1. Introduction
The group of glucose-methanol-choline (GMC) flavoprotein oxidoreductases was first outlined by Cavener [
1] and encompasses a wide variety of enzymes from prokaryotic and eukaryotic organisms. The founding enzymes of this class include glucose oxidase (GOX) from
Aspergillus niger, glucose dehydrogenase from
Drosophila melanogaster, methanol oxidase from
Hansenula polymorpha and choline dehydrogenase from
Escherichia coli. Although the sequence similarities of this family are not high and its members catalyze diverse reactions, this family of flavoenzymes contains a conserved ADP-binding motif (an approximately 30 amino acid region) in its
N-terminus and the signature 1 and 2 consensus sequences [
1]. All proteins of this family possess three FAD-binding domains and a substrate-binding domain in a contiguous sequence region [
2]. Interestingly, four flavoenzymes
p-hydroxybenzoate hydroxylase (PHBH),
d-amino acid oxidase, cholesterol oxidase and GOX contain a PHBH-like fold [
3], suggesting that the versatility of this folding topology leads to diverse functions. Surprisingly, a plant hydroxynitrile lyase demonstrated the characteristic topology of this family and was added to the GMC family [
4]. The most probable unrooted phylogenetic tree obtained from 52 selected GMC members revealed five principal evolutionary clades, among which polyethylene glycol dehydogenase (PEG-DH) is a member of the largest clade, which includes alcohol oxidase, GOX, choline and sorbose dehydrogenases [
5].
From the elucidation of the reduced flavin product of
d- and
l-amino acid oxidases with borohydride [
6,
7], the reduced flavin form created by enzymatic dehydrogenation was suggested to be 1,5-dihydroflavin. Dihydroflavin is an effective reductant and reacts readily with molecular oxygen to be re-oxidized [
8]. The GOX reaction proceeds through a ping-pong mechanism [
9]. Recent structural studies of several GMC oxidoreductases, in particular GOXs from
A. niger and
Penicillium amagasakiense have led to a thorough understanding of the catalytic mechanism. Specifically, in the reductive half-reaction, the enzyme catalyzes a two-electron oxidation of β-
d-glucose to δ-gluconolactone, which is non-enzymatically hydrolyzed to gluconic acid. The flavin ring of GOX is reduced to FADH
2, in which His559/563 acts as a catalytic residue to withdraw protons from the substrate and transfer them to FAD [
10,
11]. In the oxidative half-reaction, the same two protons and two electrons are transferred from the reduced enzyme to molecular oxygen by His516 in
A. niger GOX [
12,
13] and His520 in
P. amagasakiensis GOX [
11], yielding H
2O
2 and the re-oxidized FAD. Gadda’s group has studied the catalytic mechanism of choline oxidase (COX) from the
Arthrobacter globiformis strain ATCC 8010 [
14–
16]. They suggested that His466 (corresponding to His516 of GOX) near the flavin N(1) locus is involved in the oxidation of the alcohol substrate, but not in the reduction of oxygen. Ohta
et al.[
17] found that His467 and Asn511 of PEG-DH from
Sphingopyxis terrae correspond to His516 and His559 of GOX from
A. niger. The Asn511His mutant was generated for comparison with His516 of GOX, but almost all activity was lost. This suggests that the asparagine at position 511 is indispensable in PEG-DH.
Tasaki
et al.[
18] cloned an alcohol dehydrogenase gene from octylphenol polyethoxylate-degrading
Pseudomonas putida S-5 that has characteristics of the GMC flavoprotein alcohol dehydrogenases including PEG-DH (Here we designate this enzyme as ethoxy (EO) chain octylphenol dehydrogenase OPEO-DH). The recombinant enzyme recognized EO chains linked to bulky hydrophobic groups, but not free EO chains (PEG). We cloned a gene for a short EO chain nonylphenol dehydrogenase (NPEO-DH) from
Ensifer sp. strain AS08, which catalyzes the initial dehydrogenation of an EO chain nonylphenol [
19]. The gene encoding NPEO-DH consisted of 1659 bp, corresponding to 553 amino acid residues. The presence of an ADP-binding motif and the GMC oxidoreductase signature motifs strongly suggested that the enzyme belonged to GMC oxidoreductase family. The recombinant enzyme exhibited homology (40%–45% identity) with several PEG-DHs.
Amongst GMC oxidoreductases, GOX has been crystallized and is well characterized, and is often used as a model structure of this group, especially for the alcohol/glucose/choline/sorbose oxidoreductases clade. PEG-DH from
S. terrae has a hybrid gene structure that resembles the oxidases and dehydrogenases of this family [
20]. This enzyme has only 30.5% sequence identity with GOX, but 3D molecular modeling suggested that the secondary structures and sequence motifs were conserved in both enzymes [
17]. In this paper, we compare the 3D structures and sequence motifs of NPEO-DH, OPEO-DH and PEG-DHs, and investigate the functions of catalytic amino acid residues in the dehydrogenation of PEG residues and the reduction of FAD.
