Intrinsic anti-tumorigenic properties of the skin epithelium promote cancer resistance in naked mole-rats

Summary

Naked mole-rats (NMRs) are unique long-lived mammals that possess marked resistance to cancer including lack of any reported spontaneous skin cancer incidences. Here, we show that in comparison with mice, homeostatic epidermal keratinocytes in NMRs exhibit elevated expression of 324 tumor suppressor genes of different classes including 55 genes regulating DNA damage/repair. Furthermore, NMR skin grafted onto nude mice exhibits complete resistance to 7,12-dimethylbenz[a]-anthracene/12-O-tetradecanoylphorbol-13-acetate (DMBA/TPA)-induced skin carcinogenesis in tissue-autonomous manner. In contrast to mice, DNA-damaged cells are effectively eliminated from DMBA/TPA-treated NMR epidermis, which exhibits a unique transcription response to DMBA/TPA characterized by high expression of anti-cancer genes, activation of ferroptosis, as well as by downregulation of 80 oncogenes including components of NF-kB, Wnt, and tyrosine kinase receptor signaling pathways. Thus, intrinsic multi-level anti-cancer protective program in NMR epithelial cells serves as unique model for further analyses of natural anti-cancer defense mechanisms and their translation to humans.

Graphical abstract

Highlights

• •Skin epithelium in naked mole-rats contains highly expressed tumor suppressor genes
• •Naked mole-rat skin grafted onto nude mice shows tissue-autonomous cancer resistance
• •Mutated cells effectively eliminated from carcinogen-treated naked mole-rat epidermis
• •Carcinogen-treated naked mole-rat epidermis show downregulation of numerous oncogenes

Biological sciences; Genetics; Cell biology

Introduction

An increase in the proportion of the elderly population in modern society results in raising prevalence of age-related pathologies, including cancer, therefore, creating a significant clinical and socio-economic burden.¹ Epithelial skin cancer (i.e., basal cell carcinoma and squamous cell carcinoma [SCC]) is the most common cancer in non-African populations, and its incidence is increasing worldwide.^2,3,4,5 Over the last two decades, significant progress in delineating mechanisms underlying epithelial skin cancer development has been achieved largely through studies in genetically modified mice and experimentally induced models of skin carcinogenesis.^6,7 However, many aspects of human skin cancer pathogenesis and age-associated susceptibility remain to be defined.

SCC is one of the most common skin cancers, affecting over 1 million new patients in the United States annually.⁸ SCC development is a multi-step process initiated by carcinogen-driven activating mutations of several oncogenes, such as HRAS, KRAS, TERT, and PI3KCA, and loss-of-function mutations of tumor suppressor genes (TSGs) (TP53, CDKN2A, NOTCH1, and SRCASM), resulting in alterations of the balance between keratinocyte (KC) proliferation, differentiation, and apoptosis in the epidermis.^3,4,9 Uncontrolled proliferation of mutated cells in the epidermis leading to SCC development is promoted via deregulated activities of several pro- and anti-oncogenic signaling pathways (EGFR, FYN, NOTCH, PIK3CA, SRCASM, and WNT), transcription factors (ETS2, TP53, and TP63), and epigenetic regulators (H3K9 methyltransferase G9a, H3K27 methyltransferase EZH2, and H3K4 methyltransferase MLL4).^4,5,10

Many mechanisms involved in SCC development were discovered by using chemical skin carcinogenesis model, in which single topical application of carcinogen 7,12-dimethylbenz[a]-anthracene (DMBA) onto mouse skin, followed by repeated treatment with tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA) result in the formation of papillomas followed by their progression to SCCs.^10,11,12 In this model, DMBA induces A > T and G > T transversions and activating point mutations in Hras (Hras^Q61L), as well as in Kras and Rras2 genes, driving SCC development.^13,14

Increasing evidence of data suggests that studying animal species with natural resistance to distinct pathological conditions brings novel important information into understanding mechanisms of human diseases and is ultimately beneficial for human health.^15,16 Naked mole-rats (NMRs, Heterocephalus glaber) are long-lived mammals that possess marked resistance to cancer and other age-related pathologies (i.e., cardiovascular disease, neurodegeneration) and maintain a sustained healthy life span for over 30 years.^15,17,18 Exceptional resistance of NMRs to aging-associated pathologies is mediated by several mechanisms, including more robust DNA repair and genome stability compared with mice, a unique organization of the tumor suppressor Ink4a/b locus, production of large amounts of higher molecular weight hyaluronic acid with unusual properties, altered IGF receptor signaling, increased proteasome activity, and p53 protein stability.^{15,16,19,20,21,22,23,24,25,26,27,28,29}

