* Laboratory of Clinical and Experimental Endocrinology and Immunology, Wadsworth Center, Albany, New York 12201;
Department of Biomedical Sciences, School of Public Health, State University of New York, Albany, New York 12222;
Laboratory of Structural Pathology, Wadsworth Center, Albany, New York 12201; and
Department of Biometry and Statistics, School of Public Health, State University of New York, Albany, New York 12222
Received November 20, 2000; accepted January 28, 2001
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key Words: glutamate-cysteine ligase; GCL; GLCLC; polymorphism; trinucleotide repeat; glutathione; drug resistance.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We have previously reported identification of a polymorphic GAG trinucleotide repeat (TNR) present just upstream of the human GLCLC coding region (Walsh et al., 1996). Examination of a Caucasian population revealed three alleles with 7, 8, or 9 repeats. Multiple transcriptional start sites are used during expression of GLCLC (Mulcahy and Gipp, 1995
). The GAG repeat, located just 10 base pairs 5' to the start codon, is included in each of the four major transcripts produced. Therefore, variation in the size of this repeat, in theory, could influence transcription rate, translation efficiency, and/or mRNA stability. Additionally, the GLCLC alleles, defined by the TNR, may possess additional and potentially significant nucleotide differences in regulatory sequences or within the coding region.
Given the role that differences in the ability to synthesize GSH might play in interindividual variability in cellular responses to environmental chemicals and drugs and in susceptibility to various diseases associated with oxidative stress, it is important to determine whether polymorphism of GLCLC is functionally significant. The studies reported here further evaluate polymorphism of this locus and assess the association of GLCLC genotype with GSH levels and drug resistance in the National Cancer Institute (NCI) drug screening panel of human tumor cell lines.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Determination of GLCLC TNR allele frequencies in the 60 NCI tumor cell line drug screening panel.
Frozen cell pellets (1 x 107 cells) for each of the 60 cell lines were obtained from the NCI, and genomic DNA was isolated using a Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN). GLCLC GAG trinucleotide repeat lengths were determined as described above. The genotype of each cell line was verified at least once.
Statistical analyses.
An omnibus chi-square test was used to test the association between GLCLC allele frequencies and the four ethnic groups studied. Fisher's exact test (Fisher, 1935) was used for pairwise comparison of the frequency of each allele in two different ethnic groups.
The effect of GLCLC genotype on GSH levels was assessed with classical least squares regression (Draper and Smith, 1998) in which the effect of each allele was partitioned into hetero- and homozygotic effects (Liu, 1998
). This analysis neglects the effects of additional uncontrolled covariates, i.e., variable levels of other gene products, which may significantly influence GSH levels. To address this, a stepwise covariate selection procedure was performed using two publicly available databases on gene expression in the NCI cell lines. The NCI's Developmental Therapeutics Program (DTP) molecular target database (MTD) was downloaded from ftp://dtpsearch.ncifcrf.gov/web_hooks_prim.zip. A second data set, ftp://dtpsearch.ncifcrf.gov/web_hooks_gc.zip, containing RNA levels determined using cDNA microarray analysis was also downloaded from DTP. This molecular target microarray data (MTMD) was collected by the Brown and Botstein group at Stanford University (Ross et al., 2000
; Scherf et al., 2000
). With the GLCLC genotype included in the analysis, the additional contribution of each of 161 potential covariates (molecular targets) from the MTD were individually assessed for their contribution in explaining variation in GSH levels. The best contributor was then added to the analysis and the contributions to the new analysis of each of the remaining 160 potential covariates was again individually tested. The procedure was repeated until none of the set of remaining molecular targets made a significant contribution. Correlations between early and late entry covariates alter the significance of the former. A backwards elimination procedure, deleting the least significant covariates one at a time until only the most important contributors remain, was implemented to eliminate this effect. The entire stepwise procedure was repeated on the 9703 targets in the MTMD data set. Finally, the backwards elimination procedure was run a third time starting with all covariates from the MTD and MTMD analyses to settle on a joint set of covariates. Because not all of the molecular targets were evaluated in all of the 60 NCI cell lines, average values were imputed for missing values using the mean for each data set. The final set of covariates was selected on the basis of having the optimal number of investigated imputations (three for the MTD database and five for MTMD database). Comparison of GSH in GLCLC homozygotes was by the Mann-Whitney U-test.
