Crystallization and preliminary X-ray diffraction analysis of a flavoenzyme amine de­hydrogenase/oxidase from Pyrococcus furiosus DSM 3638

A flavoprotein amine dehydrogenase/oxidase with subunit molecular weights of 54.8 kDa (α-subunit) and 42.4 kDa (β-subunit) and specificity for L-proline was cloned from the genomic DNA of the hyperthermophilic marine archaeon Pyrococcus furiosus DSM 3638. The enzyme was overexpressed in Escherichia coli and purified to homogeneity. The enzyme was crystallized using the sitting-drop vapour-diffusion technique. Diffraction data from two different crystal forms were collected to 3.3 and 3.6 Å, respectively, using synchrotron radiation. Both crystals belonged to space group P1, with unit-cell parameters a = 91.3, b = 136.3, c = 203.8 Å, α = 94.5, β = 99.4, γ = 102.7° and a = 93.7, b = 116.3, c = 126.9 Å, α = 97.3, β = 99.9, γ = 104.6°.

1. Introduction

The mechanisms of amine oxidation catalysed by the quinoprotein amine dehydrogenase/oxidases are well established. Oxidation occurs through the formation of enzyme–substrate covalent adducts with topaquinone (TPQ), tryptophan tryptophylquinone (TTQ), cysteine tryptophylquinone (CTQ) and lysine tyrosyl quinone (LTQ) redox centres (Mure et al., 2002; Datta et al., 2001; Satoh et al., 2002; Davidson, 2000; Wang et al., 1996) and hydrogen transfer by quantum-mechanical tunnelling has been demonstrated for some quinoprotein enzymes (Masgrau et al., 2004). The mechanism of amine oxidation by flavoproteins is less well understood (Scrutton, 2004). These enzymes catalyse the two-electron oxidation of amine substrates to the corresponding iminium ion. Mechanisms involving (i) proton abstraction by an active-site base to generate a carbanion species (Rohlfs & Hille, 1994), (ii) an aminium radical cation species (Silverman, 1995), (iii) H-atom abstraction by an active-site radical species (Edmondson, 1995) and (iv) nucleophilic attack by the substrate nitrogen on the flavin C4a atom followed by proton abstraction by an active-site base or the flavin N5 atom (Kim et al., 1993) have been considered over the years, but controversy remains in the field. Further structural elucidation of flavoprotein amine dehydrogenase/oxidases coupled with detailed spectroscopic and kinetic studies are required to unravel further the mechanistic details of amine oxidation by this class of enzyme.

We have identified a putative flavoprotein amine dehydrogenase/oxidase from Pyrococcus furiosus DSM 3638. The β-subunit of this putative amine oxidase is related to the active-site region of di­methylglycine oxidase (DMGO) of Arthrobacter globiformis, for which a crystal structure has been elucidated and a mechanism of substrate oxidation proposed (Leys et al., 2003). The P. furiosus enzyme oxidizes sarcosine, L-pipecolic acid and L-proline, but is not reactive with dimethyglycine, glycine or betaine. Kinetic studies suggest that L-proline is the natural substrate, but the enzyme is not related in sequence to the structurally characterized and bifunctional proline dehydrogenase of Escherichia coli (Lee et al., 2003). The flavoprotein amine dehydrogenase/oxidase from P. furiosus appears to represent a new class of proline-oxidizing flavoprotein. To further our understanding of the mechanism and structure of flavoprotein amine dehydrogenase/oxidases, we have initiated a structural study of this novel flavoprotein amine dehydrogenase/oxidase from P. furiosus. We have expressed the recombinant enzyme in E. coli and purified the protein to homogeneity. Diffraction data are reported to 3.3 and 3.6 Å for two crystal forms both in space group P1.

2. Materials and methods

2.1. Cloning, overexpression and purification

Two open reading frames [gi|18977617 and gi|18977618; protein-extraction description and analysis tool (PEDANT) database] were identified in the genome of P. furious DSM 3638 that encode a putative flavoenzyme amine dehydrogenase/oxidase and were amplified by the polymerase chain reaction. Both genes encoding the α- and β-subunits (492 and 382 amino-acid residues, respectively) were cloned into pET11d (Novagen) and the resultant construct transformed into E. coli strain Rosetta(DE3)pLysS for recombinant expression. Details of plasmid construction and propagation will be reported elsewhere. Selected transformed cells were grown in 12 l of 2×YT medium containing chloramphenicol (34 µg ml⁻¹) and ampicillin (50 µg ml⁻¹) at 310 K to an OD₆₀₀ ≃ 0.8 and were induced by adding isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 1 mM. Incubation took place for a further 8 h, after which cells were harvested and resuspended in 50 mM ice-cold potassium phosphate buffer pH 8.0 containing a Complete protease-inhibitor cocktail tablet (Roche). The cell suspension was incubated with lysozyme (10 µg ml⁻¹) and sonicated to effect cell breakage. Deoxyribonuclease I (∼10 µg) was added and the suspension was incubated for a further 30 min at 277 K. Cellular debris was removed by centrifugation and the supernatant was incubated at 353 K for 1 h followed by centrifugation to remove denatured host-cell proteins. The supernatant was dialysed exhaustively against 50 mM Tris–HCl pH 8.0 at 277 K, loaded onto a Q-Sepharose column (75 ml bed volume) and eluted using a 0.2–0.4 M gradient of NaCl. Fractions containing recombinant protein were pooled, concentrated and applied onto a Superdex 75 column (330 ml bed volume); protein was eluted at a flow rate of 0.75 ml min⁻¹. Purified enzyme was exchanged into 10 mM Tris–HCl buffer pH 7.5 containing 100 mM KCl, concentrated to 12.2 mg ml⁻¹, sterilized (0.22 µm Millex-GP Acrodisc filter, Millipore) and stored on ice at 277 K. Protein concentration was determined by the method of Bradford (1976). Purity was >95% as judged by SDS–PAGE.

