Role of GldA in dihydroxyacetone and methylglyoxal metabolism of Escherichia coli K12


  • Editor: Bruce Ward

Correspondence: Chankyu Park, Department of Biological Sciences, Korea Advanced Institute of Science and Technology, Yusong-Ku, Taejon 305-701, Korea. Tel.: +82 42 869 2629; fax: +82 42 869 2610; e-mail:


The metabolic pathway involving dihydroxyacetone is poorly characterized although novel enzymes associated with this metabolite have recently been demonstrated. The role of GldA in dihydroxyacetone and methylglyoxal metabolism was investigated by purifying the enzyme and characterizing its catalytic ability using nuclear magnetic resonance (NMR) spectroscopy. At neutral pH, the enzyme exhibits much higher affinities towards dihydroxyacetone, methylglyoxal, and glycolaldehyde than glycerol with Km values of 0.30, 0.50, 0.85, and 56 mM, respectively. This is consistent with NMR data with crude extracts, showing that the conversion from dihydroxyacetone to glycerol by GldA is far more efficient than the reverse reaction. Dihydroxyacetone was found to be lethal at higher concentration with an LC50 value of 28 mM compared with 0.4 mM of methylglyoxal, while lactaldehyde was found to exhibit significant growth inhibition in Escherichia coli cells. The toxicity of dihydroxyacetone appears to be due to its intracellular conversion to an aldehyde compound, presumably methylglyoxal, since the glyoxalase mutant becomes sensitive to dihydroxyacetone. Based on information that gldA is preceded in an operon by the ptsA homolog and talC gene encoding fructose 6-phosphate aldolase, this study proposes that the primary role of gldA is to remove toxic dihydroxyacetone by converting it into glycerol.


Two glycerol assimilatory pathways are known to occur in members of the Enterobacteriaceae: first, glycerol is converted by glycerol kinase to glycerol-3-phosphate that is subsequently oxidized to dihydroxyacetone phosphate by a specific flavoprotein-linked glycerol-3-phosphate dehydrogenase, and second, a different two-step reaction requiring a glycerol dehydrogenase, and then a dihydroxyacetone kinase. Although the former pathway is the major glycerol assimilation pathway in Escherichia coli (Neijssel et al., 1975; Lin, 1976; Jin et al., 1983), an alternate pathway for glycerol utilization can also be used, which is mediated by glycerol dehydrogenase (GldA) (Fig. 1) converting glycerol to dihydroxyacetone (DHA) (Jin et al., 1983; Truniger & Boos, 1994). Although earlier studies in E. coli indicated that enzymatic catalysis by GldA was reversible between dihydroxyacetone and glycerol [Fig. 1 (Tang et al., 1979)], its in vivo role has not been clearly established.

Figure 1.

 Metabolism of dihydroxyacetone, methylglyoxal, and lactaldehyde in Escherichia coli. F-6-P, fructose-6-phosphate; F-1,6-P, fructose-1,6-bisphosphate; G-3-P, glyceraldehydes-3-phosphate; Glycerol-3-P, glycerol-3-phosphate; and 1,2-PD, 1,2-propanediol. The genes encoding enzymes involved in the pathway are fructose-6-phosphate aldolase 1 (fsa), fructose-6-phosphate aldolase 2 (talC), glycerol kinase (glpK), glycerol-3-phosphate dehydrogenases (aerobic glpABC and anaerobic glpD), methylglyoxal synthase (mgsA), glyoxalase I/II, III (gloA, B and C), lactaldehyde dehydrogenase (aldA), glycerol dehydrogenase (gldA), dihydroxyacetone kinases (dhaKLM), 1,2-propanediol oxidoreductase (fucO), and aldo–keto reductases (AKR). The suspected, but not characterized activity is marked [*, (Altaras & Cameron, 1999)].

