Limits...
Dissection of cell cycle-dependent dynamics of Dnmt1 by FRAP and diffusion-coupled modeling.

Schneider K, Fuchs C, Dobay A, Rottach A, Qin W, Wolf P, Álvarez-Castro JM, Nalaskowski MM, Kremmer E, Schmid V, Leonhardt H, Schermelleh L - Nucleic Acids Res. (2013)

Bottom Line: We find that transient PBD-dependent interaction directly at replication sites is the predominant specific interaction in early S phase (residence time Tres ≤ 10 s).In late S phase, this binding class is taken over by a substantially stronger (Tres ∼22 s) TS domain-dependent interaction at PCNA-enriched replication sites and at nearby pericentromeric heterochromatin subregions.We propose a two-loading-platform-model of additional PCNA-independent loading at postreplicative, heterochromatic Dnmt1 target sites to ensure faithful maintenance of densely methylated genomic regions.

View Article: PubMed Central - PubMed

Affiliation: Department of Biology and Center for Integrated Protein Science, Ludwig Maximilians University Munich (LMU), 82152 Planegg-Martinsried, Germany.

ABSTRACT
DNA methyltransferase 1 (Dnmt1) reestablishes methylation of hemimethylated CpG sites generated during DNA replication in mammalian cells. Two subdomains, the proliferating cell nuclear antigen (PCNA)-binding domain (PBD) and the targeting sequence (TS) domain, target Dnmt1 to the replication sites in S phase. We aimed to dissect the details of the cell cycle-dependent coordinated activity of both domains. To that end, we combined super-resolution 3D-structured illumination microscopy and fluorescence recovery after photobleaching (FRAP) experiments of GFP-Dnmt1 wild type and mutant constructs in somatic mouse cells. To interpret the differences in FRAP kinetics, we refined existing data analysis and modeling approaches to (i) account for the heterogeneous and variable distribution of Dnmt1-binding sites in different cell cycle stages; (ii) allow diffusion-coupled dynamics; (iii) accommodate multiple binding classes. We find that transient PBD-dependent interaction directly at replication sites is the predominant specific interaction in early S phase (residence time Tres ≤ 10 s). In late S phase, this binding class is taken over by a substantially stronger (Tres ∼22 s) TS domain-dependent interaction at PCNA-enriched replication sites and at nearby pericentromeric heterochromatin subregions. We propose a two-loading-platform-model of additional PCNA-independent loading at postreplicative, heterochromatic Dnmt1 target sites to ensure faithful maintenance of densely methylated genomic regions.

Show MeSH
Quantitative FRAP evaluation of GFP and GFP-Dnmt1 constructs. (A) Representative time frames of exemplary half-nucleus FRAP series recorded with spinning disk confocal microscopy. Scale bar: 5 µm. (B) Mean recovery curves displayed for all measured constructs and cell cycle stages. The inset illustrates cycle-dependent kinetics of GFP and GFP-Dnmt1wt alone (a), and in comparison with GFP-Dnmt1Q162E (b), GFP-Dnmt1ΔTS (c) and GFP-Dnmt1Q162E/ΔTS (d). GFP-Dnmt1wt in cells with diffuse localization shows a decreased mobility compared with GFP. The GFP-Dnmt1wt mobility decreases stepwise in early S phase and in late S phase. The mobility of GFP-Dnmt1Q162E in G1 (late G2) and early S cells (diffuse nuclear localization, pooled) is almost identical to GFP-Dnmt1wt G1 (late G2) cells. In late S phase, a moderately increased mobility is observed for both, GFP-Dnmt1Q162E and GFP-Dnmt1ΔTS mutants compared with GFP-Dnmt1wt. Of note, despite comparable overall kinetics, both curves (dark green, dark orange) are clearly different in their shape. In early S phase, the t1/2 of GFP-Dnmt1ΔTS is reduced compared to GFP-Dnmt1wt. GFP-Dnmt1Q162E/ΔTS (all interphase stages, pooled) displays kinetics almost identical to GFP-Dnmt1wt in cells with diffuse localization. For clarity, error bars are omitted here, but shown in Supplementary Figure S6. (C) Half-times of recovery (t1/2) determined for each construct and distribution pattern. Error bars represent SEM.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3643600&req=5

