Experimental mouse models
The following mouse lines were used: Tyr-NrasQ61K, Tyr-rtTA, Tyr-CreERT2, Tyr(C-2J), BrafV600E, Trp53flox, Spp1−/−, Spp1flox, tetO-Spp1, Cd44−/−, Cd44flox, K14-Cre, K14-CreERT, K14-H2B-GFP, K14-Edn3, K14-Kitl, tdTomato, TOPGAL, Nude and SCID. Tissue-specific mouse models were produced by crossing either Cre-carrying or CreER-carrying animals with flox-ed gene carrying animals, or rtTA-carrying animals with tetO-carrying animals. All animal experiments followed all relevant guidelines and regulations and were approved by the Institutional Animal Care and Use Committee at China Agricultural University (to Z.Y.) and/or the Animal Care Committee at Gifu University (to T.K.) and/or the Animal Care and Use Committee of National Taiwan University (to C.-H.K.) and/or the Institutional Animal Care and Use Committee at University of California, Irvine (to B.A. and/or A.K.G. and/or M.V.P.) and/or the Institutional Animal Care and Use Committee at Central South University (to J.L.) and/or the Institutional Animal Care and Use Committee at Kyungpook National University (to J.W.O.).
Mouse induction protocols
Tetracycline-controlled overexpression of SPP1 in melanocytes was achieved in Tyr-rtTA;tetO-Spp1 mice with 2 mg ml−1 doxycycline hyclate (Sigma) in 5% sucrose and a doxycycline-containing diet (Bio-Serv, 200 mg kg−1) provided ad libitum. Inducible conditional gene recombination was achieved in CreER-carrying and flox-ed gene-carrying animals by intraperitoneal injection of tamoxifen (Sigma) in corn oil at a dose of 75 mg kg−1. In P2 animals, inducible conditional gene recombination was achieved by topical administration of (Z)-4-hydroxytamoxifen (4-HT; Sigma) in DMSO at 75 mg ml−1.
EdU pulse and pulse-chase assays
Mice were intraperitoneally injected with EdU (5 µg g−1 body weight) daily for seven consecutive days (pulse period), followed by an 8-week chase period. A portion of harvested skin was examined histologically using an EdU imaging kit (Thermo Fisher). Remaining skin portion was used to isolate cells for flow cytometry-based quantification using an EdU flow kit (Thermo Fisher). Triple-positive CD34+CD49f+EdU+ cells were used to quantify EdU+ bulge SCs.
Protein injection procedure
Intradermal delivery of protein-soaked agarose beads was performed as previously described8,11. In brief, recombinant SPP1 protein (441-OP, R&D) was reconstituted in 0.1% BSA to a final concentration of 1.3 mg ml−1. Affi-gel blue beads (Bio-Rad) were washed three times in sterile PBS, air dried and resuspended in reconstituted recombinant protein solution. Beads were incubated on ice for 1 h before implantation. For both recombinant protein and BSA controls, beads were implanted intradermally in P51–P53 animals. Bead implantation sites were resupplied with additional protein at 24, 48 and 72 h.
Skin wounding procedure
Mice were shaved and skin was cleaned with antiseptic. Surgery was conducted under continuous isoflurane anaesthesia. A full-thickness excisional wound was created without injuring the underlying fascia with dermal biopsy punch. Mice were given post-surgical analgesia: subcutaneous ketoprofen, followed by acetaminophen in drinking water.
