(Received for publication, June 20, 1996, and in revised form, December 5, 1996)
From the Section of Molecular Therapeutics,
Department of Experimental Pediatrics, University of Texas M. D.
Anderson Cancer Center, Houston, Texas 77030 and the ¶ Department
of Medicinal Chemistry and Research Institute of Pharmaceutical
Sciences, School of Pharmacy, University of Mississippi,
University, Mississippi 38677
We report the isolation of three full-length
cDNAs corresponding to the mRNAs of closely related glutathione
S-transferase (GST) Pi genes, designated
hGSTP1*A, hGSTP1*B, and hGSTP1*C,
expressed in normal cells and malignant gliomas. The variant cDNAs
result from A G and C
T transitions at nucleotides +313 and
+341, respectively. The transitions changed codon 104 from ATC (Ile) in
hGSTP1*A to GTC (Val) in hGSTP1*B and
hGSTP1*C and changed codon 113 from GCG (Ala) to GTG (Val)
in hGSTP1*C. Both amino changes are in the
electrophile-binding active site of the GST Pi peptide. Computer
modeling of the deduced crystal structures of the encoded peptides
showed significant deviations in the interatomic distances of critical
electrophile-binding active site amino acids as a consequence of the
amino acid changes. The encoded proteins expressed in Escherichia
coli and purified by GSH affinity chromatography showed a 3-fold
lower Km (CDNB) and a 3-4-fold higher Kcat/Km for the
hGSTP1*A encoded protein than the proteins encoded by
hGSTP1*B and hGSTP1*C. Analysis of 75 cases
showed the relative frequency of hGSTP1*C to be 4-fold
higher in malignant gliomas than in normal tissues. These data provide
conclusive molecular evidence of allelopolymorphism of the human GST Pi
gene locus, resulting in active, functionally different GST Pi
proteins, and should facilitate studies of the role of this gene in
xenobiotic metabolism, cancer, and other human diseases.
Glutathione S-transferases (GSTs; EC 2.5.1.18)1 are dimeric proteins encoded by a family of distinct genes with limited structural homologies (1-4). Based on their biochemical, immunological, and structural properties, the currently known soluble human GSTs are categorized into four main classes, namely Alpha, Mu, Pi, and Theta (5). GSTs are involved in many cellular functions, the best characterized of which is their role as phase II enzymes in which they catalyze the S-conjugation of glutathione (GSH) with a wide variety of electrophilic compounds, including many mutagens, carcinogens, anticancer agents, and their metabolites (1-4, 6-13).
High GST Pi gene expression has been associated with malignant transformation, tumor drug resistance, and poor patient survival (9-17). In many human tumors and pre-neoplastic lesions, the GST Pi protein is over-expressed, even though in the corresponding normal tissues the protein either is absent or else is expressed at very low levels. The GST Pi gene has been mapped to a relatively small region of chromosome 11q13 in which is localized a number of cancer-associated genes and proto-oncogenes, including bcl1/prad1, int2, hstf1, and sea, some of which have been reported to be co-amplified with the GST Pi gene in some tumors (18, 19). In malignant gliomas, a positive correlation has been demonstrated between the level of GST Pi expression by immunocytochemistry and both the histological grade of the tumor and patient survival2 (19-21). Earlier data from our laboratory using human glioma cell lines have also shown an association between high GST Pi protein expression and 2-chloroethylnitrosourea resistance (22).
To date, the nucleotide sequences of the complete human GST Pi cDNA and GST Pi gene reported from different laboratories (23-26) have been identical, leading to the notion that only one human GST Pi gene exists (5). The isolation of two different GST Pi proteins from human placenta (27), however, strongly indicates that genetic polymorphism may exist in the human GST Pi gene locus. In the present study, we describe the isolation and molecular characterization of full-length cDNAs of three polymorphic GST Pi genes that are expressed in human malignant glioma cells and in other normal cells and tissues. The intracellular stabilities of the variant mRNAs were examined, and the proteins encoded by the variant cDNAs were expressed in Escherichia coli, purified by GSH affinity chromatography, and used in enzyme kinetic analyses to examine the functional consequences of the structural differences. Computer modeling was performed following introduction of the amino acid changes into the three-dimensional structure of the placental GST Pi protein that had been previously isolated and co-crystallized with S-hexylglutathione (28). The frequency of each of the gene variants in primary specimens and cell lines of malignant gliomas and in normal brain, normal placenta, and peripheral blood lymphocytes was investigated, and the concordance of the observed genotype and phenotype was determined by DNA- and RNA-PCR of representative specimens.
