Limitations of genotyping based on amplicon melting temperature.
We have read with interest the recent reports of Marziliano et al. (1) and Pirulli et al. (2), who used melting temperature assays to genotype different types of mutations. Because detection of the underlying mutation is only indirect in these methods, they differ in their molecular detection principle from hybridization probe-based genotyping (3) or allele-specific amplification coupled to SYBR-green I detection (4). Indirect detection methods demand extra caution in the assignment of genotypes solely on the basis of product melting temperature ([T.sub.m]). From a theoretical standpoint, some genotypes will not be detected.
The UDP-glucuronosyltransferase 1 (UGT1A1) (TA)n insertion/deletion polymorphism is of significance for the manifestation of Gilbert disease (3,5). TA repeats are intrinsically unstable, and, therefore, it is not surprising that five to eight TA repeats occur in humans and have functional significance (5). The method of Marziliano et al. (1) uses [T.sub.m] as a measure of amplicon length to discover the UGT1A1 promoter genotype. In the described assay, a 2-bp difference is detected by the resulting [T.sub.m] shift in a 130-bp amplicon. A similar technique was successfully used in the screening for a 9-bp deletion in a 55-bp PCR amplicon (6). However, for the assay of Marziliano et al. (1), it has to be noted that using only an indirect measure of strand length and composition is insufficient for the assignment of a genotype on the basis of the theoretical reasons described below.
In addition to the points stated by Marziliano et al. (1), the melting curve analysis of a DNA strand is related to (a) the ionic strength of the buffer, (b) the DNA concentration, and (c) the DNA bases' nearest neighbors (n-n) (3,7). Within an assay system, buffer conditions can be considered constant. The genotyping of heterozygous samples requires special consideration because a mixture of two homoduplexes and two heteroduplexes results after PCR amplification, denaturation, and reannealing. An insertion/ deletion polymorphism causes the formation of base bulges in the heteroduplexes. The bulge size is a major determinant of the destabilization caused by the disturbed base stacking (3). For example, the DNA base bulge size is 4 by in the case of a 6TA/8TA duplex and 2 by in the case of a 6TA/7TA or 7TA/8TA heteroduplex. Reported [T.sub.m]s for the 6TA and 7TA genotype differ by only 1.3[degrees]C in the 130-bp amplicon (1). A heterozygous 6TA/ 7TA sample has an apparent [T.sub.m] in between the 6TA and 7TA sample because the different [T.sub.m]s of the underlying homo- and heteroduplexes cannot be adequately resolved. On the basis of these respective considerations, we anticipate that in patients heterozygous for the 6TA/8TA genotype, a [T.sub.m] undistinguishable from that of a homozygous 7TA genotype results. The presence of an 8TA genotype was already reported in an Italian patient with Gilbert syndrome (8). Other allelic combinations are possible where the same problem is present. Care must also be taken to ensure a constant DNA concentration in the assay before melting curves are acquired. Variation in the purity or amount of DNA in the assay can lead to different amounts of DNA after a constant number of PCR cycles. In the described assay (1), this will add to the total error and will cause variation in the [T.sub.m] with the risk of wrong genotyping results.
This is not acceptable for a genotyping assay, and we recommend the use of more specific methods for allele assignment. Methods that ensure reliable genotyping for this locus have already been published and include polyacrylamide gel electrophoresis resolution of the PCR amplicon size (5), hybridization probes (3), sequencing, denaturing gradient gel electrophoresis, and denaturing HPLC [see references in (3)]. All of these methods also have the potential to detect the exact TA repeat numbers, which is not necessarily the case if only the amplicon [T.sub.m] is used for screening.
