Purification and properties of the glutathione S-transferases from the anoxia-tolerant turtle, Trachemys scripta elegans William G. Willmore and Kenneth B. Storey Institute of Biochemistry, Carleton University, Ottawa, Ontario, Canada The glutathione S-transferases (GSTs) belong to a multigene enzyme superfamily which catalyze the nucleophilic addition of the thiol of reduced gluta- thione (GSH) to a variety of electrophiles [1–7]. Thus, they provide protection, not only against electrophiles which tend to be toxic to the cell, but also against oxidants which they reduce. The GSTs are homodimers or heterodimers com- prised of pairings of seven different subunits [5,8]. Five main classes of GSTs exist, each containing more than one isozyme based on substrate affinity and inhibitor properties. The cytosolic classes have been named alpha (a), mu (l), pi (p) and theta (h) based on their subunit composition, substrate ⁄ inhibitor speci- ficity, primary and tertiary structure similarities and immunological identity [8]. The fifth class is the micro- somal form of the enzyme. Specific GST subunits are induced by various xenobiotics and are expressed in a tissue specific manner [9]. Expression of GST subunits is under the control of the antioxidant ⁄ electrophile response element (ARE ⁄ EpRE) to which members of the bZIP family of transcription factors (Nrf2 and Maf G ⁄ K) bind [10]. The enzymes contain two binding sites within the active site, a G-site for the binding of GSH and a H-site for the binding of an electrophile. Electrophiles have a slow spontaneous rate of reaction with GSH which is greatly enhanced in the presence of GST. Electrophilic substrates for GST include xenobiotics such as carcinogens and their metabolites, herbicides Keywords Adaptation; anoxia; glutathione S-transferases; turtle Correspondence W. G. Willmore, Institute of Biochemistry, Carleton University, Ottawa, Ontario, K1S 5B6, Canada Fax: +01 613 520 3539 Tel: +01 613 520 2600, ext. 1211 E-mail: Bill_Willmore@carleton.ca Website: http://www.carleton.ca/bwillmor (Received 28 March 05, revised 17 May 05, accepted 20 May 05) doi:10.1111/j.1742-4658.2005.04783.x Glutathione S-transferases (GSTs) play critical roles in detoxification, response to oxidative stress, regeneration of S-thiolated proteins, and cata- lysis of reactions in nondetoxification metabolic pathways. Liver GSTs were purified from the anoxia-tolerant turtle, Trachemys scripta elegans. Purification separated a homodimeric (subunit relative molecular mass ¼ 34 kDa) and a heterodimeric (subunit relative molecular mass ¼ 32.6 and 36.8 kDa) form of GST. The enzymes were purified 23–69-fold and 156– 174-fold for homodimeric and heterodimeric GSTs, respectively. Kinetic data gathered using a variety of substrates and inhibitors suggested that both homodimeric and heterodimeric GSTs were of the a class although they showed significant differences in substrate affinities and responses to inhibitors. For example, homodimeric GST showed activity with known a class substrates, cumene hydroperoxide and p-nitrobenzylchloride, whereas heterodimeric GST showed no activity with cumene hydroperoxide. The specific activity of liver GSTs with chlorodinitrobenzene (CDNB) as the substrate was reduced by 2.6- and 8.7-fold for homodimeric and hetero- dimeric GSTs isolated from liver of anoxic turtles as compared with aerobic controls, suggesting an anoxia-responsive stable modification of the protein that may alter its function during natural anaerobiosis. Abbreviations ARE, antioxidant response element; CDNB, chlorodinitrobenzene; EpRE, electrophile response element; GST, glutathione S-transferase; GSH, reduced glutathione. 3602 FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS and mutagens. In addition, GSTs bind with varying affinities to a variety of hydrophobic compounds such as heme, bilirubin, polycyclic aromatic hydrocarbons and dexamethasone [7]. Endogenous second substrates for GST are toxic products generated from tissue dam- age. These include the compounds resulting from lipid peroxidation of biological membranes such as reactive alkenes, epoxides, hydroperoxides and aldehydes. These may be the primary substrates of the micro- somal or membrane-bound GST in the same way as they are the substrates for Se-dependent glutathione peroxidases (the ‘classical’ and a more recently discov- ered phospholipid hydroperoxide glutathione peroxi- dase) [11]. Most conjugated products of GSTs are cytotoxic and therefore must be eliminated. Glutathi- one S-conjugated products are exported from cells (in particular, from liver cells where cytotoxins are con- centrated) via a membrane ATP-dependent pump known as the glutathione S-conjugate export pump [12,13], converted to mercapturic acids in the kidney and epithelial cells, and excreted in the urine [8]. Numerous lower vertebrates show well-developed tolerance for long-term oxygen deprivation and studies in recent years have demonstrated that anoxia toler- ance includes not just biochemical adjustments that deal with the metabolic and energetic consequences of survival without oxygen but also adaptations of anti- oxidant defenses that help to limit oxidative stress on cells when oxygen is reintroduced [14,15]. The best ver- tebrate facultative anaerobes are freshwater turtles of the genera Trachemys and Chrysemys. These can sur- vive for several weeks submerged in deoxygenated water at cold temperatures, an adaptation that sup- ports winter survival in ice-locked ponds [14]. Liver and heart GST activities decreased significantly after 20 h of anoxic submergence in the red-eared slider, Trachemys scripta elegans [16], indicating that the enzyme responded to anoxia stress. This change could result from one or more factors such as a change in the amount of GST protein present, a covalent modifi- cation of GST that alters its properties, or a change in the mixture of GST isozymes present in the organ to better suit the enzyme for function