Simultaneous Genotyping of GSTT1 and GSTM1 Null Polymorphisms by Melting Curve Analysis in Presence of SYBR Green I

Abstract

Due to their ability to metabolize xenobiotics, glutathione S-transferases (GSTs) play an important role in cellular protection. GST family members μ (GSTM1) and θ (GSTT1) exhibit a common polymorphism that results in the complete deletion of the gene (null allele). Homozygous deletions, which result in the absence of the enzyme, are considered a risk factor for several diseases, including cancer. We report a simple, low cost, and high throughput assay for the simultaneous analysis of the GSTM1 and GSTT1 null polymorphisms in a single step. The assay is based on multiplex real-time PCR in the presence of SYBR Green I and genotype discrimination by melting curve analysis in a LightCycler. We have genotyped 792 samples to compare this new approach with conventional PCR followed by gel electrophoresis. Comparison of the methods gave a good agreement, with κ values of 0.88 for GSTM1 and 0.64 for GSTT1. Reanalysis of discrepant samples indicated that absence of amplification of the larger GSTT1 fragment by conventional PCR accounted for most of the discrepancies. Moreover, the improved amplification efficiency of the real-time PCR results in a significant reduction of missing values. Due to its simplicity and low cost, this assay is well suited for the rapid analysis of GST-null genotypes in studies that involve large number of samples.

Glutathione S-transferases (GSTs) are an important family of enzymes that catalyze the detoxification of a wide variety of xenobiotics, including environmental carcinogens, chemotherapeutic agents, and reactive oxygen species.¹ They are involved in phase II detoxification, mediating the conjugation of reduced glutathione to electrophilic species that leads to the elimination of toxic compounds. GSTs have also been shown to play a role in kinase signaling and glutathionylation.² Therefore, GSTs play an important role in cellular protection from environmental exposures and oxidative stress and they are also implicated in cellular resistance to drugs.³

The members μ (GSTM1) and θ (GSTT1) of the multigenic family of GSTs exhibit a common polymorphism that is characterized by the complete deletion of the gene (null allele).² Homozygous deletions (null genotype) thus result in the absence of enzyme and its catalytic activity, which has been shown to decrease the cells' ability to detoxify certain genotoxic agents.⁴ The null genotypes of GSTs have been associated with enhanced genotoxicity and are believed to be key factors in determining susceptibility to diseases associated with exposure to xenobiotics, such as cancer.^2,3,5 The frequency of the null genotype varies in different populations and occurs in about 0.53 for GSTM1*0/0 and 0.20 for GSTT1*0/0 in Caucasians.⁶

Up to now, the most widely used method for genotyping null polymorphisms has been conventional multiplex PCR followed by gel electrophoresis analysis.⁷ This methodology is time consuming and, therefore, may not be the most suitable method for molecular epidemiology and clinical studies that involve large sample numbers and require simple and low cost methodologies with mid-high throughput. In recent years, other techniques based on real-time PCR have been developed. These techniques use specific hybridization probes to distinguish between GSTs homozygous deletions and the presence of at least one GST allele,⁸ or quantitative strategies to identify the three possible genotypes for each polymorphism.⁹,¹⁰,¹¹,¹² Unfortunately, none of these techniques allows simultaneous analysis of the two genes in the same display.

The objective of this work was to develop a simple, single-step assay for the simultaneous analysis of the GSTM1 and GSTT1 null polymorphisms in studies that need rapid and effective mid to high-throughput genotyping. To this end, we have developed an alternative method based on multiplex real-time PCR and genotype discrimination by melting curve analysis in the presence of SYBR Green I, a fluorescent dye that specifically binds to double stranded DNA (dsDNA). We have investigated the suitability of this method and assessed its performance as compared with conventional multiplex PCR and gel electrophoresis analysis.

