High resolution melting analysis for a rapid identification of heterozygous and homozygous sequence changes in the MUTYH gene

Colorectal carcinoma (CRC) is the second most common cause of cancer-related mortality in developed countries. About 1% of CRCs arise in individuals affected with familial adenomatous polyposis. This condition can be caused by mutations in at least 2 distinct genes: APC, implicated in the autosomal dominant form, and MUTYH, which is involved in MUTYH-associated polyposis (MAP: MIM#608456). MAP is transmitted as an autosomal recessive trait due to biallelic mutations of the MUTYH gene, whose product is a DNA glycosylase that removes adenine from A•8-oxoG as part of the base excision repair process [1].

MAP patients are homozygotes or compound heterozygotes for mutations of the MUTYH gene, that is comprised of 16 exons. Two missense base substitutions, c.536A > G (p.Tyr179Cys) and c.1187G > A (p.Gly396Asp), located in exons 7 and 13, respectively, account for about 75% of pathogenic MUTYH allelic variants reported among Caucasians [2]. In addition, other recurrent MUTYH mutations have been identified in Pakistani, Dutch, Portuguese and Japanese patients [3, 4]. The remaining fraction of MUTYH variants identified in MAP patients is highly heterogeneous and can be located along the whole coding sequence.

The identification of MUTYH germline mutations has clinical relevance because the MAP phenotype is variable and often difficult to identify on clinical grounds alone [5–7]. The first step in MUTYH molecular screening in Caucasian patients is usually represented by the search for the two common mutations. Complete analysis of the MUTYH coding region is then carried out when none or only one mutant allele is identified with the first screening.

Several scanning techniques are currently performed to identify MUTYH sequence variants. These are usually based on direct sequencing, often preceded by denaturating high-performance liquid chromatography (dHPLC), single strand conformational polymorphism (SSCP) or Tetra-ARMS-PCR assay [8–10]. Despite their high degree of accuracy, these methods are relatively time-consuming and expensive.

High resolution melting analysis (HRMA) is a mutation detection method based on the principle that the melting curves of DNA fragments vary depending on base composition, and is a potentially useful technique for fast genotyping and high-throughput mutation scanning in genetic diagnosis [11, 12]. Heterozygous samples are identified by differences in melting curve shape. When a nucleotide change is present in a target sequence, the resulting melting curve of the PCR product is a composite of both heteroduplex and homoduplex components that is visualized by the different melting curve shapes of the two species present. Homozygous base changes are not associated with heteroduplex formation. Since DNA sequences differing in a single base often have different melting temperatures (T_m), discrimination between homozygous wild-type and homozygous mutant samples can be achieved by comparing the T_m values of test samples with those of wild-type controls.

Unlike other scanning methods, mutation analysis by HRMA provides a closed-tube system that reduces the risk of contamination, decreases analytical time and requires no sample processing or separation after PCR. HRMA has been used for the detection of mutations associated with human genetic diseases, especially autosomal dominant or X-linked disorders [13–15]. More recently, this method was used to identify sequence variations in autosomal recessive disorders, such as primary carnitine deficiency, cystic fibrosis, and Zellweger syndrome [16–19]. This technique has also been used to identify specific common variants in selected MUTYH exons [20, 21]. However, there is a lack of information on the sensitivity of HRMA for the identification of a broader spectrum of MUTYH mutations.

In this study, we report on an HRMA-based approach for the rapid detection of MUTYH variations spread along the entire coding region, as well as in the 5' and 3' untranslated regions, in both the heterozygous and homozygous state.

As a first step, we assessed the specificity and sensitivity of HRMA in a set of 26 samples whose genotypes had been preliminarily determined by SSCP and direct sequencing. All samples had been investigated by both SSCP and sequencing in order to compare the results obtained with the two methods. These allowed the detection of a total of 13 MUTYH point sequence variants in the 26 samples (Table 1). All variants were identified by SSCP, with the exception of the c.64G > A and c.1014G > C substitutions; these were detected in the heterozygous and homozygous state, respectively, only by sequencing. Six variants were present in multiple samples.

In order to verify whether HRMA allowed detection of homozygous sequence changes based on T_m differences alone, the 26 samples were examined without prior mixing with the reference sample. In all samples harbouring MUTYH variants previously identified by SSCP, the shapes of the melting curves were altered compared to wild-type samples. All samples homozygous for MUTYH sequence changes recognized by SSCP analysis (c.64G > A, c.536A > G, c.1187G > A, c.1227_12289dupGG and c.1437_1439delGGA) could be reproducibly distinguished from the wild-type counterparts by a temperature shift in the melting curve. In addition, HRMA under these conditions allowed to identify one heterozygote for the c.64A allele, that had gone undetected by SSCP; this showed a melting curve clearly different from those of wild-type c.64G homozygotes (data not shown).

