Abstract
The μ class glutathione S-transferase geneGSTM1 is polymorphic in humans, with approximately half of the Caucasian population being homozygous deleted for this gene. GSTM1 enzyme deficiency has been suggested to predispose people to lung and bladder cancer. Some people in a Saudi Arabian population, however, have been described previously with ultrarapid GSTM1 enzyme activity. Here we have evaluated the molecular genetic basis for this observation. Genomic DNA from two Saudi Arabian subjects exhibiting ultrarapid enzyme activity and from 13 Swedish subjects having null, one, or two GSTM1 genes were subjected to restriction fragment length polymorphism analysis using the restriction enzymesEcoRI, EcoRV, and HindIII and combinations thereof. Hybridization was carried out using a full-length GSTM1 cDNA or the 5′ and 3′ parts of the cDNA. The restriction mapping data revealed the presence of a GSTμ cluster with two GSTM1 genes in tandem situated between the GSTM2 and GSTM5 genes. A quantitative multiplex polymerase chain reaction method, which simultaneously amplified a fragment of the GSTM1 gene and the β-globin gene, was developed, and the genomicGSTM1 copy number was determined from theGSTM1/β-globin ratio. This method clearly separatedGSTM1 +/− subjects (ratios between 0.4 and 0.7) fromGSTM1 +/+ subjects (ratios between 0.8 and 1.2). The two Saudi Arabians with ultrarapid GSTM1 activities had ratios of approximately 1.5, indicating that they carried threeGSTM1 genes. These results demonstrate the existence of a novel μ class GST cluster containing a duplicated active GSTM1 gene causing ultrarapid enzyme activity.
The GSTs are a family of enzymes involved in xenobiotic detoxification by means of conjugating glutathione to the electrophilic center of the compound. GSTs contribute in the protection against a broad range of compounds including carcinogens, pesticides, antitumor agents, and environmental pollutants (1). The mammalian cytosolic GSTs have been grouped into six classes based on amino acid sequence similarity and antibody cross-reactivity, which are designated α, μ, π, ς, θ, and κ. One member of the GST μ class,GSTM1, is polymorphic in humans and three alleles have been described: GSTM1*A, GSTM1*B, andGSTM1*O (2-4). The GSTM1*A and *Balleles differ only by a K172N amino acid exchange and seem to be functionally identical (5). The GSTM1*O allele has been shown to be the result of a deletion of the entire GSTM1gene (6). Approximately half of the Caucasian population are homozygous deleted for this allele and fail to express the protein (7). As the GSTM1 enzyme is effective at detoxifying some carcinogenic epoxides, including the highly carcinogenic diol-epoxide intermediate from benzo[a]pyrene (8), much importance has been placed on theGSTM1*O/O (GSTM1 −/−) genotype. It has been suggested that homozygous deletion of the GSTM1 gene is associated with an increased risk of developing some types of lung cancers, in particular adenocarcinomas (9, 10) and squamous cell carcinomas (11, 12). There are numerous studies that have also demonstrated a significant association between subjects lacking GSTM1 activity and the risk of developing bladder cancer (13, 14), adenocarcinoma of the stomach and colon (15, 16), and pituitary adenomas (17). Although this association has not been observed in some investigations (16, 18) and is probably confounded by other factors (1,19), it seems that the combination of cigarette smoking and lack of GSTM1 activity constitutes an increased risk for developing certain cancers.
In humans, the GSTM1 gene is situated in the GSTμ cluster, which has been localized to chromosome one in the region 1p13.3 (20, 21). The cluster contains four other μ class genes in addition to GSTM1, namely GSTM2, M3,M4, and M5 (20). A detailed physical map of the cluster has been described showing the GSTM1 gene to be situated downstream of the GSTM4 and GSTM2 genes and upstream of GSTM5 (22). The GSTM1*O deletion is postulated to be caused by an unequal crossing over event between sequences about 5 kb downstream from GSTM2 andGSTM1 (22), resulting in the deletion of the entireGSTM1 gene and spanning a region of approximately 18 kb. Evidence suggests that at least one other μ class GST gene or pseudogene exists and is found on chromosome 3, probably in the region 3p24–3pter (20, 23). With the exception of GSTM1, none of the other known GST μ genes have been found to be deleted in humans. The only other polymorphisms that have been identified in these genes are a HindIII RFLP, which probably occurs in the GSTM5 gene (20) and a 3-base deletion in intron 6 of GSTM3 (24).
