Bone marrow-derived epithelial cells and hair follicle stem cells contribute to development of chronic cutaneous neoplasms

Abstract

We used allogeneic bone marrow transplantation (BMT) and a mouse multistage cutaneous carcinogenesis model to probe recruitment of bone marrow-derived epithelial cells (BMDECs) in skin tumors initiated with the carcinogen, dimethylbenz[a]anthracene (DMBA), and promoted with 12-O-tetradecanolyphorbol-13-acetate (TPA). BMDECs clustered in the lesional epithelium, expressed cytokeratins, proliferated, and stratified. We detected cytokeratin induction in plastic-adherent bone marrow cells (BMCs) cultured in the presence of filter-separated keratinocytes (KCs) and bone morphogenetic protein 5 (BMP5). Lineage-depleted BMCs migrated towards High Mobility Group Box 1 (HMGB1) protein and epidermal KCs in ex vivo invasion assays. Naive female mice receiving BMTs from DMBA-treated donors developed benign and malignant lesions after TPA promotion alone. We conclude that BMDECs contribute to the development of papillomas and dysplasia, demonstrating a systemic contribution to these lesions. Furthermore, carcinogen-exposed BMCs can initiate benign and malignant lesions upon tumor promotion. Ultimately, these findings may suggest targets for treatment of non-melanoma skin cancers.

Fig. 1

CD34⁻, CD44⁺ BMCs express keratin after BMC/KC co-culture and BMP5 treatment. a, b All adherent BMCs are CD34-negative and CD44-positive. c, e A sub set of adherent BMCs are pan-keratin-immunoreactive (arrowheads) after 7days of BMC/KC co-culture (keratin-positive BMCs in white boxes are magnified and merged with phase image). d Pan-keratin-immunoreactive BMC (arrowhead) identified 10 days after BMP5 treatment. f K14-immunoreactive cells (arrowheads) 10 days after BMP5 treatment (white box area is magnified). g Histogram of number of keratin-expressing BMCs; BMCs without treatment, BMC/KC co-culture (pan-keratin-positive BMCs, gray bar) and BMP5 treatment (K14-positive BMCs, black bar), pan-keratin- and K14-immunoreactive BMCs are detected in KC co-cultured and BMP5-treated BMCs, but no keratin-positive cells are detected in treatment controls (n = 3, 3 different culture groups, 3 male and 3 female mice in each group; P < 0.017 as determined by Student’s t-test, mean ± s.d.) h Q-RT-PCR results show the relative level of K14 expression detected from positive (primary KCs and Tg.AC, a KC cancer cell-line) and negative (Swiss mouse 3T3 cell-line and primary BMCs) control groups, and experimental (BMC/KC co-culture and BMP5 treatment) groups. The expression levels were normalized to GAPDH expression levels, and expression mean values were converted into fold-change gene expression. Final values were normalized with log₂ transformation. (n = 3, P < 4.81 × 10E−12 as determined by Fisher’s ANOVA method, mean ± s.d.). *White scale bar, 50 µm; Red scale bar, 200 µm

Fig. 2

Long-term TPA treatment increases BMDCs in the epidermis of the skin. a BMDCs were detected in the epidermis of the skin with different treatment groups (before TPA treatment, no treatment, acetone-treated, and TPA-treated) and specific areas in white and magenta boxes are magnified. Arrows indicate keratin-negative BMDCs and arrowheads indicate keratin-positive BMDCs. b The number of GFP-positive cells in the epidermis are compared among three different treatment groups (no treatment, acetone-, and TPA-treated) at three different time points. The treatment types (no treatment, acetone, and TPA) are different across all weeks (P = 5.35 × 10E−11). The average time effects are different across all treatment levels (P = 1.32 × 10E−11). Additionally, the treatment type and time interact in a non-additive fashion (P = 6.6 × 10E−10). (n = 3, over all P value = 5.72 × 10E−13 as determined by the F-statistic, mean ± s.d.). c Keratin-positive BMDCs were observed in the basal layer, the suprabasal layer, and hair follicle. Areas in white boxes are magnified or shown in single color. Arrows indicate keratin negative BMDCs and arrowheads indicate keratin-positive BMDCs. d The number of GFP/Keratin-positive (double positive) BMDCs in the epidermis are compared among three different treatment groups (no treatment, acetone-, and TPA-treated) at three different time points. The number of double positive cells are increased in long-term TPA-treated groups. Both treatment type (acetone and TPA) and time (in weeks) have significant effects. This implies that the means of treatment types (no treatment, acetone, and TPA) have different effects across all weeks (P = 3.87 × 10E−08). The average time effects are different across all treatment types (P = 1.95 × 10E−07). Finally, there is a significant interaction term; this means that treatment type and time interact in a non-additive fashion (P = 1.53 × 10E−07). (n = 3, overall P value = 1.22 × 10E−09 as determined by the F-statistic, mean ± s.d.). *Epidermis (E), Dermis (D), Basal layer (B), Hair follicle (HF) **All counted cells are DAPI positive ***White scale bar, 50 µm

