BACKGROUND: Gilbert syndrome, a chronic nonhemolytic unconjugated hyperbilirubinemia, is associated with thymine–adenine (TA) insertions in the UGT1A1 (UDP glucuronosyltransferase 1 family, polypeptide A1) promoter. The UGT1A1 promoter genotype also correlates with toxicity induced by the chemotherapeutic drug irinotecan. Current closed-tube assays for genotyping the UGT1A1 (TA)n promoter polymorphism require multiple labeled probes and/or have difficulty classifying the (TA)5 and (TA)8 alleles.
METHODS: An unlabeled 5′ extension on one primer that creates a hairpin after asymmetric PCR was used to develop a snapback primer high-resolution melting assay for the (TA)n polymorphism. A new method that plots the local deviation from exponential decay to improve genotype clustering was used to remove background fluorescence and to analyze the data. The snapback assay was compared with small-amplicon melting and fragment length analyses in a blinded study of DNA samples from 100 African Americans.
RESULTS: Genotyping results obtained by small-amplicon melting and snapback primer melting were 83% and 99% concordant, respectively, with results obtained by fragment analysis. Reanalysis of the single discordant sample in the results of the snapback genotyping assay and the fragment analysis revealed an error in the fragment analysis. High-resolution melting was required for accurate snapback genotyping of the UGT1A1 (TA)n polymorphism. The 100% accuracy obtained with a capillary-based instrument fell to ≤81% with plate-based instruments.
CONCLUSIONS: In contrast to small-amplicon genotyping, snapback primer genotyping can distinguish all UGT1A1 promoter genotypes. Rapid-cycle PCR combined with snapback primer analysis with only 2 unlabeled PCR primers (one with a 5′ extension) and a saturating DNA dye can genotype loci with several alleles in <30 min.
UDP-glucuronosyltransferases (UGTs)4 are important enzymes involved in the glucuronidation of diverse exogenous and endogenous compounds. Glucuronidation converts water-insoluble compounds to more soluble forms that can then be excreted from the body. Glucuronidation is carried out by a multigene family of enzymes found predominantly in the liver 1. This family is generally divided into 2 distinct subfamilies, UGT1 and UGT2, which are further subdivided into isoforms 2. One of these isoforms, encoded by the UGT1A1 (UDP glucuronosyltransferase 1 family, polypeptide A1) gene, is an enzyme that conjugates bilirubin with glucuronic acid in the liver, making bilirubin more water soluble 3.
Rare mutations in UGT1A1 produce the unconjugated hyperbilirubinemias—Crigler–Najjar syndromes I and II (OMIM 218800, 606785)—with absent and severely reduced enzyme activity, respectively 4,–,6. In contrast, Gilbert syndrome (OMIM 145300), a mild inherited form of unconjugated hyperbilirubinemia that occurs in 15% of the population, is usually not treated or diagnosed because there are usually no clinical symptoms 7, 8; however, Gilbert syndrome increases the risk of drug toxicity during cancer chemotherapy with irinotecan (Camptosar®; Pfizer) 9,–,11. This finding prompted safety relabeling and a recommendation for a reduced initial dose, depending on the genotype 12.
Gilbert syndrome correlates with thymine–adenine (TA) insertions in the TATA box of the UGT1A1 promoter (rs8175347) 7, 13. The wild-type sequence contains an A(TA)6TAA segment and is designated UGT1A1*1. A single TA insertion in the promoter produces the most common disease variant, A(TA)7TAA (UGT1A1*28), which leads to an 18% to 33% reduction in transcriptional activity and a 48% reduction in bilirubin glucuronidation when the allele is homozygous 7, 14. The frequency of the (TA)7 allele varies among ethnic groups: 32%–39% in Caucasians, 16%–33% in Asians, and 40%–43% in Africans 13, 15, 16. In addition, the (TA)5 and (TA)8 alleles (UGT1A1*36 and UGT1A1*37) occur in Africans at frequencies of 3.5%–16% and 3.1%–7%, respectively, 13, 17. Enzyme production is inversely related to the number of TA repeats.
