D-type cyclins regulate DNA mismatch repair in the G1 and S phases of the cell cycle, maintaining genome stability

Abstract

The large majority of oxidative DNA lesions occurring in the G1 phase of the cell cycle are repaired by base excision repair (BER) rather than mismatch repair (MMR) to avoid long resections that can lead to genomic instability and cell death. However, the molecular mechanisms dictating pathway choice between MMR and BER have remained unknown. Here, we show that, during G1, D-type cyclins are recruited to sites of oxidative DNA damage in a PCNA- and p21-dependent manner. D-type cyclins shield p21 from its two ubiquitin ligases CRL1^SKP2 and CRL4^CDT2 in a CDK4/6-independent manner. In turn, p21 competes through its PCNA-interacting protein degron with MMR components for their binding to PCNA. This inhibits MMR while not affecting BER. At the G1/S transition, the CRL4^AMBRA1-dependent degradation of D-type cyclins renders p21 susceptible to proteolysis. These timely degradation events allow the proper binding of MMR proteins to PCNA, enabling the repair of DNA replication errors. Persistent expression of cyclin D1 during S-phase increases the mutational burden and promotes microsatellite instability. Thus, the expression of D-type cyclins inhibits MMR in G1, whereas their degradation is necessary for proper MMR function in S.

Figure 1.. D-type cyclins are recruited to DNA lesions in a p21 dependent manner.

A) U-2OS cells stably expressing EGFP-tagged cyclins, EGFP, and mPlum-PCNA were pre-sensitized with BrdU for 24 hours and subjected to 405 nm laser induced damage. DNA damage recruitment dynamics were captured by live cell imaging. Non-S phase cells (upper panel) and S phase cells (bottom panel) were identified based on the presence of PCNA foci. Relative fluorescence values and images were acquired every 5 seconds for 4 minutes. For each condition, ≥25 cells were evaluated from 3 independent experiments. Mean relative fluorescence values and standard errors were plotted against time. Representative images are shown in Fig. S1A. Times are indicated in seconds. B) U-2OS cells stably expressing either mAzGreen-cyclin D1 or NLS-mAzGreen-cyclin D1 (in green) together with mPlum-PCNA (in red) were transfected with siRNAs targeting p21, p27, p57 or a non-targeting control (NT). Cells were treated and analyzed as in (A). Representative images are shown below the graphs. C) Parental (p21^+/+) or p21^−/− U-2OS cells stably expressing mAzGreen-cyclin D1, D2, or D3 and mPlum-PCNA were treated and analyzed as in (A). D) Left: HEK 293T cells were co-transfected with the indicated FFSS-cyclin D1 constructs together with either p21-HA or EV. Cyclin D1(MV) denotes mutations M56A and V60A. Cyclin D1(HP) denotes mutations M56A, V60A, and W63A. Cell lysates were immunoprecipitated with an anti-FLAG resin, followed by elution using 3x FLAG peptide. Immunoprecipitates were probed with the indicated antibodies. Whole cell extract (WCE) is of the EV sample. Middle and right panels: U-2OS cells stably expressing the indicated mAzGreen-cyclin D1 constructs and mPlum-PCNA were treated and analyzed as in (A). E) Confocal images of U-2OS cells fixed 2 minutes after laser micro-irradiation and stained for either the DNA damage marker γH2A.X, the replication marker EdU and the indicated D-type cyclins. DAPI was used to counterstain nuclear DNA. A white dashed line denotes the border of each nucleus. Scale bar represents 5 μm. F) G1-synchronized RPE1 cells harboring fluorescent-tagged endogenous proteins (mRuby-PCNA, mVenus-cyclin D1 and p21-mTurquoise2) were treated and analyzed as in (A). Representative images are shown in the middle panel. The efficiency of depletion is shown in the right panel.

Figure 2.. D-type cyclins are recruited to oxidative DNA lesions in the proximity of BER components.