Materials and Methods

DNA Samples

DNA samples from 243 gastric cancer cases and 946 matched controls, selected from the European Prospective Investigation into Cancer and nutrition (EPIC) cohort according to a nested case-control design,¹³ were used to analyze the GSTT1 and GSTM1 null polymorphisms by real-time PCR and melting curve analysis in a LightCycler 2.0 (LC2.0). DNAs were extracted with the Puregene DNA Purification System (Gentra Systems, Minneapolis, MN) adapted to the Gentra Autopure LS DNA preparation platform (Gentra Systems) and dried for storage. Before use, DNAs were reconstituted with water.¹³ Moreover, samples of 792 subjects (cases and controls) from this study were reanalyzed by conventional multiplex PCR and gel electrophoresis. A group of 300 DNA samples from healthy subjects selected from the EPIC Spain cohort¹⁴ were used to validate the assay for performance in the LightCycler 480 (LC480). In this study, DNA was prepared by phenol-chloroform extraction and ethanol precipitation. DNA was then dissolved in distilled water.¹⁴

Genotyping by Real-Time PCR and Melting Curve Analysis with SYBR Green I

The real-time PCR was initially performed in the LightCycler (LC) 2.0 instrument, using the fluorescent dye SYBR Green I (Roche Diagnostics, Mannheim, Germany).

The PCR reaction was performed in LC capillaries. Fragments of GSTT1 (257 bp), GSTM1 (219 bp), and BLC2 (155 bp) genes, the last one used as an internal amplification control to exclude false negatives, were simultaneously amplified. The primers used were as previously described¹⁵ but the reverse primer for the GSTT1 fragment was replaced by 5′-GGAAAAGGGTACAGACTGGGGA-3′ (nt9586-9607, AB057594) to obtain a shorter amplification product and, therefore, a lower melting temperature.

Different primer and MgCl₂ concentrations were tested. In the final protocol, PCR mixtures contained 4 mmol/L MgCl₂, 0.3 μmol/L each GSTT1 primers, 0.4 μmol/L each GSTM1 primers, 0.5 μmol/L each BCL2 primers, 1× LightCycler Fast Start DNA Master SYBR Green I Master Mix (Roche Diagnostics, Mannheim, Germany), and 10 ng of DNA in a final volume of 10 μl. General PCR conditions were initial denaturation at 95°C for 10 minutes and 38 cycles of 95°C for 5 s, 62°C for 15 s, and 72°C for 20 s, at transition rates of 20°C/s. The fluorescence of the dsDNA bound SYBR Green was recorded in each sample after product extension, which allowed for monitoring of the amplification reaction.

Since each amplified sequence is characterized by its apparent melting temperature (Tm), which is a function of product length and base composition, a melting curve analysis was performed subsequent to the amplification of the target sequence. The program for the melting curve analysis was: 95°C for 5 s, 65°C for 20 s, with 20°C/s transition rate, and then ramping to 98°C at 0.2°C/s transition rate. After that, each amplified fragment was identified by converting its specific melting curves, measured as fluorescence emission decrease, to melting peaks by the LC software.

Each series of amplifications included a negative control without DNA template and a positive control with samples of known genotype, which had been previously confirmed by PCR and gel electrophoresis. We reanalyzed 9% (n = 110) of the total group of samples as quality control of the technique. The error rate was 1% in each polymorphism.

The final real-time PCR protocol was adjusted for use in the LightCycler 480 (LC480) instrument (Roche Diagnostics, Mannheim, Germany), which uses 384-well plates instead of capillaries. The primers were the same as those used for the LC2.0 assay. The PCR mixtures contained 3.5 mmol/L MgCl₂, 0.3 μmol/L each GSTT1 primers, 0.4 μmol/L each GSTM1 primers, 0.5 μmol/L each BCL2 primers, 1× LightCycler 480 SYBR Green I Master Mix (Roche Diagnostics, Mannheim, Germany), and 10 ng of DNA in a final volume of 10 μl. The fragment amplification was performed as follows: initial denaturation at 95°C for 10 minutes and 33 cycles of 95°C for 10 s, 62°C for 30 s, and 72°C for 25 s, at transition rates of 2 to 4.8°C/s. The program for the melting curve analysis was: 95°C for 10 s, 65°C for 1 minute with 2 to 4.8°C/s transition rate, and then ramping to 97°C at 5 acquisitions per °C.