Several assays for genotyping the UGT1A1 promoter (TA)n polymorphism have been described. These assays include an Invader® assay (Holologic/Third Wave Technologies) [cleared by the US Food and Drug Administration 18], pyrosequencing 19, dual hybridization probe melting 20, 21, fragment analysis of fluorescently labeled PCR products 22, hydrolysis probes 23, denaturing HPLC24, and small-amplicon genotyping by high-resolution melting 25. Most of these assays identify only the (TA)6 and (TA)7 alleles. Those that can type the (TA)5 and (TA)8 alleles require multiple manual steps and/or multiple labeled probes. Although the high-resolution melting assay is simple to perform and requires no processing, genotypes containing (TA)5 and (TA)8 alleles are not genotyped accurately 26.
To maintain the simplicity of high-resolution melting while accurately genotyping all alleles, we adapted snapback primer melting analysis 27 to genotype UGT1A1 (TA)n repeats. Snapback primers consist of unlabeled probes attached to the 5′ end of PCR primers. The probe portion is synthesized to be complementary to the primer's extension product so that an intramolecular hairpin is formed after the PCR. If the hairpin stem includes a polymorphism, its melting transition depends on the genotype. Snapback primer genotyping has previously been applied to single-base variants and small deletions. This report is the first of genotyping simple sequence repeats by snapback primer analysis.
Materials and Methods
Genomic DNA from an African American Human Variation Panel (HD100AA-1, n=100) was purchased from the Coriell Institute for Medical Research. DNA samples were diluted in 10 mmol/L Tris and 0.1 mmol/L EDTA, pH 8.0, to 40–55 ng/μL, as measured by the absorbance at 260 nm (NanoDrop 1000; Thermo Fisher Scientific).
GENOTYPING BY FRAGMENT ANALYSIS USING CAPILLARY ELECTROPHORESIS
Initial genotyping of the (TA)nUGT1A1 promoter polymorphism (GenBank accession number NG_002601) was performed by PCR amplification and fragment analysis. The PCR was performed in 10-μL reaction volumes containing 1× LCGreen® Plus (Idaho Technology), 50 mmol/L Tris (pH 8.3), 500 μg/mL nonacetylated BSA, 2.7 mmol/L MgCl2, 200 μmol/L of each deoxynucleoside triphosphate, 0.4 U Klentaq1™ polymerase (Ab Peptides), 64 ng Anti-Taq Monoclonal Antibody (eEnzyme), approximately 50 ng genomic DNA, and 0.5 μmol/L of each primer. The forward primer was 5′-MAX-ACAGTCAAACATTAACTTGGTGTATC-3′ (MAX is a fluorescent dye with emission at 524 nm and excitation at 557 nm; Integrated DNA Technologies), and the reverse primer was 5′-GTTTCTTAGGTTCGCCCTCTCCTA-3′ (University of Utah Core Facility). The reverse primer was tailed (underlined) to promote adenylation and reduce interference by double peaks in the data analysis 28. Thermal cycling was performed on a capillary thermal cycler (LS32™; Idaho Technology) with an initial denaturation step at 95 °C for 15 s followed by 35 cycles of 95 °C for 0 s, 60 °C for 0 s, and 72 °C for 3 s. A final extension at 72 °C for 10 min was performed to further promote adenylation of the PCR products. The programmed ramp rates were 20 °C/s from denaturation to annealing, 1.5 °C/s from annealing to extension, and 20 °C/s from extension to denaturation. The sizes of the amplified products ranged from 79 bp for (TA)5 to 85 bp for (TA)8. PCR products were purified by gel filtration (Performa® DTR; Edge BioSystems) to remove excess labeled primer and LCGreen Plus dye and then diluted with 50 volumes of DNase- and RNase-free water. The diluted PCR product (1 μL) was then combined with 12 μL formamide (Life Technologies) and 1 μL of an allele standard in a 96-well plate, denatured at 95 °C for 3 min on a PTC-200 thermal cycler (Bio-Rad Laboratories), and cooled on ice. Samples were separated by capillary electrophoresis on a 3130 Genetic Analyzer (Applied Biosystems/Life Technologies) using Performance Optimized Polymer 6 and a 36-cm capillary array. Data were imported into Mutation Surveyor® (SoftGenetics), and genotyping was performed visually against a known allele size standard.