In a study examining the potential association of the GSTM1 null phenotype and coronary atherosclerosis development among Saudi Arabians and Filipinos, it was observed that a small percentage of the subjects expressed very high GSTM1 activities when compared with activities expected for a GSTM1 +/− or +/+ genotype (25). To determine the molecular genetic basis for this ultrarapid GSTM1 activity, restriction mapping and multiplex PCR were used to analyze theGST μ cluster in genomic DNA from Swedish and Saudi Arabian subjects having variable levels of enzyme activity, as measured using the substrate trans-stilbene oxide. The results presented in this study describe the identification and characterization of a duplicated GSTM1 gene carried by the people who displayed ultrarapid GSTM1 activity.
Materials and Methods
Nomenclature.
This report uses the nomenclature system recommended for human GSTs (26).
Samples and GSTM1 activity.
Genomic DNA samples were obtained from Saudi Arabian subjects who participated in a GSTM1 phenotyping study (25) and Swedish subjects involved in a lung cancer susceptibility study (10). These people were phenotyped for GSTM1 activity in whole blood using trans-stilbene oxide as the substrate according to the assay procedure described previously (27). In particular, two people were selected from the Saudi Arabian population because they displayed very high GSTM1 activities. Other Saudi Arabian and Swedish control samples were chosen for analysis based on their GSTM1 activities, including those with phenotypes suggesting GSTM1 +/−, +/+, or −/− genotypes. The present study was approved by the ethical committee at Karolinska Institutet.
Genomic RFLP analysis.
After restriction enzyme digestion, samples (2 μg per lane) were subjected to gel electrophoresis at 0.8 V/cm for a period of 24 hr (EcoRI and HindIII digests) or 6 days (EcoRV), using 0.8% or 0.7% agarose gels, respectively. The DNA was then transferred to Qiabrane membranes (Qiagen GmbH, Hilden, Germany). A full-length human GSTM1*B cDNA probe described previously (6) was used for Southern blot analysis, which also cross-hybridizes with GSTM2, M4, andM5 but not GSTM3 (22). To construct a 5′ half and a 3′ half of the full-length probe, the GSTM1 cDNA was cut at aBglII restriction site located in the beginning of exon 8 (6). All hybridizations and washes were done at 61°.
Multiplex PCR analysis.
The oligonucleotide primers selected for multiplex PCR included the primers 5′-CTGGATTGTAGCAGATCATGC-3′ and 5′-CTCCTGATTATGACAGAAGCC-3′, which amplify a 625-bp fragment of the human GSTM1 gene (28). A 268-bp fragment from the human β-globin gene was amplified simultaneously using the primers 5′-CAACTTCATCCACGTTCACC-3′ and 5′-GAAGAGCCAAGGACAGGTAC-3′ (29). Approximately 125 ng of genomic DNA from each sample were added to a reaction tube containing 0.5 μm of each primer, 1 × reaction buffer IV (Advanced Biotechnologies, Surrey, UK), 0.25 mm of each dNTP, 1.6 mmMgCl2 and 1.2 units of Taq DNA polymerase (Advanced Biotechnologies, Surrey, UK) in a total reaction volume of 30 μl. PCR amplification was achieved using a Perkin Elmer Cetus thermal cycler (Norwalk, CT) with a 94° initial denaturation for 5 min, followed by 23 cycles of 94° for 48 sec, 60° for 48 sec, 72° for 1.5 min, and a final extension of 72° for 5 min. To determine the optimal cycle number, a GSTM1 +/+ sample was amplified for increasing cycles to a maximum number of 32. When quantified and plotted against cycle number, it was established that 23 cycles of PCR amplification fell in the exponential phase of the reaction and yet was easily visible on ethidium bromide stained gels. Samples were analyzed by running 12 μl of the product on a 1.5% agarose gel containing ethidium bromide and photographed under UV light using Polaroid type 665 film. Quantification of PCR band and Southern blot fragment intensities was performed using a personal densitometer (Molecular Dynamics, Sunnyvale, CA).
Results
Southern blot analysis of the GST μ cluster from people displaying very high GSTM1 activity.