We then re-examined all samples after mixing with a reference genomic DNA. All genotypes reported in Table 1, including the homozygote for the c.1014G > C (p.Gln388His) variant (Figure 1), were detected by HRMA under these conditions. Melting curves were reproducible in duplicates of each sample and among different samples carrying the same mutations.

The two most frequent Caucasian mutations (c.536A > G and c.1187G > A) were clearly detected both in the heterozygous and homozygous state and were associated with significant curve shape changes (Figures 2, 3). In addition to the c.1187G > A mutation, other sequence variations in exon 13 were recognized (Figure 2). These included one homozygous insertion (c.1227_1228dupGG), one heterozygous intronic C > T transition (c.1187-27C > T), and one heterozygous C > A transversion (c.1258C > A). The associated curves could be clearly distinguished from those of wild-type and c.1187G > A samples. The c.1187-27C > T and c.1258C > A substitutions were present in the same sample.

Abnormal melting profiles were clearly identified also for the remaining 5 variants. Moreover, reproducible melting profiles were obtained for samples carrying the polymorphic variants (c.64G > A, c.504+35G > A, c.1014G > C) located in exons 2, 6 and 12.

The mutation detection sensitivity was 100% compared to direct sequencing in samples subjected to mixing with a reference DNA. Specificity was 100%, since all abnormal profiles were associated with sequence changes. Sensitivity of HRMA performed without mixing with a reference DNA was 94% (17/18 MUTYH variant genotypes recognized).

Four different variants were identified in exon 2: the intronic polymorphism c.157+30A > G, the polymorphic missense substitution (c.64G > A; p.Val22Met), also present in the first sample series investigated, and two additional heterozygous base substitutions leading to aminoacid changes, c.37G > A (p.Ala13Thr) and c.56G > A (p.Arg19Gln). The melting profiles associated with the latter two rare variants could clearly be distinguished from those of samples homozygous and heterozygous for p.Val22Met (Figure 4).

All samples were also subjected to sequencing, and no further alterations were observed. Again, the sensitivity and specificity of mutation detection by HRMA analysis in this blinded set of samples was 100% in experiments performed on mixtures of test and reference DNAs, while it was lower (88%; 15/17 variant genotypes detected) in unmixed samples.

HRMA also allowed to identify three polymorphic variants in the heterozygous state, located in intron 8 (c.690+21C > A), intron 14 (c.1477-40G > C) and exon 16 (c.1544C > T; p.Ser515Phe), respectively. These were observed both with and without mixing the samples with the reference DNA (data not shown).

To the best of our knowledge, four of the sequence variants identified in this study (c.-205C > A, c.37G > A, c.544C > G, and c.1023_1024insGA) have not previously been reported in the literature nor are they included in the MUTYH Leiden Open Variation Database (LOVD; http://chromium.liacs.nl/LOVD2/colon_cancer/home.php) database.

We have shown that MUTYH mutations can be identified with high sensitivity by HRMA both in the heterozygous and in the homozygous state. HRMA sensitivity was higher compared to SSCP and equal to direct sequencing, the gold standard for assessing sequence variations. We have analyzed 26 samples containing 13 DNA sequence variations located in 6 exons, using both HRMA and SSCP. Sensitivity of HRMA analysis appeared superior to that of SSCP, as previously reported [27]. In particular, HRMA allowed the detection of heterozygous and homozygous sequence changes (c.64G > A and c.1014G > C) that had not been identified by SSCP.

HRMA has been introduced very recently in molecular genetic diagnostic laboratories and has shown several advantages: cost-effectiveness, simplicity, absence of post-PCR treatment, and, particularly, the swiftness to identify genetic variants or mutations. So far, this technique has been mainly used for the detection of mutations in genes involved in X-linked and autosomal dominant diseases, such as OTC, BRCA1, BRCA2 and TP53 [15, 28, 29]. Applications to autosomal recessive disorders have been more limited, as homozygous mutations, especially G > C and A > T substitutions and small insertions/deletions (e.g., p.Phe508del in CFTR) may not be associated with melting curve changes and consequently may escape detection [30]. Nevertheless, the technique has also been successfully applied to investigate genes associated with autosomal recessive conditions, such as CFTR, PEX6, ABCA4, ATP7B, C2ORF71 [18, 19, 31–33]. It has also been shown that the sensitivity limits of HRMA for the detection of homozygous sequence changes can be overcome by mixing the test DNA sample with a reference wild-type DNA [16–18, 28, 29, 34].

