Human Glx2 contains an Fe(II)Zn(II) center but is active as a mononuclear Zn(II) enzyme

Abstract

Human glyoxalase II (Glx2) was over-expressed in rich medium and in minimal medium containing zinc, iron, or cobalt, and the resulting Glx2 analogs were characterized using metal analyses, steady-state and pre-steady state kinetics, and NMR and EPR spectroscopies in order to determine the nature of the metal center in the enzyme. Recombinant human Glx2 tightly binds nearly one equivalents each of Zn(II) and Fe. In contrast to previous reports, this study demonstrates that an analog containing two equivalents of Zn(II) cannot be prepared. EPR studies suggest that most of the iron in recombinant Glx2 is Fe(II). NMR studies show that Fe(II) binds to the consensus Zn₂ site in Glx2 and that this site can also bind Co(II) and Ni(II), suggesting that Zn(II) binds to the consensus Zn₁ site. The NMR studies also reveal the presence of a dinuclear Co(II) center in Co(II)-substituted Glx2. Steady-state and pre-steady state kinetic studies show that Glx2 containing only one equivalent of Zn(II) is catalytically-active and that the metal ion in the consensus Zn₂ site has little effect on catalytic activity. Taken together, these studies suggest that Glx2 contains a Fe(II)Zn(II) center in vivo, but that the catalytic activity is due to Zn(II) in the Zn₁ site.

The glyoxalase system consists of two enzymes, lactoylglutathione lyase (glyoxalase I, Glx1) and hydroxyacylglutathione hydrolase (glyoxalase II, Glx2) (1–3). Glx1 is capable of forming S-(2-hydroxyacyl)glutathione (SLG), which is made from the thiohemiacetal produced from a spontaneous reaction of methylglyoxal and glutathione. SLG (and other related glutathione thiolesters) is then hydrolyzed by Glx2 to form D-lactate and glutathione. Glyoxalase I can utilize a number of α-ketoaldehydes; however, the primary physiological substrate of the system is thought to be methylglyoxal (MG), a cytotoxic and mutagenic compound that is formed primarily as a byproduct of carbohydrate and lipid metabolism and from triose phosphates (2, 4–6). SLG is also cytotoxic because of its ability to inhibit DNA synthesis (2, 7). While SLG can also be metabolized by γ-glutamyltransferase and dipeptidase, these processes generate N-D-lactoylcysteine, which also inhibits nucleotide synthesis (7). Therefore, the glyoxalase system, which depletes MG and SLG, plays a critical role in cellular detoxification (1, 8).

Because of its role in cellular detoxification, the glyoxalase system has received considerable attention as a possible antitumor and antiparasitic target in animal systems (1, 9–18). Increased levels of Glx1 and Glx2 mRNA and protein have been detected in tumor cells, such as in breast carcinoma cells, and glyoxalase inhibitors have been shown to inhibit the growth of tumor cells in vitro (19). Therefore, it has been proposed that the targeted inhibition of glyoxalase enzymes can be a viable anti-cancer strategy (5, 11, 13, 18, 20–26). Plasmodium falciparum and the protozoan Leishmania exhibit high rates of methylglyoxal formation and increased levels of Glx1 activity (27, 28). In Leishmania infantum, trypanothione is used instead of glutathione, and a crystal structure of L. infantum Glx2 bound to S-D-lactoyltrypanothione has recently been reported (29).

Alterations in glyoxalase activity have also been associated with several other disease states. Glx1 and 2 can inhibit the formation of hyperglycemia-induced advanced glycation end products, suggesting that these enzymes may have a role in diabetic microangiopathy (30). Glyoxalase enzymes may also play a role in the pathogenesis of Alzheimer’s disease (31, 32). Finally, Glx2 has been identified as a target of p63 and p73, and suggested to be a pro-survival factor of the p53 family of transcription factors (33).

Glx2 has been purified and biochemically characterized from many sources, such as plants, mammalian liver, Samonella, and E. coli (34–39). Cameron et al. reported the crystal structure of human Glx2 (40), which defined an overall structure and showed the presence of a dinuclear zinc active site similar to those in the enzymes of the metallo-β-lactamase superfamily (41, 42). The structure showed two domains: a four-layered α sandwich similar to that seen in metallo-β-lactamases and a predominately α-helical domain (40). As with other metallo-β-lactamase family enzymes, the metal ion in the Zn₁ site was coordinated by His54, His56, His110, bridging Asp134, and a bridging hydroxide. The metal ion in the Zn₂ site was coordinated by His59, His173, Asp58, the bridging Asp134, the bridging hydroxide, and a terminally-bound solvent molecule (Figure 1). Although the protein used for crystallography contained ~1.5 moles of zinc and 0.7 moles of iron per mole of protein, the authors concluded that human Glx2 contains a dinuclear Zn(II) active site. The issue of iron binding was not considered but this omission raises questions concerning the actual metal binding preference of human Glx2.

Figure 1.

Proposed active site of human Glx2 (40). The small spheres are solvent molecules, and the large spheres are Zn(II) ions. Figure was rendered using Raswin v. 2.7.2.2 (60) and coordinate file 1qh5.

In this manuscript we present the results of biochemical and spectroscopic studies on human Glx2. Recombinant human Glx2 was over-expressed in the presence of different combinations of zinc, iron and cobalt, and the resulting enzymes were then purified to homogeneity. Steady state kinetic studies were used to determine the catalytic properties of the purified Glx2 analogs. ICP-AES was used to determine the metal content of the purified enzymes, and nuclear magnetic resonance (NMR) and electron paramagnetic resonance spectroscopies (EPR) were used to probe the dinuclear metal centers. These biochemical and spectroscopic results provide detailed structural information on the human Glx2 metal center and insights concerning the structure and kinetic mechanism of the enzyme that may ultimately be used to design inhibitors with potential therapeutic value.

