Hsp31 of Escherichia coli K-12 is glyoxalase III


  • Krishna P. Subedi,

    1. Department of Biological Sciences, Korea Advanced Institute of Science and Technology, 335 Gwahangno, Yuseong-Gu, Daejon 305-701, Korea
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Dongwook Choi,

    1. Department of Biological Sciences, Korea Advanced Institute of Science and Technology, 335 Gwahangno, Yuseong-Gu, Daejon 305-701, Korea
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Insook Kim,

    1. Department of Biological Sciences, Korea Advanced Institute of Science and Technology, 335 Gwahangno, Yuseong-Gu, Daejon 305-701, Korea
    Search for more papers by this author
  • Bumchan Min,

    1. Analytical Science Center, Samyang Central R&D Center, 63-2, Hwaam-dong, Yuseong-gu, Daejeon 305-717, Korea
    Search for more papers by this author
  • Chankyu Park

    Corresponding author
    1. Department of Biological Sciences, Korea Advanced Institute of Science and Technology, 335 Gwahangno, Yuseong-Gu, Daejon 305-701, Korea
    Search for more papers by this author

E-mail ckpark@kaist.ac.kr; Tel. (+82) 42 350 2629; Fax (+82) 42 350 2610.


Hsp31 encoded by hchA is known as a heat-inducible molecular chaperone. Although structure studies revealed that Hsp31 has a putative catalytic triad consisting of Asp-214, His-186 and Cys-185, its enzymatic function, besides weak amino-peptidase activity, is still unknown. We found that Hsp31 displays glyoxalase activity that catalyses the conversion of methylglyoxal (MG) to d-lactate without an additional cofactor. The glyoxalase activity was completely abolished in the hchA-deficient strain, confirming the relationship between the hchA gene and its enzymatic activity in vivo. Hsp31 exhibits Michaelis–Menten kinetics for substrates MG with Km and kcat of 1.43 ± 0.12 mM and 156.9 ± 5.5 min−1 respectively. The highest glyoxalase activity was found at 35–40°C and pH of 6.0–8.0, and the activity was significantly inhibited by Cu2+, Fe3+ and Zn2+. Mutagenesis studies based on our evaluation of conserved catalytic residues revealed that the Cys-185 and Glu-77 were essential for catalysis, whereas His-186 was less crucial for enzymatic function, although it participates in the catalytic process. The stationary-phase Escherichia coli cells became more susceptible to MG when hchA was deleted, which was complemented by an expression of plasmid-encoded hchA. Furthermore, an accumulation of intracellular MG was observed in hchA-deficient strains.


Methylglyoxal (MG) is a highly reactive α-oxoaldehyde and physiological metabolite (Thornalley, 1990). MG is generated by the fragmentation and elimination of phosphate from dihydroxyacetone phosphate (DHAP) and glyceraldehyde-3-phosphate (GA3P) during glucose metabolism (Thornalley, 1996). MG is a toxic electrophile and reacts with cellular macromolecules such as proteins and nucleic acids (Ferguson et al., 1998). Since the high concentrations of MG can cause cell death, detoxification of MG is essential for cell viability and growth. There are several pathways for the cellular degradation of MG. The glyoxalase system, which consists of glyoxalase I and II, catalyses the conversion of MG to d-lactate with S-d-lactoylglutathione as an intermediate (Thornalley, 1990). The glyoxalase I/II system requires a reduced glutathione (GSH), whereas glyoxalase III catalyses the conversion of MG directly to d-lactate without the involvement of GSH (Misra et al., 1995). MG reductase also catalyses the conversion of MG into lactaldehyde, which is sequentially oxidized to l-lactate by aldehyde dehydrogenase (Misra et al., 1996). Furthermore, aldehyde compounds are reduced in cells by aldo-keto reductases, i.e. YafB, YeaE and YghZ proteins, producing acetol from MG (Ko et al., 2005).

A single enzyme, designated glyoxalase III, catalysing the GSH-independent conversion of MG into d-lactate has been reported in Escherichia coli (Misra et al., 1995; MacLean et al., 1998; Okado-Matsumoto and Fridovich, 2000; Benov et al., 2004). This enzyme was inactivated by thiol-blocking reagents and reactivated by thiol-containing agents, indicating the involvement of the thiol group for catalysis (Misra et al., 1995). The importance of thiol group was also supported by its activity reduction in the superoxide dismutase-deficient strain as well as its sensitivity to oxidizing agents, such as superoxide and hydrogen peroxide (Okado-Matsumoto and Fridovich, 2000). Glyoxalase III exhibits lower enzyme activity than the glyoxalase I/II system in vivo, whereas in vitro, the activity of glyoxalase III is higher than that of glyoxalase I/II (MacLean et al., 1998; Okado-Matsumoto and Fridovich, 2000). Interestingly, the activity of glyoxalase III is not elevated in E. coli lacking glyoxalase I (MacLean et al., 1998). The expression of glyoxalase III was dependent on the growth phase, i.e. higher at the stationary phase because of its dependence on rpoS (Benov et al., 2004). Although the glyoxalase III plays a certain role in the degradation of MG, the corresponding gene has not been reported.

