Dedifferentiation maintains melanocyte stem cells in a dynamic niche – Nature

Generation of Oca2
^creER knock-in mice

Oca2^creER knock-in line was generated following well-established protocols with slight modifications^52,53. The Oca2^creER-targeting DNA was constructed as presented in Extended Data Fig. 7a by VectorBuilder. The purified plasmid DNA of the targeting vector was linearized using the restriction enzymes NotI and SalI for embryonic stem (ES) cell targeting. The mouse MK6 (C57BL/6J) ES cells (established at New York University (NYU) Langone’s Rodent Genetic Engineering Laboratory) were grown, at passage 6, on mitotically inactivated mouse embryonic fibroblast (MEF) cells (Sigma Millipore) and passaged the day before electroporation. Linearized targeting vector DNA (25 μg ml^–1) containing the neomycin-resistance gene was added to the cell suspension, and electroporation was performed using either a Gene Pulser II or a Gene Pulser system (Bio-Rad). Following electroporation, the cells were plated onto neomycin-resistant MEFs and incubated at 37 °C, 95% humidity and 5% CO₂. After 24 h, geneticin (160 μg ml^–1 active concentration, Invitrogen) was added to the growth medium for positive selection of antibiotic-resistant ES cell colonies. The medium was changed every day, and geneticin (G418) selection was maintained for 6 days. Antibiotic-resistant ES cell colonies were counted, picked and split to grow in 96-well plates duplicated for cryopreservation, and homologous recombination events were identified through genotyping by Southern blot analysis. The mouse MK6 ES cells and MEFs were tested for mycoplasma contamination before use and were not authenticated.

ES cells with successful homologous recombination were injected into mouse blastocyst embryos. ES cells were trypsinized to obtain a single-cell suspension, and the cell suspension was kept on ice in 1 ml ES cell medium in a 15 ml tube until use. Blastocyst embryos were collected from C57BL/6-albino females (4 weeks old, NIH 562, CRL) at 3.5 days post coitum. Ten to fifteen ES cells were injected into each blastocyst embryo, and injected blastocysts were cultured in KSOM medium at 37 °C in an atmosphere of 5% CO₂ for 2 h until the blastocyst cavity was recovered. The microinjected blastocysts were transferred to the uterine horns of 2.5 days post coitum pseudopregnant females (CD-1, CRL) using the standard procedure to generate chimeric mice. The chimeric mice were bred with C57BL/6 to obtain Oca2^creERT2 mice.

Mouse experiments

All animal experiments were performed in compliance with all relevant ethical regulations for animal testing and research and in accordance with animal protocols approved by the Institutional Animal Care and Use Committee at NYU School of Medicine. Mice were housed in an animal room with a temperature range of 20–22 °C, humidity range of 30–70% and under a 12–12 h dark–light cycle.

Tyr^creER (012328), Rosa^LSL-tdTomato (007905), Wnt1^cre, K15^crePR1 (005249), Wls^fl/fl (012888) and K14^rtTA (008099) mice were purchased from The Jackson Laboratory. Dct^rtTA;tetO^H2B-GFP (iDCT-GFP) mice were obtained from the NCI Mouse Repository. Dct^lacZ mice were from P. Overbeek. Ctnnb1^fl(ex3)/+ (Ctnnb1^STA) mice were from M. M. Taketo⁴³. Mice were bred and crossed in-house to obtain experimental and control animals in mixed backgrounds. Mice from experimental and control groups were randomly selected from either sex for experiments. Data collection and analyses were not performed blind to the conditions of the experiments.

