Crystal structure of Arabidopsis thaliana 12-oxophytodienoate reductase isoform 3 in complex with 8-iso prostaglandin A₁

Abstract

12-Oxophytodienoate reductase 3 (OPR3), one of the enzymes involved in the biosynthesis of the plant hormone jasmonic acid (JA), catalyzes the reduction of the cyclopentenone ring of (9S,13S)-12-oxophytodienoate [(9S,13S)-OPDA]. However, there has been no structural information about the interaction between OPRs and the physiologically relevant (9S,13S)-OPDA. Here we report the crystal structure of Arabidopsis thaliana OPR3 in complex with 8-iso prostaglandin A1 (8-iso PGA₁) which has the same stereochemistry in the cyclopentenone ring as in the physiologically relevant 9S,13S-OPDA. This structure reveals a new binding mode for substrate that likely contributes to the relaxed stereospecificity observed for AtOPR3.

INTRODUCTION

Jasmonic acid (JA) is a plant secondary messenger that contributes to coordinated responses to many biotic and abiotic stresses, as well as other diverse processes such as pollen maturation and wound responses.^1–2 Biosynthesis of JA originates from linolenic acid in chloroplast membranes, and a number of enzymes are involved in the conversion of this polyunsaturated fatty acid into cyclopentane containing compound JA.³ 12-oxophytodienoate reductase 3 (OPR3), one of the enzymes involved in the biosynthesis of JA, uniquely catalyzes the reduction of the cyclopentenone ring of (9S,13S)-12-oxophytodienoate [(9S,13S)-OPDA].⁴ The JA-deficient phenotype of OPR3 loss-of-function mutants in Arabidopsis thaliana indicates that other OPR isoforms cannot substitute for OPR3 in JA biosynthesis.⁵

The physiologically relevant (9S,13S)-enantiomer of OPDA is not a substrate for the majority of OPRs.⁴ Thus the unique function of OPR3 in the biosynthesis of JA is a result of its relaxed stereoselectivity with respect to OPDA stereoisomers. While Arabidopsis thaliana OPR3 (AtOPR3) and Lycopersicon esculentum OPR3 (LeOPR3) reduce all four stereoisomers of OPDA, the large majority of OPR isoforms seem to resemble LeOPR1 and AtOPR1/2 in their substrate specificity.^6–7 The biological roles of OPR1 and OPR2 have not been established, however, these enzymes efficiently reduce 9R,13R-OPDA but fail to react with the enantiomeric JA precursor 9S,13S-OPDA.^7–9

To date, the several ligand- and cofactor-bound structures of OPR1 and OPR3 have been reported.^7,10–13 However, there has been no structural information about the interaction between OPRs and the physiologically relevant (9S,13S)-OPDA. Here we report the crystal structure of AtOPR3 in complex with 8-iso prostaglandin A1 (8-iso PGA₁) which has the same α,β-unsaturated carbonyl group and stereochemistry in the cyclopentenone ring as in the physiologically relevant 9S,13S-OPDA. Our structure shows that hydrogen bonds and van der Waals’ interactions between AtOPR3 and 8-iso PGA₁ stabilize the substrate molecule for reduction of the alkenyl group within the cyclopentenone ring. The positioning of 8-iso PGA₁ reveals a new binding orientation for substrate in the active site that likely contributes to the relaxed stereospecificity observed for AtOPR3 relative to other OPR’s.

MATERIALS AND METHODS

Cloning, protein expression, purification, and crystallization

Cloning, expression, purification and crystallization of AtOPR3¹² were as previously described. The calculated molar absorptivity was 43320 M⁻¹ cm⁻¹. 8-iso PGA₁ was from Cayman Chemical (Ann Arbor, MI). HPLC grade methanol and NADPH were from Sigma-Aldrich (St. Louis, MO).

X-ray data collection, structure determination, and model evaluation

X-ray diffraction data were collected using synchrotron beam line 23-ID at the Advanced Photon Source at the Argonne National Laboratory GM/CA-CAT. The diffraction images were integrated and scaled using HKL2000.¹⁴ The structure of the 8-iso PGA₁ bound AtOPR3 was determined by molecular replacement using AtOPR3 structure in complex with FMN (PDB 1q45) as a template. Cross-rotational search followed by translational search was performed utilizing the program using MolRep from CCP4.¹⁵ Subsequent manual model building and refinement were carried out using Coot and Refmac from CCP4 program suite.¹⁵ The stereochemical quality of the refined model was assessed using MolProbity.¹⁶ Structure figures were generated using PyMOL (DeLano Scientific LLC).

