Accordingly, the method was then validated recording fluorescence at each 0.3 °C step. Tm was directly measured by the internal software (StepOne Software v2.1; Applied Biosystems). Data were exported and processed according to mathematical algorithms for high-resolution DNA melting analysis (Palais & Wittwer, 2009). Briefly, the background was evaluated MEK inhibitor and removed to the negative derivative of the fluorescence data. The results obtained were then normalized and smoothed with the ‘running average’ method. Graphs were generated with Sigma

Plot 5.0 (SSI, CA). To develop the method, we used three different Map strains, carrying three, four and five repeats; unfortunately, we did not have any strains containing the allele with six repeats in our collection, so we used a synthetic single strand DNA amplicon holding six triplets. For this, 1 μg of

the reverse single strand DNA (Eurofins MWG, Ebersberg, Germany) was copied in the presence of forward primer. The synthetic double strand DNA was then diluted to 10 ng prior to PCR. The number of triplet repeats for all strains was confirmed by sequencing with ABI Prism 3100 Avant check details Sequencer (Applied Biosystems), according to Amonsin et al. (2004). The sequences were analysed using the SeqMan Module within the Lasergene Package (DNA Star, Madison, WI). Representative results of HRM analysis are shown in Fig. 1, as derivative melting curves after normalization and exponential background removal. Two melting domains for each sample were observed: one relative to the amplicon homoduplex product (DNA double strand) and another one relative to the heteroduplex single strand DNA/probe. According to the LATE-PCR strategy, the homoduplex

products were Phenylethanolamine N-methyltransferase generated during the first cycles of amplification, whereas the single strand DNA was generated during the late cycles (data not shown). This single strand DNA can match with the probe and generate the heteroduplex single strand DNA/probe. As shown in Fig. 1, analysis of homoduplex amplicons did not allow any differentiation between the various alleles. However, it did reveal approximately three degrees among the adjacent alleles, allowing an unbiased identification. In silico analysis using the software mentioned above revealed that Tm decreased theoretically by 1 °C for every triplet depletion, a difference between two adjacent alleles (ΔTm) smaller than that found in our analysis. An explanation for this discrepancy could be the formation of loops when the probe (six repeats) matches the DNA single strand of the alleles carrying three, four and five repeats. These secondary structures could contribute to further destabilize the heteroduplex, causing a larger ΔTm among two adjacent alleles with respect to that theoretically evaluated by the software (see Supporting Information, Fig. S1).