To induce Cre recombination, tamoxifen (Sigma-Aldrich) treatment was performed as previously published⁷ by intraperitoneal injection (0.1 mg g^–1 body weight) of a 20 mg ml^–1 solution in corn oil per day. For the UVB experiment, dorsal fur of 3-week-old Oca2^creER;Rosa^LSL-tdTomato mice was clipped, and mice were anaesthetized. Mice were treated every other day with 600 mJ cm^–2 UVB per day 3 times in total. Skin biopsies were taken from euthanized mice or under isoflurane anaesthesia. For isolation of McSCs and differentiated bulb MCs, Dct^rtTA;tetO^H2B-GFP mice were administered doxycycline-containing chow (1 g kg^–1, Bio-Serv) for 4 days before euthanizing the mice for cell isolation. In the c-Kit antibody injection experiment, mice were subcutaneously injected with 150 µl of 0.5 mg ml^–1 c-Kit antibody (ACK45, BD Pharmingen) into a 2 × 2cm area of back skin. Control mice were subcutaneously injected with 150 µl of PBS.

Melanocyte stem cell in vivo imaging

In vivo imaging of melanocyte stem cells was based on previously described methods of live imaging of HF stem cells^54,55. Tyr^creER;Rosa^LSL-tdTomato;K14^rtTA;tetO^H2B-GFP mice were given a single injection of 60 µg tamoxifen on P21 and maintained on a 1 g kg^–1 doxycycline-containing diet from P21. All the in vivo imaging was performed at least 3 days after tamoxifen induction. As the timing of the hair cycle is slightly variable between individual mice and individual HFs, we kept imaging the mouse to monitor the hair cycle stages. HF stages were determined according to a previous publication⁵⁶ by observing the pattern of GFP-labelled HF epithelial cells. We performed the initial imaging at telogen/anagen I to capture HFs that contain only a single tdTomato-labelled cell. To trace the fate of HG melanocytes, we focused on HFs containing a single tdTomato⁺ melanocyte in the HG and revisited the same HFs every 1–4 days to capture hair cycle stages, including early anagen, mid/late anagen, early-to-mid (early/mid) catagen, late catagen and the subsequent (second) telogen. Some mice were revisited less frequently to focus on capturing mid/late anagen and the subsequent telogen. To trace the fate of bulge McSCs, we initially visited HFs containing a single tdTomato⁺ McSC in the bulge at telogen and revisited the same HFs until mid/late anagen.

To trace the fate of melanocytes located in the TA compartment of early-anagen HFs, we placed Oca2^creER;Rosa^LSL-tdTomato;K14^ktTA;tetO^H2B-GFP mice on a doxycycline-containing diet from P21 and kept imaging the mouse to monitor the hair cycle stages. The mice were given a single injection of 1.2 mg tamoxifen when the HF stage progressed to anagen I/early-anagen II. We then performed the initial imaging at anagen II/anagen IIIa (at least 3 days after tamoxifen injection) to capture HFs that contained only a single tdTomato-labelled cell, which is located in the lowermost part of the HG. The same HFs were revisited at mid/late anagen and telogen.

Throughout the course of imaging, mice were on a warming pad and anaesthetized with vapourized isoflurane delivered through a nose cone (1.5% in oxygen and air). The ear was immobilized on a custom-made stage, and a glass coverslip was placed directly against it. An Olympus Fluoview multiphoton microscope (FVMPE-RS) equipped with a MaiTai HP DS-OL laser tuned to 940 nm and an Insight X3-OL tuned to 1,120 nm (Newport Spectraphysics) were used in line sequential mode to acquire z-stacks with a ×25 NA.

Revisitation of the same HFs in live imaging

To ensure successful revisitation of the same HFs, patterns of HF clusters and blood vessel locations were used as landmarks. Blood vessel location in the ear was first used to broadly return to the same area. Then HF cluster patterns in large areas were recorded using multiple tiled images (up to 30) with z-stack steps of 20–30 µm during the initial visit and revisits. Rather than being evenly distributed, HFs in the imaged areas showed distinct clustered patterns. A cluster usually contained three to ten HFs with a gap between different clusters. By numbering the clusters and finding the same clusters over time using a tiled map, we frequently revisited the same cluster. Owing to the low tamoxifen concentration used to achieve clonal labelling of McSCs, only a subset of HFs within a cluster contained any tdTomato-labelled cells. We verified that within a cluster, the number and pattern of HFs that contained tdTomato-labelled cells matched previous visits to a specific HF.

