Uncertain pathogenicity of MSH2 variants N127S and G322D challenges their classification
Version of Record online: 9 MAY 2008
Copyright © 2008 Wiley-Liss, Inc.
International Journal of Cancer
Volume 123, Issue 3, pages 720–724, 1 August 2008
How to Cite
Ollila, S., Bebek, D. D., Greenblatt, M. and Nyström, M. (2008), Uncertain pathogenicity of MSH2 variants N127S and G322D challenges their classification. Int. J. Cancer, 123: 720–724. doi: 10.1002/ijc.23573
- Issue online: 28 MAY 2008
- Version of Record online: 9 MAY 2008
- Manuscript Accepted: 13 FEB 2008
- Manuscript Received: 19 OCT 2007
- Academy of Finland. Grant Number: 110300
- Sigrid Juselius Foundation
- Finnish Cancer Foundation
- National Cancer Institute. Grant Number: R01 CA96536
- hereditary non-polyposis colorectal cancer;
- mismatch repair;
- functional analysis;
- unclassified variant
Hereditary non-polyposis colorectal cancer (HNPCC) is associated with germline mutations in mismatch repair (MMR) genes. Inherited missense mutations, however, complicate the diagnostics because they do not always cause unambiguous predisposition to cancer. This leads to variable and contradictory interpretations of their pathogenicity. Here, we establish evidence for the functionality of the 2 frequently reported variations, MSH2 N127S and G322D, which have been described both as pathogenic and non-pathogenic in literature and databases. We report the results of 3 different functional analyses characterizing the biochemical properties of these protein variants in vitro. We applied an immunoprecipitation assay to assess the MSH2–MSH6 interaction, a bandshift assay to study mismatch recognition and binding, and a MMR assay for repair efficiency. None of the experiments provided evidence on reduced functionality of these proteins as compared to wild-type MSH2. Our data demonstrate that MSH2 N127S and G322D per se are not sufficient to trigger MMR deficiency. This together with variable clinical phenotypes in the mutation carriers suggest no or only low cancer risk in vivo. © 2008 Wiley-Liss, Inc.
The existence of several non-truncating mutations in mismatch repair (MMR) genes, mostly in MLH1 and MSH2, is a major challenge for molecular diagnosis of hereditary non-polyposis colorectal cancer (HNPCC). They constitute already 20–30% of all MMR gene mutations.1(www.insight-group.org) Many of them can be called unclassified variants (UV) because their connection to HNPCC pathogenesis remains controversial. The 2 of the most frequently reported UVs are MSH2 G322D and MSH2 N127S.
In 2 separate studies, MSH2 N127S (c.380 A>G) was found in colorectal cancer (CRC) patients and was not found in healthy control individuals (n = 73 and n = 40).2, 3 As a result, N127S was considered pathogenic. In some other studies, MSH2 N127S was proposed to be likely polymorphic, either based on late-onset atypical cancer and no family history,4 or occurrence with another MMR gene mutation in a patient.5MSH2 N127S is also found with 0–9.2 % frequency in different study populations and documented as a single nucleotide polymorphism (SNP) in databases (www.ensembl.org, www.genome.utah.edu/genesnps/) (rs17217772).
MSH2 G322D (c.965 G>A) was first described as a non-pathogenic variant because it was found with 2% frequency in control allele population.6 To date, the variant has been reported several times. It was suggested to be polymorphic when it was not segregating with the cancer phenotype in a family or it was found in healthy control individuals.3, 5, 7–12 Contrary to these, it was interpreted as a low penetrance allele based on moderate inactivation of yeast MMR caused by a homologous yeast variant.13, 14 It was also interpreted as pathogenic when it was segregating with the cancer phenotype, and showed loss of MSH2 protein and high microsatellite instability (MSI) in the tumor,15 or when it was not found in healthy control chromosomes.16 Also MSH2 G322D is reported in SNP databases (www.ensembl.org, www.genome.utah.edu/genesnps/) (rs4987188), with population frequencies ranging from 0 to 6.5%.
In accordance with contradictory reports and interpretations, MSH2 N127S and G322D are classified in mutation databases both as pathogenic and polymorphic variations (www.insight-group.org, www.genome.utah.edu/genesnps). This and the absence of functional data on human proteins led us to study biochemical competence of MSH2 N127S and G322D to provide support either to their predisposing or non-predisposing roles in HNPCC tumorigenesis.
