Uncovering Zn2+ as a cofactor of FAD-dependent Pseudomonas aeruginosa PAO1 d-2-hydroxyglutarate dehydrogenase

Pseudomonas aeruginosa couples the oxidation of d-2-hydroxyglutarate (D2HG) to l-serine biosynthesis for survival, using d-2-hydroxyglutarate dehydrogenase from P. aeruginosa (PaD2HGDH). Knockout of PaD2HGDH impedes P. aeruginosa growth, making PaD2HGDH a potential target for therapeutics. Previous studies showed that the enzyme's activity increased with Zn2+, Co2+, or Mn2+ but did not establish the enzyme's metal composition and whether the metal is an activator or a required cofactor for the enzyme, which we addressed in this study. Comparable to the human enzyme, PaD2HGDH showed only 15% flavin reduction with D2HG or d-malate. Upon purifying PaD2HGDH with 1 mM Zn2+, the Zn2+:protein stoichiometry was 2:1, yielding an enzyme with ∼40 s−1kcat for d-malate. Treatment with 1 mM EDTA decreased the Zn2+:protein ratio to 1:1 without changing the kinetic parameters with d-malate. We observed complete enzyme inactivation for the metalloapoenzyme with 100 mM EDTA treatment, suggesting that Zn2+ is essential for PaD2HGDH activity. The presence of Zn2+ increased the flavin N3 atom pKa value to 11.9, decreased the flavin ε450 at pH 7.4 from 13.5 to 11.8 mM−1 cm−1, and yielded a charged transfer complex with a broad absorbance band >550 nm, consistent with a Zn2+-hydrate species altering the electronic properties of the enzyme-bound FAD. The exogenous addition of Zn2+, Co2+, Cd2+, Mn2+, or Ni2+ to the metalloapoenzyme reactivated the enzyme in a sigmoidal pattern, consistent with an induced fit rapid-rearrangement mechanism. Collectively, our data demonstrate that PaD2HGDH is a Zn2+-dependent metallo flavoprotein, which requires Zn2+ as an essential cofactor for enzyme activity.

Different metals have different effects and functions in enzymes. Reports suggest that metals may be involved in substrate binding in the flavin-dependent α-hydroxy acidoxidizing enzymes, whereas others suggest roles in flavin reduction (27,(39)(40)(41). During enzyme turnover, the bound flavin undergoes a reduction-oxidation cycle, in which the flavin is reduced during the first half of the reaction (39). For flavoproteins that experience a one-electron reduction yielding highly reactive semiquinone radicals, electrons are typically transferred to oxidized cytochromes and metals like Fe 3+ for flavin reoxidation (39). In a study on mitochondrial 2oxoglutarate oxygenase, inhibition of the glutathionemediated 2-hydroxyglutarate oxidation is observed when the essential Fe 2+ or Fe 3+ is substituted with Zn 2+ , Ni 2+ , Cu 2+ , Mn 2+ , or Co 2+ (32). Zn 2+ and Co 2+ increase the activity of D2HGDH from rat liver and D-LDHs from different species (26,27,30); however, they inhibit chromate reductase activity (26). Although Mg 2+ is mainly considered a neutral metal or an inhibitory metal, it has been observed to activate yeast hydroxy acid dehydrogenase and D-lactate-cytochrome (27,41,42). Mn 2+ , on the other hand, can function as an activator or an inhibitor (26,30,32,33). Native-bound metals can be removed with EDTA to render an enzyme inactive and functionally replaced by exogenous metal addition (27,41,42), as is the case for mitochondrial hydroxy acid dehydrogenase (27). The FAD-dependent D-LDH from yeast and Medasphora elsdenii has been reported to contain an essential Zn 2+ per FAD (43). Yet, no mechanisms have been identified for these metals' activation or inhibitory processes in metallo flavoenzymes, including D2HGDH.
D2HGDH from Pseudomonas aeruginosa PAO1 (PaD2HGDH) has become a flavoprotein of recent interest following its identification as a potential therapeutic target against P. aeruginosa (44). P. aeruginosa is an opportunistic multidrug-resistant bacterium (45)(46)(47), which causes fatal nosocomial infections in humans (48,49). PaD2HGDH plays a vital role in P. aeruginosa survival, with no compensatory activity after D2HGDH gene knockouts (31,50). The enzyme is an FAD-dependent dehydrogenase, which does not react with molecular oxygen, follows ping-pong bi-bi steady-state kinetics (45), and oxidizes D2HG and D-malate as substrates (30,31,44). PaD2HGDH, like the FAD-dependent LDH, FADdependent glycolate oxidase, and human D2HGDH belongs to the FAD-dependent α-hydroxy acid-oxidizing enzyme class of proteins that have been demonstrated to use metals or other coenzymes such as NAD + for maximum catalysis (39). Despite the suggestion of PaD2HGDH as a metallo flavoprotein, there is no understanding of its metal composition, binding and coordination, significance and importance, and activation mechanism.
In this study, His-tagged PaD2HGDH from P. aeruginosa PAO1 has been recombinantly expressed, purified to high levels, and investigated for its activity in the presence of Zn 2+ . Zn 2+ has been uncovered as an essential cofactor for PaD2HGDH, rather than an enzyme activator, identifying the enzyme as a metallo flavoprotein. The kinetic and inductively coupled plasma-mass spectrometry (ICP-MS) analysis data of PaD2HGDH treated with and without EDTA are discussed. This study proposes a mechanism for metal reactivation of inactive PaD2HGDH with potential protein ligands for metal binding in PaD2HGDH.

