Oligonucleotide-directed mutagenesis screen to identify pathogenic Lynch syndrome-associated MSH2 DNA mismatch repair gene variants

Abstract

Single-stranded DNA oligonucleotides can achieve targeted base-pair substitution with modest efficiency but high precision. We show that "oligo targeting" can be used effectively to study missense mutations in DNA mismatch repair (MMR) genes. Inherited inactivating mutations in DNA MMR genes are causative for the cancer predisposition Lynch syndrome (LS). Although overtly deleterious mutations in MMR genes can clearly be ascribed as the cause of LS, the functional implications of missense mutations are often unclear. We developed a genetic screen to determine the pathogenicity of these variants of uncertain significance (VUS), focusing on mutator S homolog 2 (MSH2). VUS were introduced into the endogenous Msh2 gene of mouse embryonic stem cells by oligo targeting. Subsequent selection for MMR-deficient cells using the guanine analog 6-thioguanine allowed the detection of MMR-abrogating VUS. The screen was able to distinguish weak and strong pathogenic variants from polymorphisms and was used to investigate 59 Msh2 VUS. Nineteen of the 59 VUS were identified as pathogenic. Functional assays revealed that 14 of the 19 detected variants fully abrogated MMR activity and that five of the detected variants attenuated MMR activity. Implementation of the screen in clinical practice allows proper counseling of mutation carriers and treatment of their tumors.

Keywords: DNA mismatch repair; Lynch syndrome; MSH2; site-directed mutagenesis; variants of uncertain significance.

Fig. 1.

Genetic screen for the identification of pathogenic MSH2 variants. (A) Msh2^+PUR/∆ mESCs were exposed to ssODNs that introduce the mutation of interest into the one endogenous Msh2 allele with an efficiency of 10⁻³–10⁻⁴. (B) The mESCs subsequently were exposed to 6TG. Cells that lost MMR activity form 6TG-resistant colonies. To remove cells that became MMR-deficient because of the loss-of-heterozygosity events, puromycin selection was performed simultaneously. (C) 6TG/puromycin-resistant colonies were selected and expanded. Sequence analysis was used to confirm the presence of the introduced mutation in the 6TG/puromycin-resistant cells.

Fig. S1.

Generation of Msh2^+PUR/∆ mESCs. In Msh2^+PUR/∆ mESCs one of the endogenous Msh2 alleles was deleted (∆), and a puromycin-resistance gene was introduced in front of the remaining Msh2 allele (+PUR). (A) Vectors containing a loxP site, a selection marker, and 6 kb homologous to the sequence either upstream or downstream of the Msh2 gene were created to introduce loxP sites flanking one of the Msh2 alleles. (B) The 3′ loxP vector was linearized by KpnI digestion and electroporated into wild-type 129/OLA mESCs. (C) Geneticin-resistant colonies were analyzed for integration of the 3′ loxP vector into its target site by Southern blot analysis. The genomic DNA was BglII-digested, and fragments were visualized using a probe (indicated by a yellow star). (D) Southern blot showing the wild-type 9.6-kb fragment in unmodified (Msh2^+/+) cells. The presence of an additional 8.6-kb fragment indicates correct integration of the 3′ loxP vector (LoxP integrated). (E) Cells with successful 3′ loxP vector integration were electroporated with the BamHI-linearized 5′ loxP vector. (F) Puromycin-resistant colonies were analyzed for integration of the 5′ loxP vector into the target site by Southern blot analysis. The genomic DNA was NsiI-digested, and fragments were visualized using a probe (indicated by a yellow star). (G) Southern blot showing the wild-type 9.9-kb fragment in unmodified (Msh2^+/+) cells. The presence of an additional 20.4-kb fragment indicates correct integration of the 5′ loxP vector (LoxP integrated). The 5′ loxP vector could integrate adjacent to either the wild-type or the 3′ loxP-modified Msh2 allele. (H) To delete the floxed Msh2 allele and create Msh2^+/∆ mESCs, CMV-Cre recombinase was transiently transfected into six independent clones. With the 5′ and 3′ loxP vectors flanking the same Msh2 allele, recombination between loxP sites gives rise to deletion of the complete Msh2 gene plus the tk gene and hence confers Ganciclovir resistance. If the 5′ and 3′ loxP vectors were present on different Msh2 alleles, Cre-mediated interchromosomal recombination would retain the tk gene. Therefore, Ganciclovir resistance indicates the deletion of the floxed Msh2 allele. Of the six clones tested, three gave rise to Ganciclovir-resistant colonies; in each of these colonies the presence of a single loxP site was confirmed by sequencing PCR products obtained using primers adjacent to Msh2 (arrows). (I) Msh2^+/∆ mESCs were electroporated again with the 5′ loxP vector to introduce a puromycin-resistance gene in front of the remaining Msh2 allele. (J) Puromycin-resistant colonies were analyzed by Southern blot analysis for integration of the 5′ loxP vector into its target site. The genomic DNA was NsiI-digested, and fragments were visualized using a probe (indicated by a yellow star). (K) Southern blot showing the wild-type 9.9-kb fragment in unmodified Msh2^+/∆ cells; the presence of a 20.4-kb fragment indicates correct integration of the 5′ loxP-Pur vector (Msh2^+PUR/∆). The 5′ loxP vector may integrate adjacent to either the Msh2⁺ or the Msh2^∆ allele. However, only integration into the Msh2⁺ allele generates a 20.4-kb fragment, because the Msh2^∆ allele has lost the DNA sequence that could bind the probe.