A) G1-synchronized U-2OS and RPE1 cells were treated with the indicated DNA damaging agents. Cells were then fractionated into soluble and chromatin fractions, and lysates were immunoblotted as indicated. The figure shows the chromatin fractions while Fig. S2A shows the soluble fractions. B) G1-synchronized parental (p21^+/+) and p21^−/− U-2OS cells were treated with 100 μM H₂O₂ for 20 minutes. Cells were then fractionated into soluble and chromatin fractions, and immunoblotted as indicated. C) Schematic of the site-specific DNA damage recruitment assay. 2–6-3 U-2OS reporter cells containing lac operator (LacO) repeats integrated into their genome were transfected with vectors expressing either mKillerRed-LacI or mCherry-LacI-FokI to generate local DNA oxidization and DSBs, respectively. mCherry-LacI and a nuclease FokI(D540A) inactive mutant were used as negative controls. D) 2–6-3 U-2OS cells were transfected with the indicated constructs and immunostained for γH2A.X (green) and endogenous cyclin D1 (grey). The latter was detected using the SP4 rabbit monoclonal antibody. White arrowheads indicate the position of the LacO (red) in the nucleus, which is outlined by a white dashed line. Quantification of the cyclin D1 relative mean fluorescence intensity (RMFI) was carried out from 3 independent experiments as in and visualized on violin plots. Dashed lines represent the median on the plots. Scale bar represents 2 μm. E) 2–6-3 U-2OS cells were co-transfected with mAzGreen-cyclin D1 (grey) and the indicated constructs, and subsequently immunostained for γH2A.X (green). White arrowheads indicate the position of the LacO (red) in the nucleus, which is outlined by white dashed line. Quantification of the mAzGreen-cyclin D1 relative mean fluorescence intensity (RMFI) was carried out from 3 independent experiments and visualized on violin plots. Dashed lines represent the median on the plots. Scale bar represents 2 μm. F) 2–6-3 U-2OS cells were transfected with the indicated constructs and siRNAs targeting p21 or a non-targeting control (NT), and then immunostained for γH2A.X (green) and cyclin D1 (grey). The latter was detected and quantified as in (D). G) RPE1 cells harboring fluorescent-tagged endogenous proteins (mRuby-PCNA, mVenus-cyclin D1, and p21-mTurquoise2) were micro-irradiated as described in . DNA damage recruitment dynamics were captured by live-cell imaging. Non-S phase cells were identified based on the absence of PCNA foci and analyzed as in Fig. 1A. H) Scatter plot of fold-changes in PSM (peptide spectral match) counts of H₂O₂-treated vs. non-treated cells for the control TurboID or TurboID fused to cyclin D1 expressed in U-2OS cells. Raw values are reported in Supplementary Table S1. Known DNA repair proteins are highlighted in purple, except BER proteins that are in red. Selected proteins must have a fold-change greater than or equal to 2 in the case of cyclin D1 and not have a fold-change greater than or equal to 2 in the case for TurboID on its own. Within these selected proteins, only the BER pathway was significantly enriched (p-value = 0.005711, Fisher's exact test) among the DNA repair pathways (BER, NER, MMR, HRR, NHEJ, FAN DR, TLS, and DDA).

Figure 3.. D-type cyclins inhibit the binding of MMR proteins to PCNA.

A) Scatter plot of fold-changes in PSM (peptide spectral match) counts of H₂O₂-treated vs. non-treated cells of TurboID-PCNA expressed in RPE1 cells. Cells were transfected with either non-targeting (siNT) or siRNA targeting D-type cyclins (siCCNDs) and subsequently synchronized in G1. Selected proteins must have a fold-change greater than two-fold in both cases, but to be considered enriched upon siCCNDs, the fold-changes must be at least twice as high in the case of siCCNDs compared to siNT. Raw values are reported in Supplementary Table S2. B) RPE1 cells were transfected with either non-targeting (siNT) or siRNA targeting D-type cyclins or KIPs, and subsequently synchronized in G1. Cells were then treated with 100 μM H₂O₂ before fractionated into soluble and chromatin fractions. PCNA was immunoprecipitated from the chromatin fraction and co-purified proteins were immunoblotted as indicated (left panel). Right panels show chromatin and soluble fractions. Asterix denotes a non-specific band. C) G1-synchronized parental (AMBRA1^+/+) or AMBRA1^−/− RPE1 cells were treated with 100 μM H₂O₂ before fractionated into soluble and chromatin fractions. PCNA was immunoprecipitated from the chromatin fraction and co-purified proteins were immunoblotted as indicated (left panel). Right panels show chromatin and soluble fractions. D) U-2OS cells stably expressing mCherry-tagged BER pathway components or AcGFP-tagged MMR pathway components were transfected with either non-targeting or siRNA targeting D-type cyclins, and subsequently synchronized in G1. Each row in the heatmap shows recruitment of a BER or MMR protein at micro-irradiation sites that is higher (red) or lower (blue) in D-type cyclin depleted cells when compared to control cells. Columns represent standardized recruitment values that were averaged over 15 second intervals. Column label units refer to seconds post-irradiation. Hierarchical clustering of rows in the heatmap was performed with Euclidean distance measure and Ward clustering method. The table on the right shows if the protein of interest is part of the BER or MMR pathway. The “p” column indicates if the mean intensity of the normalized difference between control and D-type cyclin-depleted cells was significant over a window of 60 seconds across the entire 10 minutes (+: p < 0.01). E) U-2OS cells stably expressing AcGFP-tagged MSH2, MSH3, and MSH6 were transfected with siRNAs targeting D-type cyclins, KIPs or a non-targeting control (NT), and subsequently synchronized in G1. DNA damage recruitment dynamics were captured by live-cell imaging. Relative fluorescence values and images were acquired every 5 seconds for 10 minutes. Mean relative fluorescence values and standard errors were plotted against time. Times are indicated in seconds.