Comparative genome analyses revealed that the NMR genome shows higher similarity (93% synteny) to the human genome compared with that of mice (83%) or rats (80%).³⁰ Similarly to human skin, NMR epidermis is thick and multi-layered,³¹ and skin aging in the NMR resembles many features of human skin aging, including histological and biochemical changes in both the epidermis and dermis.³² Despite their remarkable longevity, NMRs also exhibit a potent resistance to spontaneously developing cancers,¹⁶ with a lack of any reported skin cancer incidence, including basal cell carcinoma/SCC and malignant melanomas.³³ Recent data demonstrate that NMRs exhibit remarkable resistance to chemical skin carcinogenesis associated with dampened cutaneous inflammatory response.³⁴ However, it is still unclear how the NMR epidermal KCs serving as primary targets for carcinogen detect, repair, or eliminate DNA mutations and resist neoplastic transformation.

In this manuscript, we show that homeostatic epidermal KCs in NMRs exhibit highly elevated expression of anti-cancer genes compared with mouse KCs, while NMR skin shows complete resistance to DMBA/TPA-induced carcinogenesis in a tissue-autonomous manner. We also demonstrate that in contrast to mice, mutated cells are effectively eliminated from the NMR epidermis during time-course of DMBA/TPA treatment. Furthermore, we show that NMR epidermis exhibits a unique transcription response to DMBA/TPA treatment, including maintenance of high expression levels of anti-cancer genes, as well as downregulation of numerous oncogenes. These data provide a foundation for further analyses of the mechanisms mediating natural anti-cancer protection in NMRs and their translation to improve cancer resistance in humans.

Results

Skin epithelial cells show elevated expressions of tumor suppressor and DNA repair genes in NMRs compared with mice

To understand mechanisms mediating skin cancer resistance in NMRs, we first compared expressions of cancer-related genes (TSGs and oncogenes) between homeostatic epidermal KCs from NMRs and FVB (Friend leukemia virus B) mice. Dorsal skin was collected from normal homeostatic 6–7 month-old NMRs and 7–8-week-old FVB mice. Skin epidermis was separated from the dermis, and epidermal KCs were FACS-sorted using anti-Cd49f (integrin α6) antibody³⁵ (Figures S1A and S1B). Bulk RNA-seq analyses³⁶ revealed >2-fold differences in expression of 5,688 genes (differentially expressed genes [DEGs]) between the two species, with 4575 upregulated and 1,113 downregulated in NMR KCs compared with mouse KCs (Figure 1A; Tables S1 and S2).

Figure 1.

Homeostatic skin epithelial cells in NMRs show marked increase in the expressions of tumor suppressor and DNA repair genes compared with mice

(A) Bulk RNA-seq analysis revealed 324 tumor suppressor genes (TSGs) upregulated in FACS-sorted NMR epidermal keratinocytes (KCs) compared with mouse KCs, whereas only 50 TSGs were upregulated in mouse KCs.

(B) GO Annotation analysis shows enriched pathways among TSGs upregulated and downregulated in NMRs epidermal KCs versus mouse KCs.

(C) RNA-seq data demonstrating increased expression of selected genes encoding stimulators of apoptosis, cell cycle, and oncogenic pathway inhibitors in NMR epidermal KCs compared with mouse KCs.

(D) Number of genes regulating distinct DNA damage/repair pathways upregulated in homeostatic NMR epidermal KCs and mouse epidermal KCs.

Epidermal KCs were FACS-sorted from dorsal skin of NMRs and FVB mice against Cd49f antibody and used for bulk RNA-seq analysis. DEGs were merged with TSGs database (TSG 2.0, University of Texas, Houston, TX).

To compare a TSG profile of epidermal KCs, we merged the NMR and mouse KC transcriptomes with the TSGs database (TSG 2.0, University of Texas, Houston, TX).³⁷ Of the 1,218 genes present in the TSG database, 324 TSGs showed 2-fold and higher expression levels in NMR KCs, while only 50 genes were upregulated in mouse KCs versus NMR KCs (Figure 1A; Table S3). Gene Ontology analyses revealed enrichment of the TSGs stimulating apoptosis, inhibiting cell proliferation, and regulating DNA repair in the NMR KCs versus mouse KCs (Figure 1B; Table S4).

