Familial Cancer (2009) 8:489–500 DOI 10.1007/s10689-009-9274-4
Functional characterization of rare missense mutations in MLH1 and MSH2 identified in Danish colorectal cancer patients Lise Lotte Christensen Æ Reetta Kariola Æ Mari K. Korhonen Æ Friedrik P. Wikman Æ Lone Sunde Æ Anne-Marie Gerdes Æ Henrik Okkels Æ Carsten A. Brandt Æ Inge Bernstein Æ Thomas V. O. Hansen Æ Rikke Hagemann-Madsen Æ Claus L. Andersen Æ Minna Nystro¨m Æ Torben F. Ørntoft
Published online: 21 August 2009 Ó Springer Science+Business Media B.V. 2009
Abstract Recently, we have performed a population based study to analyse the frequency of colorectal cancer related MLH1 and MSH2 missense mutations in the Danish population. Half of the analyzed mutations were rare and most likely only present in the families where they were identified originally. Some of the missense mutations were located in conserved regions in the MLH1 and MSH2 proteins indicating a relation to disease development. In the present study, we functionally characterized 10 rare missense mutations in MLH1 and MSH2 identified in 13 Danish CRC families. To elucidate the pathogenicity of the missense mutations, we carried out in vitro functional
analyses. The missense mutations were analyzed for their effect on protein expression and repair efficiency. The results of the functional analysis were correlated with clinical data on the families carrying these mutations. Eight missense mutations resulted in proteins with expression and repair efficiency similar to the wild type. One missense mutation (MSH2 p.Met688Val) caused reduced protein expression and one (MSH2 p.Leu187Arg) caused both reduced protein expression and repair deficiency. The MSH2 p.Leu187Arg mutation was found in an Amsterdam II family presenting with high microsatellite instability and loss of MSH2 and MSH6 proteins in tumours. In conclusion, only 1/10 missense mutations displayed repair deficiency and could be classified as pathogenic. No final
The first authorship is shared by Lise Lotte Christensen, Reetta Kariola and Mari Korhonen. L. L. Christensen (&) F. P. Wikman C. L. Andersen T. F. Ørntoft Molecular Diagnostic Laboratory, Aarhus University Hospital, Skejby, Denmark e-mail:
[email protected] F. P. Wikman e-mail:
[email protected] C. L. Andersen e-mail:
[email protected] T. F. Ørntoft e-mail:
[email protected] R. Kariola M. K. Korhonen M. Nystro¨m Department of Biological and Environmental Sciences, Genetics, University of Helsinki, Helsinki, Finland e-mail:
[email protected] M. K. Korhonen e-mail:
[email protected]
L. Sunde Department of Clinical Genetics, Aarhus University Hospital, Aarhus, Denmark e-mail:
[email protected] A.-M. Gerdes Department of Clinical Biochemistry and Clinical Genetics, Odense University Hospital, Odense, Denmark e-mail:
[email protected] H. Okkels Section of Molecular Diagnostics, Department of Clinical Biochemistry, Aarhus University Hospital, Aalborg, Denmark e-mail:
[email protected] C. A. Brandt Department of Clinical Genetics, Vejle Hospital, Vejle, Denmark e-mail:
[email protected]
M. Nystro¨m e-mail:
[email protected]
123
490
conclusion can be drawn on the MSH2 p.Met688Val mutation, which caused reduced protein expression. Although, no deficiencies have been identified in the proteins harbouring the other missense mutations, pathogenicity of these variants cannot be unambiguously excluded. Keywords HNPCC Misssense mutation Mismatch repair Functional assay MLH1 and MSH2 Danish population Abbreviations CRC Colorectal cancer HNPCC Hereditary non-polyposis colorectal cancer IHC Immunohistochemistry MAPP-MMR Multivariate Analysis of Protein Polymorphisms-MisMatch Repair MMR Mismatch repair MSI Microsatellite instability TE(s) Total protein extract(s) VUS Variants of uncertain significance wt Wild type
Introduction Hereditary non-polyposis colorectal cancer (HNPCC) also known as Lynch Syndrome is the most common form of hereditary colorectal cancer (CRC) [1, 2]. HNPCC is a dominantly inherited cancer susceptibility syndrome accounting for at least 2–3% of all CRCs [3]. Diagnosis of HNPCC is generally based on kindred analysis using the Amsterdam II criteria [4]. HNPCC is often associated with loss of function germline mutations identified in one of several mismatch repair (MMR) genes, mainly MLH1 and MSH2, and in a fewer cases in MSH6 and PMS2 [5]. A population-based study has shown a considerable high
I. Bernstein Danish HNPCC Registry, Department of Gastroenterology, Hvidovre Hospital, Hvidovre, Denmark e-mail:
[email protected] T. V. O. Hansen Department of Clinical Biochemistry, Rigshospitalet University Hospital, University of Copenhagen, Copenhagen, Denmark e-mail:
[email protected] R. Hagemann-Madsen Department of Pathology, Aarhus University Hospital, Aarhus Sygehus, Denmark e-mail:
[email protected]
