Pioneer factors govern super-enhancer dynamics in stem cell plasticity and lineage choice.

Methods

Mouse lines
Female CD1 mice (8 weeks old, Charles River) were used for the purification of HFSCs. Female CD-1 mice transgenic for Krt14-H2B-GFP29 (30-32 days old) were used for the purification of TACs. Krt15-CrePGR; Sox9fl/fl; R26YFPfl/+ mice have been described16. Krt19-CreER mice have been described30. CreER was activated by intraperitoneal injection of mice with 20mg ml-1 tamoxifen (Sigma) in corn oil (Sigma) to specifically label HFSCs. For the generation of K14-H2B-iRFP mice, iRFP was first amplified from pShuttle-CMV-iRFP (Addgene plasmid 31856) and fused with H2B, before the H2B-iRFP construct was assembled with the Krt14 promoter, β-globin intron and poly(A) sequences31. Transgenic mice were generated with standard pronuclear injections. For lentiviral injections, transduced mice were confirmed by genotyping with RFP primers: forward 5′-ATCCTGTCCCCTCAGTTCCAGTAC-3′, reverse 5′-TCCACGATGGTGTAGTCCTCGTTG-3. For TRE-mycSox9 transduced mice, positive mice were fed with doxycycline-containing chow, starting either at P0 (newborn) or at P21 (adult). No formal randomization was performed, and studies were not blinded. Mice were maintained in the Association for Assessment and Accreditation of Laboratory Animal Care-accredited animal facility of The Rockefeller University (RU), and procedures were performed with Institutional Animal Care and Use Committee (IACUC)-approved protocols.

Flow cytometry
Preparation of adult mice back skins for isolation of HFSCs and TACs were done as previously described8,32. Briefly, for telogen skin, subcutaneous fat was removed with a scalpel, and skins were placed dermis side down on trypsin (Gibco) at 37°C for 35 min. Single-cell suspensions were obtained by scraping the skin gently. Anagen skin was treated with collagenase at 37 °C for 30 minutes to dissociate dermal cells and then incubated with trypsin at 37 °C for 15 minutes to detach and generate single cell suspensions of the epidermal and HF cells. Cells were then washed with PBS containing 5% of fetal bovine serum (FBS), then filtered through 70μm and 40μm cell strainers. Cell suspensions were incubated with the appropriate antibodies for 30 min on ice. The following antibodies were used for FACS: α6-PE (1:100, eBiosciences), CD34-eFluoro660 (1:100, eBiosciences) and Sca-1-PerCP-Cy5.5 (1:1000, eBiosciences). DAPI was used to exclude dead cells. Cell isolations were performed on FACSAria sorters running FACSDiva software (BD Biosciences).

ChIP-seq
Immunoprecipitations were performed on FACS-sorted populations from female mice or on cultured HFSCs8. For each ChIP-seq run, 7 × 106 to 2 × 107 cells were used. Antibodies used for ChIP-seq were anti-H3K27ac (abcam, ab4729), anti-H3K4me1 (abcam, ab8895), anti-Crsp1/Trap220 (Med1, Bethyl Laboratories, A300-793A) and anti-H3K27me3 (Millipore, 07-449). Briefly, cells were cross-linked in 1% (wt/vol) formaldehyde solution, resuspended, and lysed. To solubilize and shear cross-linked DNAs, lysates were subjected to a Bioruptor Sonicator (Diagenode, UCD-200) according to a 30x regimen of 30s sonication followed by 60s rest. The resulting whole-cell extract was incubated overnight at 4°C with 10 μL of Dynabeads Protein G magnetic beads (Life Technologies) which had been pre-incubated with 5μg of the appropriate Ab. After ChIP, samples were washed, and bound complexes were eluted and reverse–cross-linked. ChIP DNA was prepared for sequencing by repairing sheared DNA and adding Adaptor Oligo Mix (Illumina) in the ligation step. A subsequent PCR step with 25 amplification cycles added the additional Solexa linker sequence to the fragments to prepare them for annealing to the Genome Analyzer flow cell. After amplification, a range of fragment sizes between 150–300 bp was selected and the DNA was gel-purified and diluted to 10 nM for loading on the flow cell. Sequencing was performed on the Illumina HiSEq 2500 Sequencer following manufacturer protocols. ChIP-seq reads were aligned to the mouse genome (mm9, build 37) using Bowtie aligner33. ChIP-Seq signal tracks were presented by Integrative Genomics Viewer (IGV) software.

