A single mutation in the castor Δ⁹-18:0-desaturase changes reaction partitioning from desaturation to oxidase chemistry

Abstract

Sequence analysis of the diiron cluster-containing soluble desaturases suggests they are unrelated to other diiron enzymes; however, structural alignment of the core four-helix bundle of desaturases to other diiron enzymes reveals a conserved iron binding motif with similar spacing in all enzymes of this structural class, implying a common evolutionary ancestry. Detailed structural comparison of the castor desaturase with that of a peroxidase, rubrerythrin, shows remarkable conservation of both identity and geometry of residues surrounding the diiron center, with the exception of residue 199. Position 199 is occupied by a threonine in the castor desaturase, but the equivalent position in rubrerythrin contains a glutamic acid. We previously hypothesized that a carboxylate in this location facilitates oxidase chemistry in rubrerythrin by the close apposition of a residue capable of facilitating proton transfer to the activated oxygen (in a hydrophobic cavity adjacent to the diiron center based on the crystal structure of the oxygen-binding mimic azide). Here we report that desaturase mutant T199D binds substrate but its desaturase activity decreases by ≈2 × 10³-fold. However, it shows a >31-fold increase in peroxide-dependent oxidase activity with respect to WT desaturase, as monitored by single-turnover stopped-flow spectrometry. A 2.65-Å crystal structure of T199D reveals active-site geometry remarkably similar to that of rubrerythrin, consistent with its enhanced function as an oxidase enzyme. That a single amino acid substitution can switch reactivity from desaturation to oxidation provides experimental support for the hypothesis that the desaturase evolved from an ancestral oxidase enzyme.

Nonheme diiron-containing four-helix-bundle proteins possess the ability to functionalize unactivated C-H groups and mediate a diversity of chemical reactions including oxidation, hydroxylation, desaturation, and epoxidation (1, 2). A wealth of mechanistic information is available from various diiron-containing proteins including methane monooxygenases, Δ⁹ desaturases, ribonucleotide reductases, rubrerythrins, alternate oxidases, ferritins, and bacterioferritins (1–3).

The diiron-containing proteins are highly divergent in their amino acid sequences, with identities typically falling below that necessary for conventional phylogenetic analysis. However, when the analysis is restricted to the four helices that coordinate the diiron active site, the amino acid identity rises to 16–31% (4). A shared diiron-binding motif within the conserved four-helix bundle is involved in oxygen chemistry. The reactions have been described as occurring in two phases, an oxygen activation phase followed by reaction phases (1). Oxygen activation likely placed strong evolutionary constraints on the organization of the diiron center, whereas the reaction phases exhibit great diversity of functional outcome. In addition to their individual catalytic reactions, rubrerythrin, methane monooxygenase, ribonucleotide reductase, and the Δ⁹ desaturase have also been shown to reduce dioxygen to water (4–6). Based on these similarities, Gomes et al. (4) proposed that the four-helix bundle diiron proteins arose from a common ancestor that bound activated oxygen species and reduced them to water. This hypothetical oxidase enzyme is thought to have appeared at the transition from anaerobic to aerobic environment, ≈2.5 billion years ago.

We previously performed a structural comparison of the active site of the Δ⁹ desaturase with that of rubrerythrin, an NAD(P)H peroxidase, which revealed remarkable similarity of the diiron ligands (7). Based on this structural analysis we proposed that residue 199, which occupies a location adjacent to the diiron site and abuts the hydrophobic substrate binding cavity, plays a key role in determining the chemical outcome of the enzyme (7). In the desaturase it is occupied by threonine, and in the rubrerythrin it is occupied by a glutamic acid. In this work we report that the T199D mutant of the Δ⁹ desaturase shows greatly reduced desaturation activity but increases its oxidase activity by >31-fold with respect to the WT desaturase. A crystal structure of the T199D mutant is presented that shows very close active-site similarity to rubrerythrin, consistent with its change in functionality.