123
L. L. Christensen et al.
frequency of MSH6 mutations in Danish HNPCC families although mutations in MLH1 and MSH2 are still predominant [6]. The MMR genes are involved in post-replicative DNA repair and defects in these genes result in an increased rate of mutations, especially in mono-, di- and trinucleotide repeats leading to microsatellite instability (MSI), a hallmark of MMR defective tumours (reviewed by Marra and Boland) [7]. In most HNPCC families, the identified mutations are nonsense, frameshift, non-coding mutations affecting splice sites or large genomic rearrangements and thus clearly pathogenic [8]. However, numerous missense, silent and non-coding MMR variants have also been identified in HNPCC and HNPCC suspected families (families that do not fulfil the strict Amsterdam criteria). The influence of these variants on cancer risk is often difficult to predict. Some well defined criteria must be considered to distinguish disease causing missense mutations from non-pathogenic variants: (1) co-segregation of the variant with the disease, (2) presence/absence of the variant in the healthy population, (3) MSI in tumour tissue of variant carriers, (4) lack of immunohistochemical staining of the relevant MMR protein in the tumour tissue of variant carriers, and (5) in silico functional analysis. However, segregation analyses are often not feasible due to limited family sizes and unavailability of clinical specimens. In addition, segregation of the variant with the disease may simply reflect that the variant is in linkage disequilibrium with a yet unidentified pathogenic mutation. Further, population studies are laborious since a significant number of controls must be screened. MSI and absence of the relevant MMR proteins in tumour tissue do not necessarily provide evidence of pathogenicity since these conditions may be caused by another unidentified mutation in a MMR gene or its regulatory sequences or by somatic inactivation of the MMR gene. It has also been shown that missense mutations (e.g., MLH1 p.Glu578Gly) associated with CRC do not correlate with MSI in tumour tissue [9]. Several studies have shown that missense mutations causing MMR deficiency do not result in the simultaneous loss of the corresponding protein in tumor tissue [10, 11]. Finally, the results of the in silico functional analysis does not always reflect the functional consequence in vivo. Accordingly, in vitro functional analyses of individual missense variants may be carried out to provide further knowledge about the functional effects of the variants at the protein level. Different in vitro functional assays have revealed loss of function mutations that are most likely pathogenic, variants with reduced activity, and variants that do not affect protein functionality e.g., [10, 12–18]. None of the above described indicators of pathogenicity, including the functional assays, are informative alone but must be combined to be able to draw final conclusions about the pathogenicity of a given missense mutation.
Functional characterization of rare missense mutations
Genotyping of MLH1 and MSH2 in Danish HNPCC families or in non-HNPCC families having a familiar accumulation of CRC have led to the identification of several missense mutations of unknown significance in addition to the clearly pathogenic mutations. Recently, we have performed a population based study to determine the frequency of previously identified unclassified variants in MLH1 and MSH2 in the Danish population and to analyze whether some of the common variants were involved in an increased susceptibility to CRC in the Danish population [19]. That study demonstrated that 16 out of 20 analyzed missense mutations were rare, since they were not identified in the 380 Danish patients with sporadic CRC nor in a sub-cohort of 770 Danish individuals. The current study was carried out to determine whether some of these clinically identified rare missense mutations in MLH1 and MSH2 give rise to dysfunctional proteins and thus confirm their pathogenicity. A total of 10 missense mutations identified in 13 Danish families with inherited risk of CRC were analyzed. Subsequently, the results of the functional analyses were combined with clinical and genetic information to draw final conclusions about the pathogenicity of the individual missense mutations.
Materials and methods MLH1 and MSH2 missense mutations and associated CRC families The present study comprised 2 MLH1 missense mutations (p.Glu460Ala and p.Arg687Trp) and 8 MSH2 missense mutations (p.Thr44Met, p.Ala45Val, p.Leu187Arg, p.Phe519Ile, p.Met688Val, p.Val722Ile, p.Ala848Ser and p.Glu886Gly) identified in 13 Danish CRC families who underwent genetic testing for hereditary CRC. Four missense mutations (MLH1: p.Glu460Ala and MSH2: p.Leu187Arg, p.Phe519Ile and p.Glu886Gly) were identified in 6 families fulfilling the Amsterdam II criteria. Four of these families also carried pathogenic mutations in either MSH2 or MSH6 (see Table 1 for details). Five missense mutations (MLH1: p.Arg687Trp and MSH2: p.Thr44Met, p.Ala45Val, p.Val722Ile and p.Ala848Ser) were identified in five families not fulfilling the Amsterdam II criteria. Finally, the MSH2 p.Met688Val missense mutations was identified in two families, of which one fulfilled the Amsterdam II criteria and carried a mutation in MLH1 (Table 1). Six of the missense mutations (MLH1: p.Glu460Ala and MSH2: p.Thr44Met p.Ala45Val, p.Leu187Arg, p.Met688Val and p.Ala848Ser) had previously been included in a population study demonstrating that they were not present in a subcohort of 770 Danish individuals [19]. The missense mutations affected both conserved and non-conserved amino acid
491