Flow cytometry and FACS procedures
Dorsal skin was digested into single cells with Dispase II solution (Roche), followed by collagenase I solution (Life Technologies). Cells were filtered first through 70-µM and then 40-µM strainers. Viability dye (BioLegend) was used to exclude dead cells. Cell suspension was stained with primary antibodies in FACS staining buffer (1% BSA in PBS with 2 mM EDTA) for 30 min on ice before sorting. The following antibodies were used: mouse anti-γH2AX (1:100; 564718, BD Biosciences), mouse anti-TRP2 (1:50; sc-74439 AF647, Santa Cruz Biotechnology), rat anti-Ki67 (1:50; 58-5698-82, Thermo Fisher), rat anti-CD117 (1:100; 105812, BioLegend), rat anti-CD45 (1:50; 103108, BioLegend), rat anti-CD34 (1:50; 560230, BD Biosciences), rat anti-CD49f (1:100; 555736, BD Biosciences) and rabbit anti-SPP1 (1:100; 702184, Thermo Fisher). Cells were sorted on FACSAria II sorters (BD Biosciences) and flow cytometry analysis was performed on LSRII flow cytometer (BD Biosciences). Data were analysed with FlowJo software (version 10.8.0). Expression of SPP1 protein was detected using staining of both permeabilized cells (permeabilized condition) and non-permeabilized cells (surface-bound condition). Under permeabilized condition, we measured total SPP1 present in cells, whereas under surface-bound conditions, we measured SPP1 present on the cell surface, such as bound to its receptors. For permeabilization, cells were washed in PBS and resuspended at 1 million cells per 100 μl, permeabilization buffer was added and cells were stained following Fixation/Permeabilization kit instructions (BD Biosciences).
Primary melanocyte culture assay
Melanocytes were purified from P0 mouse skin by FACS as CD117+CD45neg populations. Sorted cells were then cultured in complete primary melanocyte media (RPMI 1640, 5% FBS, antibiotic–antimycotic, 2.5 ng l−1 basic human fibroblast growth factor, 10 μM ethanolamine, 1 mg ml−1 insulin, 1 μM O-phosphoethanolamine, 5 nM endothelin, 25 nM α-MSH and 50 ng ml−1 murine SC factor) at 37 °C with 5% CO2.
H2O2 treatment procedure
Cultured melanocytes in culture dishes or chamber slides were treated with H2O2 (Sigma) at 100 mM or vehicle (medium 254 and HMGS-2) for 2 h at 37 °C. Treated cells were rinsed twice with PBS.
DiI labelling procedure
Cells were labelled with DiI dye (Thermo Fisher) following the manufacturer’s instructions. In brief, cells were incubated for 15 min at 37 °C in culture medium supplied with 5 µl of the cell-labelling solution per 1 ml. After labelling, cells were dissociated with Accutase (Stemcell Technologies), followed by two washes with PBS.
Cell injection procedure
Cells were counted using a haemocytometer and then diluted to 2,000 cells per microlitre in cell culture medium. Of cell suspension, 10–50 µl was slowly injected intradermally into the dorsal skin of recipient mice using a 29-G needle.
Skin micro-grafts containing four to six anagen HFs were transplanted to the dorsal skin of 6-to-8-week-old female SCID or Nude mice, as previously described8. Thirty days post-grafting, 10 µl of recombinant protein or saline was microinjected to the HF grafting site for 3 consecutive days. Host mice were euthanized on post-grafting day 50 and skin was analysed on wholemount.
ABT-737 treatment procedure
Mice were subcutaneously injected twice (on days P10 and P12) with ABT-737 (Cayman Chemical) or vehicle control at a dose of 75 mg kg−1.
For β-galactosidase staining, thick sections (20 µm) were incubated in 1 mg ml−1 X-gal substrate in PBS with 1.3 mM MgCl2, 3 mM K3Fe(CN)6 and 3 mM K4Fe(CN)6 at 37 °C overnight. For senescence-associated β-gal staining, cells were stained using a kit (Cell Signaling) according to the manufacturer’s instructions. In brief, cells were fixed with fixative solution provided by the manufacturer for 15 min at room temperature, followed by acidic β-gal detection using pH 6.0 staining solution overnight at 37 °C.
