Our work supports the high sensitivity of HRM for the detection and quantification of DAE. We have shown that HRM is able to detect DAE associated with NMD in LCLs carrying a non-sense mutation in CHEK2. Although no DAE was observed in the patients who do not carry the 1100delC mutation, the series investigated here was limited, and we cannot rule out that cis-regulatory variants in CHEK2 may lead to DAE in a tissue specific manner . However, this later hypothesis could not be tested since no breast tissue was available from these patients.
The approach used in our study relies on subjects who are heterozygous for a coding SNP and allows relative quantification of allelic transcripts. This methodology has major advantages over more conventional methods for investigating DAE based on the comparison of gene expression between individuals as discussed elsewhere [8, 10, 20]. Since they come from the same tissue sample and have therefore been subjected to the same environmental influences (such as genetic trans-acting factors and experimental exposures, including mRNA degradation) both alleles should be equally expressed in the absence of cis-acting sequence variation or epigenetic effects affecting the expression of the target mRNA. Thus, the strength of this approach is that each allele acts as an internal control for confounding factors, disclosing cis-variation effects without being confounded by any trans-variation effects.
Here, we report a complete solution for HRM analysis that can be used on both the HR-1™ (1 single capillary) and LightScanner® (384-well plate format) instruments, with the format depending on the required throughput. Access to DAE assessment technology can be cost prohibitive for many laboratories. HRM provides a good alternative when compared to methodologies based for instance, on the use of capillary electrophoresis for single-base extension assays, such as SnapShot assays , allele-specific quantitative real-time PCR  and microarray platform . Advantages offered by HRM analysis include its rapidity, cost-effectiveness and security due to its closed-tube nature. Though the HR-1™ is reported to provide a better accuracy , both instruments performed well to identify the 4 carriers of the CHEK2*1100delC variant showing DAE in the absence of puromycin treatment in our study. However, given the number of samples to test, analysis with the HR-1™ instrument ended up being much more time consuming (Table 1). The results obtained with the LightScanner® instrument showed that this methodology can be applied in larger-scale studies, provided that LCL material is available, while maintaining high accuracy and remaining cost-effective. Indeed, the protocol is relatively inexpensive since it only requires standard PCR reagents and a small amount of fluorescent probe.
The script we developed using R computing software was made compatible with both instruments and greatly reduces the time of analysis. Once HRM data are acquired, the normalization of the curves, peak heights measurements, ratios calculations and statistical analysis are performed automatically within less than 15 minutes for a set of 96 samples when using the LightScanner® instrument. The output consists in a summary table of the peak heights, relative allelic ratios, and the Student's t-test values, as well as a plot on which DAE carriers are highlighted. The script can also display other information on demand, such as melting curve profiles which can be displayed for each replicate or by average of 4 replicates for each individual (see examples in Figure 3 and 4).
In DAE analysis by HRM, the peak heights obtained from the melting curve reflect the relative abundance of each allele's transcript. The reproducibility and precision of the assay are reasonable as seen in the small standard deviations associated with the calculations. The accuracy of the method was illustrated by the consistency of the allelic expression estimates across multiple replicates assay within the same individual sample. Genomic DNA ratios varied within a very narrow range, showing the excellent reproducibility and precision of the assay on DNA derived from LCL. The intra-sample variation in replicate analysis was higher for mRNA ratios than for DNA ratios, possibly owing to RNA stability. In addition, at low copy numbers of mRNA, the stochastic distribution of the RNA templates may be a major source of variation and hence affect the accuracy of DAE analysis, by generating disagreeing replicate results for instance .
In a DAE study, the main optimization issue is the ability to select a subset of 2-3 marker SNPs so that as many individuals as possible are heterozygous for at least one of the markers. Subsets of individuals giving the most heterozygotes at 2 loci should be chosen in order to maximize redundancy, and to self-check for error reduction. Unfortunately, in the present study, no individual heterozygous for both SNPs showed evidence of DAE. Detection of DAE in a candidate gene may be indicative of the presence of a coding or regulatory variant altering expression of the gene product. However, DAE-based approaches can point out the presence of a regulatory causative variant only if the subjects are heterozygous for the causative variant (and of course for the coding SNP serving as marker). In some situations, the coding SNP used to distinguish both alleles may be itself responsible for the observed DAE, or it can be in linkage disequilibrium (LD) with it, i.e. on the same haplotype. In the case of no LD between the marker and the functional variant, it is still possible to map the variant, as previously reported by others [30, 31].