residues causing both conservative and non-conservative amino acid changes (Table 1). The index patients’ mean age at tumour diagnosis was low, 48 years in MLH1 variant carriers and 40 years in MSH2 variant carriers. Table 1 summarizes the genetic and clinical characteristics of the CRC families. In some of the CRC families, the most probable disease causing mutation had already previously been identified. Nevertheless, missense mutations identified in these families were functionally characterized to interpret their pathogenicity. The patients have given informed consent to genetic testing and to the additional analyses performed in this study. Site-directed mutagenesis and generation of expression vectors Wild-type MLH1 and wild-type MSH2 cDNAs had previously been cloned into the pFastBac1 vector (Invitrogen, Carlsbad, CA) [20, 21]. Two mutations were introduced into the MLH1 cDNA and eight mutations into the MSH2 cDNA using QuickChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) according to the instructions provided by the manufacturer. The primer sequences and the PCR conditions are available from the authors on request. The entire reading frame of the variant cDNA was verified by sequencing (ABIPrism, 3100 genetic Analyzer, Applied Biosystems, Foster City, CA) prior to further use. Subsequently, recombinant baculoviruses carrying each of the variant cDNAs were generated using the Bac-to-Bac baculovirus expression system (Invitrogen, Carlsbad, CA) according to the manufactures instructions. The MLH1-wt, MLH1-Arg687Trp and PMS2-wt cDNAs were further cloned, as described previously, into the EGFP-N1 vector (BD biosciences, Palo Alto, CA) to generate constructs expressing MLH1-wt-EGFP, MLH1-Arg687Trp-EGFP and PMS2-wt-EGFP fluorescent fusion proteins [10]. The MLH1-wt-N1, the MLH1-Arg687Trp-N1 and the PMS2-wtN1 plasmids, are modified forms of the EGFP-N1 vector (BD biosciences, Palo Alto, CA) in which the EGFP gene has been replaced by either the MLH1-wt, the MLH1Arg687Trp or the PMS2-wt genes [22]. These constructs generates the expression of MLH1-wt, MLH1-Arg687Trp and PMS2-wt proteins, respectively. Expression of recombinant proteins in insect cells The recombinant proteins were produced in Spodoptera frugiperda (Sf9) insect cells using Bac-to Bac baculovirus expression system (Invitrogen, Carlsbad, CA) as previously described [20, 21]. Recombinant baculoviruses (Bacmid DNAs) containing the following cDNAs: MLH1-wt, MLH1Glu460Ala, MLH1-Arg687Trp, MSH2-wt, MSH2-Thr44 Met, MSH2-Ala45Val, MSH2-Leu187Arg, MSH2-Phe519
123
123
c.2062 A[G
c.2062 A[G
c.2164 G[A
c.2542 G[T
c.2657 A[G
c.2657 A[G
p.Met688Val
p.Met688Val
p.Val722Ile
p.Ala848Ser
p.Glu886Gly
p.Glu886Gly
NA not available, Red reduced
c.1555 T[C
p.Phe519Leu
41/rectum
34/rectum
NA/colon
30/colon
46/colon
54/colon
40/rectum
31/rectum
45/colon
c.134 C[T
c.560 T[G
p.Ala45Val
NA/colon (adenomas)
p.Leu187Arg
p.Thr44Met
48/endometrie
34/rectum 53/colon
Index patient: age of onset/tumour site
c.131 C[T
c.2059 C[T
p.Arg687Trp
MSH2
c.1379 A[C c.1379 A[C
Nucleotide change
p.Glu460Ala p.Glu460Ala
MLH1
Missense mutation
Table 1 Genetic and clinical data of the CRC families
MSH6 (NA)
MSH6 (NA)
No
MLH1 (c.1039 -8 T[A)
MSH6 (p.Ala1339Val
MLH1 (p.Thr117Met)
No
No
No
No
No
No
MSH2 (del ex8) MSH2 (p.Met663 fs)
Other MMR gene mutations in the family
Yes
Yes
No
No
Yes
No
Yes
Yes
No
No
No
Yes Yes
Amsterdam criteria II
Polar to nonpolar
Polar to nonpolar
Nonpolar to polar
Nonpolar to nonpolar
Nonpolar to nonpolar
Nonpolar to nonpolar
Nonpolar to nonpolar
Nonpolar to polar
Nonpolar to nonpolar
Polar to nonpolar
Polar to nonpolar
Polar to nonpolar Polar to nonpolar
Type of AA change
NA
NA
Stable
Stable
High
NA
Stable
High
NA
NA
High
NA NA
MSI status
?
?/-
NA
NA
-
?
?
?
NA
NA
-/Red.
? NA
MLH1
IHC
Red/-
?
NA
?
?
?
?
-
NA
NA
?
NA
MSH2
-
?/Red
NA
NA
NA
?
?
-
NA
NA
Red.
NA
MSH6
[47]
[47]
[19]
This study
[19]
[19]
This study
[19]
[47]
[47]
[35–37]
[19] [19]
References
492 L. L. Christensen et al.
Functional characterization of rare missense mutations
Ile, MSH2-Met688Val, MSH2-Val722Ile, MSH2-Ala848Ser and MSH2-Glu886Gly were used to infect Sf9 insect cells. For protein production, Sf9 cells were co-infected with MLH1 and PMS2 baculoviruses to generate the stable heterodimeric MutLa (MLH1-PMS2) protein complex required for mismatch repair activity [23]. Likewise the MSH2 and MSH6 baculoviruses were co-infected to generate the stable and active heterodimeric MutSa (MSH2-MSH6) complex [23, 24]. The total protein extracts (TEs) were prepared as described previously [20]. Expression and localization of fluorescent proteins in human cells The expression of MLH-wt, MLH1-Arg687Trp and PMS2wt proteins and MLH1-wt-EGFP, MLH1-Arg687TrpEGFP and PMS2-wt-EGFP fluorescent fusion proteins in 293T or HCT116 cells were performed using 3 different transfection combinations expressing the following fusion proteins: (1) MLH1-wt-EGFP, MLH1-Arg687Trp-EGFP or PMS2-wt-EGFP alone (2) MLH1-wt-EGFP or MLH1Arg687Trp-EGFP with PMS2-wt and (3) MLH1-wt or MLH1-Arg687Trp with PMS2-wt-EGFP wt. The cells lines 293T and HCT116 both lack the expression of MLH1 and PMS2 protein. 1 9 105 cells were seeded onto glass coverslips and transfected with 1 lg of MLH1-EGFP (wt or variant), 1 lg MLH1-N1 (wt or variant) and 1 lg of PMS2-wt-EGFP or PMS2-wt-N1 vectors using 4 ll of TurboFectTM in vitro transfection reagent (Fermentas, Germany). Following transfection, the cells were cultured for 24 h. For fluorescence detection, 24 h after transfection the cells were washed twice with PBS and fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. After fixation, the cells were washed with PBS, and the nuclei were stained incubating the cells in PBS with 300 nM 40 ,60 -diamidino-2-phenylindole (DAPI)(Sigma– Aldrich, St.Louis, MO) for 3 min. Slides were mounted with Fluorescence Mounting Medium (DAKO, Carpinteria, CA). Subcellular localization of recombinant proteins were analyzed by direct fluorescence using Axiplan 2 microscope (Carl Zeiss, Thorwood, NY) with 639 objective. Each transfection was performed in triplicate in both cell lines, and at least 100 cells from each replicate were analyzed from randomly sampled microscope fields of view. Representative images were taken with Isis 3.4.3 software (Metasystems, Altlussheim, Germany). Western blotting analysis The expression of the different recombinant MutLa and MutSa complexes were analyzed by Western blotting. Total protein extracts from the Sf9 cells (1,5 lg MutLa-wt,
493
