Click here to close
Hello! We notice that you are using Internet Explorer, which is not supported by Xenbase and may cause the site to display incorrectly.
We suggest using a current version of Chrome,
FireFox, or Safari.
Cell Syst
2020 Dec 16;116:653-662.e8. doi: 10.1016/j.cels.2020.11.003.
Show Gene links
Show Anatomy links
H4K20 Methylation Is Differently Regulated by Dilution and Demethylation in Proliferating and Cell-Cycle-Arrested Xenopus Embryos.
Schuh L
,
Loos C
,
Pokrovsky D
,
Imhof A
,
Rupp RAW
,
Marr C
.
???displayArticle.abstract???
DNA replication during cell division leads to dilution of histone modifications and can thus affect chromatin-mediated gene regulation, raising the question of how the cell-cycle shapes the histone modification landscape, particularly during embryogenesis. We tackled this problem by manipulating the cell cycle during early Xenopus laevis embryogenesis and analyzing in vivo histone H4K20 methylation kinetics. The global distribution of un-, mono-, di-, and tri-methylated histone H4K20 was measured by mass spectrometry in normal and cell-cycle-arrested embryos over time. Using multi-start maximum likelihood optimization and quantitative model selection, we found that three specific biological methylation rate constants were required to explain the measured H4K20 methylation state kinetics. While demethylation is essential for regulating H4K20 methylation kinetics in non-cycling cells, demethylation is very likely dispensable in rapidly dividing cells of early embryos, suggesting that cell-cycle-mediated dilution of H4K20 methylation is an essential regulatory component for shaping its epigenetic landscape during early development. A record of this paper's transparent peer review process is included in the Supplemental Information.
Figure 1H4K20 Methylation Kinetics during Xenopus Embryogenesis Are Altered upon HUA-Induced Cell-Cycle Arrest
(A) Xenopus eggs are fertilized in vitro at time point 0. For the next 5 hpf, the embryonic cell-cycle consists of S and M phases only. At 5.5 hpf, G1 and G2 phases appear. At 11 hpf, half of the embryos are incubated with hydroxyurea/aphidicolin (HUA), arresting cells at the G1/S boundary. Mass spectrometry measurements of H4K20 methylation (H4K20me) are performed at 14.75, 19.75, 27.5, and 40 hpf in embryos with dividing (mock) or non-dividing cells (HUA). HUA incubated embryos are viable and visually remarkably similar to mock embryos (scale bar 1 mm).
(B) H4K20me kinetics differ significantly between mock (gray) and HUA treated (green) embryo populations (two-sample t test for all three biological replicates of mock and HUA for each time point resulted in p values < 0.05 for 15 out of 16 time points). In HUA H4K20 un- and mono-methylation is decreased while H4K20 di- and tri-methylation (see inset) is increased.
Figure 2Demethylation Is Not Necessary to Explain Data of Cycling Mock Cells
(A) Model of cycling mock population composed of four H4K20 states: un- (me0), mono- (me1), di- (me2), and tri-methylation (me3). m1, m2, and m3 represent the mono-, di-, and tri-methylation rate constants and d1, d2, and d3 represent the demethylation rate constants. An overall dilution of methylation happens due to cell division, parametrized with population growth rate g(t), which is dependent on the cell-cycle function c(t).
(B) All possible parameter combinations result in 5 models without demethylation and 25 models with demethylation. Rate constants specific to a particular methylation or demethylation step are indicated in color, rate constants shared between methylation or demethylation steps are shown in gray. The number of rate constants ranges between 1 for the simplest model with no demethylation and shared methylation rate constant and 6 for the most complex model, where each methylation and demethylation rate constant is specific.
(C) Only a constrained scaled Hill function with Hill coefficient 1 and offset 0.5 gives an average cell-cycle duration in the expected range of 8 h (marked by the black box). All other cell-cycle functions c(t) predicted average cell-cycle durations of at least 70 h, which is biologically not meaningful and reflects a population of non-cycling cells.