The size standard was generated by PCR amplification of a known (TA)6/(TA)7 heterozygote as described above, except that the forward primer was labeled with fluorescein (emission at 495 nm, excitation at 520 nm; University of Utah Core Facility). After gel filtration, the purified allele standard was diluted with 100 volumes of water and stored at 4 °C until use.
The PCR was performed as described for fragment analysis except that the forward primer was not fluorescently labeled, the reverse primer was not tailed, and a final extension was not performed. Each sample was run in duplicate, and every run included a no-template control and all available genotype controls (Table 1). Amplified products ranged in size from 72 bp for (TA)5 to 78 bp for (TA)8.
SNAPBACK PRIMER GENOTYPING
PCR was performed as described for small-amplicon genotyping except that the concentration of the reverse primer was reduced 5-fold (to 0.1 μmol/L), 45 cycles were performed at an annealing temperature of 65 °C, and the forward snapback primer was 5′-ggCCTACTTATATATATATATGGCAAACAGTCAA ACATTAACTTGGTGTATC-3′, which included a self-complementary probe sequence (underlined) and a 2-bp mismatch at the 5′ end (lowercase). Each sample was run in duplicate, and every run included no-template and (TA)6/(TA)7 controls. Amplified products ranged in size from 98 bp for (TA)5 to 104 bp for (TA)8.
Snapback primer genotyping was also performed on a plate-based real-time PCR instrument (LC480; Roche Applied Science) in a second blinded study that used 95 of the original 100 DNA samples. Equivalent amplification was achieved after 20 s of initial denaturation at 95 °C followed by 45 cycles of annealing at 67 °C for 3 s, extension at 76 °C for 10 s, and denaturation at 95 °C for 1 s.
Snapback primer genotyping on the 7500 Fast Dx instrument (Applied Biosystems) was also attempted; however, the fixed data-acquisition capabilities (approximately 3 data points/°C) of this instrument and the supplied Sequence Detection Software (version 1.4) were not adequate for evaluating snapback genotyping.
HIGH-RESOLUTION MELTING ANALYSIS
High-resolution melting was performed on the LS32 and LC480 instruments immediately after the PCR without user intervention. For the LS32, samples were held for 10 s at 95 °C, cooled to 40 °C, and then heated continuously at 0.3 °C/s with fluorescence acquisition from 50 °C to 95 °C. For the LC480, samples were held for 30 s at 95 °C, cooled to 50 °C for 1 min, and then heated continuously at approximately 0.2 °C/s (25 fluorescence acquisitions/°C) from 50 °C to 95 °C. Small-amplicon data were displayed as derivative and difference plots after normalization and exponential background removal 29. Snapback melting data were analyzed with a new “deviation” method for separating signal from background. The deviation method performs a running exponential fit on the experimental data: E(T)=CeA(T − Tl), where E(T) is the best least squares fit of the data within a running temperature window of lower bound Tl, C is the amplitude constant, and A is the exponential decay factor. The deviation melting curve (analogous to the derivative melting curve) is plotted as A(T) vs T. Deviation analysis appears better at identifying melting transitions that occur at low temperatures (such as those of hairpin structures) than derivative or exponential background-subtraction methods 29. Genotypes were determined by both the position (temperature) and the shape of the melting curves.