1,5 lg MutLa-Glu460Ala, 1,5 lg MutLa-Arg687Trp, 3 lg MutSa-wt, 2 lg MutSa-Thr44Met, 3 lg MutSa-Ala45Val, 23 lg MutSa-Leu187Arg, 2 lg MutSa-Phe519Ile, 8 lg MutSa-Met688Val, 4 lg MutSa-Val722Ile, 4 lg MutSaAla848Ser and 4 lg MutSa-Glu886Gly) were loaded on a 6% SDS–PAGE gel, blotted to nitrocellulose membranes and detected with anti-MLH1 (MLH1 clone 168-15 BD Biosciences, 0,2 lg/ml), anti-MSH2 (MSH2 Ab-2, NA27 Calbiochem Oncogene Research, San Diego, CA, 0,4 lg/ml), anti-PMS2 (PMS2 Ab-1 Calbiochem/Oncogene Research, San Diego, CA, 0,2 lg/ml) and anti-MSH6 (MSH6 clone 44, BD Transduction Laboratories, San Jose´, CA 0,17 lg/ml) antibodies. In vitro mismatch repair (MMR) assay The in vitro MMR assay was performed as described previously [10, 14]. Briefly, 75 lg of nuclear extracts from the MMR deficient cell lines HCT116 (MLH1-/-) and LoVo (MSH2-/-) were incubated with MutLa or MutSa recombinant proteins, respectively, in the presence of 100 ng of circular DNA heterodublexes containing a GT mismatch 370 bp 50 downstream from a single-strand nick. Given that the recombinant proteins were not expressed at equal levels, the total protein amounts were adjusted to contain similar quantities of recombinant MutLa or MutSa proteins (3 lg MutLa-wt, 3 lg MutLa-Glu460Ala, 3 lg MutLaArg687Trp, 3 lg MutSa-wt, 2 lg MutSa-Thr44Met, 3 lg MutSa-Ala45Val, 23 lg MutSa-Leu187Arg, 2 lg MutSaPhe519Ile, 8 lg MutSa-Met688Val, 4 lg MutSa-Val722Ile, 4 lg MutSa-Ala848Ser and 4 lg MutSa-Glu886Gly). Successful repair converts the GT heterodublex to a AT homoduplex generating a BglII restriction site, which allows the repair activity to be visualized using restriction analysis. Consequently, the repair efficiency can be measured by the cleavage efficiency of BglII. Nuclear extracts from Hela cells (MLH1?/?and MSH2?/?) without complement were used as positive control whereas nuclear extracts from HCT116 (/MLH1-/-) and LoVo (MSH2-/-) cells without complement were used as negative controls. The repair percentages were analyzed using Image-Pro 4.0 (Media Cybernetics) and calculated as an average of 3 independent experiments. In silico analysis In silico prediction of the functional consequence of the missense mutations was performed using SIFT (Sorting Intolerant From Tolerant): http://blocks.fhcrc.org/sift/SIFT. html [25], Polyphen: http://coot.embl.de/PolyPhen/ [26], PMut: http://mmb2.pcb.ub.es:8080/PMut/ [27] and MAPPMMR (Multivariate Analysis of Protein PolymorphismsMisMatch Repair): http://mappmmr.blueankh.com/ [28].
123
494
Results Expression of MLH1 and MSH2 protein variants in Sf9 insect cells Initially, we tested the expression of the MutLa (MLH1PMS2) wt/variant and MutSa (MSH2-MSH6) wt/variants in Sf9 insect cells using Western blotting analysis. All the variants and the wt proteins were expressed with success (Fig. 1). Two variants in MSH2 (Met688Val and Leu187Arg) demonstrated reduced protein expression of MSH2 and MSH6, whereas the others were expressed at levels comparable to the wt proteins. Repair efficiency of MutLa and MutSa protein variants Subsequently, we analyzed the ability of the recombinant MutLa and MutSa variants to complement MMR-defective HCT116 (MLH1-/-) or LoVo (MSH2-/-) extracts in repairing GT mismatches in vitro (Fig. 2). The MutLa and
L. L. Christensen et al.
MutSa protein complexes were not expressed at equal levels, consequently the TEs used were adjusted to contain similar quantities of MutLa or MutSa. Nuclear extracts from Hela cells were used as positive control whereas nuclear extracts from HCT116 and LoVo cells without complement were used as negative controls. Nine out of 10 analyzed missense mutations demonstrated repair efficiency at levels comparable to the wt protein complexes (i.e. more than 30% (MLH1) and app. 10% (MSH2) of the added heterodublex DNA was repaired) (Fig. 2). MutSaLeu187Arg completely lacked the ability to repair the added heterodublex DNA, whereas the other variants demonstrated repair activity comparable to the wt protein (Fig. 2). It has been shown, that the pathogenicity of variants demonstrating diminished expression levels but with intact repair activity, is linked to shortage of the functional protein [22]. Consequently the MMR assay was repeated using only 1.6 lg of MSH2 p.Met688Val recombinant protein. Lowering the amount of variant protein did not change the repair efficiency (data not shown).
Fig. 1 Western blot analysis of MutLa and MutSa wild type proteins and variants in total protein extracts (TEs) of Sf9 cells. MutLa-WT, MutLa- Glu460Ala and MutLa-Arg687Trp contain similar amounts of MLH1 and PMS2 proteins. MutSa-WT and six MutSa variants (Thr44Met, Ala45Val, Phe519Ile, Val722Ile, Ala848Ser and Glu886Gly) contain similar amounts of MSH2 and MSH6 proteins.
Five times more of the TEs of MutSa-Leu187Arg and MutSaMet688Val were loaded compared to the wt and other variants. In these complexes the amounts of MSH2 and MSH6 proteins was clearly decreased compared to the expression levels of the wt proteins. Protein production in insect cells was repeated three times
Fig. 2 In vitro mismatch repair assay in vitro MMR efficiency of nuclear extracts (NEs) from HCT116 (MLH1-/-) and LoVo (MSH2-/-) complemented with MutLa wt or variant and Mutsa wt or variant complexes, respectively. NEs from Hela cells without complementation were used as positive control whereas NEs from HCT116 and LoVo cells without complement were used as negative controls. The upper fragment represents unrepaired linearized G.T
mismatch-containing plasmid DNA. The two lower fragments show repaired and double-digested DNA (BsaI (linearization of DNA) and BglII). Relative repair efficiency (%) calculated as the ratio of double digested DNA relative to total DNA added to the reaction. The repair efficiencies (%) represent the average of three independent experiments
123
Functional characterization of rare missense mutations
Subcellular localization of fluorescent MutLa-Arg687Trp The MLH1 p.Arg687Trp variant demonstrated similar results as MLH1-wt with regard to stability and repair efficiency. However, the missense mutations co-segregates with the disease in the family with at lod score of 1.5 (data not shown) (pedigree is shown in Fig. 3b). In addition,
495
immunohistochemical (IHC) analysis of the tumour tissue from selected family members all demonstrated lack/ reduced staining of MLH1 protein. Therefore, further functional analysis of this missense mutation was undertaken. To study the subcellular localization, the MLH1 wt and variant cDNAs were fused to the EGFP cDNA followed by transient expression of the fluorescent proteins in 293T and HCT116 human cells. Three different
Fig. 3 Pedigrees of the family carrying the MSH2 p.Leu187Arg missense mutation (a) and of the family carrying the MLH1 p.Arg687Trp (b). The proband is indicated with an arrow. The IHC data are stated as ?/-/reduced. Ca. Cancer, Ad. Adenoma and Papy. Papyloma