under anoxic condi- tions. Stress-related changes in the maximal activities of GST are known to occur in many stress-tolerant organisms. For example, the maximal activities of GST increased during anoxia exposure in brain of the leopard frog Rana pipiens [17] but decreased during freezing in kidney and heart of the wood frog Rana sylvatica [18]. A decrease in maximal GST activity also occurred during estivation in liver and four other organs of the spadefoot toad Scaphiopus couchii [19]. In the present study, two GST isoforms were puri- fied from liver of the anoxia tolerant turtle, T. s. ele- gans. Analysis of kinetic and inhibitory properties characterized these as alpha class GSTs but the two forms showed a variety of distinctive differences. The specific activities of both were reduced in anoxic liver suggesting anoxia-responsive regulation of GST. Results GST Purification Table 1 summarizes the purification of turtle liver GSTs using Matrix Red dye ligand chromatography, Sephacryl S-200 gel filtration and hydroxylapatite ion exchange chromatography. GST activity from liver of both aerobic and anoxic turtles eluted from the Matrix Red column in a single peak at  440 mm KCl (data not shown). Elution from Matrix Red gave a 3.7-fold purification with 86% yield of the control liver enzyme. With the anoxic enzyme, however, this col- umn gave no purification (no change in specific activ- ity) but it was used anyway so that both enzymes were treated alike. Typical elution profiles from Sephacryl Table 1. Purification of GST from liver of control and anoxic turtles. Enzyme activity was assayed using optimal CDNB and GSH concentra- tions. Results are from a single purification but all other trials yielded similar results. Column Control Liver Anoxic Liver Specific activity Specific activity (UÆmg )1 protein) Fold purification % yield (UÆmg )1 protein) Fold purification % yield Supernatant 0.319 1.00 100 0.0408 1.00 100 Matrix Red 1.17 3.67 86.0 0.0335 0.821 95.9 Sephacryl S-200 11.6 36.4 108 5.91 145 113 Hydroxylapatite (Peak 1) 7.31 22.9 103 2.80 68.6 117 Hydroxylapatite (Peak 2) 55.4 174 6.37 156 W. G Willmore and K. B. Storey GST function in anoxia-tolerant turtles FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS 3603 S-200 are shown in Fig. 1 for GSTs from control and anoxic turtle liver. Both enzymes eluted in a single peak with a calculated mean molecular mass of the native dimer being 59.8 ± 3.25 kDa. Elution from Sephacryl S-200 resulted in an activation of the enzyme (125 and 118% yield as compared with the activity after the Matrix Red column and 108 and 113% as compared with the crude extract for control and anoxic enzymes, respectively). The increase in spe- cific activity at this purification step suggested the possible loss of a low molecular mass repressor of the enzyme. A typical elution profile for control and anoxic turtle liver GSTs off hydroxylapatite is shown in Fig. 2. Two peaks eluted in both cases at about 98 and 131 mm KPi, respectively. The portion of total activity that was present in Peak 2 was higher in sam- ples from anoxic liver than in aerobic liver, the Peak 1 ⁄ Peak 2 ratio being 1.46 : 1 for the enzyme from con- trol preparations and 1.03 : 1 for anoxic preparations (assayed with CDNB as the substrate). The combined yield of GST activity in the two peaks was 103% for control and 117% for anoxic enzymes, respectively, compared with the crude supernatant (Table 1). No activity with H 2 O 2 as a substrate was detected in either of the CDNB-utilizing GST peaks that were eluted from the hydroxylapatite column indicating that this column had separated Se-dependent GPOX activity from GST. No new peaks of GST activity were seen in the elution profiles off any column when isolations from anoxic liver were compared with aerobic liver. It was therefore concluded that no new isozymes of GST were produced during anoxia exposure. Subsequent kinetic studies characterized the properties of GST in hydroxylapatite Peaks 1 and 2 from control liver. Isoelectric focusing Isoelectric focusing of GSTs from liver of aerobic and anoxic turtles is shown in Fig. 3. In both cases, turtle liver GSTs separated into two peaks; pI values were 8.5 and 8.7 for the larger peak and 6.1 and 6.8 for the smaller peak in aerobic vs. anoxic preparations, respectively, using CDNB as a substrate. The larger shift in pI values for the smaller peak possibly repre- sents an anoxia-dependent stable modification of the enzyme. When cumene hydroperoxide was used as a substrate, the glutathione peroxidase activity of turtle liver GSTs was tested. The ratio of cumene hydroper- oxide to CDNB activities was 0.37 to 0.39 for Peak 1 and either 0.58 or 0.81 for control and anoxic turtle Fig. 1. Typical profiles of GST elution from Sephacryl S-200 for the liver enzyme from control and anoxic T. s. elegans. Activities are expressed relative to peak fractions which were set at 100%. GST activity from control and anoxic turtle liver pools eluted in one peak at the same molecular mass (between 56.5 and 63.0 kDa). Stand- ards were Blue Dextran (BD; 2000 kDa), phosphofructokinase (PFK; 360 kDa), pyruvate kinase (PK; 238 kDa), aldolase (ALD; 150 kDa), hexokinase (HK; 102 kDa), hemoglobin (Hb; 64.5 kDa), and cyto- chrome c (Cyt c; 13.7 kDa). d, s, control and anoxic isolations, respectively. Fig. 2. Typical profiles for GST elution from hydroxylapatite for the liver enzyme from control and anoxic T. s. elegans. Activities are expressed relative to peak fractions which were set at 100%. The column was eluted with a 0–250 m M gradient of potassium phos- phate. GST activity eluted in two peaks at 98 and 131 m M KP i for Peak 1 and Peak 2, respectively. The percentage of total GST activ- ity present in Peak 2 increased during anoxia. d, s, control and anoxic isolations, respectively. GST function in anoxia-tolerant turtles W. G Willmore and K. B. Storey 3604 FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS liver