The DTP Human Tumor Cell Line Screen for anticancer activity has produced data on the sensitivities of 60 tumor cell lines to a standard set of 169 commonly used chemotherapeutic agents (Monks et al., 1991). This publicly available agent sensitivity data, August 1999 release, was downloaded from http://dtp.nci.nih.gov at ftp://dtpsearch.ncifcrf.gov/gi50_a99.bin. The influence of GLCLC genotype on tumor cell sensitivity to the growth inhibitory effects of these agents was estimated by classical least squares regression.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
Using data reported by Tew et al. (1996), we assessed the association of GLCLC genotype with total cellular GSH levels. Only 59 cell lines were used in this analysis, as GSH levels were not reported for one cell line. An initial regression analysis of GSH levels on GLCLC alleles explains a significant amount of the observed GSH variability (31%, p = 0.003). Alleles 1 and 2 are significantly associated with higher GSH levels (p < 0.0009 and 0.0139, respectively). Interestingly, this influence of allele 2 appears to be reversed in the homozygous condition, with the 2,2 genotype significantly associated (p < 0.0089) with decreased levels of GSH. Similarly, total GSH levels in cells homozygous for allele 3 are significantly lower than GSH levels in 1,1 homozygotes (p < 0.0034). A calculation of the contributions of individual alleles with respect to total GSH levels is represented in Figure 2. All GSH data were simultaneously used to estimate haploid effects and a single variance. Homozygote GSH levels were modeled by appropriate additions of haploid effect estimates. Although only two cell lines homozygous for GLCLC allele 2 exist, the significant p-value results from their large departure from the haploid prediction, which is based on 118 alleles.
|
|
Association of GLCLC Genotype with Drug Resistance in the NCI Cell Lines
We next asked whether GLCLC genotype was significant with respect to drug susceptibility/resistance. The NCI's DTP in vitro anticancer drug screen assesses the growth-inhibitory effects of test agents on a panel of 60 tumor cell lines derived from leukemias, melanomas, and cancers of the breast, CNS, colon, lung, kidney, ovary, and prostate (Monks et al., 1991). Analysis of drug sensitivity is measured as the GI50 (Skehan et al., 1990
). Over 70,000 agents have been screened and among these is a set of "standard agents" that includes most of the drugs approved for clinical use as chemotherapeutic agents. A publicly available database that contains information on the inhibitory effects of 169 standard agents was used to determine whether GLCLC genotype influences drug susceptibility. Sensitivity to 26 different agents was significantly influenced by GLCLC genotype, with allele 2 affecting sensitivity to the largest number of agents (Table 2
). In 169 regressions, the occurrence of 19 agents negatively associated with a single allele is significant (p < 0.0002). The association of GLCLC allele 2 with increased sensitivity to these agents did not correlate with any single putative drug mechanism of action.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In an effort to determine whether the described polymorphism of GLCLC is functionally significant, the 60 tumor cell lines of the NCI drug screening panel were genotyped and the results analyzed with respect to existing data on GLCLC transcript levels, GSH levels, and drug sensitivity/resistance. These studies indicate that GLCLC alleles 1 and 2 are significantly associated with increased levels of total GSH in the proliferating tumor cells. In contrast, allele 2 in the homozygous condition is associated with decreased levels of GSH in the two cell lines with this genotype. Likewise, GLCLC allele 3 homozygotes have decreased levels of GSH relative to GLCLC allele 1 homozygotes. A reasonable hypothesis would be that an allele producing low levels of GCL catalytic subunit or producing a catalytic subunit with decreased stability or activity would have minimal effects in the heterozygous condition, where the second allele could compensate for the "inadequate" allele. In the homozygous condition, this inadequate allele would be associated with a deficit of GCL activity leading to decreased levels of GSH. However, no correlation between reported GLCLC transcript levels and genotype was observed. Thus, it is not simply that alleles associated with lower GSH levels express lower levels of GLCLC mRNA.