123
496
L. L. Christensen et al.
combinations of vectors: MLH1-wt-EGFP or MLH1-Arg 687Trp-EGFP alone, MLH1-wt-EGFP or MLH1-Arg687 Trp-EGFP with PMS2-wt-N1 and PMS2-wt-EGFP with MLH1-wt-N1 or MLH1-Arg687Trp-N1 were used in the transfection experiments, as described previously [10]. The results of the subcellular localization analysis in 293T cells are shown in Fig. 4a–g (similar results were obtained with the HCT116 cells, data not shown). The MLH1-wt-EGFP was detected almost entirely in the nucleus in the absence
of PMS2-wt (Fig. 4a). The addition of PMS2-wt resulted in complete localization of MLH1-wt-EGFP to the nucleus (Fig. 4b). The PMS2-wt-EGFP protein, on the contrary, was located in the cytoplasm in the absence of MLH1-wt (Fig. 4c). Co-expression of PMS2-wt-EGFP with MLH1wt resulted in nuclear import of PMS2-wt-EGFP (Fig. 4d). The MLH1-Arg687Trp variant acted similar to the MLH1wt (Fig. 4e–g) and could hence be classified as normal with regard to subcellular localization.
Fig. 4 Sub-cellular localization of MLH1-wt-EGFP and MLH1Arg687Trp-EGFP fusion protein in 293T cells was detected using direct fluorescence analysis. Staining of the nuclei with DAPI is shown on separate photos. a Nuclear expression of MLH1-wt-EGFP without PMS2-wt. b Coexpression of MLH1-wt-EGFP with PMS2wt. c Cytoplasmic expression of PMS2-wt-EGFP without MLH1-wt.
d Coexpression of PMS2-wt-EGFP with MLH1-wt. e Nuclear expression of MLH1-Arg687Trp-EGFP without PMS2-wt. f Coexpression of MLH1- Arg687Trp-EGFP with PMS2-wt. g Coexpression of PMS2-wt-EGFP with MLH1-Arg687Trp. (Original magnification 639)
123
Functional characterization of rare missense mutations
497
Table 2 Summary of the in vitro functional and in silico analyses Missense mutation
Expression in Sf9 cells
In vitro MMR assay
SIFT
Polyphen
PMut
MAPP-MMR
p.Glu460Ala
Normal
Normal
Tolerated
Benign
Pathological
NA
p.Arg687Trp
Normal
Normal
Not-tolerated
Possible damaging
Pathological
Neutral
p.Thr44Met
Normal
Normal
Not tolerated
Possible damaging
Neutral
Neutral
p.Ala45Val
Normal
Normal
Tolerated
Benign
Neutral
Neutral
p.Leu187Arg
Reduced
Deficient
Not tolerated
Probable damaging
Neutral
Deleterious
p.Phe519Leu
Normal
Normal
Tolerated
Benign
Neutral
Neutral
p.Met688Val
Reduced
Normal
Not tolerated
Probable damaging
Neutral
Deleterious
p.Val722Ile
Normal
Normal
Not tolerated
Benign
Neutral
Neutral
p.Ala848Ser p.Glu886Gly
Normal Normal
Normal Normal
Not tolerated Tolerated
Possible damaging Benign
Neutral Pathological
Deleterious Neutral
MLH1
MSH2
NA not available
In silico analysis In silico analysis were carried out to compare the in silico prediction of pathogenicity with the results of the in vitro functional analysis. Four in silico prediction programs; SIFT, Polyphen, PMut and MAPP-MMR were used for this analysis [25–28]. SIFT predicted 6/10 variants to be ‘‘not tolerated’’. Using polyphen 3/10 variants were classified as ‘‘possibly damaging’’ and 2/10 as ‘‘probably damaging’’, whereas 3/10 were ‘‘pathological’’ according to PMut. Finally, MAPP-MMR predicted 3/10 variants to be deleterious using a MMP-MMR score of 4.55 as threshold. Chao et al. found that known deleterious Lynch syndrome missense variants presented with a mean MAPP-MMR score of 16.5 whereas a mean MAPP-MMR score of 13.5 was found for predicted deleterious variants in subjects with familial CRC [28]. In contrast, the mean MAPP-MMR score of neutral variants was 3.5. MSH2 p.Leu187Arg and p.Met688Val analyzed in the present study showed a MAPP-MMR score of 33.8 and 17.4, respectively. These two variants were also the only ones predicted to be ‘‘probably damaging’’ by Polyphen. In addition, the MSH2 p.Leu187Arg and p.Met688Val variants were the only ones demonstrating reduced protein expression in insect cells. Nevertheless, only the p.Leu187Arg variant was classified as clearly pathogenic due to its additional lack of repair activity. A summary of the results of the in vitro and in silico analyses is shown in Table 2.
