Validation of dot blot hybridization and denaturing high performance liquid chromatography as reliable methods for TP53 codon 72 genotyping in molecular epidemiologic studies

Background Mutations in TP53 are common events during carcinogenesis. In addition to gene mutations, several reports have focused on TP53 polymorphisms as risk factors for malignant disease. Many studies have highlighted that the status of the TP53 codon 72 polymorphism could influence cancer susceptibility. However, the results have been inconsistent and various methodological features can contribute to departures from Hardy-Weinberg equilibrium, a condition that may influence the disease risk estimates. The most widely accepted method of detecting genotyping error is to confirm genotypes by sequencing and/or via a separate method. Results We developed two new genotyping methods for TP53 codon 72 polymorphism detection: Denaturing High Performance Liquid Chromatography (DHPLC) and Dot Blot hybridization. These methods were compared with Restriction Fragment Length Polymorphism (RFLP) using two different restriction enzymes. We observed high agreement among all methodologies assayed. Dot-blot hybridization and DHPLC results were more highly concordant with each other than when either of these methods was compared with RFLP. Conclusions Although variations may occur, our results indicate that DHPLC and Dot Blot hybridization can be used as reliable screening methods for TP53 codon 72 polymorphism detection, especially in molecular epidemiologic studies, where high throughput methodologies are required.


Background
The tumour suppressor protein p53 is a key transcription factor that participates in numerous homeostatic functions such as cell cycle checkpoint control, repair of DNA damage and apoptosis induction [1]. Tumour-associated mutations in TP53, typically single nucleotide substitutions in the coding sequence, are a hallmark of most human cancers and cause dramatic defects in p53 function [2]. By contrast, only a small fraction of the naturally occurring sequence variations (single nucleotide polymorphisms, SNP) of TP53 in human populations are expected to cause measurable perturbation of p53 function [3]. The best studied TP53 SNP is on codon 72. In human populations, codon 72 of TP53 has either the sequence CCC, which encodes proline, or CGC, which encodes arginine. TP53 codon 72 polymorphism alters the ability of the p53 protein to induce apoptosis, influences the behaviour of mutant p53 and decreases DNA repair capacity [4][5][6].
The role of this polymorphism in carcinogenesis has been well studied but is a matter of controversy. A decade ago, a dramatic effect of the TP53 codon 72 polymorphism on the risk of cervical cancer was reported. This effect was explained by the finding that the E6 oncoprotein from high-risk mucosal human papillomaviruses (HPVs) causes more efficient degradation of the Arg form of the p53 protein than of the Pro form, thus reducing cellular levels of p53 protein and increasing the risk of HPV-associated cancers in TP53-Arg homozygotes [7]. This sparked an intensive investigation into the potential effect of TP53 codon 72 polymorphism on susceptibility to cervical cancer or cancer-related phenotypes. Many of these studies have reported significant associations. However, the inconsistency in findings suggests that results should be interpreted with caution. For cervical cancer, a meta-regression analysis identified that the most important factor contributing to heterogeneity among results for invasive lesions was departure from Hardy-Weinberg equilibrium in the control group. However, various methodological features can contribute to departures from Hardy-Weinberg equilibrium and consequently to less than ideal circumstances for the examination of this polymorphism [8]. Therefore, confirming genotype results with a second method can reduce misclassification of genotype [9]. In doing so, deviations from Hardy-Weinberg equilibrium can be avoided. With this objective in mind, we aimed to compare different methodologies to define reference-quality methods for TP53 codon 72 genotyping. We developed two new methods for TP53 codon 72 polymorphism detection: Denaturing High Performance Liquid Chromatography (DHPLC) and Dot Blot hybridization. These methods were compared with Restriction Fragment Length Polymorphism (RFLP), a well establish method for TP53 codon 72 polymorphism genotyping, using two different restriction enzymes. We report herein the findings obtained with the validation of these methods and propose that they could be used for TP53 codon 72 polymorphism detection in molecular epidemiologic studies.