Peripheral blood lymphocytes were isolated from the peripheral blood of healthy donors using a single step Ficoll-Hypaque gradient centrifugation (29). Full-term human placentas were obtained after normal vaginal delivery. Primary malignant glioma specimens were obtained incidental to surgery. All specimens were acquired on institutionally approved protocols. Glioma cell lines were established in our laboratories from primary specimens, as we previously described (30), and were in less than 40 in vitro passages. Unless otherwise stated, all chemicals were from Sigma. Restriction enzymes and biochemicals were from Boehringer Mannheim, and PCR reagents from Perkin-Elmer. Polyclonal antibodies against human GST Alpha, GST Mu, and GST Pi were obtained from Biotrin Inc. (Dublin, Ireland).
cDNA Library Synthesis and ScreeningPolyadenylated RNA
was purified on oligo(dT) cellulose columns from total RNA isolated
from glioma cells by the acid guanidinium thiocyanate phenol-chloroform
method (31) and used in gt 11 cDNA library synthesis according
to the modified procedure of Gubler and Hoffman (32), using the
protocol of Clontech (Palo Alto, CA). After first and second strand
cDNA synthesis, the cDNA pool was size-fractionated at a 500-bp
cut-off to reduce the proportion of truncated GST Pi cDNAs. The
double-stranded cDNAs were then blunt-ended with T4 DNA polymerase,
methylated with EcoRI methylase, and ligated to
EcoRI linkers. Following EcoRI digestion to
remove excess linkers, the cDNAs were ligated into bacteriophage
gt 11 EcoRI arms and packaged using the Gigapack II Gold
packaging extract (Stratagene, La Jolla, CA). Serial dilutions of the
resulting cDNA library was screened using a rapid PCR screening
procedure (33), which we had previously modified (34) by the use of the
ExpandTM PCR system (Boehringer Mannheim). Positive cDNA pools were plated on E. coli strain Y109 Or
,
screened with a 32P-labeled GST Pi cDNA probe, and GST
Pi positive clones were amplified. The DNA was isolated, purified, and
sequenced.
Nucleotide sequencing was performed with the 35S-labeled dideoxynucleotide chain termination method (35) using the circumvent thermal cycle sequencing protocol (New England Biolabs, MA) either directly or after subcloning into Bluescript phagemid II. Sequencing primers were designed to overlap internal GST Pi cDNA regions as well as to the vector. Each clone was sequenced twice in both directions.
PCRAmplification of GST Pi cDNAs for subcloning and for structural analyses was performed by PCR on templates of isolated GST Pi cDNAs or first strand cDNA synthesized from total RNA prepared, as described earlier, from cells and tissues. For the latter, first strand cDNA was synthesized by reverse transcription (RT) in a 20-µl reaction mixture containing 100 ng of reverse primer, 1 µg of total RNA, 250 µM dNTP, 3.2 mM sodium pyrophosphate, and 0.4 unit/ml each of placental RNase inhibitor and avian myeloblastosis virus reverse transcriptase. After 2 min at 94 °C, followed by 1 h at 42 °C, the mixture was heated to 95 °C for 2 min and rapidly cooled to 25 °C. The PCR reaction mixture contained approximately 200 ng of template cDNA (or 20 µl of the first strand cDNA reaction mixture), 500 ng of the appropriate forward and reverse GST Pi primers, 200 µM dNTP, 1.5 mM MgCl2, 0.025 unit/ml of Amplitaq polymerase and PCR buffer (to 100 µl). PCR amplification was performed for 34 cycles of 94 °C (1 min), 58 °C (2 min), and 72 °C (3 min).
Restriction Site Mapping of Variant GST Pi cDNAsComputerassisted analysis of restriction
endonuclease sites in the variant GST Pi cDNAs was used to identify
restriction enzymes, which had gained or lost restriction motifs as a
consequence of the nucleotide transitions identified from the sequence
data of the cDNAs. From these, two low frequency cutting enzymes,
MaeII and XcmI, were selected for restriction
site mapping of the cDNAs. A 484-bp cDNA fragment, spanning
positions +112 to +596 of the GST Pi cDNA, was amplified by RT-PCR
from each specimen, as described earlier, using the primers:
5-ACGTGGCAGGAGGGCTCACTC-3
(forward) and 5
-TACTCAGGGGAGGCCAGCAA-3
(reverse). The cDNA product was purified and after restriction with
MaeII and XcmI was electrophoresed in 2% agarose
and stained with 0.5 µg/ml ethidium bromide, and the restriction
pattern was photographed under ultraviolet illumination. Fragment sizes
were determined relative to marker DNA (
X174 DNA-HaeIII digest) and the structures were confirmed, in representative cases, by
nucleotide sequencing.
Oligodeoxynucleotide probes were designed to cover the
region +312 to +342 of the GST Pi cDNA and to contain the
nucleotide changes specific to each of the GST Pi cDNAs. The probe
sequences were: 5-CATCTCCCTCATCTACACCAACTA-TGAGGCG-3
(hGSTP1*A), 5
-CG*TCTCCCTCATCTACACCAACTATGAGGCG-3
(hGSTP1*B), and 5
-CG*TCTCCCTCATCTACACCAACTATGAGGT*G-3
(hGSTP1*C). The asterisks indicate the transition
nucleotides. Each probe was end-labeled with [
-32P]ATP
to a specific activity of approximately 4 × 106
cpm/ml using T4 polynucleotide kinase. Southern hybridization was
performed using standard methods (36) but at the higher, more stringent
temperature of 65 °C. Following autoradiography photodocumentation,
the membranes were stripped off the hybridized oligonucleotide probe
and rehybridized with the next probe.