As a homodimeric member of the DJ-1/ThiJ/PfpI superfamily, E. coli Hsp31, a hchA gene product, is known as a heat-inducible chaperone with weak aminopeptidase activity (Malki et al., 2005). It is also reported to play a role in the acid resistance of E. coli cells (Mujacic and Baneyx, 2007). The hchA is transcribed from dual promoters recognized by the sigma D and S subunits of RNA polymerase, which is repressed during the exponentially growing cells by H-NS (Yoshida et al., 1993), and the thermal induction of RNA polymerase relies primarily on the relief of the H-NS-mediated silencing of sigma d-dependent transcription (Mujacic and Baneyx, 2006). Additionally, increasing transcript stability was proposed as a mechanism underlying temperature-dependent control of hchA expression (Rasouly et al., 2007).

The crystal structure of Hsp31 indicates a homology with the PfpI family of putative intracellular proteases of archaebacteria and the ThiJ family of eubacteria (Lee et al., 2003). Furthermore, it has an evolutionarily conserved domain with human protein DJ-1, which is associated with early-onset Parkinson's disease (Bonifati et al., 2003). Three classes of Hsp31 orthologues, i.e. Classes I, II and III, have been described based on their sequence similarity (Quigley et al., 2003). The putative catalytic structure, consisting of Asp(Glu)–His–Cys, was found in the single subunit of the Classes I and II, and in the subunit interface of Class III (Quigley et al., 2003). Hsp31 has an alternative catalytic triad (His–Cys–Glu) consisting of His-186, Cys-185 and Glu-77 residues that partially overlap with the previous structure containing Asp-214 instead of Glu-77. (Lee et al., 2003; Quigley et al., 2003). Glu-77 residue, which is strictly conserved among all of the DJ-1/ThiJ/PfpI proteins (Wei et al., 2007), forms a hydrogen bond with Cys-185 (Zhao et al., 2003). Although each Hsp31 monomer was reported to contain a poorly accessible catalytic triad that is responsible for a broad specificity aminopeptidase activity (Malki et al., 2005), its biochemical function and physiological role have remained unknown.

We demonstrate here that E. coli Hsp31 functions as glyoxalase III, catalysing GSH-independent conversion of MG. The nucleophilic cysteine (Cys-185) and adjacent glutamic acid (Glu-77) residues are critical in enzyme catalysis, forming the core of the active site. Our findings provide evidence that there are multiple pathways for removing intracellular oxoaldehydes.


Proteomic characterization of E. coli glyoxalase III

Glyoxalase III was purified from E. coli K-12 strain deficient in the gloA gene through a series of steps as described in Experimental procedures. The procedure involved ammonium sulphate fractionation, heat treatment, gel filtration chromatography, anion exchange chromatography and hydrophobic interaction column chromatography with n-butyl sepharose (Table 1). The protein after final purification had near homogeneous purity (Fig. 1A); thus, the protein was directly used for analysing its biochemical property. Specific activity of purified protein was increased almost 25-fold over the initial crude extract. To identify a gene encoding glyoxalase III, the purified protein was used for peptide mass fingerprint analysis. The possibility of minor contaminants with the same molecular weight was excluded by two-dimensional electrophoresis (2-DE) analyses. The result of the mass analysis for the candidate protein matched with the amino acid sequence of E. coli Hsp31.

Table 1.  Purification of glyoxalase III.
ProcedureTotal protein (mg)Total activity (UI)aSpecific activityb (mUI mg−1)Purificationc (fold)Recoveryd (%)
  • a. 

    UI is defined as the amount of enzyme converting 1 µmol of MG per minute.

  • b. 

    Specific activity = UI per mg protein.

  • c. 

    Purification = Specific activity at given step/Specific activity of initial extract.

  • d. 

    Recovery = (Total activity at given step/Total activity of initial extract) × 100.

Crude extract1205.511.19.21.0100
First ammonium sulphate552.
Heat treatment429.14.911.51.244.4
Second ammonium sulphate353.94.512.71.440.5
First Superdex-200142.
Second Superdex-20063.92.843.94.825.3
Figure 1.