The E. coli GldA (EC, also known as NADH-linked reductase, has broad substrate specificity, capable of reducing hydroxyacetone (Tang et al., 1979; Lee & Whitesides, 1986). The enzyme exhibits sequence similarity (49% identity) to glycerol dehydrogenase of Bacillus sterothermophilus, a member of iron-containing alcohol dehydrogenase family (Ruzheinikov et al., 2001). Although the enzyme has been characterized previously (Asnis & Brodie, 1953), the gene was mapped and cloned much later (Truniger & Boos, 1994). Earlier studies of E. coli GldA indicated that it is identical to d-1-amino-2-propanol oxidoreductase and able to oxidize 1,2-propanediol (1,2-PD) as well as glycerol (Kelley & Dekker, 1984). The crystal structure of GldA from Thermotoga maritima has been determined (Srinivasan et al., 2002).

Dihydroxyacetone and methylglyoxal (MG) are toxic metabolites being produced during glycolysis in different types of cells, although the metabolic pathway involving dihydroxyacetone is poorly understood in E. coli. Dihydroxyacetone could be generated by aldol cleavage from fructose-6-phosphate in E. coli by the action of fructose-6-phosphate aldolases, Fsa and TalC [Fig. 1 (Schurmann & Sprenger, 2001)]. Recent findings in E. coli indicate that the Dha kinase of E. coli, consisting of DhaK, DhaL, and the PTS homolog of DhaM (Daniel et al., 1995; Gutknecht et al., 2001) is involved in uptake and phosphorylation of dihydroxyacetone [Fig. 1 (Paulsen et al., 2000; Gutknecht et al., 2001)]. Although dihydroxyacetone is produced as a metabolic intermediate, accumulation of this compound is presumed to be toxic. Spontaneous but inefficient conversion of dihydroxyacetone to methylglyoxal occurs with t1/2 of 7 h in 100 mM phosphate buffer, pH 7.4 at 40 °C (Riddle & Lorenz, 1968). Dihydroxyacetone and short-chain trioses are known to have increased propensities to react with proteins by Maillard-type reaction (Tessier et al., 2003), and thus induce DNA damage, cell-cycle block, and apoptosis in cultured HaCaT keratinocytes (Petersen et al., 2004). The toxic effect of dihydroxyacetone and its metabolic conversion in E. coli cells is poorly understood.

Methylglyoxal is a reactive 2-oxaldehyde derived from glycolysis, inhibiting growth of an organism. In E. coli, dihydroxyacetone phosphate (DHAP) is converted to methylglyoxal by methylglyoxal synthase (MgsA) (Hopper & Cooper, 1971), releasing inorganic phosphate (Pi, Fig. 1, mgsA). The glyoxalase system, aldo–keto reductases, and alcohol dehydrogenases are well established routes for methylglyoxal detoxification in E. coli. The glyoxalase I/II system of E. coli converts methylglyoxal into d-lactate in the presence of glutathione (Fig. 1, gloAB) (Vander Jagt, 1993; MacLean et al., 1998). In contrast, glyoxalase III generates d-lactate directly (Fig. 1, gloC) (Misra et al., 1995). The glutathione-independent pathways of E. coli includes aldo–keto reductases, which are known to convert methylglyoxal to acetol (Fig. 1, AKR) through multiple pathways, consisting of yafB, yeaE, and yghZ genes (Grant et al., 2003; Ko et al., 2005). In addition, GldA is able to reduce methylglyoxal to lactaldehyde and further to 1,2-PD, and thus has been used for enhanced production of 1,2-PD [Fig. 1 (Altaras & Cameron, 1999)].

l-Lactaldehyde, one of the major products of methylglyoxal metabolism, was originally identified as an intermediate in the metabolic pathway of l-fucose and l-rhamnose utilization. Under aerobic conditions, l-lactaldehyde is oxidized to l-lactate by aldehyde dehydrogenase (aldA), while under anaerobic conditions, l-lactaldehyde is reduced to l-1,2-PD by the propanediol oxidoreductase [fucO in Fig. 1 (Cocks et al., 1974; Zhu & Lin, 1989)]. Accumulation of lactaldehyde is suggested to cause growth inhibition (Zhu & Lin, 1989).