Sedimentation equilibrium studies

Sedimentation equilibrium studies were performed with a Beckman Optima XL-A analytical ultracentrifuge in the Biophysics Instrumentation Facility at the University of Wisconsin-Madison. AtOPR1 was dissolved in 5 mM Tris-Hcl at pH 8, containing 50 mM NaCl, while AtOPR3 was dissolved in 5 mM MES-NaOH at pH 6, containing 50 mM NaCl. From a single stock of each protein, dilutions were made to obtain three samples with UV absorbance readings of 0.2 (4.6 μM), 0.5 (11.5 μM) and 0.8 (18.5 μM) at 280 nm. Solvent densities measured using an Anton Paar DMA5000 at 4°C were 1.00241 and 1.00257 g mL⁻¹, respectively. The partial specific volumes for AtOPR1 and AtOPR3 based on their sequences, neglecting contributions from flavin, were the same, 0.732 mL g⁻¹. For each protein, ~100 μL of each of the three dilutions were placed in 1.2 cm pathlength double-sector, charcoal-filled, epon centerpieces with the appropriate buffers as references. Samples were spun to equilibrium at various speeds at 4°C. Superimposed gradients recorded at 278 nm 2–3 h apart were taken to indicate equilibrium. Data were collected at 278 nm and in most cases at 461 nm and/or 378 nm. A high-speed depletion of protein material from the cells was used as a measure of non-sedimented baseline material. For a given protein, data from all speeds and/or wavelengths were fit to various models including a fixed contribution from the non-sedimented material.

Enzyme reaction

Reactions of AtOPR1, AtOPR2 and AtOPR3 with 8-iso PGA₁ and NADPH were monitored by optical spectrometry. A blank was obtained with buffer containing NADPH (150 μM), substrate was added, and the reaction was initiated by addition of enzyme (~1 μM). The oxidation of NADPH resulted in a decrease in absorbance around 340 nm, while reduction of the substrate caused a decrease in the absorption band at ~220 nm. Substrate and product were separated using a Shimadzu VP Series HPLC equipped with a photodiode array spectrophotometer and an ACE C18 column (100 × 4.6 mm, 3 micron particle size) using 70% methanol and 30% water (70:30 v/v) pumped isocratically at 1 mL/min. The substrate and product were monitored and quantified using UV absorption at 218 nm (8-iso PGA₁, ε₂₁₈ = 8200 M⁻¹ cm⁻¹) and 220 nm (reduced product, ε₂₂₀ = 1700 M⁻¹ cm⁻¹).

The masses of 8-iso PGA₁ and the reduced product were determined using a Shimadzu LCMS-2010 with an HPLC inlet, an electrospray ionizer, and a quadrupole analyzer at the Paul Bender Chemical Instrument Center at the Department of Chemistry, University of Wisconsin-Madison.

Data deposition

Atomic coordinates and structure factors have been deposited in the RCSB Protein Data Bank (accession code 2G5W).

RESULTS AND DISCUSSION

In order to study the molecular properties that underlie the distinct substrate preferences of OPR3, we analyzed the structure AtOPR3 in complex with 8-iso PGA₁ which has the same α,β-unsaturated carbonyl group and stereochemistry in the cyclopentenone ring as in the physiologically relevant (9S,13S)-OPDA (Fig. 1a and 1b). We soaked the AtOPR3 crystals in a crystallization solution containing 5 mM 8-iso PGA₁ and the crystal structure of 8-iso PGA₁ bound AtOPR3 was solved to 2.6 Å by molecular replacement using AtOPR3 structure as a phasing model (PDB ID: 1q45).¹² The final model includes two AtOPR3 monomers per asymmetric unit, and each AtOPR3 monomer contains one FMN and one 8-iso PGA₁ molecule. The R and R_free factors were 20.2% and 26.4%, respectively (Table I). The N-terminal (7 residues in chain A; 8 residues in chain B) and C-terminal regions (3 residues in chain A; 5 residues in chain B) of the protein, as well as the L6 loop region (Try285—Glu301 in chain A; Gly289—Asp299 in chain B) in the crystal have not been included in the refined model. The two chains in the asymmetric unit are highly similar to each other, with r.m.s. deviation of 0.25 Å. When compared with the ligand-free AtOPR3 structure,¹² the structure of 8-iso PGA₁ bound AtOPR3 has no significant conformational change (Fig. 1c). The previously suspected substrate binding loop region (residues Ile133 to Tyr150), which corresponds to the Lβ3 substrate-binding loop of OPR1, was found to neither move nor interact with 8-iso PGA₁. Likewise, comparison of the ligand-free and (9R,13R)-OPDA-bound LeOPR1¹⁰ showed no significant conformational changes.

Figure 1.

(A) Schematic representation of jasmonic acid biosynthesis.

(B) Structures of 8-iso-PGA1 (top) and (9S,13S)-OPDA (bottom).

