A novel aldo-keto reductase (AKR) from Escherichia coli has been cloned, expressed and purified. This protein, YghZ, is distantly related (<40%) to mammalian aflatoxin dialdehyde reductases of the aldo-keto reductase AKR7 family and to potassium channel β-subunits in the AKR6 family. The enzyme has been placed in a new AKR family (AKR14), with the designation AKR14A1. Sequences encoding putative homologues of this enzyme exist in many other bacteria. The enzyme can reduce several aldehyde and diketone substrates, including the toxic metabolite methylglyoxal. The Km for the model substrate 4-nitrobenzaldehyde is 1.06 mM and for the endogenous dicarbonyl methylglyoxal it is 3.4 mM. Overexpression of the recombinant enzyme in E. coli leads to increased resistance to methylglyoxal. It is possible that this enzyme plays a role in the metabolism of methylglyoxal, and can influence its levels in vivo.
The aldo-keto reductases (AKR) are a superfamily of enzymes that catalyse the reduction of aldehydes and ketones to alcohols (reviewed in ). Over 50 AKR enzymes have been identified, and these have been grouped within families (AKR1 to AKR14) if they share more than 40% identity [1,2]. Many mammalian and eukaryotic AKR enzymes have been studied, but only 4 bacterial enzymes have been characterised as enzymes. These include: morphine dehydrogenase from Pseudomonas putida and a 2,5-diketo-d-gluconate reductase (2,5-DKGR) from Corynebacterium sp.  The recent availability of bacterial genome sequence data has revealed the presence of many open reading frames (ORFs) encoding putative AKR. Two ORFs in the Escherichia coli genome that are similar to Corynebacterium 2,5-DKGR have recently been cloned and expressed [5,6], but the functions of the vast majority of the remaining putative AKR are completely unknown .
A putative AKR identified in the genome of E. coli is similar to mammalian AKR of the AKR7 family  as well as to several eukaryotic K+ channel β-subunits of the AKR6 family . The AKR7 family contains mammalian aflatoxin dialdehyde reductases, which show a preference for acid-aldehydes, dialdehydes and diketones . The structure of AKR7A1 has recently been determined, and although it has a simlar α/β-barrel structure to other AKR, it is the first dimeric enzyme to be crystallised . The underlying catalytic mechanism of AKR7A1 is similar to other AKR, requiring four amino acids (Tyr, His, Lys, Asp) in which a tyrosine residue acts as the general acid, forming a proton relay from the histidine. However, the substrate-binding pocket of AKR7A1 differs from other AKR in that it contains several charged amino acids (Arg-17, Arg-231 and Arg-327) which may account for its specificity for 4-carbon acid-aldehydes such as succinic semialdehyde (SSA) and 2-carboxybenzaldehyde, as well as aflatoxin B1 dialdehyde . The K+ channel β-subunits were first noted as being similar to the AKR superfamily by McCormack and McCormack , and this degree of similarity led to them being included in the AKR superfamily, forming the AKR6 family . Subsequently the structure of the K+ channel β-subunit Kvβ2 from rat (AKR6A2) has been determined . Although these K+ channel β-subunits share many features of the AKR superfamily, they have not been shown to have any activity as AKR. This may be because these proteins lack the catalytic histidine that active AKR possess. It is not known why they are similar to AKR, but it has been proposed that the binding of NADPH to the β-subunit causes a conformational change, allowing these proteins to interact with the other components of the potassium channel and possibly conferring regulation upon the channel . The K+ channel β-subunits are also different from other AKR in that they are tetrameric, a feature which is suggested allows them to associate with the tetrameric α-subunits of the channel .
In this paper, we describe the cloning, expression and characterisation of the E. coli YghZ enzyme which appears to share features of both the AKR6 and AKR7 protein families. It is hoped that increased knowledge of this bacterial enzyme may provide valuable information about both of the related protein families. Unlike AKR6 proteins, YghZ is catalytically active, and we demonstrate its ability to contribute to methylglyoxal metabolism in vitro and in vivo.