GSTs, respectively, for Peak 2 (Table 2). The increase in the ratio of activities for anoxic liver GSTs for Peak 2 was due primarily to a decrease in CDNB activity. No activity with H 2 O 2 as the substrate was detected in either peak. SDS/PAGE The results of SDS ⁄ PAGE of turtle liver GSTs, puri- fied to homogeneity, are shown in Fig. 4A; lane 3 shows the Peak 1 enzyme and lane 4 shows the Peak 2 enzyme eluted from the hydroxylapatite column. A comparison with equine liver GST is also shown in lane 2. Using the standard curve constructed from the protein standards (Fig. 4B), the molecular mass of the Peak 1 GST subunit was determined to be 34.0 kDa (Table 2). Peak 2 showed 2 subunits of 36.8 and 32.6 kDa. All turtle liver subunits were larger than the two equine liver subunits (28.9 and 21.1 kDa). It was concluded that GST in Peak 1 is a homodimer with an approximate molecular mass of 68 kDa (homGST) whereas GST in Peak 2 is a heterodimer with an approximate mass of 69.4 kDa (hetGST); both are lar- ger than equine liver GST which is a heterodimer of 50.0 kDa. Fig. 3. Isoelectric focusing profiles of liver GST from control (A) and anoxic (B) T. s. elegans. Activities are expressed relative to peak fractions which were set at 100%. Two peaks of GST activity were found in both control and anoxic situations. In both cases the activ- ity profile with cumene hydroperoxide activity as the substrate (h) matched the profile for CDNB activity (d). s, pH. Table 2. General characteristics of GSTs purified from turtle liver. Results are means ± SEM, n ¼ 3 determinations on independent preparations; otherwise n ¼ 1. Units are corrected for the volume assayed. Hydroxylapatite (Peak 1) Hydroxylapatite (Peak 2) Arrhenius activation energy (E a ) (kJÆmol )1 ) 36 ± 2.2 40 ± 3.7 pH optimum 7.2 7.2 Molecular mass (Da) 34 000 36 800 (subunit 1) 32 600 (subunit 2) Specific activity using CDNB (UÆmg protein )1 ) Control 7.3 ± 0.38 55 ± 5.8 a Anoxic 2.8 ± 0.051 b 6.4 ± 0.098 a,b Specific activity using cumene hydroperoxide (UÆmg protein )1 ) Control 1.8 ± 0.054 0 Anoxic 0.94 ± 0.0086 b 0 Isoelectric Focusing (Peak 1) Isoelectric Focusing (Peak 2) pI Control 8.5 ± 0.24 c 6.1 ± 0.26 Anoxic 8.7 ± 0.10 c 6.8 ± 0.26 CDNB activity (UÆmL )1 of peak fraction assayed) Control 1.9 0.46 Anoxic 1.7 0.26 Cumene hydroperoxide activity (UÆmL )1 of peak fraction assayed) Control 0.72 0.26 Anoxic 0.61 0.21 Ratio of cumene hydroperoxide to CDNB activities Control 0.39 0.58 Anoxic 0.37 0.81 a Significantly different from Peak 1-values as assessed by a two- tailed Student’s t-test, P < 0.005; b significantly different from aero- bic control values P < 0.005; c major peak activity using CDNB. W. G Willmore and K. B. Storey GST function in anoxia-tolerant turtles FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS 3605 Kinetic and inhibition characteristics The specific activity of purified GST in Peaks 1 and 2 changed dramatically between aerobic and anoxic states, in all cases decreasing significantly (P<0.005) in the anoxic state (Table 2). With CDNB as the sub- strate, the specific activity of purified Peak 1 GST was 7.3 ± 0.38 UÆmg )1 protein for the aerobic control enzyme and fell by 62% to 2.8 ± 0.05 UÆmg )1 protein in anoxia. Peak 1 activity using cumene hydroperoxide as the substrate similarly decreased by 47% from an aerobic value of 1.8 ± 0.054 UÆmg )1 protein to an anoxic value or 0.94 ± 0.0086 UÆmg )1 protein. Activity of Peak 2 GST with CDNB changed even more dramatically, decreasing by 89% from 55.4 ± 5.8 UÆmg )1 protein for the aerobic enzyme to 6.4 ± 0.098 UÆmg )1 protein in anoxia. Activity using cumene hydroperoxide as a substrate was not detected in Peak 2 off hydroxylapatite. Substrate and inhibitor profiles of Peaks 1 and 2 GST isozymes off of hydroxylapatite from aerobic control liver are summarized in Table 3. Peak 1 GST had a greater affinity for GSH, with a K m that was only 63% of the Peak 2-value. By contrast, Peak 2 GST had a greater affinity for CDNB with a K m that was 67% of the Peak 1 enzyme. Peak 1 GST could use A B Fig. 4. SDS ⁄ PAGE of purified GSTs from liver of control T. s. ele- gans. (A) A 15% acrylamide gel was run, lane 1, mass standards; 2, horse liver GST (Sigma); 3, turtle liver GST from Peak 1; 4, turtle liver GST from Peak 2. The standards were myosin (200 kDa), b-ga- lactosidase (116 kDa), phosphorylase B (97.4 kDa), serum albumin (66.2 kDa), ovalbumin (45 kDa), carbonic anhydrase (31 kDa), trypsin inhibitor (21.5 kDa), lysozyme (14.5 kDa) and aprotinin (6.5 kDa). (B) Standard curve used to determine the subunit molecular mass of turtle GSTs. The positions of GST subunits are shown (s). Table 3. Kinetic parameters of GST isozymes purified from aerobic turtle liver. Results are means ± SEM, n ¼ 3 independent determi- nations. Effector Hydroxylapatite (Peak 1) Hydroxylapatite (Peak 2) K m GSH (mM) 0.38 ± 0.019 0.60 ± 0.064 a K m CDNB (mM) 1.7 ± 0.15 1.14 ± 0.040 a K m Cumene hydroperoxide (m M) 0.11 ± 0.021 No activity I 50 GSSG (mM) 2.0 ± 0.19 2.6 ± 0.57 I 50 Cibacron Blue (lM) 48 ± 0.97 8.4 ± 0.38 b I 50 Rose Bengal (lM) 0.31 ± 0.016 0.47 ± 0.085 I 50 S-hexylglutathione (lM) 0.31 ± 0.036 0.39 ± 0.19 I 50 iodoacetamide (mM) 40 ± 0.46 8.7 ± 0.33 b I 50 KCl (M) 0.33 ± 0.039 0.18 ± 0.020 a I 50 NaCl (M) 0.332 ± 0.0341 0.17 ± 0.012 a I 50 Na 2 SO 4 (M) No inhibition 0.19 ± 0.045 I 50 NH 4 Cl (M) 0.20 ± 0.020 0.12 ± 0.010 a I 50 Na acetate (M) No inhibition No inhibition Other substrates tested (specific activity in UÆmg )1 ; n ¼ 1 deter- mination) 1,2-dichloro-4-nitrobenzene (1 m M) 0.0035 0.0090 p-Nitrobenzylchloride (1 m M) 0.18 0.28 p-Nitrophenylacetate (1 m M) 0.15 1.2 a Significantly different from the corresponding Peak 1-value via the Student’s t-test P < 0.05; b P < 0.001. GST function in anoxia-tolerant turtles W. G Willmore and K. B. Storey 3606 FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS cumene hydroperoxide as a substrate but Peak 2 GST could not. Neither enzyme showed