Variation in the levels of expression of 15 genes, as assessed by cDNA microarray analysis performed by the DTP and the Stanford group, is shown to significantly influence the association between GLCLC genotype and GSH levels (Table 1). Both positive effects, increased levels of the covariant being associated with higher levels of GSH, and negative effects, increased levels of the covariant being associated with lower levels of GSH, were observed. The association, in some cell lines, of higher levels of GLCLC mRNA with increased cellular GSH is understandable. However, correction for this influence, which may be related to non-GLCLC effects such as the level of a transcription factor, tightens the association between GLCLC genotype and GSH levels. On the other hand, high levels of mRNA encoding the regulatory subunit, GLCLR, is associated with lower GSH levels, but again, eliminating this influence of GLCLR transcript levels strengthens the association between GLCLC genotype and GSH levels. It is unclear at this time what the connection is between expression of the other 13 genes and GSH levels. Among these genes are five encoding tyrosine kinases and one encoding a tyrosine phosphatase, possibly implicating signaling pathways involved in regulation of GSH synthesis, turnover, or export. Interestingly, 6 of the 15 genes having an effect (EIF1A, PTP4A2, LCK, TIE, EST H15054, and GLCLR) map to a less than 100 centimorgan (cM ) region on chromosome 1 (according to the GB4 map, http://www.ncbi.nlm.nih.gov/genemap). Based on a total genome size of 3488 cM (Broman et al., 1998
), the likelihood of this occurring by chance is remote (p = 0.0003). Therefore, this clustering of genes influencing GSH levels may have biological significance. This covariance analysis illustrates a novel use of gene expression data obtained using cDNA microarrays. The inclusion of such data in the evaluation of allelic effects can significantly increase the sensitivity of the analysis by accounting for the influence of covariants.
Tew et al. (1996) have previously reported an analysis of these tumor cell lines with respect to GSH levels, expression of several GSH-related enzymes, and drug sensitivity. Their findings indicated that expression of GLCLC transcripts in the proliferating but untreated tumor cells was positively correlated with resistance to alkylating agents. GSH levels in these cells, however, were not observed to significantly correlate with drug sensitivity. In accordance with these findings, our data show that the increase in GSH levels associated with alleles 1 and 2 in untreated cells does not translate to increased protection of cells possessing these alleles upon exposure to the standard agents. In fact, allele 2 is significantly associated with increased sensitivity to certain chemotherapeutics, i.e., the concentration of drug required to result in 50% growth inhibition is decreased in those cells expressing GLCLC allele 2 in either the heterozygous or homozygous form (Table 2). These observations likely reflect a difference in regulation of GSH synthesis in proliferating, untreated cells versus in cells responding to exposure to a chemotherapeutic agent. In the latter case, compensation by a second GLCLC allele may not be adequate. GSH may interfere with the drugmolecular target interaction by conjugation with the drug, by participation in export of the drug by the MRP family of transporters, or by influencing DNA repair (Schroder et al., 1996
). In each case, the ability to replenish intracellular GSH would be required for a continued contribution to drug resistance. Therefore, induced expression of GLCLC and GCL, which is rate-limiting for GSH biosynthesis, may be critical in the cellular drug response. Induction of GLCLC and MRP1 in response to exposure of cells to a number of agents has been demonstrated, and recent studies have shown that overexpression of GLCLC can lead to decreased sensitivity to drugs (Gomi et al., 1997
; Ishida et al., 1997
; Ishikawa et al., 1996
; Mulcahy et al., 1997
; Yamane et al., 1998
). The association of GLCLC allele 2 with increased sensitivity to a number of agents suggests that allele 2 is impaired in its ability to direct expression of the GCL catalytic subunit in response to exposure to certain chemotherapeutic agents. It will be important to specifically determine how GLCLC genotype influences induction of GLCLC and/or GCL activity in cells exposed to drugs or oxidative stress.
The association of GLCLC allele 2 with sensitivity to those agents listed in Table 2 suggests a cis effect of GLCLC polymorphism as opposed to variation in the levels of some trans-activating factor involved in regulation of GLCLC synthesis. An alternative explanation is that some other polymorphic locus in close physical linkage with GLCLC (with each of its alleles inherited in tandem with a particular GLCLC allele) is producing the effect. Because genes encoding the alpha class of GSTs map to the same region of chromosome 6p12 as GLCLC, and because polymorphism of GST
-encoding genes has been suggested (van Dyke et al., 1991
), we examined the association between GLCLC genotype and expression of GST
mRNA and GSTA1 protein. No correlation was observed between GLCLC genotype and expression of the GST
genes. Therefore we do not believe the association between drug sensitivity and GLCLC genotype can be explained by the variable expression of linked GST
genes.