To determine the concordance
between the GST Pi genotype and phenotype in a given specimen, we
examined the nucleotide sequences of exons 5 and 6, which contained the
nucleotide transitions observed in the GST Pi mRNA variants. For
this, nine representative samples were selected to represent each of
the three GST Pi mRNA variants. Using genomic DNA, a 305-bp DNA
fragment spanning nucleotides +1219 to +1524 of the GST Pi gene and
containing the entire exon 5 and its flanking regions in introns 4 and
6 as well as a 321-bp fragment spanning nucleotides +2136 to +2467
including all of exon 6 and its flanking regions in introns 5 and 7 were amplified by PCR. Primers for these amplifications were as
follows: 5-CCAGGCTGGGGCTCACAGACAGC-3
(forward) and
5
-GGTCAGCCCAAGCCACCTGAGG-3
(reverse) for exon 5 and
5
-TGGCAGCTGAAGTGGACAGGATT-3
(forward) and
5
-ATGGCTCACACCTGTGTCCATCT-3
(reverse) for exon 6.
Full-length cDNAs of the GST Pi gene variants were
amplified by PCR from the respective GST Pi cDNA clones. To
facilitate directional subcloning into the expression vector, forward
and reverse primers were designed to contain restriction sites for EcoRI and XbaI at the 5- and 3
-termini,
respectively, of the resulting cDNA probes. Both restriction sites
are absent in the GST Pi cDNA. After nucleotide sequencing to
ensure the absence of PCR-induced mutations, the cDNA products were
ligated into the pBK-CMV phagemid vector (Stratagene, La
Jolla, CA), in which prokaryotic gene expression is driven by the
lac promoter and eukaryotic expression by the immediate
early CMV promoter. Strain XL1 Blue bacteria were
transformed with the resulting cDNA constructs, screened for
positive clones and bacterial cultures of these grown overnight in LB
broth. Isopropyl-
-thiogalactopyranoside was added to 1 mM in the last hour of culture. The bacteria were pelleted by centrifugation, resuspended in 50 mM Tris-HCl (pH 7.4)
containing 2 µg/ml lysozyme for 1 h at 37 °C, and
ultrasonicated. The crude homogenates were centrifuged at 30,000 × g for 20 min, and the supernatants were concentrated
10-fold by membrane filtration at a 10-kDa molecular mass cut-off.
Protein content of the supernatants was determined (37) and GST enzyme
activity was assayed as described previously (22) using CDNB as
substrate. SDS-PAGE and Western blotting with polyclonal antibodies
against human GST Alpha, Mu, and Pi were performed as described
previously (22).
Functional consequences of the structural differences in the GST Pi variant proteins was determined by the differential ability of the recombinant GST Pi proteins to catalyze the conjugation of GSH with 1-chloro-2,4-dinitrobenzene (CDNB), a universal GST substrate. After expression in E. coli, the three GST Pi proteins were purified by GSH affinity chromatography on S-hexylglutathione linked to epoxy-activated Sepharose 6B as described previously (38) and then used for enzyme kinetic analysis.
For determining the enzyme kinetic constants for CDNB, reaction mixtures (25 °C) in 100 mM potassium phosphate buffer (pH 6.5) contained CDNB (over the range of 0.5-4 mM), 2.5 mM GSH, and 0.015 unit of purified GST Pi protein. Similar reaction mixtures were set up for determining the Km of GSH as a co-substrate, except that the concentration of GSH was varied from 0 to 5 mM, with the concentration of CDNB maintained at 0.1 mM. Changes in absorbance were monitored at 340 nm over two minutes in a Beckman DU 60 spectrophotometer equipped with a kinetic module and used to compute reaction rates. The rates of the spontaneous reactions of GSH with CDNB, determined with reaction mixtures in which the GST Pi enzyme was replaced with buffer, were subtracted from the rates of the enzyme catalyzed reactions. The resulting reaction rates were used to generate double reciprocal plots. Km, Vmax, and Kcat values were determined using standard methodology (39).
Computer Structural Modeling of Variant GST Pi ProteinsTo determine possible effects of the amino acid changes on the three-dimensional architecture of the three GST Pi proteins, the x-ray crystallographic co-ordinates (2.8 Å resolution) of the placental GST Pi (GSTP1a) co-crystallized with S-hexylglutathione (28) were imported from the Brookhaven Protein Data Bank into the SYBYL molecular modeling program (version 6.2, 1995; Tripos Associates, St. Louis, MO) running on a Silicon Graphics Indigo 2 workstation (IRIX 5.2, 64 MB). GSTP1b was created by substituting Ile with Val at amino acid residue 104, and GSTP1c was obtained by substituting Val for Ile at amino acid residue 104 and Val for Ala at 113, using the SYBYL BIOPOLYMER module. Each monomeric subunit of 209 amino acids was energy minimized using the amber all-atom force field in SYBYL. The energy minimized structures were superimposed using the match function of SYBYL. Changes in atomic co-ordinates and in interside chain distances of amino acid residues lining the putative H-site were examined to determine how these structural changes might predict changes in functional activity.