The UGT1A1 (TA)n repeat region was genotyped in 100 African American DNA samples by 3 different methods. DNA samples were first genotyped via a reference method: capillary electrophoresis of fluorescently labeled PCR products (Table 1). Among the DNA samples analyzed, all (TA)n genotypes were present except homozygous (TA)5 and (TA)8. The most common genotype was (TA)6/(TA)7, which represented 33% of the samples studied. The homozygous (TA)6/(TA)6 and (TA)7/(TA)7 genotypes were also common (22% and 16%, respectively) in this African American population. Genotypes other than the wild-type (TA)6/(TA)6 genotype were found in 78% of the studied samples. Of the 100 DNA samples, 29 contained alleles other than (TA)6 and (TA)7. Allele frequencies were 0.075 for (TA)5, 0.450 for (TA)6, 0.395 for (TA)7, and 0.080 for (TA)8.
Samples genotyped by fragment analysis were blinded and analyzed by 2 high-resolution melting techniques: small-amplicon melting and snapback primer genotyping. The probe element of the snapback primer hybridized perfectly to the (TA)5 allele; consequently, the (TA)6, (TA)7, and (TA)8 alleles form increasingly larger bulges (Fig. 1) and thus less stable hairpins. Fig. 2A shows representative deviation plots for all of the genotypes studied. Genotypes with lower repeat numbers were more stable than those with higher repeat numbers. For example, the homozygous (TA)6/(TA)6 genotype was 1.5 °C more stable than the (TA)7/(TA)7 genotype. Homozygous genotypes had sharper transitions (thinner peaks) than heterozygous genotypes. Among the heterozygous genotypes, the overall apparent melting transition (including transitions of both alleles) became broader as the difference between repeat numbers increased. Only in the case of the (TA)5/(TA)8 genotype is there a hint of 2 separate peaks. Although the 2 allele peaks are never resolved, the 6 heterozygotes are still easily distinguishable because of the absolute temperature precision of the measurement system and the various shapes of the melting curves. This capability is further demonstrated in Fig. 2B, in which the clustering of multiple samples with the same genotype is evident. Genotype clustering was better with deviation analysis than with derivative analysis, with or without exponential background subtraction (see Fig. 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol57/issue9).
The results of genotyping of the UGT1A1 (TA)n repeat polymorphism with snapback primers were 99% concordant with those obtained by fragment analysis. The single discordant sample was reported as (TA)6/(TA)7 by fragment analysis but as (TA)7/(TA)7 by snapback genotyping (Fig. 3). Although the snapback genotype was clear, the fragment analysis result was difficult to read because of “stutter” bands, which are often seen after amplification of dinucleotide repeats. The correct genotype could be inferred only after careful comparison with standards (see Fig. 2 in the online Data Supplement). Repeat analysis by snapback genotyping and fragment analysis confirmed that the discordance was due to incorrect interpretation of the fragment length analysis.
The accuracy of snapback primer genotyping depended on the instrument resolution. Although 100% accuracy was achieved with the high-resolution capillary format of the LS32 instrument, genotyping accuracy fell to 81% with the high-resolution plate-based LC480 machine (see Fig. 3 in the online Data Supplement); analysis was not even possible on the Applied Biosystems 7500 Fast Dx instrument. The genotypes most frequently miscalled on the LC480 were (TA)6/(TA)7, (TA)6/(TA)8, and (TA)7/(TA)7 (see Table 1 in the online Data Supplement).
In addition to snapback primer analysis, blinded samples were also genotyped by small-amplicon melting on the LS32 instrument. Differences between genotypes in melting temperature (Tm) were small, with the (TA)6/(TA)6Tm being about 0.28 °C greater than for (TA)7/(TA)7. The Tm for the heterozygous (TA)6/(TA)7 genotype was intermediate between the Tms for (TA)6/(TA)6 and (TA)7/(TA)7. Derivative and difference plots for all tested genotypes are shown in Fig. 4. Although most genotypes were adequately separated in difference plots, (TA)6/(TA)7 was difficult to distinguish from (TA)6/(TA)8. Genotyping accuracy after blinded analysis was 84%. The most frequently miscalled genotypes were (TA)6/(TA)7, (TA)6/(TA)8, and (TA)7/(TA)8 (Table 2).