Results
Using a combination of both PCR strategies described above we were able to amplify TP53 exon 4 of 962 samples, approximately 98.3% of all subjects tested. Samples were submitted to TP53 codon 72 genotyping using four different methods: DHPLC, Dot Blot hybridization and RFLP using two different restriction enzymes(BsaJI and BstUI).
DHPLC analysis of TP53 exon 4 at 59°C revealed a small shoulder only in heterozygous samples. The first assessment allowed the distinction of homozygous and heterozygous samples, as illustrated in Figure 1B. We were able to identify that 428 samples (44.5%) were heterozygous for TP53 codon 72 polymorphism. The second assessment which included a previously known PCR product in the assays revealed that 377 (39.2%) of samples were homozygous Arg/Arg and 157 (16.3%) were homozygous Pro/Pro (Table 1). Once the same profile was observed, we concluded that samples with the same genotype of the PCR product added presented a single peak, while samples with a different genotype of this added product presented a small shoulder, indicating heteroduplex formation.
Using Dot-blot hybridization we observed that 418 samples (43.4%) were heterozygous for TP53 codon 72 polymorphism ( Table 1). The hybridization patterns observed both for heterozygous and homozygous samples are shown in Figure 1C. Homozygous Arg/Arg represented 374 (38.9%) samples while homozygous Pro/Pro represented 170 (17.7%) of all samples tested.
TP53 exon 4 was also subjected to RFLP using the restriction enzymes BsaJI and BstUI. For both enzymes a unique digestion profile is obtained when TP53 exon 4 alleles codes for proline or arginine. BsaJI leads to generation of five fragments when TP53 codon 72 encodes the amino acid proline (94, 66, 44, 41 and 34 base pairs). However, only four fragments are obtained when TP53 codon 72 encodes the amino acid arginine (138, 66, 41 and 34 base pairs). Heterozygous forms present a digestion profile representative of both alleles ( Figure 1D). BstUI leads to generation of two fragments when TP53 codon 72 encodes for arginine (160 and 119 base pairs). However, this enzyme is not able to cleave TP53 encoding homozygous proline in the codon 72. Heterozygous forms present a digestion profile representative of both alleles ( Figure 1E).
The agreement between methods is summarized in table 2. We observed high agreement in all combinations tested. Dot-blot hybridization and DHPLC results were more highly concordant with each other (kappa coefficient 0.92). Of the total of 962 samples analyzed, only 43 (4.5%) presented discordant results. The higher discordance was observed between DHPLC and RFLP using BstUI (kappa coefficient 0.83). We observed disagreement between 102 samples, approximately 10% of all samples tested. The comparison between RFLP using BstUI with all methods tested revealed that this enzyme may lead to misclassification. However, this disagreement was not sufficient to lead to departure from Hardy-Weinberg equilibrium. With all methods tested, the majority of samples were genotyped as heterozygous, followed by a high percentage of homozygous arginine and a lower percentage of homozygous proline.

