Figure 3. Best regulated on-off-slide model compatible with all the experimental data sets in stage 3.

All fits were done simultaneously (see Materials and methods). (A) Regulated on-off-slide model with the regulated process A (global assembly, thick arrows) and the constitutive processes D (global disassembly), D1-4 (disassembly from configuration 1 to 4, overrules D), S2* (sliding away from N-2) and S3-4 (sliding from configuration 3 to 4). (B, C, D) Combined fits to the steady-state promoter nucleosome configuration occurrences in repressed, weakly activated, and activated state. Only a change in the rate of the regulated global assembly process accounts for differences in the three distributions. The other processes (D, D1-4, S2*, and S3-4) are constitutive and their rates do not change. The model fits in stage 1 (ignoring all other data) and stage 2 (ignoring Flag-/Myc-tagged histone exchange data) are only slightly better (data not shown). (E) Fit to the sticky N-3 RE accessibility data of two mutants in the activated state (error bars with standard deviation of rescaled RE accessibility). Only reaction rates involving the N-3 were allowed to divert from the previously fitted parameter values. (F) Fit to the Rufiange et al., 2007 data of Flag amounts at N-1 over N-2 after 2 h of Flag expression (shifted by 0.25 h lag time determined in Dion et al., 2007). The error bar corresponds to the standard deviation of two measurements. (G) Fit to the Dion et al., 2007 data of Flag over Myc amounts at N-1 at four time points after Flag expression (shifted by 0.25 h lag time). y-axis points are normalized by their mean to account for a sloppy fitted parameter in the treatment of the data in Dion et al., 2007. Error bars are estimated experimental standard deviations used in the fit.

Figure 3—figure supplement 1. Stage 3: Flag amount ratios of tested models.

Dynamics of Flag amount ratios after Flag expression (shifted by 0.25 h lag time) of different sites for repressed (left column) and activated state (right column), calculated for all models with satisfactory likelihood after stage 2. Green lines represent the satisfactory models of stage 3. The dashed lines show the steady state ratio that eventually all models should reach closely. (A) Ratio of N-1 over N-2 in repressed state with available experimental data. The three groups of green lines match the three model groups of the main text and Figure 4: almost constant lines (group 3), slightly rising lines (group 1), and quickly rising lines (group 2). (B) As panel A but for the ratio of N-3 over N-2 and without experimental data. (C) As panel B, but N-3 over N-1. (D, E, F) All three ratios now for the activated promoter state.

Figure 3—figure supplement 2. Stage 3: Flag over Myc amount ratios of tested models.

Dynamics of the normalized Flag over Myc amount ratio (shifted by 0.25 h lag time) at N-1 position, calculated for all models with satisfactory likelihood after stage 2. Green lines represent the satisfactory models of stage 3.

Figure 3—figure supplement 3. Stage 3: logarithmic likelihood ratio histograms.

Histograms of the logarithmic likelihood ratio with respect to the perfect fit likelihood in stage 3, ${\displaystyle R_3}$. 0 on the x-axis corresponds to a perfect fit. (A) The seven satisfactory models with $R_3>6$, all of them with one regulated assembly process. Dashed green line: value of the best model. (B) Final results for the same analysis initially also including all models with one additional fitted parameter, a total of 837,435 models, to check for the occurrence of differently regulated models. Of the now 508 satisfactory models, 501 also have a regulated assembly process and only 7 a regulated disassembly process.