López-Villar et al. [20] have recently used HRMA to detect heterozygous and homozygous variants in MUTYH exons 7, 12 and 13 in a series of 82 patients with a phenotype suggestive of MAP. The experiments were set up without mixing samples with a reference DNA. Hence, homozygotes for common variants in these exons, such as c.1014G > C, as well as for rare nucleotide changes, could have gone undetected, since heteroduplexes could not be formed. In addition, the melting profiles of samples heterozygous for the common c.1187G > A and for the rare c.1276C > T (p.Arg426Cys) mutation appeared to be overlapping.

Partial MUTYH mutation screening by HRMA limited to exons 7, 13 and 15, has been recently performed in a series of cutaneous melanomas [21]; however, also in this latter study, the samples had not been premixed with reference DNA. Therefore, so far HRMA-based approaches for the detection of MUTYH mutations have been limited to few targeted regions of the gene. In addition, the sensitivity of HRMA compared to that of other mutation screening methods was not assessed in either study.

Although the majority of MUTYH mutations present in Caucasian MAP patients are located in exons 7 and 13, several additional pathogenic variants are scattered over all other exons of the gene in the Caucasian population [35]. In addition, the distribution of MUTYH mutations differs among populations [36]. Therefore, in order to allow maximum sensitivity, mutation scanning needs to be performed on all MUTYH exons.

A similar situation applies to the CFTR gene, where population-specific mutations are observed in different populations. For this gene, it has been demonstrated that identification of all point mutations, both at the heterozygous and homozygous state, can be achieved by HRMA when the whole coding sequence is investigated [18]. In addition, it has been observed that homozygotes for the common Caucasian p.Phe508del mutation can only be identified when the formation of heteroduplexes is induced by mixing with wild-type DNA [18].

We therefore developed an improved HRMA-based MUTYH analysis protocol that scans the entire coding region and the UTRs and increases sensitivity by mixing the DNA test with a reference DNA. Our results confirm that the common Caucasian mutations in the MUTYH gene can be reliably identified by HRMA even when present in the homozygous state. We also found that the T_ms of samples homozygous for either of the two common MUTYH mutations are consistently different from those of their wild-type counterparts; this allows simple and rapid detection of homozygotes for these frequent variations in native test DNA samples, without need to add reference wild-type DNA.

In addition, complete HRMA scanning of MUTYH clearly differentiated all samples containing DNA sequence variations, including small insertions and deletions, from wild-type DNA. Importantly, samples heterozygous for different variants located within the same amplicon displayed distinct melting profiles. However, it has been pointed out that occasionally the melting curves of samples containing two linked variants located in the same amplicon may be undistinguishable from those observed in simple heterozygotes [37]. Hence, we advocate that the mixing strategy should be used also for the purpose of identifying all possible heterozygote combinations.

The availability of a rapid and inexpensive method for the detection of MUTYH mutations is important for the identification of individuals at increased CRC risk. Since the phenotype associated with MUTYH mutations is highly variable, the diagnosis of MAP is difficult on clinical grounds alone [5–7, 38, 39]. In addition, several studies have shown that monoallelic MUTYH mutation carriers may have a moderately increased risk of CRC compared to the general population [40–43]. Since the frequency of simple heterozygotes for MUTYH mutations may be as high as 0.02 in Caucasian populations [44], MUTYH testing could become a widespread screening test to identify individuals at risk for CRC in the general population.

We have set up a very fast, simple and non-expensive mutation scanning method that identifies the most common MUTYH mutations and known polymorphic variants, as well as rare DNA sequence changes spread in the coding sequence and 5' UTR. In Caucasian patients, molecular screening by HRMA can be performed in two steps: the first one to identify common variants in exons 7 and 13 without mixing with a reference DNA; the second one should be performed using a mixture of test and control DNA, to detect more rare changes in the whole coding sequence when no or a single mutation has been identified in the first stage. The implementation of a simple and sensitive method for MUTYH mutation scanning is relevant for the diagnosis of CRC susceptibility and for the identification of at risk individuals, who should be advised intensive surveillance.