Results
Substrate-induced reduction of PaD2HGDH as purified without Zn 2+ To characterize PaD2HGDH as purified in its kinetic properties, the reductive-half reaction of the enzyme purified in the absence of ZnCl 2 (vide infra) in the purification buffers was investigated by following the substrate-induced absorbance changes of the enzyme-bound flavin at 450 nm. Timeresolved absorption spectroscopy at varying concentrations of D-malate showed incomplete flavin reduction, that is, 15%, in two reaction phases ( Fig. 1) and irrespective of the substrate concentration used. Similar results were obtained with D2HG ( Fig. 1). These data are consistent with the data previously reported for the human enzyme, for which no explanation was provided to account for the partial flavin reduction (51).

ICP-MS analysis of PaD2HGDH
ICP-MS was carried out to determine whether metals were bound to the enzyme as purified after recombinant expression in Escherichia coli. The data revealed significant, although nonstoichiometric, amounts of Mg 2+ and Zn 2+ bound to the enzyme ( Table 2). In contrast, other metals like Co 2+ , Mn 2+ , Fe 2+ , Cd 2+ , and Ni 2+ were present only in trace amounts. Thus, Zn 2+ was the only metal present in significant quantities in the recombinantly expressed PaD2HGDH that also yielded an increase in enzyme catalytic activity. Subsequent preparations of PaD2HGDH were carried out in the presence of 1 mM ZnCl 2 in all purification buffers to evaluate whether the Zn 2+ :protein stoichiometry could be increased. When the ICP-MS analysis was performed on the enzyme prepared in the presence of ZnCl 2 , labeled E-Zn 2+ , the mol Zn 2+ :mol protein ratio was 2.0, whereas that of Mg 2+ was 0.2 ( Table 2).
The E-Zn 2+ enzyme was then treated with either 1 mM EDTA or 100 mM EDTA to explore whether Zn 2+ was labile and could be removed from the enzyme. A molar Zn 2+ :protein stoichiometry of 1.0 was determined using ICP-MS for the enzyme mildly treated with EDTA, labeled E-Zn 2+ 1 mM EDTA, and of 0 for the enzyme harshly treated with EDTA, labeled E-Zn 2+ 100 mM EDTA . Apart from Mg 2+ , all other metals were identified in trace amounts, irrespective of the EDTA treatment (Table 2).

Steady-state kinetics of PaD2HGDH with various Zn 2+ :protein stoichiometries
To investigate the effects of Zn 2+ loading on PaD2HGDH activity, the steady-state kinetic parameters with D-malate and 1 mM PMS as substrates for E-Zn 2+ , E-Zn 2+ 1 mM EDTA, and E-Zn 2+ 100 mM EDTA were measured with the method of the initial rates at pH 7.4 and 25 C. The data showed a Michaelis-Menten pattern with both E-Zn 2+ and E-Zn 2+ 1 mM EDTA , allowing for the determination of the apparent k cat , K m , and k cat /K m values. As shown in Table 3, the kinetic parameters for E-Zn 2+ and E-Zn 2+ 1 mM EDTA were comparable. No enzymatic activity was determined for E-Zn 2+ 100 mM EDTA, irrespective of the substrate concentration.
To investigate the effect of Zn 2+ loading on the ability of PaD2HGDH to utilize molecular oxygen as an electron acceptor, E-Zn 2+ was tested for its reactivity with O 2 by mixing the enzyme with various concentrations of D-malate without PMS in a Clark-type oxygen electrode at pH 7.4 and 25 C. There was no oxygen consumption under these conditions, consistent with previous data on the enzyme purified without exogenous Zn 2+ (44). Thus, PaD2HGDH is a dehydrogenase lacking the ability to use molecular oxygen as an electron acceptor irrespective of the presence of Zn 2+ .

Substrate-induced reduction of E-Zn 2+
The reduction of the enzyme-bound FAD in E-Zn 2+ was followed by monitoring the absorbance at 450 nm upon mixing the enzyme with D-malate to investigate whether the enzyme fully loaded with Zn 2+ could be reduced further than 15%. With 5.4 mM D-malate, there was a 92% reduction of the E-Zn 2+ -bound flavin (Fig. 2). With E-Zn 2+ 100 mM EDTA briefly preincubated with 1 mM ZnCl 2 before the addition of D-malate, a similar 90% flavin reduction was observed (data not shown). In contrast, flavin reduction was only 20% when D-malate was added to E-Zn 2+ 100 mM EDTA before the addition of ZnCl 2 , suggesting that the order of substrate and metal binding to the enzyme is important.