Discussion
Traditional studies that investigated individual TP53 polymorphisms in case-control studies of limited size have not given definitive answers and new approaches are required. High-throughput methodologies and consortium studies investigating large number of individuals may provide the robustness these investigations require while minimizing genotyping errors [3].  For cervical cancer, a meta-analysis identified that the most important factor contributing to heterogeneity among results for invasive lesions was departure from Hardy-Weinberg equilibrium in the control group [8]. According to these authors, various methodological features can contribute to departures from Hardy-Weinberg equilibrium and consequently to less than ideal circumstances for the examination of this polymorphism. To avoid these pitfalls, confirming genotype results with a second method can significantly reduce genotyping misclassification [9].
The results presented in this report indicate that DHPLC and Dot Blot can be used as reliable methods to identify TP53 codon 72 polymorphism. The comparison between these methods and RFLP, a well established methodology, indicate that Dot-blot and DHPLC results are more highly concordant with each other than when either of these methods was compared with RFLP only.
RFLP analysis was one of the first techniques to be widely used for detecting variation at the DNA sequence level [10]. The principle behind the technology rests on the possibility of comparing band profiles generated after restriction enzyme digestion in DNA molecules of different individuals. Diverse restriction sites that might have occurred affect DNA molecules in different ways, producing fragments of variable lengths. These differences in fragment lengths can be seen after gel electrophoresis [11]. However, several conditions need to be optimized in order to achieve the high sensitivity (the amount of DNA, the time for complete digestion, acrylamide concentration, etc). Additionally, if the DNA fragment is only partially digested the interpretation of band patterns may lead to misclassification. Compared with RFLP, dot-blot hybridization can be considered a moderately demanding technique. However, Dot blot is much less time consuming since a large number of samples can be hybridized on a single membrane. On the other hand, DHPLC can be considered a quick and easy screening method. This method allows the identification of mutations and polymorphisms based on detection of heteroduplex formation between mismatched nucleotides in double stranded PCR amplified DNA. Sequence variation creates a mixed population of heteroduplexes and homoduplexes during reannealing of heterozygous and homozygous DNA. When this mixed population is analyzed by HPLC under partially denaturing temperatures, the heteroduplexes elute from the column earlier than the homoduplexes because of their reduced melting temperature [12]. Thus, the sensitivity of this method is dependent on the temperature at which the analysis is completed. For the analysis of TP53 codon 72 polymorphism we have relied on the software-based predictions and we also selected overlapping temperatures to increase the chance of heteroduplex detection. Although we have not detected any false negative or false positive cases using DHPLC and Dot Blot, it is advisable to sequence a small percentage of ran-  domly selected samples as a standard procedure for quality control purposes. The high concordance between DHPLC and Dot Blot reveal that both methodologies are sensitive, reliable, fast and reproducible. Our results demonstrate that these methods can be used in combination with each other or in combination with RFLPBsaJI and RFLP-BstUI to specifically detect TP53 codon 72 polymorphism. Even with some level of disagreement, observed when RFLP-BstUI was used as genotyping method, we did not observe departure from Hardy-Weinberg equilibrium, one of the most important factors contributing to disagreement observed in the literature.

Conclusions
Because TP53 codon 72 polymorphism has been associated with cancer susceptibility, prognosis, response to treatment, and even survival it is of fundamental importance to efficiently and accurately detect this polymorphism in biological specimens. The use of appropriate methods may have a remarkable importance for defining the real role of TP53 codon 72 polymorphism in many types of human neoplasias.

Samples
We tested for TP53 codon 72 polymorphism specimens from a sub-cohort of 978 Brazilian women enrolled in the Ludwig-McGill cohort study [13]. The Ludwig-McGill Cohort Study is an ongoing investigation of the natural history of HPV infection and the risk of cervical neoplasia. This study adhered to the tenets of the Declaration of Helsinki. All participants entered the study only after giving signed informed consent. All study procedures and the informed consent were approved by the institutional review boards and ethical committees of the Ludwig Institute for Cancer Research and the Maternidade Escola Vila Nova Cachoeirinha clinic, both in São Paulo, Brazil [13]. DNA was extracted from exfoliated cervical cells and purified by spin column chromatography. Four different methodologies were used to detect TP53 codon 72 polymorphism: DHPLC, RFLP using two different restriction enzymes -BstUI and BsaJI and Dot-blot hybridization. Because TP53 codon 72 polymorphism is located at TP53 exon 4, previous amplification of this region was performed. The resultant amplicon was then tested by the methodologies described below.