Material and methods
MSH2 variations N127S and G322D
The variations MSH2 N127S and G322D were chosen to our study based on the frequent reports in literature and contradictory classifications in genetic databases. The published clinical and population data of MSH2 N127S and G322D are collected in Table I.
|MSH2 variant||Tumor/age1||Other cancers in family||MSI2||IHC3/MSH2||Variant found in healthy controls||Another MMR gene mutation found in a mutation carrier||References|
|N127S||CRC<50, EC<50||+||+||ND||−||MSH2 A328P||5|
|N127S||CRC 40||+||+||ND||−||MLH1 frameshift c 1877||5|
|N127S||CRC<60||−||+||ND||−||MSH2 N108N, MLH1 IVS15-5T>C, APCI1307K||2|
|N127S||PC 71, BI C 78||−||ND||ND||ND||−||4|
|G322D||CRC and EC||ND||+||ND||ND||−||12|
|G322D||EC49||+||+||+/−||+||MLH1 D203N5, MSH2 exon 4 frameshift5||24|
Site-directed mutagenesis and expression of recombinant proteins
The cDNAs of wild-type (WT) MSH2 and N-terminally (His)6-tagged wild-type MSH6 were cloned in pFastBac1 vector (Invitrogen, Carlsbad, CA) between BamHI and XhoI or BamHI and NotI restriction sites, respectively. Site-directed mutagenesis for MSH2-N127S and MSH2-G322D constructs was applied by a PCR-based method according to instructions of the manufacturer (QuikChange Site-Directed Mutagenesis, Stratagene, La Jolla, CA). The primer sequences and PCR parameters are available upon request. The entire MSH2 cDNA sequences were verified by sequencing (ABIPrism 3100 Genetic Analyzer, Applied Biosystems, Foster City, CA) and transferred to baculovirus vectors as recommended by manufacturer of the Bac-to-Bac baculovirus expression system (Invitrogen). The recombinant proteins were produced in Spodoptera frugiperda 9 (Sf9) insect cells according to manufacturer's instructions, and the total protein content was extracted as described.25 All MSH2 proteins were produced together with MSH6 to give rise to the heterodimeric MutSα (MSH2-MSH6) protein complex, whose both partners are required for active MMR.26, 27
Western blot and coimmunoprecipitation
The expression of recombinant MutSα complexes was verified by Western blotting. 3 μg of total protein extracts from Sf9 cells were electrophoresed on 6% SDS-PAGE gels, blotted to nitrocellulose membranes (Amersham Biosciences) and detected with anti-MSH2 (MSH2 Ab-2, NA27, Calbiochem, dilution 1:250) and anti-MSH6 (MSH6/GTBP, Clone 44, BD Transduction Laboratories, dilution 1:1,000) antibodies. The interaction between MSH2 and MSH6 was verified by immunoprecipitation by rotating 30 μg of Sf9 extracts with MSH6-antibody-coated agarose A/G beads (Santa Cruz Biotechnology, Santa Cruz, CA) as previously described.28, 29 The interaction, demonstrated by the presence of both proteins in the immunoprecipitate, was detected with Western blot as detailed above.
Purification of MutSα
MutSα has previously been successfully purified using a polyhistidine tag in the N-terminus of MSH6.30 For purification of recombinant MutSα complexes, 100 μl of Ni-NTA agarose matrix (Qiagen, Hilden, Germany) was used for every 1 ml of total protein extract. The matrix was equilibrated with PBS and mixed with total protein extracts. The mixtures were rotated in +4°C for 2 hr and loaded into 1.5 ml polypropylene columns (Qiagen). The MutSα-bound matrix was washed with wash buffer (25 mM HEPES, 300 mM NaCl, 20 mM imidazole, 1 μl/ml leupeptin, 1× complete EDTA free, 0.5 mM PMSF) and MutSα was eluted with increasing imidazole concentrations. The MutSα-containing fractions were pooled and dialyzed for 4 hr against 25 mM HEPES, 110 mM NaCl, 3 mM DTT, 1 mM EDTA, 10% sucrose, 1 μg/ml leupeptine and 0.5 mM PMSF. The proteins were aliquotted, snap-frozen in liquid nitrogen and stored at −80°C. The purities of different MutSα preparations were compared to SDS-PAGE and Coomassie staining, and the concentrations were assessed with Bradford assay, using bovine serum albumin (BSA) as a standard.