4. Discussion
The phylogenic tree and alignment of amino acid sequences of NPEO-DH, OPEO-DH and PEG-DH suggested that they share common features of the GMC oxidoreductases family and belong to the same clade that includes GOX, alcohol oxidase, glucose dehydrogenase and choline dehydrogenase in the family [
5]. As NPEO-DH appears to be a type of PEG-DH that acts on free PEGs [
19], while OPEO-DH does not, we expected the closer relationship between NPEO-DH and PEG-DH than between OPEO-DH and PEG-DH. In fact, the opposite result was obtained, although OPEO-DH did not act on free PEG (the reaction is a dehydrogenation of the terminal hydroxyl group in an EO chain). Therefore, we compared the 3D structures of the three enzymes and found that the 3D structure around the active site cavity in OPEO-DH is distinctly different from that of the other two enzymes. First, the size of the active cavity in OPEO-DH is smaller than that of the other two enzymes (
Figure 3). Second, OPEO-DH has hydrophobic β-strand and loop in the entrance of the active site opposite to the flavin (
Figures 3 and
S1–S3 and
Table 1), which would not accommodate PEG molecules. PEG is considered to be a random coil in water, binding approximately 2–4 waters per EO unit to make the polymer-hydrate and become completely hydrophilic. The resultant PEG-water complex loosely aggregates to become far bigger in size than the molecular mass of PEG itself [
29]. Short EO chain alkylphenols probably form different structures from free PEG. Hydrophobic strand and loop would recognize and bind to a hydrophobic alkyl phenol residue, but repel a hydrophilic PEG-water complex. NPEO-DH has a hydrophilic strand as well as a hydrophobic strand, which could explain its activity on NPEO and PEG [
19]. The two loops of PEG-DH would cause substrate size flexibility in the active site cavity,
i.e., accommodation of an oligomeric EO chain to PEG 20000 [
30]. Differences in active site cavity sizes, the secondary structures of the substrate-binding regions, and hydropathy seemed to explain the differences in substrate specificities of the enzymes against EO chain alkyl phenols and PEGs.
His465 and Asn507 in the active site cavity of NPEO-DH are well conserved among the three enzymes and other PEG-DHs [
19]. Comparison of wild type and mutant NPEO-DHs revealed that Asn507 mediates the transfer of proton from a substrate to FAD and His465 mediates the transfer of proton from reduced FAD to an electron acceptor. These results are in accord with the roles of His559 and His516 in GOX of
A. niger[
12,
13]. As histidine does not work at high pH [
12] and the optimal pHs of NPEO-DH and PEG-DH are both 8.0 [
17,
19], His559 in the GOX protein of
A. niger must have been replaced with asparagines in both dehydrogenases. On the other hand, Ghanem and Gadda [
14] showed that, in COX from
A. globiformis, the positive charge brought by His466 near the FAD N(1) locus is necessary for catalytic activity and suggested the involvement of His466 in the reductive half-reaction at pH 6, although the enzyme had another active site Asn510, instead of His559 in the GOX [
10]. Histidines and histidine/asparagine are two combinations of important catalytic residues in the GMC oxidoreductase family, but their roles in the reductive and oxidative half reactions might be different in oxidases and dehydrogenases. As asparagine gives a positive charge at high pH; thus, replacement of histidine with asparagine in PEG-DH and NPEO-DH is considered reasonable.
Taking together, in the reductive half-reaction, NPEO-DH catalyses a two-electron oxidation of a substrate to a corresponding aldehyde; the flavin ring of NPEO-DH is reduced to FADH
2, which should be assisted by Asn507 as the potential proton acceptor. In the oxidative half-reaction, the same two protons and two electrons are transferred by His465 from FADH
2 to an electron acceptor coupled to a respiratory chain, yielding a reduced electron acceptor and regenerating the oxidized flavin (
Figure 6). The presence of quinone-binding motif (
Figure 1) and activity of PEG-DH on Coenzyme Q
10 strongly indicated that an electron acceptor is most probably Coenzyme Q. The typical quinone-binding motif CxxC is located on the second sequence of OPEO-DH and PEG-DH, which is related to the substrate-docking (
Table 1) and close to the membrane-anchoring motif (
Figure S1–3). NPEO-DH does not have the typical quinone-binding motif on the second sequence, but its second sequence is analogous to others, which might act as a quinone-binding site even in NPEO-DH. It is reasonable to assume that electrons released from FADH
2 by His465 are transferred to a ubiquinone in which the isoprene chain is located on the cytoplasmic membrane, but the benzoquinone ring is bound to the second sequence and acts as an actual electron acceptor. In NPEO-DH, another probable quinone-binding motif is shown in
Figures 1 and
S1, where the motif is closer to Asn507 than the second sequence. Asn507 in NPEO-DH corresponds to Asn511 in PEG-DH and Asn508 in OPEO-DH. The fact that Asn511 is indispensable in PEG-DH [
17] supports the crucial role of the active Asn in a PEG-DH group. The reaction mechanisms in dehydrogenases and oxidases of the GMC oxidoreductases family might be different, an idea, which would be elucidated by further experiments.