Computational protocol: Large scale genomic reorganization of topological domains at the HoxD locus

Similar protocols

Protocol publication

[…] 3D DNA fluorescent in situ hybridization was as previously described [, ]. E12.5 mouse embryos were fixed in 4% paraformaldehyde, embedded in paraffin blocks, and cut at 6 μm. Sections were oriented such that cells belonging to either the distal (autopod) or proximal (zeugopod) parts of the growing limb bud could be unambiguously identified. Probes were prepared by nick-translation with directly labeled nucleotides (Ulysis alexa 647, Life Technologies) biotin- or digoxigenin-UTP using fosmid clones obtained from the BACPAC Resources Center ( and listed in Additional file : Table S2. DNA (100 ng) was used with 7 μg of Cot1-DNA and 10 μg of sonicated salmon sperm DNA. They were labeled using either digoxigenin- or biotin-dUTP by nick translation with fluorescent revelations as described in [], using either Alexa 647, Alexa 568, or Alexa 488 as fluorophores. Slides were stained with DAPI and mounted in ProLong Gold (Life Technologies). Images were acquired using a B/W CCD ORCA ER B7W Hamamatsu camera associated with an inverted Olympus IX81 microscope. The image stacks with a 200-nm step were saved as TIFF stacks. Image reconstruction and deconvolution were performed using FIJI (NIH, ImageJ v1.47q) and Huygens Remote Manager (Scientific Volume Imaging, version 3.0.3). Distance measurements between probe signals were determined using an automated spot/surface detection algorithm followed by visual verification and manual correction using IMARIS version 6.5, Bitplane AG, and Matlab 7.5, MathWorks SA. Data from Fig.  were evaluated using only manual measurements. Statistical significance analyses of distances were performed using Mann–Whitney test (Figs. , , ; Additional file : Figures S2b and S3b), or using the Kruskal–Wallis test followed by Dunn’s post test (Fig. ). The images can be found on Figshare (Fig. , doi:10.6084/m9.figshare.5198161; Fig. , doi:10.6084/m9.figshare.5198242; Fig. , doi:10.6084/m9.figshare.5198326; Additional file : Figure S3a, doi:10.6084/m9.figshare.5198386). [...] Micro-dissected E12.5 proximal or distal limb bud tissues were dissociated, fixed with 2% formaldehyde, lysed, and stored at −80 °C. The nuclei from ten pairs of distal or proximal forelimbs were then digested with a sequence of NlaIII and DpnII, followed by amplification according to []. The ligation steps were performed using high concentrated T4 DNA ligase (M1794, Promega) and the inverse PCRs for amplification were carried out using primers specific for the various viewpoints []. Hoxd13 amplification primers were previously described []. PCR products were multiplexed and sequenced with a 100-bp single-end Illumina HiSeq flow cell. Demultiplexing, mapping to the mouse assembly GRCm38 (mm10), and 4C-Seq analysis were performed using the BBCF HTSstation ( and []), according to []. Briefly, the 4C-Seq fragments directly surrounding the viewpoints (2 kb) were excluded for the rest of the analysis. Fragment scores were normalized to the total number of reads mapped and smoothed (running mean with a window size of 11 fragments). For comparison purposes, the 4C-seq profiles were normalized to the mean score of fragments falling into a region defined as the bait coordinates plus or minus 1 Mb. In Fig. , a profile correction method similar to that described in [] was applied to emphasize the relevance of long-range interactions with the islands. This was done using a fit with a slope −1 in a log-log scale []. For quantification of 4C-seq profiles in specific islands or regions of interest (box plots in Additional file : Figure S2 and Fig. ), the smoothed data, with or without profile correction, were used. When appropriate (e.g., signals in Fig. ), replicates were combined by averaging the resulting signal densities. In Fig  and Additional file : Figure S4, quantitative log2 ratios were calculated by dividing the fragment scores with the means in WT1, WT2, and WT3. The number of mapped reads for each sample were as follows: in Fig.  (wild type only), 19,160,126 for autopods and 13,257,140 for zeugopods; in Figs.  and (inversions) 9,050,395 for Hoxd13 and 4,134,080 for island II; in Fig.  (deletion series), 3,722,913 for WT1, 3,812,342 for del1, 8,943,042 for WT2, 6,112,348 for del2, 6,494,171 for WT3, and 6,702,116 for del3. 4C-Seq data are available from the GEO repository under accession number GSE98861. In order to detect tripartite interactions, one of the 4C libraries was re-sequenced as 250-bp single-end. The reads were de-multiplexed using fastx_barcode_splitter ( and the viewpoint sequence was removed except for CATG (first cutter sequence) with seqtk ( Then they were trimmed for low quality and the presence of GATC (second cutter sequence) with cutadapt (cutadapt -q 10 -a GATC) []. Next, they were split if a CATG was present. The 5′ part of the split reads (hereafter referred to as “mid”) and the 3′ part of the split reads (hereafter referred to as “third”) were mapped independently with bowtie [] (version 0.12.9) on mm10 (bowtie -p 5 -S -k 1 -m 1 -I 0 --best –strata). If the third read did not map, they were split again for CATG and only the first part was now considered as third and mapped. Reads for which the mapping of mid and third were consecutive (undigested situation) were not considered in the analysis. The reads were then pooled according to the mapping position and strand of the mid and the mapping position and strand of the third to remove potential PCR duplicates, resulting in a list of unique tripartite interactions. Each tripartite interaction in the 73.8–74.7 Mb region of chromosome 2 was assigned to a 20-kb bin and the matrix showing the number of different tripartite interactions was plotted. […]

Pipeline specifications

Software tools HTSstation, FASTX-Toolkit, Seqtk, cutadapt, Bowtie
Application 4C-seq analysis
Organisms Homo sapiens, Danio rerio