Figure 4.. D-type cyclin levels promote the repair of oxidative DNA lesions during G1.

A) Two CCND1^−/−; CCND2^−/− (D1/D2^−/−) U-2OS clones (clones #KO1 and #KO2) and parental U-2OS cells (D1/D2^+/+) were transfected with siRNAs targeting D-type cyclins, cyclin D3, or a non-targeting control (siNT), and subsequently synchronized in G1. Cells were then treated with H₂O₂ for 15 minutes or left untreated (UT). Half of the samples treated with H₂O₂ were allowed to recover in media without H₂O₂ for 4 hours (Recov.), after which all samples were subjected to comet assay. Comet tail moment values for were normalized to the siNT “H₂O₂” samples. Lines represent mean values on the plots from 4 independent experiments. The efficiency of depletion for cyclin D1, cyclin D2, and cyclin D3 is shown on the right panel. B) U-2OS cells were transfected with siRNAs targeting KIPs, or a non-targeting control (siNT), and subsequently synchronized in G1. Cells were then treated with H₂O₂ for 15 minutes or left untreated (UT). Half of the samples treated with H₂O₂ were allowed to recover in media without H₂O₂ for 4 hours (Recov.), after which all samples were subjected to comet assay. Where indicated, cells were pretreated for 2 hours with palbociclib (CDK4/6i) and kept in palbociclib during the 4 hours recovery. Comet tail moment values were normalized to the siNT “H₂O₂” samples. Lines represent mean values on the plots from 3 independent experiments. The efficiency of p21, p27, p57 depletion and CDK4/6 inhibition is shown on the right panel. C) RPE1 cells were transfected with either siRNAs targeting the indicated proteins or a non-targeting siRNA control (siNT), and subsequently synchronized in G1. Where indicated, cells were treated with H₂O₂ for 30 minutes and then allowed to recover for 1.5 hours in DMEM supplemented with EdU. The EdU signal was normalized to the untreated siNT samples. Measurements were carried out from 3 independent experiments and visualized on violin plots. Dashed lines represent the median on the plots. The efficiency of the siRNA knockdowns is shown using immunoblotting on the right. D) RPE1 cells were transfected with either siRNAs targeting D-type cyclins, KIPs, MLH1, MSH2, or a non-targeting siRNA control (siNT), and subsequently synchronized in G1. Cells were then treated and evaluated as in (C). E) RPE1 cells were transfected with siRNAs targeting D-type cyclins, KIPs, or a non-targeting control (siNT) and serum-starved to arrest them in G0. Cells were then nucleoporated with the indicated FM-HCR reporter plasmids and released into serum containing media. Ten hours later, cells, now in G1, were subjected to FACS analysis. DNA repair activity for each siRNA was normalized to siNT and was represented as normalized repair activity. Graphs show average and standard deviation from at least 3 independent experiments. Schematic of the transcriptional mutagenesis-based fluorescent reporters is shown next to the graphs. The base pairs shown correspond to sites that code for a key amino acid of the chromophores of the fluorescent proteins. The transcribed strand is on the top. The efficiency of the siRNA knock-downs is shown in the right panel. F) Parental (AMBRA1^+/+) and AMBRA1^−/− RPE1 cells (clones #KO1 and #KO2) were treated and analyzed as in (E).

Figure 5.. D-type cyclins stabilize p21, which competes with MMR proteins to bind PCNA.