For paraffin-embedded sections, skin samples were fixed with 4% (vol/vol) paraformaldehyde overnight at 4 °C. Histological sections were permeabilized for 15 min in PBS + 0.1% Triton X-100 (PBST) and blocked for at least 1 h at room temperature with PBST + 3% BSA. Mouse antibodies were blocked with the M.O.M. block kit (Vector Laboratories). Primary antibodies were incubated overnight at 4 °C and secondary antibodies were incubated for 1 h at room temperature. The following primary antibodies were used: rabbit anti-γH2AX (1:300; 9718, Cell Signaling), rabbit anti-TRP2 (1:200; ab74073, Abcam), rabbit anti-TRP2 (1:200; ab103463, Abcam), mouse anti-PCNA (1:1,000; ab29, Abcam), rat anti-CD34 (1:100; 14-0341-82, Thermo Fisher), rabbit anti-SOX9 (1:200; AB5535, Millipore), goat anti-SPP1 (1:100; AF808, R&D), goat anti-SPP1 (1:300; AF1433, R&D), rabbit anti-KRT14 (1:2,000; ab119695, Abcam), rabbit anti-CD44 (1:100; PA5-94934, Thermo Fisher), rabbit anti-SOX10 (1:100; ab180862, Abcam), rabbit anti-KRT5 (1:1,000; 905501, BioLegend) and goat anti-Pcad (1:200; AF761, R&D Systems). The following secondary antibodies were used: donkey anti-rat AF555 (1:1,000; ab150154, Abcam), donkey anti-rabbit AF555 (1:1,000; A31572, Thermo Fisher), donkey anti-mouse AF555 (1:1,000; A31570, Thermo Fisher), donkey anti-rabbit AF488 (1:1,000; A21206, Thermo Fisher), donkey anti-goat AF488 (1:1,000; A11055, Thermo Fisher), goat anti-rat AF488 (1:1,000; A11006, Thermo Fisher), goat anti-rabbit AF488 (1:1,000; 4412s, Cell Signaling), goat anti-mouse AF555 (1:1,000; 4409s, Cell Signaling) and goat anti-rabbit AF555 (1:1,000; 4413s, Cell Signaling).
RNA staining was performed using the Multiplex Fluorescent v2 kit (Advanced Cell Diagnostics). In brief, skin was frozen in OCT compound and sectioned at 12–15 µm. Sections were fixed at room temperature for 1 h with 4% paraformaldehyde in PBS, followed by standard manufacturer’s protocols (Advanced Cell Diagnostics). RNA probes for hybridization were purchased from Advanced Cell Diagnostics and included Mm-Spp1 (catalogue no. 435191-C1), Mm-Dct-C2 (Trp2; 460461-C2), Mm-Cdkn2b (p15; 458341-C1), Mm-Cdkn2a (p16; 411011-C1), Mm-Mki67-C3 (Ki67; 416771-C3) and Mm-Aurkb (461761-C1).
Western blot assay
Single sorted melanocytes or cells from mouse whole-back skin were lysed in RIPA buffer (Sigma) containing a cocktail of protease inhibitors (Thermo Fisher). Of each cell lysate, 25 µg was loaded onto a 12% separating Bis-Tris gel. Proteins were transferred to a nitrocellulose membrane. Membrane was incubated with primary goat anti-mouse SPP1 antibody (1:100; AF808, R&D) or rabbit anti-β-actin antibody (1:1,000; 4967, Cell Signaling) at a concentration of 2.5 μg ml−1. The blot was developed with Enhanced Chemiluminescence Plus Developer (Fisher Scientific).
SPP1 levels in the supernatant of cell cultures were measured by a mouse OPN/SPP1 ELISA kit (Thermo Fisher) according to the manufacturer’s instructions. In brief, SPP1 concentration was calculated by generating a standard curve from recombinant SPP1 protein diluted between 0 and 2,000 pg ml−1. Microplates were measured using a Synergy microplate reader (BIO-TEK) at a wavelength of 450 nm.
Real-time PCR assay
Total RNA from sorted cells was extracted using RNeasy Micro Kit (Qiagen) coupled with its on-column DNase digestion protocol. Total RNA was then reverse-transcribed with Superscript III (Life Technologies) in the presence of oligo-dT. Full-length cDNA was normalized to an equal amount using housekeeping genes GAPDH or 18S. Primers are listed in Supplementary Table 6.
Sorted GFP-expressing HF bulge SCs and hair germ progenitors from K14-H2B-GFP mice were plated onto 3T3 fibroblast feeder layer cells, pre-treated with mitomycin C to induce cell cycle arrest. Cells were co-cultured at 37 °C in William’s E medium supplemented with calcium and antibiotic–antimycotic. Medium was replaced after 48 h, and the attachment rate was evaluated following an additional 12 h of culture. Attached cells were passaged upon confluence, which was achieved every 4–6 days. Calcium-supplemented culture medium was changed every 2–3 days. In other experiments, bulge SCs were FACS sorted as CD34+CD49f+ cells and cultured at a concentration of 1,000 cells per squared centimetre, in the presence of mitomycin C inactivated 3T3 fibroblasts. After 2 weeks, 0.5% crystal violet (Sigma) solution made in 1:1 ratio of water:methanol was added to each culture well. Stained plates were then rinsed with water, air dried and imaged.