In this study, an attempt was made to characterize the role of GldA in dihydroxyacetone and methylglyoxal metabolism in E. coli by purifying the enzyme and characterizing its catalytic ability using UV and nuclear magnetic resonance (NMR) spectroscopy. A range of substrates were analyzed at physiological pH with their products identified. Toxicity of dihydroxyacetone was also studied. Based on these findings, it was proposed that the primary role of E. coli GldA is to remove dihydroxyacetone by converting it to glycerol. It was also observed that the lethal effect of dihydroxyacetone is mediated by the enzymes involved in dihydroxyacetone metabolism. Toxicity of d-lactaldehyde, a product from methylglyoxal by GldA, and its metabolism by AldA is discussed.

Materials and methods


d-Lactaldehyde was synthesized by the reaction of ninhydrin with l-threonine (Zagalak et al., 1966). Other chemicals and substrates were obtained from Sigma-Aldrich or from Wako (Osaka, Japan).

Bacterial strains and plasmids

All strains used in this study are derivatives of E. coli K12. MG1655 was used as wild type strain for gene amplification and gene disruption. Standard procedures were used for plasmid purification, restriction analysis, ligation, and transformation. Disruption of aldA gene was performed by the one-step method for inactivation of chromosomal genes using PCR (Datsenko & Wanner, 2000). Deletion was designed to be in-frame to eliminate polar effects of the expression of downstream genes. The gldA∷Km, fucO∷Km, gloA∷Km and aldB∷Km mutants were obtained from University of Wisconsin (E. coli Genome Project), S.C. Kim, Reid C. Johnson, and I.R. Booth and were transferred to MG1655 using P1. Other strains were from lab collection or made during this work. P1 of fucO∷Km mutant was transduced to ΔaldA strain to make aldA fucO double mutant. The BL21 (DE3) strain was used for overexpression and purification of protein.

Cell growth and media

Unless otherwise specified, cells were grown at 37 °C in Luria–Bertani (LB) broth containing appropriate antibiotics.

Sample preparation for NMR analysis

To analyze metabolites and their conversions from wild type and mutant strains, cells were cultured overnight in LB medium. Cells were then harvested by centrifugation, washed twice with 100 mM potassium phosphate (pH 7.0), and resuspended in the same buffer. The cells were disrupted by sonication, and debris was removed by centrifugation at 15 000 g for 30 min. The resulting supernatants were dialyzed for 15 h against three consecutive changes of 100 mM potassium phosphate buffer (pH 7.0). The supernatants were stored at −70 °C until use.

NMR analyses of metabolites

The Bruker AVANCE-400 NMR spectrometer equipped with a temperature controller was used for NMR experiments. All measurements were carried out in 5 mm NMR tube using 600 μL of solution with 10% D2O as a locking substance. For characterization of metabolites from enzymatic reactions of methylglyoxal, dihydroxyacetone, lactaldehyde (3 mM each), and glycerol (8 mM) with crude extracts, NMR measurement was carried out for the mixture of crude proteins (1–2 mg), substrate, and coenzymes (1 mM NADH, 1 mM NADPH, 5 mM NAD) in buffer (100 mM potassium phosphate, pH 7.0) with D2O. The methods used to measure the metabolic conversion of dihydroxyacetone by purified GldA were identical, except for the amount of protein used (10 μg). The NMR data were collected at indicated times.

Cloning of E. coli gldA by PCR

The genomic DNA of E. coli MG1655 was used as template for amplification of gldA. The PCR primer pairs used were 5′-cgcatatgccgcatt tggcacta-3′ and 5′-cgctcgagttcccactcttgcag-3′ (the NdeI and XhoI sites are underlined). The PCR products were digested with NdeI and XhoI, which was introduced into NdeI and XhoI restriction sites of pET21b (Novagen). Thus, the GldA protein was expressed with C-terminal His tag.