The fragments observed in EcoRI Southern blot analysis have been assigned to the known μ class GST genes that can cross-hybridize with the full-length GSTM1 probe used (22). The 8.0-kb fragment contains theGSTM1 gene and is absent in people who are homozygous deleted for the gene (6). Genomic DNA samples from 13 Swedish control subjects with variable GSTM1 enzyme activities indicating the presence of null, one, or two GSTM1 genes (Table1) were subjected to EcoRI RFLP analysis to determine the actual genotypes (Fig.1). The Southern blot revealed the absence of the 8.0-kb fragment in one subject (S5) with no GSTM1 activity. The other samples carried the 8.0-kb EcoRI fragment. The intensity of this fragment was compared with that of the 9.0-kb fragment (GSTM4) used as an internal standard, and a ratio was calculated from the densities of the two fragments to determine the GSTM1 genotype of the people. As shown in Table 1, the homozygous and heterozygous carriers were clearly separated into two groups whereby GSTM1 +/− subjects gave ratios between 0.5 and 0.8 and those with a GSTM1 +/+ genotype between 1.2 and 1.5. It was also observed that the two Saudi Arabian people with very high GSTM1 activities (A1 and A2) had 8.0-kbEcoRI fragments approximately 30 and 50% more intense than the mean of the four GSTM1 +/+ samples (see Fig. 1), suggesting the presence of an extra GSTM1 gene in these people.
To further characterize the locus from subjects having ultrarapid enzyme activity, Southern blot analysis using the restriction enzymeEcoRV was performed as shown in Fig.2. Using the same full-length GSTM1 probe that was employed for EcoRI RFLP analysis, a 29-kb fragment was observed in all samples. A 38-kb fragment was seen inGSTM1 +/+ and +/− subjects but was absent in subjects homozygous for a gene deletion, indicating that the GSTM1gene was located on this fragment. Samples homozygous or heterozygous deleted for the GSTM1 gene demonstrated a 20-kb fragment. An additional fragment of approximately 56 kb was present only in the two Saudi Arabians with ultrarapid GSTM1 activity. Previous data have shown the absence of EcoRV restriction sites located in the complete sequences of the GSTM1 and M4 genes (30,31) or in the cDNA from the GSTM2 gene (32). The only reported EcoRV restriction site has been found in the 3′-noncoding region of the cDNA from GSTM5 (33). The extra fragment of 56 kb was thus suggestive of a GST μ cluster containing two GSTM1 genes on the fragment (Fig. 2).
Determination of the approximate locations of theEcoRV restriction sites in the GSTμ cluster.
To characterize the cluster containing twoGSTM1 genes with respect to the location of theEcoRV restriction sites, genomic DNA from two Swedes, one being GSTM1 +/+ (S1) and the other GSTM1 −/− (S9) were digested with selected combinations of restriction enzymes and hybridized with the GSTM1 probe. When the combination ofEcoRV and EcoRI restriction enzymes were used on the two genomic DNA samples, the previously assigned EcoRI fragment of 20 kb (6) was lost in both cases, and the 9.0-kb fragment appeared with only half of the intensity (Fig.3A). Hybridization with the 3′ half of the GSTM1 probe (6) (see Materials and Methods) caused the 9.0-kb fragment to disappear from samples cleaved with EcoRI andEcoRV, indicating that it was part of the original 20-kb fragment and that the 9.0-kb EcoRI fragment was cut byEcoRV (Fig. 3A). An additional fragment of approximately 5 kb was also observed in samples that were digested with the combination of enzymes when hybridized with the full-length or the 5′ half of the probe. It represents one of the two fragments created byEcoRV digestion of the 9.0-kb EcoRI fragment, which contains the GSTM4 gene.
The two samples were also digested with HindIII alone and with a combination of EcoRV and HindIII (Fig.3B). The fragments obtained after HindIII digestion of genomic DNA have been assigned to the μ class GST genes (22). In particular three 5.5-kb fragments are observed, one of which contains the 5′ end of GSTM4 and another the 3′ end ofGSTM2. As shown in Fig. 3B, the 5.5-kb fragment, which was not shortened using both EcoRV and HindIII, hybridized with the 5′ half and not the 3′ half of the probe, showing that it contained the 5′ part of the GSTM4 gene and had noEcoRV site. One of the 5.5-kb HindIII fragments was shortened to 3.8 kb when the two samples were digested with the combination of EcoRV and HindIII. The 3.8-kb fragment hybridized with the 3′ half of the probe and not with the 5′ half, indicating that it was created by EcoRV digestion of the 5.5-kb fragment that contained the 3′ end of the GSTM2gene. The other 5.5-kb fragment that was shortened to 5.1 kb using the enzyme combination probably contains another μ class GSTgene. Based on the GST μ cluster map (22) and the present Southern blot results, maps of clusters containing a duplicatedGSTM1 gene, one GSTM1 gene, and a deletedGSTM1 gene were determined and are shown in Fig.4, A and B.