Fig. 2.

Identification of pathogenic Msh2 variants by sequencing 6TG-resistant colonies. (A) Proven pathogenic mutations in the proof-of-principle study. (B) Proven nonpathogenic variants in the proof-of-principle study. (C) Newly identified pathogenic mutations among 59 tested VUS. The “Variant” and “Nucleotide change” columns display the amino acid change as well as the location and the one- or two-base change introduced by either sense (upper bars) or antisense (lower bars) ssODNs. The bars in the “Sequenced colonies carrying mutation” column illustrate the number of 6TG/puromycin-resistant colonies sequenced per sense or antisense ssODN tested. We always aimed to sequence 12 colonies unless fewer survived the 6TG/puromycin selection. Each box in the bars represents one sequenced colony. Gray boxes represent colonies carrying the mutation of interest; white boxes represent colonies in which the wild-type Msh2 sequence was maintained. NA indicates the targeting was not performed.

Fig. S2.

Sequences of pathogenic mutations detected in the proof-of-principle study. One-letter amino acid codes are given below the nucleotide sequences.

Fig. S3.

Sequences of the detected Msh2 variants. Nucleotide and one-letter amino acid codes are noted.

Fig. 3.

6TG toxicity in Msh2 mutant mESCs. The colony-forming capacity of the detected pathogenic Msh2 variant cell lines, Msh2^+PUR/∆ mESCs, and the Msh2^P622L/Δ pathogenic control was determined in response to increasing doses of 6TG. The 6TG tolerance of mutant cell lines should be compared with the MMR-proficient Msh2^+PUR/∆ and MMR-deficient Msh2^P622L/Δ mESCs in the same experiment because slight differences in the 6TG concentrations result in small interexperiment variances.

Fig. S4.

MNNG clonogenic survival assay. The colony-forming capacity of the detected Msh2 mutant mESCs and the Msh2^+PUR/∆ and Msh2^P622L/Δ control cell lines was determined in response to increasing doses of the methylating agent MNNG. Msh2^+PUR/∆ cells appeared to die at slightly different MNNG doses in various experiments. This variance was caused by small differences in MNNG concentrations in the experiments. Hence the MNNG tolerance of the variant cells lines should be compared with the MMR-proficient Msh2^+PUR/∆ and MMR-deficient Msh2^P622L/Δ mESCs in the same experiment.

Fig. 4.

Western blot analysis of detected Msh2 VUS. MSH2, MSH6, and CDK4 levels were analyzed in whole-cell lysates. CDK4 functioned as the loading control. Relative MSH2 levels compared with Msh2^+/+ mESCs are shown as percentages. Msh2^−/− and Msh2^P622L/Δ cells were used as pathogenic controls.

Fig. S5.

Western blot analysis of the Msh2^+PUR/∆ cell line. MSH2 levels in Msh2^+PUR/∆ mESCs were quantified with respect to Msh2^+/+ cells and compared with levels in Msh2^+/∆ and Msh2^−/− mESCs. MSH6 and CDK4 levels were analyzed also. CDK4 functioned as the loading control.

Fig. 5.

MSI analysis of detected Msh2 mutant mESCs. A slippage reporter composed of a neo gene that was rendered out of frame by a (G)₁₀ repeat, was introduced into the Msh2 mutant mESCs. The relative slippage rates could be calculated by the number of cells that became Geneticin-resistant because of a slippage event bringing the neo gene in-frame. Slippage rates were compared with the MMR-proficient Msh2^+PUR/∆ cell line, the MMR-deficient control P622L, and the partially pathogenic control Y165D. Statistical differences were calculated using an unpaired t test with Welch’s correction. Asterisks indicate values significantly higher than those in the MMR-proficient Msh2^+PUR/∆ control: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Fig. S6.

Alignment of human (Upper Rows) and mouse (Lower Rows) MSH2 amino acid sequences demonstrating conservation of studied mutations. Asterisks mark differences between the two sequences. (A) The 12 pathogenic mutations (red) and 10 polymorphisms (green) used in the proof-of-principle study are highlighted in both the human and mouse sequences to demonstrate their conservation. At the yellow-colored residues, both a pathogenic and nonpathogenic mutation was detected. (B) The 59 studied VUS are highlighted, illustrating their conservation between humans and mice. Detected pathogenic variants are colored red, undetected variants are green, and residues in which both detected and undetected variants were identified are yellow.

Fig. S7.

Location of the studied mutations in the MSH2 protein. The MSH2 domains are displayed in different colors (–41) and are annotated according to their amino acid number and change. (A) The location of the variants used in the proof-of-principle study. The 12 pathogenic mutations are indicated by red lines above the MSH2 domains, and the 10 polymorphisms are indicated by green lines below the MSH2 domains. (B) The location of the 59 studied VUS is depicted: Pathogenic mutations are indicated by red lines above the MSH2 domains, and the undetected variants are indicated by green lines below the MSH2 domains.

Fig. S8.

Location of variants V63E, G162R, D603N, G674A, S723F, and G759E in MSH2. The MSH2–MSH6 heterodimer bound to G:T mispaired DNA, taken from Warren et al. (39), is presented in a ribbon diagram. MSH2, lilac; MSH6, gray; DNA, orange/blue/green; ADP, yellow/orange/red/blue. The location of residues Val63, Gly162, Asp603, Gly674, Ser723, and Gly759 are enlarged and colored green to indicate why substitutions may lead to MMR attenuation.