activity with H 2 O 2 indicating that Se-dependent GPOX activity was not present in either peak. Several other potential GST substrates were also tested for catalytic activity. Nei- ther enzyme showed activity with ethacrynic acid, trans-4-phenyl-3-buten-2-one or 1,2-epoxy-3-(p-nitro- phenoxy) propane. Peak 2 GST showed 2.56-, 1.56-, and 7.93-fold greater activity than Peak 1 GST using 1,2-dichloro-4-nitrobenzene, p-nitrobenzylchloride, and p-nitrophenylacetate, respectively. Responses to inhibi- tors also characterize different GST isoforms. Ciba- cron Blue and iodoacetamide were both much stronger inhibitors of Peak 2 GST with I 50 values that were just 18–22% that of the corresponding Peak 1-values. Peak 2 GST was also more strongly inhibited by chloride salts (KCl, NaCl, NH 4 Cl) with I 50 values that were 52–60% of the corresponding values for Peak 1 GST. Sodium acetate did not inhibit either enzyme and sodium sulfate inhibited only Peak 2 GST. Rose ben- gal, hexylglutathione, and the GSSG product of the GST reaction inhibited both turtle liver GST isozymes to a similar extent. Temperature and pH dependence Fig. 5 shows pH curves for Peak 1 and Peak 2 GSTs from aerobic turtle liver. The pH optimum of both enzymes was 7.2 (Table 2). Activity declined relatively slowly on the acidic side so that about 40% of activity still remained at pH 6 whereas activity fell sharply at higher pH values with almost no activity remaining at pH 7.6 and above. Arrhenius plots for Peaks 1 and 2 GSTs are shown in Fig. 6. Both enzymes showed a straight line rela- tionship over the full range of temperatures tested (5–40 °C). Calculated activation energies (Ea) were 36 ± 2.2 and 40 ± 3.7 kJÆmol )1 for Peak 1 and Peak 2 GST, respectively, and were not significantly different. Discussion Freshwater species of turtles (T. s. elegans and Chryse- mys picta bellii) can survive extended periods of sub- mergence past the point at which internal oxygen reserves are exhausted. These species tolerate oxygen deprivation for a day or more at 20 °C and at least 3 months at 3 °C [20]. Such conditions occur during overwintering hibernation in ice-covered rivers and ponds where the water becomes quite hypoxic and tur- tle bury themselves in anoxic mud [21]. The hallmarks of anoxia tolerance in turtles include a profound lowering of metabolic rate and a buffering of lactic acidosis [22]. The magnitude of metabolic depression can be 10–20% of the normoxic rate and can be further decreased to 0.1% due to Q 10 effects of tem- perature. During hibernation, plasma lactic acid load can climb to as high as 150–200 mm [23,24]. In non- tolerant organisms, the drop in plasma pH can be as large as a full pH point [25]. Freshwater turtles coun- ter this acid load by buffering it with bicarbonate, Ca 2+ , and Mg 2+ ions from the shell [26]. In terms of their biochemistry, the enzymes of freshwater turtles must work optimally at low pHs during acid load. The current study shows that turtle GSTs function optimally under acidic conditions occurring under anaerobiosis. Turtles, being ectotherms, will have lower metabolic rates than those of endotherms of comparative sizes [22]. A drop in environmental temperature will lower an ectotherm’s metabolism even further due to Q 10 effects [22]. Therefore, the activities of turtle enzymes normally differ from those of mammalian vertebrates at their respective biological temperatures and oxygen exposure. With temperature differences taken into account, Na + ⁄ K + ATPase and creatine kinase activit- ies are two- to threefold higher in rat than in turtle brain, whereas hexokinase and lactate dehydrogenase Fig. 5. pH profiles of GSTs purified from liver of control turtles. Data are means ± SEM for n ¼ 3 trials performed on a single enzyme preparation. Where error bars are not visible, they are con- tained within the symbol. Phosphate buffer was used and pH was confirmed immediately prior to and following the assay; the pH val- ues shown are the average of these two values. Peak 1 (d, lines) and Peak 2 (s, dotted lines) GST had pH optima of 7.20 and 7.21, respectively. W. G Willmore and K. B. Storey GST function in anoxia-tolerant turtles FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS 3607 activities were found to be similar [27]. This is consis- tent with the idea that lower rates of Na + and K + pump fluxes result in lower rates of aerobic energy metabolism in turle brains compared with rat brains. Superoxide dismutase activities in turtle brain, lung and skeletal muscle, but not liver or cardiac muscle were found to be significantly lower than those found in mouse and rabbit [28]. This shows the relationship between SOD activities and oxygen exposure in verte- brate species. The GST activities in the current study were measured at room temperature to remove any temperature effects on enzyme activities that normally occur during over-wintering hibernation. The maximum activity of GST in T. s. elegans liver was previously found to decrease by 25% over 20 h of anoxia exposure and this suggested a possible role for changes in GST activity in the support of anaerobiosis [16]. The mechanism of GST modulation in anoxia could take one of several forms and, hence, this study of turtle liver GST was undertaken to identify any anoxia-responsive changes in isozymic forms, specific activities, and kinetic properties of the enzyme. The current data document the presence of two isozymes of GST in turtle liver that are separable by column chro- matography and isoelectric focusing but did not find evidence of a change in the expression pattern of either isozyme during anoxia or of the expression of novel GST isozymes under anoxia. However, the effect of anoxia exposure on liver GST activity was profound when the purified enzymes were examined; the specific activities of purified Peak 1 and 2 GSTs from anoxic liver were only 38 and 11%, respectively, of the corres- ponding values for the aerobic enzymes (Table 2). This