Determination of GSTM1 genomic copy number from DNA samples.
A quantitative multiplex PCR method was used to simultaneously amplify a fragment of the GSTM1 gene and the β-globin gene from genomic DNA. By calculating the densitometric ratio between the intensity of the GSTM1 band and the constant β-globin band, the GSTM1 copy number carried by an individual could be determined. Genomic DNA from the Swedish control subjects, which had been analyzed for GSTM1 activity and genotyped byEcoRI RFLP densitometric analysis, was amplified and subsequently quantified to determine the GSTM1 copy number. A representative gel showing the PCR products is shown in Fig.5. The two Saudi Arabians with ultrarapid enzyme activity had GSTM1 bands that were much more intense than that of GSTM1 +/+ subjects. Using theGSTM1/β-globin ratios calculated from the control samples, a ratio falling between 0.40 and 0.70 was defined as representing aGSTM1 +/− genotype and one between 0.80 and 1.20 indicated a GSTM1 +/+ genotype. This approach allowed for the clear separation between the two different genotypes (Table2 and Fig.6A). A good correlation between theGSTM1 copy number determined by multiplex PCR and the assigned EcoRI RFLP genotype was observed (Table 2). TheGSTM1/β-globin ratios for the ultrarapid Saudi Arabian subjects were approximately 1.5, higher than the range representing aGSTM1 +/+ genotype and indicative of three GSTM1genes (Table 2 and Fig. 6A).
Twenty additional Saudi Arabian people with phenotypes suggestingGSTM1 +/− or +/+ genotypes were analyzed forGSTM1 copy number. This was done to ensure that the comparison between the two ultrarapid Saudi Arabians and the Swedish samples was not influenced by factors such as interethnic differences in primer binding sequences. The GSTM1/β-globin ratios for the subjects are shown in Fig. 6B. No differences were observed in the calculated ratios between the two populations and all Saudi Arabian samples fell into the ranges indicative of a GSTM1 +/− or +/+ genotype.
Discussion
The results presented indicate strongly that the molecular genetic basis for ultrarapid GSTM1 enzyme activity, demonstrated previously among Saudi Arabians, is caused by the presence of aGST μ cluster containing two functional GSTM1genes in tandem. This conclusion is based on several observations: (i)trans-stilbene oxide is a substrate specific for GSTM1, and this compound was used in the phenotyping study; (ii) the 8.0-kbEcoRI fragment contains the entire GSTM1 gene and was more intense in subjects with the duplicated GSTM1 gene; (iii) the 56-kb EcoRV fragment corresponds to a haplotype that has been generated by the addition of the 18-kb fragment lost inGSTM1 deleted haplotypes; (iv) restriction mapping using various combinations of enzymes showed that ultrarapid activity subjects carried a GST μ cluster with two GSTM1genes in tandem; and (v) the PCR primers used in the multiplex PCR reactions were specific for the GSTM1 gene and gave ratios indicative of an extra gene in both subjects having ultrarapid GSTM1 activity.
Quantification of GSTM1 copy number using genomic DNA from the subjects was performed using a multiplex PCR method. The coamplification of the β-globin fragment with the variableGSTM1 fragment allowed the determination of theGSTM1 copy number carried by the individual and also acted as an internal amplification control for the PCR reaction. It was found that the calculated number of GSTM1 copies determined by this method corresponded with the genotypes determined byEcoRI RFLP analysis. The two Saudi Arabian people with ultrarapid enzyme activities had ratios indicative of threeGSTM1 genes, supporting the Southern blot data that revealed the presence of a duplicated GSTM1 gene. The multiplex PCR approach proved to be a simple and convenient way of determiningGSTM1 genotype and genomic copy number and could be applied easily in the routine analysis of GSTM1 copy number in molecular epidemiology studies as well as for the analysis of any other gene of interest.