A) Parental (AMBRA1^+/+) and AMBRA1^−/− RPE1 cells (clones #KO1 and #KO2) were treated with cycloheximide (CHX) and MG132 as indicated, and lysates were blotted with the indicated antibodies. The graph at the bottom shows the quantification of p21 levels from three independent experiments. Error bars indicate standard error. B) AMBRA1^−/− RPE1 cells (clones #KO1 and #KO2) were transfected with siRNAs targeting D-type cyclins, or a non-targeting control (siNT) before being treated with cycloheximide (CHX) and MG132 as indicated. Lysates were blotted with the indicated antibodies. The graphs on the right show the quantification of p21 levels from three independent experiments. Error bars indicate standard error. C) RPE1 cells were transfected with siRNAs targeting D-type cyclins or a non-targeting control (siNT), and subsequently synchronized in G1. Cells were then treated with cycloheximide (CHX) and either MG132 or MLN4924 as indicated. Lysates were blotted with the indicated antibodies. The graph on the right shows the quantification of p21 levels from three independent experiments. Error bars indicate standard error. D) HEK 293T cells were transfected with either an empty vector (EV) or HA-p21, together with increasing amounts of FFSS-cyclin D1, FFSS-cyclin D2, or FFSS-cyclin D3 vectors. p21 was immunoprecipitated from the lysates using HA-beads and co-purified proteins were immunoblotted as indicated. E) Stably transduced RPE1 cells harboring a doxycycline-inducible HA-p21, HA-p21(ΔPCNA), or EV were synchronized into G0 using serum starvation, during which the transgenes were induced by doxycycline. After serum release, G1 phase cells were treated with 100 μM H₂O₂ before fractionating them into soluble and chromatin fractions. PCNA was immunoprecipitated from the chromatin fraction and co-purified proteins were immunoblotted as indicated (left panel). Right panels show chromatin and soluble fractions.

Figure 6.. D-type cyclin degradation in S phase is necessary to limit mutational burden and microsatellite instability.

A) S phase-synchronized parental (AMBRA1^+/+) or AMBRA1^−/− RPE1 cells (clones #KO1 and #KO2) were fractionated into soluble and chromatin fractions. PCNA was immunoprecipitated from the chromatin fraction and co-purified proteins were immunoblotted as indicated (left panel). Right panels show chromatin and soluble fractions. Asterix denoted a non-specific band. B) S phase-synchronized parental (AMBRA1^+/+) or AMBRA1^−/− RPE1 cells (clones #KO1 and #KO2) were nucleoporated with the indicated FM-HCR reporter plasmids. Eight hours later, the cells were subjected to FACS analysis. DNA repair activity for each sample was normalized to AMBRA1^+/+ cells and is represented as normalized repair activity. Graphs show average and standard deviation from at least 3 independent experiments. C) DNA damage recruitment dynamics were captured by live-cell imaging in S phase-synchronized parental (AMBRA1^+/+) or AMBRA1^−/− U-2OS cells (clones #KO1 and #KO2) stably expressing AcGFP-tagged MSH2, MSH3 and MSH6. Relative fluorescence values and images were acquired every 5 seconds for 10 minutes. Mean relative fluorescence values and standard errors were plotted against time. Times are indicated in seconds. D) S phase-synchronized parental (AMBRA1^+/+) or AMBRA1^−/− RPE1 cells (clones #KO1 and #KO2) were treated with 100 μM H₂O₂ before being fractionated into soluble and chromatin fractions. PCNA was immunoprecipitated from the chromatin fraction and co-purified proteins were immunoblotted as indicated (left panel). Right panels show chromatin and soluble fractions. Asterix denoted a non-specific band. E) Schematic representation of the CherryOFF reporter assay. The CherryOFF reporter expresses two fluorescence proteins, mCherry and EGFP, driven by a CMV promoter (blue) and separated with an internal ribosome entry site (IRES). The mCherry contains a point mutation, converting its Trp98 residue (TGG) into a premature STOP codon (TGA). The resulting protein does not have fluorescence. Only a reversion of A to G in the sense strand or a T to C reversion in the antisense strand restores a functional mCherry ORF, allowing its expression. F) Quantification of the mutational frequencies in U-2OS (left) and MCF10Am (right) cells measured with the mCherryOFF reporter. mCherry signal was evaluated 18 days after infecting the cells with lentiviruses expressing either the stable cyclin D1(T286A) mutant, wild-type cyclin D1, or an empty vector (EV). Graphs show average and standard error from at least 3 independent experiments. Cyclin D1 and p21 levels are shown in the panels below the graphs. G) Schematic representation of the (CA)₁₈-NanoLuc reporter assay. The CMV promoter-driven (blue) NanoLuc enzyme is out-of-frame due to a microsatellite region, composed of 18 (CA)-dinucleotide repeats. In the absence of proper MMR, random mutations that introduce frameshifts in this region can restore the NanoLuc ORF, leading to luminescence. H) Graphs show NanoLuc signal in U-2OS (left) and MCF10Am (right) cells normalized to cell viability. NanoLuc signal was evaluated 144 hours after infecting the cells with lentiviruses expressing either the stable cyclin D1(T286A) mutant, wild-type cyclin D1, or an empty vector (EV). Graphs show average and standard deviation from at least 3 independent experiments. Cyclin D1 and p21 levels are shown in the panels next to the graphs.

Figure 7.. Schematics representing how D-type cyclins regulate MMR and genome stability during the G1 and S phases of the cell division cycle.