Crystal Structure of a Novel FAD-, FMN-, and ATP-containing l-Proline Dehydrogenase Complex from Pyrococcus horikoshii

Abstract

Two novel types of dye-linked l-proline dehydrogenase complex (PDH1 and PDH2) were found in a hyperthermophilic archaeon, Pyrococcus horikoshii OT3. Here we report the first crystal structure of PDH1, which is a heterooctameric complex (αβ)4 containing three different cofactors: FAD, FMN, and ATP. The structure was determined by x-ray crystallography to a resolution of 2.86 Å. The structure of the β subunit, which is an l-proline dehydrogenase catalytic component containing FAD as a cofactor, was similar to that of monomeric sarcosine oxidase. On the other hand, the α subunit possessed a unique structure composed of a classical dinucleotide fold domain with ATP, a central domain, an N-terminal domain, and a Cys-clustered domain. Serving as a third cofactor, FMN was located at the interface between the α and β subunits in a novel configuration. The observed structure suggests that FAD and FMN are incorporated into an electron transfer system, with electrons passing from the former to the latter. The function of ATP is unknown, but it may play a regulatory role. Although the structure of the α subunit differs from that of the β subunit, except for the presence of an analogous dinucleotide domain with a different cofactor, the structural characteristics of PDH1 suggest that each represents a divergent enzyme that arose from a common ancestral flavoenzyme and that they eventually formed a complex to gain a new function. The structural characteristics described here reveal the PDH1 complex to be a unique diflavin dehydrogenase containing a novel electron transfer system. Two novel types of dye-linked l-proline dehydrogenase complex (PDH1 and PDH2) were found in a hyperthermophilic archaeon, Pyrococcus horikoshii OT3. Here we report the first crystal structure of PDH1, which is a heterooctameric complex (αβ)4 containing three different cofactors: FAD, FMN, and ATP. The structure was determined by x-ray crystallography to a resolution of 2.86 Å. The structure of the β subunit, which is an l-proline dehydrogenase catalytic component containing FAD as a cofactor, was similar to that of monomeric sarcosine oxidase. On the other hand, the α subunit possessed a unique structure composed of a classical dinucleotide fold domain with ATP, a central domain, an N-terminal domain, and a Cys-clustered domain. Serving as a third cofactor, FMN was located at the interface between the α and β subunits in a novel configuration. The observed structure suggests that FAD and FMN are incorporated into an electron transfer system, with electrons passing from the former to the latter. The function of ATP is unknown, but it may play a regulatory role. Although the structure of the α subunit differs from that of the β subunit, except for the presence of an analogous dinucleotide domain with a different cofactor, the structural characteristics of PDH1 suggest that each represents a divergent enzyme that arose from a common ancestral flavoenzyme and that they eventually formed a complex to gain a new function. The structural characteristics described here reveal the PDH1 complex to be a unique diflavin dehydrogenase containing a novel electron transfer system. Dye-linked dehydrogenases catalyze the oxidation of various amino acids, organic acids, amines, and alcohols in the presence of an artificial electron acceptor such as ferricyanide or 2,6-dichloroindophenol (DCIP). 1The abbreviations used are: DCIP, 2,6-dichloroindophenol; CAPS, N-cyclohexyl-3-aminopropanesulfonic acid; CPR, NADPH-cytochrome P450 reductase; GR, glutathione reductase; HPLC, high performance liquid chromatography; MSOX, monomeric sarcosine oxidase; OABA, o-aminobenzaldehyde; P5C, Δ1-pyroline-5-carboxylate; PDH, l-proline dehydrogenase; r.m.s.d., root mean square deviation; TD, trimethylamine dehydrogenase; TSOX, heterotetrameric sarcosine oxidase. We have identified several novel dye-linked dehydrogenases among the hyperthermophilic archaea, including d-proline dehydrogenase (1Satomura T. Kawakami R. Sakuraba H. Ohshima T. J. Biol. Chem. 2002; 277: 12861-12867Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar) and l-proline dehydrogenase (PDH) (2Sakuraba H. Takamatsu Y. Satomura T. Kawakami R. Ohshima T. Appl. Environ. Microbiol. 2001; 67: 1470-1475Crossref PubMed Scopus (38) Google Scholar, 3Kawakami R. Sakuraba H. Ohshima T. Extremophiles. 