Results and Discussion

Structural alignment of the reduced azide complexes of Δ⁹ desaturase and rubrerythrin (8) revealed similarities with respect to the position and identity of iron binding ligands and the position of the azide adduct (7) (Fig. 1). The single major difference in the active site is the identity of the residue corresponding to threonine-199 in the desaturase, which is a glutamic acid in rubrerythrin. The side chain of the residue occupying this position faces the bound azide that mimics the binding site of molecular oxygen. Thus, the desaturase contains threonine, a poor proton donor, whereas rubrerythrin contains a glutamic acid, which facilitates proton transfer. We previously hypothesized that the presence or absence of a proton donor in this position might influence the partitioning of chemical reactivity of the diiron site between desaturation and oxidase chemistry (7). Thus, we engineered mutations at position 199 into the desaturase to replace threonine with either glutamic or aspartic acid and compared the desaturase and oxidase activity of these mutants to those of WT Δ⁹ desaturase.

Fig. 1.

Crystal structures of the reduced azide complexes of desaturase (Upper) and rubrerythrin (Lower).

T199D and T199E mutants were expressed as soluble proteins in Escherichia coli and purified with similar yields as WT Δ⁹ desaturase. Purified proteins were straw yellow in appearance, and their spectra showed absorption features in the 300- to 500-nm range characteristic of ligand-to-metal charge transfer bands characteristic of oxidized WT Δ⁹ desaturase (9). Mutants T199D and T199E showed reductions of ≈2 × 10³ in their rates of desaturation (Table 1). Because T199D and T199E mutations are adjacent to the substrate binding cavity, we tested for possible changes in chain length specificity; however, neither mutant showed any increased preference for either 16:0- or 14:0-ACP.

Table 1.

Activities of desaturase enzymes

Enzyme	Desaturation with 18:0-ACP		Oxidation with H2O2
	kcat,* min−1	Fold WT	Rate constant,† M−1·s−1	Fold WT
T199 (WT)	42.3 (1.6)	—	3.6 × 103	—
T199E	0.022 (0.013)	<10−3	4.5 × 103	1.3
T199D	0.021 (0.011)	<10−3	1.1 × 105	31

*Desaturase assays, with mean standard error in parentheses (n = 3).

^†Oxidation assay. Rate constants were obtained from the slope of curves in Fig. 2, and each estimate is composed of 14 or more separate experiments.

The introduction of a carboxylate at position 199 introduces a charged residue into a primarily hydrophobic substrate binding channel, raising the possibility that desaturation is prevented because the 18:0-ACP substrate is unable to bind to the T199D mutant desaturase. We therefore tested whether T199D is able to bind substrate with the use of HPLC size-exclusion chromatography performed in the presence of high salt to prevent nonphysiological electrostatic enzyme–substrate association (10). WT desaturase and the T199D mutant both show an ≈5-kDa increase in molecular mass when incubated with 18:0-ACP (Table 2) but no change when incubated with unacylated holo-ACP, indicating that both WT and T199D are capable of binding substrate. These data suggest that loss of desaturation activity of T199D does not result from an inability to bind substrate.

Table 2.

Apparent molecular masses of desaturase preparations

Enzyme	Desaturase plus			Δ (18:0-ACP, buffer)
	Buffer	Holo-ACP	18:0-ACP
T199 (WT)	72.81 (0.59)	71.48 (1.60)	77.30 (0.73)	4.50 (0.46)
T199D	73.20 (0.28)	72.26 (0.99)	78.36 (0.17)	5.16 (0.40)

Data are mean apparent molecular mass in kilodaltons, with standard deviation in parentheses (n = 3).

If replacement of the hydroxy-containing threonine for carboxylate functionality in mutants T199E or T199D increases rubrerythrin-like catalysis, we predicted they should exhibit enhanced capacity to reduce peroxide to water. Thus, we considered various approaches to measuring the rate of peroxide reduction in the WT Δ⁹ desaturase and the T199 mutants. To perform the experiment physiologically requires the presence of the natural electron donor ferredoxin and its reductase, ferredoxin NADPH(+) oxidoreductase. However, ferredoxin and ferredoxin NADPH(+) oxidoreductase contain chromophores that mask the ligand-to-metal charge transfer bands of the Δ⁹ desaturase, preventing the monitoring of their rate of appearance upon reoxidation of the desaturase. In addition, the use of the physiological electron transport chain was discounted because uncoupling of the electron transport chain when a nonnatural substrate was provided to the desaturase has been reported (11). However, an unusual property of the desaturase is that its autooxidation rate in the absence of substrate is >10³-fold slower that those of other diiron proteins such as the R2 component of ribonucleotide reductase or the hydroxylase component of methane monooxygenase (6). The relative stability of reduced desaturase allowed us to separate it from excess reductant by size-exclusion chromatography. Time-resolved single-turnover reoxidation experiments were then performed by reacting the reduced desaturase with various concentrations of peroxide in a stopped-flow spectrophotometer. The peroxide-dependent reoxidation rate of the desaturase was determined for WT, T199E, and T199D (see Table 1). A previous report established that 4e⁻-reduced Δ⁹ desaturase–substrate complex is capable of reducing dioxygen to water (6); in this study we observed a peroxide-dependent rate of desaturase reoxidation in the absence of bound substrate (Fig. 2). The reoxidation rate increased only modestly, by ≈30%, upon substitution of a glutamic acid at position 199. However, the introduction of an aspartic acid at position 199 resulted in a >31-fold increase in the reoxidation rate.