Experimental Procedures

Over-expression and purification of human Glx2

PCR was conducted on a plasmid, which contains the gene for Glx2 from H. sapiens, which was kindly provided by Dr. Bengt Mannervik, using the primers CCTCCATGGTAAAAATCGAACTGGTGC and GAGTCGACTCGAGCTCTAGATCTTTTTTTTTT that generated NdeI and HindIII restriction sites at the 5′ and 3′ ends of the glx2 gene. The PCR fragment was subcloned into pET26b using the NdeI and HindIII restriction sites, and the sequence of the resulting pGlx2/pET26b plasmid was confirmed by DNA sequencing. The plasmid pGlx2/pET26b was transformed into E. coli BL21(DE3) Rosetta cells, and small scale cultures were used to maximize the recovery of soluble protein at different temperatures (15 °C, 22 °C, 30 °C, and 37 °C). A large scale over-expression of human Glx2 was performed as follows. A 10 mL overnight culture of E. coli BL21(DE3) Rosetta cells containing pGlx2/pET26b was used to inoculate 1L of LB (Luria Bertani) medium containing 25 μg/mL kanamycin and 25 μg/mL chloramphenicol. The cells were allowed to grow at 37 °C with shaking until they reached an optical density at 600 nm of 0.6–0.8. Protein production was induced by making the cultures 0.5 mM in isopropyl-β-D-thiogalactopyranoside (IPTG), and the cells were shaken at 22 °C for 24 h. The cells were collected by centrifugation (15 min at 7000 ×g), and the cell pellets were stored at −80 °C until further use.

The cell pellet was thawed and re-suspended in 15 mL of 10 mM MOPS, pH 6.5, containing 0.1 μM phenylmethylsulphonyl fluoride (PMSF). The cells were French pressed four times at 16,000 psi and centrifuged for 30 min at 15,000 ×g and 4 °C. The supernatant was dialyzed overnight at 4 °C versus 2L of 10 mM MOPS, pH 6.5. The dialyzed crude protein sample was centrifuged at 15,000 ×g and was subjected to FPLC using a SP-Sepharose column (1.5×12 cm with a 25 mL bed volume) that was equilibrated with 10 mM MOPS, pH 6.5. Bound proteins were eluted with a 0–500 mM NaCl gradient in 10 mM MOPS, pH 6.5, at 2 mL/min. Fractions containing human Glx2 were identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), pooled, and concentrated by using an Amicon ultrafiltration cell equipped with a YM-10 membrane. Enzyme concentrations were determined by measuring the absorbance at 280 nm and using a molar extinction coefficient of 23,080 M⁻¹cm⁻¹ (40).

Human Glx2 was over-expressed in minimum medium consisting of 2.5 g glucose, 5 g casamino acids, 5.5 g KH₂PO₄, 10.8 g K₂HPO₄, 1 g ammonium sulfate, and 10 g NaCl per 1 L of distilled H₂O in the presence of 100 μM Zn(II), Fe(II), Mn(II), or Co(II) to evaluate its metal binding preference. The resulting enzyme samples were purified as described above.

Metal analyses

The metal content of Glx2 samples was determined using a Varian-Liberty 150 Inductively Coupled Plasma spectrometer with atomic emission spectroscopy detection (ICP-AES), as described previously (43). Protein samples were diluted to 10 μM with 10 mM MOPS, pH 6.5, prior to analysis. A calibration curve with four standards and a correlation coefficient of greater than 0.99 was generated using Fe, Zn(II), Mn, and Co reference solutions. The following emission wavelengths were chosen to ensure the lowest detection limits possible: Fe, 259.940 nm; Zn, 213.856 nm; Mn, 257.610 nm; and Co 238.892nm.

To further evaluate metal binding to Glx2, a 3-fold molar excess of Fe(NH₄)₂(SO₄)₂, Zn(SO₄)₂, or Fe(NH₄)₂(SO₄)₂ + Zn(SO₄)₂ was added directly to purified as-isolated human Glx2, and the mixtures were allowed to incubate on ice for 1 hour. Unbound metal ions were removed by 4 × 1L dialysis steps against 10 mM MOPS, pH 6.5, at 4 °C (12 h for each step). The metal content of these protein samples was determined using ICP-AES as described above.

Steady state kinetics

Steady state kinetic parameters (K_m, k_cat) of human Glx2 were determined using S-D-lactoylglutathione (SLG) as a substrate. Thioester hydrolysis was monitored at 240 nm over 30 s at 25 °C as previously reported (43). The concentration of Glx2 analogs was typically 1–10 nM, and substrate concentrations used were 30 – 600 μM. The buffer used in the steady-state kinetic studies was 10 mM MOPS (pH 6.5), containing either no added metals, 100 μM ZnCl₂, or 100 μM Fe(NH₄)₂(SO₄)₂.

Stopped-flow kinetic studies

Stopped-flow UV-Vis studies were conducted on an Applied Photophysics SX.18-MVR stopped flow spectrophotometer at 2 °C. The reaction of Glx2 analogs (final concentration of 32.5 μM) and SLG (final concentration of 60.5 μM) was monitored at 240 nm for 200 milliseconds. Stopped-flow absorbance data were converted to concentration data using the SLG extinction coefficient (−3,100 M⁻¹cm⁻¹). All reactions were conducted in triplicate, and reaction rates were determined by fitting the progress curves to a first order exponential equation.