Genotyping by Multiplex PCR and Gel Electrophoresis

The primers used to simultaneously amplify the fragments of the GSTT1 (480 bp), GSTM1 (219 bp), and BLC2 (155 bp) genes had been previously described.¹⁵ The PCR reaction was performed in a GeneAmp PCR System 9700 instrument (Applied Biosystems), in a total volume of 20 μl with 50 ng of genomic DNA, 0.2 μmol/L of each primer pairs, 200 μmol/L dNTPs, 4 mmol/L MgCl₂, 4 μl of 10× PCR Buffer, and 1 U of TaqDNA polymerase (Roche Diagnostics, Mannheim, Germany). Conventional PCR conditions were: initial denaturation at 95°C for 5 minutes, 40 cycles of 94°C for 90 s, 62°C for 90 s, 72°C for 90 s, and a final step of 72°C for 5 minutes. The PCR products were analyzed by electrophoresis in 2% agarose gels and ethidium bromide staining.

Validation Analysis

Samples with discrepant results between both methods were reanalyzed by singleplex and multiplex long-PCR using combinations of previously described primers.^16,17

Statistical Analysis

The genotyping results were classified as GST null (homozygous deletion), GST present (heterozygous or homozygous genotype for the gene presence) and missing (there was no or very poor internal control fragment amplification). Comparison of proportions was done by means of the McNemar test for paired data.¹⁸ The agreement between results obtained with each method was assessed by means of the kappa index (κ) with its corresponding 95% confidence intervals (CI).¹⁸ This index takes into account the expected proportion of agreement, with a range from −1 to +1; the value 0 corresponds to no agreement beyond what is expected by chance, while +1/−1 correspond to perfect agreement or complete disagreement respectively. A scale has been proposed for the interpretation of κ,¹⁹ according to which a value below 0.20 indicates poor agreement, for 0.21 to 0.40 the agreement is fair; 0.41 to 0.60, moderate; 0.61 to 0.80, good; and 0.81 to 1, very good. Finally, the proportion of null alleles for both GST genes were compared¹⁸ to the theoretical proportion in the population.⁶

Results

Figure 1 shows the melting peaks of the GSTT1, GSTM1, and BCL2 fragments amplified in the LC2.0 (Figure 1A) and in the LC480 (Figure 1B) from four individual DNAs having different combinations of the GSTM1 and GSTT1 genotypes. Similar patterns were obtained when the initial DNA amount varied between 3 to 25 ng. Analysis at different primer concentrations indicated that the best melting curve pattern was obtained by increasing the primer concentration of the shorter fragments (data not shown).

Figure 1.

GSTM1 and GSTT1 null polymorphism genotyping by real-time PCR and melting curve analysis in the presence of SYBR Green. Tm melting peaks of the GSTT1, GSTM1, and the BCL2 amplification control gene in: The Light Cycler 2.0 (A), and The Light Cycler 480 (B). Red line corresponds to GSTM1 and GSTT1 present genotype, blue to GSTM1 null and GSTT1 present genotype, brown to GSTM1 present and GSTT1 null genotype and green to GSTM1 and GSTT1 null genotype. “+” corresponds to the heterozygous or homozygous genotype for the gene presence. “−” corresponds to the homozygous null genotype.