gkt191-F3: Quantitative FRAP evaluation of GFP and GFP-Dnmt1 constructs. (A) Representative time frames of exemplary half-nucleus FRAP series recorded with spinning disk confocal microscopy. Scale bar: 5 µm. (B) Mean recovery curves displayed for all measured constructs and cell cycle stages. The inset illustrates cycle-dependent kinetics of GFP and GFP-Dnmt1wt alone (a), and in comparison with GFP-Dnmt1Q162E (b), GFP-Dnmt1ΔTS (c) and GFP-Dnmt1Q162E/ΔTS (d). GFP-Dnmt1wt in cells with diffuse localization shows a decreased mobility compared with GFP. The GFP-Dnmt1wt mobility decreases stepwise in early S phase and in late S phase. The mobility of GFP-Dnmt1Q162E in G1 (late G2) and early S cells (diffuse nuclear localization, pooled) is almost identical to GFP-Dnmt1wt G1 (late G2) cells. In late S phase, a moderately increased mobility is observed for both, GFP-Dnmt1Q162E and GFP-Dnmt1ΔTS mutants compared with GFP-Dnmt1wt. Of note, despite comparable overall kinetics, both curves (dark green, dark orange) are clearly different in their shape. In early S phase, the t1/2 of GFP-Dnmt1ΔTS is reduced compared to GFP-Dnmt1wt. GFP-Dnmt1Q162E/ΔTS (all interphase stages, pooled) displays kinetics almost identical to GFP-Dnmt1wt in cells with diffuse localization. For clarity, error bars are omitted here, but shown in Supplementary Figure S6. (C) Half-times of recovery (t1/2) determined for each construct and distribution pattern. Error bars represent SEM.

Mentions: In our analysis, we estimate models with different numbers of MCs. Because the models are nested, the inclusion of more MCs always leads to a better or at least equally good fit. However, one may ask whether the additional computational effort for multiple MCs is worth the improved matching of the data. At first glance, model choice criteria like the Akaike information criterion (AIC) (38) seem appropriate. In our application, however, the difference in the mean squared residuals for different models is typically small owing to parameter redundancies. Because of the large number of model parameters, the AIC will often favor less MCs although the curvature of the recovery curves is better described by more complex models. For that reason, we developed a model selection criterion that penalizes complexity less rigorously and is specific to our application. Due to the relatively small noise in the FRAP curves (Figure 3B and Supplementary Figure S6), we do not expect to overfit the data. The criterion reads as follows.


Dissection of cell cycle-dependent dynamics of Dnmt1 by FRAP and diffusion-coupled modeling.

Schneider K, Fuchs C, Dobay A, Rottach A, Qin W, Wolf P, Álvarez-Castro JM, Nalaskowski MM, Kremmer E, Schmid V, Leonhardt H, Schermelleh L - Nucleic Acids Res. (2013)

Quantitative FRAP evaluation of GFP and GFP-Dnmt1 constructs. (A) Representative time frames of exemplary half-nucleus FRAP series recorded with spinning disk confocal microscopy. Scale bar: 5 µm. (B) Mean recovery curves displayed for all measured constructs and cell cycle stages. The inset illustrates cycle-dependent kinetics of GFP and GFP-Dnmt1wt alone (a), and in comparison with GFP-Dnmt1Q162E (b), GFP-Dnmt1ΔTS (c) and GFP-Dnmt1Q162E/ΔTS (d). GFP-Dnmt1wt in cells with diffuse localization shows a decreased mobility compared with GFP. The GFP-Dnmt1wt mobility decreases stepwise in early S phase and in late S phase. The mobility of GFP-Dnmt1Q162E in G1 (late G2) and early S cells (diffuse nuclear localization, pooled) is almost identical to GFP-Dnmt1wt G1 (late G2) cells. In late S phase, a moderately increased mobility is observed for both, GFP-Dnmt1Q162E and GFP-Dnmt1ΔTS mutants compared with GFP-Dnmt1wt. Of note, despite comparable overall kinetics, both curves (dark green, dark orange) are clearly different in their shape. In early S phase, the t1/2 of GFP-Dnmt1ΔTS is reduced compared to GFP-Dnmt1wt. GFP-Dnmt1Q162E/ΔTS (all interphase stages, pooled) displays kinetics almost identical to GFP-Dnmt1wt in cells with diffuse localization. For clarity, error bars are omitted here, but shown in Supplementary Figure S6. (C) Half-times of recovery (t1/2) determined for each construct and distribution pattern. Error bars represent SEM.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3643600&req=5