Pirulli et al. (2) claim "sensitivity" and "specificity" of the DNA melting assay for the detection of different types of alanine:glyoxylate aminotransferase (AGXT) mutations. In addition to what was already mentioned above, we want to point out some results in Table 1 of Ref. (2). Both sample 9 and sample 13 share a homozygous G [right arrow] A mutation. However, the resulting [T.sub.m] shifts are 1.6[degrees]C and -0.9[degrees]C, respectively. In general, n-n pairs containing guanosine are more stable than n-n pairs containing adenosine. When a mutation in a strand causes the change of a guanosine n-n pair to an adenosine n-n pair, we would expect this to be destabilizing in most of the cases. Sample 2 is, in contrast to sample 11, stabilized by the presence of mismatches. The observed [T.sub.m] of sample 2 is 0.8[degrees]C higher than for the wild type, although the duplex is destabilized by two mismatches in both cases. This illustrates the deviation from n-n behavior in longer DNA strands and makes the results difficult to interpret. Furthermore, sometimes a product [T.sub.m] differs by only 0.1[degrees]C, whereas mean SDs are 0.14[degrees]C (samples 15 and 2; samples 6 and 11). This makes it impossible to "easily distinguish different types of AGXT mutations" as claimed by Pirulli et al. (2). This is exemplified in Fig. 1, which does not show melting curves, but the expected probability distribution defined by 88.0[degrees]C with an SD of 0.14[degrees]C and 88.1[degrees]C with an SD of 0.18[degrees]C. The large overlap in the distributions is obvious and substantiates that it is impossible to assign certain [T.sub.m]s to underlying genotypes.
[FIGURE 1 OMITTED]
Our conclusion is that DNA [T.sub.m] assays based on SYBR Green I melting curves are powerful tools for screening various types of mutations but lack both sequence and mutation specificity. A lack of sensitivity must be expected if stable mismatches occur among stable neighboring bases (high GC content). Results obtained with these methods at highly polymorphic loci must be confirmed by a mutation-specific detection method. Interpretation of results arising from only the inspection of SYBR Green I melting curves requires great caution. In our opinion, such methods should not be used for routine genotyping applications.
(1.) Marziliano N, Pelo E, Minuti B, Passerini I, Torricelli F, Da Prato L. Melting temperature assay for a UGT1A gene variant in Gilbert syndrome. Clin Chem 2000;46:423-5.
(2.) Pirulli D, Boniotto M, Puzzer D, Spano A, Amoroso A, Crovella S. Flexibility of melting temperature assay for rapid detection of insertions, deletions, and single-point mutations of the AGXT gene responsible for type 1 primary hyperoxaluria. Clin Chem 2000;46:1842-4.
(3.) von Ahsen N, Oellerich M, Schutz E. DNA base bulge versus unmatched end formation in probe-based diagnostic insertion/deletion genotyping. Genotyping the UDP-glucuronosyltransferase 1 (UGT1A1) (TA)n polymorphism by real time fluorescence PCR. Clin Chem 2000; 46:1939-45.
(4.) Germer S, Higuchi R. Single-tube genotyping without oligonucleotide probes. Genome Res 1999;9:72-8.
(5.) Beutler E, Gelbart T, Demina A. Racial variability in the UDP-glucuronosyltransferase 1 (UGT1A1) promoter: a balanced polymorphism for regulation of bilirubin metabolism? Proc Natl Acad Sci U S A 1998;95:8170-4.
(6.) Aoshima T, Sekido Y, Miyazaki T, Kajita M, Mimura S, Watanabe K, et al. Rapid detection of deletion mutations in inherited metabolic diseases by melting curve analysis with LightCycler. Clin Chem 2000;46:119-22.
(7.) Wetmur JG. DNA probes: applications of the principles of nucleic acid hybridization. Crit Rev Biochem Mol Biol 1991;26:227-59.
(8.) Iolascon A, Faienza MF, Centra M, Storelli S, Zelante L, Savoia A. (TA)8 allele in the UGT1A1 gene promoter of a Caucasian with Gilbert's syndrome. Haematologica 1999;84:106-9.
Nicolas von Ahsen
Department of Clinical Chemistry
37075 Goettingen, Germany
* Author for correspondence. Fax 49551-39-8551; e-mail email@example.com.
|Printer friendly Cite/link Email Feedback|
|Author:||von Ahsen, Nicolas; Oellerich, Michael; Schutz, Ekkehard|
|Article Type:||Letter to the editor|
|Date:||Jul 1, 2001|
|Previous Article:||How to improve total error modeling by accounting for error sources beyond imprecision and bias.|
|Next Article:||Use of heterophilic antibody blocking agent (HBT) in reducing false-positive hCG results.|