Comparison of GSTM1 null polymorphisms analysis by real-time PCR (LC2.0) and by conventional PCR in 792 samples is shown in Table 1. GSTM1 genotyping by multiplex PCR and gel electrophoresis produced missing results in 166 samples (21%), while only 10 subjects (1.3%) had missing results by real-time PCR, which is significantly lower (P value <0.0001). Overall, the agreement between both methods is acceptable, with κ (95%CI) 0.59 (0.54 to 0.59). However, it is much better when only the 619 subjects with defined (non-missing) results by both methods were taken into account: the observed proportion of agreement was 94% with a κ value of 0.88 (0.84 to 0.92). Validation analysis revealed that 30 out of 37 discrepant genotypes agreed with those obtained by real-time PCR, while only three agreed with the null genotype obtained by conventional PCR (Table 1). Four samples couldn't be amplified. The proportion of GSTM1 null genotypes as measured by real-time PCR was 51.4% (402 subjects out of 782 genotyped), while for conventional PCR it was 53.7% (336 out of 626). None of them differed significantly from the 53.1% reported in Caucasian populations.⁶ Since most missing values were due to absence of amplification, all but one samples with a missing value for GSTM1 were the same as for GSTT1 (Table 2). Therefore, the pattern of GSTT1 missing values was similar to that described previously. Overall, the agreement in GSTT1 results was lower than for GSTM1. We found an observed proportion of agreement of 88.7%, with a κ (95% CI) of 0.64 (0.56 to 0.71) among 620 subjects with defined (non-missing) results by both methods, while the overall observed agreement (including missing values) was 70% with κ 0.36 (0.33 to 0.36). Reanalysis of discrepant samples showed that 60 out of 70 genotypes were in agreement with those obtained by real-time, while only four agreed with conventional PCR (Table 2). Six samples couldn't be amplified. The prevalence of GSTT1 null genotypes obtained by real-time PCR was 15.6% (122 out of 782 genotyped), which was significantly lower (P value = 0.004) than the 19.7% reported in Caucasian populations.⁶ On the contrary, the proportion of GSTT1 null genotypes obtained by conventional PCR was 24.2%, significantly higher (P value = 0.001) than reported 19.7%.⁶

Table 1.

Comparison of GSTM1 Null Polymorphism Genotyping Results Obtained by Real-Time PCR in the Presence of SYBR Green I and by Conventional PCR and Gel Electrophoresis

	Real-time PCR (LC2.0)
Conventional	Missing	Present	Null	Total
Missing	3	65	98	166
	(0.4%)	(8.2%)	(12.4%)	(21.0%)
Present	0	284	6*	290
	(0.0%)	(35.9%)	(0.8%)	(36.6%)
Null	7	31†	298	336
	(0.9%)	(3.9%)	(37.6%)	(42.4%)
Total	10	380	402	792
	(1.3%)	(48.0%)	(50.8%)	(100.0%)

^*Reanalysis of these 6 samples by a third method indicated that 4 had a null genotype. Two did not amplify.

^†Reanalysis of these 31 samples by a third method indicated that 26 carried the GSTM1 gene, 3 were null alleles and 2 did not amplify.

Table 2.

Comparison of GSTT1 Null Polymorphism Genotyping Results Obtained by Real-Time PCR in the Presence of SYBR Green I and by Conventional PCR and Gel Electrophoresis

	Real-time PCR (LC2.0)
Conventional	Missing	Present	Null	Total
Missing	3	131	31	165
	(0.4%)	(16.5%)	(3.9%)	(20.8%)
Present	0	467	8*	475
	(0.0%)	(59.0%)	(1.0%)	(60.0%)
Null	7	62†	83	152
	(0.9%)	(7.8%)	(10.5%)	(19.2%)
Total	10	660	122	792
	(1.3%)	(83.3%)	(15.4%)	(100.0%)

^*Reanalysis of these 8 samples by a third method indicated that 7 had a null genotype and 1 carried the GSTT1 gene.

^†Reanalysis of these 62 samples by a third method indicated that 53 carried the GSTT1 gene, 3 were null alleles and 6 did not amplify.

Furthermore, 300 samples of healthy subjects from the EPIC Spain cohort, previously analyzed for GSTM1 and GSTT1 by real-time PCR in the LC2.0,¹⁴ were reanalyzed in the LC480. Among 287 subjects with non-missing values, we found a proportion of observed agreement of 96%, with a κ value of 0.92 for GSTM1, and 99.7% (only one discordant result) with κ of 0.99 for GSTT1. Reanalysis of the 12 samples with discordant GSTM1 results (8 GSTM1 present and 4 null alleles, in the LC480) indicated that 50% of the results agreed with those obtained with the LC480 and 50% agreed with the LC2.0, suggesting that the discrepancy between the instruments was due to chance or sampling errors. The proportion of GSTM1 null genotypes in this population of Spanish controls was 47.2%, which was similar to the 49.7% reported by Garte⁶ for the Spanish population. By contrast, the proportion of GSTT1 null genotypes was found to be of 15.0%, which again is lower that the 20.5% reported by Garte.⁶

Discussion

Molecular epidemiology studies, as well as molecular diagnosis, need rapid and low cost methods for the analysis of a great number of samples. Real-time PCR and melting curve analysis in the presence of SYBR Green I offers the opportunity for rapid and specific analysis of the presence/absence of DNA fragments of different melting temperatures and requires only small amounts of DNA.