Bandshift assays were used to assess the ability of MutSα variants to bind homo- or heteroduplexed DNA. 350 ng (total 65 nM) of each MutSα variant was incubated with 32P-labeled 36 bp double-stranded DNA oligomers, containing either a central G·T mismatch (GT heteroduplex) or an A·T base pair (AT homoduplex), in conditions previously described.25 The binding was visualized by running the reactions in 5% nondenaturing acrylamide gels, which were dried and exposed to phosphoscreen (Fujifilm BAS-1500). The data were analyzed with TINA software, version 2.09 (OY Tamro AB).
In vitro MMR assay
MMR assays were conducted as previously described.28, 29 Briefly, 75 μg of MSH2 deficient LoVo CRC cell line nuclear extracts (NE) were incubated with 1 μg of purified recombinant MutSα proteins in the presence of 100 ng of substrate plasmid containing an unpaired thymidine loop and a single-stranded nick in the opposite DNA strand, 370 base pairs 5′ from the mismatch site. In case of successful repair, i.e. addition of an adenine opposite the unpaired thymidine in the heteroduplex plasmid by functional MMR, a BglII restriction site was created, enabling the visualization of the repair by restriction analysis (For details, see Ref. 29). MutSα WT was used as a positive control and MutSα E749K variant, which was previously shown to be deficient in the in vitro MMR assay29 was used as a negative control. The repair percentages were analyzed with Image-Pro 4.0 (Media Cybernetics) and calculated as an average of 3 independent experiments.
Comparative sequence alignment
Expression and purification of MutSα variants
The production of MutSα (MSH2-MSH6 heterodimer) variants was successful and the expression levels were comparable to WT protein. MutSα was eluted from Ni-NTA matrix with 100–150 mM imidazole concentrations. The purity and yield of all recombinant MutSα complexes was similar, as visualized in Coomassie staining (Fig. 1a).
Interaction of MSH2 variants with MSH6
Combined immunoprecipitation and Western blot assay were used to detect the ability of mutant MSH2 proteins to interact with their functional partner MSH6. As demonstrated in Figure 1b (left panel, WB), the total protein extracts of MutSα WT, MutSα N127S and MutSα G322D expressing Sf9 cells contained equivalent amounts of MSH2 and MSH6 proteins. Furthermore, the anti-MSH6 antibody used in the assay immunoprecipitated comparable amounts of MSH2 protein, indicating the intact MSH2-MSH6 interaction capacity of WT and mutated proteins (Fig. 1b, right panel, IP). Extracts derived from cells expressing the MSH2-MSH3 (MutSβ) complex gave no detectable signals in the immunoprecipitation assay, confirming that only MSH6 antibody-bound protein complexes were present in the immunoprecipitate (data not shown).
Mismatch binding activities of MutSα variants
In bandshift assays, the MutSα WT as well as both of the variants bound homoduplex AT-oligomers with only low levels, as expected. GT heteroduplex oligomers were, however, bound in similar quantities, causing a clear bandshift, as demonstrated by slower migration of the protein-bound oligomers (Fig. 2).
In vitro MMR assay was used to assess the repair activity of WT and mutated MutSα. WT and both mutated proteins were able to correct the unpaired T mismatch with repair percentages of WT 30% (STD ±8%), N127S 32% (STD ±6%), and G322D 37% (STD ±7%) (Fig. 3). Non-complemented LoVo (MSH2−/−) cell line as well as the MMR deficient MutSα E749K variant failed to produce any detectable repair (0%), verifying the specificity of the assay.
According to the SIFT program, N127S was predicted to be “not tolerated” whereas G322D was “tolerated”. SIFT calculated that the normalized probabilities (SIFT scores) were 0.01 and 0.53, respectively, for these substitutions to appear in an alignment of orthologs.
The regularly applied means to differentiate between pathogenic and non-pathogenic variations of MMR genes, such as presence in healthy control chromosomes, MSI status and staining of the respective protein in tumor tissue, segregation with disease phenotype and conservation of the altered amino acid often fail to provide exclusive evidence on the actual role of the variants in the pathogenesis of HNPCC. Functional assays are available in limited quantities and differences between various assays may lead to contradictory results.9, 32 MMR gene mutation databases and SNP databases provide tools for researches and clinicians by collecting and classifying the vast amounts of published data on gene variations. However, data in different databases are not always in agreement. For instance, in the genetic databases MSH2 N127S and MSH2 G322D variants are reported both as polymorphic and pathogenic (http://www.insight-group.org/, http://www.genome.utah.edu/genesnps/). In a functional database, MSH2 G322D is classified as being likely pathogenic, based on functional assays in yeast (www.mmrmissense.net).