gkt191-F3: Quantitative FRAP evaluation of GFP and GFP-Dnmt1 constructs. (A) Representative time frames of exemplary half-nucleus FRAP series recorded with spinning disk confocal microscopy. Scale bar: 5 µm. (B) Mean recovery curves displayed for all measured constructs and cell cycle stages. The inset illustrates cycle-dependent kinetics of GFP and GFP-Dnmt1wt alone (a), and in comparison with GFP-Dnmt1Q162E (b), GFP-Dnmt1ΔTS (c) and GFP-Dnmt1Q162E/ΔTS (d). GFP-Dnmt1wt in cells with diffuse localization shows a decreased mobility compared with GFP. The GFP-Dnmt1wt mobility decreases stepwise in early S phase and in late S phase. The mobility of GFP-Dnmt1Q162E in G1 (late G2) and early S cells (diffuse nuclear localization, pooled) is almost identical to GFP-Dnmt1wt G1 (late G2) cells. In late S phase, a moderately increased mobility is observed for both, GFP-Dnmt1Q162E and GFP-Dnmt1ΔTS mutants compared with GFP-Dnmt1wt. Of note, despite comparable overall kinetics, both curves (dark green, dark orange) are clearly different in their shape. In early S phase, the t1/2 of GFP-Dnmt1ΔTS is reduced compared to GFP-Dnmt1wt. GFP-Dnmt1Q162E/ΔTS (all interphase stages, pooled) displays kinetics almost identical to GFP-Dnmt1wt in cells with diffuse localization. For clarity, error bars are omitted here, but shown in Supplementary Figure S6. (C) Half-times of recovery (t1/2) determined for each construct and distribution pattern. Error bars represent SEM.
Mentions: In our analysis, we estimate models with different numbers of MCs. Because the models are nested, the inclusion of more MCs always leads to a better or at least equally good fit. However, one may ask whether the additional computational effort for multiple MCs is worth the improved matching of the data. At first glance, model choice criteria like the Akaike information criterion (AIC) (38) seem appropriate. In our application, however, the difference in the mean squared residuals for different models is typically small owing to parameter redundancies. Because of the large number of model parameters, the AIC will often favor less MCs although the curvature of the recovery curves is better described by more complex models. For that reason, we developed a model selection criterion that penalizes complexity less rigorously and is specific to our application. Due to the relatively small noise in the FRAP curves (Figure 3B and Supplementary Figure S6), we do not expect to overfit the data. The criterion reads as follows.

Bottom Line: We find that transient PBD-dependent interaction directly at replication sites is the predominant specific interaction in early S phase (residence time Tres ≤ 10 s).In late S phase, this binding class is taken over by a substantially stronger (Tres ∼22 s) TS domain-dependent interaction at PCNA-enriched replication sites and at nearby pericentromeric heterochromatin subregions.We propose a two-loading-platform-model of additional PCNA-independent loading at postreplicative, heterochromatic Dnmt1 target sites to ensure faithful maintenance of densely methylated genomic regions.

View Article: PubMed Central - PubMed

Affiliation: Department of Biology and Center for Integrated Protein Science, Ludwig Maximilians University Munich (LMU), 82152 Planegg-Martinsried, Germany.

ABSTRACT
DNA methyltransferase 1 (Dnmt1) reestablishes methylation of hemimethylated CpG sites generated during DNA replication in mammalian cells. Two subdomains, the proliferating cell nuclear antigen (PCNA)-binding domain (PBD) and the targeting sequence (TS) domain, target Dnmt1 to the replication sites in S phase. We aimed to dissect the details of the cell cycle-dependent coordinated activity of both domains. To that end, we combined super-resolution 3D-structured illumination microscopy and fluorescence recovery after photobleaching (FRAP) experiments of GFP-Dnmt1 wild type and mutant constructs in somatic mouse cells. To interpret the differences in FRAP kinetics, we refined existing data analysis and modeling approaches to (i) account for the heterogeneous and variable distribution of Dnmt1-binding sites in different cell cycle stages; (ii) allow diffusion-coupled dynamics; (iii) accommodate multiple binding classes. We find that transient PBD-dependent interaction directly at replication sites is the predominant specific interaction in early S phase (residence time Tres ≤ 10 s). In late S phase, this binding class is taken over by a substantially stronger (Tres ∼22 s) TS domain-dependent interaction at PCNA-enriched replication sites and at nearby pericentromeric heterochromatin subregions. We propose a two-loading-platform-model of additional PCNA-independent loading at postreplicative, heterochromatic Dnmt1 target sites to ensure faithful maintenance of densely methylated genomic regions.

Show MeSH