Materials and methods
E. coli strain 1400 (supE, supF, recA56, met, thi, L512)  was used as the wild-type strain. For expression and toxicity studies BL21(DE3)/pLysS (F−ompT r−B m−B, λDE3 lysogen) was used.
Cloning and sequencing of YghZ
Genomic DNA was prepared from E. coli strain 1400 using standard procedures . The following primers were used to amplify the ORF encoding YghZ: 5′-CCGGAATTCATATGGTCTGGTTAGCGAATC-3′ and 5′-CGCGGATCCATAATCACTGATCTGATTCG-3′. These contained 5′Eco RI and Nde I sites and a 3′Bam HI site. After digestion with Eco RI and Bam HI, the PCR product was cloned into the Eco RI/Bam HI sites of pTZ19r (APBiotech) to give pEE87. The DNA insert was sequenced to verify fidelity of amplification. Multiple sequence alignments of YghZ with AKR family members was carried out using Clustal-X .
Expression and purification of YghZ
The Nde I/Bam HI fragment from pEE87 was subcloned into the Nde I/Bam HI site of pET15b to give plasmid pGS1, placing the ORF under the control of the T7 promoter, which is inducible by IPTG in the expression host BL21pLysS. The ORF is also in frame with an N-terminal 6-His tag, allowing it to be expressed as a fusion protein. Expression of recombinant YghZ protein and purification of the protein using a His-tag affinity column was achieved as described previously . Essentially, bacterial cells were lysed using a French press in 300 mM NaCl, 20 mM Tris pH 8.0, and after removing cell debris through centrifugation the soluble extract was applied to a nickel-agarose column. Fractions were eluted using an imidazole stepped gradient, pure fractions were pooled and dialysed against 20 mM sodium phosphate buffer pH 7.0. The His-tag was removed using human thrombin. Purified recombinant protein was separated by SDS–PAGE using 12% polyacrylamide resolving gels with the buffer system described by Laemmli  and gels were stained using Coomassie blue.
Protein and enzyme assays
Protein concentrations were measured using the method of Bradford  and standardised using bovine serum albumin, using a kit from Bio-Rad Laboratories Ltd. (Hemel Hempstead, UK). Aldehydes and ketone substrates were obtained from Sigma Chemical Co. (Poole, Dorset, UK) or from Aldrich Chemical Co. (Gillingham, Dorset, UK). Aldehyde- and ketone-reducing activity was routinely measured with a Beckman DU650 UV single-beam recording spectrophotometer by following the initial rate of oxidation of NADPH at 340 nm (ɛ=6270 M−1 cm−1) in 0.1 M phosphate buffer pH 6.0 as described previously . For specific activities, the NADPH concentration used was 10 µM, and the concentration of substrate was 0.1 or 1 mM. Apparent Km and Kcat values were determined by measuring the initial reaction rate over a range of substrate concentrations and were calculated using Ultrafit curve-fitting software (Biosoft, Cambridge, UK) using the Marquardt alogrithm.
E. coli cells containing the expression construct pGS1 or control pET15b were grown at 37°C in LB to the early exponential phase (OD600= 0.1), and the expression of YghZ was induced by the addition of 0.5 mM IPTG. After 75 min, 1-ml aliquots of cells were exposed to a range of concentrations of methylglyoxal (0–10 mM) for 1 h, after which time cells were harvested by centrifugation for 5 min at 9000×g and washed twice in LB, before plating out serial dilutions in order to estimate the number of viable bacteria remaining.
Cloning of YghZ
Six ORFs in the E. coli genome have significant similarity to members of the AKR superfamily . One of these, YghZ (previously ORF_o346 or B3001), was of particular interest because of its potential role as a dicarbonyl reductase, as well its similarity to a potassium channel regulator . We therefore used specific primers to amplify the gene from E. coli genomic DNA. Verification of the sequence, and its predicted amino acid sequence, confirmed its identity to YghZ (accession no. Q46851). A multiple sequence alignment of YghZ reveals similarities which can be mapped to structural features of the AKR7A1 and AKR6A2 proteins (Fig. 1). At the amino acid level, it is 33% identical to AKR7A1  and 37% identical to AKR6A2 . YghZ possesses four amino acids which could form part of the catalytic tetrad of the active site. These are Tyr-66, Hys-137, Lys-97 and Asp-61. In comparison to other AKR, it possesses an extended loop similar to the corresponding loop regions present in AKR6A2 (between β9 and α8) and AKR7A1 (between β7 and α7). At the C-terminal end, where there are considerable differences between AKR7A1 and AKR6A2, YghZ being more similar to AKR7A1. However, unlike AKR7A1, it does not possess an extreme C-terminal arginine, although it does have a positively charged lysine in a similar position. Neither of the other two arginine residues that form part of the AKR7A1 active site is conserved.