(C) Aligned structures of FMN-bound AtOPR3 (colored in pink) and 8-iso-PGA1-FMN-AtOPR3 complex (colored in green). The bound PGA1 is shown in sticks.

Table I.

Statistics for data collection and model refinement

Data collectiona	AtOPR3 in complex with 8-iso PGA1
Space group	P212121
Cell dimensions
a, b, c (Å)	79.70, 86.61, 123.58
α, β, γ (°)	90, 90, 90
Resolution (Å)b	50.00–2.58 (2.67-2.58)
No. reflections	22750
Rmerge (%)c	15.7 (36.6)
I/σId	9.55 (3.42)
Completeness (%)	82 (50)
Redundancy	6.1 (3.9)
Refinement e,f
Resolution (Å)	48.83 - 2.58 (2.64-2.58)
No. reflections	22704 (927)
Rwork / Rfree	0.202 / 0.264 (0.290 / 0.430)
No. of atoms
Protein	5627g
Ligand/ion	110h
Water	108
Mean B-value (overall)	39.82
Ramachandran analysis
Most favored regions	273 (89.2%)
Additional allowed regions	33 (10.8%)
Disallowed regions	0 (0%)
R.m.s deviations
Bond lengths (Å)	0.006
Bond angles (°)	0.977

^aData collected at Sector 32 of the Advanced Photon Source.

^bNumbers in parentheses indicate the highest resolution shell of 20.

^cRmerge = Σ_h Σ_i |I(h)_i − <I(h)>| / Σ_h Σ_i I(h)_i, where I(h) is the observed intensity of relection h, and < I(h) > is the average intensity obtained from multiple measurements.

^dThe root-mean-squared value of the intensity measurements divided by their estimated standard deviation.

^eR_work = Σ | | F_o | − | F_c | | / Σ | F_o |, where | F_o | is the observed structure factor amplitude and | F_c | is the calculated structure factor amplitude.

^fR_free = R-factor based on 4.96% of the data excluded from refinement.

^gNumber of non-hydrogen protein atoms included in refinement.

^hThese include FMN and 8-iso PGA₁.

The cyclopentenone ring of 8-iso PGA₁, which is structurally comparable to the corresponding cyclopentenone ring of OPDA, forms stacking interactions with the isoalloxazine ring of the FMN (Fig. 2a), and its carbonyl oxygen is in close contact with the two catalytic histidine residues His186 and His189 (Fig. 2b). The double bond formed by C10 and C11 carbon atoms of the cyclopentenone ring of the bound 8-iso-PGA₁ is located closely to both hydride (FMN-N5) and proton (Tyr191-OH) donors, suggesting that the cyclopentenone ring in 8-iso-PGA₁ has been effectively oriented for redox chemistry in this enzyme-substrate complex (Fig. 2a). While the gatekeeper residues Tyr78 and Tyr246 of AtOPR1 form a narrow entrance into the cavity above the FMN, the corresponding amino acids of AtOPR3, Phe74 and His245, turn away from the active site, giving rise to a widening of the cavity in OPR3 (Fig. 2b). The importance of this position in stereoselectivity is also supported by substitution of the corresponding phenylalanine and histidine residues in LeOPR3 with tyrosine residues, which shifted the specificity of the mutated enzymes toward (9R,13R)-OPDA over (9S,13S)-OPDA.¹³ These geometrical differences most likely account for the high degree of substrate stereoselectivity of OPR1 and the more relaxed specificity of OPR3 as proposed earlier.¹³

Figure 2.

(A) Orientation of the bound 8-iso-PGA1 relative to the FMN and the general acid Tyr191. The Fo − Fc electron density of the bound 8-iso-PGA1 is contoured at 2.5 σ. The interactions of the double bond-forming carbon atoms of 8-iso-PGA1 with N5 nitrogen atom of FMN and the general acid Tyr191 are indicated by dotted lines.

(B) Stereo view of the substrate binding site of the 8-iso-PGA1-FMN-AtOPR3 complex structure (green) overlaid with the structure of (9R,13R)-OPDA-FMN-LeOPR1 complex (light green) after structural alignment. Residues are numbered according to the AtOPR3 sequence. The FMN molecules of the enzymes superimpose perfectly with each other and are left out for clarity.

(C) Enzymatic assay results of three A. thaliana OPR isoforms using 8-iso-PGA1 as a substrate. Time-dependent change in A340 observed from AtOPR1, AtOPR2, and AtOPR3 are indicated by circles, rectangles, and triangle, respectively.