To our knowledge, this is the first reported evidence of functionally significant polymorphism of GLCLC. We propose that certain allelic forms of GLCLC may influence GSH synthesis and interindividual variation in response to drug or toxin exposure, and therefore suggest that GLCLC be considered a candidate susceptibility gene. Given the hypothesized involvement of oxidative stress and associated GSH deficiency in such far-ranging disease processes as aging, macular degeneration, chronic obstructive pulmonary disease, HIV infection, neurodegenerative disorders, and cancer, further analysis of GLCLC polymorphism is warranted. Studies to determine the mechanistic bases of functionally relevant GLCLC polymorphism are ongoing in our laboratory. In epidemiology studies designed to test the association of alleles of GLCLC with susceptibility to cellular damage upon chemical exposure or with risk of disease, consideration should be given to the fact that GLCLC allele frequencies can vary significantly depending on the ethnic origin of the population studied.
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
NOTES |
---|
2 Current address: Southwestern Graduate School of Biomedical Sciences, University of Texas Southwestern Medical Center, Dallas, Texas 75390.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Broman, K. W., Murray, J. C., Sheffield, V. C., White, R. L., and Weber, J. L. (1998). Comprehensive human genetic maps: individual and sex-specific variation in recombination. Am J. Hum. Genet. 63, 861869.[ISI][Medline]
Caggana, M., Conroy, J. M., and Pass, K. A. (1998). Rapid, efficient method for multiplex amplification from filter paper. Hum. Mutat. 11, 404409.[ISI][Medline]
Draper, N. R., and Smith, H. (1998). Applied Regression Analysis. Wiley, New York.
Fisher, R. A. (1935). The logic of inductive inference. J. Roy. Stat. Soc. 98, 3954.
Gomi, A., Shinoda, S., Masuzawa, T., Ishikawa, T., and Kuo, M. T. (1997). Transient induction of the MRP/GS-X pump and gamma-glutamylcysteine synthetase by 1-(4-amino-2-methyl- 5pyrimidinyl)methyl-3-(2-chloroethyl)-3-nitrosourea in human glioma cells. Cancer Res. 57, 52925299.[Abstract]
Griffith, O. W., and Mulcahy, R. T. (1999). The enzymes of glutathione synthesis: Gamma-glutamylcysteine synthetase. Adv. Enzymol. Relat. Areas Mol. Biol. 73, 209267.[ISI][Medline]
Huang, C. S., Anderson, M. E., and Meister, A. (1993). Amino acid sequence and function of the light subunit of rat kidney gamma-glutamylcysteine synthetase. J. Biol. Chem. 268, 2057820583.
Ishida, T., Nishio, K., Kurokawa, H., Arioka, H., Fukumoto, H., Fukuoka, K., Nomoto, T., Yokote, H., Hasegawa, S., and Saijo, N. (1997). Circumvention of glutathione-mediated mitomycin C resistance by a novel mitomycin C analogue, KW-2149. Int. J. Cancer 72, 865870.[ISI][Medline]
Ishikawa, T., Bao, J., Yamane, Y., Akimaru, K., Frindrich, K., Wright, C. D., and Kuo, M. T. (1996). Coordinated induction of MRP/GS-X pump and gamma-glutamylcysteine synthetase by heavy metals in human leukemia cells. J. Biol. Chem. 271, 1498114988.
Liu, B. H. (1998). QTL mapping: Single-marker analysis. In Statistical Genomics: Linkage, Mapping, and QTL Analysis, pp. 406407. CRC Press, Boca Raton.
Meister, A. (1988). Glutathione metabolism and its selective modification. J. Biol. Chem. 263, 1720517208.