High-resolution melting, an established technique for the homogeneous genotyping and scanning of PCR products 30,–,32, combines precise temperature control and fluorescence monitoring of a saturating dye for observing DNA melting. High-resolution instrumentation and targeted software for data analysis are required to generate reproducible melting curves 33. Small-amplicon melting is the simplest method of genotyping by high-resolution melting and requires only 2 standard PCR primers 34. Genotyping is based on subtle changes in amplicon Tm and curve shape. When greater sequence resolution is required, unlabeled probes may be used in addition to amplicon melting 35. Unlabeled probes are blocked at the 3′ end and are typically complementary to 1 allele in a region of variation. To generate a sufficient signal from probe melting, asymmetric PCR overproduces the product strand complementary to the unlabeled probe. Snapback primers combine the simplicity of amplicon genotyping with the increased sequence resolution of unlabeled probes 27. By adding the probe to the 5′ end of one of the PCR primers, only 2 unlabeled PCR primers are used, and no 3′ blocking is required.
Before the introduction of high-resolution melting, both UGT1A1 (TA)n genotyping by SYBR Green® I melting of PCR products and dual hybridization probe melting had been described. Genotyping a 132-bp product with SYBR Green I was reported to produce a 1.3 °C higher Tm for the (TA)7 homozygote compared with the (TA)6 homozygote 36; however, this method was criticized as being indirect and inadequate for genotyping all combinations of (TA)5 and (TA)8 alleles 37. Greater discrimination was possible with sequence-specific fluorescently labeled dual hybridization probes 20, 21. One such assay 20 successfully genotyped all allelic combinations with 2 sequential reporter probes, although genotypes containing (TA)5 and (TA)8 alleles were amplified from synthetic plasmids. Each reaction required 2 fluorescently labeled probes.
Recently, a high-resolution melting assay successfully genotyped UGT1A1 (TA)6 and (TA)7 alleles from a 70-bp product 25 The accuracy of genotyping (TA)5 and (TA)8 alleles was not evaluated, however, and the Tm for the (TA)7 homozygote was 0.20 °C to 0.42 °C lower than for the (TA)6 homozygote. Compared with the SYBR Green I assay, the ΔTm between homozygotes decreased and reversed direction, both surprising results. With a 47% decrease in amplicon length, the ΔTm is expected to increase, not decrease. In a follow-up report, the same high-resolution melting assay failed to accurately genotype populations containing (TA)5 and (TA)8 alleles 26. The temperature homogeneity of the instruments used (SD, 0.144 °C to 0.160 °C) also suggests that genotyping may be difficult via small-amplicon melting 33.
To test the limits of small-amplicon genotyping, we genotyped the UGT1A1 promoter polymorphism with the recently available LS32 thermal cycler. The LS32 performs both real-time rapid-cycle PCR and high-resolution melting, with a Tm precision similar to that of the gold standard HR-1 instrument 38. DNA samples from an African American population included (TA)5 and (TA)8 alleles at frequencies of 7.5% and 8.0%, respectively. The ΔTm between the common (TA)6 and (TA)7 homozygotes was 0.28 °C, which is similar to prior high-resolution results. Even when all available genotypes were used as controls, (TA)6/(TA)7 and (TA)6/(TA)8 were not easily genotyped (Table 2). (TA)7/(TA)8 heterozygotes were also frequently miscalled as (TA)7/(TA)7. The overall error rate was 16%, which is not acceptable for genotyping the African American population.