Human skin samples
Collection of human skin samples followed all relevant guidelines and regulations and was approved by the Research Ethics Committee at National Taiwan University Hospital and/or the Medical Ethics Committee at Kyungpook National University Hospital and/or the Ethics Committee of Xiangya Hospital, Central South University and comply with guidelines from the Ministry of Science and Technology (MOST) of the People’s Republic of China. All participants provided written informed consent. No identifiable images of human research participants are shown.
Bulk and single-cell RNA-seq for mouse tissue
For bulk RNA-seq, total RNA was extracted from FACS-sorted cells in biological triplicates with an RNA integrity number of more than 9.1, and 1 ng of mRNA was used for full-length cDNA synthesis, followed by PCR amplification using Smart-seq2. The libraries were sequenced on the Illumina Next-Seq500 system to an average depth of 10–30 million reads per library using paired 43-bp reads.
For single-cell RNA-seq, cells were captured using the Fluidigm C1 chips as per the manufacturer’s protocol. A concentration of 200,000–350,000 cells per millilitre was used for chip loading. After cell capture, chips were examined visually under the microscope to determine the capture rate, and empty chambers or chambers with multiple cells were excluded from the analysis. cDNA was synthesized and amplified on the Fluidigm C1 Single-Cell Auto Prep System with the Clontech SMARTer Ultra Low RNA kit and the ADVANTAGE-2 PCR kit (Clontech). Single-cell RNA-seq libraries were constructed in 96-well plates according to the Fluidigm C1 manual. Multiplexed libraries were analysed on Agilent 2100 Bioanalyzer for fragment distribution and quantified using Kapa Biosystem’s universal library quantification kit. Libraries were sequenced as 75-bp paired-end reads on the Illumina Next-Seq500 platform.
For both bulk and single-cell RNA-seq, reads were first aligned using STAR v.2.4.2a with parameters ‘–outFilterMismatchNmax 10 –outFilterMismatchNoverReadLmax 0.07 –outFilterMultimapNmax 10’ to the reference mouse genome (mm10/genocode,vM8). Gene expression levels were quantified using RSEM v.1.2.25 with expression values normalized into fragments per kilobase of transcript per million mapped reads (FPKM). Samples with more than 1,000,000 uniquely mapped reads and more than 60% uniquely mapping efficiency were used for downstream analyses. Differential expression analysis was performed using edgeR v.3.2.2 on protein-coding genes and long non-coding RNAs. Differentially expressed genes were selected by using fold change ≥ 2, false discovery rate < 0.05 and counts per million reads ≥ 2.
Bulk RNA-seq for human tissue
RNA was extracted from human hairy nevus skin as well as normal skin from nevus edge using the Qiagen RNA extraction kit. cDNA was synthesized using the Superscript III First-strand synthesis system (Invitrogen) and quantified using the Agilent Bioanalyzer. Bulk RNA-seq analysis was performed using the standard pipeline. In brief, pair-end RNA-seq reads were aligned using STAT/2.5.1b to the human reference genome hg38. Gene expression was measured using RESM/1.2/25 with expression values normalized into FPKM.
Single-cell data analysis
For all single-cell data analysis, low-quality cells were filtered out and the same normalization was performed to eliminate cell-specific biases. For each cell, we calculated three quality control metrics: the number of expressed genes, the total number of transcripts and the proportion of transcripts in mitochondrial genes. The single-cell data matrix was column-normalized (divided by the total number of transcripts and multiplied by 10,000) and then log-transformed with pseudo-count +1.