Effects of Zn 2+ on the enzyme-bound flavin
To elucidate the impact of metal on the spectral properties of the enzyme-bound FAD, the UV−visible absorption properties of E-Zn 2+ , E-Zn 2+ 1 mM EDTA , and E-Zn 2+ 100 mM EDTA were investigated and compared. At pH 7.4, the UV−visible absorption spectra of all three enzyme species showed peaks at 380 and 450 nm, a shoulder at 475 nm, and valleys at 320 and 405 nm (Fig. 3). The data are consistent with the reported absorption spectrum of PaD2HGDH purified in the absence of exogenous ZnCl 2 (44). While the spectra of E-Zn 2+ and E-Zn 2+ 1 mM EDTA generally overlapped, the spectrum of E-Zn 2+ 100 mM EDTA showed some differences around the 300 to 450 nm region. To further elucidate whether Zn 2+ affected the flavin microenvironment of PaD2HGDH, the enzyme-bound  flavin for E-Zn 2+ , E-Zn 2+ 1 mM EDTA , and E-Zn 2+ 100 mM EDTA was extracted by heat denaturation and its extinction coefficient when enzyme bound was determined. The resulting flavin ε 450 values decreased with increasing Zn 2+ mole ratio as shown in Table 4. To further provide insights on the effect of Zn 2+ on the electronic properties of the PaD2HGDH-bound flavin, the flavin N 3 atom pK a values of E-Zn 2+ and E-Zn 2+ 100 mM EDTA determined by NaOH titration of the enzyme-bound flavin with simultaneous monitoring of the UV-visible absorption spectra of the flavin species were determined and compared with that of free FAD as shown in Figure 4. The difference spectra for E-Zn 2+ , E-Zn 2+ 100 mM EDTA , and free FAD generated by subtracting a reference spectrum at low pH from all subsequent spectra are shown in Figure 5. The observed absorbance changes at 385 nm as a function of pH were used to determine the pK a values of the flavin N 3 atoms (Fig. 6). The highest pK a value was recorded for the flavin N 3 atom of E-Zn 2+ ( Table 4).

Kinetics of activation of E-Zn 2+
100 mM EDTA with divalent metals The activity of E-Zn 2+ 100 mM EDTA was determined in the presence of chloride salts of Zn 2+ , Co 2+ , Mn 2+ , Cd 2+ , or Ni 2+ at pH 7.4 and 25 C. In all cases, the enzyme activity dependence on the concentration of metal showed a sigmoidal pattern (Fig. 7). From the fit of the data to Equation 5, the concentration of metal yielding a half-maximal increase of enzyme activity, K act , the maximal and limiting enzyme activity at saturating metal concentration, k lim , and the activation coefficient, n, could be estimated ( Table 5). The highest extent of activation, to 19 to 26 s −1 , was observed in the presence of saturating concentrations of Zn 2+ , Co 2+ , or Mn 2+ ( Fig. 7 and Table 5). In contrast, saturating concentrations of Cd 2+ or Ni 2+ yielded increases in the initial rate of reaction to 6 to 8 s −1 . With Zn 2+ , Cd 2+ , and Co 2+ , the transition from inactive to full enzyme activity was sharp, with n values ≥9, and occurred at 60 μM with Zn 2+ and Co 2+ , and 110 μM with Cd 2+ . Mn 2+ and Ni 2+ yielded shallow transitions with n values ≤5, with K act values of 95 μM and 160 μM, respectively.

Surface electrostatic potential map of PaD2HGDH
To determine possible metal-binding pockets in PaD2HGDH, the electrostatic potential map of the PaD2HGDH homology model was constructed using the UCSF Chimera Coulombic Surface Coloring tool. As seen in Figure 8, an electronegative pocket was identified in the enzyme active site, close to H 374 , H 380 , E 420 , and the N 3 -C 4 O region of the flavin. Analysis of the enzyme's surface revealed multiple regions of solvent-accessible electronegative surface pockets around the active site entrance (Fig. 8).