To our knowledge, the application of this technology to the analysis of GSTT1 and GSTM1 null polymorphisms has not been previously described, and the adaptation of the LC2.0 protocol to a platform of higher throughput such as the LC480 instrument is of great advantage for studies involving large sample numbers. The genotyping method is performed in solution from beginning to end and no separation or purification steps are needed, which minimizes handling errors. The closed system for amplification and genotyping avoids end-product contamination²⁰ and the improved amplification efficiency of the real time PCR versus the conventional PCR²¹ results in a significant reduction of missing values, which is another important point to be considered. In comparison with other real-time methods that use hybridization probes, the cost of genotyping using real-time PCR with SYBR Green I is four times lower when the LC480 is used. Furthermore, genotype discrimination by Tm peaks reduces the time and effort during assay setup and removes the need for downstream relative quantification, allowing for the simultaneous analysis of the two GST genes in the same display.

A disadvantage of this method, compared with those based on quantitative or on long-PCR for the detection of the null GSTs alleles,^9,11,12,16 is that it does not discriminate between the presence of one or two copies of the gene. Nevertheless, the role of the heterozygote genotype for null GST polymorphisms on disease risk is poorly understood. If association between GST null alleles and disease risk follows a recessive model (risk is only associated with homozygosis for the null allele), then the discrimination between homozygotes and heterozygotes for the gene presence will not be necessary. In this case the application of more economic and rapid methods for the efficient measurement of null allele homozygotes will be preferable.

Although all of the fragments are efficiently amplified with the same primer concentration, the shortest fragments bound less SYBR Green I, which results in low fluorescence emission and a small, hardly detectable, melting peak. Using different primer concentration, we obtained a larger amount of amplification product of the shortest fragments and, therefore, an increment of the fluorescent emission that resulted in a better discrimination of the peaks.

The proportion of agreement in the genotyping results of the two genes, comparing real-time and conventional PCR, was good, although the κ index for GSTT1 was lower than GSTM1. While the frequency of GSTM1 null genotypes obtained by either method were very similar to the ones previously reported in Caucasian and in Spanish populations, the GSTT1 null genotype frequencies obtained by the two methods were significantly different from each other and from those previously reported.⁶ The differences obtained when both methods were applied to the same sample group (EPIC samples) could be explained by the fact that the conventional multiplex PCR method used here to amplify a GSTT1 fragment larger than the one amplified by real-time PCR, gave a higher number of false negatives due to a reduced efficiency in the amplification of large fragments. The results obtained when the discrepant samples were reanalyzed by a third method are in agreement with this hypothesis. We do not know if this could also explain the differences obtained between null GSTT1 genotype frequencies in our populations and in those analyzed by Garte et al. The null GST genotype frequencies obtained in our study by the melting curve method were similar to those recently obtained by real-time PCR in a large Danish population.¹¹

In summary, the analysis of melting peaks after real-time PCR allows clear discrimination, in one-step and simultaneously for the GSTM1 and GSTT1 genes, between homozygotes for the null allele and carriers of any of these genes, with a good proportion of agreement between real-time and conventional PCR methods. Furthermore, the improved amplification efficiency of real-time PCR results in a significant reduction of missing values. Finally, the method described here is relatively inexpensive and labor-saving. Because of its high sensitivity and specificity, rapid performance and robustness, this assay is well suited for the analysis of GST null polymorphisms in large scale analysis, and could help overcome the controversy regarding the association between the GSTT1 and GSTM1 null polymorphisms and a wide variety of diseases.