Fig. 3

Keratin-expressing BMDCs are in the basal layer of the papilloma epithelium. a, b GFP-positive BMDCs (red) were detected in the a papilloma epithelium and b junctional area of papilloma and adjacent hyperplastic epidermis in paraffin sections of a papilloma. c, d Groups of pan-keratin-positive BMDCs (areas of yellow broken line and white arrowheads) were detected in basal parts of the papilloma epithelium in paraffin sections (keratin-negative BMDCs: white arrows). Areas in white and magenta box in figure d are magnified. e A group of K14-positive BMDCs (areas of yellow broken line) were detected in basal epithelium of frozen papilloma section (40 µm) using a multiphoton confocal microscope (note: tumor stroma was washed out in the staining processes). Area in white box is magnified. f, g BMDCs were detected in tumor-adjacent f hyperplastic epidermis, and g hair follicles. h Pan-keratin-positive BMDCs were identified in HFs (white boxes) under the tumor mass. i BMDCs were identified in deregulated HF structure under the tumor mass. *Black and white scale bars, 50 µm

Fig. 4

BMDECs are recruited in ulcer-associated dysplasia. a Dysplastic skin shows extremely abnormal epithelial structure (red box area is magnified). Non-dysplastic (mildly hyperplastic) epidermis adjacent to dysplasia (blue box area is magnified). b BMDCs (red) are observed in the dysplastic epithelium (red box area is magnified). c BMDCs (red) are identified in dysplastic epidermis (red box, c-I) but not in non-dysplastic-adjacent epidermis (blue box, c-II). BMDCs are not observed in non-dysplastic epidermis adjacent to dysplasia (c-III). d Clusters of keratin expressing BMDCs are identified in the basal region of dysplastic epithelium (white box area is magnified). e, f A significant contribution of e pan-keratin-expressing and f K14-expressing BMDECs are detected in dysplastic epithelium using a multiphoton confocal microscope. g Keratin-expressing BMDCs are identified in HF adjacent to a dysplastic wound (broken line boxes). White box area is magnified. h A group of GFP-positive cells (white box) are directly detected in HF in the dysplastic area of frozen skin sections using the GFP-channel. i The number of Y chromosome-positive donor-derived BMCs (white arrowheads) are identified in the frozen dysplastic skin section of female recipients. j, k K14/Y chromosome-positive (double positive) cells (white arrowheads) were identified in the dysplastic epithelium of frozen sections of female recipients. j A micrograph with high magnification and k a micrograph with low magnification of different samples and white box area is magnified. *Black and white scale bar, 50 µm; Red scale bar, 200 µm

Fig. 5

BMDECs in the papillomas and dysplasia are proliferating. a BrdU-positive BMDECs (arrowheads, white box area is magnified) are identified in the outer root sheath area of a deregulated HF under the tumor mass. b BMDECs in epithelium of papilloma (a serial section of Fig. 3a). b-I A group of Ki67-expressing BMDECs (area of yellow broken line and arrowheads) are identified in the basal epithelium of the papilloma. b-II Cytoplasmic beta-catenin expression was detected from BMDECs (areas of red broken line) in the epithelium of the papilloma. c BMDECs in deregulated HF structure under the tumor mass (a serial section of Fig. 3i). c-I Few Ki67-positive BMDCs (arrowheads) were identified in the section and c-II BMDCs in this area show membrane beta-catenin expression (GFP-channel in white box shows membrane expression). d BMDECs in tumor-adjacent hyperplastic epidermis (a serial section of Fig. 3f). d-I Ki67-expressing BMDCs (arrowheads) are located in HF and d-II strong membrane beta-catenin expression is detected overall in the epidermis, but low levels of cytoplasmic beta-catenin expression are detected in BMDCs in the HF (GFP-channel in white box). e–i BMDECs in dysplastic skin. e BrdU-positive BMDCs (arrowheads) are identified in the basal region of dysplastic epithelium. White box area is magnified. f f-I Ki67-expressing BMDCs (arrowheads) are identified in the basal region of dysplastic epithelium. f-II Higher magnification micrograph from different dysplastic skin sample. g Small Ki67-expressing BMDCs (white box area is magnified) are detected in the neck region (white line, located at junction between down-growing deregulated HF and interfollicular epidermis). h KCs and BMDCs in dysplastic epidermis (white boxes, magnified in h-I and h-III) show cytoplasmic beta-catenin expression, but cells in normal epidermis (yellow boxes, magnified in h-II and h-IV, orange broken line indicates basal layer of epidermis) show membrane expression. i BMDCs in dysplasia (white box area and area of white broken line in i-I) show cytoplasmic beta-catenin expression. *White scale bar, 50 µm