Among TSGs upregulated in the NMR KCs were genes encoding stimulators of apoptosis (Apaf1, Bcl2l11, Bclaf1, Bmf, and Dapk2), inhibitors of cell proliferation (Btg2, Cdkn1c, Dlec1, and Gas1), inhibitors of the Wnt (Dkk3, Sfrp2, and Sfrp5), IGF (Igfbp2), and Hedgehog (Ptch2) signaling pathways, regulators of the DNA methylation/demethylation (Dnmt3a, Dnmt3b, and Tet1), and Polycomb-mediated gene silencing (Pcgf2) (Figure 1C; Table S3). Conversely, mouse KCs preferentially expressed TSGs linked to cell adhesion (Cdh1, Cdh4, Cldn23, and Gjb2), growth factors and cytokine signaling (Bmp4, Ngfr, Tnfrs18, and Socs3), nuclear receptors (Pprg and Vdr), and transcription regulators (Foxc1, Hic1, and Per2) (Table S3).

The largest functional group of TSGs upregulated in NMR KCs (55 genes) was comprised of several classes of DNA repair machinery genes: encompassing interstrand cross-link DNA repair/Fanconi anemia pathway, DNA double-strand break repair by homologous recombination, DNA base excision repair, mismatch repair, and non-homologous end joining pathways (Figure 1D; Table S5). Notably, key genes involved in the control of interstrand cross-link repair, homologous recombination and base excision repair (Fanca, Fancd2, Fanci, and Neil2) showed >10-fold higher expression levels in NMR KCs compared with mouse KCs (Table S5). By contrast, mouse KCs upregulated only 6 DNA damage/repair genes controlling interstrand cross-link repair (Exo5 and Fancb), base excision repair (Rpa3), and double-strand break (Rnf168) pathways or involved in DNA damage response (Parp2 and Rhno1) (Table S5).

Next, we intersected the NMR and mouse KC transcriptome with the human Cancer Gene Census (CGC) database (Welcome Sanger/EBI, Hinxton, UK).³⁸ Detailed analyses of 58 cancer-related genes upregulated in NMR epidermis revealed oncogenes promoting carcinogenesis in the organs outside of the skin (Acsl3, Acvr1, Akt2/3, Bcl3, Ccr7, Cxcr4, Flt4, Hey1, Mn1, Mpl, Mycn, Src, and Zeb1), as well as a number of genes involved in the control of epidermal proliferation (Ccnd3 and Cdk6) and KC differentiation in normal homeostatic epidermis (Egfr and Jun), while stimulating SCC development under experimental and pathological conditions⁵ (Figure 1A; Table S6).

However, in contrast to mice, NMR KCs showed downregulation of 19 cancer-related genes including key components of the Hedgehog pathway Gli1 and Smo promoting development of basal cell carcinoma in the skin,⁷ Wnt pathway component Ctnnb1 involved in neoplastic transformation of epithelial stem cells during SCC formation,³⁹ as well as of a number of oncogenes stimulating cancer development in the skin and other organs (Cbl, Ccr4, Ddr2, Etv5, Fgfr1, Irf4, Jak3, Lmo1, Mafb, Ptk6, and Tbx3) (Figure 1A; Table S6).

Thus, homeostatic epidermal KCs in NMRs maintain high expression of TSGs that inhibit cell proliferation and activity of pro-oncogenic Wnt, IGF, and Hedgehog signaling pathways, stimulate apoptosis and DNA repair, which is associated with downregulation of several key oncogenes stimulating skin carcinogenesis. In contrast to NMRs, mouse KCs predominantly express TSGs linked to cell adhesion, growth factor signaling, nuclear receptors, and transcriptional regulators, as well as show decreased expression of oncogenes driving carcinogenesis outside of the skin.