Stability of Variant GST Pi Gene TranscriptsTranscriptional block by actinomycin D has been shown in previous studies to be a reliable method with which to determine mRNA turnover and stability in cells (40). We therefore applied this technique to investigate differences in the intracellular stability of the transcripts of the different GST Pi genes. Three glioma cell lines expressing different GST Pi gene variants were grown to approximately 70% confluency and then refed with fresh culture medium containing 5 µg/ml actinomycin D. Controls were similarly set up but without actinomycin D treatment. At 0, 6, 12, and 24 h after actinomycin D exposure, total RNA was isolated from each culture and electrophoresed as described earlier at 7.5 µg of RNA/lane. After electrophoresis, the gels were stained with ethidium bromide, viewed under ultraviolet light to ensure equal RNA loading of the lanes, and transferred to nylon filters. Northern hybridization for GST Pi transcript levels was performed as described previously. Hybridization bands were quantitated by densitometry and plotted against time.
Thermostability of Variant GST Pi ProteinsTo further
determine the effects of the amino acid changes on the variant GST Pi
proteins, we compared the thermal stabilities of the enzymatic function
of the three variant GST Pi proteins based in part on the expected
differences in -helix stability of the region of the GST Pi peptides
containing the amino acid changes and on a previous study (41) that
showed that a recombinant GST Pi enzyme corresponding to GSTP1b-1b,
created by site-directed mutagenesis, was functionally more heat-stable
than the parent enzyme. Each variant GST Pi protein was incubated at
approximately 0.1 unit/ml at 45 °C in phosphate-buffered saline (pH
7.2) in a water bath. Every 15 min, over 1 h, a 50-µl aliquot
was removed from each incubate, and total GST activity was determined
as described previously using CDNB as substrate. Residual GST activity
was computed relative to the activity of controls maintained at
25 °C and plotted against time. SDS-PAGE and Western blotting were
performed as described earlier to determine if degradation of the GST
Pi peptides had occurred during the incubation.
We
analyzed approximately 3 × 106 plaque forming units
of each cDNA library using the rapid PCR procedure and obtained GST Pi positive clones, from which selected clones were subjected to
secondary and tertiary screening and subsequent sequencing. Three
clones, Pi 3A-7, Pi 3B-2, and Pi 3C-9, corresponding to each of three
GST Pi mRNA variants, were obtained containing full-length GST Pi
cDNAs and including both 5- and 3
-termini. The cDNAs were
designated hGSTP1*A, hGSTP1*B, and
hGSTP1*C, in accordance with the proposed nomenclature for
allelic GST gene variants (5). The nucleotide transitions giving rise
to the three cDNA variants are shown in the sequence
autoradiographs in Fig. 1 . The sequencing strategy is
shown in Fig. 2A, and the complete nucleotide
and deduced amino acid sequences of the cDNAs are shown in Fig.
2B. Each cDNA contained an open reading frame of 630 nucleotides, encoding 210 amino acids, including the initiating
methionine. hGSTP1*A was completely identical in nucleotide
sequence to the previously reported human GST Pi cDNA (23, 24). It
consisted of 712 nucleotides, of which 9 were in the 5
noncoding
region. hGSTP1*B differed from hGSTP1*A by having
an A
G transition at nucleotide +313, thus changing codon 104 from
ATC (Ile)
GTC (Val). Of the 719 nucleotides, 12 were in the 5
noncoding region. hGSTP1*C was characterized by two active
nucleotide transitions, the A
G transition at +313 observed in
hGSTP1*B and a C
T transition at +341, resulting in
changes of ATC (Ile)
GTC (Val) GTC in codon 104 and of GCG (Ala)
GTG (Val) in codon 113. hGSTP1*C consisted of 723 nucleotides, 13 of which were 5
of the ATG start codon. The 3
noncoding regions of all three cDNAs were similar and contained the
AATAAA polyadenylation signal at +689 to +696. In 8 cases examined,
there was complete concordance between the nucleotide sequences of
exons 5 and 6 in the genomic DNA and that of the corresponding regions
of the mRNA. In addition to the transitions at nucleotide positions
+313 and +341, a silent C
T transition at +555 was observed in
hGSTP1*B and hGSTP1*C. This transition, also
previously observed in the GST Pi cDNA isolated by Moscow et
al. (24), does not alter the encoded amino acid (serine) in the
affected codon 185.