Fig. 6

K15-positive bulge stem cells contribute to development of ulcer-associated dysplasia. a Contribution of K15-positive bulge stem cells (Beta-gal-positive) in hyperplastic epidermis. a-I Beta-gal-positive (white arrows and white E) and negative (yellow arrow and yellow E) HF are identified. a-II Beta-gal-positive cells are identified in HFs (red and white arrows) and interfollicular epidermis in hyperplastic epidermis. a-III Beta-gal-positive cells (red line) are identified in the bulge and lower HF in anagen. a-IV Beta-gal-positive cells (red IHC staining and black arrows) in HF bulge area and lower HF. b Contribution of Beta-gal-positive cells in papilloma epithelium. The junction of Beta-gal-positive and negative area (white box) is magnified. c Contribution of K15 bulge stem cells in dysplasia. c-II-IV Serial sections of red box area in c-I. c-I and c-II Beta-gal-positive cells are mainly located in the HF structure in dysplastic epidermis. c-III Serial sections show beta-gal-positive area is also K14-positive, and (c-IV) cytoplasmic beta-catenin expression with Ki67-positive cells located in basal region. d Beta-gal-positive cells located in both HF (white arrow) and interfollicular epidermis (white E) in dysplastic epidermis. e Ki67- and beta-catenin-expressing cells located in the neck region of dysplastic HF. f Cellular sources from six different dysplastic HFs (three beta-gal-positive, white E’s, and three beta-gal-negative, yellow E’s) participate in developing dysplastic interfollicular epidermis. g Beta-gal-positive cells (white E) and KC-shaped BMDCs (arrowheads) were observed together in hyperplastic interfollicular epidermis. White box area is magnified. g-I KC-shape BMDCs in white box-I and white box-II are magnified. h, i IHC of serial sections identified beta-gal-positive cells and BMDCs in close area of dysplastic epidermis. h Beta-gal-positive cells are in dysplastic HFs. Red box area is magnified. i A group of BMDCs were identified in a HF adjacent to dysplasia. Red box area is magnified. i-I Area of BMDCs is magnified further in black box. *Black and white scale bar, 50 µm; Red scale bar, 200 µm

Fig. 7

BMDCs initiate squamous lesions in the skin. a Overview of DMBA-exposed BMT to naive recipients: a-I Carcinogen-exposed donor BMC preparation: FVB/EGFP male donor mice were given one treatment of DMBA (1 mg in 0.5 ml corn oil) via oral gavage and DMBA-exposed whole toxic BMCs were isolated for BMT. a-II Naive female FVB/N mice were prepared as BMT recipients, and twice-weekly topical TPA (5 nmol for 3 weeks followed by 17 nmol in 200 µl of acetone until 18 weeks) promotion followed 5 weeks after BMT for 18 weeks (1. DMBA (BMT donor) ► 2. BMT ► TPA (BMT recipient)). Controls included corn oil/acetone, DMBA/acetone, and corn oil/TPA. None of the controls developed tumors. b K14-positive BMDCs identified from the squamous cell carcinoma sample from DMBA-exposed BMT recipient. All five of the tumors demonstrated co-expression of GFP and K14. Some heterogeneity was observed; however, most of the tumor epithelium was composed of cells of donor origin. Black scale bar, 50 µm. c Flow cytometric analysis: blood, bone marrow, and two disaggregated squamous tumor samples from DMBA-exposed BMT recipients. Native GFP fluorescence is shown in the Y-axis, and EpCAM, an epithelial cell marker, is shown along the X-axis. Note the GFP⁺/EpCAM⁺ double positive cells in quadrant III, and that there are more double positive cells in the blood than in the bone marrow. Note further, GFP^high/EpCAM^high double positive cells are increased in both tumors (red boxes). The single EpCAM⁺ cells in quadrant IV suggest heterogeneity of the tumor epithelium. *Some images in Fig. 7a were generated with AutoDraw