Bioinformatics analysis
H3K27ac peaks were called by the program MACS34 (v 1.4.2, default parameters) from the ChIP-seq data with the input as controls. The peaks were associated to genes using the mouse RefSeq annotations; those located within 2kb of transcription start sites were called as “promoter” peaks and the rest were “enhancer” peaks. The H3K27ac enhancer peaks were used for the identification of super-enhancers, using the algorithm described previously, wherein enhancer peaks were stitched together if they are located within 12.5kb of each other and if they don't have multiple active promoters in between. Enhancers were then ranked according to increasing H3K27ac signal intensity5. Enhancer-gene assignments were performed using the following criteria to make gene assignments: (1) proximity of genes to the SE of stem cells; (2) high transcriptional activity in stem cells (by RNA-seq and by ChIP-seq for presence of H3K4me3/H3K79me2 marks and no H3K27me3 marks in the promoter/typical enhancer and/or gene body; (3) correlation between loss of the SE (or shift in its epicenter peaks), loss of gene transcription and loss of H3K79me2 mark ± H3K27me3 mark in proliferative short-lived progenitors; (4) correlation between loss of the SE (or shift in its epicenter), loss of gene transcription and loss of H3K79me2 mark ± H3K27me3 mark in proliferative cultured stem cells. The overlap of super-enhancers with ChIP-seq peaks for MED1 and other TFs was defined by >= 1 base overlap. For TF enrichment analysis at super-enhancers, H3K27ac peaks not located at super-enhancers (i.e., typical enhancers) were randomly picked, extended to match the sizes of super-enhancers and used as background controls. GO function enrichment analyses were carried out by the software GREAT35 using the list of super-enhancer coordinates and the default setting. For motif analysis of enhancers located in superenhancers, 1-kb sequences under the H3K27ac peaks were searched for enriched motifs using the software HOMER (v4.6) with the default setting (PMID 20513432). Epicenters were defined as 1kb-regions flanking either side of the H3K27ac peaks. 1-kb was chosen based on our analysis of the distances of H3K27ac peaks to their nearest transcription factor ChIP-seq peaks in HFSCs in vivo (distance of the two peak centers, Extended Data Fig. 3e), which showed an enrichment of TF binding within 1-kb regions of H3K27ac peaks. Overlapping epicenters were merged during this analysis. To analyze epicenter shifting, for each of the overlapping super-enhancers between HFSCs in vivo and in vitro, we determined the number of epicenters that were not overlapping in the two samples and considered them as shifting epicenters. To generate the heatmap (Extended Data Fig. 3c), the program seqMiner36 was used to calculate the ChIP-seq read densities, which were the maximal numbers of overlapping ChIP-seq reads in 50-bp bins from -5 kb to +5 kb of the H3K27ac peak summits. The density matrix was clustered based on the H3K27ac ChIP-signal and then used to generate a heatmap.

Antibodies
The following antibodies and dilutions were used: SOX9 (rabbit, 1:1000, Millipore), NFIB (rabbit, 1:1000, Active Motif), LHX2 (rabbit, 1:2000, Fuchs lab), K6 (guinea pig, 1:5000, Fuchs Lab), K24 (rabbit, 1:5000, Fuchs lab), CD34 (rat, 1:100, BD-Pharmingen), LEF1 (rabbit, 1:100, Fuchs lab), NFATc1 (mouse, 1:100, Santa Cruz), TCF3 (guinea pig, 1:200; Fuchs laboratory), TCF4 (rabbit, 1:300; Cell Signaling Technology), FHL2 (rabbit, 1:100, abcam), PRRG4 (rabbit, 1:100, abcam), CUX1 (rabbit, 1:200, Santa Cruz), β4-Integrin (rat, 1:100, BD-Pharmingen), GFP (chicken, 1:2000, Abcam), RFP (rabbit, 1:5000, MBL; or guinea pig, 1:3000, Fuchs lab). Secondary Abs coupled to Alexa488, RRX, or Alexa647 were from Life Technologies. Nuclei were stained using 4’6’-diamidino-2-phenylindole (DAPI).