suggests that the GST protein may undergo a stable modification in response to anoxia that lowers its spe- cific activity and may also affect other kinetic pro- perties. To date, there have been no reports in the literature that GSTs are regulated by post-translational modification. Most GSTs are regulated by a change in isozyme form and specific GST subunits are induced by various xenobiotics and are expressed in a tissue specific manner [5,9]. GSTs are often purified using an affinity column which has GSH attached to the stationary phase, either S-hexylglutathione or sulfobromophthalein gluta- thione [29], but neither of these worked for turtle liver GSTs which either bound irreversibly to the resins (and could not be eluted with very high concentrations of GSH) or were denatured. The glutathione S-transf- erases contain two sites for substrate binding; a G-site for the binding of glutathione and a H-site for the binding of hydrophobic substrate. S -hexylglutathione has previously been shown to bind to the H-site of the enzyme [30] while sulfobromophthalein, a noncompeti- tive inhibitor of GSTs, has been shown to bind to a site other than the active site [31]. In both cases, elu- tion with GSH would not be possible. Interestingly the large relative molecular mass GSTs from the yeast Yarrowia lipolytica [32] were also found not to bind to GSH affinity columns. Studies on crystallized turtle liver GSTs would provide information on the proxi- mity of the G-, H- and inhibitor sites in relation to GSTs from other organisms. Turtle liver GST was purified with a combination of three chromatography methods: dye ligand, gel filtra- tion and ion exchange. The purification scheme devel- oped for turtle GSTs resulted in final specific activities of 7.31 and 55.4 UÆmg )1 protein for Peaks 1 and 2 GST from aerobic control liver and 2.80 and 6.37 UÆmg )1 protein for the enzymes from anoxic liver (Table 1). Specific activities for both enzymes were in the range of values reported for a class GSTs in ham- ster liver (8.0–8.1 UÆmg )1 protein) [33] but, with the exception of the specific activity of Peak 2 GST from control animals, were lower than activities reported for human liver (16–37 UÆmg )1 protein) [1], various mam- malian tissues (20–357 UÆmg )1 protein) [34] and adult Fig. 6. Arrhenius plots for GSTs purified from liver of control T. s. elegans. Data are means ± SEM for n ¼ 3 trials performed on a single enzyme preparation. Where error bars are not visible, they are contained within the symbol. Phosphate buffer was used and cuvette temperature was checked immediately prior to and follow- ing the assay; the temperatures shown are the mean these two values. Peak 1 (d, solid line) and Peak 2 (s, dotted line) enzymes from hydroxylapatite. For both isozymes, heat denaturation was evi- dent at 40 °C. GST function in anoxia-tolerant turtles W. G Willmore and K. B. Storey 3608 FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS toad (Bufo bufo) liver (24–55 UÆmg )1 protein) [35]. However, SDS ⁄ PAGE of the pooled peak fractions revealed that the enzymes were purified to homogen- eity (Fig. 4A). Turtle liver GSTs showed a higher molecular mass than most known GSTs. SDS ⁄ PAGE of Peak 1 GST showed a subunit with a mass of 34 kDa, whereas Peak 2 GST was composed of two subunits of 36.8 and 32.6 kDa. This indicated that the Peak 1 enzyme was a homodimer and the Peak 2 enzyme a heterodi- mer. Native molecular masses for both would be about 68 kDa which is somewhat higher than the 60 kDa estimated from the Sephacryl column. This is consider- ably higher than the masses of 45–50 kDa that have been reported for toad, rabbit, rat or human liver [1,35,36]. The native molecular mass of some yeast (Y. lipolytica) GSTs, however, is 110 kDa [32]. Thus GSTs can vary widely in their subunit size. The larger molecular mass of turtle GSTs may arise as a result of post-translational modifications of the subunit pro- teins. Cloning and sequencing of turtle GST subunits would confirm their size, identify potential sites of post-translational modification and establish their place within the classification of nonmammalian GSTs. Cytosolic GSTs can generally be assigned to one of four classes (a, l, p and h) based on their pI and kin- etic characteristics [37]. The isozymes a, l, and p have basic, near-neutral, and acidic isoelectric points, respectively. Isoelectric focusing separated turtle liver GSTs into one major and one minor isozyme exhibit- ing activity with CDNB. Subsequent characterization of these peaks revealed that each had some cumene hydroperoxide activity. Class a isozymes exhibit strong activity with cumene hydroperoxide so it is likely that the major peak with basic pI values of 8.5–8.9 repre- sents an a class GST in turtle liver. The cumene hydro- peroxide activity exhibited by the minor isoform also suggests an a class although the pI (6.5–6.8) is more suggestive of l class which typically shows only minor activity towards cumene hydroperoxide. The classification of GSTs also depends on their responses to substrates and inhibitors. Like all GSTs, turtle liver GSTs showed activity with the nonspecific substrates including CDNB, 1,2-dichloro-4-nitroben- zene, and p-nitrophenylacetate (Table 3). The Peak 1 enzyme also used cumene hydroperoxide, a prominent a class substrate. GSTs in both peaks also showed good activity with p-nitrobenzylchloride which is spe- cifically used by the a1 isozyme (but not a2 in rats) but did not utilize ethacrynic acid (a p GST substrate), trans-4-phenyl-3-buten-2-one (a l class substrate) or 2- epoxy-3-(p-nitrophenoxy)propane (a l and p class sub- strate) [38]. Overall, then, the substrate specificities of the turtle liver GSTs are consistent with their classifi- cation as a class enzymes. Responses to inhibitors also generally supported this conclusion. Cibacron Blue causes greatest to least inhibition (lowest to highest I 50 )ofl, p and a isozymes, respectively [38]. Rose Bengal