Discussion
Kinetic studies, UV-visible absorption analyses, and ICP-MS have been used to investigate the role of metals in PaD2HGDH activity and demonstrate that Zn 2+ is a cofactor required for the enzyme's activity, rather than an activator. This study also demonstrates that PaD2HGDH can use Co 2+ , Mn 2+ , Cd 2+ , or Ni 2+ as alternative metals and is inactive when only the flavin cofactor is bound. Evidence to support these conclusions and the kinetic mechanism of metal binding to the metalloapoenzyme are provided later.
Zn 2+ is essential for PaD2HGDH activity This conclusion is drawn from the results of the activity assays and ICP-MS analyses of E-Zn 2+ , E-Zn 2+ 1 mM EDTA, and E-Zn 2+ 100 mM EDTA (Tables 2 and 3). The data demonstrate that upon treatment of PaD2HGDH with 100 mM EDTA to yield E-Zn 2+ 100 mM EDTA , the metalloapoenzyme is completely inactive. Thus, Zn 2+ is an essential cofactor for PaD2HGDH activity. Although Zn 2+ was found in the active site of human D2HGDH (52) and was reported as an activator for D2HGDHs from other sources and in other FAD-dependent α-hydroxy acid dehydrogenases (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38), our study is the first to provide unequivocal evidence of the essentiality of Zn 2+ for D2HGDHs. Despite other divalent metals providing enzymatic activity to PaD2HGDH, Zn 2+ was the only one found in significant amounts in the enzyme purified without the addition of any exogenous metal, suggesting that Zn 2+ is the physiological cofactor for the enzyme. Thus, PaD2HGDH, like the FAD-dependent D-LDHs (25,43), is a Zn 2+ -dependent metallo flavoenzyme dehydrogenase. Independent evidence that Zn 2+ is a cofactor for PaD2HGDH activity comes from the substrate reduction of E-Zn 2+ and the enzyme purified in the absence of ZnCl 2 (Figs. 1 and 2, Table 2), showing a six fold increase in the yield of flavin reduction in the presence of Zn 2+ . In addition, the 90% yield of flavin reduction seen only when the metalloapoenzyme was preincubated with ZnCl 2 before substrate addition, but not vice versa, suggests that Zn 2+ binding must precede substrate binding to the metalloapoenzyme for catalysis to occur.  (34) and FAD-dependent histone demethylase (LSD2) (35), which contain a 2:1 Zn 2+ :protein ratio. The observed 2:1 Zn 2+ :protein ratio for E-Zn 2+ can be explained as nonspecific loose metal binding to high-density solventaccessible electronegative surface pockets of PaD2HGDH (Fig. 8). The excess loosely bound metal ions could easily be stripped off by mild concentrations of EDTA, leaving behind the essential and tightly bound Zn 2+ in the enzyme's active site as identified through the electrostatic analysis of the PaD2HGDH homology model (Fig. 8). This phenomenon has been reported for the binuclear Zn 2+ -binding creatininase from P. putida (53). The recently published crystal structure of human D2HGDH provides independent evidence of the presence of an active site Zn 2+ ion for the enzyme (52). This conclusion is supported by the identical active-site topologies of the PaD2HGDH homology model (44) and the human D2HGDH crystal structure (52) (Fig. 9). The crystal structure of human D2HGDH reveals a nitrogen-oxygen-nitrogen (N-O-N) Zn 2+ coordination with active site H 434 , H 441 , E 475 , an active site water molecule, and the flavin O 4 atom (Fig. 9A). Given the full conservation of the PaD2HGDH and human D2HGDH active-site residues (44), and the electrostatic potential map of PaD2HGDH (Fig. 8), an identical Zn 2+ -binding pocket comprising the topologically equivalent active site H 374 , H 381 , E 420 , and the flavin O 4 atom ( Fig. 9B) with water/ hydroxide coordination in the absence of substrate can be proposed for PaD2HGDH. This N-O-N binding pocket has been described for thermolysin and carboxypeptidase A (24). Thermolysin binds D-Ala-D-Ala through an interaction between Zn 2+ and the peptide carbonyl group. Thermolysin's transition state involves Zn 2+ coordination with an oxyanion generated by the nucleophilic addition of water to the peptide carbonyl carbon, resembling the coordination between a carboxylate and an alcohol (54). In the case of carboxypeptidase A, the active site Zn 2+ coordinates between carbonyl groups and N-H or O-H bonds of peptides or esters (55). By comparison,   (56), coordinate metals using three cysteine residues (S-S-S) and water (57).
Notably, other Zn 2+ -dependent enzymes have different binding pockets like the histidine triad (N-N-N) of carbonic anhydrase II and the histidine-cysteine triad (N-N-S) of bacteriophage T7 lysozyme and peptide deformylase (57,58).

Zn 2+ alters the electronic properties of the bound FAD in PaD2HGDH
Evidence to support this conclusion stems from the flavin N 3 atom deprotonation studies of E-Zn 2+ and E-Zn 2+ 100 mM EDTA and the UV-visible absorption spectra of the different enzyme species (Table 4, Figs. [3][4][5]. The UV-visible absorption spectrum for E-Zn 2+ showed a decreased intensity of the 450 nm peak with respect to E-Zn 2+ 100 mM EDTA , with no significant changes around the 380 nm peak and 400 nm valley, consistent with Zn 2+ hydrate perturbing the electronic environment of the flavin. The observed increase in the flavin N 3 atom pK a value in the presence of Zn 2+ can be explained by a polarizing effect of Zn 2+ hydrate on the enzyme-bound flavin. Such a perturbation of the flavin N 3 atom pK a value is likely because of a Zn 2+ -mediated lowering of the pK a value of an active-site water molecule coordinating Zn 2+ to 7.0 to form a hydroxide ion, which is the prevalent species coordinated to Zn 2+ at high pH (57,(59)(60)(61)(62)(63)(64)(65)(66). The proximity of the hydroxide ion's negative charge to the flavin will reduce the likelihood of flavin deprotonation, resulting in an increased flavin N 3 atom pK a value. In addition, the hydroxide ion's proximity to the oxidized flavin results in a charge-transfer complex (67), which is expressed as a long-wavelength absorption of the oxidized flavin above 550 nm, as shown in Figure 4A. Independent evidence for the formation of a Zn 2+coordinated hydroxide ion in the enzyme's active site comes from pL profiles of the k cat and k cat /K m values of E-Zn 2+ with D-malate in aqueous and deuterated buffers, which is presented in the accompanying article (68). The observed decrease in the enzyme-bound flavin ε 450 value with Zn 2+ at pH 7.4, on the other hand, is likely because of an electrostatic effect of the metal, because the active-site hydrated Zn 2+ is only partially neutralized by the coordinated water/hydroxide at low pH. A recent study that investigates the effects of a negative charge on the absorption wavelengths and oscillator strengths of flavins (69) demonstrates an increase in the flavin 450 nm peak intensity. Conversely, a positive charge is expected to decrease the intensity of the 450 nm peak of the flavin, which is expressed as a decreased flavin ε 450 value upon Zn 2+ binding to PaD2HGDH. All taken together, the biophysical data presented show that Zn 2+ alters the electron  distribution across the isoalloxazine ring of the enzyme-bound FAD in PaD2HGDH.