Purification and catalytic activity of endogenous glyoxalase III. A. Proteins separated on 12% SDS-PAGE: lane 1, molecular weight size marker; lane 2, wild-type MG1655; lane 3, MG1655 ΔgloAΔhchA (KS604); lane 4, BL21(DE3)/pET21b hchA (KS628); lane 5, purified glyoxalase III (after n-butyl sepharose purification); and lane 6, purified glyoxalase III. The glyoxalase band at approximately 40 kDa is shown with an asterisk. B and C. (B) NMR spectra indicating the production of lactate by glyoxalase. The reaction mixtures containing MG (3 mM) and the total soluble extracts of wild-type MG1655 and MG1655 ΔgloAΔhchA (KS604) strains (1 mg each) were analysed after incubations of 30 min, which were compared with that of purified Hsp31 (10 µg, C). The reaction was carried in 50 mM sodium phosphate buffer (pH 7.5), incubated for 30 min and analysed by NMR. The standard of methylglyoxal reagent contains some amount of acetate with a peak around 1.9 ppm.

In order to further confirm our finding, we generated a hchA-deficient strain from E. coli MG1655 (Table 2). To avoid glyoxalase activity contribution of glyoxalase I/II requiring GSH, the crude extract of this strain was extensively dialysed in sodium phosphate buffer containing 1 mM DTT, and subjected to a measurement of glyoxalase III activity. As a positive control, extract from the hchA+ΔgloA strain was used. The glyoxalase activity in the control strain was approximately 50 mUI per mg of protein, whereas it was not detected in the strain lacking hchA. This result was consistent with the observation that lactate, a product of the glyoxalase reaction, was undetected in the hchA-deficient strain (Fig. 1B). These data indicate that the hchA gene is responsible for glyoxalase III activity in E. coli, and the existence of other genes encoding glyoxalase is highly unlikely, at least in MG1655.

Table 2. E. coli strains used.
Strain/plasmidRelevant genotypeaReference
  • a. 

    Resistance markers: cat, chloramphenicol; kan, kanamycin.

MG1655F-, λ-Lab collection
BL21(DE3)F-ompT hsdSB (rB- mB-) gal dcm (DE3)Novagen
MJF388MJF274 ΔgloA::kanMacLean et al. (1998)
KS601MG1655 ΔgloA::kanThis work
KS603MG1655 ΔhchA::catThis work
KS604MG1655 ΔgloA::kanΔhchA::catThis work
KS618MG1655/pQE30This work
KS619MG1655/pQE30 hchAThis work
KS620KS603/pQE30This work
KS621KS603/pQE30 hchAThis work
KS622KS604/pQE30This work
KS623KS604/pQE30 hchAThis work
KS628BL21(DE3)/pET21b hchAThis work
pKS5pQE30 hchAThis work
pDO1pET21b hchAThis work

Enzymatic properties and catalytic residues of glyoxalase III

The purified recombinant Hsp31 protein was subjected to molecular and biochemical characterizations. The conversion of MG into lactate by the recombinant Hsp31 was measured by HPLC and NMR analyses (Fig. 1B). The HPLC chromatograms showed the reaction product, i.e. lactate, which was produced from the reaction of MG with purified Hsp31 (data not shown). The stoichiometry of the reaction catalysed by Hsp31 demonstrated an equimolar generation of lactic acid from MG (Fig. 2A). Neither the utilization of MG nor the formation of d-lactate was observed in the hchA mutant extract or any of the control reaction mixtures.

Figure 2.

Catalytic properties of Hsp31. A. Stoichiometry of a glyoxalase reaction. The amounts of MG left after a specified incubation time (closed triangle) correspond to that of d-lactate formed (open square). B. The enzyme activity was assayed at 37°C in the range of pH 4–10, as described in Experimental procedures. C. The activity was assayed at 4–50°C. D. The purified Hsp31 (10 µg), free of EDTA, was incubated in 100 µl of 25 mM sodium phosphate, pH 7.5, in the presence of different concentrations of metal ions (30 min at 4°C), and their activities were measured at 37°C with 1 µM MG. The control without an added cation was set to 100%.

