MATERIALS AND METHODS

HNPCC kindreds and a novel MLH1 mutation

A novel mutation, MLH1 del71, was found in two unrelated HNPCC families (A and B). The pedigrees and the clinical characteristics of the patients are shown in fig 1. In kindred A, three sibs and their uncle had CRC and one distant relative had an endometrial cancer (EC). All the patients carried the MLH1 del71 mutation, all the tumours displayed high MSI phenotypes, and the mean age at onset was low. In family B, there were seven CRCs in two successive generations, young age at onset, and the index tumour from the mutation carrier showed a high MSI phenotype. All human investigations were performed after approval of the local ethical committee or in accordance with DNA diagnostics (patient care).

Analyses of MSI and mutations

In family A, a panel of four microsatellite markers, one mononucleotide repeat (BAT 26) and three dinucleotide repeats (D1S158, D2S123, D5S346), was studied. In family B, a panel of five examined markers included two mononucleotide (BAT25, BAT26) and three dinucleotide repeats (D2S123, D5S346, D17S250).¹⁰ Tumours were classified as MSI-high when two or more markers showed instability.

The genomic DNA of the index patients of families A and B were analysed for a mutation in the MLH1 gene using a DGGE or a DHPLC based method (WAVE apparatus, Transgenomic Inc), respectively.^11–14 The mutation was confirmed and screened in other family members by direct sequencing.

Site directed mutagenesis and production of recombinant baculoviruses

The human MLH1 cDNA was cloned into the plasmid pFastBac1 (Gibco BRL). The mutagenesis of wild type (WT) cDNA was carried out using a PCR based protocol as described previously.⁸ In the first PCR, fragment A was created using the primers 5′- CGG ATT ATT CAT ACC GTC CC -3′ (FA) and 5′- CAA TAT CCA GAT CTT TCC TGA TCC -3′ (RA), and fragment B was created using the primers 5′- GGA TCA GGA AAG ATC TGG ATA TTG -3′ (FB) and 5′- TTC TCC TCG TGG CTA TGT TG -3′ (RB). In the second PCR, fragments A and B were mixed in a 1:1 ratio and primers FA and RB were used to complete the PCR products. The reactions were carried out for 30 cycles, at 94°C for one minute, at 60°C for one minute, and at 72°C for 2.5 minutes. The product of the second PCR was cloned into the MLH1/pFastBac1 cDNA vector between the BamHI and PvuII sites. The insert was sequenced in order to ensure that the del71 mutation was present and that no other mutations were introduced into the PCR fragment during amplification (ABI Prism 3100 Genetic Analyzer, Applied Biosystems/Hitachi) (data not shown). The recombinant baculoviruses were generated using the Bac to Bac system according to the manufacturer’s instructions (Gibco BRL).

Production of recombinant proteins

For protein production, 2 × 10⁷Spodoptera frugiperda 9 (Sf9) cells were coinfected with wild type or mutated MLH1 together with wild type PMS2 recombinant baculoviruses. The coinfection was performed because the PMS2 protein was reported to be unstable in the absence of its partner MLH1.¹⁵ After 48 hours of culture, the total protein extracts including overexpressed MLH1 and PMS2 proteins were prepared as described previously,¹⁶ except that 100 mmol/l salt was used. The MLH1 expression levels were examined by SDS-PAGE and western blot analysis.

Combined co-immunoprecipitation and western blot analysis

For co-immunoprecipitation, the total protein extracts were estimated to contain equal quantities of recombinant MLH1. A total of 85 μg of protein extract including the mutated protein MLH1del71 and WT PMS2 (MutLα-del71) and 8.0 μg of extract including WT MLH1 and WT PMS2 (MutLα-WT) were incubated for one hour at 4°C on a rotating wheel with 0.5 μg of anti-MLH1 antibody (PharMingen, 13291A) in a total volume of 1 ml in RIPA lysis buffer (150 mmol/l NaCl, 1% NP-40, 0.5% DOC, 0.1% SDS, 50 mmol/l Tris pH 8.0). A total of 20 ml of protein A/G agarose suspension (Santa Cruz) were added and incubation was continued for a further 1.5 hours. The precipitates were centrifuged for five minutes at 2500 g, washed three times with cold RIPA buffer, run on a 7.5% SDS polyacrylamide gel, and transferred onto PVDF nitrocellulose membrane. The interactions between MLH1 and PMS2 were detected with western blot analysis with both anti-MLH1 (PharMingen, 13271A) and anti-PMS2 (Calbiochem, NA30-100UG) antibodies.