Discussion HNPCC syndrome associates with an inherited predisposition to cancer, predominantly colorectal cancer. HNPCC is primarily caused by deficiency of DNA mismatch repair,
due to inherited deleterious mutations in the MMR genes; MLH1, MSH2, MSH6 and PMS2 [5]. However, when screening the MMR genes in HNPCC or HNPCC suspected families, the geneticists are constantly challenged by the identification of germline mutations of uncertain significance (e.g., missense mutations). Functional consequences of single amino acid changes may vary from none to complete dysfunction of the protein, which makes the assessment of pathogenicity and hence the estimation of the cancer risk of the carriers difficult. The overall aim of the present study was to use functional analyses to elucidate whether 10 missense mutations identified in Danish patients with CRC could contribute to cancer development by altering protein expression and/or MMR efficiency. In addition, the subcellular localization was analyzed for one of the missense mutations. The missense mutations affected both conserved and non-conserved amino acid residues and resulted in both conservative and non-conservative amino acid changes. The missense mutations were included independently on the MSI status of the tumour or expression of the corresponding proteins in tumour tissue. Consequently, exclusion of low penetrant variants not presenting with a clearly MMR defective phenotype was avoided. Six out of 10 missense mutations had previously been included in a population study demonstrating that they were not present in a sub-cohort of 770 healthy Danish individuals [19]. Three missense mutations MLH1 p.Glu460Ala, MSH2 p.Met688Val and MSH2 p.Glu886Gly were identified in individuals that also carried a deleterious mutation in one of the other MMR genes (Table 1). These deleterious mutations co-segregate with the HNPCC-related cancers in these families and thus most likely explain the HNPCC phenotypes of these families. However, family members carrying only the missense mutations were also identified
123
498
in these families and hence we speculated that these missense mutations could be low penetrant variants causing e.g., CRC at an older age. In addition, a previous paper has suggested a compound effect of two MMR mutations occurring simultaneously in the same individual [29]. In that study the presence of a missense mutation (Asn127Ser) in MSH2 in combination with another truncating deleterious mutation in MSH2 lead to the manifestation of CRC at an earlier age. Similarly, biallelic mutations in MSH6 in combination with a missense mutation in APC have also been shown to be associated with a more severe phenotype in a patient with HNPCC [30]. Compound heterozygosity for mutations in other MMR genes e.g., MSH6 and PMS2 resulting in early onset of disease has been demonstrated in other families as well e.g., [31–33]. In the present functional study, we initially analyzed the MLH1/PMS2 and MSH2/MSH6 protein expression levels to determine the effects of the missense mutations on protein expression. The expression of 2 missense mutations (MSH2 p.Leu187Arg and p.Met688Val) was diminished compared to the expression levels of the wt proteins. The remaining 8 missense mutations demonstrated expression levels comparable to the wt proteins. Subsequently, we carried out the in vitro MMR assay to study the repair efficiency of each variant. Only one missense mutation, MSH2 p.Leu187Arg, demonstrated repair deficiency. In a recent functional study, human MSH2 missense mutations were introduced at cognate positions in yeast Msh2. That study showed that the MSH2 Thr44Met missense mutations demonstrated wt phenotype in all assays, including yeast in vivo MMR, expression level and subunit interaction, supporting the results in the present study [17]. The MSH2 p.Glu886Gly missense mutations was found to behave like the wt with regard to expression and subunit interaction, whereas the results of the yeast in vivo MMR assay were inconclusive [17]. MSH2 p.Leu187Arg has to our knowledge not been characterized functionally. However, another missense mutation (p.Leu187Pro) affecting the same codon has previously been shown to be MMR deficient and to exhibit low expression [11, 14]. MSH2 p.Leu187Arg, analyzed in the present study, was identified in a Danish HNPCC family (Fig. 3a). The proband had rectal cancer at the age of 31 years. The father and uncle of the proband were also affected with HNPCC-related cancer. All tumours from the three affected individuals lacked MSH2 and MSH6 at the protein level. In addition, MSI-H was demonstrated in tumour tissue from the father. These observations corroborate that the MSH2 p.Leu187Arg missense mutation is disease causing in this family. The MLH1 p.Arg687Trp missense mutation has been identified in several CRC families from different populations, e.g., Spain, Japan, Poland and Sweden [34–37]. The missense mutation was detected in a Spanish proband with
123
L. L. Christensen et al.
HNPCC and in 3 affected siblings and therefore considered pathogenic [37]. In addition, 3 children affected with gastrointestinal cancers and neurofibromatosis type 1 carried germline homozygous mutations of this missense mutation [38]. On the contrary, the missense mutation did not segregate consistently with disease in two Swedish families and was thus considered of unclear biologic relevance [34]. At the functional level, MLH1 p.Arg687Trp demonstrated pathogenic phenotype in three functional assays in yeast [18]. The same study also showed that MLH1 p.Arg687Trp exhibited reduced MMR activity and expression using a human cell-based system indication a role in disease development. In contrast, our study demonstrated that MLH1 p.Arg687Trp behaved as wt MLH1, both with regard to expression and MMR efficiency using a similar expression system. However, the MLH1 p.Arg687Trp was found to segregate with disease (LOD score 1.5) and to lead to loss/reduction of MLH1 protein in the tumours of 6 family members carrying the missense mutation, suggesting a role in disease development. Therefore, the subcellular localization of p.Arg687Trp was analyzed, to further assess its pathogenic role. The results of the sub-cellular localization analysis demonstrated that the MLH1 p.Arg687Trp protein acted similar to wt MLH1. Consequently, our results do not support that the MLH1 p.Arg687Trp missense mutation is in itself disease causing. We cannot rule out that the missense mutation is in linkage with another