Restriction Endonuclease Mapping
The partial restriction
endonuclease maps of the 484-bp cDNA region containing the
nucleotide transitions in the three GST Pi cDNAs are shown in Fig.
3A. The A G transition at +313 created a
MaeII recognition sequence (A
CGT) in hGSTP1*B
and hGSTP1*C, whereas the C
T transition at +341
resulted in the gain of an XcmI recognition sequence
(CCANNNNN
NNNNTGG) in hGSTP1*C. The 484-bp GST Pi cDNA
was amplified by RNA-PCR from representative normal and malignant
specimens and after purification was subjected to MaeII or
XcmI digestion. As expected from the restriction map (Fig.
3A), MaeII or XcmI did not restrict
the cDNAs from specimens that express hGSTP1*A alone
(lanes 1, 2, 3, and 8),
because both restriction endonuclease sites are absent in this cDNA
variant. However, following MaeII digestion, cDNAs from
specimens that expressed hGSTP1*B (lane 6) and
hGSTP1*C genes (lane 9) alone were cleaved
completely to yield the two fragments, 199 and 285 bp in size, as
expected from the restriction endonuclease map in Fig. 3A.
Similarly, XcmI digestion of hGSTP1*C cDNA
(lane 9) yielded the expected 224- and 260-bp fragments. In contrast to the specimens that expressed only one GST Pi gene variant, specimens expressing two different GST Pi genes yielded, upon digestion of their
cDNAs with MaeII or XcmI, restriction
fragments representative of the mixture of mRNAs. Thus, cDNA
from specimens that express a mixture of hGSTP1*A and
hGSTP1*B genes (lane 5) is partially restricted
by MaeII and is unrestricted by XcmI. On the
other hand, cDNAs from specimens expressing a mixture of
hGSTP1*A and hGSTP*1C genes (lanes 4,
7, and 10) are partially restricted by both
MaeII and XcmI. Fig. 4 shows
nucleotide sequence autoradiographs obtained by direct sequencing of
cDNAs obtained by RT-PCR from specimens expressing a mixture of
hGSTP1*A and hGSTP1*B (left panel) and
hGSTP1*A and hGSTP1*C (right
panel).
Detection of Expressed GST Pi Genes by Southern Hybridization
The results of the Southern hybridizations of the
cDNAs generated by RT-PCR from specimens expressing different GST
Pi genes are shown in Fig. 5. The lane designations are:
lanes 1 and 2, hGSTP1*A; lane
3, hGSTP1*B; lane 4, hGSTP1*A and
hGSTP1*C; lanes 5 and 6,
hGSTP1*C. Oligonucleotide probes specific for both
hGSTP1*A and hGSTP1*C genes hybridized strongly
to their respective target DNAs. When either of the two GST Pi gene
variants was expressed alone in a given specimen, as in lanes
1 and 2 (hGSTP1*A) and lanes 5 and 6 (hGSTP1*C), no significant
cross-hybridization with each other's specific probe was observed. In
contrast, GSTP1*B (lane 3) hybridized not only to
its own specific probe but also to the probes specific for
hGSTP1*A and hGSTP1*C. Similar to its target DNA,
the hGSTP1*B specific probe also showed little specificity and hybridized to all three GST Pi DNA variants as shown by the positive bands in lanes 1-6 of Fig. 5 (second row of
Southern blots). These results demonstrate that under our hybridization conditions hGSTP1*A and hGSTP1*C can be detected
by Southern analysis with specific oligonucleotide probes but that
detection of hGSTP1*B is limited by a high degree of probe
and DNA target cross-hybridization, possibly because
hGSTP1*B differs from hGSTP1*A and
hGSTP1*C by only one nucleotide, whereas the latter two
differ from each other by two nucleotides.
Expression, Purification, and Functional Analysis of Variant GST Pi Proteins
The GST Pi peptides encoded by the three GST Pi genes
were designated GSTP1a, GSTP1b and GSTP1c, as previously recommended (5). The cDNAs were expressed in E. coli, and the
expressed proteins were purified by GSH affinity chromatography. We
achieved approximately a 200-fold purification for all three GST Pi
proteins, with yields of 78, 63, and 72% for GSTP1a-1a, GSTP1b-1b, and
GSTP1c-1c, respectively. Lanes a, b, and
c in panels A, B, and C of
Fig. 6 represent the proteins encoded by
hGSTP1*A, hGSTP1*B, and hGSTP1*C, respectively. SDS-PAGE of each isolated protein showed a single band
after Coomassie Blue staining (Fig. 6B), and Western
analysis showed that they were all immunoreactive with anti-GST Pi
antibodies (Fig. 6C) but not with antibodies against GST Mu
or GST Alpha (data not shown). The enzyme kinetic analysis of the GSH
affinity purified recombinant proteins yielded, for CDNB,
Km and Vmax values of 0.98 mM ± 0.06 and 62.53 ± 5.3 µmol·min1·mg
1, respectively, for
GSTP1a-1a, 2.7 ± 0.23 mM and 50.25 ± 4.8 µmol·min
1·mg
1 for GSTP1b-1b, and
3.1 ± 0.17 mM and 53.81 ± 6.3 µmol·min
1·mg
1 for GSTP1c-1c. For the
co-substrate GSH, the Km values were 0.45 ± 0.05, 0.42 ± 0.03, and 0.51 ± 0.06 mM,
respectively, for GSTP1a-1a, GSTP1b-1b, and GSTP1c-1c.