2004; 8: 99-108Crossref PubMed Scopus (27) Google Scholar), which catalyzes the dehydrogenation from l-prroline to Δ 1The abbreviations used are: DCIP, 2,6-dichloroindophenol; CAPS, N-cyclohexyl-3-aminopropanesulfonic acid; CPR, NADPH-cytochrome P450 reductase; GR, glutathione reductase; HPLC, high performance liquid chromatography; MSOX, monomeric sarcosine oxidase; OABA, o-aminobenzaldehyde; P5C, Δ1-pyroline-5-carboxylate; PDH, l-proline dehydrogenase; r.m.s.d., root mean square deviation; TD, trimethylamine dehydrogenase; TSOX, heterotetrameric sarcosine oxidase.-pyrroline-5-carboxylate (P5C). The gene sequence and primary structure of the first PDH isolated from a hyperthermophilic archaeon, Thermococcus profundus DSM9503, have been determined (2Sakuraba H. Takamatsu Y. Satomura T. Kawakami R. Ohshima T. Appl. Environ. Microbiol. 2001; 67: 1470-1475Crossref PubMed Scopus (38) Google Scholar, 3Kawakami R. Sakuraba H. Ohshima T. Extremophiles. 2004; 8: 99-108Crossref PubMed Scopus (27) Google Scholar). The complete gene is formed by an operon comprised of four genes, pdhA, pdhB, pdhF and pdhX, in a tandem arrangement in the order pdhA-F-X-B. The purified enzyme has a native molecular mass of 120 kDa and forms an unusual αβγδ heterotetrameric complex (α (54 kDa), β (43 kDa), γ (19 kDa), and δ (11 kDa)) encoded by pdhA, pdhB, pdhF, and pdhX, respectively. In addition, functional analysis of each subunit has shown that the α and β subunits possess NADH and l-proline dehydrogenase activities, respectively (3Kawakami R. Sakuraba H. Ohshima T. Extremophiles. 2004; 8: 99-108Crossref PubMed Scopus (27) Google Scholar). Because this multifunctional dye-linked dehydrogenase complex appears to representative of a new enzyme group (3Kawakami R. Sakuraba H. Ohshima T. Extremophiles. 2004; 8: 99-108Crossref PubMed Scopus (27) Google Scholar), we have been seeking others by screening dye-linked dehydrogenases expressed in hyperthermophilic archaea. So far, we have identified two different types of dye-linked PDH in the hyperthermophilic archaeon Pyrococcus horikoshii OT-3: PDH1 and PDH2 (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). Biochemical and gene analyses of PDH1 and PDH2 revealed PDH2 to be similar to the T. profundus PDH. By contrast, the gene encoding PDH1 formed an operon comprised of two consecutive genes, PH1363 and PH1364, and the molecular masses of the encoded α and β subunits were determined to be about 56 and 43 kDa, respectively. The native molecular mass of PDH1 was determined to be 440 kDa by gel filtration study (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar), indicating that the enzyme exists as an (αβ)4 heterooctamer with a subunit structure that is different from that of PDH2. In addition, high performance liquid chromatography (HPLC) analysis recently showed that PDH1 contains three cofactors: FAD, FMN, and ATP (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). The β subunit contains FAD and exhibits proline dehydrogenase activity, while the α subunit contains ATP but exhibits no proline dehydrogenase activity (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). Although both PDH1 and PDH2 genes are annotated as sarcosine oxidase in the P. horikoshii genome, because of their high homology with a monomeric sarcosine oxidase (MSOX), neither of these enzymes exhibits any sarcosine oxidase activity. The amino acid sequences of the α and β subunits of PDH1 showed 31 and 56% similarity with those of the α and β subunits of PDH2, respectively. Analysis of the genome database revealed that both types of PDH complex are widely distributed among the Thermococcales family of hyperthermophilic archaea, including the species Pyrococcus and Thermococcus. PDH has also been identified in Escherichia coli and Salmonella typhimurium, and some of the fundamental properties of those enzymes have been investigated (5Menzel R. Roth J. J. Biol. Chem. 1981; 256: 9762-9766Abstract Full Text PDF PubMed Google Scholar, 6Menzel R. Roth J. J. Biol. Chem. 1981; 256: 9755-9761Abstract Full Text PDF PubMed Google Scholar, 7Graham S.B. Stephenson J.T. Wood J.M. J. Biol. Chem. 1984; 259: 2656-2661Abstract Full Text PDF PubMed Google Scholar, 8Maloy S.R. Roth J.R. J. Bacteriol. 