Metal binding to the metallo-apo PaD2HGDH occurs through an induced fit rapid rearrangement mechanism
Evidence to support this conclusion comes from the plots of the concentration dependence of Zn 2+ , Co 2+ , Mn 2+ , Cd 2+ , or Ni 2+ on the activity of E-Zn 2+ 100 mM EDTA using Equation 1 (Fig. 7 and Table 5). The data plots for all five metals indicate a sigmoidal binding pattern with an abrupt increase in enzyme activity. In a study describing protein-ligand interactions, such a steep-sloped sigmodal pattern has been associated with an induced fit rapid rearrangement model (70). This model describes an enzyme's kinetic behavior that involves two distinct phases: a fast phase and a slow phase. As illustrated in Figure 10, before metal binding, the enzyme exists in the inactive E "off" state. The binding of the metal generates an enzyme-metal EM complex, which undergoes a concentration-dependent rapid conformational change, with the forward and reverse rate constants k f and k r , respectively, to yield the active EM* "on" state. At any given time, the enzyme-metal complex population is split into two states in an equilibrium process. The metal concentration dictates the percentage of the enzyme-metal complex population in the on state or off state. Thus, the relative contributions of the enzyme-metal complex populations in the on and off states, at any given time, create the two distinct slow and fast kinetic phases. The rate constant for metal binding k on and dissociation k off and the probability of the EM complex to partition forward to yield the EM* complex define the slow phase, as shown in Equation 1.
In contrast, the fast phase is given by the sum of the forward and reverse rate constants of enzyme isomerization of the enzyme-metal complexes after metal binding (Equation 2). As the metal concentration increases, the fast phase overcomes the slow phase, and the on state becomes more favored, resulting in the optimal generation of EM* that undergoes turnover. The resulting rapid generation of the EM* is observed as a sharp rise in enzyme activity (Fig. 7). In Equation 5 and for any given metal, the extent of activation is directly measured from the steepness of the burst phase through the activation coefficient n of the sigmoidal plot. A metal that requires the lowest concentration to reach a halfmaximal activity, which is probed by the K act value, would be the most active. Thus, Zn 2+ with an n value of 50 and a K act value of 62 μM is the most preferred metal for PaD2HGDH (Table 5). Of the other four metals restoring the activity of the metalloapoenzyme, Co 2+ is the only one yielding a K act value of 61 μM, which is comparable to that seen with Zn 2+ . In addition, the activity of the metalloapoenzyme increased to a k lim value of 20 s −1 for the maximal enzyme activity at saturating metal concentration with both Zn 2+ and Co 2+ , suggesting that the two metals play an interchangeable role in the enzyme's function. An increase in the enzyme's activity to a similar extent by Zn 2+ and Co 2+ was previously demonstrated in carbonic anhydrase (66). Similar sigmoidal patterns have been reported for the activation, that is, opening, of gated ion channels in response to metals and tissue responses to irreversible electroporation during tumor suppression (71)(72)(73)(74).
In conclusion, this study establishes that PaD2HGDH is a metallo flavoenzyme dehydrogenase that requires an essential and tightly bound Zn 2+ for activity with a metal:protein ratio of 1:1. The enzyme is inactivated upon removal of the bound Zn 2+ and is restored in its kinetic properties when exogenous amounts of the chloride salts of Co 2+ , Zn 2+ , Mn 2+ , Cd 2+ , or Ni 2+ are added. The binding of the divalent metals to PaD2HGDH occurs through an induced-fit rapid rearrangement mechanism. For most alcohol dehydrogenases, such as liver alcohol dehydrogenase (56,57), that employ metals as cofactors, the alcohol substrates oxidized during catalysis are usually simple alcohols, such as butanol. In the case of alcohol dehydrogenases that utilize flavins as cofactors, the alcohol substrates oxidized are generally primary and secondary alcohols such as isopropanol (75). In NADH-dependent α-hydroxy acid oxidases such as the Plasmodium vivax LDH, the enzyme's active site is designed with positively charged amino acid residues to coordinate the lactate carboxylate group (76). In PaD2HGDH, nature employs a metal besides a flavin for catalysis to yield a neutral active site for the oxidation of the negatively charged dicarboxylic acid substrate. The study also demonstrates that Zn 2+ alters the electronic properties of the bound flavin through a metal-mediated polarizing effect that results in altered UV-visible absorption spectra, increased flavin N 3 atom pK a value, and decreased ε 450 value of the flavin with Zn 2+ . This study provides a remedy for the incomplete flavin reduction observed in D2HGDHs and a rationale for the need for two cofactors, that is, a flavin and a divalent metal. The study also demonstrates mechanisms of PaD2HGDH activation and inactivation that can serve as a direction for future designs of new therapeutic targets against P. aeruginosa infections.  Table 5. PaD2HGDH, D2HGDH from Pseudomonas aeruginosa; PMS, phenazine methosulfate.

Materials
The PaD2HGDH pET20b(+) plasmid harboring the PA0317 gene was designed in-lab and purchased from GenScript. The plasmid was sequenced to verify the presence of the wildtype gene. E. coli strain Rosetta(DE3)pLysS was from Novagen. Bovine serum albumin was purchased from Promega. Luria-Bertani (LB) agar, LB broth, chloramphenicol, IPTG, lysozyme, sodium hydrosulfite (dithionite), PMS, and PMSF were obtained from Sigma-Aldrich. Ampicillin was purchased from ICN Biomedicals. D-2-hydroxyglutarate was purchased from MilliporeSigma. D-malate was purchased from Alfa Aesar. EDTA, glycerol, and all other reagents were of the highest purity commercially available.

