【Science】重磅发现：超85%的化学诱导皮肤肿瘤，竟然起源于上毛囊的“长寿”干细胞#

(A) Illustration showing the skin SC-based lineage tracing approach used in combination with DMBA / TPA carcinogenesis. (B) Diagram of the location of specific skin SC markers of the HF and IFE and corresponding CreER mouse strains used for lineage tracing. (C) Representative images of the initial tdTomato labeling (red) of Krt19+ and Lgr5+ bulge hfSCs, _Lgr6_GFP+ SCs in the Mid HF and overlying IFE and Lrig1+ SCs of the Up HF, one day after TAM treatment. (D) Representative brightfield and fluorescence images of tdTomato fluorescence (red) in a fully labeled (left panel) and a largely unlabeled PAP of unknown origin (right panel) after chronic TPA treatment. Lower panels show representative cryosections of abundant tdTomato+ fluorescence within fully labeled PAP epithelium (left, lower) and red epithelial “streaks” (right, below), respectively. (E) Quantification of the percentage of PAPs labeled with tdTomato+ fluorescence from each SC compartment after chronic TPA promotion (number of mice per group: Krt19, n = 7; Lgr5, n = 18, Lgr6, n = 30, Lrig1, n = 9). The unpaired, two-tailed Student’s t test was used to calculate p values with * p < 0.05, ** p < 0.01 and *** p < 0.001. (F) Percentage of fully labeled tdTomato+ PAPs derived from each SC compartment after chronic TPA treatment with corresponding bars showing the percentage of labeled PAPs per individual mouse. DAPI counterstaining (blue) was used to label cell nuclei [(C) and (D)]. Abbreviations – Bu: bulge, DP: dermal papillae, HG: hair germ, DAPI: 4’,6-diamidino-2-phenylindole, SG: sebaceous gland, IFE: interfollicular epidermis, Up HF: upper hair follicle, Mid HF: mid-hair follicle, +TAM: tamoxifen treated, hfSCs: hair follicle stem cells, PAP: papilloma, SC: stem cell. Scale bar = 50 μm. Elements of images were created using BioRender.#

(A) Schematic showing the DOX-inducible, Lgr6-targeted KD approach during skin carcinogenesis. (B) PAP development over time in WT and Lgr6 KD-Het mice. (C) Mean CA burden in WT and Lgr6 KD-Het mice after chronic TPA promotion (total number of CAs present per treatment group: Lgr6 WT + DOX = 43 /19 mice; Lgr6 KD-Het + DOX = 20 CAs / 15 mice; Lgr6 WT + sucrose = 40 CAs / 16 mice and Lgr6 KD-Het + sucrose = 33 CAs / 10 mice). The unpaired, two-tailed Student’s t test was used to calculate p values with ns: non-significant, * p < 0.05, ** p < 0.01 and *** p < 0.001. (D) Graph representing the percentage of WT and Lgr6 KD-Het mice lacking CA development over time (CA-free survival - defined as the time point in weeks after the first TPA treatment was administered at which the 1st CA was observed on the dorsal skin). Abbreviations - CA: carcinoma, DOX: doxycycline, KD: knockdown, Het: heterozygous, PAP: papilloma, SC: stem cell, WT: wild type. Elements of images were created using BioRender.#