unidentified mutation affecting MLH1 expression nor can we rule out that methylation defects are causing the lack of MLH1 protein expression in tumour tissue. Indeed, germline epimutations of MLH1 have been demonstrated in two individuals with no germline mutations in any of the MMR genes but showing both molecular and clinical indications of HNPCC [39]. The remaining missense mutations analyzed in the present study have, to our knowledge, not previously been characterized functionally. In the present study, in silico variant classifications were generated using four different prediction algorithms; SIFT, Polyphen, PMut and MAPP-MMR [28]. The results of the in silico predictions were compared to the results of the functional analyses. MSH2 p.Leu187Arg and p.Met688Val were the only variants demonstrating abnormal behaviour in the functional analyses. SIFT, PolyPhen and MAPPMMR classified these variants as ‘‘not tolerated’’, ‘‘probably damaging’’ and ‘‘deleterious’’, respectively. In contrast both variants were classified as neutral by PMut. Yet, both SIFT, PolyPhen and MAPP-MMR also classified other variants as deleterious. However, p.Leu187Arg and p.Met688Val were the only variants classified as probably damaging using Polyphen. In addition, they alone demonstrated a MAPP-MMR score comparable to the scores obtained with deleterious variants from families with
Functional characterization of rare missense mutations
Lynch syndrome or familial CRC. Consequently, the predictions generated using both Polyphen and MAPP-MRR correlated well with the results of the functional assays in the present study. However, Chao et al. [28] have demonstrated that MAPP-MMR generally outperformed PolyPhen and SIFT with improved specificity and sensitivity. The functional assays used in our study do not reveal missense mutations that have an impact on splicing. Recently, several papers have dealt with the association of unclassified variants in MLH1 and MSH2 with splicing defects e.g., [40–43]. These studies have identified several unclassified variants including missense mutations that cause aberrant splicing. The majority of these variants are present within 30 or 50 splice sites whereas few were located at positions distinct from those defining exon boundaries probably disrupting splicing regulatory elements such as the ESEs (exon splicing enhancers) [42]. The missense mutations in the present study are not present within 30 or 50 splice sites but some of them either abolish or introduce ESEs (analyzed using SNAP [44] (data not shown). However, the bioinformatic prediction of sequence changes in regulatory elements lack specificity and it has become clear that in vivo or ex vivo testing of the effect of individual missense mutations on spicing is required [43]. The MLH1 p.Arg687Trp and the MSH2 p.Leu187Arg mutations included in the present study have been shown not to affect splicing using patient derived cell lines [41]. Regarding the other missense mutations, we cannot rule out that some of those may indeed cause aberrant splicing. In addition to the MMR activity the MMR proteins are also known to be involved in other cellular processes, such as DNA damage signaling, apoptosis and recombination (reviewed by Jiricny et al. [45]). The analyses performed in the present study do not exclude pathogenic effects on those processes. Recently, Couch et al. [46] proposed a decision tree to improve the classification of MMR gene variants identified in HNPCC or HNPCC suspected families. The decision tree included three steps: (1) an initial ‘‘biochemical diagnosis’’ using IHC and MSI analysis followed by sequencing of relevant MMR genes, (2) in silico analysis and in vitro MMR assays of identified missense variant and (3) complex functional analysis including protein stability, protein–protein interaction and cellular localization of missense mutations displaying normal MMR activity. However, a thoroughly validation of especially the more complex functional assays (step 3) is needed prior to their use in clinical interpretation. In summary, 2/10 analyzed missense mutations resulted in proteins with aberrant behaviour compared to the wt protein. The MSH2 p.Met688Val missense mutations only caused reduced protein expression and hence further investigations of that variant is needed. The MSH2
499
p.Leu187Arg missense mutation caused both reduced protein expression and repair deficiency. Supported by the clinical data of the family carrying the MSH2 p.Leu187Arg, we conclude that this mutation is most likely disease causing. It should, however, be emphasized that although, no obvious defects have been identified in the proteins harbouring the other missense mutations, pathogenicity of these variants cannot be unambiguously excluded. Acknowledgments We are especially grateful to Bente Devantie´ and Inge Lis Thorsen for skilful technical assistance. We thank Dr. Saara Ollila for a critical reading of the manuscript. This work was supported by grants from Sigrid Juselius Foundation and Academy of Finland (Grant number: 110300).
References 1. Weitz J, Koch M, Debus J et al (2005) Colorectal cancer. Lancet 365(9454):153–165 2. Garber JE, Offit K (2005) Hereditary cancer predisposition syndromes. J Clin Oncol 23(2):276–292 3. Allen BA, Terdiman JP (2003) Hereditary polyposis syndromes and hereditary non-polyposis colorectal cancer. Best Pract Res Clin Gastroenterol 17(2):237–258 4. Vasen HF, Watson P, Mecklin JP et al (1999) New clinical criteria for hereditary nonpolyposis colorectal cancer (HNPCC, Lynch syndrome) proposed by the International Collaborative group on HNPCC. Gastroenterology 116(6):1453–1456 5. Peltomaki P (2005) Lynch syndrome genes. Fam Cancer 4(3):227–232 6. Nilbert M, Wikman FP, Hansen TV et al (2008) Major contribution from recurrent alterations and MSH6 mutations in the Danish Lynch syndrome population. Fam Cancer 8:75–83 7. Marra G, Boland CR (1995) Hereditary nonpolyposis colorectal cancer: the syndrome, the genes, and historical perspectives. J Natl Cancer Inst 87(15):1114–1125 8. Mecklin JP (2008) The implications of genetics in colorectal cancer. Ann Oncol 19(suppl 5):v87–v90 9. Liu T, Tannergard P, Hackman P et al (1999) Missense mutations in hMLH1 associated with colorectal cancer. Hum Genet 105(5):437–441 10. Raevaara TE, Korhonen MK, Lohi H et al (2005) Functional significance and clinical phenotype of nontruncating mismatch repair variants of MLH1. Gastroenterology 129(2):537–549 11. Ollila S, Dermadi BD, Jiricny J et al (2008) Mechanisms of pathogenicity in human MSH2 missense mutants. Hum Mutat 29(11):1355–1363 12. Kondo E, Suzuki H, Horii A et al (2003) A yeast two-hybrid assay provides a simple way to evaluate the vast majority of hMLH1 germ-line mutations. Cancer Res 63(12):3302–3308 13. Ellison AR, Lofing J, Bitter GA (2001) Functional analysis of human MLH1 and MSH2 missense variants and hybrid humanyeast MLH1 proteins in Saccharomyces cerevisiae. Hum Mol Genet 10(18):1889–1900 14. Ollila S, Sarantaus L, Kariola R et al (2006) Pathogenicity of MSH2 missense mutations is typically associated with impaired repair capability of the mutated protein. Gastroenterology 131(5):1408–1417 15. Trojan J, Zeuzem S, Randolph A et al (2002) Functional analysis of hMLH1 variants and HNPCC-related mutations using a human expression system. Gastroenterology 122(1):211–219