Expression and purification of PaD2HGDH
To obtain pure enzyme for kinetic studies, a 10 ml LB broth medium containing 100 μg/ml ampicillin and 34 μg/ml chloramphenicol was inoculated with frozen stocks of E. coli cells Rosetta(DE3)pLysS harboring the PaD2HGDH pET 20b(+) plasmid. The cell cultures were used to inoculate 1 l LB broth and incubated on a rotatory plate at 37 C and 180 rpm for 18 h. Protein expression was induced with 100 μM IPTG when cell density reached an absorbance at 600 nm of 0.6. The temperature of the culture was then lowered to 18 C while shaking on a rotatory plate at 180 rpm. After 17 h of expression, the cells were harvested by centrifugation for 30 min at 2800g and 4 C.
The lysis buffer containing 1 mM PMSF, 2 μg/ml DNase or RNase, 4 mg/ml lysozyme, 5 mM MgCl 2 , 300 mM NaCl, 10 mM imidazole, 10% glycerol, and 20 mM NaPO 4 , pH 7.4, was used to resuspend the wet cell paste in a ratio of 1 g of the wet cell paste to 4 ml of lysis buffer. The suspended cells were then incubated for 30 min on ice while stirring. The resulting slurry was sonicated in five cycles of 5 min each with 5 min off intervals, and then the cell debris was removed by centrifugation at 11,200g for 30 min. The resulting cell-free extract supernatant was purified to homogeneity using a nickelnitrilotriacetic acid column, equilibrated with buffer A (20 mM Tris-Cl, 10 mM imidazole, 300 mM NaCl, and 10% glycerol, pH 7.4). The purification was carried out using a Unicorn ÄKTA Start purification system. Elution of the bound protein was through a gradient from 0 to 100% buffer B (20 mM Tris-Cl, 500 mM imidazole, 300 mM NaCl, and 10% glycerol, pH 7.4), with PaD2HGDH eluting at 40% buffer B. The solution containing the purified protein, typically 15 ml, was dialyzed against five 2 l changes of 10% glycerol, 20 mM Tris-Cl, pH 7.4, for 24 h, at 4 C. The purified enzyme was stored in single-use aliquots in 10% glycerol, 20 mM Tris-Cl, pH 7.4, at −20 C, and was stable for at least 6 months.  Figure 7 to Equation 5. b k lim is the limiting rate of catalysis at saturating metal concentrations. c K act is the metal concentration yielding half k lim . d n is the activation coefficient associated with the steepness of the sigmoidal activity change in Figure 7. with FAD (yellow sticks) and D2HG (green sticks). C, highly electronegative protein surface pockets suitable for nonspecific Zn 2+ binding to PaD2HGDH. Red, electronegative region; blue, electropositive region; white, neutral region. Images were generated after structural analyses of the PaD2HGDH homology model using UCSF Chimera software (77). PaD2HGDH, D2HGDH from Pseudomonas aeruginosa.
To obtain the Zn 2+ -bound enzyme, purifications were carried out by following the aforementioned protocol using 25 mM NaPO 4 as a buffer with 1 mM ZnCl 2 in all purification buffers.
To rid the enzyme of all bound metals, dialysis of the purified enzyme was performed in 2 l volumes of 100 mM NaPO 4 , 1 or 100 mM EDTA, 1 mM dithiothreitol, pH 7.4, for 72 h with no buffer changes at 4 C.

Reductive-half reaction
To determine the K d values for D2HG and D-malate with the recombinantly expressed PaD2HGDH without Zn 2+ , the reduction of the enzyme-bound flavin was followed by monitoring the decrease in absorbance at 450 nm upon mixing PaD2HGDH anaerobically with varying concentrations of the reducing substrate. The time-resolved absorption spectroscopy of the reduction of the enzyme-bound flavin with D2HG or D-malate was carried out with an SF-61DX2 Hi-Tech KinetAsyst high-performance stopped-flow spectrophotometer (TgK-Scientific) thermostated at 25 C and equipped with a photomultiplier detector under anaerobic conditions. The reductive half-reaction was performed under pseudo-firstorder conditions where the enzyme concentration after mixing with the substrate was 9 μM, and that of the reducing substrate was between 80 and 800 μM for D2HG or 0.6 and 60 mM for D-malate. The enzyme purified without Zn 2+ was equilibrated with 25 mM Tris-Cl, pH 7.4, using a PD10 column. Equal volumes of the enzyme and the reducing substrate were mixed in the stopped-flow spectrophotometer in singlemixing mode following established procedures. The instrument dead time for mixing was 2.2 ms.
The stopped-flow traces were fit to Equation 3 using the KinetAsyst 3 (TgK-Scientific) software, which describes a double-exponential process where A represents the absorbance at 450 nm at time t, B 1 , and B 2 represent the amplitudes of the decrease in absorbance, k obs1 and k obs2 define the observed rate constants for the change in absorbance associated with flavin reduction. C is an offset value accounting for the nonzero absorbance of the enzyme-bound reduced flavin at an infinite time. . Images were generated after structural analyses of the protein files using UCSF Chimera software (77). PaD2HGDH, D2HGDH from Pseudomonas aeruginosa. Figure 10. General kinetic scheme of the induced fit rapid isomerization mechanism.
PaD2HGDH activity with divalent cations To investigate the effect of divalent cations on the activity of natively expressed PaD2HGDH purified without metals, the enzyme's activity was analyzed by adding exogenous amounts of 100 μM chloride salts of Co 2+ , Ca 2+ , Zn 2+ , Mn 2+ , Mg 2+ , Fe 2+ , Cd 2+ , Cu 2+ , or Ni 2+ to an enzyme reaction mixture. The enzyme reaction mixture comprised of buffer, PaD2HGDH, reducing substrate, and an artificial electron acceptor. Activity assays for PaD2HGDH with 5 mM D-malate as a substrate and 1 mM PMS as the electron acceptor were measured by monitoring the initial oxygen consumption rates with a computer-interfaced Oxy-32 oxygen-monitoring system (Hansatech Instruments Ltd) after the exogenous addition of metals in 25 mM Tris-Cl, pH 7.4 and 25 C. The data were analyzed using Microsoft Excel (data not shown).