Fig. 2.

Pseudo first-order rate constants and H₂O₂ concentration dependency for the WT (●), T199E (○), and T199D (○) mutant proteins determined at 10°C.

A scheme representing the oxidase activity described in these experiments is shown in Fig. 3.

Fig. 3.

A schematic to describe the reaction of the desaturase T199D.

The result that the aspartic acid substitution had a substantial effect whereas substitution with glutamic acid had little effect suggests that the active-site geometry attained by the T199D mutant is better suited to reducing peroxide. To investigate the relative position and orientation of the aspartic acid side chain in T199D with that of WT desaturase and of rubrerythrin we crystallized T199D and solved its structure at 2.65 Å (Table 3 and Fig. 4). No significant conformational changes beyond the active-site region are seen between the T199D model (Fig. 4) and previously published models of the reduced native castor desaturase. The electron density in the region of the active site is of excellent quality; however, as reported for previous desaturase models, the N terminus and the regions comprising residues 205–215 and 338–348 show less well defined electron density. The iron–iron distance of 4.2 Å is that of the diferrous iron center, presumably resulting from reduction by the x-ray exposure as also observed in previous desaturase structures (7, 19). The electron density for the threonine-to-aspartic acid mutation at position 199 is clearly visible in the active site of each monomer. As shown in Fig. 4, the carboxyl group of D199 occupies a similar, although not identical, position to that of E97 in rubrerythrin and is well situated to facilitate proton transfer. Because the main chain of the desaturase is ≈1 Å closer to the diiron site at residue 199 than the equivalent residue 97 of the rubrerythrin structure, the shorter side chain length of the aspartate positioned its carboxylate in an approximately equivalent position to that of E97 of rubrerythrin with respect to the diiron site.

Table 3.

Crystallography data collection and refinement summary

Measurement	Value
Data collection
Space group	P212121
Cell axis a, Å	82.05
Cell axis b, Å	145.77
Cell axis c, Å	193.25
No. of molecules in asymmetric unit	6
Resolution, Å	2.65
Rsym	0.093 (0.474)
I/σ	9.6 (2.3)
Completeness	98.4 (98.4)
Refinement
Refinement program	REFMAC5
TLS model	6 TLS groups
Reflections in working set	68,033
Reflections in test set	3,388
R-factor, %	24.0
Rfree, %	27.1
No. of atoms modeled	17,049
No. of irons	12
No. of waters	150
Average B-factor protein	36.2
Average B-factor solvent	17.3
rmsd from ideals
Bonds, Å	0.016
Angles, °	1.37
Ramachandran plot
Most favored, %	90.3
Additionally allowed, %	9.1
Generously allowed, %	0.3
Disallowed, %	0.3

Statistics for the highest-resolution shell are given in parentheses where appropriate.

Fig. 4.

A view of the superimposed active sites of the desaturase T199D mutant (green) and of reduced rubrerythrin (blue), showing the similar position of the putative proton donor groups.

During refinement, difference density in the active site (Fig. 5), corresponding to a ligand bound both by the mutated D199 residue and the diiron center became apparent, coincident with the azide binding position of the azide–desaturase complex. The density was initially modeled as a water molecule, but significant positive density remained after refinement. It was subsequently found that the density could be described almost equally well by modeling either two waters or a dioxygen molecule. The most accurate description appears to lie somewhere between the two as, unrestrained, the distance between the two oxygen atoms refined to 1.4–1.5 Å. Precedence exists for both models, with the reduced form of rubrerythrin (12) containing two waters, and sulerythrin (13) and rubredoxin:oxygen oxidoreductase (14) each describing a putative dioxygen coordinating iron center. Although the most likely explanation is a combination of the two states, it was ultimately decided to model the density as two waters because of the limited resolution of the structure. After refinement this resulted in a relatively short OO distance of between 2.2 and 2.4 Å. We do not rule out the possibility that the density could be that of a dioxygen species, particularly as the binding site is very similar to that of the peroxo-mimic azide in both desaturase and rubrerythrin.