Purification of GldA

The pET-gldA-His containing E. coli BL21(DE3) was grown in LB broth with ampicillin (100 μg mL−1) at 30 °C to mid-exponential phase (0.5 at OD600 nm), and the protein was induced by adding isopropyl-β-d-thiogalactopyranoside (0.5 mM). The cells were continued to grow further for 3 h. Cells were harvested and resuspended in binding buffer (20 mM Tris-Cl, pH 7.9, 5 mM imidazole, 500 mM NaCl). After disruption by sonication, cell debris was removed by centrifugation at 15 000 g for 15 min, and the protein was purified by standard procedures for Ni-NTA column (Novagen). The concentration and purity was determined with the Bradford reagent and with sodium dodecylsulfate-polyacrylamide gel electrophoresis, respectively.

Enzyme assays

Enzyme activities of the purified E. coli GldA was measured at 25 °C using a Beckman Coulter DU800 spectrophotometer by monitoring the initial rate at 340 nm with oxidation of NADPH/NADH or reduction of NADP/NAD. The substrate specificity was assayed in 1 mL of 100 mM potassium phosphate buffer (pH 7.0) with 0.1 mM NADH or 5 mM NAD as coenzyme. In all reactions, nonenzymatic rates were subtracted from the observed initial reaction rates. The apparent Km and kcat values were determined by measuring the initial rate over a range of substrate concentrations. The kinetic parameters were determined with Sigmaplot (SPSS Inc., Chicago, IL) by fitting to the Michaelis–Menten equation. To measure specific activities of crude proteins, the same assay was adopted except that 100 μg of total protein was used with substrates dihydroxyacetone and glycerol at 5 mM concentration. Crude extracts were prepared from wild type and gldA mutant grown in LB media for 12 h with shaking (200 r.p.m.) at 37 °C or in modified Fraser and Jerrel medium as described (Kelley & Dekker, 1984) in order to ensure expression of GldA enzyme. The cells were disrupted by sonication, and debris was removed by centrifugation at 15 000 g for 30 min. The resulting supernatants were dialyzed for 15 h against three consecutive changes of 100 mM potassium phosphate buffer (pH 7.0). The supernatants were stored at −70 °C until use.

Effect of lactaldehyde and dihydroxyacetone in cell viability

Cell viability after being exposed to lactaldehyde was determined for two different strains. Growth was compared by streaking fresh colonies of wild-type and aldA mutant on LB plates with or without 1.5 mM d-lactaldehyde. To test toxic effect of dihydroxyacetone and to compare it with that of methylglyoxal, the growth was compared on LB plate containing different concentrations of dihydroxyacetone and methylglyoxal. Fresh colonies of wild-type and mutant strains were grown overnight in LB broth, diluted 100-fold in the same medium and incubated with shaking until OD600 nm reaches to 1.0. The cells were diluted from 10−1 to 10−6 and spotted (4 μL) onto LB plates containing different concentrations of dihydroxyacetone and methylglyoxal. The growth was compared after 12–14 h of incubation at 37 °C.

Measurement of lethal concentrations (LC50)

Cells were grown overnight at 37 °C in M9 minimal medium with 0.4% glycerol and resuspended in the same medium to an OD600 nm of 0.01. After growing to an early exponential phase, cells were diluted 10-fold in fresh M9 medium containing 0.4% (w/v) glycerol. The cells were serially diluted and spreaded onto LB agar plates with indicated concentrations of methylglyoxal/dihydroxyacetone/HA. After 15 h of incubation at 37 °C, colonies were counted to measure LC50.