Kcat (CDNB) values were 53.99 ± 5.1 s
1, 45.1 ± 3.3 s
1, and 41.01 ± 4.7 s
1, respectively, for GSTP1a-1a, GSTP1b-1b, and
GSTP1c-1c. The utilization ratio
(Kcat/Km) of GSTP1a-1a was 3- and 4-fold higher than that for GSTP1b-1b and GSTP1c-1c, respectively.
These results are summarized in Table I.
|
Data from the
previously reported x-ray crystallographic structure of GSTP1a (28, 45)
have shown the key amino acid residues lining the putative H-site of
human placental GST Pi to consist of Tyr7,
Tyr108, Val10, Val35,
Phe8, and Gly205. We modeled the effects of the
amino acid substitutions in the variant GST Pi proteins on the
structure of the resulting peptides, particularly in the H-site pocket.
The superimposed energy-minimized structures of the H-site region are
shown in Fig. 7A for GSTP1a and GSTP1b and in
Fig. 7B for GSTP1a and GSTP1c. The substitutions of
Val104 Ile104 in GSTP1b and GSTP1c and of
Val113
Ala113 in GSTP1c caused significant
deviations in the atomic co-ordinates of side chains of the key H-site
residues (Table II). The deviations caused by the
Val104 for Ile104 substitution are magnified in
the same direction as but to a lesser degree than those caused by the
Ala113 to Val113 substitution. The highest
deviations involved the side chains of Tyr108 and
Tyr7, which, relative to GSTP1a, had shifted by 0.153 and
0.116 Å in GSTP1b and 0.242 and 0.185 Å in GSTP1c. Phe8
was the least affected by the amino acid changes. Overall, the deviations in atomic co-ordinates in the active site residues were
larger going from GSTP1a to GSTP1c than from GSTP1b to GSTP1c.
|
Changes in interside chain distances within the three-dimensional structure of the variant peptides are summarized in Table III, and are also evident in Fig. 7 (A and B). The distances between the residues Tyr108 and Val10 and between Tyr108 and Val35 decreased progressively going from GSTP1a to GSTP1b and GSTP1c. From the superimposed structures in Fig. 7 (A and B), it is apparent that in GSTP1b and GSTP1c, the methyl group of Val104 proximal to Tyr108 is closer to several of the putative H-site residues than is the secondary methyl group of Ile104 in GSTP1a. The orientation of the methyl group of Val104 in GSTP1b and GSTP1c toward Tyr108 causes a decrease in the distances between Val104 and both Val10 and Val35 and a restriction of the region of the active site bordered by Tyr108, Val10, and Val35. The replacement of Ile104 with the less bulky Val104 also opens up the region lined by Tyr7, Tyr10, and Phe8. The distances between side chains of the paired residues Tyr7 and Val10 is shorter in GSTP1a than in both GSTP1b and GSTP1c, whereas the distances between Tyr7 and Tyr108 and between Tyr7 and Phe8 are longer.
|
The
intracellular decay of the three variant GST Pi mRNAs in malignant
glioma cell lines, each of which expressed different GST Pi mRNAs,
was determined following inhibition of de novo RNA synthesis
by exposure of the cells to actinomycin D. The decay curves (Fig.
8) showed only a modest difference in the intracellular stabilities of transcripts of the three variant GST Pi genes under these conditions. The decay of the GST Pi message in each cell line
followed first order kinetics with half-lives of 9.4, 14.1, and
11.8 h for hGSTP1*A, hGSTP1*B, and
hGSTP1*C mRNA, respectively.
Thermostability of Variant GST Pi Proteins
The
time-dependent loss of enzymatic activity of the three
variant recombinant GST Pi proteins at 45 °C followed first order kinetics and was relatively rapid. As summarized in Fig.
9, the rates of decay of GST enzyme activity was
significantly different between the three GST Pi variant proteins. The
decay rates were 1.81·h1 for GSTP1a-1a,
1.01·h
1 for GSTP1b-1b, and 0.89·h
1 for
GSTP1c-1c, with half lives of 23, 47, and 51 min, respectively. SDS-PAGE and Western blotting showed no detectable degradation in the
GST peptides associated with the loss in enzyme activity under these
conditions.