1983; 154: 561-568Crossref PubMed Google Scholar), and the crystal structure of E. coli PutA has been resolved (9Lee Y.H. Nadaraia S. Gu D. Becker D.F. Tanner J.J. Nat. Struct. Biol. 2003; 10: 109-114Crossref PubMed Scopus (98) Google Scholar). Notably, the primary structure and molecular properties of PDH1 and PDH2 are different from those of PutA. Still, dye-linked dehydrogenases generally have been found to function in the uptake of electrons from a reduced substrate into an electron transfer system. Little is currently known about the physiological function of these enzymes in hyperthermophilic archaea nor is much known about their mechanism catalysis or their electron transfer systems. This makes structural analysis of the archaeal PDH complex particularly interesting, as it could shed new light on the evolution of energy and amino acid metabolism in hyperthermophilic archaea such as P. horikoshii and T. profundus. Here we describe the unique structure of the PDH1 complex, which is a novel type of archaeal enzyme containing FAD, FMN, and ATP. Protein Expression and Purification—PH1363 and PH1364 genes, which, respectively, encode the α and β subunits of PDH1, were amplified and inserted separately into the pET11a expression plasmid. To construct the co-expression plasmid pEPDH1, the region containing PH1363 gene and its T7 promoter was cleaved from pET11a/PH1363 and inserted into pET11a/PH1364 upstream of the region containing the T7 promoter. E. coli strain BL21 CodonPlus (DE3) cells transformed with pEPDH1 were grown at 37 °C in SB medium containing ampicillin. After 6 h, isopropyl β-d-thiogalactopyranoside (1 mm) was added, and cultivation was continued for 3 h at 37 °C. The cells were then collected by centrifugation (10,000 × g, 20 min) and suspended in 10 mm potassium phosphate buffer (pH 7.0). Crude extract was prepared by ultrasonication, and then three purification steps were applied: heat treatment at 90 °C for 10 min, hydrophobic chromatography on Butyl Toyopearl, and gel filtration chromatography on Superdex 200. Active fractions containing (αβ)4 complex were pooled and dialyzed against 10 mm potassium phosphate (pH 7.0). Enzyme Assay—Enzyme activity was spectrophotometrically assayed as described previously using a Shimadzu UV-1200 spectrophotometer (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). It was also assayed by determining levels of the l-proline dehydrogenation product P5C coupled with o-aminobenzaldehyde (OABA). In this case, the reaction mixture was composed of 200 mm Tris/HCl (pH 7.5), 50 mm l-proline, 2.5 mm OABA, and enzyme. The amount of P5C coupled with OABA was monitored at 440 nm at 50 °C, and a millimolar absorption coefficient of 2.7 mm-1 cm-1 was used. The native enzyme had specific activities of 1.09, 2.01, and 0.324 (μmol/min/mg) using DCIP, ferricyanide, and oxygen as electron acceptors, respectively. Crystallization—After concentrating the enzyme to 10 mg/ml, the crystallization conditions were screened using Wizard 1, 2 and Cryo 1, 2 (Emerald Biostructure). A drop containing 2 μl of enzyme solution and 2 μl of reservoir solution was equilibrated against 200 μl of reservoir solution (0.1 m N-cyclohexyl-3-aminopropanesulfonic acid (CAPS)/NaOH (pH 10.5), 1.2 m NaH2PO4/0.8 M K2HPO4 and 0.2 m Li2SO4) by vapor diffusion using the sitting drop method. Crystals were obtained within 3-5 days at room temperature. The crystals were then placed in mother liquor containing 40% trehalose, briefly soaked, and flash-frozen in liquid nitrogen. Heavy atom derivatives were made by soaking the crystals in mother liquor containing a heavy atom compound (1 mm) for 24 h at 20 °C. PDH1 crystallizes in the hexagonal space group P6422 or its enantiomorph P6222 with lattice constants a = b = 172.22 Å and c = 175.39 Å, and it diffracts to 2.8 Å resolution. The asymmetric unit consists of one αβ heterodimer with a solvent content of 67%. Phase Determination and Model Building—Phase was determined using the multiple isomorphous replacement with anomalous scattering method by collecting two native data sets and four derivative data sets (K2Pt(SCN)6, (NH3)2PtCl4, phenyl mercury acetate, and mersalyl acid) at KEK beamline NW12 using monochromatized radiation at λ = 1.0 Å and an Area Detector Systems Corp. CCD detector (supplemental Table 1). The crystal-to-detector distance was 290 mm, and the scan angle was 1.0°. Data processing was carried our using HKL2000 (10Otwinowski Z. Minor W. Methods Enzymol. 1996; 276: 307-326Crossref Scopus (38572) Google Scholar), and phase calculations were carried out using SOLVE and RESOLVE (11Terwilliger T. J. Synchrotron Radiat. 