Spectroscopic studies

¹H NMR spectra were collected on a Bruker Avance 500 spectrometer operating at 500.13 MHz, 298 K, magnetic field of 11.7 T, recycle delay (AQ) of 41 ms, and sweep width of 400 ppm. Chemical shifts were referenced by assigning the H₂O signal the value of 4.70 ppm. A modified presaturation pulse sequence (zgpr) was used to suppress the proton signals originating from solvent and amino acids not coupled to the metal center. Line broadening of 50 Hz was used for all of the spectra. The protein concentration was ~ 1 mM, and 10% D₂O was included in samples for locking.

EPR spectra were recorded using a Bruker E600 EleXsys spectrometer equipped with an Oxford Instruments ESR900 helium flow cryostat and ITC503 temperature controller, and an ER4116DM cavity operating at 9.63 GHz in perpendicular mode. Other recording parameters are given in the figure legend. Quantitation of Fe(III) signals was carried out by double integration of spectra recorded at non-saturating power (2 mW) at 12 K. A 2 mM Cu(II)-EDTA standard in HEPES buffer, pH 7.5 recorded at 60 K, 50 μW was used. Integration limits and correction factors for S = ½, and S = ⁵/₂ signals, where D is assumed to be small compared to temperature, were employed (44–46). Co(II) signals were quantified by double integration, with reference to a frozen aqueous reference sample containing 2 mM Co(II), 50 mM imidazole, and 10 % by volume glycerol, recorded at 12 K, 0.8 mW. EPR simulations were performed using XSophe (Bruker Biospin), assuming S = ³/₂ and |D| ≫ hν.

Results

Over-expression and purification of human Glx2

High levels of soluble human Glx2 were produced in E. coli BL21(DE3) Rosetta cells grown in LB medium at 22 °C. The colorless recombinant protein eluted from SP-Sepharose chromatography at 100 mM NaCl in 10 mM MOPS, pH 6.5. Approximately 20 mg of purified Glx2 per liter of culture were obtained using this method. Human Glx2, over-expressed in LB medium (Glx2-LB), was shown to bind 0.4 ± 0.1 equivalents of Zn(II) and 0.5 ± 0.1 equivalents of Fe (Table 1). After incubation with 1.5 equivalents of Zn(II) and 1.5 equivalents of Fe(II), followed by exhaustive dialysis, human Glx2 (Glx2-LB+1.5Zn+1.5Fe) was shown to bind 1.2 ± 0.1 equivalents of Zn(II) and 0.9 ± 0.1 equivalents of Fe (Table 1).

Table 1.

Metal content and steady-state kinetic constants for human Glx2 analogs

Enzyme	Zn(II) (eq)	Fe (eq)	Co or Mn (eq)	kcat (s−1)	Km (μM)
Glx2-LB	0.4 ± 0.1	0.5 ± 0.1	ND (Mn)	570 ± 99	660 ± 190
Glx2-LB+1.5Zn+1.5Fe	1.2 ± 0.1	0.9 ± 0.1	ND (Mn)	740 ± 40	780 ± 68
Glx2-Znmin	1.1 ± 0.2	ND	ND (Mn)	407 ± 13	81 ± 11
Glx2-Znmin + 3 equivalents Zn(II)	0.9 ± 0.2	ND	ND (Mn)	262 ± 24	53 ± 22
Glx2-Znmin + 3 equivalents Fe(II)	1.0 ± 0.1	0.7 ± 0.1	ND (Mn)	281 ± 28	81 ± 30
Glx2-Znmin in 100 μM Fe buffer	N/A	N/A	N/A	355 ± 24	105 ± 24
Glx2-Znmin in 100 μM Zn(II) buffer	N/A	N/A	N/A	384 ± 9	109 ± 8
Glx2-Femin + Zn(II)	0.5 ± 0.2	0.6 ± 0.2	ND (Mn)	240 ± 5	256 ± 19
Glx2-Comin	0.1 ± 0.1	0.2 ± 0.1	1.0 ± 0.1 Co ND (Mn)	815 ± 36	110 ± 17
Glx2-Comin + 1 eq. Zn(II)	1.0	0.2 ± 0.1	1.0 ± 0.1 Co	565 ± 30	65 ± 13
Glx2-Znmin + 1 eq. Ni	0.1 ± 0.1	0.2 ± 0.1	1.0 Ni	439 ± 6	91 ± 9

ND – none detected

N/A – not applicable

Metal binding to human Glx2

Metal binding to human Glx2 was further evaluated by expressing the enzyme in minimum medium containing various metal ions, and the metal content of the resulting enzymes was analyzed by ICP-AES (Table 1). Glx2, over-expressed in the presence of 100 μM ZnCl₂ (Glx2-Zn_min) bound 1.1 ± 0.2 equivalents of Zn(II) and < 0.005 equivalents of Fe or Mn. When 3 equivalents of Zn(II) were added to this enzyme, followed by exhaustive dialysis, the resulting enzyme (Glx2-Zn_min + 3 equivalents Zn(II)) was shown to bind 0.9 ± 0.2 equivalents of Zn(II) and < 0.005 equivalents of Fe or Mn. When 3 equivalents of Fe(II) were added to Glx2-Zn_min followed by dialysis, the resulting enzyme (Glx2-Zn_min + 3 equivalents Fe(II)) was shown to bind 1.0 ± 0.1 equivalents of Zn(II) and 0.7 ± 0.1 equivalents of Fe.

