Open Access Highly Accessed Research

Modulated contact frequencies at gene-rich loci support a statistical helix model for mammalian chromatin organization

Franck Court1, Julie Miro2, Caroline Braem1, Marie-Noëlle Lelay-Taha1, Audrey Brisebarre1, Florian Atger1, Thierry Gostan1, Michaël Weber1, Guy Cathala1 and Thierry Forné1*

Author Affiliations

1 Institut de Génétique Moléculaire de Montpellier (IGMM), UMR5535 CNRS, Universités Montpellier 1 et Montpellier 2. 1919, Route de Mende, 34293 Montpellier Cedex 5, France

2 Current address: INSERM U827, Laboratoire de Génétique des Maladies Rares, IURC, 64, avenue du Doyen G Giraud, 34093 Montpellier Cedex 5, France

For all author emails, please log on.

Genome Biology 2011, 12:R42  doi:10.1186/gb-2011-12-5-r42

Published: 10 May 2011

Additional files

Additional file 1:

Random collision frequencies in gene-rich regions for large separations distances. Random collision frequencies were determined by 3C-qPCR after a primer extension step (see Materials and methods) at two Usp22 genomic sites (sites F1 and F-28) (Figure 1a) in liver samples from 16.5-days-post-coitus embryos (grey data points) or 30-day-old mice (white data points). Data analysis was as described in the legend of Figure 1b. Red squares represent the floating mean (45-kb windows, shift of 22.5 kb). We determined the higher and the lower points of the floating mean for site separations above 40 kb and calculated the average random collision frequencies (values are indicated in the figure) of sites located 40 kb around these points (horizontal black bars). P-values (Mann-Whitney U-test) account for the significance of the differences observed between these averages. Error bars are standard error of the mean.

Format: PDF Size: 21KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 2:

Collision frequencies at the human β-globin locus. Collision frequencies at the human β-globin locus (a gene-rich region on chromosome 11p15.4) were obtained from several published 5C experiments performed in GM06990 cells, an EBV-transformed lymphoblastoid cell line where this locus is not expressed and where only a very weak/residual interaction was detected (Supplemental Tables 6 and 7 in [13]). Data from each experiment were normalized according to a previously published algorithm [19] and plotted into a single graph. Statistical analyses were performed as explained in the legend of Figure 1b.

Format: PDF Size: 97KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 3:

Fitting the circular polymer model to mouse gene-rich loci. The circular polymer model (Equations 1 and 2b) was fitted to 3C-qPCR data obtained at gene-rich loci. The best fit curve is shown in red and best fit parameters are as follows: R2 = 0.50 with K = 725,785 ± 66,540; S = 2.515 ± 0.092 kb; c = 110.515 ± 2.028 kb. The black curve depicts the best fit obtained with the linear polymer model (Equations 1 and 2a; R2 = 0.18).

Format: PDF Size: 91KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 4:

Gene expression at loci investigated by 3C-qPCR. Total RNA from 30-day-old mouse liver was prepared and mRNA levels were determined by RT-qPCR relative to Gapdh mRNA level. The Usp22, LnP and Mtx2 genes were found to be expressed. Very low levels of expression were found for the Gtlf3b, Aldh3a2 and Emb genes. The other genes (Kcnj12, Tnfref13b, Gtl2, Dlk1 and HoxD13) are fully repressed.

Format: PDF Size: 22KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 5:

Random collisions at silent versus expressed loci. Data points represent collision frequencies determined at silent (Dlk1/Emb/Lnp; black circles) or expressed (Usp22/Mtx2; red circles) loci. Best fit of the statistical helix model (Equations 1 and 5) was performed for each dataset (black curve = silent loci; red curve = expressed loci). The values of best fit parameters for each data set are indicated in the graph. Both the diameter (D) and the step (P) of the helix are larger in the expressed loci compared to the silent ones.