For single-cell RNA-seq data on bulge SCs, cells from P30 WT, P56 WT and P56 Tyr-NrasQ61K samples were combined, and the expression of genes with multiple Ensembl IDs was averaged. For quality control, cells with the total number of TPM counts of less than 750,000, with the proportion of TPM counts in mitochondrial genes of more than 20% and with the number of expressed genes of more than 7,000 or less than 2,000 were removed. In summary, 20 cells were removed, leading to 256 cells for downstream analyses. Clustering of cells was performed using the Seurat R package (V2.3). Principle component analysis (PCA) was first performed using highly variable genes, which were identified with an average expression of more than 0.01 and dispersion of more than 1. We regressed out the effects of the total number of transcripts and the transcripts in mitochondrial genes. The top 17 PCs were selected based on the Jackstraw method (JackStraw function). Using these top PCs, the Louvain modularity-based community detection algorithm was used to obtain cell clusters with resolution being 1.1, giving five clusters. The likelihood-ratio test was used to perform differential gene expression analysis between the clusters. Genes with a P value of less than 0.01 and a log fold change greater than 0.25 were considered as differentially expressed. To visualize cells onto a two-dimensional space, we performed t-distributed stochastic neighbour embedding. The relatedness of cell clusters was determined by performing unsupervised hierarchical clustering of average gene expression of cell clusters using the highly variable genes (correlation distance metric and average linkage). To determine the cell cycle phase of each cell, we used cell cycle-related genes, including a core set of 43 G1/S and 54 G2/M genes. For each cell, a cell cycle phase (G1, S and G2/M) was assigned based on its expression of these cell cycle-related genes using the CellCycleScoring function in Seurat.
Statistics and reproducibility
Sample size calculations were not performed for mouse experiments, but n = 3 is a standard minimal sample size that in our previous studies was found to be sufficient to assess changes in hair growth in mice. Group sizes in animal experiments were derived from the power analysis performed on preliminary experimental data. Animals of both sexes were used, and analyses were not segregated by sex. Age of animals is defined in all experiments in postnatal days. Statistical analyses were performed using unpaired one-tailed or two-tailed (defined in the figure legends) Student’s t-tests. In all bar charts shown in figures, error bars are mean ± s.d. Statistical significance degree in figures is defined as follows: P ≥ 0.05 (not significant), *P ≤ 0.05 and **P ≤ 0.01; exact P values are provided in the figure legends. Differentially expressed gene analysis on RNA-seq data, reported in Supplementary Tables 1, 3, 4 and 5, was done using the edgeR package. When comparing gene expression between groups, the exact test (exactTest() function, two-sided) was performed for P value calculation after the negative binomial models were fitted and dispersion was calculated. P values were adjusted by using Benjamini and Hochberg’s approach for false discovery rate output. For gene ontology terms reported in Supplementary Tables 1, 3, 4 and 5, analysis was done using Metascape. P values were calculated using hypergeometric test, and then adjusted by using Benjamini and Hochberg correction. Exact P values are reported in the above-mentioned tables. All experiments were repeated independently with similar results of three times or more, and data shown in the figures are from representative experiments. The number of independent repeats for the representative experiments shown as micrographs are as follows: Fig. 1c (n = 3), Fig. 1d (n = 5), Fig. 1e (n = 3), Fig. 1k (n = 5), Fig. 2j (n = 3), Fig. 3g (n = 5), Fig. 4a–d (n = 3 each), Fig. 5d–h (n = 3 each), Extended Data Fig. 1a,b,d,f–h (n = 3 each), Extended Data Fig. 1c (n = 6), Extended Data Fig. 1e (n = 7), Extended Data Fig. 1j,k (n = 3 each), Extended Data Fig. 2a (n = 4), Extended Data Fig. 2d (n = 4), Extended Data Fig. 3j (n = 4), Extended Data Fig. 3k (n = 4), Extended Data Fig. 4a (n = 5), Extended Data Fig. 4l (n = 3), Extended Data Fig. 5a (n = 6), Extended Data Fig. 5f–j (n = 3 each), Extended Data Fig. 6a,b (n = 3 each), Extended Data Fig. 7i (n = 3), Extended Data Fig. 8a–c (n = 4 each), Extended Data Fig. 8d (n = 3), Extended Data Fig. 8e (n = 5), Extended Data Fig. 8f (n = 5), Extended Data Fig. 8g (n = 4), Extended Data Fig. 9e (n = 3), Extended Data Fig. 9l,m (n = 3 each). Experiments were not randomized or performed in a blinded manner, except where noted.
Schematics were prepared using Adobe Illustrator.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.