Expression and Distribution of Variant GST Pi mRNA in Normal Cells, Tissues, and Tumors
The frequencies with which the three GST Pi gene variants were observed in malignant gliomas and normal specimens (lymphocytes and placentas) are summarized in Table IV. The frequency of hGSTP1*A homozygosity was 0.22 for gliomas compared with 0.51 for normal specimens. In contrast, hGSTP1*C homozygosity was at a frequency of 0.07 in normal specimens and 0.25 in gliomas. hGSTP1*A/hGSTP1*C heterozygosity was observed at frequencies of 0.34 for gliomas and 0.09 for normal specimens. Thus, overall, hGSTP1*C was present at a frequency of 0.59 in gliomas compared with 0.16 in normal specimens. Two of 43 normal specimens and none of the gliomas were homozygous for hGSTP1*B. However, the frequency of hGSTP1*A/hGST1*B heterozygosity was significantly higher, 0.19 in gliomas and 0.28 in normal specimens, respectively. None of the 75 specimens were heterozygous for hGSTP1*B and hGST1*C. These differences between gliomas and normal specimens in the frequency of expression of the variant GST Pi gene variants are statistically significant (p value = 0.001) by the Fisher exact test.
|
Amino acid sequence analyses of GST Pi proteins isolated from
human placenta have indicated the existence of at least two variant GST
Pi proteins, one with Ile and the other with Val at codon 104 (27).
Despite these findings, the complete GST Pi cDNA and gene isolated
by different laboratories (23-26) have been identical, both in
nucleotide sequence and in the sequence of the encoded peptide. This
study describes for the first time the isolation and characterization
of three different full-length GST Pi cDNA variants corresponding
to mRNAs of GST Pi genes expressed in human cells and tissues and
provides conclusive molecular evidence for polymorphism of the human
GST Pi gene locus. The presence and expression of the variant GST Pi
genes in both normal and malignant specimens indicate that they
represent allelopolymorphism of the GST Pi gene locus, rather than
tumor-associated or random mutations. One of the placental tissues
examined in this study expressed hGSTP1*B exclusively. In a
prior report (27), a GST Pi protein was isolated from human placenta
and shown to have the same amino acid sequence as that encoded by
hGSTP1*B. However, because neither the gene encoding this
GST Pi peptide nor the cDNA corresponding to its mRNA had been
identified, it was inconclusive as to whether this protein was the
result of a point mutation in the GST Pi gene or represented a
naturally occurring polymorphism of the GST Pi gene. Similarly, until
now, there had been no report of the isolation of the full-length
hGSTP1*C cDNA or gene, although a truncated cDNA
encoding 176 amino acids of the N terminus of a peptide homologous to
hGSTP1*C had been described (42). Comparison of the
nucleotide sequences of the human GST Pi cDNAs reported here with
that of the rat orthologue, GSTP (43), showed codons 104 and 113 to be
among those altered in the GST Pi cDNAs of the two species, with
changes from human to rat of Ile104 Gly104
and Ala113
Asn113, suggesting that these
positions are among the evolutionarily least conserved in the GST Pi
gene.
Our data show that the alterations in restriction endonuclease sites caused by the nucleotide transitions in the GST Pi gene variants provide a simple, rapid, and specific technique for determining the GST Pi gene variant(s) expressed in cells and tissues. The method is suitable for screening large numbers of specimens. Alternatively, GST Pi phenotype/genotype can be determined by Southern or Northern hybridizations with oligonucleotide probes specific for the different genes, as was also demonstrated in this study. However, because of cross-hybridization, the specificity of the latter procedure is limited, particularly for differentiating hGSTP1*B from the other two GST Pi gene variants. Furthermore, Southern hybridization is not suitable for detection of mixtures of GST Pi genes in a single specimen. Genotype analysis is best achieved by restriction mapping, single strand conformational analysis3 and nucleotide sequence determination. In the absence of variant specific antibodies, GST Pi phenotype is best determined by RT-PCR of the polymorphic region, followed by restriction site mapping and/or sequencing of the resultant cDNA as described in this study. In this regard, it is an advantage that transcripts of the three variant GSTP1 genes were found to be relatively stable intracellulary, with half-lives of the mRNAs between 9.4 and 11.8 h, which are slightly longer than those previously reported for GST Pi transcripts in other cell types (44).
The Km (CDNB) of GSTP1a-1a observed in this study for the conjugation of GSH with CDNB was approximately 3-fold lower than that of either GSTP1b-1b or GSTP1c-1c. The Km for the co-substrate GSH, on the other hand, did not differ significantly between the variant GST Pi proteins. These results are similar to those in a previous report in which a recombinant GSTP1b-1b protein expressed from a cDNA obtained artificially by site-directed mutagenesis of hGSTP1*A, was shown to have a 4-fold higher Km (CDNB) than the hGSTP1*A encoded protein (41). At high substrate (CDNB) concentration, Vmax of the reaction catalyzed by all three GST Pi proteins differed only modestly and was in the range reported in other studies of both tissue and recombinant GST Pi proteins (41, 45-47). Kcat for CDNB was also similar for all three proteins, thus resulting in utilization ratios, Kcat/Km, for GSTP1b-1c and GSTP1c-1c that were 3.3- and 4-fold lower, respectively, than that of GSTP1a-1a. These data indicate that the process of conjugation of GSH with CDNB was unaffected by the amino acid substitutions in the proteins, as was previously observed with recombinant GSTP1a-1a and GSTP1b-1b (41), but that the catalytic center activity was significantly higher for GSTP1a-1a than for GSTP1b-1c and GSTP1c-1c.