Over-expression of human Glx2 in minimum medium containing 100 μM Fe(NH₄)₂(SO₄)₂ or 100 μM MnCl₂ did not result in appreciable amounts of enzyme at any temperature. However when 5 μM ZnCl₂ was added to the minimal medium containing Fe(II), human Glx2 was over-expressed, and the purified enzyme (Glx2-Fe_min + Zn(II) in Table 1) was found to contain 0.5 ± 0.2 equivalents of Zn(II) and 0.6 ± 0.2 equivalents of Fe. This result suggests that human Glx2 needs Zn(II) to be over-expressed and that the enzyme may bind mole each of iron and zinc per mole enzyme. We tested this hypothesis by incubating Glx2-LB with 1.5 equivalents of Zn(II) and Fe(II), and the resulting enzyme (Glx2-LB+1.5Zn+1.5Fe) was exhaustively dialyzed. Consistent with our hypothesis, Glx2-LB+1.5Zn+1.5Fe was shown to bind 1.2 ± 0.1 equivalents of Zn(II) and 0.9 ± 0.1 equivalents of Fe.

Since none of the human Glx2 samples contained significant amount of Mn, we did not determine the effect of adding Zn(II) to minimal medium containing Mn(II) on Glx2 over-expression. We did however find that human GLX2 does bind cobalt. When the enzyme was over-expressed in minimal medium containing 100 μM CoCl₂, Glx2-Co_min bound 0.1 ± 0.1 equivalents of Zn(II), 0.2 ± 0.1 equivalents of Fe, and 1.0 ± 0.1 equivalents of Co. Therefore, human Glx2 preferentially binds one mole each of Fe and Zn(II) per mole enzyme and can bind one mole of Co, but it does not bind Mn.

Steady state kinetics

Steady-state kinetic studies were performed on the different forms of human Glx2 in order to evaluate the effect of different metal ions on the catalytic activity of the enzyme. Glx2-LB, which contained 0.4 ± 0.1 equivalents of Zn(II) and 0.5 ± 0.1 equivalents of Fe, exhibited a k_cat of 570 ± 99 s⁻¹ and a K_m of 660 ± 190 μM (Table 1). Interestingly, Glx2-LB+1.5Zn+1.5Fe, which contained 1.2 ± 0.1 equivalents of Zn(II) and 0.9 ± 0.1 equivalents of Fe, exhibited relatively similar values (k_cat of 740 ± 40 s⁻¹ and K_m of 780 ± 68 μM, Table 1). Glx2-Zn_min exhibited a k_cat of 407 ± 13 s⁻¹ and a K_m of 81 ± 11 μM, while purified Glx2-Zn_min, which was incubated with 3 equivalents or Zn(II) or Fe(II) to increase the Zn(II) or Fe(II) content, exhibited k_cat values of 262 ± 24 and 281 ± 28 s⁻¹ and K_m values of 53 ± 22 and 81 ± 30 μM, respectively. Inclusion of 100 μM Fe(II) or Zn(II) in the assay buffer for Glx2-Zn_min in an attempt to further load the enzyme with metal, did not significantly affect the kinetic constants (Table 1). These results suggest that human Glx2 requires only one equivalent of Zn(II) for activity. However, Glx2-Co_min, was also shown to be active with SLG (k_cat of 815 ± 36 s⁻¹ and a K_m of 110 ±17 μM). The direct addition of 1 equivalent of Co(II) to Glx2-Zn_min or Glx2-Co_min resulted in the formation of a yellow precipitate, most likely due to oxidation of Co(II) to Co(III).

Stopped-flow kinetic studies

To further probe the kinetic behavior of the different human Glx2 analogs, stopped-flow kinetic studies were conducted in which the disappearance of substrate SLG was monitored over time (Figure 2). Each reaction contained 33 μM SLG and 60 μM Glx2 analog. Glx2-LB, which contained 0.4 equivalents of Zn(II) and 0.5 equivalents of Fe, exhibited a rate of 30 ± 1 s⁻¹. Glx2-Zn_min, Glx2-Zn_min + 3 equivalents Fe(II), Glx2-Zn_min + 3 equivalents of Zn(II), and Glx2-LB+1.5Zn+1.5Fe analogs exhibited essentially the same rate of 49 ± 3 s⁻¹. The Glx2-Fe_min + Zn(II) analog, which contains 0.5 and 0.6 equivalents of Zn(II) and Fe, respectively, exhibited the slowest rate of 15 ± 1 s⁻¹. On the other hand, Glx2-Co_min, which contains 0.1 equivalents of Zn(II), 0.2 equivalents of Fe, and 1.0 equivalents of Co, exhibited the fastest rate of 139 ± 2 s⁻¹.

Figure 2.

Stopped-flow kinetic studies of the reaction of human Glx2 analogs with SLG at 2 °C. The concentration of Glx2 analogs was 32.5 μM, and the concentration for SLG was 60.5 μM. The progress curves shown are: (hexagons) Glx2-Fe_min + Zn(II), (□) Glx2-LB, (●) Glx2-Zn_min, (◆) Glx2-Zn_min + 3 equivalents Zn(II), (◇) Glx2-Zn_min + 3 equivalents Fe(II), (○) Glx2-LB+1.5Zn+1.5Fe, and (unfilled triangles) Glx2-Co_min. The progress curves for Glx2-Zn_min, Glx2-Zn_min + 3 equivalents Zn(II), Glx2-Zn_min + 3 equivalents Fe(II), and Glx2-LB+1.5Zn+1.5Fe were nearly superimposable.

Spectroscopic studies

¹H NMR spectroscopy was utilized to probe the metal binding sites of human Glx2. A ¹H NMR spectrum of Glx2 containing 1.2 ± 0.1 equivalents of Zn(II) and 0.9 ± 0.1 equivalents of Fe (Glx2-LB+1.5Zn+1.5Fe) showed two solvent exchangeable peaks at 47 ppm and 71 ppm (Figure 3(A)). Given the linewidths of these signals and the fact that the predicted Zn₂ site has two histidines while the Zn₁ site has 3 histidines (40), we predict that there is a Fe(II) bound to the Zn₂ site in this sample. Nonetheless, we cannot completely discount the possibility that Fe(II) is binding to a histidine from each site or the possibility that Fe(II) is binding to the Zn₁ site and one of the histidine NH protons is in fast exchange with solvent.