Fig. 5.

The active site of the T199D mutant, showing an omit map (contoured at 3σ) of the difference density that was ultimately modeled as two water molecules.

In the model (Fig. 4), the waters W2 interacts with OD1 of the mutated Asp-199 residue at a distance of 2.4 Å, whereas W1 interacts very weakly at a distance of 3.2 Å. Both also interact with the diiron center, W2 binding at a distance of 2.3 Å to Fe2 and W1 interacting more weakly with a distance of 2.7–2.9 Å to Fe1. W2 is also within hydrogen bonding distance of OE2 of residue Glu-229. In this model, the apparent difference in water binding relative to WT and its similarity to rubrerythrin correlates with the observed change in functionality and supports Yoon and Lippard's suggestion (15) that the amount of accessible water in nonheme diiron(II) enzymes might act as a control element for achieving diverse functions using a shared structural motif. Beyond the mutated residue and the putative waters, no further structural changes are seen in the active site when compared with the native desaturase structure.

Gomes et al. (4) previously proposed that the four-helix bundle diiron protein family evolved from an ancestral rubrerythrin-like oxidase enzyme that was responsible for reducing oxygen to water. Correspondence of the identity and relative orientation of residues in the actives site of the desaturase and rubrerythrin are remarkable in light of the absence of overall detectable homology between the two enzyme families. Results presented here demonstrate that conversion of the hydroxy functionality of T199 to carboxylate functionality in the T199D mutant diminished desaturase activity by ≈2 × 10³-fold and increased the oxidase activity by >31-fold. Effecting a profound change in chemical reactivity of an enzyme by a single amino acid substitution, i.e., loss of desaturase activity accompanied by a large increase in oxidase activity, provides experimental support for the hypothesis that the desaturase evolved from an ancestral oxidase enzyme.

Materials and Methods

Desaturase Expression, Purification, and Enzyme Assay.

Castor recombinant desaturase was generated by expression in plasmid pET9d in E. coli BL21(DE3) that were grown in LB media in a New Brunswick Scientific (Edison, NJ) G25 incubator shaker at 37°C until OD₆₀₀ ≈ 0.5, at which time isopropyl-β-d-thiogalactopyranoside was added to 0.1 mM (16). The temperature was lowered to 30°C, and the culture was shaken at 275 rpm for a further 4 h. Cells were collected by centrifugation, resuspended in 5 vol of 7 mM Hepes, 7 mM Mes, 7 mM NaOAc, 4 mM MgCl₂, and 6 Kunitz units/ml DNase I (pH 7.4), and lysed by passage through a French pressure cell with a 10⁴-psi pressure drop. The lysate was clarified by centrifugation at 45,000 × g for 30 min. The supernatant was applied to a 12-ml Poros 20 CM column equilibrated with 7 mM Hepes, 7 mM Mes, and 7 mM NaOAc (pH 7.4) (equilibration buffer). After loading, the column was washed with 10 vol of equilibration buffer before elution with a linear gradient of 0–600 mM NaCl in equilibration buffer. The resulting desaturase was judged to be >90% pure by SDS/PAGE. The resulting enriched desaturase was concentrated with the use of an Amicon PM30 ultrafilter (Milliport, Framingham, MA) and subjected to HPLC size-exclusion chromatography with the use of a preparative G-3000SW (Toso Haas, Montgomeryville, PA) developed with 20 mM Hepes/70 mM NaCl (pH 7.0). Castor Δ⁹-18:0-ACP desaturase variants were assayed with [1-¹⁴C]18:0-ACP substrate with the use of recombinant spinach ACP-I (17). Methyl esters of fatty acids were analyzed by argentation TLC, and radioactivity in products was quantified as previously described (18). Δ⁹-18:0-ACP desaturase assays were performed in triplicate.

Stopped-Flow Kinetic Experiments.

