a manner affecting reaction outcomes. Although we obtained crystal structures of 1449B PsEFE in complex with Mn and 4‐methyl‐2OG or 3‐methyl‐2OG to test this proposal, their resolutions (2.05 and 1.85 Å, respectively) were insufficient to unambiguously assign the precise 2OG derivative binding modes at the active site.

An extensive water network was observed in the active site of the 1449B PsEFE:Mn:2OG structure, this network likely stabilizes 2OG binding and may affect reaction outcomes (Figure 6). Some of the 1449B PsEFE residues involved in stabilizing the water network at the active site are also likely involved in binding L‐arginine (Table S1) [35]. All EFE crystal structures [34, 45] and ACCO structures [49] also feature active site water networks, suggesting such networks may be an important general characteristic of ethylene‐producing enzymes of the 2OG superfamily. It is therefore of interest to investigate the effects of variations in residues which are involved in stabilizing the water network on the currently accepted PsEFE mechanism (Figure S1) [18, 27, 40, 86, 87, 88, 89], including on PsEFE fold, 2OG binding mode and conformation, catalytic efficiency, and reaction outcome.

4. Conclusion

The combined spectroscopic and crystallographic evidence demonstrates that both the glycinea and 1449B PsEFEs have capacity to produce ethylene from 2OG (Figure 3), however, apparently less efficiently than PK2 PsEFE and, at least in the case of 1449B PsEFE, with an altered product ratio, favoring ethylene over succinic acid formation. The results provide further evidence for variations in the binding modes of 2OG and 2OG derivatives at the active sites of 2OG oxygenases.

5. Experimental Procedures

5.1. Protein Production

Plasmid DNA encoding for glycinea PsEFE (i.e., the R326S PK2 PsEFE variant) was obtained by standard site‐directed mutagenesis of our reported PK2 PsEFE construct [34] (Table S3). Glycinea PsEFE was produced and purified as reported for PK2 PsEFE [28, 34]. Plasmid DNA encoding for 1449B PsEFE in a pET‐28a(+)‐TEV vector was codon optimized for expression in E. coli cells (prepared by GenScript Biotech). 1449B PsEFE was produced and purified as reported for PK2 PsEFE [28, 34]; however, the Tobacco Etch Virus (TEV) protease was used to remove the N‐terminal His₆‐tag of 1449B PsEFE instead of the SENP2 protease. Thus, His₆‐tagged 1449B PsEFE was incubated overnight with the TEV protease (∼100:1 ratio) in cleavage buffer (50 mM Tris, 100 mM NaCl, 10%_v/v glycerol, pH 7.8) at 4˚C, followed by reverse Ni(II) affinity chromatography purification and size exclusion chromatography as reported for PK2 PsEFE [28, 34]. Glycinea and 1449B PsEFE were >90% pure, as measured by SDS‐PAGE analysis, and had the anticipated masses (Figure S3).

5.2. 2OG Derivatives

2‐Oxoglutarate (2OG), 2‐oxoadipate (2OA), and racemic 4‐hydroxy‐2OG (5) were from Sigma–Aldrich. Racemic 4‐methyl‐2OG (3) and 3‐methyl‐2OG (4) were synthesized as reported [82].

5.3. ¹H NMR Assays

¹H NMR assays were performed as reported [28], typically using: 2OG or a 2OG derivative (400 µM), L‐arginine (500 µM), L‐ascorbic acid (500 µM), Fe(II) (50 µM), TMSP‐d ₄ (800 µM, used as an internal standard), and a PsEFE (10 µM for 1449B PsEFE and 30 µM for glycinea PsEFE) in buffer (50 mM sodium phosphate, pH 7.4, 10%_v/v D₂O), if not specified otherwise. Freshly thawed glycinea and 1449B PsEFE aliquots were used for all experiments. Assays were performed using a BrukerAVIII 700 MHz NMR spectrometer with a 5 mm inverse triple‐resonance‐inverse cryoprobe. The water peak in ¹H NMR experiments was suppressed using the perfect echo‐modified WATERGATE solvent suppression method [90].

5.4. Crystallography

Full‐length 1449B PsEFE (11 mg/mL) was incubated with MnCl₂ (3 mM), and 2OG (10 mM) in buffer (50 mM Tris, pH 7.5) at 4°C for 15 min and centrifuged using a MicroCL 21R centrifuge (Thermo Fisher Scientific; 18,800 × g, 4°C, 1 min). Crystallisations were performed using the sitting drop vapor diffusion method (3 μL drops) at 17 °C using 2:1, 1:1, and 1:2 well solution (μL): protein solution (μL) ratios. Conditions: 0.03 M citric acid, 0.07 M Bis‐Tris propane/pH 7.6, 20%_w/v polyethylene glycol (PEG) 3350. Crystals were cryo‐protected with 20‐25%_v/v glycerol prior to flash‐freezing with liquid N₂.

Diffraction data for the 1449B PsEFE single crystals were collected at the Diamond Light Source (beamline: I03). The data were indexed, integrated, and scaled using the Xia2 pipeline [91, 92]. Structures were solved by molecular replacement with Phaser [93], using a reported PK2 PsEFE structure as a search model (PDB ID: 5MOF [34]). Structures were iteratively refined using PHENIX [94] and manual model building using COOT [95].

Author Contributions

Yihong Sun and Siddhant Dhingra produced enzymes with help of Ethan S. Z. Cheng; Yihong Sun, Siddhant Dhingra, and Ethan S. Z. Cheng performed assays; Yihong Sun, Siddhant Dhingra, and Ethan S. Z. Cheng performed crystallizations, Mark D. Alle n solved and refined the structure with help from Yihong Sun; Siddhant Dhingra and Lennart Brewitz conceived the project together with Yihong Sun and Christopher J. Schofield; Lennart Brewitz, Zhihong Zhang, and Christopher J. Schofield supervised the research; all authors analyzed data; Lennart Brewitz, Siddhant Dhingra and Christopher J. Schofield wrote the manuscript with help from all authors. The authors thank the Diamond Light Source (beamline I03) and staff for allocation of beam time and support.

Supporting Information

Additional supporting information can be found online in the Supporting Information Section.

Funding

This study was supported by Wellcome Trust (Grant 106244/Z/14/Z), Cancer Research UK (Grant C8717/A18245), and Biotechnology and Biological Sciences Research Council (Grant BB/J003018/1 and BB/R000344/1).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material

Data Availability Statrement

Crystal structure data for 1449B PsEFE complexed to Mn and 2OG (1449B PsEFE:Mn:2OG) are deposited in the protein data bank with PDB accession code: 9SCF.