Proteasome inhibition rescues clinically significant unstable variants of the mismatch repair protein Msh2

Abstract

MSH2 is required for DNA mismatch repair recognition in eukaryotes. Deleterious mutations in human MSH2 account for approximately half of the alleles associated with a common hereditary cancer syndrome. Previously, we characterized clinically identified MSH2 missense mutations, using yeast as a model system, and found that the most common cause of defective DNA mismatch repair was low levels of the variant Msh2 proteins. Here, we show that increased protein turnover is responsible for the reduced cellular levels. Increasing gene dosage of more than half of the missense alleles fully restored function. A titration experiment revealed that raising the expression level of one variant to less than wild-type levels restored mismatch repair, suggesting that overexpression is not always required to regain function. We found that the ubiquitin-mediated proteasome degradation pathway is the major mechanism for increased turnover of the Msh2 variants and identified the primary ubiquitin ligase as San1. Deletion of San1 restored protein levels for all but one variant, but did not elevate wild-type Msh2 levels. The unstable variants interacted with San1, whereas wild-type Msh2 did not. Additionally, san1Δ suppressed the mismatch repair defect of unstable variants. Of medical significance, the clinically approved drug Bortezomib partially restored protein levels and mismatch repair function for low-level variants and reversed the resistance to cisplatin, a common chemotherapeutic. Our results provide the foundation for an innovative therapeutic regime for certain mismatch-repair-defective cancers that are refractory to conventional chemotherapies.

Fig. 1.

Low-level Msh2 variants are unstable and degraded by the proteasome. (A) msh2∆ (AGY75) bearing pRS423 (VEC), pGAL–MSH2 (WT), or pGAL–MSH2–A618V (A618V) was used to permit controlled expression of Msh2. Cells were harvested at the indicated time after repression of MSH2. Throughout, protein extracts were subjected to immunoblotting methods to detect Msh2 (Upper) and Kar2 or PGK (Lower). (B) pre1-1 pre2-2 (MY11719) bearing pGAL–MSH2–G711D was grown to exponential phase at 23 °C, and the culture was split and grown at 23 °C and 37 °C for 3 h. The Msh2 variant (Msh2^G711D) and a loading control (PGK) are indicated. (C) pre1-1 pre2-2 strain and AGY227, a wild-type (WT) strain harboring pGAL–MSH2–G711D, were grown to early exponential phase at 30 °C in galactose-containing medium and shifted to 37 °C for 30 min. Glucose (2%) was added to each culture to repress synthesis and harvested at the indicated times. The Msh2 variant (Msh2^G711D) and a loading control (PGK) are shown. (D) Turnover experiments using an msh2Δ erg6Δ strain (MY11331) expressing GAL10–msh2–R542P were conducted in the presence of 50 μM MG132 (▲) to inhibit the proteasome, 1 mM PMSF to inhibit vacuolar proteases (■), or DMSO, the solvent for MG132 (♦). Time points after repression of synthesis and chemical addition were processed for immunoblotting. The data were normalized for growth and are expressed as a percentage of the zero time point (% Msh2^R542P). (E) msh2Δ cells from the yeast deletion collection (131-B-1) encoding the vector (VEC), the low-copy centromere-based plasmid expressing msh2–A618V from the endogenous promoter (CEN), or the high-copy (2μ) plasmid expressing msh2–A618V from the GAL10 promoter were grown overnight in selectable medium containing 2% galactose. Protein extracts were immunoprecipitated, processed for immunoblotting, and probed with α-ubiquitin antibody (α-Ubq). The Ig band is indicated. The membrane was stripped and reprobed to detect the Msh2 variant (Msh2^A618V).

Fig. 2.

Increasing or restoring expression rescues the mismatch-repair defect of unstable Msh2 variants. (A) An msh2Δ pol3-01 strain kept alive by a URA3-based plasmid expressing wild-type MSH2 was transformed with HIS3 encoding plasmids expressing no MSH2 (vector), endogenously expressed wild-type MSH2 (pMSH2), and variants (pMSH2–allele, e.g., D524Y), as well as high-copy, P_GAL10-driven MSH2 (pGAL–MSH2), and variants (pGAL–MSH2–allele). Stains were grown overnight in medium with galactose (GAL) to activate P_GAL10, with glucose (GLC) to repress P_GAL10, or with raffinose (RAF) that neither activates nor represses P_GAL10. Selective pressure for the wild-type MSH2 URA3 plasmid was removed to allow for plasmid loss. Fivefold serial dilutions were delivered to plates lacking histidine, with 2% galactose and FOA (−HIS+GAL FOA). The 5-FOA selects for cells that were able to lose the MSH2 URA3 plasmid during growth, indicating suppression of the mismatch-repair defect. Proper plating and dilution were confirmed by growth on −HIS medium. (B) Stains were pregrown overnight in medium with GAL to induce expression with varying percentages of GLC to titrate expression. Twofold serial dilutions of the cultures were delivered to −HIS+GAL FOA selective plates and −HIS plates control plates. (C) Aliquots of the cultures described in B were processed for immunoblotting. The data were quantified by using ImageJ open-source software and normalized to wild-type Msh2 levels (fold increase in Msh2). The functional data presented in B are summarized, and the rescue of the mismatch-repair defect (+) and the failure to rescue (−) are denoted above the protein-level data corresponding to the relevant strain and growth condition.