2.2. Crystallization

Crystallization trials were performed using the sitting-drop vapour-diffusion technique. Drops were prepared by mixing 2 µl of 6.1 mg ml⁻¹ protein solution with 2 µl reservoir solution. After 3–4 d incubation at 292 K, yellow crystals were observed in two conditions containing 0.2 M potassium thiocyanate, 8% PEG 20K, 8% PEG 550 MME pH 7.5, 100 mM Tris (crystal form 1; Fig. 1) and 0.2 M sodium formate, 8% PEG 20K and 8% PEG 550 MME pH 8.5, 100 mM Tris (crystal form 2; Fig. 2) in the reservoir solution. The average dimensions of these crystals were 40 × 250 × 250 µm (form 1) and 50 × 50 × 200 µm (form 2).

Figure 1 Crystals (form 1) of the recombinant flavoprotein amine oxidase/dehydrogenase from P. furiosus.

Figure 2 Crystals (form 2) of the recombinant flavoprotein amine oxidase/dehydrogenase from P. furiosus.

Prior to data collection at 100 K, selected crystals were soaked in mother liquor with 10% PEG 200 as cryoprotectant and flash-cooled in liquid nitrogen.

2.3. X-ray diffraction analysis

Diffraction data were collected from cryocooled crystals (100 K) using a Quantum ADSC (Area Dectector Systems Corporation) CCD detector on beamline ID 14-3 at the European Synchrotron Radiation Facility, Grenoble, France. Data were collected using 1° oscillations with the crystal-to-detector distance set to 230 mm. All data were indexed, integrated and scaled using the programs DENZO and SCALEPACK from the HKL software package (Otwinowski & Minor, 1997). Crystals of form 1 diffracted to a resolution of 3.6 Å and belonged to space group P1, with unit-cell parameters a = 93.7, b = 116.3, c = 126.9 Å, α = 97.3, β = 99.9, γ = 104.6°. Crystals of form 2 diffracted to a resolution of 3.3 Å and belonged to space group P1, with unit-cell parameters a = 91.3, b = 136.3, c = 203.8 Å, α = 94.5, β = 99.4, γ = 102.7°. Diffraction data-collection statistics are given in Table 1. Using the program MOLREP (Vagin, 1997), a general self-rotation function was computed for a number of κ angles (60, 90, 120 and 180°) to test for the presence of non-crystallographic twofold, threefold, fourfold and sixfold axes. A self-rotation function of the form 1 data set indicates the presence of three perpendicular twofold axes in the unit cell, which strongly suggests the protein is a heterooctamer, with heterodimers arranged in 222 non-crystallographic symmetry (Table 2, Fig. 3). Four heterodimers (one heterooctamer) per unit cell, totalling 388.8 kDa, corresponds to a Matthews coefficient of 3.34 ų Da⁻¹ and a solvent content of ∼62.8% (Matthews, 1968). Self-rotation functions of data from form 2 did not reveal any clear indication of the presence of non-crystallographic symmetry. However, two heterooctamers per unit cell, totalling 777.6 kDa, corresponds to a Matthews coefficient of 3.12 ų Da⁻¹ and a solvent content of ∼60.2%.

Table 1
Data-collection and processing statistics for the form 1 and form 2 crystals of recombinant P. furiosus flavoprotein amine oxidase/dehydrogenase

Crystal form	1	2
Synchrotron radiation	Beamline ID 14-3, ESRF	Beamline ID 14-3, ESRF
Space group	P1	P1
Wavelength (Å)	0.931	0.931
Unit-cell parameters
a (Å)	93.7	91.3
b (Å)	116.3	136.3
c (Å)	126.9	203.8
α (°)	97.3	94.5
β (°)	99.9	99.4
γ (°)	104.6	102.7
Matthews coefficient (Å^3 Da^−1)	3.34	3.12
Resolution (Å)	30–3.6 (3.8–3.6)	30–3.3 (3.4–3.3)
Rmerge (%)	0.120 (0.432)	0.084 (0.357)
Completeness (%)	96.4 (95.0)	96.0 (93.0)
Average I/σ(I)	6.2 (1.8)	8.2 (2.0)
Average redundancy	2.5 (2.3)	1.9 (1.8)
Unique reflections	55367	132489

Table 2
Peaks observed in the self-rotation analyis of the form 1 crystals

The polar angles θ, φ and χ are defined as follows: θ is the angle between the rotation axis and z, φ is the angle in the xy plane between x and the projection of the rotation axis and χ is the rotation angle around the rotation axis.

Peak	θ (°)	φ (°)	χ (°)	Rf/σ
1	86.6	−30.9	180	10.6
2	127.9	57.1	180	10.5
3	38.3	64.3	180	9.7

Figure 3 Self-rotation function calculated using crystal form 1 data between 30 and 3.6 Å with a 50.0 Å radius of integration. Twofold symmetry is evident from the very strong peaks in the χ = 180° plot.

Molecular-replacement trials performed with the available structures of similar flavin oxidases as the starting model failed to identify a solution. Therefore, MIR (multiple isomorphous replacement)/MAD (multiwavelength anomalous diffraction) experiments are planned to provide sufficient experimental phasing information. A search for heavy-atom derivatives and the production of selenomethionine-substituted protein are in progress.