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
Trang 1Purification 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.
Trang 2and 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
(UÆmg)1protein) Fold purification % yield (UÆmg)1protein) Fold purification % yield
Hydroxylapatite
(Peak 1)
Hydroxylapatite
(Peak 2)
Trang 3S-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 H2O2as 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.
Trang 4liver 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 H2O2 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 activactiv-ity 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
(Ea) (kJÆmol)1)
(subunit 1)
32 600 (subunit 2) Specific activity using CDNB (UÆmg protein)1)
Specific activity using cumene hydroperoxide (UÆmg protein)1)
Isoelectric Focusing (Peak 1)
Isoelectric Focusing (Peak 2) pI
CDNB activity (UÆmL)1of peak fraction assayed)
Cumene hydroperoxide activity (UÆmL)1of peak fraction assayed)
Ratio of cumene hydroperoxide to CDNB activities
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.
Trang 5Kinetic 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)1protein
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)1protein 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 Km that was only 63% of the Peak 2-value By contrast, Peak 2 GST had a greater affinity for CDNB with a Km 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 (m M ) 0.38 ± 0.019 0.60 ± 0.064 a
K m CDNB (m M ) 1.7 ± 0.15 1.14 ± 0.040a
KmCumene hydroperoxide (m M )
0.11 ± 0.021 No activity
I50Cibacron Blue (lM) 48 ± 0.97 8.4 ± 0.38 b
I50Rose Bengal (lM) 0.31 ± 0.016 0.47 ± 0.085
I 50 S-hexylglutathione (l M ) 0.31 ± 0.036 0.39 ± 0.19
I50iodoacetamide (m M ) 40 ± 0.46 8.7 ± 0.33 b
I50KCl ( M ) 0.33 ± 0.039 0.18 ± 0.020 a
I 50 NaCl ( M ) 0.332 ± 0.0341 0.17 ± 0.012a
I50Na2SO4( M ) No inhibition 0.19 ± 0.045
I50NH4Cl ( 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 )
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.
Trang 6cumene hydroperoxide as a substrate but Peak 2 GST
could not Neither enzyme showed activity with H2O2
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 I50values 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, NH4Cl) with I50 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
(5–40C) Calculated activation energies (Ea) were
36 ± 2.2 and 40 ± 3.7 kJÆmol)1for 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 20C and at least
3 months at 3C [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 Q10 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,
Ca2+, and Mg2+ 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 Q10 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 valval-ues Peak 1 (d, lines) and Peak 2 (s, dotted lines) GST had pH optima of 7.20 and 7.21, respectively.
Trang 7activities 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 purificafiltra-tion 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)1protein 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.
Trang 8toad (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-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
I50) of l, 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 I50than 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 25C) [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)1protein 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)1protein) [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)1protein) [38] or hamster liver (2.7–3.4 UÆmg)1 protein) [33] Peak 1 GST showed a significantly lower Km for GSH than did the Peak 2 enzyme but the opposite was true of the Km for CDNB The Km values for CDNB were higher than that of human lung GST (Km¼ 0.033–0.042 mm) [45] Both enzymes were inhibited by GSSG, the oxidized form of GSH, with I50 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
Trang 9during 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
I50 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 3C 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 7C 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
5C 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 4C using an Eppendorf
Trang 10microcentrifuge 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)
(e¼ 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 21C
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 fracfrac-tions 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 4C (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 5C 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 (Km) for GSH, CDNB and cum-ene hydroperoxide as well as I50values (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
I50 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 KPibuffer under optimal substrate concentrations Temperature dependence was assessed over