Figure 3.

¹H NMR spectra of human Glx2 analogs in 10 mM MOPS, pH 6.5, containing 10% D₂O: (A) Glx2-LB+1.5Zn+1.5Fe, (B) Glx2-Co_min, (C) Glx2-Co_min with 1 equivalents of Zn(II) added, and (D) ZnNi-Glx2. The enzyme concentrations in these samples were ~ 1 mM. The * represents peaks that were solvent-exchangeable. The peak at 42 ppm in the spectrum of Glx2-Co_min decreased by 1/2 when the sample was exchanged in 90% D₂O.

The ¹H NMR spectrum of Glx2 containing 1.0 ± 0.1 equivalent of Co(II) (Table 1), showed 5 paramagnetically-shifted resonances (Figure 3(B)). The peaks at 55, 71, and 87 ppm completely disappear and the peak at 42 ppm decreased by 1/2 when the sample was exchanged in 90% D₂O. This result indicates that there are at least 4 solvent exchangeable peaks in the Glx2-Co_min sample. The peak at 87 ppm (spectrum (B) in Figure 3) is larger than the other peaks, suggesting that this peak may be due to 2 protons both of which are solvent-exchangeable. The NMR spectra of Glx2-Co_min suggests that Co(II) binds to 4 or 5 histidines and, therefore, that Co(II) binds to both the Zn₁ and Zn₂ sites in human Glx2.

To probe whether Zn(II) could displace Co(II) from one of the metal binding sites, 1 equivalent of Zn(II) was added to Glx2-Co_min to generate Glx2-Co_min + 1 equivalent Zn(II). The resulting analog exhibited a k_cat of 565 ± 30 s⁻¹ and a K_m of 65 ± 13 μM, when using SLG as the substrate (Table 1). The NMR spectrum of this analog showed two solvent exchangeable peaks at 47 and 71 ppm (Figure 3(C)), which are the same positions as the solvent exchangeable peaks for the Glx2-LB+1.5Zn+1.5Fe sample (Figure 3(A)). This result suggests that Glx2 preferentially binds Zn(II) over Co(II) in the Zn₁ site, but not in the Zn₂ site.

To test whether the Zn₂ site can bind other metal ions in addition to Co(II) and Fe(II), we added 1 equivalent of Ni(II) to Glx2-Zn_min to generate a ZnNi-analog of human Glx2. Glx2-Znmin + 1 equivalent Ni exhibited a k_cat of 439 ± 6 s⁻¹ and K_m of 91 ± 9 μM when using SLG as substrate (Table 1). A ¹H NMR spectrum of ZnNi-Glx2 showed two relatively sharp, solvent-exchangeable peaks at 57 and 73 ppm (Figure 3(D)). In agreement with the other NMR studies, this result suggests that Ni(II) exhibits a preference to bind to the consensus Zn₂ site.

UV-Vis spectra of Co(II)-containing Glx2 samples were obtained in an effort to obtain information about the coordination number of Co(II) in these samples (Figure 4). The spectrum of Glx2-Co_min showed three peaks between 500 and 600 nm, which we assign to ligand field transitions of high-spin Co(II) and a weaker feature at 410 nm, which we assign to the presence of Co(III) (47). The extinction coefficient of the ligand field transitions ranged from 28 – 42 M⁻¹cm⁻¹, which suggests that the Co(II)’s are 5/6 coordinate. The addition of 1 equivalent of Zn(II) to the Glx2-Co_min sample did not result in a change in the intensities of the ligand field transitions nor the feature assigned to Co(III). This result suggests that both Co(II)’s in Glx2-Co_min are 5/6 coordinate. There was an increase in the 280 nm peak, which resulted in a relatively higher absorption of the feature at 410 nm and slightly higher absorbance of the ligand field transitions at 500 and 550 nm, was due to enzyme precipitation when Zn(II) was added to the sample.

Figure 4.

UV-Vis spectra of Glx2-Co_min and Glx2-Co_min + 1 equivalent of Zn(II). The buffer in these samples was 10 mM MOPS, pH 6.5, and the enzyme concentration was 1.6 mM.

Essentially indistinguishable EPR spectra were observed for both Glx2-LB and Glx2-LB+1.5Zn+1.5Fe, and were dominated by a g_eff. = 4.3 signal due to Fe(III) (Figure 5). Unlike the rich spectra observed from other Glx2 species, such as Arabidopsis mitochondrial Glx2-5 (34, 48, 49), the S = 5/2 signals from human Glx2 contained no well-resolved features other than the g_eff. = 4.3 line. This result suggests very high strains in the rhombic zero-field splitting term, E/D, and provides no confirmatory evidence for binding of Fe(III) to human Glx2 in a well-defined tight binding site. The intensities of the spectra from human Glx2 were equivalent and accounted for ≤ 0.2 Fe(III) per Glx2 molecule, suggesting that the majority of iron in the samples was present as Fe(II). A small feature at 3600 G (360 mT) suggested the possibility of an Fe(III)Fe(II) center, as in Glx2-5 (34), in a small proportion of the molecules (Figure 3C). This was the only evidence for bona fide binding of Fe(III) to Glx2. The spectrum also revealed a complex pattern of lines in the g_eff. = 2 region that, from comparison with a standard Mn(II) signal, suggested trace amounts of Mn(II), likely adventitiously bound.

Figure 5.