very strongly inhibits p class GSTs [38] while iodoacetamide, a reagent directed against thiol groups, is a nonspecific inhibitor of all classes. S-hexylglutathi- one shows highest to lowest inhibition of a, l and p GST isozymes, respectively [38]. Both Peak 1 and 2 GSTs showed low inhibition by Cibacron Blue although Peak 2 had a substantially lower I 50 than did Peak 1. Rose Bengal inhibition of turtle liver GSTs was in the range seen for inhibition of human a and l class GSTs [39]. Inhibition by S-hexylglutathione was the same for Peak 1 and Peak 2 isozymes and was stronger than the inhibition of human a GST [39]. Hence, both substrate and inhibition responses suggest that Peak 1 GST is an a class enzyme while Peak 2 may be an a-like isozyme without peroxidase activity. Peak 1 and 2 GSTs from turtle liver also differed in several other ways. Specific activities of turtle liver GSTs from crude extracts, using CDNB as a substrate, were comparable to those of rat liver (0.254 UÆmg )1 protein), brain (0.034 UÆmg )1 protein) and cultured glial cells (0.093 UÆmg )1 protein measured at 25 °C) [40]. Activities of the purified enzymes were compar- able to those found in human liver microsomes (21.6 and 3.8 UÆmg )1 protein for CDNB and cumene hydro- peroxide, respectively) [41]. Specific activities of puri- fied turtle liver GSTs were much lower than those for Xenopus liver GST (207 and 2.1 UÆmg )1 protein for CDNB and cumene hydroperoxide, respectively) [42], but were comparable to largemouth bass (7.0 and 0.5 UÆmg )1 protein for CDNB and cumene hydroper- oxide, respectively) [43] and salmonid species (17–28 and 22–37 UÆmg )1 protein for liver and kidney purified enzymes, respectively) [44]. Specific activities of Peak 1 enzyme using cumene hydroperoxide as a substrate were in the range of a GSTs found in human lung (1.84 UÆmg )1 protein) [45] which play a protective role in lipid peroxidation. Specific activities using cumene hydroperoxide were not as high as a GSTs found in human liver (10.6 UÆmg )1 protein) [38] or hamster liver (2.7–3.4 UÆmg )1 protein) [33]. Peak 1 GST showed a significantly lower K m for GSH than did the Peak 2 enzyme but the opposite was true of the K m for CDNB. The K m values for CDNB were higher than that of human lung GST (K m ¼ 0.033–0.042 mm) [45]. Both enzymes were inhibited by GSSG, the oxidized form of GSH, with I 50 values of 2–2.6 mm; however, this is about 100-fold higher than GSSG levels in vivo so inhibition by this compound, which accumulates W. G Willmore and K. B. Storey GST function in anoxia-tolerant turtles FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS 3609 during oxidative stress, may not be a significant influ- ence on enzyme activity in vivo. Both enzymes were also strongly inhibited by S -hexylglutathione; this strong inhibition (high affinity binding) may be the reason that high concentrations of GSH could not elute turtle liver GSTs from an S-hexylglutathione matrix. The two turtle liver GSTs responded differ- ently to various other inhibitors. For example, both iodoacetamide and Cibacron blue were poor inhibitors of Peak 1 GST but inhibited the Peak 2 enzyme with I 50 values 5–6 fold lower than those of the Peak 1 enzyme. Peak 2 GST was also more strongly inhibited by all chloride salts than was the Peak 1 isozyme. The Peak 1 and Peak 2 GSTs separated by hydroxyl- apatite chromatography did not differ in their pH optima or activation energies. Furthermore, the lack of a break in the Arrhenius relationship shows that enzyme structure and conformation was not compro- mised over the range of temperatures tested for either enzyme. This range covers the physiological tempera- ture range over which the animal normally functions. Peak 1 and Peak 2 GSTs both had a pH optimum of around 7.2. This pH optimum is on the acidic side of the pH optima of most known GSTs, including human l class enzymes [46]. The adaptive significance of this is that turtle GSTs may function normally under the acidotic cellular conditions that develop over the course of long-term anoxia. Previous studies [25] have shown that the blood pH of turtles can drop from 8 to 7 over the course of 130 days of anoxic submergence at 3 °C. Enzymes that are crucial for cell survival dur- ing metabolic depression would be required to function under acidic conditions. Turtle GSTs may represent one class of enzymes that function normally in the face of metabolic acidosis occurring during over-wintering. Likewise, keeping the pH optima of enzymes that are inactivated during anoxia high would provide a signal for shutting down entire biochemical pathways during hibernation. Determination of the pH optima of other purified turtle enzymes would reveal if this is a general mechanism of anoxia survival in freshwater turtles. In conclusion, the lower specific activities of GSTs in liver from anoxic turtles (using either CDNB or cumeme hydroperoxide as substrates) suggest a possible specific suppression of GST activity during anaerobiosis, perhaps caused by a stable modification of the protein. However, the elution profiles from the various columns demonstrate that anoxia exposure did not stimulate the synthesis of any new isozymic forms of GST. Based on SDS ⁄ PAGE as well as kinetic and inhibition properties, the Peak 1 GST eluted from hyd- roxylapatite was identified as a homodimeric a-class GST whereas the Peak 2 isozyme appears to be a heterodimeric a-class enzyme that lacked peroxidase activity. Reduced activities using both substrates were also documented for the anoxic, compared with the aerobic, enzyme forms separated by isoelectric focus- ing. For the Peak 2 enzyme retrieved by isoelectric focusing, the decrease in CDNB activity was much greater than the decrease in cumene hydroperoxide activity during anoxia, suggesting that peroxidase activity of this second isozyme was more conserved during turtle hibernation. The GST isozyme(s) in Peak 2 of isoelectric focusing may play an important