Histology, Immunofluorescence and Imaging
Back skins from mice were embedded in OCT (Tissue Tek), frozen, cryosectioned (10-20 μm) and fixed for 10 min in 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS). For lentivirally transduced mice, head and backskins were pre-fixed in 4% PFA for 4h at 4 degrees, followed by washes in PBS and incubation in 30% sucrose, before embedding in OCT. Sections were blocked for 1hr in gelatin block (5% normal donkey serum, 1% BSA, 2% fish gelatin, 0.3% Triton X-100 in PBS). Primary antibodies were diluted in blocking buffer and incubated at 4°C overnight (O/N). MOMBasic kit (Vector Laboratories) was used for blocking when primary antibodies were generated from mouse. After washing with PBS, secondary antibodies, were added for 1hr at room temperature (RT). Slides were washed with PBS, counterstained with 4’6’-diamidino-2-phenylindole (DAPI) and mounted in Prolong Gold (Invitrogen). Images were acquired with an Axio Observer.Z1 epifluorescence microscope equipped with a Hamamatsu ORCA-ER camera (Hamamatsu Photonics), and with an ApoTome.2 (Carl Zeiss) slider that reduces the light scatter in the fluorescent samples, using 20x objective, controlled by Zen software (Carl Zeiss). Z stacks were projected and RGB images were assembled using ImageJ. Panels were labeled in Adobe Illustrator CS5.

Lentiviral expression constructs
Lentiviral super-enhancer reporters were generated by PCR amplification of selected enhancer regions from BAC clones, followed by insertion into KpnI and BsaBI restriction sites of the Rbpj-EGFP construct37. To generate the Sox9 expression construct, Sox9 cDNA was PCR amplified, and inserted into the LV-TRE-PGKH2BmRFP1 construct18. The resulting LV-TRE-mycSox9-PGK-H2BmRFP was used for in utero injections.

Partial Thickness Wound (Dermabrasion) and HFSC transplantation
Animals were anesthetized with Ketamine/Xylazine and administered Bupenorphine analgesia. Skin was shaved and remaining hair cleared with hair removal cream. Skin was gently stretched between two fingers and epidermis removed using a small rotary drill (Dremel) with a polishing wheel attachment (model 520), to create a partial-thickness wound. HFSC transplantations were described previously26.

Cell Culture
Primary HFSCs were isolated from P52-60 K14-H2B-iRFP mice and plated onto mitomycin C-treated dermal fibroblasts in E-media supplemented with 15% (vol/vol) serum and 0.3 mM calcium32. For colony formation assays, equal numbers of Sox9-deficient live cells were plated. After 14 days in culture, cells were fixed and stained with 1% (wt/vol) Rhodamine B (Sigma). Colony diameter was measured from scanned images of plates using Image J and colony numbers were counted. For viral infections, HFSCs were spun with lentivirus for 30 min at 1100g in the presence of polybrene (100 μg/ml)23. For Sox9 over-expression studies, the PGK-Sox9-IRES-H2BYFP construct was transfected into cultured HFSCs or epidermal keratinocytes. 72h later, YFP+ and YFP- cells were purified by FACS. Luciferase assays were performed as described15.

RNA extraction and qRT-PCR
FACS-isolated cells were sorted directly intro TrizolLS (Invitrogen). Total RNA was purified using the Direct-zol RNA MiniPrep kit (Zymo Research) per manufacturer's instructions. DNase treatment was performed to remove genomic DNA (RNase-Free DNase Set, Qiagen) Equal amounts of RNA were reverse-transcribed using Oligo-dT primers (Superscript III, Life Technologies). qRT-PCR was performed on an Applied Biosystems 7900HT Fast Real-Time PCR system. cDNAs were normalized to equal amounts using primers against Ppib2.

Statistics
For all measurements, 3 biological replicates and 2 or more technical replicates were used. Experiments were independently replicated twice, and representative data are shown. To determine the significance between two groups, comparisons were made using unpaired two-tailed Student's t test in Prism6 (GraphPad software). For all statistical tests, the 0.05 level of confidence was accepted for statistical significance.

Extended Data

Supplementary Material
supp info guidesupp table 1supp table 2supp table 3