123
500 16. Brieger A, Trojan J, Raedle J et al (2002) Transient mismatch repair gene transfection for functional analysis of genetic hMLH1 and hMSH2 variants. Gut 51(5):677–684 17. Gammie AE, Erdeniz N, Beaver J et al (2007) Functional characterization of pathogenic human MSH2 missense mutations in Saccharomyces cerevisiae. Genetics 177(2):707–721 18. Takahashi M, Shimodaira H, Andreutti-Zaugg C et al (2007) Functional analysis of human MLH1 variants using yeast and in vitro mismatch repair assays. Cancer Res 67(10):4595–4604 19. Christensen LL, Madsen BE, Wikman FP et al (2008) The association between genetic variants in hMLH1 and hMSH2 and the development of sporadic colorectal cancer in the Danish population. BMC Med Genet 9:52 20. Nystrom-Lahti M, Perrera C, Raschle M et al (2002) Functional analysis of MLH1 mutations linked to hereditary nonpolyposis colon cancer. Genes Chromosomes Cancer 33(2):160–167 21. Kariola R, Raevaara TE, Lonnqvist KE et al (2002) Functional analysis of MSH6 mutations linked to kindreds with putative hereditary non-polyposis colorectal cancer syndrome. Hum Mol Genet 11(11):1303–1310 22. Raevaara TE, Vaccaro C, Abdel-Rahman WM et al (2003) Pathogenicity of the hereditary colorectal cancer mutation hMLH1 del616 linked to shortage of the functional protein. Gastroenterology 125(2):501–509 23. Chang DK, Ricciardiello L, Goel A et al (2000) Steady-state regulation of the human DNA mismatch repair system. J Biol Chem 275(24):18424–18431 24. Marra G, Iaccarino I, Lettieri T et al (1998) Mismatch repair deficiency associated with overexpression of the MSH3 gene. Proc Natl Acad Sci U S A 95(15):8568–8573 25. Ng PC, Henikoff S (2003) SIFT: predicting amino acid changes that affect protein function. Nucleic Acids Res 31(13):3812–3814 26. Ramensky V, Bork P, Sunyaev S (2002) Human non-synonymous SNPs: server and survey. Nucl Acids Res 30(17):3894–3900 27. Ferrer-Costa C, Gelpi JL, Zamakola L et al (2005) PMUT: a webbased tool for the annotation of pathological mutations on proteins. Bioinformatics 21(14):3176–3178 28. Chao EC, Velasquez JL, Witherspoon MS et al (2008) Accurate classification of MLH1/MSH2 missense variants with multivariate analysis of protein polymorphisms-mismatch repair (MAPPMMR). Hum Mutat 29(6):852–860 29. Tanyi M, Olasz J, Lukacs G et al (2006) Pedigree and genetic analysis of a novel mutation carrier patient suffering from hereditary nonpolyposis colorectal cancer. World J Gastroenterol 12(8):1192–1197 30. Okkels H, Sunde L, Lindorff-Larsen K et al (2006) Polyposis and early cancer in a patient with low penetrant mutations in MSH6 and APC: hereditary colorectal cancer as a polygenic trait. Int J Colorectal Dis 21(8):847–850 31. Plaschke J, Linnebacher M, Kloor M et al (2006) Compound heterozygosity for two MSH6 mutations in a patient with early onset of HNPCC-associated cancers, but without hematological malignancy and brain tumor. Eur J Hum Genet 14(5):561–566
123
L. L. Christensen et al. 32. Rahner N, Hoefler G, Hogenauer C et al (2008) Compound heterozygosity for two MSH6 mutations in a patient with early onset colorectal cancer, vitiligo and systemic lupus erythematosus. Am J Med Genet A 146A(10):1314–1319 33. Auclair J, Leroux D, Desseigne F et al (2007) Novel biallelic mutations in MSH6 and PMS2 genes: gene conversion as a likely cause of PMS2 gene inactivation. Hum Mutat 28(11):1084–1090 34. Lagerstedt RK, Liu T, Vandrovcova J et al (2007) Lynch syndrome (hereditary nonpolyposis colorectal cancer) diagnostics. J Natl Cancer Inst 99(4):291–299 35. Furukawa T, Konishi F, Shitoh K et al (2002) Evaluation of screening strategy for detecting hereditary nonpolyposis colorectal carcinoma. Cancer 94(4):911–920 36. Jakubowska A, Gorski B, Kurzawski G et al (2001) Optimization of experimental conditions for RNA-based sequencing of MLH1 and MSH2 genes. Hum Mutat 17(1):52–60 37. Caldes T, Godino J, de la Hoya M et al (2002) Prevalence of germline mutations of MLH1 and MSH2 in hereditary nonpolyposis colorectal cancer families from Spain. Int J Cancer 98(5):774–779 38. Gallinger S, Aronson M, Shayan K et al (2004) Gastrointestinal cancers and neurofibromatosis type 1 features in children with a germline homozygous MLH1 mutation. Gastroenterology 126(2):576–585 39. Suter CM, Martin DI, Ward RL (2004) Germline epimutation of MLH1 in individuals with multiple cancers. Nat Genet 36(5):497–501 40. Lastella P, Resta N, Miccolis I et al (2004) Site directed mutagenesis of hMLH1 exonic splicing enhancers does not correlate with splicing disruption. J Med Genet 41(6):e72 41. Auclair J, Busine MP, Navarro C et al (2006) Systematic mRNA analysis for the effect of MLH1 and MSH2 missense and silent mutations on aberrant splicing. Hum Mutat 27(2):145–154 42. Tournier I, Vezain M, Martins A et al (2008) A large fraction of unclassified variants of the mismatch repair genes MLH1 and MSH2 is associated with splicing defects. Hum Mutat 29(12):1412–1424 43. Lastella P, Surdo NC, Resta N et al (2006) In silico and in vivo splicing analysis of MLH1 and MSH2 missense mutations shows exon- and tissue-specific effects. BMC Genomics 7:243 44. Li S, Ma L, Li H et al (2007) Snap: an integrated SNP annotation platform. Nucl Acids Res 35(suppl_1):D707–D710 45. Jiricny J (2006) The multifaceted mismatch-repair system. Nat Rev Mol Cell Biol 7(5):335–346 46. Couch FJ, Rasmussen LJ, Hofstra R et al (2008) Assessment of functional effects of unclassified genetic variants. Hum Mutat 29(11):1314–1326 47. Bisgaard ML, Jager AC, Myrhoj T et al (2002) Hereditary nonpolyposis colorectal cancer (HNPCC): phenotype-genotype correlation between patients with and without identified mutation. Hum Mutat 20(1):20–27