X-gal staining

X-gal staining was done as previously published³⁸. Dorsal skin from Dct^lacZ mice was collected, and subcutaneous fat was removed using scalpel blades (Miltex). Tissues were fixed in 4% paraformaldehyde (PFA) for 30 min at 4 °C, and whole-mount X-gal staining was performed. After X-gal staining, skin samples were fixed again in 4% PFA at 4 °C overnight. The tissues were then subjected to 3D whole-mount analyses.

3D whole-mount niche analyses

3D whole-mount niche analysis was done as previously published³⁸. Whole skin samples of Dct^LacZ mice were stained with X-gal and sequentially treated with 25% glycerol–PBS, 50% glycerol–PBS and 100% glycerol for 3 h at room temperature or overnight at 4 °C. Skin was then cut into thin strips using a scalpel blade. Single HFs were then dissected and isolated using a dissection microscope (Axiovision Discovery V12).

Skin samples of Oca2^creER;Rosa^LSL-Tomato mice, Tyr^creER;Rosa^LSL-Tomato mice and Dct^rtTA;tetO^H2B-GFP mice were collected and fixed in 4% PFA for 30 min at room temperature to preserve tdTomato signals. Tissues were then incubated in 30% sucrose at 4 °C overnight and embedded in OCT compound (Sakura). Then 70–100-µM-thick skin sections, including whole-mount HFs, were made and counterstained with 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI).

Whole HFs were then imaged as described in the section ‘Microscopy’.

Immunofluorescence

Immunofluorescence was done as previously published but with slight modifications^7,24. Skin tissues were fixed overnight in 4% PFA at 4 °C. After sequential dehydration in increasing concentrations of ethanol and xylene, tissues were embedded in paraffin. Sections were cut at 6 μm, deparaffinized and microwaved in 10 mM Tris and 1 mM EDTA (pH 8.0) for antigen retrieval. Tissue sections were then incubated with the following primary antibodies for 2 h at room temperature or overnight at 4 °C in PBT (PBS plus 0.1% Triton-X100) with 10% FBS, followed by Alexa-488-conjugated or Alexa-594-conjugated secondary antibodies (1:200; Thermo Fisher): goat anti-Dct (1:100; Santa Cruz, sc-10451), rabbit anti-Tomato (1:1,000; Rockland, 600–401–379); mouse anti-Tomato (RF5R) (1:500; Thermo Fisher, MA5–15257); rabbit anti-Tyrp1 (1:100; Sigma-Aldrich, SAB2102617); mouse anti-E-cadherin (1:100; BD Transduction, 610181); rabbit anti-Ki67 (1:100; Abcam, ab15580); mouse anti-β-catenin (1:400; Sigma-Aldrich, C7207); and mouse anti-MITF (1:100; Abcam, ab12039). The following secondary antibodies from Thermo Fisher were used: Alexa Fluor 594 donkey anti-mouse IgG (1:200; A21203); Alexa Fluor 488 donkey anti-mouse IgG (1:200; A21202); Alexa Fluor 594 donkey anti-rabbit IgG (1:200; A21207); Alexa Fluor 488 donkey anti-rabbit IgG (1:200; A21206); Alexa Fluor 594 donkey anti-goat IgG (1:200; 11058); and Alexa Fluor 488 donkey anti-goat IgG (1:200; A11055). Skin sections were counterstained with DAPI.