Monks, A., Scudiero, D., Skehan, P., Shoemaker, R., Paull, K., Vistica, D., Hose, C., Langley, J., Cronise, P., Vaigro-Wolff, A., Gray-Goodrich, M., Campbell, H., Mayo, J., and Boyd, M. (1991). Feasibility of a high-flux anticancer drug screen using a diverse panel of cultured human tumor cell lines. J. Natl. Cancer Inst. 83, 757766.[Abstract]
Mulcahy, R. T. and Gipp, J. J. (1995). Identification of a putative antioxidant response element in the 5'-flanking region of the human gamma-glutamylcysteine synthetase heavy subunit gene. Biochem. Biophys. Res. Commun. 209, 227233.[ISI][Medline]
Mulcahy, R. T., Wartman, M. A., Bailey, H. H., and Gipp, J. J. (1997). Constitutive and ß-naphthoflavone-induced expression of the human gamma-glutamylcysteine synthetase heavy subunit gene is regulated by a distal antioxidant response element/TRE sequence. J. Biol. Chem. 272, 74457454.
Ristoff, E., Augustson, C., Geissler, J., de Rijk, T., Carlsson, K., Luo, J., Anderson, K., Weening, R. S., van Zwieten, R., Larsson, A., and Roos, D. (2000). A missense mutation in the heavy subunit of gamma-glutamylcysteine synthetase gene causes hemolytic anemia. Blood 95, 21932196.
Ross, D. T., Scherf, U., Eisen, M. B., Perou, C. M., Spellman, P., Iyer, V., Jeffery, S. S., Van de Rijn, M., Waltham, M., Pergamenschikov, A., Lee, J. C. F., Lashkari, D., Shalon, D., Meyers, T. G., Weinstein, J. N., Botstein, D., and Brown, P. O. (2000). Systematic variation in gene expression patterns in human cancer cell lines. Nat. Genet. 24, 227235.[ISI][Medline]
Scherf, U., Ross, D. T., Waltham, M., Smith, L. H., Lee, J. K., Tanabe, L., Kohn, K. W., Reinhold, W. C., Myers, T. G., Andrews, D. T., Scudiero, D. A., Eisen, M. B., Sausville, E. A., Pommier, Y., Botstein, D., Brown, P. O., and Weinstein, J. N. (2000). A gene expression database for the molecular pharmacology of cancer. Nat. Genet. 24, 236244.[ISI][Medline]
Schroder, C. P., Godwin, A. K., O'Dwyer, P. J., Tew, K. D., Hamilton, T. C., and Ozols, R. F. (1996). Glutathione and drug resistance. Cancer Invest. 14, 158168.[ISI][Medline]
Seelig, G. F., Simondsen, R. P., and Meister, A. (1984). Reversible dissociation of gamma-glutamylcysteine synthetase into two subunits. J. Biol. Chem. 259, 93459347.
Skehan, P., Storeng, R., Scudiero, D., Monks, A., McMahon, J., Vistica, D., Warren, J. T., Bokesch, H., Kenney, S., and Boyd, M. R. (1990). New colorimetric cytotoxicity assay for anticancer-drug screening. J. Natl. Cancer Inst. 82, 11071112.[Abstract]
Tew, K. D., Monks, A., Barone, L., Rosser, D., Akerman, G., Montali, J. A., Wheatly, J. B., and Schmidt, D. E. (1996). Glutathione-associated enzymes in the human cell lines of the National Cancer Institute Drug Screening Program. Mol. Pharmacol. 50, 149159.[Abstract]
Van Dyke, D. C., Roby, J., and Tu, C.-P. D. (1991). New Sac I RFLP for human liver glutathione-S-transferase. Nucleic Acids Res. 19, 199.
Walsh, A. C., Li, W., Rosen, D. R., and Lawrence, D. A. (1996). Genetic Mapping of GLCLC, the human gene encoding the catalytic subunit of gamma-glutamylcysteine synthetase, to chromosome band 6p12 and characterization of a polymorphic trinucleotide repeat within its 5' untranslated region. Cytogenet. Cell Genet. 75, 1416.[ISI][Medline]
Yamane, Y., Furuichi, M., Song, R., Van, N. T., Mulcahy, R. T., Ishikawa, T., and Kuo, M. T. (1998). Expression of multidrug resistance protein/GS-X pump and gamma-glutamylcysteine synthetase genes is regulated by oxidative stress. J. Biol. Chem. 273, 3107531085.