The Hsp31 protein showed the enzymatic activity at pH range of 6.0–8.0 with a significant inhibition below pH 5.0 (Fig. 2B), which is in agreement with an earlier report (Misra et al., 1995), and at temperatures ranging from 30°C to 50°C with a maximum around 37°C (Fig. 2C). The glyoxalase activity of Hsp31 was significantly inhibited by Cu2+ and Zn2+, and enhanced by Fe2+, whereas Mg2+ and Ni2+ had little effect (Fig. 2D). The inhibition by Zn2+ was maximally observed in more than 0.3 mM, and the Cu2+ inhibition below 25 µM (data not shown). An extensive dialysis of Hsp31 with 10 mM EDTA did not significantly decrease the glyoxalase III activity of more than 30% (data not shown). These data indicate that, unlike glyoxalase I/II system, the glyoxalase III activity does not seem to require any cation or metal as cofactor. Thus, the effects of some metal ions might be due to a change in oxidation state of enzyme. The inhibition of glyoxalase III activity by Zn2+ was observed in a concentration-dependent manner with approximately 10% at 25 µM and more than 50% at 100 µM (data not shown). Although Zn2+ was suggested as a cofactor for Hsp31 in its peptidase activity (Malki et al., 2005), our results on glyoxalase III did not indicate Zn2+ as a requirement.

Our finding of Hsp31 as glyoxalase prompted us to re-evaluate the previously suggested role of Asp(Glu)–His–Cys triad in enzymatic catalysis. As a matter of fact, a mutation in Asp-214 did not completely abolish the glyoxalase activity (data not shown). Thus, we focused on residues in close proximity to Cys-185 and found a highly conserved residue Glu-77 in all the members of the Hsp31 family (Fig. 3A), leading us to generate mutants containing single amino acid substitutions of E77A, C185A and H186A. The distances between the putative catalytic residues are between 3.0 and 3.7 Å (Fig. 3A). The substrate saturation curve and the kcat, Km values of the purified Hsp31 proteins are shown (Fig. 3B). As in Fig. 3C, the E77A and C185A mutants almost completely abolished glyoxalase activity, indicating that Glu-77 and Cys-185 residues directly participate in catalysis. The H186A mutant shows approximately 17% remaining activity compared with that of the wild-type protein, suggesting that it is less crucial in catalysis.

Figure 3.

The catalytic residues of Hsp31 and enzyme kinetics for MG. A. A ribbon diagram of Hsp31 monomer [PDB code: 1N57 (Quigley et al., 2003)] is shown with the putative catalytic residues modelled in stick and coloured by atomic types: oxygen (red), nitrogen (blue) and sulphur (yellow). The distances between catalytic residues are presented in angstroms. B. Substrate saturation curves of enzyme reactions with MG in wild-type and mutant proteins. The kinetics of C185A and E77A were not determined because of their extremely low activities below the resolution of assay. C. Comparison of kinetic constants for MG in wild-type and mutant enzymes. The catalytic constant (kcat) and specificity constant (kcat/Km) are presented by the lengths of vertical bars: WT, Km (mM) = 1.43 ± 0.12; WT, kcat (m−1) = 156.90 ± 5.48; H186A, Km (mM) = 0.19 ± 0.02; H184A, kcat (m−1) = 5.71 ± 0.19. The double asterisk (**) represents low activity below the resolution of assay.

In vivo role of Hsp31 in protecting E. coli cells from MG toxicity

Using hchA- and/or gloA-deficient strains, we observed that Hsp31 efficiently detoxified exogenously added MG. The stationary-phase cells of hchA-deficient mutant were more sensitive to MG than that of hchA-proficient strain (Fig. 4A). The MG sensitivity was complemented by an in vivo expression of Hsp31 on plasmid (pQE30 hchA) that produces as much Hsp31 protein as the wild type containing a chromosomal copy (data not shown). In all the genetic backgrounds tested, cells harbouring the Hsp31 plasmid exhibited enhanced levels of MG detoxification, as shown in Fig. 4A, which is also apparent in the gloA-deficient background. To gain further insights into the role of hchA in detoxifying MG, we compared growths of strains in different concentrations of MG. The growths of the hchA-deficient strains were slower with increased concentrations of MG than that of wild type (Fig. 4B). The protective effect by endogenous hchA was more pronounced in the gloA-deficient background than in gloA-positive strain. The growth of gloA-deficient strain was suppressed by approximately 0.5 mM of MG concentration, while about 0.3 mM MG was sufficient to inhibit the growth of gloA hchA double mutant (Fig. 4B).

Figure 4.

Hsp31 protects E. coli cells against externally added MG. A. MG susceptibilities of gloA- and/or hchA-deficient strains and the complementation with a hchA plasmid were observed as described in Experimental procedures. Cells containing Hsp31 plasmids in different genetic backgrounds were grown overnight in LB broth, diluted (1/100-fold) with fresh media and grown to OD600 = 1.0. The cultured cell was then serially diluted and spotted (4 µl aliquot) onto LB media containing different concentrations of MG. Growth was observed after 12–15 h of incubation at 37°C. B. Concentration-dependent susceptibilities of MG in different mutant strains; the wild-type MG1655 (filled triangle), MG1655 ΔgloA (KS601, open triangle), MG1655 ΔhchA (KS603, open square) and MG1655 ΔgloAΔhchA (KS604, filled square).