Results and discussion

Substrate specificity of E. coli glycerol dehydrogenase

His-tagged GldA was purified and its activity was measured with a variety of substrates in both the forward and reverse direction. As shown in Table 1, activity was detected for seven substrates, glycerol and 1,2-PD for oxidation and dihydroxyacetone, methylglyoxal, glycolaldehyde, hydroxyacetone, and dl-glyceraldehyde for reduction. The highest specific activity was obtained for dihydroxyacetone followed by glycolaldehyde, glycerol, methylglyoxal, HA, and dl-glyceraldehyde (Table 1). Reaction products for the substrates were characterized by 1H-NMR spectroscopy (Table 1). The kinetics of purified GldA was measured for highly specific substrates at pH 7.0 in 100 mM potassium phosphate buffer. The Km for dihydroxyacetone and methylglyoxal were 0.3 and 0.5 mM, respectively, relative to 56 mM for glycerol (Fig. 2 and Table 1) with kcat/Km for dihydroxyacetone more than 100-fold higher than that of glycerol (Table 1). When crude extracts of wild-type and gldA deletion strains were mixed with dihydroxyacetone in the presence of NADH, significant amount of glycerol was obtained from the wild type, whereas none were detected from the gldA deletion strain (Fig. 3a, dihydroxyacetone). In contrast, extremely low level of dihydroxyacetone was detected from glycerol even in wild type (Fig. 3a, glycerol). This is consistent with specific activities measured from crude extracts of wild-type and gldA deletion strains grown in modified media (Kelley & Dekker, 1984), 0.20±0.02 vs. 0.075±0.004 μmole min mg−1 of protein for dihydroxyacetone (5 mM, details in the ‘Materials and methods’), respectively. However, specific activity of 0.019±0.04 vs. <0.001 μmole min mg−1 was detected for wild-type and gldA deletion strains, respectively, for glycerol at 5 mM concentration. No activity was detected for glycerol for both strains grown in LB.

Table 1.   Substrate specificities and kinetics of GldA
SubstrateProductSpecific activity
(nmol min−1 mg−1)
Km (mM)kcat (s−1)kcat/Km (s−1 M−1)
  1. Enzyme activities were measured at 25°C using a Beckman Coulter DU800 spectro-photometer by monitoring the initial rate at 340 nm with NADH oxidation or NAD reduction. The reaction mixture contained 2 μg of purified protein, various concentrations of substrates, and 100 μM of NADH or 5 mM NAD+ in 100 mM potassium phosphate buffer (pH 7.0). For measuring specific activity, 100 mM glycerol was used, and all other substrates were 1.0 mM concentrations. To measure kinetic properties; dihydroxyacetone, glycolaldehyde, methylglyoxal, and glycerol were used in the range of 0.01–10, 0.1–10, 0.1–10, and 0.1–700 mM concentrations, respectively. The reaction products were determined by 1NMR analysis. The authors were unable to measure kinetics for the substrates with low or no activities. The compounds incapable of serving as substrates are d-lactaldehyde, benzaldehyde, 3-nitrobenzaldehyde, 4-nitrobenzaldehyde, and 2-carboxybenzaldehyde.

Reduction (+NADH)
 DihydroxyacetoneGlycerol55 7000.3017.25.73 × 104
 GlycolaldehydeEthylene glycol50 5000.8531.73.73 × 104
 MethylglyoxalLactaldehyde90000.506.11.22 × 104
Oxidation (+NAD)
 GlycerolDihydroxyacetone33 0005622.44.00 × 102
 dl-1,2-PropanediolNot detected1100
Figure 2.

 Enzyme kinetics. Michaelis–Menten kinetics of GldA for dihydroxyacetone, methylglyoxal, and glycerol. Substrate saturation curves for dihydroxyacetone, methylglyoxal, and glycerol are obtained with at least 10 different concentrations. The Lineweaver–Burk plots are shown as insets. Velocity is expressed as μmol min−1, and substrate concentration as mM. Reaction conditions and substrate concentrations used are as described in Table 1. Details are described in ‘Materials and methods’.

Figure 3.

1H-NMR spectrum showing metabolic conversion of dihydroxyacetone and d-lactaldehyde. (a) 1H-NMR spectra indicating dihydroxyacetone-converting activity of crude extracts. Crude extracts of wild type and gldA deficient strains were mixed with dihydroxyacetone or glycerol in the presence of NADH or NAD, respectively, and the metabolites were analyzed with 1H-NMR at indicated time periods (3 and 30 min after mixing substrate, enzyme, and cofactor). (b) 1H-NMR spectrum indicating dihydroxyacetone-converting activity of purified GldA. The same method was used to measure conversion of dihydroxyacetone by purified GldA (10 μg) in the presence of NADH. (c) 1H-NMR spectra showing metabolic conversion of d-lactaldehyde to d-lactate by crude extracts of wile-type and mutant Escherichia coli strains in the presence of NAD. The strains, metabolites, and peaks representing different metabolites are designated; dihydroxyacetone (▴) and glycerol (- - -) in (a) and (b).