In vitro mismatch repair assay

In vitro MMR assay was carried out according to a previous report.⁸ A total of 75 μg of nuclear extract (NE) of HCT116 colorectal cancer cell line, which lacks functional MLH1 and is deficient in MMR, was complemented with the total protein extract including overexpressed MLH1 and PMS2 proteins. The circular DNA heteroduplexes, which contain a G•T mispair 369 bp downstream from a single strand nick, were used as substrates.^17–19 For the assay, the total protein extracts were estimated to contain equal quantities of recombinant MLH1 (MutLα-WT, 2.3 μg; MutLα-del71, 25.0 μg). NE of the MMR proficient cell line TK6 (75 μg), purified MutLα (0.3 μg), and MutLα-WT were used as positive controls in the assay. MMR deficient HCT116 NE was a negative control. The presence of a G•T mispair makes the heteroduplexes not susceptible to cleavage by the endonuclease BglII. The repair reaction converts the G•T heteroduplex to the A•T homoduplex, whereafter BglII can cleave the molecule and the repair efficiency can be measured by the cleavage efficiency. Since the heteroduplex molecules are not all repairable they are always added in excess in the assay, and the repair percentage of the proficient wild type controls is used as a reference level.

RESULTS

All five tumours available for MSI analysis from family A, as well as the index tumour from family B, showed high instability. Mutation analyses and sequencing showed an in frame mutation, MLH1 del71, which deletes bases GAA (nucleotides 211-213) in exon 3 corresponding to a glutamic acid in the polypeptide. This mutation had not been reported before in HNPCC families (http://www.nfdht.nl).

The co-immunoprecipitation assay was performed in order to study the effect of mutation MLH1 del71 on heterodimerisation of MLH1 and PMS2 proteins. As shown in fig 2, the mutation might have reduced but did not abolish MutLα assembly. However, the complex MLH1 del71/PMS2 (MutLα-del71) failed to complement MMR deficient HCT116 nuclear extract (fig 3). While proficient extracts TK6, HCT116 + purified MutLα, and HCT116 + MutLα-WT repaired 18%, 17%, and 37 % of G•T heteroduplexes, respectively, HCT116 + MutLα-del71 repaired 0% of them. The result of the MMR assay thus confirmed the pathogenicity of the mutation.

DISCUSSION

The present study was undertaken to test the pathogenicity of a novel mutation MLH1 del71, which was found in two unrelated HNPCC families. Both families displayed typical clinical features, such as early age at onset, predominance of colon carcinomas, and high MSI in tumours. However, the alteration was a 3 bp deletion maintaining the coding sequence in frame and deleting only one amino acid from the protein. Moreover, the deleted codon was not evolutionarily conserved and the mutation has not been reported before in HNPCC families. Thus, interpretation was difficult and reliable assessment of pathogenicity required a functional test. In vitro MMR assay, the method used here, provides an opportunity to find out if a mutation prevents the protein from repairing mismatches. If tumours from the mutation carriers are MSI positive, the method can confirm that MSI results from the mutation in question. If tumours are MSI positive but the mutation found functions in the in vitro MMR assay it suggests that either there is another germline mutation or an epigenetic inactivation of a MMR gene, which causes MSI. If tumours are MSI negative, the MMR mechanism is functioning and not associated with disease. The in vitro MMR assay was recently used to test six MLH1 mutations⁸ and four MSH6 mutations.²⁰ The results suggested that the mutations linked to typical HNPCC imparted a MMR deficient phenotype on the altered polypeptide, whereas the mutations associated with atypical or putative HNPCC did not. However, the results did not exclude the possibility that the latter mutations affect biochemical events preceding the MMR function in vivo.

Here, we showed that MLH1 del71 might have reduced but did not abolish MutLα assembly. However, the protein complex MLH1 del71/PMS2 (MutLα-del71) was totally deficient in MMR reaction. Accordingly, the analysis of the interaction between MLH1 and PMS2 can be indicative, but the result can be difficult to interpret when used as a functional test. This is important since over 60% of MLH1 mutations are not located in the known MLH1/PMS2 interaction region (http://www.nfdht.nl),^21,22 and thus presumably do not or only slightly hamper their heterodimerisation. The results of the in vitro MMR assay confirmed the pathogenicity of MLH1 del71. Codon 71 is located in the region, which has been reported to function in ATP binding and hydrolysis^23,24 and, accordingly, can be expected to be sensitive to alterations. Our result showed that a mutation can be pathogenic irrespective of the conservational status of the altered sequence. Finally, the confirmed pathogenicity of MLH1 del71 is compatible with the typical HNPCC phenotype linked to the mutation, emphasising the importance of comparing clinical and experimental data.