role in removing the products of lipid peroxidation during anoxia as some oxidative stress may occur in turtle liver during anoxia (indicated by changes in the GSH ⁄ GSSG ratio) [16]. Conservation of GST activity in turtle liver also provides the animal the means to deal with oxidative stress during the reoxygenation after anoxic excursions. Experimental procedures Chemicals and animals All chemicals were purchased from Sigma Chemical Co. (St. Louis, MO) or Boehringer Mannheim Corp. (Montreal, Quebec, Canada) and were of the highest purity available. Winter acclimated adult red-eared sliders (T. s. elegans) were obtained from Wards Natural Science, Mississauga, Ontario and were maintained in large tanks of dechlori- nated water at 7 °C for at least 3 weeks prior to experimen- tation. Turtles had access to deep water and a dry platform supplied with a heat lamp and were fed ad libitum on a diet of trout pellets, lettuce and egg shells. Control (normoxic) turtles were sampled directly from the tank. Anoxia was imposed by submerging turtles at 5 °C in sealed tanks of deoxygenated water that had been bubbled previously with 100% nitrogen gas for 1 h [16]. A wire mesh placed 20 cm below the surface of the water pre- vented turtles from surfacing. Turtles were sampled after for 20 h of anoxic submergence. All animals were killed by decapitation and organ samples were removed quickly, frozen in liquid nitrogen and then transferred to )80 °C for storage. Preparation of tissue extracts and GST assay Frozen tissue samples were quickly weighed and homogen- ized 1 : 5 (w ⁄ v) in ice-cold 50 mm potassium phosphate buffer (pH 7.5, containing 1 mm EDTA) and with phenyl- methylsulfonyl fluoride (1 mgÆmL )1 ) added immediately before homogenizing using an Ultra-Turrax (Tekmar) tissue homogenizer. Homogenates were then sonicated for 10 s on ice with a Kontes microultrasonic cell disrupter and centri- fuged at 16 000 g for 15 min at 4 °C using an Eppendorf GST function in anoxia-tolerant turtles W. G Willmore and K. B. Storey 3610 FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS microcentrifuge. Supernatants were removed and desalted by passage through a small column (1 · 5 cm) of Sephadex G-25 (equilibrated in homogenizing buffer) with centrifuga- tion for 1 min in an IEC benchtop centrifuge at full speed [47]. GST was assayed by monitoring the formation of the thioether product of the reaction between reduced gluta- thione (GSH) and 1-chloro-2,4-dinitrobenzene (CDNB) (e ¼ 9.6 mM )1 ) at 340 nm [1]. Standard assay conditions in a 1 mL volume were 50 mm potassium phosphate (KPi) buffer (pH 6.5), 1 mm EDTA, 6 mm GSH and 1 mm CDNB. Blanks were run in the absence of either GSH or enzyme. One unit of activity is defined as the amount of enzyme that formed 1 lmol of product per min at 21 °C. Turtle liver GST purification The purification procedure was developed using liver extracts from control turtles but also used for purification of GSTs from anoxic liver. Four milliliters of crude supernatant was applied to a Matrix Red column (2 cm length · 2.8 cm diameter) equilibrated in homogenization buffer. A 30 mL void volume was collected (containing no GST activity) and then GST was eluted with a KCl gradient (0–1 m) with 37 · 1 mL fractions collected. Ten microliters of each frac- tion was assayed for GST activity. Peak fractions were pooled and concentrated in dialysis tubing (Spectra ⁄ Por molecular porous membrane tubing, relative molecular mass cut-off at 12–14 000, Spectrum Medical Industries, Inc., Houston, Texas, USA) surrounded by solid polyethylene glycol 20 000. The concentrated enzyme was then applied to a Sephacryl S-200 gel filtration column (45 cm length · 1.8 cm diameter) equilibrated in homogenization buffer (pH 6.0). The column was eluted with homogenization buffer and, after a 34 mL void volume, 40 · 1 mL fractions were gathered and assayed for GST activity. Peak fractions were pooled, concentrated as above and then applied to a hydroxylapatite column (2 cm length · 1.8 cm diameter) equilibrated in homogenization buffer (pH 6.0). A 3 mL void volume was collected and then a gradient of 0–250 mm KPi was run. Forty-five fractions of 1 mL each were collec- ted and assayed for GST activity. Peak fractions were com- bined and used for subsequent studies. Stability tests revealed that the pure enzyme retained 27–64% activity after 8 days at 4 °C (or 2 days of freezing at )80 °C). For long- term storage, glycerol was added to the pure enzyme to a final concentration of 50%. For native molecular mass deter- mination, the same Sephacryl S-200 column used for purifi- cation was calibrated using Blue Dextran to determine the void volume and six protein standards. Isoelectric focusing of turtle liver GSTs Samples of crude supernatant were subjected to isoelectric focusing [48] using an LKB 8101 isoelectric focusing column (110 mL) with a sucrose gradient containing pH 3.5–10 ampholines. The column was run for 14–18 h at 450 V constant voltage at 5 °C. After focusing, the column was drained into 2 mL fractions and the elution profile of enzyme activity and the pH gradient were measured. Peak fractions were tested for activity using both CDNB and cumene hydroperoxide substrates, the latter testing for Se-independent glutathione peroxidase activity which is catalyzed by GST. SDS/PAGE of turtle liver GSTs Peak fractions from the hydroxylapatite column were sub- jected to discontinuous SDS ⁄ PAGE. Samples of purified GSTs were mixed 1 : 1 (v:v) with 2· SDS ⁄ PAGE loading buffer (100 mm Tris ⁄ HCl, pH 6.8, 4% w ⁄ v SDS, 20% v ⁄ v glycerol, 0.2% w ⁄ v bromophenol blue) and boiled for 5 min. Turtle enzyme preparations were then loaded into wells of a 0.75 mm thick gel and run adjacent to broad range standards (Bio-Rad, Hercules, CA) and horse liver GST (Sigma, Oakville, Ontario) using 1· Tris-glycine run- ning buffer (3.02 gÆL )1 Tris-base, 18.8 gÆL )1 glycine, 0.1% w ⁄ v SDS). The stacking gel was 5% w ⁄ v acrylamide (30 : 0.8 w ⁄ w acrylamide:bisacrylamide) and the separating gel was 15% w ⁄ v acrylamide. The gel was run at 200 V for 1 h and then fixed in 30% v ⁄ v methanol, 10% v ⁄ v acetic acid for 1 h at room temperature on a rotary sha- ker. The gel was stained for 2 h in 0.25% Coomassie Bril- liant Blue R, 50% v ⁄ v methanol, and 7.5% v ⁄ v acetic acid, destained overnight in 30% methanol, 10% v ⁄ v acetic acid and then photographed using a Polaroid DS34 Direct Screen Instant Camera (Bio ⁄ Can Scientific, Mississauga, Ontario, Canada). Kinetic and inhibition characteristics of turtle liver GSTs Substrate affinity constants (K m ) for GSH, CDNB and cum- ene hydroperoxide as well as I 50 values (the concentration of inhibitor that reduces activity by 50%) for various salts and a range of known inhibitors of GST were determined for the Peak 1 and 2 enzymes from the hydroxylapatite column. I 50 determinations were performed at optimal GSH and CDNB concentrations. Specific substrates for known classes of GSTs were tested for activity (at 1 mm each) including ethacrynic acid, trans-4-phenyl-3-buten-2-one, 1,2-epoxy- 3-(p-nitrophenoxy) propane, 1,2-dichloro-4-nitrobenzene, p-nitrobenzylchloride, and p-nitrophenylacetate. Temperature and pH dependence of turtle liver GSTs The temperature and pH dependence of Peak 1 and Peak 2 GSTs were assessed in KP i buffer under optimal substrate concentrations. Temperature dependence was assessed over W. G Willmore and K. B. Storey GST function in anoxia-tolerant turtles FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS 3611 [...]... Comparative study of the purification and characterization of the cytosolic glutathione S-transferases from two salmonid species: Atlantic salmon (Salmo salar) and brown trout (Salmo trutta) Comp Biochem Physiol C 131, 207–213 Singhal SS, Saxena M, Ahmad H, Awasthi S, Haque AK & Awasthi YC (1992) Glutathione S-transferases of human lung: characterization and evaluation of the protective role of the a-class... Bladeren PJ (1992) Purification and characterization of eight glutathione S-transferase isoenzymes of hamster Comparison of subunit composition of enzymes from liver, kidney, testis, pancreas and trachea Biochem J 286, 383–388 34 Addya S, Mullick J, Fang J-K & Avadhani NG (1994) Purification and characterization of a hepatic mitochon- FEBS Journal 272 (2005) 3602–3614 ª 2005 FEBS GST function in anoxia-tolerant. .. (1993) Glutathione S-transferases, structure, regulation, and therapeutic implications J Biol Chem 268, 11475–11478 8 Sheehan D, Meade G, Foley VM & Dowd CA (2001) Structure, function and evolution of glutathione transferases: implications for classification of non-mammalian members of an ancient enzyme superfamily Biochem J 360, 1–16 9 Daniel V (1993) Glutathione S-transferases: gene structure and regulation... the freshwater turtle, Pseudemys scripta elegans Comp Biochem Physiol B 58, 377–379 29 Nishinaka T, Kodaka R, Nanjo H, Terada T, Mizoguchi T & Nishihara T (1992) Purification and characterization of glutathione S-transferase isozymes in dog lens Int J Biochem 24, 1737–1742 30 Ji X, Tordova M, O’Donnell R, Parsons JF, Hayden JB, Gilliland GL & Zimniak P (1997) Structure and function of the xenobiotic... substrate-binding site and location of a potential non-substrate-binding site in a class p glutathione S-transferase Biochem 36, 9690– 9702 31 Kolobe D, Sayed Y & Dirr HW (2004) Characterization of bromosulphophthalein binding to human glutathione S-transferase A1–1: thermodynamics and inhibition kinetics Biochem J 382, 703–709 32 Foley V & Sheehan D (1998) Glutathione S-transferases of the yeast Yarrowia...GST function in anoxia-tolerant turtles the range from 5 to 40 °C using increments of 3 °C and analyzed using an Arrhenius plot of log V vs 1 ⁄ K For pH curves, different buffer pH values were obtained by mixing appropriate amounts of KH2PO4 and K2HPO4 with further adjustment using HCl or KOH as needed Assay pH in cuvettes was measured prior to and immediately following the enzyme assay Protein... anoxic softshell and painted turtles Physiol Biochem Zool 73, 290–297 27 Suarez RK, Doll CJ, Buie AE, West TG, Funk GD & Hochachka PW (1989) Turtles and rats: a biochemical comparison of anoxia-tolerant and anoxia-sensitive brains Am J Physiol 257, R1083–R1088 28 Simon LM, Liu J, Theodore J & Robin ED (1977) Some comparative aspects of the organ distribution of superoxide dismutase activity in the freshwater... FEBS W G Willmore and K B Storey 21 Ultsch GR (1989) Ecology and physiology of hibernation and overwintering among freshwater fishes, turtles, and snakes Biol Rev 64, 435–516 22 Jackson DC (2000) Living without oxygen: lessons from the freshwater turtle Comp Biochem Physiol A 125, 299–315 23 Jackson DC & Heisler N (1982) Plasma ion balance of submerged anoxic turtles at 3°C: the role of calcium lactate... Mechanisms and Biological Significance (Sies, H & Ketterer, B, eds), pp 157–174 Academic Press, London 5 Pickett CB & Lu AYH (1989) Glutathione S-transferases: gene structure, regulation and biological function Annu Rev Biochem 58, 743–764 6 Coles B & Ketterer B (1990) The role of glutathione and glutathione transferases in chemical carcinogenesis Crit Rev Biochem Mol Biol 25, 47–70 3612 W G Willmore and K... 35 36 37 38 39 40 41 42 43 44 45 46 drial glutathione S-transferase exhibiting immunochemical relationship to the a-class of cytosolic enzymes Arch Biochem Biophys 310, 82–88 Aceto A, Dragani B, Bucciarelli T, Sacchetta P, Martini F, Angelucci S, Amicarelli F, Miranda M & Di Ilio C (1993) Purification and characterization of the major glutathione transferase from adult toad (Bufo bufo) liver Biochem . Purification and properties of the glutathione S-transferases from the anoxia-tolerant turtle, Trachemys scripta elegans William G. Willmore and Kenneth. study of the purification and char- acterization of the cytosolic glutathione S-transferases from two salmonid species: Atlantic salmon (Salmo salar) and