NMR skin exhibits resistance to chemically induced carcinogenesis in a tissue-autonomous manner

NMRs possess remarkable tumor resistance in two-stage chemically induced skin carcinogenesis model and remain tumor-free for up to 55 weeks.³⁴ To understand whether this resistance is intrinsic to the NMR skin rather than linked to their unique endocrine or metabolic features,^16,18 we grafted dorsal NMR skin onto immunodeficient nude mice and topically applied a single dose of carcinogen DMBA followed by tumor promoter TPA twice weekly (Figures 2, S2, and S3).^6,40 Earlier research demonstrated that despite changes in tissue microenvironment, mouse skin retains high sensitivity to DMBA/TPA after transplantation onto nude mice.⁴¹ By week 10–12, host mice developed numerous papillomas, while the NMR skin grafts remained tumor-free for the duration of 20 weeks of the DMBA/TPA treatment (Figures 2A, 2B, and 2E). Correspondingly, the skin of nude mice developed marked epidermal thickening and hyper-proliferation, whereas NMR grafts showed no such changes (Figures 2F, 2G, and S2A). Also, the number of Cd11b+ macrophages increased in the skin of DMBA/TPA-treated nude mice, while lack of differences in their number was seen in NMR skin grafts compared with controls (Figures 2H and S2B).

Figure 2.

NMR skin transplanted onto nude mice shows marked resistance to chemically induced carcinogenesis in a tissue-autonomous manner

(A) Acetone-treated nude mice (control group) bearing NMR skin grafts. NMR grafts are characterized by increased epidermal thickness, lack of hair follicles and dermal pigmentation (inset, H&E staining, large arrows). Neighboring mouse skin shows thinner epidermis and hair follicle in the dermis (inset, H&E staining, small arrows, graft area is demarcated by dotted line).

(B) DMBA/TPA-treated nude mice (upper left image) show numerous papillomas, some located adjacent to the transplanted NMR skin (inset, H&E staining, small arrows). NMR skin graft (inset, H&E staining, demarcated by dotted line) show lack of tumor development (large arrows).

(C) Control (left image) and DMBA/TPA-treated (right image) FVB mice. Numerous papillomas are present on the back skin after 12 weeks of DMBA/TPA treatment (top) and H&E (bottom).

(D) Control (left image) and DMBA/TPA-treated (right image) NMRs. Lack of papillomas and visible effects of DMBA/TPA on the back skin after 25 weeks of DMBA/TPA treatment (top) or H&E (bottom).

(E) Number of epidermal tumors in nude mice (per mouse) 13–20 weeks after DMBA/TPA treatment. NMR grafts show lack of tumor development.

(F) Significant increase of the epidermal thickness in DMBA/TPA-treated nude mice and lack of differences in NMR skin grafts compared with controls.

(G) Significant increase of epidermal Ki-67+ cells in DMBA/TPA-treated nude mice and lack of differences in NMR skin grafts versus the controls.

(H) Significant increase of Cd11b+ macrophages in DMBA/TPA-treated nude mouse skin and lack of differences in the NMR skin grafts compared with controls.

(I) Number of epidermal tumors in FVB mice (per mouse) 12 weeks after DMBA/TPA treatment. Lack of tumors in NMR skin for 25 weeks after DMBA/TPA treatment.

(J) Increase of Ki-67+ cells in both NMR and mouse epidermis after DMBA treatment alone and after DMBA/TPA treatment.

(K) Moderate increase of Caspase-3+ cells in the NMR epidermis after DMBA treatment (p < 0.05) and marked increase in mouse epidermis after DMBA/TPA treatment.

(L) Decrease of Cd3+ T-cells in the NMR skin after DMBA treatment alone (p < 0.01) and marked increase in DMBA/TPA-treated mouse skin.

(M) Increase of Cd11b+ macrophages in both NMR and mouse skin after DMBA treatment alone or after DMBA/TPA treatment.

Dorsal skin of 6–7-month-old NMRs was dissected and grafted onto 7–8-week-old nude mice. DMBA/TPA treatment was applied to entire dorsal skin of nude mice 8 weeks after grafting according to established protocol for 13–20 weeks.⁶ 7–8-week-old FVB mice and 6–7-month-old NMRs were treated by DMBA/TPA or acetone (control) for 12–25 weeks. Skin was harvested after DMBA treatment alone (10 days after single application) or after DMBA/TPA treatment and processed for morphological and immunohistochemical analyses.

Mean ± SEM; N = 3, Student’s t test: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001.

Scale bars, images of animals – 10 mm (A–D) and microphotographs – 100 μm.