The WT and mutant desaturase proteins at concentrations between 5 and 8 mg/ml in 0.5 M CAT buffer {CAT designates equal proportions of Mes [2-(N-morpholino)ethanesulfonic acid 4-morpholineethanesulfonic acid], Hepes [4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid] and sodium acetate}, i.e., 167 mM Mes, 167 mM Hepes, and 167 mM sodium acetate (pH 7.5) were used for these experiments. Desaturase preparations were made anaerobic by repeated cycles of vacuum and equilibration with oxygen-free argon with the use of a Schlenk line. The resulting desaturase solutions were reduced, as monitored by the decrease in absorption at 340 nm, by titration with sodium dithionite in the presence of 0.4 M CAT (pH 7.5) supplemented with 0.25 mM methyl viologen. The reduced protein was then applied to a PD10 column (Amersham Pharmacia, Uppsala, Sweden) equilibrated with 50 mM CAT/70 mM NaCl (pH 7.5) to eliminate excess sodium dithionite and methyl viologen. Protein eluting from this column was used for stopped-flow spectrometry. Reoxidation of desaturase by hydrogen peroxide was monitored by an increase in absorption of the 340-nm ligand-to-metal charge transfer band, with the use of a KinTek Stopped-Flow SF-2001. All measurements were made by using 0.1–5 mM H₂O₂ solutions in 50 mM CAT/70 mM NaCl (pH 7.5) at 10°C. Data analysis was performed by using IGOR Pro computer software (WaveMetrics, Lake Oswego, OR).

Desaturase–Substrate Complex.

Purified WT or T199D mutant desaturase (≈50 μM) was incubated with 18:0-ACP substrate in 20 mM Hepes/450 mM NaCl (pH 7.0) for 30 min. The elution times of either native enzyme or enzyme incubated with substrate were estimated after passage through a G3000SWXL size-exclusion column developed with the same buffer. Apparent molecular masses were estimated based on comparison of elution times of proteins of known masses.

Crystallization and Data Collection.

Crystallizations were performed by the hanging drop vapor diffusion method, using conditions very similar to those described previously for the native desaturase (19). Before crystallization, the protein was concentrated to 14–16 mg·ml⁻¹ in 20 mM Hepes (pH 7.0)/70 mM NaCl. Crystals were obtained at 20°C from a well solution of 0.08 M cacodylate buffer (pH 5.4), 0.2 M magnesium acetate, 75 mM ammonium sulfate, 16–18% (wt/vol) polyethylene glycol 4000, and 0.2% β-octyl glucoside, using a drop consisting of 5 μl of the well mixed with 5 μl of protein solution. Under these conditions crystals grew in 2–4 days, reaching a final size of ≈200 × 300 × 20 μm. Crystals were cryoprotected by soaking for ≈30 sec in well solution supplemented with 25% (vol/vol) glycerol. The crystals belong to the orthorhombic space group P2₁2₁2₁, with unit cell dimensions a = 82.05, b = 145.77, and c = 193.25 Å, and contain six monomers per asymmetric unit.

Data were collected at cryogenic temperatures by using beamline ID14-3 of the European Synchrotron Research Facility (Grenoble, France). The data were collected at a wavelength of 0.931 Å with an oscillation angle of 0.2° and were processed by using MOSFLM (20) and SCALA (21) from the CCP4 suite (22). Data collection and processing statistics are summarized in Table 3.

Structure Determination and Refinement.

The structure of the T199D mutant was solved by molecular replacement implemented in the program MOLREP (23) by using the original Δ9 desaturase structure (19) (Protein Data Bank ID code 1AFR) as the search model. The model obtained was refined by using a combination of simulated annealing with the use of the Crystallography and NMR System (CNS) software suite (24) and refinement by the maximum-likelihood method in REFMAC5 (15). Atomic displacement parameters were refined in REFMAC by the TLS (translation, liberation, screw) method, with each of the six monomers in the asymmetric unit treated as a single TLS group. Tight 6-fold NCS restraints were used throughout refinement to maximize the observation-to-parameter ratio. Graphics operations were performed in COOT (25), and water molecules were manually added to the model in COOT by using the 2F_o − F_c map. Annealed omit maps were calculated in CNS (24) and used to confirm the content of the active site, as well as the reduced state of the diiron center.

The geometry of the refined structure was checked with PROCHECK (26), and the refined parameters are summarized in Table 3. The coordinates of the final model, in addition to the structure factors, have been deposited in the Protein Data Bank with the ID code 2J2F. All structural figures were produced by using PyMol (27).