For detection of CD34, P-cadherin and DCT, skin tissues were collected, processed and embedded in OCT as described above (‘3D Whole mount niche analyses’) to preserve tdTomato or GFP signals. Next, 50–70-µM-thick skin sections were made. The skin sections were incubated in PBS plus 0.5% Triton-X100 for 1 h at room temperature and incubated with primary antibody against DCT (1:100; Santa Cruz, sc-10451), CD34 (1:50; BD Pharmingen, 553731) and P-cadherin (1:100; Invitrogen, 13-2000Z) overnight at 4 °C. Skin sections were then washed 3 times with PBS for 10 min at room temperature. For DCT, sections were incubated with Alexa-488-conjugated secondary antibody (1:200) for 1 h at room temperature. For CD34 and P-cadherin, sections were incubated with biotinylated anti-rat IgG (1:100; Vector Laboratories, BA-9400) for 1 h followed by incubation with streptavidin Alexa 647 conjugate (1:200, Invitrogen, S32357) for 30 min at room temperature. Skin sections were counterstained with DAPI.

In situ hybridization

RNAscope in situ hybridization was performed using a Leica Bond III automated staining platform (Leica Biosystems) according to the manufacturer’s protocol. Mouse probes of Oca2 (ACDBio, 1072511), Gpr143 (ACDBio, 535108), Col12a1 (ACDBio, 312638) and Txnip (ACDBio, 457228) for the Leica System were used, and a RNAscope LS Multiplex Fluorescent assay (ACDBio) was used for detection. The double detection of immunofluorescence was performed after finishing in situ hybridization by applying the primary and secondary antibodies as described in the section ‘Immunofluorescence’. Skin sections were then scanned using Vectra Polaris (Akoya Biosciences) at ×20 for Opal fluors 570 (Akoya Biosciences, FP1488001KT), 690 (Akoya Biosciences, FP1497001KT) and DAPI (Akoya Biosciences, FP1490). Scanned images were processed using Inform (v.2.6.0) software (Aloya Biosciences). For quantitative analysis, HALO (v.3.5) software (module Indica Labs-FISH v.3.2.3) was utilized. FISH probe cell intensity (average intensity of FISH probe (×) spots and clusters per cell) was measured in tdTomato⁺ cells in the mid-anagen bulge/ORS^up and the bulb. The relative intensity of bulge/ORS^up tdTomato⁺ cells to bulb tdTomato⁺ cells within the same sample was compared across multiple samples.

Microscopy

For thin sections, images were taken at a single focal plane. For whole-mount tissues and thick sections, serial z-images were collected throughout the depth of each entire HF. Wide-field fluorescence images were taken with standard narrow-pass filters with an Eclipse Ti inverted microscope (Nikon) or an upright Axioplan (Zeiss). Images were processed to reconstruct a focused image using ImageJ/Fiji, Adobe Photoshop and the extended depth of focus function in the NIS-Elements software (Nikon, v.5.20.02).

The z-stack fluorescent images in Figs. 2a and 3b were taken using a LSM 880 confocal microscope with a ×63 NA/1.4 Plan Apochromat lens (Zeiss). The z-stacks were taken at 0.3 µm steps. Volumes were reconstructed using Imaris 9.5 software (Oxford Instruments).

Single-cell dissociation

Single melanocyte isolation was performed as previously published²⁴ but with slight modifications. To isolate a single melanocyte from anagen II HFs, Dct^rtTA;tetO^H2B-GFP mice were depilated at 8 weeks old and fed a doxycycline-containing diet (1 g kg^–1) for 4 days. At 4 days after depilation, when the HFs are at anagen II, mice were killed. The mice were then rinsed in betadine followed by 70% ethanol. The back skin of mice was collected. Scalpel blades were used to remove subcutaneous fat, and skin was rinsed in PBS and cut into 1 × 1 cm pieces followed by incubation in 0.25% trypsin for 1 h 30 min at 37 °C. Epidermis was separated from the dermis using forceps and scalpel blades, and the epidermis was finely chopped and incubated in 0.25% trypsin for 30 min at 37 °C while shaking at 100 r.p.m., followed by gentle pipetting to obtain a single-cell suspension. The obtained McSC suspension was filtered through a 70 µm nylon filter and centrifuged at 200 r.c.f. for 5 min and resuspended in medium A (DMEM, 10% FBS and 1× penicillin–streptomycin).