To increase genotyping accuracy and both maintain the simplicity of a high-resolution melting assay and avoid labeled probes, we performed the PCR with a primer containing a 5′ addition complementary to the UGT1A1 (TA)n repeat region. Asymmetric amplification overproduces a single-stranded product that “snaps back” to form an intramolecular hairpin. A 2-bp mismatch at the 5′ end of the addition prevents extension of the minor PCR product hairpin that has an extendable 3′ end 27. The intramolecular probe includes 5 bp on each side of the smallest repeat to anchor the edges so that longer alleles form a bulge within the repeat. The stability of the hairpin is dependent on the genotype and is assessed by high-resolution melting. Larger bulge loops progressively decrease the hairpin Tm.
Methods for analyzing fluorescence melting curves continue to evolve. Fluorescence intensity is almost always normalized to adjust for sample volume and eliminate optical differences. More-advanced techniques subtract the fluorescence background so that melting curves are horizontal outside of the melting transitions, permitting comparison with curve predictions 39. Although linear background extrapolation is adequate for the melting of PCR products, exponential background subtraction better displays melting curves with both amplicon and unlabeled-probe transitions 29. Finally, we have introduced a “deviation” method for analyzing hairpin melting curves. Hairpin melting, as we have described for (TA)n genotyping, clusters better by deviation analysis than by exponential background subtraction or direct display of the derivative after normalization. An additional advantage of deviation analysis is that it is not necessary to define temperature regions for analysis. Although not yet available on commercial instruments, the deviation method we have described can be implemented on exported data.
When run on a capillary high-resolution instrument, the snapback primer assay with deviation analysis correctly genotyped all blinded DNA samples, including 1 discordant sample misclassified by capillary electrophoresis. Although some genotypes are difficult to classify by Tm alone, all genotypes are distinct when both Tm and curve shape are considered on deviation plots. A further benefit of the snapback primer assay compared with small-amplicon genotyping is that only the (TA)6/(TA)7 control is necessary. Tm and curve shapes are sufficiently different to allow accurate genotyping in the absence of other controls. Instrument resolution is critical for successful genotyping, and plate-based high-resolution instruments did not perform as well as the capillary format.
In addition to their application to single-base variants and small deletions, snapback primers can be used to genotype repeat regions. UGT1A1 promoter genotyping is an example in which all combinations of 4 alleles (4 homozygotes and 6 heterozygotes) are distinguishable. As the number (n) of alleles increases, the number of possible diploid combinations [n + n(n − 1)/2] also increases. The number of combinations that can be distinguished depends on the resolution of the melting analysis and should improve as the instrumentation improves. Snapback primer genotyping of repeats may also be limited by the length of the repeat region. Long repeats will require long primer tails, which are more difficult and expensive to synthesize. In addition, long repeats will produce long hairpin stems that may be less sensitive to allele differences, making genotyping more difficult.
The correlation between UGT1A1 promoter genotype and irinotecan-induced drug toxicity requires genotyping assays that are rapid, accurate, and inexpensive; however, the high prevalence of the (TA)5 and (TA)8 promoter alleles among persons of African descent and the increasing ethnic admixture in the global community create particular concern for assays that have not been validated for these alleles. Additionally, the (TA)8 allele, although rare, has been found in Caucasians and can arise via spontaneous mutation 40. High-resolution melting offers considerable time and cost savings compared with other genotyping methods. Using the LS32 thermal cycler permits the completion of both the PCR and high-resolution melting in <30 min without user intervention. In contrast to small-amplicon genotyping, snapback primers accurately identify all genotypes in a population with the (TA)5, (TA)6, (TA)7, and (TA)8 alleles.
↵4 Nonstandard abbreviations:
- melting temperature
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest:
Employment or Leadership: C.T. Wittwer, Idaho Technology and Clinical Chemistry, AACC.
Consultant or Advisory Role: None declared.
Stock Ownership: C.T. Wittwer, Idaho Technology.
Honoraria: None declared.
Research Funding: C.T. Wittwer, Idaho Technology.
Expert Testimony: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
- Received for publication March 25, 2011.
- Accepted for publication June 27, 2011.
- © 2011 The American Association for Clinical Chemistry