EPR spectra of human Glx2 analogs. (A) Glx2-LB and (B) Glx2-LB+1.5Zn+1.5 Fe. Trace C shows the g ~ 2 region of trace B expanded (thin line). Overlaid are a spectrum of Mn(II) in modeling wax (heavy line) and a signal from an Fe(III)Fe(II) center in Glx2-1 from Arabidopsis thaliana (heavy line with circular markers). A signal due to Cu(II) in the spectrometer cavity was subtracted from the experimental traces of human glyoxylase, and imperfect subtraction is likely responsible for the poor correlation of the Mn(II) reference signal with the glyoxylase signal in the g ~ 2.01 region (3100 – 3200 G; this is where the intense g(perpendicular) feature of Cu(II) is observed). Spectra were recorded at 2 mW microwave power, 10 K, and 12 G (1.2 mT) magnetic field modulation at 100 kHz.

EPR spectra of Glx2-Co_min indicated that only ~ 25 % of the Co(II) in Glx2 was EPR-visible. The EPR spectrum itself, recorded under non-saturating conditions, was complex (Figure 6A). Individual species were deconvoluted by preferential saturation of spectral components, and by the collection and analysis of spectra recorded under rapid passage conditions. the presence of three distinct Co(II) species. The EPR spectrum recorded under partially saturating conditions (Figure 6B) indicated that the derivative feature at 2040 G is not associated with either the large absorption-shaped component at 1110 G or the derivative feature at 2660 G. Comparison of two rapid-passage spectra recorded at different microwave powers (Figure 6C and D) indicated that the derivative feature at 2660 G was associated with the broad absorption ‘tail’ centered around 4000 G, and likely with some absorption in the 800 – 1600 G range. Subtraction of appropriate amounts of Figure 6D from 6C yielded an apparently single-component axial signal that was readily simulated as an M_S = ± ½ species with E/D = 0.1. The experimental data, therefore, clearly identified that three distinct species contributed to the spectrum, and provided a full set of parameters for one of them. Using the axial species, the two associated features at 2660 G and 4000 G from Figure 6D, and the distinct feature at 2040 G from Figure 6B as a basis for three species, the experimental spectrum recorded under non-saturating conditions was best simulated using the parameters described in the legend to Figure 6. The simulations indicate that two of the species are likely five-coordinate and the third is either five- or six coordinate. It is of note that for one of the species, that of Figure 6B, the g_real values are unusually high for Co(II) in a metalloprotein, though not excessively so. Our assignment to M_S = ± ½, and five-coordinate geometry, is based on comparison of the resultant g_real values of 2.9, 2.9 and 3.0 with those returned by assuming the alternative M_S = ± ³/₂ manifold, and hence distorted tetrahedral geometry, of 3.4, 3.4 and 3.0. These latter values would be unprecedented for Co(II) in an environment of the types found in the Glx2 active site.

Figure 6.

EPR spectra of human Glx2 containing Co(II). (A), Experimental EPR spectrum of 1.7 mM Glx2 incubated with 1.7 mM Co(II), recorded at 12 K, 0.8 mW. The doubly integrated intensity of the spectrum corresponded to 0.4 mM Co(II). A simulated spectrum is overlaid; the simulated spectrum was generated by adding 0.21, 0.44 and 0.35 fractional spin equivalents of the computed spectra B, D and E, respectively. (B), Experimental spectrum recorded at 7.6 K, 100 mW, and a computed spectrum with spin Hamiltonian parameters S = ³/₂, M_S = ½ (E ≫ gβSH), g|| = 3.0, g⊥ = 2.9, E/D = 0.28. (C) Rapid-passage δχ″/δH EPR spectrum recorded at 7.6 K, 100 mW. The experimental pseudo-absorption spectrum was collected with second-harmonic out-of-phase modulation-phase-sensitive detection. The derivative spectrum shown was generated by differentiating the experimental spectrum, applying 20 G pseudomodulation. (D) Rapid-passage δχ″/δH EPR spectrum recorded at 7.6 K, 10 mW. A computed spectrum is overlaid, with the parameters S = ³/₂, M_S = ½ (E ≫ gβSH), g|| = 2.15, g⊥ = 2.3, E/D = 0.29. (E) Difference spectrum generated by subtraction of a fraction of the experimental spectrum D from C, and an overlaid computed spectrum with the parameters S = ³/₂, M_S = ½ (E ≫ gβSH), g|| = 2.9, g⊥ = 2.6, E/D = 0.10. All spectra were recorded with 12 G magnetic field modulation at 100 kHz.

Discussion

The metallo-β-lactamase fold consists of an αβ/βα sandwich motif, made up of a core unit of two β-sheets surrounded by solvent-exposed helices (41, 42). Members of this superfamily contain a conserved HXHXD motif that has been shown to bind Zn(II), Fe, and Mn. There are several enzymes in metallo-β-lactamase fold family, including metallo-β-lactamases, glyoxalase 2, lactonase, rubredoxin:oxygen oxidoreductase (ROO), arylsulfatase, phosphodiesterase, and tRNA maturase (50). Most of the metallo-β-lactamase superfamily members (metallo-β-lactamases, tRNA maturase, phosphodiesterase, arylsulfatase, and lactonase) appear to contain dinuclear Zn(II) centers. On the other hand, rubredoxin:oxygen oxidoreductase (ROO) appears to contain a dinuclear iron center (51). Glx2 from E. coli was recently reported to contain a dinuclear Zn(II) center (35), while plant mitochondrial Glx2 (Glx2-5) has been shown to contain a FeZn center (34). Interestingly, plant cytoplasmic Glx2 (48, 49) and Glx2 from Samonella typhinurium (39) can exist with a number of possible metal centers, including dinuclear Fe, FeZn, MnZn, and presumably dinuclear Zn(II). Based on a crystal structure, human Glx2 was reported to contain a dinuclear Zn(II) center although the enzyme used for the crystallization studies contained 1.5 Zn(II) and 0.7 Fe (40).