A BLAST search of sequences in the databases reveals that the predicted protein sequence is similar to many other ORFs present in bacterial genomes, in some cases with over 60% amino acid identity. These include ORFs in: Xanthomonas axonopodis, Agrobacterium tumefaciens, the nitrogen-fixing symbiotic bacterium Mesorhizobium loti, the legume symbiont Sinorhizobium meliloti, Streptomyces coelicolor, Corynebacterium glutamicum and Salmonella spp.  (Fig. 1). In many cases, these unknown ORFs have been classified as either putative oxido-reductases or putative ion channel proteins. Because of the high degree of similarity, it is likely that these ORFs encode proteins which carry out a similar function to YghZ.
YghZ is an active AKR
To determine whether the YghZ protein has catalytic activity towards aldehydes and ketones, and to identify possible endogenous substrates, the protein was expressed and purified from bacterial cells (Fig. 2). A single band with a molecular mass of approximately 39 kDa was observed on SDS–PAGE gels (Fig. 2, lane 7), and after removal of the His-tag, the activity of the protein was assayed. The enzyme is active with NADPH as cofactor, but no activity was detected with NADH as cofactor under the conditions tested. The results in Table 1 show that the enzyme is capable of catalysing the reduction of a variety of aldehydes, dialdehydes, acid-aldehydes and diketones at rates that are comparable to rat liver AKR7A1. Of the physiological substrates, YghZ shows relatively high specific activity with methylglyoxal, but low activity towards the AKR7A substrate SSA. The enzyme showed negligible activity towards sugars such as d-xylose, though it was able to reduce d, l-glyceraldehyde to some extent. This range of activities clearly distinguishes YghZ from the AKR6 family, as these ion channel-associated proteins have not been shown to reduce any of these aldehyde substrates .
|Substrate||Specific activity (nmol min−1mg−1)|
|Substrate||Specific activity (nmol min−1mg−1)|
Enzyme assays were carried out using purified recombinant YghZ at 25°C, pH 7.0 by measuring the rate of decrease of NADPH with substrates at 1 mM. Values represent the mean±S.E.M. of triplicate assays.
To characterise the enzyme further, apparent kinetic constants were determined for those substrates for which YghZ had highest specific activity (Table 2). The affinity for methylglyoxal (Km= 3.24 mM) suggests it could function to reduce methylglyoxal over the range of physiological levels found within the cell (up to 1.4 mM) .
Constants were calculated by measuring the initial rate over a range of substrate concentrations. Values represent the mean±S.E.M.
YghZ can detoxify methylglyoxal in vivo
Overexpression of YghZ from the IPTG-inducible promoter in E. coli gives rise to a protein band that constitutes approximately 1% of total cell protein (lanes 2 and 3, Fig. 2). These cells have significantly increased levels of NADPH-dependent methylglyoxal reductase activity compared to control cells (Fig. 3). The ability of YghZ to detoxify methylglyoxal in vivo was assessed by examining the contribution of the overexpressed enzyme to protection against methylglyoxal toxicity. Exposure of control cells to concentrations of up to 5 mM methylglyoxal for 60 min led to substantial loss of viability (Fig. 4), and gave rise to variable size colonies, suggesting cells are accumulating damage to DNA at the effective concentration used. However, cells expressing recombinant YghZ were more resistant to high concentrations of methylglyoxal than control cells (Fig. 4). This increased resistance correlates with the increased methylglyoxal reductase activity seen in cells expressing high levels of YghZ, and suggests that the enzyme is capable of playing a role in methylglyoxal metabolism/detoxification in vivo.