(A) Representation of the Lgr6+ SC-specific six month delay, lineage tracing approach used with DMBA / TPA carcinogenesis. (B) The percentage of fully labeled tdTomato+ PAPs per mouse following delayed chronic TPA treatment of DMBA-initiated, TAM-labeled _Lgr6_GFP / tdTomato mice (n = 7). (C) Ras mutation analysis of tdTomato+ PAPs generated from regular DMBA / TPA (n = 21) and delayed DMBA / TPA treatment (n = 27). (D) Trinucleotide signature of FACS-purified, non-exposed and DMBA-initiated, viable skin cell populations two weeks and one year after the initial exposure. The trinucleotide mutational signature induced by DMBA (25, 41) (grey bars; A > T / T > A) persisted for at least one year after treatment. (E) Trinucleotide signature analysis of non-exposed and DMBA-initiated, FACS purified, viable _Lgr6_GFP+ SCs two weeks and one year after the initial exposure. Red bars indicate the C > T trinucleotide signature corresponding to the classical SBS1 “clock” signature that increased over one year, independent of treatment with DMBA [(D) and (E)]. (F) Duplex-seq of DMBA-initiated, viable skin cells and _Lgr6_GFP+ SCs following four weeks of TPA promotion. A blue circle indicates the presence of a rare “singlet” mutation (single mutant cell) within the sample, a pink circle indicates a “multiplet” mutation indicating two or more cells with a particular mutation (potential clonal expansion) and the observed VAF value is displayed in parentheses below each mutation. Multiplet DMBA-associated mutations are highlighted in bold. (G) Representative cryosection of an early skin lesion labeled by Lgr6-derived, tdTomato+ cells (red) after six weeks of TPA treatment (DAPI, blue, was used to label cell nuclei; scale bar = 100 μm). Abbreviations – HF: hair follicle, DAPI: 4’,6-diamidino-2-phenylindole, +TAM: tamoxifen treated, PAP: papilloma, SC: stem cell, VAF: variant allele frequency. In (B) the unpaired, two-tailed Student’s t test was used to calculate p values with ns: non-significant, * p < 0.05, ** p < 0.01 and *** p < 0.001. Elements of images were created using BioRender.#

从深层机制与细胞动态来看，这些潜伏在上毛囊的受损干细胞展现出了惊人的长寿与休眠特性，在受到化学致癌物引发突变后，它们能在皮肤中隐匿长达一年之久而不被免疫系统清除，且在重新接触肿瘤促进剂（如TPA）时依然能被精准唤醒并疯狂扩张。单细胞测序揭示，这与上毛囊干细胞中高度富集 Cyp1b1 和 Ephx1 这两种致癌物代谢关键酶密切相关，使其天生具备更高的致癌物敏感性。此外，研究还意外揭示了一种此前不为人知的克隆竞争抑制机制：在正常状态下，体内自发产生 Kras 突变的细胞虽能受到促进剂的短暂刺激而扩张，但随后会被携带 Hras 突变的上毛囊起源细胞无情地竞争并清除；只有当 Hras 基因被彻底敲除后，这些自发突变的 Kras 克隆才能摆脱束缚，大肆发展为恶性皮肤癌。

(A) Normal dorsal mouse skin transcriptome clustering using Seurat. (B) Enrichment and replotting of Hras+ cells from normal mouse skin. (C) Clustering of Hras+ cell populations using Seurat. (D) UMAP plot of Hras+ skin cells with subcluster expression of a DMBA metabolism enzyme, Ephx1. (E) qPCR of Ephx1 expression in viable global skin cells, _Lgr5_GFP- bulge hfSCs and _Lgr5_GFP+ SCs isolated from unexposed skin (pale blue) and skin 24 hours after DMBA exposure (dark blue) in vivo. (F) qPCR of Ephx1 expression in viable global skin cells, parental _Lgr6_GFP+ SCs and sub-fractionated _Lgr6_GFP+ SCA1- HF SCs and _Lgr6_GFP+ SCA1+ IFE SCs isolated from unexposed skin (pale blue) and skin 24 hours after DMBA exposure (dark blue). (G) UMAP of Hras+ cells isolated from unexposed mouse skin with subcluster expression of a DMBA metabolism enzyme, Cyp1b1. (H) qPCR of Cyp1b1 expression in viable global skin cells, _Lgr5_GFP- bulge hfSCs and _Lgr5_GFP+ SCs isolated from unexposed skin (pale green) and skin 24 hours after DMBA exposure (dark green). (I) qPCR of Cyp1b1 expression in viable global skin cells, parental _Lgr6_GFP+ SCs and sub-fractionated _Lgr6_GFP+ SCA1- HF SCs and _Lgr6_GFP+ SCA1+ IFE SCs isolated from unexposed skin (pale green) and skin 24 hours after DMBA exposure (dark green). (J) Bar chart showing the percentage of cells displaying CYP1B1 protein expression in purified, _Lgr5_GFP- bulge hfSCs, _Lgr5_GFP+ SCs, the whole parental _Lgr6_GFP+ SC fraction, sub-fractionated Lgr6+ SCA1- HF SCs, _Lgr6_GFP+ SCA1+ IFE SCs and LRIG1+ SCs isolated from unexposed skin (pale green) and skin 24 hours following DMBA exposure (dark green) in vivo (n = 5 independent biological replicates per SC type and exposure). (K) Illustration representing the in vivo expression of DMBA metabolism enzymes, Ephx1 and Cyp1b1, in HF _Lgr6_GFP+ HF SCs mouse skin. Abbreviations – SG: sebaceous gland, Bu: bulge, HF: hair follicle, hfSCs: hair follicle stem cells, IFE: interfollicular epidermis, LCs: Langerhans cells. The unpaired, two-tailed Student’s t test was used to calculate p values with * p < 0.05, ** p < 0.01 and *** p < 0.001. Elements of images were created using BioRender.#