The generation of a GST μ cluster containing a duplicatedGSTM1 gene probably involved an unequal recombination between two homologous but nonallelic sequences flanking theGSTM1 gene. This may have occurred because of a chromosomal misalignment followed by an unequal crossing over event or by unequal crossing over between sister chromatids. The outcome of such an event would be the generation of a μ cluster containing twoGSTM1 genes in tandem on one cluster and a deletedGSTM1 gene on the other (Fig. 4B). The break points of theGSTM1 deletion have been reported to be about 5 kb downstream from the GSTM2 gene and 5 kb downstream from theGSTM1 gene (22), resulting in a deletion spanning approximately 18 kb. EcoRV RFLP analysis showed that a cluster containing one GSTM1 gene gave a 38-kb fragment; therefore, a cluster consisting of two GSTM1 genes would be expected to be roughly 18 kb longer because of the unequal crossover event. Both Saudi Arabians with ultrarapid GSTM1 activity carried anEcoRV fragment of about 56 kb, which was the expected length of a fragment consisting of two GSTM1 genes in tandem.
A potential outcome of having extra GSTM1 genes could be an increased protective effect against some carcinogenic chemicals because of the more rapid detoxification of these compounds. One carcinogen found in cigarette smoke is benzo[a]pyrene, which can be metabolized by phase I enzymes to a diol-epoxide, one of its most carcinogenic forms. This carcinogen can be further conjugated with glutathione by GSTM1, resulting in a deactivated compound (8). Therefore, people carrying extra GSTM1 genes have the ability to rapidly conjugate and inactivate compounds that are substrates for this enzyme, including potential carcinogens.
Excluding the human immunoglobulin genes, which are subject to complex somatic recombination, reports describing inherited duplication events of an entire gene causing a phenotypic change are uncommon (34-36). One well studied example of this involves CYP2D6, a cytochrome P450 enzyme involved in the metabolism of many clinically used drugs and implicated as a risk factor for cancer (19). People carrying duplicated, multiduplicated, and amplified (13-fold) functional CYP2D6 genes have been shown (34, 37-39). These subjects display ultrarapid metabolism of probe drugs for this enzyme and require higher than normal doses of drugs to achieve therapeutic levels, indicating that the extra gene(s) result in higher levels of protein to be expressed (34, 37). Also in the present case, the duplicated GSTM1 gene seemed to be functional and a gene dosage effect was observed in the subjects with this genotype.
In the Saudi Arabian population phenotyped, 56% of the subjects were homozygous for the GSTM1 deleted allele (25), which is roughly the same frequency as reported in several other studies in Caucasian populations. Because over half of the Saudi Arabian population were homozygous for the GSTM1*O allele it would be reasonable, based on the unequal crossover event as the cause for this genotype, to expect a high frequency of people carrying duplicatedGSTM1 genes. This was not the case, as only 3% of the Saudi population had very high GSTM1 activities, some of which probably carried duplicated GSTM1 genes. However, because of the high frequency of the GSTM1*O allele in the population, it is possible that some people had a ++/− genotype, which would have been included in the +/+ phenotypic group. The low frequency of potential duplicated GSTM1 genes is in contrast to the high frequency (21%) of people from the same Saudi Arabian population who were found to carry duplicated CYP2D6 genes (39). It has been postulated that the CYP2D6 gene has undergone a selective environmental pressure in some populations, including the Saudi Arabians (40). The high frequency of the GSTM1*O allele suggests that this gene has not encountered strong environmental selection pressure during evolution, although the rapidly expanding number of compounds found in the environment make it likely that the GSTM1 enzyme might be increasingly involved in chemical detoxification.
In conclusion, this study identified a duplicated functionalGSTM1 gene as the genetic basis for the ultrarapid enzyme activity observed in some Saudi Arabian subjects. The finding that this polymorphic detoxification enzyme has undergone a stable duplication event implicates that other enzymes involved in the metabolism of foreign compounds may also be found in multiple copies and cause phenotypic alterations. The global distribution of people carrying duplicated or multiduplicated GSTM1 genes and the consequences of multiple GSTM1 gene copies for environmentally related diseases remain to be determined.
Footnotes
- Received July 16, 1997.
- Accepted August 26, 1997.
-
Send reprint requests to: Dr. Magnus Ingelman-Sundberg, Division of Molecular Toxicology, Institute of Environmental Medicine, Karolinska Institutet, Box 210, S-171 77 Stockholm, Sweden. E-mail:magnus.ingelman-sundberg{at}imm.ki.se
-
This work was supported by grants from the European Union, Biomed 2, the Swedish Medical Research Council, and the Research and Ethical Committee, Riyadh Armed Forces Hospital.
Abbreviations
- GST
- glutathioneS-transferase
- RFLP
- restriction fragment length polymorphism
- PCR
- polymerase chain reaction
- kb
- kilobase pair(s)
- bp
- base pair(s)
- The American Society for Pharmacology and Experimental Therapeutics