2004; 11: 49-52Crossref PubMed Scopus (361) Google Scholar). When the phase was calculated using the P6422 space group, the quality of the electron density map was excellent, even at 2.86 Å resolution. The model was built manually using XTALVIEW (12McRee D.E. J. Struct. Biol. 1999; 125: 156-165Crossref PubMed Scopus (2022) Google Scholar). Model Refinement and Quality of the Model—The initial R-factor for a model consisting of 860 residues (α1-94 and α101-490 and β2-375) without cofactor was 41% in the resolution range 2.86-20.0 Å. The model was refined using REFMAC5 (13Potterton E. Briggs P. Turkenburg M. Dodson E. Acta Crystallogr. Sect. D Biol. Crystallogr. 2003; 59: 1131-1137Crossref PubMed Scopus (1070) Google Scholar). Several cycles of inspection of 2Fo - Fc and Fo - Fc Fourier maps calculated with 2.86-20.0 Å data were carried out, which the model was FAD, FMN, ATP, CAPS, and were in that in the α subunit had electron density and formed a to the β and using P. J. M. T. Acta Crystallogr. Sect. D Biol. Crystallogr. PubMed Scopus Google Scholar) and of and respectively. In both the α and β were no residues in the region of the were using of the of the (αβ)4 structure of PDH1 as a of the αβ heterodimer had a α and one β subunit of PDH1 were in an asymmetric a heterodimer 1, A and The (αβ)4 heterooctamer had of × × 90 Å. of the complex was the presence of a formed between two α The structure that one by the with from the to the The of the αβ heterodimer was while that of the (αβ)4 heterooctamer was a The structure with interface the of PDH1, which showed that no activity was at 90 °C for 120 (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). It that one of FAD is located in the of the β subunit, one of ATP is located in the α subunit, and one of FMN is located at the interface between the α and β subunits The of FAD and FMN were in within the as the distance between the two was Å The of the two of FAD and FMN were within about Å of one and the of the two was The between the α and β FMN was found between the α and β The of the subunits this FMN appears for the function the of the between the FMN and the be The other between the subunits were is in a β subunit hydrophobic formed by and is within the β subunit hydrophobic residues and and a hydrophobic is formed by and In addition, the were and of the β β subunit consists of two (supplemental and is an domain and containing the fold observed in enzymes to the glutathione family D. Protein 2001; 10: PubMed Scopus Google Scholar). The fold is composed of a central β and by two α and on one and a β and on the The other domain is an interface domain formed by an β two and using Biochem. Full Text PDF PubMed Scopus Google Scholar) showed that the β subunit has similarity with Å for sequence oxidase Å for sequence oxidase Å for sequence and acid oxidase Å for sequence is known to to the structural of flavoenzyme found in and contains FAD as the cofactor P. Struct. 1999; Full Text Full Text PDF Scopus Google Scholar). FAD in the β subunit is in a similar to that in MSOX, we found several FAD to it with in but one of the residues on the of the of FAD from those in residues are for of a new electron from FAD to FMN, which is located at the interface between the α and β subunits is no FMN in on the is between and The of makes with the of FAD, while makes a with the of By contrast, residues on the of the of FAD, which a as are and are in both and a unique is between the interface domain of the β subunit and the Cys-clustered domain of the α subunit and of the α of the J. Biol. Chem. Full Text PDF PubMed Google Scholar) family data that the α subunit of PDH1 has a dinucleotide domain, the structure is The crystal structure revealed that the α subunit consists of four (supplemental 2 and which an domain containing residues and with a a central domain containing an N-terminal domain containing and a Cys-clustered domain containing The domain, which contains a similar to the one in the β subunit, consists of a β and on one by two and and by a β and on the Analysis using showed this domain to be similar to the domain of an to the family Å for sequence Biochem. Full Text PDF PubMed Scopus Google Scholar). The central domain consists of a β and on one by two and and by a β and on the The has some similarity with the domain of GR, but the domain of the to The structure of is unique because the N-terminal domain and the Cys-clustered domain were the former domain has an and the domain consists of four α and FMN to the α subunit at the interface with the β HPLC, we the presence of ATP in the α subunit of PDH1 (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). The of the electron density of the enzyme the of an ATP