Basal cell carcinomas arise from hair follicle stem cells in Ptch1(+/-) mice

Abstract

Basal cell carcinomas (BCCs) are hedgehog-driven tumors that resemble follicular and interfollicular epidermal basal keratinocytes and hence long have been thought to arise from these cells. However, the actual cell of origin is unknown. Using cell fate tracking of X-ray induced BCCs in Ptch1(+/-) mice, we found their essentially exclusive origin to be keratin 15-expressing stem cells of the follicular bulge. However, conditional loss of p53 not only enhanced BCC carcinogenesis from the bulge but also produced BCCs from the interfollicular epidermis, at least in part by enhancing Smo expression. This latter finding is consistent with the lack of visible tumors on ears and tail, sites lacking Smo expression, in Ptch1(+/-) mice.

Figure 1. K15- or K14-expressing cells are specifically marked with YFP in adult Y15 or Y14 mice

One day after the last dose of RU-486 or Tamoxifen, the skin of Y15 or Y14 mice were biopsied and immunostained with anti-K15, anti-K14 or anti-YFP antibodies. Color coding indicates antibody labeling. Scale bars = 50μm. (A) A selective image shows that K15-expressing cells are restricted in the bulge region and the K14-positive cells line the whole hair follicle including the isthmus region. (B) In Y15 mice, YFP was present in K15-expressing follicles but absent from IFE (dashed line). (C) In Y15 mice, some K15-expressing follicles failed to express YFP (arrows). (D) In Y14 mice, YFP was present in both follicles and in the IFE. (E) In Y14 mice, occasional regions of IFE or follicles (arrows) were free of YFP expression.

Figure 2. YFP-expressing microscopic BCCs arise from Y15 and Y14 mice

(A and B) The skin biopsies of Y15 (A) and Y14 (B) mice were immunostained with anti-YFP antibody (brown) and counterstained with hematoxylin (blue). Arrows denote the representative YFP-positive microscopic BCCs. Scale bars = 50μm. (C to F) Representative images of serial sections of a skin biopsy. A total of 15 serial sections of 5 μm in thickness per section were stained with H & E and examined for any hair follicle connection of every single microscopic BCC on all serial sections. Selective images represent sections from one sample that show no connection to the follicle in sections C and D and surrounds a follicle in sections E and F. Arrows indicate microscopic BCC. Scale bars = 50μm.

Figure 3. YFP-expressing macroscopic BCC tumors arise from Y15 mice

(A) A representative macroscopic BCC generated from Y15 mice was stained with anti-YFP antibody (brown) and counterstained with hematoxylin (blue). Scale bar = 50μm. (B) Examination of the normal-looking skin surrounding the tumor confirmed that follicles expressed YFP while the IFE was free of YFP expression. Scale bar = 50μm.

Figure 4. Deletion of p53 specifically in either K15- or K14-expressing cells accelerates BCC tumorigenesis in Ptch1+/− mice

(A) Tumor latency of macroscopic BCC development in PF15, PF14 and p53WT (combined data of Y15 and Y14) were analyzed with Kaplan-Meier curve (p<0.01). (B) The tumor multiplicity of microscopic and macroscopic BCC was assessed at mouse age 5 to 8 months. Fold differences between PF14 and PF15 mice and p-values are shown. Data are shown as means ± SD (n=8).

Figure 5. The ratio of IFE microscopic BCCs is increased in PF14 mice and appears associated with type I macroscopic BCCs

The fraction of microscopic BCCs that were free of follicle connection, i.e. the IFE microscopic BCCs, is plotted with the percentage of type II macroscopic tumors. Data are normalized against total number of tumors and shown as mean ± SD (n=4) with p values <0.05 (*) or <0.01 (**).

Figure 6. Two distinct histopathologic subtypes of macroscopic BCCs

(A and B) Representative images of type I histologic feature show branching, radiating tumor structure (arrow) and patchy expression of SMA (brown) in tumor cells. Scale bar = 50μm. (C and D) Representative images of type II histologic feature show nests of tumor cells and positive SMA-staining limited to the stroma. Scale bar = 50μm.

Figure 7. Differential expression of Smo

(A to C) Representative images of immunostaining with anti-Smo antibody (brown) and counterstained with hematoxylin (blue) show absence of staining in ear or tail and positive staining in dorsal skin follicles. (D and E)The skin biopsy of Y14 or Y15 mouse that is immunostained with anti-Smo antibody (green) shows that Smo is detected in follicles and some microscopic BCCs (arrows) but not in IFE (dashed line). (E) No DAPI counter staining. Scale bars = 50μm. (F and G)The skin biopsy of PF14 mouse that is immunostained with anti-Smo antibody (green) shows that Smo is detected in follicles and some microscopic BCCs (arrows) as well as in IFE (dashed line). (G) No DAPI counter staining. Scale bars = 50μm.

Figure 8. Differential localization of Cyclin B1 in BCCs developed from Ptch1+/− or Smo/M2 mouse model

Representative immunostained images with anti-Cyclin B1 antibody (red). Insets are images in higher magnification. Scale bars = 50μm. (A) Cutaneous tumor of Smo/M2 mouse shows cells with generally weak and cytoplasmic Cyclin B1 (arrows). (B) BCC of Ptch1+/− mouse has cells with strong nuclear Cyclin B1.