Previous studies on the E. coli GldA (Tang et al., 1979; Jin et al., 1983) were inconclusive about its physiological substrate, for the enzyme exhibited broad specificity and different optimum conditions for different substrates, i.e. pH 5.5–6.0 for dihydroxyacetone and 9.5–10.0 for glycerol. The observation on GldA for its affinity and specificity towards dihydroxyacetone at physiological pH (Fig. 2) suggest that the enzyme serves primarily as a dihydroxyacetone converting enzyme. This is further substantiated by an observation that the crude E. coli extract displayed the catalytic activity to dihydroxyacetone, not to glycerol (Fig. 3a), and that the gldA mutant became sensitive to dihydroxyacetone (Fig. 4), presumably because of inefficient removal of dihydroxyacetone incorporated into the cell. The reaction for dihydroxyacetone by GldA is well known in some fungal species including Hypocrea jecorina (Liepins et al., 2006), Aspergillus nidulans, and Aspergillus niger (Schuurink et al., 1990), although the pathway is poorly understood in bacteria including E. coli.

Figure 4.

 Susceptibilities of Escherichia coli cells to dihydroxyacetone. Cells from OD600 nm=1.0 were diluted from 10−1 to 10−6, spotted (4 μL) onto LB plates containing dihydroxyacetone at the indicated concentrations, and incubated at 37°C for 12–14 h.

Intracellular conversion of d-Lactaldehyde

As shown above (Table 1), GldA catalyzes the conversion of methylglyoxal to d-lactaldehyde, a reactive aldehyde. The externally added d-lactaldehyde at 1.5 mM concentration resulted in growth inhibition of E. coli MG1655 (not shown). The metabolic conversion of l-lactaldehyde has previously been described as part of the l-fucose pathway (Cocks et al., 1974; Zhu & Lin, 1989). Here the authors addressed the question as to whether the compound is a substrate for GldA or AldA previously known to catalyze l-lactaldehyde. For l-lactaldehyde, it was shown that GldA is involved in its reduction, while it does not have specificity to d-lactaldehyde (Altaras & Cameron, 1999), which is consistent with what was observed in Table 1. In order to assess activity of AldA to d-lactaldehyde, crude extracts of wild type and aldA mutant were prepared and mixed with d-lactaldehyde in the presence of 1 mM NAD, which were analyzed by 1H-NMR (Fig. 3c). The results indicate that AldA is responsible for some level of d-lactaldehyde conversion to lactate (Fig. 3c), which is consistent with the previous observation that the purified lactaldehyde dehydrogenase exhibited about 5% activity to d-lactaldehyde relative to that of l-lactaldehyde (Sridhara & Wu, 1969).

Toxicity of dihydroxyacetone

Although the toxicity of dihydroxyacetone has been reported for several organisms, its effect on E. coli is still unclear. This study demonstrated that externally added dihydroxyacetone is toxic to E. coli cells. In order to assess toxic effects of dihydroxyacetone and HA relative to methylglyoxal, survival assay of E. coli cells was carried out with variable concentrations of the dihydroxyacetone, HA, and methylglyoxal. As described in ‘Materials and methods’ section, cells were plated onto LB agar with different concentrations of methylglyoxal/dihydroxyacetone/HA, and viable cells were counted after incubation at 37 °C for 15 h. Dihydroxyacetone exhibited lethality to E. coli cells with the LC10, LC50 and LC90 value of 14, 28 and 48 mM, respectively, as compared with 0.04, 0.4 and 0.7 mM of methylglyoxal, whereas no significant toxicity was observed for HA (LC10 value of 300 mM). Approximately 70-fold higher concentration of dihydroxyacetone as compared to methylglyoxal was thus required to kill 90% of cells. Consistent with the finding, dihydroxyacetone was shown to be toxic in yeast cells, and its detoxification requires functional dihydroxyacetone kinase (Molin et al., 2003). Furthermore, a decreased cell yield was reported for Zymomonas mobilis (Viikari & Korhola, 1986), and also for Gluconobacter oxydans (Boris et al., 1991), with increased concentrations of external dihydroxyacetone.