(A) Illustration showing the approach used to collect carcinogen-exposed mouse skin samples for targeted, Duplex-seq to detect the presence of Ras codon G12, G13, and Q61 mutations. (B) Duplex-seq detection of Ras hotspot mutations in individual control, unexposed skin samples (n = 10, red boxes highlight the presence of spontaneous codon 12 and 13 mutations). (C) Detection of Ras hotspot mutations in individual DMBA-initiated skin samples (n = 10, red boxes highlight the presence of codon 61 mutations). (D) Duplex-seq detection of Ras hotspot mutations in DMBA-initiated and TPA-promoted skin samples (n = 10, red boxes highlight the presence of codon 61 mutations). In (B) to (D), a blue circle indicates the presence of a rare “singlet” mutation within the sample (a single mutant cell) and a pink circle indicates a “multiplet” mutation, indicating two or more cells with a particular mutation. Note: in a single control animal [(B), top right], mutations were observed in Hras Q61L, Kras G12D, and Kras G13R, each as very small multiplets revealing minimal clonal expansion without any treatment - this was the only animal to display this pattern, which was not typical. Elements of images were created using BioRender.#

(A) Targeted Duplex-seq reveals the outgrowth of cells with spontaneous Kras G12D and Kras G13R mutations in response to short term TPA promotion in vivo. Each plot represents the presence of hotspot Ras mutations observed in an individual TPA-treated mouse skin sample (n = 10 biological replicates, a blue circle indicates the presence of a rare “singlet” mutation and a pink circle indicates the presence of a “multiplet” mutation (red boxes highlight codon 12 and 13 mutations). (B) Plot representing the estimated number of mutant cells per million for the most common Ras hotspot mutations observed in uninitiated and DMBA-initiated skin in the absence or presence of TPA promotion, and the effect observed after withdrawal of TPA promotion for four weeks. (C) Comparison of the frequency of Q61H and Q61L mutations in Nras (light and dark blue, respectively), Kras (olive and dark green, respectively) and Hras (yellow and orange, respectively) in each of the treatment groups. (D and E) Duplex-seq detection of the frequency (D) and mutated reads per sample (E) for Hras Q61L (orange) and Kras G13R (light green) mutations in each of the treatment groups. (F) Ras driver mutations observed in tumors generated from each of the DMBA-initiated and TPA-promoted groups (the number of tumors analyzed per group is displayed above each bar in parentheses). (G) Detection of Kras mutations observed in Hras KO papillomas (n = 22 tumors) and carcinomas (n = 5 tumors) generated from DMBA + TPA treatment. The unpaired, two-tailed Student’s t test was used to calculate p values with * p < 0.05, ** p < 0.01 and *** p < 0.001. Abbreviations – VAF, variant allele frequency, KO: knockout, WD: withdrawal. Elements of images were created using BioRender.#