Format: PDF Size: 124KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 6:

Fitting the statistical helix model to the yeast Saccharomyces cerevisiae genome. In order to test whether a statistical helix organization may be valid for other organisms, we fitted the statistical helix polymer model to the 3C data obtained in the yeast S. cerevisiae [24]. For both AT-rich and GC-rich regions (Additional file 7 and 7b, respectively), correlation coefficients (R2 = 0.82 and 0.80, respectively) were similar to those obtained from published models (R2 = 0.81 and 0.79, respectively) [24]. For AT-rich regions, consistent with previous findings [24], the statistical helix model predicts a linear polymer organization (Additional file 7). However, data obtained in GC-rich domains are fully compatible with a statistical helix organization. Compared to mammals, chromatin dynamics in yeast can be described as a statistical helix that would have a slightly smaller diameter (212.62 ± 31.73 nm) but a much wider step (310.94 ± 54.86) (Additional file 7). Finally, using these best-fit parameters and Equation 4c, we calculated how, according to this statistical helix model, the spatial distances should vary as a function of genomic site separations. We found that spatial distances calculated from the statistical helix model are in good agreement with those measured in high-resolution FISH analyses performed in living yeast cells (Additional file 7) [37]. Therefore, the statistical helix model may also be valid to describe chromatin dynamics in GC-rich domains of the S. cerevisiae genome.

Format: PDF Size: 50KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 7:

Fitting the statistical helix model to the yeast Saccharomyces cerevisiae genome. Data published by Dekker for the yeast S. cerevisiae [24] were normalized using the previously published algorithm [19] and the statistical helix polymer model (Equations 1 and 5 was fitted to normalized data. (a) For AT-rich regions, consistent with previous findings [24], the statistical helix model (red curve) predicted a linear polymer organization (black curve). In this case, the best fit values obtained for the diameter D and the step P are not relevant, as indicated by large standard deviations. (b) In GC-rich regions, the statistical helix model (red curve), fits with a distended helical shape. Best-fit parameters are indicated above the graph. They were calculated using a linear mass density of 11.1 nm/kb [11]. The black curve depicts the best fit of the linear polymer model and the green curve the best fit of the circular polymer model. Note that the lengths of the statistical fragments obtained from the statistical helix model (S = 6.060 ± 0.519 kb and 4.558 ± 0.503 kb for AT-rich and GC-rich domains, respectively) are compatible with the parameters previously obtained with the linear or circular polymer models (S = 6.4 ± 0.34 kb and 4.7 ± 0.45 kb, respectively) [24]. (b) Using the best-fit parameters obtained for the yeast S. cerevisiae (b), we calculated the expected mean spatial distances (in nm) for increasing site separation distances (0 to 140 kb) for both the statistical helix (Equation 4c; red curve) and the linear polymer (Equation 4a; black curve) models. The experimental spatial distances (in nm) obtained by Bystricky et al. (Table 1 and Supplementary Table of [37]) from high-resolution FISH experiments were plotted into this graph (open squares, adjusted average distances; black diamonds, average peak distances). The statistical helix model is in good agreement with these experimental data.

Format: PDF Size: 65KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 8:

An upper limit of validity for the statistical helix model. Expected spatial distances (in nm) were calculated as a function of increasing genomic distances (in kb) using either Equation 4a (linear polymer model, black curve, with L = 9.6 nm/kb) or Equation 4c and the biophysical parameter given in Figure 4 (statistical helix model, red curve). Dashed lines represent the expected deviations due to standard errors on the measured biophysical parameters (Figure 4). Details about mathematical equations are given in the Materials and methods section. Data points (blue diamonds) depict spatial distances measured by FISH experiments as reported by van den Engh et al. [32]. These data points were obtained from a gene-rich chromosomal region containing the Huntington disease locus.

Format: PDF Size: 29KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 9:

3C-qPCR dataset for gene-rich regions.

Format: XLS Size: 69KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 10:

3C-qPCR dataset for the gene-desert region.

Format: XLS Size: 35KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 11:

3C-qPCR primers.

Format: XLS Size: 38KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data