Accumulation of intracellular MG in hchA-deficient cells

We expected that the hchA gene contributes to a removal of MG, thereby affecting intracellular level of MG. As described in Experimental procedures, MG levels were measured from samples after derivatization of the compound with o-phenylenediamine (o-PD) and analysis with HPLC. The total and free intracellular MG concentrations were measured in four different strains, including the wild type, ΔgloA, ΔhchA and ΔgloAΔhchA mutants. Figure 5A shows the HPLC chromatograms of total MG levels at two different growth stages (OD600 = 0.5 and 1.2) Strains lacking the hchA gene accumulated higher amounts of intracellular MG than the parent strains in both stages. As in Fig. 5B, the total and free intracellular MG concentrations were much higher in OD600 = 1.2 than in OD600 = 0.5 cells, which were more pronounced when Hsp31 was absent. Two- to threefold accumulation of intracellular MG was observed in hchA mutants relative to that of the wild type. A considerably high level of intracellular MG was observed in strains lacking glyoxalase I.

Figure 5.

Accumulation of intracellular MG in strains lacking hchA gene. A. HPLC chromatograms showing an accumulation of MG in wild type (WT, MG1655) and MG1655 ΔhchA (KS603) strains at two different growth stages (OD600 = 0.5 and 1.2). Equal numbers of cells were used in all cases, and the samples were treated and analysed equally. The strains and growth stages are as indicated with peaks of MG (▾) and an internal standard (5-MQ, ▿). B. Comparison of free and total intracellular MG concentrations in different strains of E. coli collected at different growth stages (OD600 = 0.5 and 1.2). An accumulation of MG, shown as the peak of 2-MQ, was measured by HPLC after derivatizing with o-PD, in which 5-MQ was used as an internal standard.


The E. coli Hsp31 belongs to the DJ-1/ThiJ/PfpI family (Wei et al., 2007) whose function is not well characterized. Hsp31 homologues are found in both eukaryotes and prokaryotes (Sastry et al., 2002; Quigley et al., 2003). Hsp31 was originally characterized as a heat-inducible chaperone, which shows a weak aminopeptidase activity (Sastry et al., 2002; Malki et al., 2003; 2005). Recently, it was reported that Hsp31 plays an important role in protecting E. coli from severe heat shock and starvation (Mujacic et al., 2004; Mujacic and Baneyx, 2006), as well as in acid resistance of stationary-phase cells (Mujacic and Baneyx, 2007). In the present study we demonstrated that Hsp31 catalyses the conversion of MG directly into lactic acid, thereby reducing cellular toxicity of MG. Furthermore, an accumulation of intracellular MG in hchA-deficient cells supports the notion that Hsp31 functions as glyoxalase, which protects cells against dicarbonyl stress.

Based on the increased sensitivity of the gloA-deficient E. coli strain to MG and the absence of other genes compensating for this activity (MacLean et al., 1998), it was concluded that the glutathione-dependent glyoxalase pathway, i.e. glyoxalase I/II, is the most important route for the in vivo detoxification of MG (Ferguson et al., 1998). The level of GSH that is necessary for the glyoxalase I/II system is variable depending on the availability of glucose in minimal medium, becoming lower in starved than in rich condition (Loewen, 1979). Thus, the glyoxalase III may play a critical role in conditions with limiting carbon source. Previous studies indicate that the activity of glyoxalase I/II is constitutive irrespective of growth phases, while the glyoxalase III activity is variable, reaching a maximum in the stationary phase (Benov et al., 2004). Unlike glyoxalase I/II, the expression of glyoxalase III was not affected upon changes in major nutrients of growth medium (Misra et al., 1995). The measurements of each glyoxalase activity revealed that the glyoxalase III was the most abundant enzyme in E. coli (Okado-Matsumoto and Fridovich, 2000). The reason for variability in glyoxalase III activity may be because the glyoxalase III is a stationary-phase enzyme, and its activity is regulated by rpoS (Benov et al., 2004). Thus, it is likely that Hsp31 plays an important role in protecting stationary-phase cells against carbonyl toxicity. In this situation, cells may need an additional route for the detoxification of dicarbonyls, presumably due to their inefficient metabolic activity. In the case of aldehyde and aldo-keto reductases, either NADPH or NADH is required as a cofactor. On the other hand, there is an evidence that the glyoxalase III protects E. coli cells against short-chain sugars, which are known to yield toxic carbonyls due to an oxidation by superoxide (Okado-Matsumoto and Fridovich, 2000).

