Genomic deletions of MSH2 and MLH1 in colorectal cancer families detected by a novel mutation detection approach

1Hereditary non-polyposis colorectal cancer (HNPCC) is an autosomal dominant predisposition for early-onset colorectal cancer, endometrial cancer and other malignant tumours. Many HNPCC families fulfil the ‘Amsterdam criteria’ which require three colorectal cancer patients, vertical transmission and young age at diagnosis. In about 50% of families that meet these criteria, germline mutations in one of the DNA-mismatch-repair (MMR) genes MSH2 or MLH1 are detected. Virtually all colorectal tumours from MSH2 or MLH1 mutation carriers show microsatellite instability (MSI), which reflects the defect in DNA-mismatch-repair, and absence of expression of the MMR gene product involved. MSI is an important but not a specific marker of a germline MMR gene defect: the instability can also be due to acquired hypermethylation of MLH1. A minority of HNPCC families has a MSH6 defect or, exceptionally, a mutation in one of the other MMR genes. Tumours due to MSH6 mutations may or may not show MSI. Germline mutations in MMR genes can lead to a variety of clinical presentations, including sporadic early-onset colorectal cancer and familial endometrial cancer. Therefore, additional clinical criteria were developed which may indicate an MMR gene defect, in particular the Bethesda and Amsterdam II criteria. Identification of a germline defect in colorectal cancer patients is crucial to establish the etiology of the disease and to direct clinical decision making for patients and family members.

2At present, mutation detection of DNA-mismatch-repair genes is often performed by DGGE (Denaturing Gradient Gel Electrophoresis), DHPLC (Denaturing High Performance Liquid Chromatography), SSCA (Single Strand Conformation Analysis) and direct DNA sequencing. With these methods single base substitutions and small deletions and insertions can be detected. However, genomic deletions, i.e. deletions of one or more entire exons as well as duplications of exons will not be identified by these techniques. Detection of this class of mutations is important since they are an important cause of hereditary non-polyposis colorectal cancer.

3Conventionally, the main technique used for the detection of genomic deletions in MMR genes is Southern blotting. However, Southern blot analysis has several drawbacks: the method is laborious and time-consuming and requires a relatively large amount of high-quality DNA. Therefore, alternative methods were developed based on (semi-) quantitative PCR.

4Here we describe the value of a new technique, called Multiplex Ligation-dependent Probe Amplification (MLPA), for determination of the relative copy numbers of DNA sequences, and application of this method for the analysis of genomic deletions in MSH2 and MLH1.

5We studied a cohort of colorectal cancer families investigated at the Cancer Family Clinics of the VU University Medical Center and the Netherlands Cancer Institute, Amsterdam, the Netherlands. The aim of this study was to assess the value of the MLPA test for the detection of genomic MSH2- or MLH1-deletions.

6In 126 colorectal cancer families germline mutation analysis of MSH2, MLH1 and/or MSH6 was performed. MSH1 and MLH2 were analysed in 103 kindreds. MSH6 was analysed in most families negative for MSH2 and MLH1. In the other 23 families, all with MSI-stable or MSI-low tumours, only MSH6 was examined.

7The pedigree data either fulfilled the Amsterdam criteria I or II for HNPCC or less stringent criteria, usually one or several of the Bethesda criteria (‘suspected HNPCC’). Germline mutation screening of MMR genes was performed using DGGE or SSCA in combination with DHPLC for MLH1 and MSH2 and DGGE for MSH6. Investigation of MSI in tumours was performed according to standard procedures. Southern blotting of MSH2 and MLH1 was performed in one institution (VUMC) for families with MSI-high tumours and a previous negative screening for MMR gene defects. DNA samples from all 126 families were subjected to MLPA analysis. Informed consent was obtained from all patients who underwent DNA-based diagnosis.

8In MLPA, illustrated in Figure 1A, probes are used that consist of two oligonucleotides, both having a sequence complementary to a part of a target sequence at one site and a universal primer annealing sequence at the other site. Both hemi-probes can hybridize directly adjacent to each other to a target genomic sequence. If hybridized, the hemi-probes can be ligated and subsequently be amplified by PCR. Non-hybridized hemi-probes do not have to be removed since they can not be ligated and consequently will not be amplified. By using probes with amplification products of different lengths, multiple probes can be used in one reaction and be separated by gel electrophoresis for quantification.