We performed additional DMBA/TPA experiments on FVB mice and NMRs that confirmed the results obtained on nude mice with NMR skin grafts and previously published observations³⁴: FVB mice developed numerous papillomas by 12 weeks of DMBA/TPA treatment, while none of the NMRs developed papillomas for at least 25 weeks (Figures 2C, 2D, and 2I). Histologically, NMR skin remained largely unaffected, lacking neoplastic changes after DMBA/TPA treatment, in sharp contrast to mouse skin (Figures 2C, 2D, and 2I).

Both species showed an increase in epidermal proliferation already 10 days after single DMBA application (Figures 2J and S2D). By week 7 of DMBA/TPA treatment, just before papilloma emergence, epidermal proliferation, and thickness were dramatically increased in FVB mice, while NMR epidermis displayed only a modest increase in the number of Ki-67+ cells and thickness (Figures 2J and S3A). Compared with controls, the number of Caspase-3+ cells increased in DMBA-treated NMR epidermis on day 10, which returned to baseline by 7 weeks of DMBA/TPA treatment (Figures 2K and S2E). In contrast, no such response was observed in FVB mice treated with DMBA alone; whereas marked increase of Caspase-3+ cells was seen after 7 weeks of DMBA/TPA-treatment in mouse epidermis (Figures 2K and S2E).

DMBA/TPA treatment of FVB mice resulted in a marked expansion of Krt14-expressing cells to suprabasal epidermal layers, while Krt14 expression in the basal layer was decreased compared with controls (Figure S3B). By contrast, Krt14+ expressing cells were restricted to basal epidermal layer, without apparent expansion to suprabasal layers in DMBA/TPA-treated NMRs (Figure S3B). Also, DMBA/TPA treatment of NMRs resulted in a decrease of epidermal Krt10 expression, while a marked expansion of the Krt10+ epidermal layers was seen in DMBA/TPA-treated FVB mice (Figure S3C). In addition, hyaluronic acid receptor CD44 and hyaluronic acid binding protein (HABP) increased in DMBA-treated NMR epidermis (Figures S3D and S3E).

Analysis of immune cells revealed opposite dynamics in the number of skin Cd3+ T-cells between NMRs and mice. Less numerous Cd3+ cells were detected in NMR skin after DMBA treatment compared with controls, while in mouse skin, a significant increase in Cd3+ cell number was seen after DMBA alone and DMBA/TPA treatment (Figures 2L and S2G). However, a quite similar trend in the number of Cd11b+ macrophages was observed between NMRs and FVB mice after DMBA and DMBA/TPA treatments (Figures 2M and S2F).

DMBA-induced somatic mutations and DNA-damaged cells are eliminated faster in the NMR epidermis compared with mouse epidermis

To compare the dynamics of DMBA-induced somatic mutations between mouse and NMRs epidermal KCs, we performed whole-genome sequencing (WGS) of isolated epidermis from FVB mice and NMRs collected 10 days after a single DMBA exposure and 46 days (6.5 weeks) after DMBA/TPA treatment together with their corresponding controls. Consistent with previous data,^13,42 DMBA induced A > T and T > A transversions in both mouse and NMR epidermis (Figure 3A). However, the frequency of these transversions increased further in mouse epidermis by day 46 after DMBA/TPA treatment compared with day 10 after DMBA treatment alone, while no such increase was detected in NMR epidermis (Figure 3A).

Figure 3.

DMBA-induced mutations and DNA-damaged cells are eliminated faster in NMR epidermis compared with mouse epidermis

(A) Whole-genome sequencing analysis revealed an increase in DMBA-specific A>T and T>A transversions in mouse epidermis over the course of DMBA/TPA treatment, while lack of such increase was observed in NMR epidermis.

(B) Somatic mutation profile of the NMR and mouse epidermis at day 46 of DMBA/TPA treatment compared with day10 of DMBA treatment. Marked increase of the total mutation variants, as well as of the distinct mutation categories in mouse epidermis after DMBA/TPA treatment. Decrease of total insertions/deletions in the NMR epidermis was seen in the DMBA/TPA-treated NMR epidermis compared with epidermis treated by DMBA alone.

(C) Immunofluorescence analysis of KCs with DNA damage using anti-pH2AX antibody: significant increase of pH2AX + cells in DMBA-treated mouse and NMR epidermis of day 10 of experiment and marked decrease of pH2AX + cells in NMR epidermis by day 46 of DMBA/TPA treatment.

Mean ± SEM; N = 3, Student’s t test: ∗p < 0.05, ∗∗∗∗p < 0.0001.