Hsp31 has been known as a molecular chaperone, whose function is accomplished by binding substrate protein to its surface region called ‘bowl’ (Quigley et al., 2003). Further, it was revealed that Hsp31 undergoes a temperature-induced conformational change in its flexible regions, i.e. linker and a neighbouring loop, which are involved in an exposure of hydrophobic surface for binding a substrate protein (Sastry et al., 2004; 2009). In this study, we demonstrated that Hsp31 has a function of glyoxalase III in normal situation. However, it is possible that Hsp31 exhibits a chaperone activity, which may or may not be dependent on glyoxalase activity. An analogous situation would be the DegP protein, in which the protein undergoes a temperature-dependent switching from chaperone to protease (Krojer et al., 2008).

Previous studies on glyoxalase III suggested a role of the thiol group(s) in its catalysis, based on the fact that the enzyme activity was sensitive to thiol-blocking reagents (Misra et al., 1995). Our study shows that the enzyme itself has a catalytic cysteine. It is likely that the thiol group of Cys-185 serves as an electrophile to attack the aldehyde group of MG to trigger catalysis as in glyoxalase I that utilizes the cysteine thiol of GSH in its enzymatic action (Thornalley, 2003). The thiol group of Cys-185 residue appears to play a similar role to that of glutathione in the glyoxalase I/II system, in which the Cys-S-MG hemithioacetal is formed non-enzymatically prior to enzyme catalysis. Structure studies of Hsp31 revealed an existence of oxyanion hole, which may stabilize substrate during transition state, in the putative catalytic pocket of the protein (Quigley et al., 2003; Zhao et al., 2003). The oxyanion hole is formed by the backbone amino groups of His-186 and Gly-154. The Glu-77 residue located very close to the side-chain of Cys-185 may be involved in proton transfer as in the glutamate residue of glyoxalase I (Cameron et al., 1999), which appears to provide a catalytic base for proton, forming an enediolate intermediate. His-186 can serve as a base detaching a proton from water molecule, thereby facilitating the hydrolysis of thioester substrate. The positive charge of His-186 may be stabilized by the nearby Asp-214.

The DJ-1 superfamily consists of a number of proteins with diverse functions, even though they share basic sequence similarities (Wei et al., 2007). The family can be subdivided into several groups, i.e. PfpI, ThiJ and GATase, etc. (Wilson et al., 2003), while each group contains potential catalytic residues including conserved cysteine (Horvath and Grishin, 2001). The PfpI subgroup is the closest structural neighbour of Hsp31 (Quigley et al., 2003), which has an active site associated with proteolytic function utilizing the putative His–Cys–Glu catalytic triad. Hsp31 contains a pocket consisting of Asp-214, His-186 and Cys-185 (Asp–His–Cys), which is structurally similar to PfpI catalytic residues, but it did not exhibit an efficient proteolytic activity (Malki et al., 2005). The Hsp31 homologues, present in other species, possessing similar catalytic residues may also show glyoxalase III activity. The residues, Cys-185 and Glu-77, of Hsp31 are conserved in the DJ-1 orthologues of eukaryotes (Wei et al., 2007), increasing the likelihood of functional convergence.

Experimental procedures

Bacterial strains and plasmids

All strains used are derivatives of E. coli K-12. MG1655 was used as a wild type for the gene disruption and amplification of hchA. The disruption of the hchA gene was carried out by the one-step inactivation method using PCR (Datsenko and Wanner, 2000). The whole reading frame of hchA gene, from the initiation to stop codons (2 033 859–2 034 710), was replaced by the chloramphenicol resistance (cat) gene. The gloA-deficient strains were constructed from MJF388 by transferring the mutant allele by P1 transduction. The MJF388 strain is a gift from I. R. Booth, University of Aberdeen. The strains used in this study are listed in Table 2.

Purification and proteomic identification of glyoxalase III

The enzyme was purified by the method described earlier (Misra et al., 1995) with modifications. In brief, 6 l of saturated E. coli culture (KS601) grown in Luria–Bertani (LB) broth was centrifuged, resuspended in 25 mM sodium phosphate buffer (pH 7.5) containing 1 mM dithiothreitol (DTT) and 0.1 mM PMSF, and disrupted by sonication. All other methods were as described earlier (Misra et al., 1995) with the exception that Sephadex G-100 and DEAE Sephacell columns were replaced by Superdex-200 and Q-sepharose, respectively, and the final purification was carried out using n-butyl sepharose (GE Healthcare) instead of pHMB agarose. The high activity fractions obtained from the Q-sepharose purification were pooled and equilibrated in 50 mM sodium phosphate buffer (pH 7.0) containing 5% sucrose, 1 mM DTT and 1 M ammonium sulphate. The enzyme solution was then concentrated by ultrafiltration through a Centricon-30 filter (Millipore). A portion of concentrated enzyme solution was applied to a Hi-Trap Butyl FF column pre-equilibrated with the same buffer. The enzyme was adsorbed to the column, and the bound protein was eluted by continuously lowering the concentration of ammonium sulphate. The high activity fractions were pooled, equilibrated with 25 mM sodium phosphate buffer (pH 7.5) containing 1 mM DTT, and concentrated using Centricon-30. The enzyme was then frozen under liquid nitrogen and stored at −80°C until further use.