Fig. 3.

San1 is the ubiquitin ligase for the Msh2 low-level variants. (A) Strains encoding knockouts of ubiquitin ligases (san1Δ, pib1Δ, doa10Δ) or BY4742, the wild-type strain (WT) expressing Msh2^G711D, were transformed with the P_CUP1–UBI4 plasmid (27) or the vector (LYS2 2μ) to allow for ubiquitin overexpression (Ubq OE) (+) or no overexpression (−), respectively, when treated with 50 μM CuSO₄ for 3 h before harvesting for immunoblotting. Protein extracts were probed to detect Msh2^G711D and a loading control (PGK). The blot was stripped and reprobed with α-ubiquitin (Ubq). (B) A san1Δ strain and the isogenic wild-type (SAN1) expressing Msh2^G350R were induced with 2% galactose for 30 min and repressed with 2% glucose for the times indicated. Msh2 and a loading control (PGK) are indicated. (C) Temperature-sensitive proteasome mutant strains (pre1-1 pre2-2) with wild-type SAN1 (MY11719) or with a deletion of san1, san1Δ (MY12516) were transformed with pGAL–MSH2–A618V. The variant protein was overexpressed for 2 h at 37 °C to enrich for ubiquitinated forms of the protein. The immunoprecipitated proteins were processed for immunoblotting and probed with α-ubiquitin antibody (Ubq). The membrane was stripped and reprobed to detect Msh2^A618V. (D) Yeast two-hybrid strain PJ69-4A was transformed with pGBD-C2 (no Msh2), pGBD–MSH2 (Msh2), and all of the pGBD–MSH2 low-level variant plasmids including pGBD–MSH2–T347I (Msh2^T347I) shown. Yeast two-hybrid strain PJ69-4α was transformed with pGAD–SAN1–C279S (San1^C297S). Diploids were formed and selected for growth (Growth) on medium lacking leucine and tryptophan (−LEU−TRP) and on selective medium also lacking histidine (−LEU−TRP−HIS) for the yeast two-hybrid interaction (Interaction). (E) A strain with a deletion of the SAN1 ubiquitin ligase (−) and the isogenic wild-type strain (+) expressing Msh2 or variants from a centromere-based plasmid were processed for immunoblotting. (F) MY12336, an msh2Δ san1Δ (san1Δ), and MY12064, an msh2Δ with wild-type SAN1 (SAN1), were transformed with the vector (msh2Δ), with pMSH2 (MSH2), and with plasmids to express the msh2 missense substitutions (msh2–D524Y and –A618V are shown). Equally grown patches of cells from the strains were replica-printed to medium with canavanine (CAN) to determine qualitatively the mutation frequency in the CAN1 forward mutation assay. The cells were also transferred to medium lacking methionine (−MET) to select for cells that had undergone a reversion at the hom3-10 polynucleotide run. (G) The mutation rates for the strain described above were conducted by using the CAN1 forward mutation assay and fluctuation assays. The mutation rates and 95% confidence limits (error bars) are shown.

Fig. 4.

The clinically approved proteasome inhibitor, Bortezomib, rescues the mismatch-repair defect of low-level Msh2 variants. (A) MY12078 (pdr5Δ erg6Δ msh2Δ) was transformed with centromere-based plasmids expressing msh2–D524Y or wild-type MSH2. Cultures were grown to midexponential phase and then grown in the absence (DMSO) or presence of Bortezomib in DMSO at the indicated dose for 3 h and processed for immunoblotting. Msh2^D524Y levels were quantified by using ImageJ and are expressed as a percentage of the wild-type Msh2 levels (% WT Msh2 protein level). Error bars indicate SEM of replicate experiments. (B) Wild-type (WT), pdr5Δ, erg6Δ, and pdr5Δ erg6Δ were tested for sensitivity to Bortezomib. Approximately 10⁴ cells were plated, and filter discs were placed on the plates. Bortezomib (in DMSO) or DMSO was applied to the disk at the indicated concentrations. (C) MY12078 (pdr5Δ erg6Δ msh2Δ) was transformed with centromere-based plasmids expressing no MSH2 (msh2Δ), msh2–D524Y, msh2–A618V, or wild-type MSH2. Cultures were grown in the absence (DMSO) or presence of 5 μM Bortezomib in DMSO overnight and processed to calculate the CAN1 forward mutation rate as described in Materials and Methods. Error bars indicate 95% confidence limits. (D) msh2–A618V or wild-type MSH2 cultures from above were grown in a range of cisplatin concentrations (400 µM shown) in the absence or presence of 1 μM Bortezomib overnight in a microtiter dish, and the optical density at 600 nm was detected every 15 min for 48 h.