(D) Sedimentation equilibrium data for three concentrations of AtOPR3 (blue, 4,6 μM; green, 11.5 μM; red, 18.5 μM) and 5 speeds (4.2k, 6.6k, 8.6k, 12k, 17.5k) detected at 278 nm. Global curve fitting was consistent with the presence of a single species at all concentrations observed. The average molecular weight from the global fit of data to the 278 nm data was 44,101 ± 41 Da for AtOPR3. These results are in good agreement with the molecular weight from sequence of 42,694 Da. Data obtained at other wavelengths (378 nm or 461 nm, corresponding to FAD absorption peaks) were consistent with this analysis and their inclusion in global fits minimally impacted the calculated weight average molecular weights.

Compared to (9R,13R)-OPDA bound to LeOPR1,¹⁰ the binding mode of the cyclopentenone ring of 8-iso-PGA₁ differs in the distance of the double bond and its angle relative to FMN-N5 (Fig. 2a). The small differences in the positions of the cyclopentenone rings propagate into much larger differences in the positions of the alkyl chains as they emerge from the interior of the active site (Fig. 2b). These discrepancies imply that contacts of the alkyl chains with distal residues may have important contributions to the differences in substrate specificity observed for OPR isoforms. As a result, the 8S tail of 8-iso-PGA₁ has van der Waals interactions with the side chains of His189 and Ile243 (Fig. 2b). These interactions potentially mimic hydrophobic interactions with the 13S pentenyl side chain of the natural substrate. The methyl end of the 12S tail of the bound 8-iso-PGA₁ locates within 4 Å of the aliphatic part of the Arg367 side chain and within 6 Å of the guanidinium group. Since the methyl group of the 12S tail of 8-iso-PGA₁ is replaced by a carboxyl group in the natural substrate (9S,13S)-OPDA, (Fig. 2b), a hydrogen bonding interaction between the natural substrate and the guanidinium group of Arg367 may stabilize the natural substrate-enzyme complex. In addition, the strong hydrogen bond between the PGA1 carboxyl group and the main chain nitrogen of Arg284 that might pull the substrate more out of the cavity than (9S,13S)-OPDA and that leads to very weak hydrogen bonds of the carbonyl oxygen to the two activating histidines.

In order to confirm that 8-iso-PGA₁ was an effective substrate for AtOPR3, we performed an enzymatic assay using 8-iso-PGA₁ as a substrate (Fig. 2c). The result indicated that 8-iso-PGA₁ was an active substrate for AtOPR3 with apparent k_cat and K_M values of 0.23 s⁻¹ and 22 μM, respectively. Of the three Arabidopsis OPR isoforms tested, only AtOPR3 reduced 8-iso-PGA₁, while AtOPR1 and AtOPR2 did not (Fig. 2c). This result is consistent with the stereochemical preferences previously established for the JA pathway and with the known reactions of the AtOPR isoforms with the different OPDA stereoisomers.¹⁷

While most OPRs have been described as monomers, LeOPR crystallizes as a self-inhibited dimer, in which the substrate binding pocket is occupied by loop L6 of the neighboring protomer.^7,13 In contrast to the self-inhibited dimeric structure of LeOPR3, the structure of substrate-free and 8-iso-PGA₁-bound structures of OPR3 showed that the protein exists as a monomer in the crystal. Incorporation of 8-iso-PGA₁ into the active site was not affected by crystal packing while soaking the OPDA substrate into LeOPR3 crystals or cocrystallization trials were unsuccessful. Furthermore, sedimentation equilibrium measurements on AtOPR1 and AtOPR3 showed that both proteins were monomers at concentrations up to ~30 μM, at which LeOPR3 forms a self-inhibited dimer (Fig. 2d).¹³

In the crystal structure of LeOPR3, the sulfate is located at the dimer interface, close to Tyr364, and is positioned perfectly to mimic a phosphorylated tyrosine residue.⁷ The bound sulfate ion directly interacts with the protruding loop L6 from the protomer, stabilizing the dimer formation of the protein. Thus, it has been suggested that dimerization and hence OPR3 activity may be regulated by the reversible phosphorylation of Tyr364 in vivo.⁷ In the structures of substrate-free and 8-iso-PGA₁ bound AtOPR3 the two monomers observed in the asymmetric unit of AtOPR3 crystals can be superimposed with the corresponding protomers of the LeOPR3 dimer, suggesting a similar dimeric structure for AtOPR3. Although the L6 loop of AtOPR3 is poorly defined in the refined structure, it clearly protrudes toward the active site of the molecular neighbor in a similar fashion as in LeOPR3. Considering that the amino acid sequence of the loop L6 is highly conserved among the OPR3-like enzymes, it is still intriguing that OPR3 activity is regulated by self-inhibitory dimerization, which might be controlled by reversible phosphorylation of the protein, although it awaits further characterizations to confirm this hypothesis.

Abbreviations

OPR3

12-oxophytodienoate reductase isoform 3

PGA₁

prostaglandin A1

JA

jasmonic acid