The hchA gene coding for a polypeptide with 283 amino acids was amplified by PCR using E. coli genomic DNA as a template. The gene was inserted downstream of the T7 promoter of the expression plasmid pET-21b (Novagen), and the plasmid was introduced into an E. coli BL21 (DE3) strain. The cells were grown to an OD600 of approximately 0.5 in LB media containing 0.1 mg ml−1 ampicillin, and the expression of Hsp31 was induced by 1 mM isopropyl-β-d-thiogalactoside (IPTG). After 4 h of induction, cells were harvested and resuspended in 20 mM sodium phosphate buffer, pH 7.5, containing 0.5 mM DTT. After sonication and subsequent centrifugation, the supernatant was loaded onto a Ni2+-NTA column (Qiagen), and the protein was eluted with 0–0.5 M imidazole gradient in 20 mM sodium phosphate, pH 7.5. In some cases, Hsp31 was further purified by Q-sepharose column chromatography (GE Healthcare). The protein was applied to the Q-sepharose column equilibrated with 20 mM sodium phosphate (pH 7.5) containing 1 mM DTT. The proteins were eluted with a linear gradient of 0–0.5 M NaCl in the same buffer. Fractions that contained Hsp31 were pooled, equilibrated with sodium phosphate buffer, pH 7.5, containing 1 mM DTT, and stored at −80°C. The same method was employed to purify mutant proteins described in this study. The proteins purified with n-butyl sepharose were separated by 2D IEF/SDS-PAGE. An immobilized dry-strip of pI 3–10 was used for the first dimension separation. After two-dimensional separation on SDS-PAGE, the gel was silver stained, and a spot with a Mr of ∼31 kDa was subjected to MALDI/TOF MASS fingerprint analysis. The gene corresponding to glyoxalase III was identified by comparing the peptide fingerprint with those of databases in the SWISS-PROT and NCBI databanks.

Mutagenesis of the catalytic residues

Putative catalytic residues of Hsp31 were altered to alanines by using PCR with mutagenic primers. The following primer sequences were used to construct the corresponding mutants: E77a (5′-CCA TCC GAT TGC AAC GTT GCT GCC G-3′ and 5′-CGG CAG CAA CGT TGC AAT CGG ATG G-3′), C185a (5′-GTT ATC TCC CTT GCC CAC GGC CCG G-3′ and 5′-CCG GGC CGT GGG CAA GGG AGA TAA C-3′), and h186a (5′-ATC TCC CTT TGC GCC GGC CCG GCG G-3′ and 5′-CCG CCG GGC CGG CGC AAA GGG AGA T-3′). Using the restriction enzymes NdeI and XhoI, the mutant gene was subcloned into the expression vector pET21b.

Enzymatic characterization of Hsp31

Methylglyoxal was determined either colorimetrically by using the 2,4-dinitrophenylhydrazine-alkali (DNPH) reaction (Cooper, 1975) or by HPLC (Chaplen et al., 1996; 1998). The glyoxalase III was assayed by monitoring the utilization of MG or lactate formation. Unless otherwise stated, the assay mixture contained 50 µmol of sodium phosphate buffer, pH 7.5, 1 µmol of MG, and the enzyme, in a total volume of 1 ml. After a certain period of incubation, a 50 µl sample was transferred to DNPH solution, incubated for 15 min and measured (A550) by adding an alkali. For lactate measurement, the Simadzu HPLC system with the Prevail C18 column (Alltech) was used. The operating conditions for HPLC determination of lactic acid were as described earlier (Bai et al., 2000). In brief, 0.01 mol l−1 phosphoric acid solution (pH 2.5) was used as a mobile phase with a flow rate of 1.0 ml min−1, followed by UV detection at 210 nm at room temperature. A sample volume of 20 µl for d-lactate or the reaction product was used. The retention time of d-lactate was around 4.6 min. The quantification was made by running different concentrations of d-lactate (Sigma). One unit of glyoxalase III was defined as the amount of enzyme required to utilize 1 µmol of MG or to form 1 µmol of lactate min−1. In all reactions, non-enzymatic rates were subtracted from the observed initial reaction rates. The generation of lactate as a reaction product was further confirmed by analysing the reaction product with NMR (nuclear magnetic resonance) spectroscopy as described earlier (Kim et al., 2004). Kinetic constants for the wild-type and mutant enzymes were determined using the condition for determining enzyme activity, except that substrate concentrations were varied. Each data points (initial velocities) were obtained in triplicate, and at least eight different substrate concentrations were examined. Kinetic constants were determined by fitting directly to the hyperbolic form of the Michaelis–Menten equation with unweighted least-square analysis or Lineweaver–Burk plots using the Sigma-plot program (SPSS).

