High resolution melting analysis for gene scanning

Abstract

High resolution melting is a new method of genotyping and variant scanning that can be seamlessly appended to PCR amplification. Limitations of genotyping by amplicon melting can be addressed by unlabeled probe or snapback primer analysis, all performed without labeled probes. High resolution melting can also be used to scan for rare sequence variants in large genes with multiple exons and is the focus of this article. With the simple addition of a heteroduplex-detecting dye before PCR, high resolution melting is performed without any additions, processing or separation steps. Heterozygous variants are identified by atypical melting curves of a different shape compared to wild-type homozygotes. Homozygous or hemizygous variants are detected by prior mixing with wild-type DNA. Design, optimization, and performance considerations for high resolution scanning assays are presented for rapid turnaround of gene scanning. Design concerns include primer selection and predicting melting profiles in silico. Optimization includes temperature gradient selection of the annealing temperature, random population screening for common variants, and batch preparation of primer plates with robotically deposited and dried primer pairs. Performance includes rapid DNA preparation, PCR, and scanning by high resolution melting that require, in total, only 3h when no variants are present. When variants are detected, they can be identified in an additional 3h by rapid cycle sequencing and capillary electrophoresis. For each step in the protocol, a general overview of principles is provided, followed by an in depth analysis of one example, scanning of CYBB, the gene that is mutated in X-linked chronic granulomatous disease.

Fig. 1

Consensus sequences for splicing and branch sites of human exons. The branch site is typically 20–50 bases upstream of the exon. Splice sites bracket each exon with consensus bases reaching at least 15 bases upstream and 6 bases downstream. The colored bars indicate the proportion of each nucleotide at the position given (A, C, T, G_■). Primers should be designed to avoid consensus bases where variation is likely to effect splicing.

Fig. 2

Ninety-five melting curves of CYBB exon 1 after PCR from the DNA of 95 healthy individuals. After exponential background subtraction, melting curves are shown normalized and overlaid (top) and as a difference from the average (bottom). No sequence variations were detected in the 146 bp product with two domains, typical of the low variation found in CYBB. Melting curves were acquired on the LightScanner in the presence of the saturating DNA dye, LCGreen Plus. The overall reproducibility of PCR and melting analysis can be assessed from the tightness of such clusters, typically +/− 1% on difference plots.

Fig 3

Workflow diagram for CYBB mutation scanning by high resolution melting analysis. Wild type samples containing no variants can be reported within 3 hours. Rare samples with aberrant melting profiles are sequenced for variant identification and can be reported within 6 hours. (A) DNA preparation from whole blood in 60 min (Roche MagNA Pure Compact) with quantification and dilution in 15 min (NanoDrop ND-8000 Spectrophotometer), (B) PCR preparation in 15 min with PCR cycling in 45 min (BioRad C1000), (C) Mutation scanning in 15 min (Idaho Techonology LightScanner 96), (D) Scanning data analysis in 30 min → if no variants detected, report within 3 hours, (E) Cycle sequence preparation (PCR product dilution, master mix preparation, and plate loading) in 30 min followed by cycle sequencing in 45 min (BioRad C1000), (F) Capillary electrophoresis preparation (ABI 3130xl) and product clean-up in 30 min with capillary electrophoresis in 45 min, (G) Sequencing data analysis in 45 min → report variant(s) within 6 hours. The hatched area indicates overlap of cycle sequencing and capillary electrophoresis preparation.

Fig. 4

Normalized and overlaid high resolution melting curves for all 13 CYBB exons. All traces were generated on a single 96-well plate using 3 samples run in duplicate. The samples analyzed were an unknown male, a wild type male (WT) and a mixed sample containing both unknown and WT. The traces for exon 11 (circled) suggested a heterozygous sequence in the mixed sample. The traces for the other 12 exons were normal. Additional analysis of exon 11 for this unknown male sample is shown in Fig. 5.

Fig. 5

High resolution melting analysis of CYBB exon 11 in wild type DNA (left panels) and variant DNA (right panels). Melting curves are displayed normalized (top), normalized and overlaid (middle) and as difference curves (bottom). In X-linked disorders, variants are best detected after heteroduplex formation by mixing unknown DNA with wild type (WT) DNA. Less reliable is detecting the unknown sample without mixing, although in this case the unmixed hemizygous sample (Unknown) can also be distinguished from wild type by difference curves. The unknown male sample had a T duplication at c.1456 by sequencing.