Dihydroxyacetone toxicity is due to a production of methylglyoxal

Since the toxic mechanism of dihydroxyacetone was not clearly established, the authors addressed this question by testing various mutant cells for their susceptibility to dihydroxyacetone. Although dihydroxyacetone was shown to react with proteins (Tessier et al., 2003) like methylglyoxal, the effect of dihydroxyacetone might be indirect since dihydroxyacetone can be converted to methylglyoxal nonenzymatically (Riddle & Lorenz, 1968; Inoue & Kimura, 1995). Alternatively, externally added dihydroxyacetone is converted to DHAP, which may lead to an enhanced methylglyoxal level. Indeed, the gloA mutant strain lacking glyoxalase I, the major pathway detoxifying methylglyoxal, displayed sensitivity to dihydroxyacetone. The mutant strain lacking gloA gene, encoding glyoxalase I, increases sensitivity to dihydroxyacetone at concentration of 10 mM (Fig. 4, upper panel), suggesting that the dihydroxyacetone toxicity was due to a metabolic conversion of dihydroxyacetone to a toxic aldehyde compound, e.g. methylglyoxal. On the other hand, it was observed that the gldA-deficient strain became sensitive to dihydroxyacetone (Fig. 4, lower panel), presumably due to an inability to remove intracellularly accumulated dihydroxyacetone. Furthermore, gloA mgsA double deficient mutant became more resistant to dihydroxyacetone than gloA single mutant (Fig. 4, upper panel). This observation again supports the earlier notion that the in vivo conversion of dihydroxyacetone to glycerol does occur. Thus, it seems likely that the toxicity of dihydroxyacetone in E. coli cells is due to an intracellular conversion to methylglyoxal. In contrast, yeast glo1 deletion strain displayed the same level of dihydroxyacetone sensitivity as wild type (Molin et al., 2003), suggesting a different mechanism of dihydroxyacetone toxicity in yeast.

In conclusion, GldA was shown to have broad substrate specificity, e.g. reduction of dihydroxyacetone to glycerol and HA to 1,2-PD, in addition to detoxification of methylglyoxal (Fig. 1). However, the primary role of gldA is likely to regulate an intracellular level of dihydroxyacetone by converting it into glycerol. This is further supported by the fact that the E. coli gldA gene lies immediately downstream of the talC gene that encodes fructose-6-P aldolase, overlapping 28 bp in its 3′ region (Schurmann & Sprenger, 2001). The gene encoding putative phosphotransferase system enzyme I (ptsA) is located upstream of the talC gene, thereby forming an operon with other two genes (Moreno-Hagelsieb & Collado-Vides, 2002; Huerta & Collado-Vides, 2003). TalC had been classified earlier as a transaldolase (Reizer et al., 1995) without experimental evidence. Schurmann & Sprenger (2001), however, demonstrated that E. coli TalC exhibited a lower fructose-6-phosphate aldolase activity than Fsa, which shares 68% amino acid identity with fructose-6-phosphate aldolase (Fsa). Thus, it is conceivable that these genes are metabolically associated, possibly with dihydroxyacetone. Since TalC was shown to have an aldolase activity generating dihydroxyacetone as one of its cleavage products (Schurmann & Sprenger, 2001), the primary role of GldA might be to regulate an intracellular level of dihydroxyacetone by converting it into glycerol. Further investigation on the metabolic regulation of dihydroxyacetone may unravel the physiological significance of this reactive metabolite.


This work was supported by the 21C Frontier Microbial Genomics and Application Center Program, Ministry of Science & Technology (Grant MG05-0202-2-0), Republic of Korea to C. Park. The authors thank S.C. Kim, Reid C. Johnson, and I.R. Booth for the strains.