Measuring MG susceptibility

In order to measure the MG susceptibility of the hchA-deficient strain, fresh colonies of wild-type and mutant strains were grown overnight in LB broth, diluted 100-fold in the same medium and incubated for 15 h with shaking. Cultures of OD600 = 1.0 were serially diluted from 101 to 106, which were spotted (4 µl) onto LB plates containing different concentrations of MG, and growths were examined after 12–14 h of incubation at 37°C. For the strains overexpressing Hsp31, fresh colonies of indicated strains were grown overnight in LB broth, diluted 100-fold in the same medium and incubated with shaking until their OD600 reached to 1.0. The cells were diluted from 101 to 106, which were spotted (4 µl) onto LB plates containing different concentrations of MG. Growths were observed after 12–14 h of incubation at 37°C. In order to measure susceptibility to MG at different concentrations, cells were grown overnight in LB medium and diluted to OD600 = 0.05 in M9CA medium containing varied concentrations of MG. Cell growths were monitored at 600 nm after 10 h of incubation at 37°C with shaking.

Measurement of intracellular amount of MG

A minor modification was made to the method used earlier for measuring an amount of intracellular MG in animal cells with HPLC (Chaplen et al., 1996; 1998). Briefly, cells grown overnight in LB medium were diluted 100-fold in M9-glucose medium and incubated at 37°C to reach OD600 of 0.5 and 1.2. Calculated volumes of cells from two growth stages were centrifuged and washed immediately in ice-cold PBS. The cells were resuspended in ice-cold PBS, and an aliquot of cell suspension was removed for measuring OD at 600 nm to determine the cell number. Subsequently, samples with an equal number of cells from different strains were suspended in 20 ml volumes, which were lysed over ice by sonication. The degree of lysis and the remaining cell numbers were further verified by measuring the protein concentrations of each sample. The sample was then divided into 10 ml each and used for measurements of free and total MG concentrations.

The assay conditions for free MG were as described earlier (Chaplen et al., 1996). Perchloric acid (PCA; Aldrich Chemical) was added to a final concentration of 0.45 M, which was incubated on ice for 10 min and centrifuged at 20 000 g for 10 min. The samples were derivatized at 20°C for 4 h with 500 nmol of o-PD (Aldrich Chemical) and loaded with an internal standard, 2.5 nmol of 5-methylquinoxaline (5-MQ, Aldrich Chemical). Solid-phase extraction was performed with a C18 SPE cartridge (Alltech), pre-activated with 8 ml of acetonitrile, followed by 8 ml of 10 mM KH2PO4 (pH 2.5). The cartridge was rinsed with 5 ml of 10 mM KH2PO4, pH 2.5, and the retentate was eluted with 3 ml of acetonitrile. Elutes were filtered through a 0.2-µm-pore-size PVDF membrane (Millipore). The quinoxaline derivative of MG (2-MQ) was detected with an internal standard (5-MQ) at 315 nm by HPLC (Simadzu). The mobile phase was an 82:18 (v/v) solution of 10 mM KH2PO4 (pH 2.5) and HPLC-grade acetonitrile with a flow rate of 1.0 ml min−1, and 20 µl of this solution was injected into HPLC. Here, the retention times of 2-MQ and 5-MQ were approximately 17 and 28 min respectively. Quantification was performed by using standard curve of different concentrations of MG, which were treated and analysed in the same manner. The total MG concentration was measured as described earlier (Chaplen et al., 1998). Cellular extracts were treated with PCA to a final concentration of 0.45 M, and 500 nmol of o-PD and 2.5 nmol of 5-MQ were added as an internal standard and derivatizing agent respectively. The samples were derivatized at 20°C for 24 h before centrifugation (20 000 g, 10 min) and removal of PCA precipitate, which was followed by concentration and analysis of 2-MQ content by HPLC as described above. To estimate MG formation from nucleic acid degradation, DNAs obtained from an equal number of cells were treated with 0.5 M PCA and quantified in triplicate.


The authors thank Ian Booth for generously providing E. coli strains. This work was supported by the 21C Frontier Microbial Genomics and Application Center Program, Ministry of Education, Science and Technology, Republic of Korea to C. Park.