In order to more clearly define the metal binding properties of human Glx2, the protein was over-expressed in either LB or minimum medium containing different metal ions. Human Glx2 over-expressed in LB medium, bound roughly equal, albeit substoichiometric, amounts of Zn(II) and Fe, and exhibited a k_cat of 570 s⁻¹ and a very large K_m value of 660 μM (Table 1). Addition of Zn(II) and Fe to this enzyme followed by dialysis resulted in enzyme (Glx2-LB+1.5Zn+1.5Fe) that bound 1.2 ± 0.1 of Zn(II) and 0.9 ± 0.1 equivalents of Fe and exhibited a ~20% higher k_cat and a similar K_m (within error) value. The similar K_m values suggest a common active species, yet a 2-fold increase in metal content did not correspond to a 2-fold increase in k_cat. This result is most likely due to the as yet unexplained drop in activity when Glx2 samples are dialyzed (43, 52). There were also drops in k_cat when Glx2-Zn_min was incubated with Zn(II) or Fe, followed by dialysis (Table 1).

The large K_m values exhibited by human Glx2 samples prepared from LB medium were not observed for any of the Glx2 analogs prepared from minimal medium (Table 1). This result suggests that the samples prepared in LB contain a competitive inhibitor, perhaps a peptide from the LB medium. The presence of a competitive inhibitor in the enzymes cultured in LB is supported by the stopped-flow studies that show a lower activity for Glx2-LB (less Zn(II)) and similar activity for Glx2-LB+1.5Zn+1.5Fe analogs as compared to the Glx2 samples prepared in minimal medium (Figure 2). The ¹H NMR spectrum of Glx2-LB+1.5Zn+1.5Fe did not reveal any unassigned peaks (Figure 3), suggesting that a peptide does not bind directly to the Fe(II) center. However, it is possible that resonances from protons on a metal-bound peptide may not shift to downfield positions greater than 30 ppm. In addition, the crystal structure of human Glx2 identified a number of active site residues that interact with groups on glutathione (40), so the binding of a peptide to the metal may not be required for a Glx2-peptide complex to form. MALDI-TOF mass spectrometry was used to compare multiple Glx2 samples prepared in LB and minimal medium in an effort to identify a bound competitive inhibitor; however, no differences in the masses of the different GLX2 analogs were identified. It is possible that the ionization process coupled with the acidic matrix used in the MALDI technique resulted in the loss of the inhibitor.

In an effort to obtain human Glx2 containing only one metal ion, the enzyme was over-expressed in minimal medium containing either Zn(II) or Fe(II). Glx2 over-expressed in minimal medium containing Zn(II) (Glx2-Zn_min) was shown to bind 1.1 ± 0.2 equivalents of Zn(II) and no detectable Fe or Mn (Table 1). When Glx2-Zn_min was incubated with a 3-fold excess of Zn(II), and unbound Zn(II) was removed by dialysis, the resulting enzyme contained 0.9 ± 0.2 equivalents of Zn(II). It is unlikely that the samples of Glx2-Zn_min contain a mixture of ZnZn-Glx2 and apo-Glx2, since we are unable to prepare an analog of Glx2 containing 2 equivalents of Zn(II) by adding Zn(II) to the sample. This result is in contrast to the crystallographic conclusion that human Glx2 contains a dinuclear Zn(II) metal center (40). Regarding iron in Glx2, both NMR and EPR indicated that iron in the iron-containing forms of Glx2 was largely in the Fe(II) state. However, sufficient iron (≈ 20 %) was present as Fe(III) to provide for an easily observed EPR signal. Almost all of the Fe(III), however, was present as mononuclear Fe(III), and only a very small signal that was suggestive of an Fe(III)Fe(II) center was observed. These data argue against any significant proportion of di-iron Glx2. On the other hand, human Glx2 can bind two metal ions, and metal analyses and spectroscopic studies suggest that Zn(II) binds in the consensus Zn₁ site, while Ni(II), Co(II), and Fe(II) can bind in the consensus Zn₂ site. The formation of mixed-metal analogs is not surprising since mixed-metal analogs of several metallo-β-lactamases have been reported (53–55). However, a surprising result is that a CoCo-analog of human Glx2 can be prepared (Figure 3), while the biophysically/biochemically similar dinuclear Zn(II) analog cannot be prepared (see metal analyses data above). Given the bioavailability of the metal ions tested (56), we hypothesize that human Glx2 contains a Zn(II)Fe(II) center in vivo. Unlike Arabidopsis Glx2-2 (49, 57), dinuclear Fe or Zn(II) containing analogs of human Glx2 cannot be prepared.

The steady-state kinetic studies on Glx2 samples prepared in minimal medium revealed some surprising results with respect to metal content and activity. Glx2-Zn_min, which contains ~ 1 equivalents of Zn(II), is the most active Zn(II)-containing form of these analogs (Table 1), and the presence of Zn(II) or Fe(II) in the assay buffer does not greatly affect the activity of the enzyme. This result demonstrates that a dinuclear metal center is not required for full catalytic activity of human Glx2 and that the second metal ion does not play a large role in catalysis. The steady-state kinetic data did reveal that Glx2-Zn_min, which was incubated with 3 equivalents of Zn(II) or Fe(II), exhibited similar albeit lower activities. This lower activity is probably due to the dilute (low nM) enzyme being somewhat unstable after dialysis. At the higher concentrations used in the stopped-flow studies, the Glx2-Zn_min + Zn, and Glx2-Zn_min + Fe analogs exhibited similar rates as Glx2-Zn_min (Figure 2). The steady state and stopped-flow kinetic data on the Glx2-Zn_min samples after incubation with Zn(II) or Fe(II) and dialysis show that Fe has little or no effect on the activity of the enzyme. These data clearly show that human Glx2 is active when there is a mononuclear Zn(II) bound to the enzyme, presumably in the Zn₁ site. This result is important for the rational design of inhibitors that target the metal binding site.