PaD2HGDH metal component analysis
To determine the concentrations of divalent metals incorporated into PaD2HGDH after recombinant expression in E. coli Rosetta(DE3)pLysS cells after purification with or without the exogenous addition of Zn 2+ , the metal composition of PaD2HGDH was investigated using ICP-MS analysis. Aliquots of all enzyme species, namely the enzyme purified without exogenous Zn 2+ (E), the enzyme purified and stored in NaPO 4 and ZnCl 2 (E-Zn 2+ ), the enzyme treated with 1 mM EDTA (E-Zn 2+ 1 mM EDTA ), and the enzyme treated with 100 mM EDTA (E-Zn 2+ 100 mM EDTA ), were dialyzed against 2 l of deionized water for 24 h. The resulting enzyme solutions were then submitted for ICP-MS analyses at the Center for Applied Isotope Studies, University of Georgia, Athens, GA. The resulting data were analyzed using Microsoft Excel.

Oxygen reactivity
To investigate the effect of Zn 2+ incorporation on the O 2 reactivity of PaD2HGDH, the enzymatic activity of E-Zn 2+ with molecular oxygen was measured by monitoring the initial oxygen consumption rate with a computer-interfaced Oxy-32 oxygen-monitoring system (Hansatech Instruments Ltd) at 25 C. To test for oxygen reactivity, the reaction mixture contained 40 nM enzyme and D-malate between 1.6 and 40 mM in 25 mM NaPO 4 , pH 7.4. After 3 min of reaction time, 1 mM PMS was added to the reaction mixture to regenerate the oxidized state of the enzyme at the expense of O 2 . The instantaneous reoxidation of the enzymatically reduced PMS by molecular oxygen was then observed (data not shown).

Enzyme activity assay
To investigate the kinetic properties of the various PaD2HGDH species, the apparent steady-state kinetic parameters of E-Zn 2+ , E-Zn 2+ 1 mM EDTA , and E-Zn 2+ 100 mM EDTA with D-malate as a substrate and PMS as an artificial electron acceptor were determined by monitoring the initial rates of oxygen consumption with a computer-interfaced Oxy-32 oxygen-monitoring system (Hansatech Instruments Ltd). Dmalate concentrations were between 1.6 and 40 mM, and PMS concentration was fixed at a saturating concentration of 1 mM.
The enzyme concentration was 7 nM, and the buffer was 25 mM NaPO 4 , pH 7.4, and 25 C. Data analysis was performed using the KaleidaGraph software (Synergy Software) or Enzfitter software (Biosoft). For the apparent steady-state kinetics, the Michaelis-Menten equation was used.

Substrate-induced reduction of E-Zn 2+
To determine the effect of Zn 2+ incorporation on the reductive properties of the enzyme-bound FAD, the percent flavin reduction of E-Zn 2+ in the presence of D-malate was investigated. The reaction followed the UV−visible absorption peak of E-Zn 2+ oxidized flavin at 450 nm in a 1 cm path length quartz cuvette using an Agilent Technologies model HP 8453 PC diode-array spectrophotometer equipped with a thermostated water bath. The experiment was carried out under anaerobic conditions to avoid any potential effects of O 2 on flavin reduction. The enzyme was initially incubated in 25 mM NaPO 4 , pH 7.4, followed by the repeated addition of D-malate aliquots until a final D-malate concentration of 5.4 mM was achieved. The quenching of the 450 nm peak associated with flavin reduction was observed with each addition of D-malate. To determine the end point of flavin reduction, sodium dithionite was added to the reaction mixture as a chemical reducing agent. The absorbance at 450 nm after adding sodium dithionite was marked as the endpoint of flavin reduction. The spectra data were analyzed using the KaleidaGraph software.