9The MLPA test for MSH2 and MLH1 was obtained from MRC-Holland, Amsterdam, the Netherlands. The probe mix contains 16 exon probes for MSH2, 19 exon probes for MLH1, and seven control probes specific for DNA sequences outside the MSH2 and MLH1 genes. Details on probe sequences can be found on http://www.mrc-holland.com. All incubations were performed in a PCR machine with heated lid. Genomic DNA (50–500 ng) in 5 μl TE was denatured at 98°C for 5 min. Next, 3 μl probe mix was added and the mix was heated at 95°C for 1 min. and incubated at 60°C for 16 h (‘overnight’). Ligation was performed using heat-stable Ligase-65 enzyme at 54°C for 15 min, followed by ligase inactivation at 98°C for 2 min. Next, 10 μl ligation mix was added to 40 μl PCR buffer containing dNTPs, Taq polymerase and PCR primers (one unlabelled and one FAM-labelled primer). The reaction mixture was preheated at 95°C for 1 min. followed by 32 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s and extension at 72°C for 1 min. A final extension was performed at 72°C for 20 min. Fragment analysis was carried out on ABI model 310 or 3700 capillary sequencer (Applied Biosystems, Forster City, CA, USA) using TAMRA-500 or ROX-500 as size standards. A peak pattern of 42 peaks ranging in size from 130 to 472 nt is obtained.

10Data analysis was performed using Genescan and Genotyper software (Applied Biosystems). For analysing a small series of samples, visual inspection of the peak pattern of a patient's sample superimposed over a peak pattern of a control is very suitable (see Figure 1B). For analysing large series of samples, peak areas were imported into Excel spreadsheets and peak fractions were calculated by dividing the peak area of a certain probe by the sum of peak areas of all control probes in a certain sample. Subsequently, this relative peak area of each probe was divided by the average relative peak area of this probe in control samples. In normal individuals this calculation will result in a value of 1.0 representing two copies of the target sequence in the sample.

11The yield of germline mutation analysis according to the clinical subgroup and the MSI status of tumours is given in Table 1. We found germline mutations in 37 (29.4%) of the 126 colorectal cancer families. The characteristics of these families are summarised in Table 2. Thirty-eight germline mutations were found, including a double mutation in one kindred (family C222: a splice defect and an unclassified variant of MLH1), 20 in MSH2, 15 in MLH1 and three in MSH6. Seven (six different) DNA alterations were of unknown clinical significance. Thirteen genomic deletions were identified in MSH2 and four in MLH1, including a deletion of the entire MLH1 gene in two families (Figure 1B). These two kindreds were found to have a common ancestor in the eighteenth century. The clinical picture is summarised in Figure 2. All genomic deletions detected by Southern blotting were confirmed by MLPA.

12We found a high frequency of genomic deletions in MSH2 and MLH1 in a large cohort of colorectal cancer families. The new MLPA technique efficiently detected these deletions. In general, the yield of MMR gene mutation analysis in colorectal cancer families depends upon: (1) the selection of families based on clinical criteria, (2) the MSI status of tumours and/or immunohistochemical expression of MMR gene products, (3) the methods used for germline mutation analysis and (4) the interpretation of mutations as pathogenic vs non-pathogenic.

13The families we studied include a wide range of clinical conditions with varying indications of HNPCC. They were classified into two main groups: those that fulfil the Amsterdam criteria (I or II) and other ‘suspected’ families. Families from the latter group generally satisfied one or more of the Bethesda criteria. It should be noted that the description of a given family according to the Amsterdam or Bethesda criteria does not always give the most complete clinical information. For example, Muir–Torre syndrome skin lesions were observed in several families from our study group and strongly indicate HNPCC; however, they are not mentioned in the above-mentioned criteria. In addition, many families seen in the clinic have a strong but unconfirmed family history of colorectal cancer whereas the Amsterdam and Bethesda criteria require histologically confirmed diagnoses.