The most active form of human Glx2 is Glx2-Co_min (Table 1 and Figure 2). NMR studies show that Co(II) binds to both metal binding sites in this analog (Figure 3). EPR spectra of Glx2-Co_min revealed the presence of three distinct Co(II) species, two rhombic and one axial. The exhibition of both an axial and a rhombic signal from a single Co(II) binding site has been observed in a number of instances and, in some cases, been shown to be due to the different effects of water and hydroxyl ligands, in a pH-dependent equilibrium, on EPR strain parameters (58). However, the presence of two highly rhombic species, and three species overall, indicates at least two Co(II) binding sites, and the spin Hamiltonian parameters indicate five- or sixfold coordination in each case. These data are entirely in accord with the NMR results, showing binding to both Zn₁ and Zn₂ sites, and with the five- or sixfold coordination indicated by the electronic absorption spectrum. The Co(II) EPR signals accounted for only ~ 25 % of the total Co(II) and corresponded to a population of Glx2 in which either the Zn₁ or Zn₂ site was occupied. Most of the Co(II) in Glx2, then, is likely present in a dinuclear site that is EPR-silent due to antiferromagnetic coupling. This coupling need only be very weak, of the order of a wavenumber, to preclude observation of an EPR signal. When 1 equivalent of Zn(II) is added to this enzyme, Zn(II) presumably displaces Co(II) from the Zn₁ site, and the resulting ZnCo analog exhibits similar activity as Glx2-Zn_min. These results suggest that the most active analog of human Glx2 has Co(II) in the Zn₁ site and that the metal ion in the Zn₂ site plays little role in catalysis. We have attempted to prepare a CoCo analog human Glx2 by adding Co(II) to Glx2-Co_min; however, the enzyme precipitates upon addition of Co(II). It is not clear why Glx2-Co_min is more active than the corresponding Glx2-Zn_min or ZnCo-analogs; however, the greater Lewis acidity of Co(II) as compared to Zn(II) may explain some of the differences (59). Co(II)-substituted liver alcohol dehydrogenase is 140% more active than the corresponding Zn(II)-containing enzyme (59). In spite of greater activity, the relatively lower bioavailability of Co(II), as compared to Zn(II) or Fe, strongly suggests that human Glx2 is not a Co(II)-containing enzyme.

The results present herein demonstrate that human Glx2 is active as a mononuclear Zn(II)-containing enzyme and that Zn(II) binds preferentially to the consensus Zn₁ site. This finding is similar to a recent study that reported that metallo-β-lactamase L1 is active as a mononuclear Zn(II) enzyme when Zn(II) is bound in the Zn₁ site (53). The metal content of the active forms of other enzymes belonging to the metallo-β-lactamase superfamily is not clear. Unlike the other enzymes in this superfamily, the metal ion in the Zn₂ site does not appear to play a significant role in human Glx2. Nonetheless given the bioavailability of Zn(II) and Fe in cells (56), we predict that human Glx2 contains a Zn(II)Fe(II) metal binding site in vivo and not a Zn(II)Zn(II) site as previously reported (40).

It has been proposed that the enzymes in the metallo-β-lactamase superfamily arose due to a gene duplication event (50), and the presence of a dinuclear metal binding site may offer some of the enzymes increased activity. Since Glx2 appears to be a critical enzyme involved in the cellular detoxification of 2-oxoaldehydes (1), it may have evolved to be active with only one metal ion, so that the enzyme is active in the presence of low Zn(II) concentrations. The ability to utilize only the single, non-redox active Zn(II) site for catalysis would better position the enzyme to react with the oxidizing substrates.

Abbreviations

EDTA

ethylenediaminetetraacetic acid

FPLC

fast performance liquid chromatography

Glx2-Co_min

recombinant human glyoxalase 2 over-expressed in minimal medium containing Co(II)

Glx2-Co_min + Zn(II)

recombinant human glyoxalase 2 over-expressed in minimal medium containing Co(II) to which was added 1 eq. of Zn(II)

Glx2-Fe_min + Zn(II)

recombinant human glyoxalase 2 over-expressed in minimal medium containing Fe and Zn(II)

Glx2-LB

recombinant human glyoxalase 2 over-expressed in LB medium

Glx2-LB+1.5Zn+1.5Fe

recombinant human glyoxalase 2 over-expressed in LB medium which was incubated with 1.5 eq. of Zn(II) and Fe(II) and dialyzed

Glx2-Zn_min

recombinant human glyoxalase 2 over-expressed in minimal medium containing Zn(II)

Glx2-Zn_min + Ni(II)

recombinant human glyoxalase 2 over-expressed in minimal medium containing Zn(II) to which was added 1 eq. of Ni(II)

Glx2-Zn_min + 3 eq. Fe(II)

recombinant human glyoxalase 2 over-expressed in minimal medium containing Zn(II) which was incubated with 3 eq of Fe(II) and dialyzed

Glx2-Zn_min + 3 eq. Zn(II)

recombinant human glyoxalase 2 over-expressed in minimal medium containing Zn(II) which was incubated with 3 eq. of Zn(II) and dialyzed

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

ICP-AES

inductively-couple plasma with atomic emission spectroscopy detection

IPTG

isopropyl-β-D-thiogalactopyranoside

LB

Luria-Bertani

MG

methylglyoxal

MOPS

3-morpholinopropanesulfonic acid

PMSF

phenylmethylsulfonylfluoride

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

SLG

S-(2-hydroxyacyl)glutathione