Here, the functionality of MSH2 N127S and G322D, 2 frequently reported MMR gene UVs, was for the first time assessed in human homologous system. We used 3 different functional analyses to evaluate their pathogenic significance and mechanism in vitro. Neither variant showed defects in interaction with MSH6, mismatch binding activities, or MMR efficiency compared to wild-type MSH2. Furthermore, our preliminary experiments on kinetic properties of MSH2 variants indicate that the MSH2 N127S and G322D mutations do not interfere with the release of MutSα from the mismatch, which has been shown to occur upon ATP uptake33(data not shown). Overall, our functional results demonstrate that these variants do not compromise MMR at least when being the only MMR variation in a cell. However, it is worthy of consideration that in many families a mutation carrier was shown to have also another inherited MMR gene mutation (Table I). Based on our assays, we cannot exclude the possible compound effects of 2 MMR mutations when they occur simultaneously in the same individual. Indeed, Tanyi et al. reported an HNPCC family with a truncating mutation in MSH2, which when in combination with MSH2 N127S leads to HNPCC phenotype in very early age.17 Although the majority of the published data discusses MSH2 G322D as a neutral polymorphism,3, 5, 7–12 it has also been hypothesized to be a low penetrance allele.24 This interpretation was supported by the functional analyses conducted with yeast assays, where the putative homologous mutation Msh2 D317D in yeast was reported to cause a moderate, as yet not complete inactivation of MMR.14, 34 In contrast with these interpretations, a recently published population-based study raises a possibility of G322D being instead of pathogenic a protective allele35 further complicating its classification.
Our comparative sequence alignment analysis (SIFT) predicted N127S to be “not tolerated” and G322D “tolerated”. According to our functional analyses, the prediction was correct for G322D, but incorrect for N127S. The positive predictive value of SIFT in predicting deleterious germline mutations in MMR genes was previously found to be between 82 and 90%.36 The amino acid N127 is highly but not completely conserved, so it is surprising but not inconceivable that the N127S variant retains its MMR capability.
To date, we have functionally characterized altogether 62 non-truncating mutations in MMR genes28, 29, 37, 38 (our study). According to our functional assays, 25 of them (10/34 in MLH1, 5/17 in MSH2 and 10/11 in MSH6) could have been classified as non-pathogenic variations. Indeed, the assays distinguished alterations with severe functional defects from those not or slightly impaired in protein function. Furthermore, the severe biochemical defects were mirrored by typical HNPCC characteristics such as early age at onset and high MSI, whereas variants with no or mild defects in functionality were associated with variable clinical phenotypes. The results indicate the importance to interpret the biochemical results together with clinical data and that many of minor mutations in MMR genes cannot be strictly classified as pathogenic or non-pathogenic. Taken to account that in addition to MMR function, the MMR proteins are known to be involved in other cellular processes, such as DNA damage signaling, apoptosis and recombination (reviewed e.g. in Ref. 39) and that our analyses do not exclude pathogenic effects on those functions, an important question to address is whether a group of variants with possible low cancer risk should be distinguished from clearly pathogenic and harmless variants. Keeping the clinical phenotype of mutation carriers as a basic criterium for pathogenicity determination, the non-pathogenic MMR gene variations would include only the ones, which have never been associated with cancer phenotype. As discussed in our study, MSH2 N127S and MSH2 G322D represent variations, which are associated with variable cancer phenotypes together with occurrence in healthy individuals. Although they have been shown to retain the repair activity in contrast to typical MMR-deficient mutations providing support to their non-predisposing role in HNPCC tumorigenesis, their low cancer risk in vivo cannot be excluded.
The authors thank Dr. Josef Jiricny for providing the MSH2 and MSH6 complementary DNAs.
- 12Mutation analysis of the MLH1. MSH2 and. MSH6 genes in patients with double primary cancers of the colorectum and the endometrium: a population-based study in northern. Sweden. Int J Cancer 2004; 109: 370–6., , , , , , .
- 35Classification of ambiguous mutations in DNA mismatch repair genes identified in a population-based study of colorectal cancer. Hum Mutat 2007., , , , , , , , , , .