Amplification of TP53 exon 4
To establish and optimize TP53 codon 72 genotyping we performed a previous amplification of TP53 exon 4. Since this exon encodes a polyproline domain [14] GC-rich regions are frequently found. This condition favoured the generation of primer-dimers during amplification leading to low yield of PCR product. To avoid this problem we used hot start PCR to increase specificity [15]. In addition, because optimal primers annealing temperatures differ for more than 6°C we also used touchdown PCR to prevent non-specific extensions. The combination of both PCR strategies improved the yield of amplification and a single fragment of 279 base pairs was obtained (Figure 1A). The total reaction volume for DNA amplification was 50 μL. Each reaction was carried out using 0.5 μM of each of the two primers: TP53 +, 5'-TCCCCCTTGC-CGTCCCAAG-3'; and TP53 -, 5'-CGTGCAAGTCACA-GACTT-3'; in a mixture containing 10 mM Tris-HCl (pH 8.5), 50 mM KCl, 1.5 mM MgCl 2 , 200 mM of each deoxiribonucleotides, and 2 U of Taq Gold DNA Polymerase (Applied Biosystems --Roche, New Jersey, NY). All other reagents were purchased from GIBCO BRL (Gaithersburg, MD). One microliter of DNA was added to the reaction. A negative control containing no DNA was also included with each series of reactions to check for contamination. For controls, we included 50 ng of DNA from SW756 cells and from a penile condyloma, known to be heterozygous and homozygous TP53 -Pro in codon 72, respectively. Reactions were conducted in a TC-341 thermal controller (Amersham Pharmacia Biotech, Buckinghamshire, England) with the following amplification profile: The following amplification profile was employed: 95°C for 10 min, 10 cycles of 95°C for 1 min, 58°C for 1 min and 72°C for 1 min followed by 15 cycles of 95°C for 1 min, 56°C for 1 min and 72°C for 1 min and an additional round of 15 cycles of 95°C for 1 min, 53°C for 1 min and 72°C for 1 min. A final extension step was added including 72°C for 3 min followed by 95°C for 10 min. Amplification products were run on 1.5% agarose gels containing 3 mg/ml of ethidium bromide at 90--100 V for 20--30 min. Bands were visualized and photographed on a UV transilluminator. Positive reactions were selected for TP53 codon 72 polymorphism detection.

Denaturing HPLC Analysis (DHPLC)
PCR products were heated at 95°C for 10 min and allowed to cool down to room temperature for approximately 30 min. Five to 10 μL of each sample were run on a Wave DNA Fragment Analysis System (Transgenomic, Omaha, NE) using a DNASep column and the run was monitored by ultraviolet light (UV) (260 nm). Optimum DHPLC temperatures were determined by an incremental temperature scan, using the software-predicted melting profile as a starting point (Transgenomic, Omaha, NE). Samples were run at more than one temperature due to the heterogeneity in the distribution of GC-rich regions that results in a heterogeneous melting temperature distribution throughout the amplified DNA fragments to be analyzed. By using 59°C as a melting temperature we were able to identify the formation of heteroduplexes in a subset of samples submitted to the analysis. The heteroduplexes are frequently found in samples with differences in alleles. Thus, heterozygous samples were identified due to a small shoulder peak in the chromatogram. After distinguishing the heterozygous samples, the subset of homozygous samples were mixed in approximately equimolar proportions with a control sample with TP53 codon 72 polymorphism previously identified as homozygous proline. This mixture was heated at 95°C for 10 min and allowed to cool down to room temperature for approximately 30 min. Five to 10 μL of each sample were then submitted to a new DHPLC analysis. This allowed differentiating the two homozygous genotypes. Heteroduplexes were observed in samples with genotype homozygous arginine, since the PCR product added to the sample was known to be homozygous proline. The homoduplexes, or samples with a single peak, were observed in samples with genotype proline, since it was the same genotype presented by the control sample added to the mixture. As an additional control, 10% of the samples were randomly selected and mixed with the same equimolar proportions of a control sample previously identified as homozygous arginine. Results were compared and no discordant results were observed between the two pulled analyses.

Restriction Fragment Length Polymorphism (RFLP)
After TP53 exon 4 amplification, 7 μL of the amplified reaction products were digested in a 10 μL volume of 1 × NEB buffer 2 and 5 U of BsaJI or BstUI endonucleases (New England BioLabs, Newton, MA) at 60°C for at least 2 hours. Products digested by BsaJI were run in 12% polyacrylamide gel and products digested by BstUI were run in 8% polyacrylamide gel. In both cases silver staining was used to identify band profile [16].
To ensure quality control of all genotyping methods, the samples were blindly read by two persons independently (TR, LLV). Despite the high percentage of agreement, around 95% for each method, the samples with discordant results were submitted to a new round of amplification and genotyping.

Statistical Analysis
Cohen's kappa coefficients of agreement were computed to evaluate the inter-methodology variability in genotyping.