To isolate single bulb melanocytes, Dct^rtTA;tetO^H2B-GFP were fed a doxycycline-containing diet for 4 days at 5 weeks old, when the HFs are at the anagen VI stage. Isolation of bulb melanocytes was performed according to previously described methods⁵⁷ but with slight modifications. Mice were euthanized and rinsed in betadine followed by 70% ethanol. Skin was cut into 0.5 cm² sections and incubated in 5 mM EDTA (pH 8) and PBS for 2 h at 37 °C. Following incubation, connective tissue, the adipocyte layer and dermis were removed using forceps, and 0.5 cm² skin samples were cut into single rows of HFs and kept in medium A. HF bulbs were microdissected with a surgical blade, collected into medium A and centrifuged for 5 min at 200 r.c.f. Medium A was removed and hair bulbs were incubated in 1 ml of 0.2% collagenase II and 50 U ml^–1 dispase (9:1 solution) and shaken at 100 r.p.m. for 25 min at 37 °C. Next 400 U ml^–1 DNase I was added and incubated for 5 min at room temperature. Five volumes of medium A was added and then filtered through a 100 µm cell strainer. Cell suspension was pelleted by centrifuging at 200 r.c.f. for 5 min at 4 °C and resuspended in medium A.

Single GFP⁺ and DAPI-excluded live melanocytes were then isolated by cell sorting on a Sony SY3200 cell sorter equipped with a WinList 3D Analyzer (v.8.0) with a 100 µm nozzle. Single-cell suspensions from four mice of each condition were combined for subsequent scRNA-seq analyses. FlowJo 10.8.2 (Mac only) was used to plot the FACS gating strategy.

scRNA-seq and data analysis

Single melanocyte suspensions were loaded on a 10x Genomics Chromium instrument to generate single-cell gel beads in emulsion (GEMs). Approximately 5,000–10,000 cells were loaded per channel. The scRNA-seq library for differentiated bulb melanocytes was prepared using the following Chromium Single Cell 3′ v2 reagent kits: Chromium Single Cell 3′ Library & Gel Bead kit v2 PN-120237; Single Cell 3′ Chip kit v2 PN-120236; and i7 Multiplex kit PN-120262 (10x Genomics). The Single Cell 3′ Reagent kits v2 User Guide (Manual Part CG00052 RevA) was followed⁵⁸. The scRNA-seq library for anagen II melanocytes was prepared using the following Chromium Single Cell 3′ v3 reagent kits: Chromium Single Cell 3′ Library & Gel Bead kit v3 PN-1000075; Single Cell 3ʹ Feature Barcode Library kit PN-1000079; Single Cell B Chip Kit PN-1000073; and i7 Multiplex Kit PN-120262 (10x Genomics). The Single Cell 3′ Reagent kits v3 User Guide (Manual Part CG000201 RevA) was followed. Libraries were run on an Illumina NovaSeq 6000. The Cell Ranger Single-Cell software suite (v.6.0.1) was used to perform sample de-multiplexing, barcode processing and single-cell 3′ gene counting. The cDNA insert was aligned to the mm10/GRCm38 reference genome (mm10-2020-A). The undifferentiated telogen melanocyte dataset was downloaded from the NCBI Gene Expression Omnibus (identifier GSE113502) and re-processed with the same version of the Cell Ranger Single-Cell Software Suite (v.6.0.1) and mapped to the same version of reference genome (mm10-2020-A). Further analysis and visualization were performed using Seurat package (v.4.1.0)⁵⁹, using R Studio Desktop (v.1.4.1717) and R (v.4.1.2).