(D) The 12 best-performing models are ordered by increasing BIC. All models with ΔBIC < 10 require either three specific methylation rate constants (m1, m2, and m3) or a specific tri-methylation rate constant. However, if present, demethylation may take on any of the 5 possible rate constant combinations. The best-performing models without demethylation perform similarly well as the best-performing models with demethylation (ΔBIC = 0 and 1). The estimated average cell-cycle duration <c(t)> is in a biologically realistic range of around 8 h.
(E) All 12 best-performing models fit the data. The model with three specific methylation rate constants but with no demethylation is shown in black.
(F) Model prediction of the cell-cycle duration (median, 25th and 75th percentiles of MCMC samples of the cell-cycle parameter of the model with three specific methylation rate constants but with no demethylation (inset)) agrees with experimental measurements of different papers.
(G) The model with three specific methylation rate constants but with no demethylation (inset) predicts an increase of cell numbers from roughly 20,000 cells after 10 h to 300,000 cells after 40 h (using the median, 25th and 75th percentiles of the MCMC samples of the cell-cycle parameter of the model with three specific methylation rate constants but with no demethylation (inset)) in a developing Xenopus embryo.
(H) The model with three specific methylation rate constants but with no demethylation is able to predict the effects on H4K20me upon morpholino knockdowns of the di- and tri-methyltransferases SUV4-20H1/2 (KD) assuming a reduction to 10% of the original di- and tri-methylation rate constants. The dotted lines are the H4K20me kinetics predictions corresponding to 0%, 5%, and 15% of the original di- and tri-methylation rate constants. The solid line shows the previous fit with 100% of the original di-and tri-methylation rate constants.
See also Table S1.
Figure 3Demethylation Is Essential to Explain Data of Cell-Cycle-Arrested HUA Cells
(A) Model of cell-cycle-arrested HUA population. In contrast to the mock model (Figure 2A), the HUA cells do not divide ( g(t) = 0), and no dilution of methylated H4K20 is required.
(B) The 5 best-performing HUA models with ΔBIC < 10 all require 3 specific methylation rate constants (m1, m2, and m3) and demethylation. However, demethylation may take on any of the 5 possible rate constant combinations. The single best-performing HUA model without demethylation (right) is outperformed by the HUA models with demethylation (ΔBIC = 13).
(C) Model fits of top 5 HUA models with demethylation overlap strongly and show the ability to explain the HUA data. The best-performing model is highlighted in black.
See also Table S1.
Figure 4 Joint Computational Modeling Allows Direct Comparisons between Mock and HUA Rate Constants and Reveals that Demethylation Is Overshadowed by HUA
(A) Joint model allows for three methylation and one demethylation rate constants for both mock and HUA as suggested by the best models for mock and HUA.
(B) We fit 16 models with demethylation and 8 models each for without demethylation in mock and/or HUA to the joint data to infer mock- and HUA-specific rate constants. The joint rate constants of mock and HUA are shown in orange, the rate constants present in both the mock and HUA models but taking on mock- and HUA-specific values are indicated in gray/green, the rate constants only present in the mock or HUA model are shown in gray and green half-circles, respectively. The model structure of the most complex of models is shown in (A). The number of rate constants ranges between 3 and 8.
(C) The best-performing models on the combined dataset are ordered according to their BIC value. All models require HUA-specific mono- and di-methylation rate constants but are indecisive about tri- methylation and demethylation. Joint models where demethylation is present in either only HUA or in both mock and HUA perform equally well. Joint models where demethylation is not present in either only HUA or in both mock and HUA perform considerably worse. Model IDs of all considerably best-performing models are given (I–VI).
(D) Model structure of the simplest best-performing joint model with demethylation in only HUA (model I).
(E) All best-performing joint models are able to explain both the mock and HUA data. The estimated initial conditions vary between the models. Joint model I is highlighted.
(F) The violin plots of the marginal distributions of all best-performing joint models show high consistency between the estimated methylation and demethylation rate constants. HUA-specific mono- and di-methylation rate constants are considerably decreased. Tri-methylation rate constants between mock and HUA have strongly overlapping marginal distributions. Demethylation seems to be dominated by the HUA population and is negligible in the mock population if a mock-specific demethylation rate is allowed.
See also Table S1.