in the α subunit (supplemental it as a third cofactor with FAD and FMN This is as it is the first report of ATP as a cofactor in a ATP is within the α subunit in the domain known as the residues in with the of the ATP, as is observed in and and of the ATP with the group of the of In of the of a cofactor, to trimethylamine dehydrogenase J. Biol. Chem. Full Text PDF PubMed Google Scholar), the physiological function of the is for in the in GR, TD, and we found that they have a between was Å Å and Å of the three sequences of the of the and showed that and of the residues are are a was observed in the that and in In TD, and ATP was found from the and Fo - Fc electron CAPS, which was in the crystallization a between the domain and the N-terminal domain the consists of four residues and and This that four of the residues in the α subunit are in this region to a hydrophobic The four α of the Cys-clustered domain four residues and The crystal structure revealed that two of and a that is in to Although the R. S. M. S. S.R. 2004; PubMed Google Scholar) family database the of a domain within the sequence of we could the presence of an electron density at the was to the (supplemental Table absorption for PDH1 showed a for with at and the were to be FAD and FMN (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). Although it was from the analysis the is in the β or α subunit, in the study we found the electron density for FMN between the Cys-clustered domain in the α subunit and the interface domain in the β subunit, with the of FMN by two hydrophobic and and and the phosphate group by three and as we this of FMN using the interface of two subunits is FAD is known to between two in enzymes to the family J. Biol. PubMed Scopus Google Scholar), but the of the in the in PDH is Notably, one of the NADPH-cytochrome P450 a of electrons from as a with FAD and FMN as the of and respectively, and to a of P450 M. R. J.J. S. PubMed Scopus Google Scholar). both PDH1 and are diflavin they in the of FAD and electron from l-proline to oxygen as in our to ferricyanide or be the to oxygen be unusual for this as Pyrococcus and the archaeal in which it is In the physiological of electron the structural of the two FAD and FMN, to be The of their suggests that an electron from FAD to FMN is a similar be in PDH1 is as a different of diflavin enzyme because of the between the two are by a in CPR, but by two in between the are in but are and in and the of the diflavin In addition, the oxidase activity and dye-linked PDH activity could be even with a separately expressed β subunit, without the α subunit and FMN (4Kawakami R. Sakuraba H. Tsuge H. Goda S. Katunuma N. Ohshima T. Eur. J. Biochem. 2005; (in press)Google Scholar). to into the α subunit and FMN, we an for electron The between FAD and FMN and between FMN and the are of those in other and are with electron transfer J.M. PubMed Scopus Google Scholar). we the presence of within the PDH1 crystal were within the Cys-clustered it be in an to transfer electrons from FAD in a the electron could be to an as electron In this the α subunit to the but also to electrons to a specific electron acceptor the hydrophobic a in the electron with this is the the region is by of the (αβ)4 heterooctameric So far, four different FAD enzyme have been identified on a their analysis D. Protein 2001; 10: PubMed Scopus Google Scholar). In the of PDH1, both subunits to the structural but each to different In the two were identified D. Protein 2001; 10: PubMed Scopus Google which is by J. Biol. PubMed Scopus Google Scholar), and which is by acid oxidase H. Y. R. J. Biochem. 1996; PubMed Scopus Google Scholar, M. M. S. 1996; PubMed Scopus Google Scholar). a similar structure in their domain as as at one sequence from the the domain, those to in their to the a Z. Z. W. PubMed Scopus Google Scholar) showed that has no similarity with the both except the α subunit of heterotetrameric sarcosine oxidase M. P. 2001; PubMed Scopus Google Scholar). It is that and a different in In addition, we the structure of the α and β subunits on the common fold with different Å for we found that the structure of the α subunit from the β subunit The structural characteristics of PDH1 suggest that each represents a divergent enzyme that arose from a common ancestral flavoenzyme and that they eventually formed a complex to gain a new function. We that our a structural for this unique which is an and archaeal l-proline We N. M. N. and S. for their in data at KEK beamline H. T. H. for with