We have cloned and characterised a member of the AKR superfamily from E. coli, and have shown that it has activity towards a range of substrates including methylglyoxal. Because it is less than 40% identical to currently characterised AKR, it has been placed in a new family (AKR14) with the designation AKR14A1. It is likely that related ORFs identified in other microbial genomes (Fig. 1) are also part of the AKR14 family and share similar functions. Several other ORFs which show between 30 and 40% amino acid identity to YghZ are present in the Arabidopsis thalania and Schizosaccharomyces pombe genomes (data not shown). Given the differences that may be associated with divergence of species, these ORFs may also encode enzymes with related activities.
We examined the activity of YghZ towards several physiological substrates. One of these, methylglyoxal, is a biologically active dicarbonyl that is known to be produced by a range of microorganisms, including E. coli[25–27]. Because of its reactive carbonyl groups, it causes damage to DNA and protein [28,29], leading to loss of viability [25,30]. In E. coli, detoxification of methylglyoxal has been demonstrated previously through two related routes: (1) the activation of K+ channels KefB and KefC by a methylglyoxal–glutathione conjugate which leads to cytoplasmic acidification that protects against the electrophile ; and (2) metabolism of methylglyoxal–glutathione conjugates via glyoxalase I and II to d-lactate and glutathione [26,32]. However, several other enzymes have also been reported as being capable of metabolising methylglyoxal. It can be reduced to acetol through the action of an alcohol dehydrogenase or aldehyde reductase ; or it can be oxidised to pyruvate by an aldehyde dehydrogenase . AKR appear to have evolved predominantly in favour of alcohol production at the expense of NADPH . Although an aldehyde reductase and an alcohol dehydrogenase capable of reducing methylglyoxal have been purified from E. coli previously , these enzymes have been only partially characterised, and nothing is known about the genes that encode them. YghZ is a possible candidate for such a methylglyoxal reductase.
Methylglyoxal is synthesised endogenously in bacteria from dihydroxyacetone phosphate by methylglyoxal synthase [26,35]. Several possible functions for methylglyoxal in the cell have been put forward, including as a growth regulator , as a bypass for glycolysis, and as a way of preventing the accumulation of certain phosphorylated sugars . In support of this latter role, if entry of glucose-6-phosphate into the cell is not regulated, then methylglyoxal accumulates intracellularly and cells will die . Regulation of methylglyoxal levels is therefore an important determinant in cell growth and survival. In addition to factors that control methylglyoxal synthesis, its break-down needs to be similarly controlled. It is possible that a methylglyoxal reductase such as YghZ could make an important contribution to methylglyoxal removal in vivo. The glyoxalase I enzyme has been shown to be important in protecting cells against methylglyoxal concentrations of up to 0.8 mM , but it is not known what happens when this pathway is saturated.
We have shown that when YghZ is expressed at high levels it can increase resistance to methylglyoxal at effective concentrations that are physiological. Little is currently known about the endogenous expression levels of this protein, though it is anticipated that information from gene arrays will lead to a wealth of information on its transcriptional regulation by carbon source and other environmental factors. Additionally, the construction of a strain in which the YghZ gene has been knocked out will in the future allow the enzyme's function in a variety of pathways to be investigated, including metabolism of methylglyoxal, mediating cell growth and regulation and the effects of phosphorylated sugars.
The similarity of YghZ to eukaryotic potassium channel-associated β-subunits is intriguing and at first suggests the possibility that it may also associate with a potassium channel in vivo. A putative potassium channel (Kch) that shares some similarity to the α-subunit of the mammalian K+ channel has also been identified in E. coli, though little is known of its function [38,39]. It is possible however that the observed similarity of YghZ to K channel β-subunits is vestigial, and does not relate to its current role in the cell.
This research was initiated through the award of a Society for General Microbiology Vacation Studentship (ref Vac 98/44) to H.W. We are grateful to Derek Jamieson for helpful comments and advice and to Prof Iain Hunter for use of the spectrophotometer. We would also like to thank Dr Adrian Lapthorn for assistance with the structural alignment and information resourcing.