Effects of Zn 2+ on the biophysical properties of PaD2HGDHbound FAD
To understand the impact of Zn 2+ on the spectral properties of PaD2HGDH, the various enzyme species were subjected to UV−visible absorption studies. The spectral properties of E-Zn 2+ , E-Zn 2+ 1 mM EDTA , and E-Zn 2+ 100 mM EDTA were investigated using an Agilent Technologies model HP 8453 PC diode-array spectrophotometer equipped with a thermostated water bath at 18 C. The spectroscopic fingerprints for all enzyme species were determined by obtaining the absorption spectra of the enzyme-bound flavin between 280 and 800 nm. The flavin absorbance data were observed using a 1 cm pathlength quartz cuvette. The spectra data were analyzed using the KaleidaGraph software.
To understand the impact of Zn 2+ on the flavin microenvironment of PaD2HGDH, the extinction coefficient of the enzyme-bound FAD was determined. The extinction coefficients of E-Zn 2+ , E-Zn 2+ 1 mM EDTA , and E-Zn 2+ 100 mM EDTA were determined after enzyme heat denaturation at 100 C for 20 to 40 min. The absorbance values of the extracted flavins followed at 450 nm were used to calculate the molar extinction coefficients for the respective enzyme species. The extinction coefficient of free FAD at 450 nm was used as the reference in determining the molar extinction of the FAD bound to E-Zn 2+ , E-Zn 2+ 1 mM EDTA , or E-Zn 2+ 100 mM EDTA . To understand the impact of Zn 2+ on the electronic properties of the PaD2HGDH-bound flavin, the pK a values of the flavin N 3 atoms of E-Zn 2+ and E-Zn 2+ 100 mM EDTA were investigated and compared with that of free FAD. E-Zn 2+ 1 mM EDTA was not investigated in this analysis because the enzyme is expected to behave like E-Zn 2+ after observing similar spectral and kinetic properties for E-Zn 2+ 1 mM EDTA and E-Zn 2+ (vide supra). The pK a values of the flavin N 3 atoms of E-Zn 2+ and E-Zn 2+ 100 mM EDTA were investigated by NaOH titration of the enzyme-bound flavin with simultaneous monitoring of the UV-visible absorption spectra of the flavin species from pH 9.0 to 12.5. The enzymes were stable at all pH values tested. The resulting spectra were then analyzed using the KaleidaGraph software. Flavin difference spectra were generated by subtracting the spectrum of a reference pH, that is, 9.0 for E-Zn 2+ and E-Zn 2+ 100 mM EDTA and 8.0 for free FAD, from all subsequent spectra. The changes in the absorbance of the high energy valley for each flavin species at 385 nm for E-Zn 2+ and 387 nm for E-Zn 2+ and free FAD were plotted as a function of pH. The resulting plots were fit to Equation 4, which describes the sigmoidal decay of the absorbance at 385 nm as a function of pH. Y is the absorbance at 385 nm. C L and C H represent low and high limiting values describing lower and upper offsets in absorbance changes. pK a is the resulting pK a value of the flavin N 3 atom. The spectra data were analyzed using the KaleidaGraph software.

Reactivation of inactive PaD2HGDH by metals
To investigate the effects of divalent metals on PaD2HGDH activity, the concentration dependence of Co 2+ , Zn 2+ , Mn 2+ , Cd 2+ , or Ni 2+ on the activity of E-Zn 2+ 100 mM EDTA was investigated in 25 mM NaPO 4 , pH 7.4, at 25 C, with fixed concentrations of D-malate and PMS at 100 mM and 1 mM, respectively. The initial enzyme reaction rates were monitored after adding exogenous chloride salts or the metals to yield final concentrations between 0.01 mM and 0.5 mM. The resulting data were fit to Equation 5, in which Y is the fraction of PaD2HGDH concentration bound by the metal, [M] is the total metal concentration, n is the activation coefficient, K act is the metal concentration producing half-saturation of PaD2HGDH, and k lim is the maximal and limiting enzyme activity at saturating metal concentration. The data were analyzed using the KaleidaGraph software.
Y ¼ k lim ½M n K act n þ½M n (Eq 5)

Electrostatic potential map of PaD2HGDH
To determine possible metal-binding pockets in PaD2HGDH, the electrostatic properties of the PaD2HGDH homology model (44) were investigated using the UCSF Chimera software (77). The PaD2HGDH electrostatic potential revealing the surface and binding properties of the protein was generated using the UCSF Chimera Coulombic Surface Coloring tool, which calculates the electrostatic potential according to Coulomb's law. The color code was set to present red for negative potential, white for neutral, and blue for positive potential. The resulting protein surface and activesite potential maps were then saved as images for graphical comparison. To better understand the role of enzyme electrostatic potential in the binding of ligands to PaD2HGDH, the physiological substrate D2HGD and Zn 2+ were modeled into the PaD2HGDH homology model and analyzed in light of the electrostatic potential map generated for the enzyme's active site.

Data availability
All data are contained within the article.
Acknowledgments-A special thanks to Dominic Dawutey, Dr Samer Gozem, Theresa Akoto, Bilkis Mehrin Moni, Jessica Eyram Kugblenu, Trea Roosevelt Martin, Kendall Wood, and all previous members of the Gadda group for their encouragement, support, and insightful contributions toward this study. We acknowledge the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from the National Institutes of Health P41-GM103311, for the molecular graphics and analyses performed with UCSF Chimera. Protein Data Bank accession codes for PaD2HGDH (Q9I6H4) and hD2HGDH (6LPP).
Author contributions-J. A. Q. and G. G. methodology; J. A. Q. investigation; J. A. Q. and G. G. data curation; J. A. Q. and G. G. writing-original draft.
Funding and additional information-This work was supported in part by Georgia State University, Department of Chemistry. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.