The functional differences between the variant GST Pi proteins observed
in the enzyme kinetic studies are supported by the results of the
analyses of the three-dimensional structures obtained by modeling the
amino acid substitutions into the crystal structure of the GST Pi
peptide. These data suggest that the functional consequences of the
nucleotide transitions are at least in part the result of steric
effects caused by the resultant changes of Ile104 to
Val104 and of Ala113 to Val1113 in
the GST Pi H-site. The Ile104 Val104
substitution in GSTP1b-1b and GSTP1c-1c caused significant shifts in
the side chains of several active site amino acid residues and resulted
in a steric restriction of the region of the H-site bordered by
Tyr108, Val35, Val10, and to a
lesser extent Phe8, while opening up the region bordered by
Tyr7 and Val10. These findings suggest that
bulky substrates may fit better in the space lined by these residues
and Ile104, whereas less bulky ones may bind better in the
larger space lined by Val104 and might explain previous
findings that the Km of GSTP1a-1a for CDNB was lower
than that for GSTP1b-1b, whereas the opposite was true for the bulkier
bromosulphthalein (41). Additional structural deviations were observed
with the amino acid changes of alanine to valine at codon 113 in
hGSTP1c-1c; however, these were of a lower magnitude than the codon 104 changes and did not have a major additional impact in catalyzing the
GSH-CDNB conjugation reaction.
An additional basis for the functional differences observed between the
variant GST Pi proteins could reside in the fact that domain II of the
GST Pi peptide in which the H-site is localized, contains five
-helices, two of which,
D (AA81-107) and
E
(AA109-132), contain the amino acid substitutions of
Ile104
Val104 and Ala113
Val113 (28, 48). A right-handed superhelix exists in this
region, generated in part by an up-down arrangement of
D and
E
and a cross-over connection between
E and a third helix,
F to
form a superhelix (28, 48). In a previous study, it had been shown that
the thermodynamic propensities of Ile, Ala, and Val to contribute to
the
-helical structure in a protein differ significantly (49), as
indicated by the computed free energy,
G, when the protein folds
to the native
-helix structure. Consequently, the changes of Ile or
Ala to Val could result in subtle alterations in the
-helical and/or
superhelical structure that can affect H-site architecture and
ultimately result in differences in substrate binding affinities and
catalytic activities between the GST Pi enzymes. This will be
consistent with the different enzyme kinetic data observed for the
variant proteins in this study. At 45 °C, the enzymatic activity of
recombinant GSTP1a-1a was lost at a rate significantly faster than that
of GSTP1b-1b or GSTP1c-1c, as had been previously reported for
GSTP1a-1a and GSTP1b-1b (41). We speculate that this might be due to
differential H-site stability due to differences in the free energy,
G, of the a-helix formed with the different amino acids at codons
104 and 113 at the higher temperature.
The 4-fold higher frequency of hGSTP1*C observed in malignant gliomas than in normal tissues warrant further studies to determine if this represents clonal selection or a loss of heterozygosity associated with the malignant process and whether a similar high hGSTP1*C frequency is present in other tumor types, such as breast carcinoma, in which GST Pi expression is not only high but is also associated with poor prognosis. In contrast to the other two GST Pi genes, hGSTP1*B appears to be a relatively rare allele, and intriguingly, none of the specimens examined was heterozygous for hGSTP1*B and hGSTP1*C. The functionally different GST Pi proteins may provide a mechanism for fine tuning/regulation of cellular GST Pi function, and as such, because of the similar catalytic properties of GSTP1b-1b and GSTP1c-1c, no biological advantage exists in co-expressing these two GST Pi proteins in a single cell or tissue. It will be interesting, in future studies, to determine whether the different GST Pi peptides dimerize to yield GST Pi proteins with different catalytic activities for different substrates.
It is reasonable to expect that the ability of the variant GST Pi proteins to catalyze the conjugation of different mutagens, carcinogens, and alkylating anticancer agents with GSH will differ; however, the contribution of this to the biological behavior, therapeutic response of gliomas, and increased cancer risk remains to be determined. Since the isolation of the cDNA variants reported in this study, we have also isolated and characterized the complete hGSTP1*C gene from a cosmid genomic library of a glioblastoma cell line (GenBankTM/EMBL Data Bank accession number U21689[GenBank]). We believe that the findings and the data presented in this report represent an important advance in the study of this important gene and of its function and its role in human malignancy.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U30897[GenBank] and U62589[GenBank].