The Seurat object for each condition was generated from digital gene expression matrices. In the quality control step, the parameter of subset cells is nFeature_RNA (200–2000) for telogen melanocytes, nCount_RNA>10000 for anagen II melanocytes and nCount_RNA > 5000 for bulb melanocytes and percentage of mitochondria genes < 0.05 for all conditions. Different thresholds in filtering out low-quality cells were set for each condition owing to the differences of sequencing depth among the conditions. Data were then log scaled, centred and normalized to the number of Unique Molecular Identifier (nUMI). Principal components (PCs) were calculated using Seurat’s RunPCA function. The top 3,000 variable genes were used for calculating PCs.

For each condition, UMAP dimension reduction was performed on the normalized, centred, scaled nUMI count matrices using the first ten PCs. We then performed unsupervised clustering using the Seurat SNN clustering package, using a resolution of 0.6. Most clusters were identified as melanocytes based on expression for Dct, whereas each condition contained a minor cluster of epidermal cells positive for Krt10 or Krt14. The epidermal cells were excluded from subsequent analysis.

For comparative analysis, we merged melanocyte matrices from three conditions (telogen melanocyte, anagen II melanocyte and differentiated melanocyte) followed by a standard workflow of Seurat. Data were log normalized, scaled and centred after regressing out the effect of nCount_RNA (the total number of molecules detected within a cell from sequencing) and percent.mt (mitochondria ratio, defined by the PercentageFeatureSet function in Seurat). As the three conditions showed differences in sequencing depth (nCount_RNA varies greatly) and sequencing depth is a major cause of batch effects in scRNA-seq, the effect of nCount_RNA was regressed out before identifying PCs to account for such batch effects. The top 3,000 variable genes were used for calculating PCs. UMAP dimension reduction was performed on the normalized, centred, scaled nUMI count matrices using the first three PCs.

To estimate the cell cycle stage of a cell, we used Seurat’s cell cycle scoring. In brief, averaged relative expression of cell cycle related genes were used to calculate G2/M and S scores, which were used for binning cells into G2/M, S and G1/G0 bins. The cell cycle scores were regressed out in ScaleData step in the analysis of the anagen II melanocyte dataset alone, followed by PC calculation, UMAP dimension reduction and unbiased clustering as described above.

Pseudotime analysis was performed using the slingshot package (v.2.2.0)⁶⁰. The Seurat object was imported into slingshot using the as.SingleCellExperiment function. Then a pseudotime trajectory was constructed using the slingshot function with UMAP dimensional reduction.

Gene set enrichment analysis

Genes differentially expressed between anagen II melanocytes compared with telogen melanocytes (P < 0.05, log₂(fold change) > 0.25) were rank-ordered from high to low on the basis of their fold change. The pre-ranked gene list as queried for its enrichment in two annotated gene sets acquired from The Molecular Signature Database (MSigDB)—GOBP_DENDRITE_DEVELOPMENT and GOBP_DENDRITE_MORPHOGENESIS—using the preranked gene set enrichment analysis (GSEA) analysis tool^61,62. A false discover rate q-value of <0.25 was deemed significant.

Quantification and statistical analyses

The measurement of quantifications can be found in the y axes of bar plots in the figure and in figure legends. The statistical details of each plot can be found in the figure (N number). The exact meaning of N is described in the corresponding figure legend. No statistical methods were used to predetermine sample sizes, but our sample sizes were similar to those reported in previous publications^{7,24,38,39,40,55}. Pairwise comparisons between two groups were performed using two-tailed unpaired t-test. Comparisons of multiple groups were performed using one-way analysis of variance (ANOVA) or two-way ANOVA followed by multiple comparison test. Details of the statistical test are specified in the figure legends. Statistical significances were considered significant if P < 0.05. Exact P values are indicated in the figures and legends. Experimental data are shown as the mean ± standard deviation or mean ± standard error of the mean. Statistical analyses and plotting were done using GraphPad Prism (v.9.2.0) and Microsoft Excel (v.2016).

Material availability

Materials generated in this study can be provided upon reasonable requests by contacting the corresponding author.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.