14In families that meet the stringent Amsterdam criteria, pathogenic MSH2 or MLH1 mutations have been detected in about 50% of kindreds. The same yield of about 50% is found for families with less clinical indications of HNPCC but which, in addition, have MSI-high tumours. Among families that only fulfil clinical criteria, usually not as stringent as the Amsterdam criteria, mutations are detected in about 30% of kindreds. Depending on the specific subgroup which is tested, the frequency of mutations may be lower than 10% or higher than 90%.

15Our mutation yield was 5.8% for families with MSI-S or MSI-L tumours and 65.7% for families with MSI-H tumours (Table 1). Our data support the notion that screening for MSI in tumours is an efficient prescreening tool for mutation analysis.

16Since we included a technique for the detection of genomic deletions which revealed a high frequency of this class of mutations, one might argue that a frequency of non-deletion mutations of 20 out of 126 kindreds (15.9%) is lower than one would expect. Probably there are two main reasons for the low percentage of non-deletion mutations. First, we performed mutation analysis in a relatively large group of families which did not meet the Amsterdam criteria (I or II) and, in addition, had MSI-stable or MSI-low tumours. In 64 of these families we found only four (6.3%) germline mutations. These mutations include three unclassified variants in MSH2 and MSH1 and one pathogenic MSH6 mutation. It has been observed before that deleterious MSH6 mutations may be accompanied by MSI-low tumours. Second, in several families all affected individuals had died and, consequently, only at-risk family members were available for testing.

17The methods we used for the analysis of point mutations were DGGE or SSCA in combination with DHPLC for MSH2 and MLH1 and DGGE for MSH6. These techniques are highly sensitive and specific for the detection of mutations in the coding regions of these genes. Mutations in the promoter regions of MSH2 and MLH1 will not be detected by these methods; however, they seem to play a limited role. We found six different mutations in MSH2 and MLH1 categorised as unclassified variants; their clinical significance is uncertain. However, the Thr117Met variant in MLH1 most probably has a pathogenic effect as demonstrated in a recently developed in vitro assay.

18In our cohort of colorectal cancer families more than 50% of the pathogenic mutations were genomic deletions. Evidently, genomic deletions in MLH1 and MSH2 are an important cause of HNPCC in Dutch colorectal cancer families. It has been reported previously that MSH2 genomic deletions are frequent in this group. Haplotype analysis of the kindreds sharing the same deletions failed to show evidence of a founder effect. Only two MLH1 deletions had been reported at that time, one of which was a founder mutation that accounted for a large proportion of HNPCC kindreds in the Finnish population. In a recent study of German HNPCC kindreds a predominance of MLH1 deletions was found. Recently, deletion mutations have also been demonstrated in series of European and US colorectal cancer kindreds, but at a much lower frequency than in our study group (Wijnen JTh, personal communication). We found 17 genomic deletions, 13 in MSH2 and four in MLH1. These included at least nine different subtypes, which might imply that the high frequency of deletions in our study group is not caused by a strong founder effect. However, five deletions occurred in multiple kindreds and a common ancestor was demonstrated for the MLH1 deletion. Future studies will clarify the frequency and background of genomic deletions in MSH2 and MLH1 in different populations.

19A deletion of the entire MLH1 gene has not been described before. In the two kindreds involved, the clinical expression included Muir–Torre syndrome and very-early-onset colonic cancer. A deletion of the complete MSH2 gene in an HNPCC family was recently described by.

20In summary, the MLPA technique is a fast and efficient test for the detection of genomic deletions in MMR genes. In contrast with the PCR- based techniques described by and deletion screening of MSH2 and MLH1 by MLPA can be performed in a single procedure. Compared with the MAPH method developed recently MLPA is easier to perform. The technique allows the detection of deletions that might escape detection using Southern blot analysis. In particular, interpretation of Southern blotting results may be difficult in case of the deletion of an entire gene.

21Clearly, screening for genomic deletions in MSH2 and MLH1 is an essential procedure for the molecular diagnosis of HNPCC. Due to the advantages of the MLPA technique, screening for these deletions might be considered as an early step in MMR gene mutation analysis.