Further analyses of the somatic mutation profile in mouse epidermis revealed a sharp increase (by 117.22%) in total single-nucleotide polymorphism (SNP) by day 46 after DMBA/TPA treatment compared with a single DMBA exposure (Figure 3B; Table S7). This increase included insertion/deletions (by 51.79%), SNP transitions (by 86.23%) and, most strikingly, SNP transversions (by 156.15%) (Figure 3B; Table S7). By contrast, only marginal changes were detected in DMBA/TPA-treated NMR epidermis: total SNPs were increased by 4.09%, SNP transitions by 4.03%, and SNP transversions by 4.22%, while insertions/deletions fell by 9.27% compared with a single DMBA treatment (Figure 3B; Table S8).

These data were mirrored by immunofluorescence analyses of epidermal pH2AX expression. In mice, the percentage of pH2AX -positive cells increased sharply by day 46 DMBA/TPA treatment compared with day 10 after DMBA alone (Figures 3C and S2C). By contrast, the number of pH2AX-positive cells significantly decreased in NMR epidermis at day 46 following DMBA/TPA exposure relative to day 10 after DMBA alone (Figures 3C and S2C). Together, these data indicate that although DMBA induced somatic mutations in both mouse and NMR epidermis, DNA-damaged cells were cleared significantly faster from NMR epidermis than from mouse epidermis.

Carcinogen exposure induces unique anti-cancer transcriptional response and activation of ferroptosis in the NMR epidermis

To dissect transcriptional response to DMBA, we treated 6–7-month-old NMRs and 7–8-week-old FVB mice with a single dose of DMBA, and 10 days later isolated epidermis was processed for RNA-seq analyses. In mouse epidermis, DMBA altered 365 genes by > 2-fold, 152 upregulated and 213 downregulated, while in NMR epidermis, 180 genes were upregulated and 110 downregulated relative to corresponding controls (Figure 4A; Tables S9, S10, S11, and S12). Remarkably, of the 655 differentially-expressed genes (DEGs) across both species, only 11 genes (or 1.6%) overlapped: 6 genes were upregulated in both (Edn2, H2Bc12, Mgmt, Serpine2, Spc25, and Sprr1b), 3 genes were downregulated in both (Krtap1-5, Krtap-19-3, and Paqr7), while 2 genes (Scd and Soat1) were upregulated in NMR and downregulated in mouse epidermis (Tables S9, S10, S11, and S12). These results highlight the highly species-specific transcriptional response induced by DMBA in NMR versus mouse epidermis.

Figure 4.

DMBA treatment induces unique transcriptional response in the NMR epidermis compared with mice

(A) Differentially expressed genes (DEGs; up- and downregulated) in DMBA-treated mouse (7–8-week-old) and NMR (6–7-month-old) epidermis compared with respective controls at day 10 after a single DMBA dose.

(B) Ingenuity Pathway Analysis of up- or downregulated DEGs in mouse and NMR epidermis in response to DMBA treatment versus control.

(C) Selected DEGs in distinct functional categories up- and downregulated in DMBA-treated mouse epidermis compared with controls.

(D) Selected DEGs in distinct functional categories up- and downregulated in DMBA-treated NMR epidermis compared with controls.

Skin epidermis was harvested from 7–8-week-old FVB mice and 6–7-month-old NMRs 10 days after single DMBA or acetone (control) application. RNA was isolated and processed for bulk RNA-seq analyses.

Ingenuity pathway analysis (IPA) revealed that the genes upregulated in DMBA-treated mouse epidermis were strongly enriched for immune response pathways, whereas downregulated genes were dominated by extracellular matrix and wound healing programs (Figure 4B; Tables S13 and S14). Notably, transcripts of several chemokines known to attract monocytes (Ccl2 and Ccl7) and T-cells (Ccl22, Cxcl9, and Cxcl10) were upregulated in DMBA-treated mouse epidermis versus controls (Figure 4C; Table S9). Furthermore, several components of T cell receptor (TCR) complex showed mixed regulation after DMBA treatment in mice, thus suggesting a re-arrangement of the epidermal TCR repertoire accompanying T cell activation (Figure 4C; Tables S9 and S10). These results support the immunofluorescence data showing an increase of Cd3+ cells in mouse skin after DMBA treatment (Figure 2L). In contrast, numerous extracellular matrix-associated genes (Col1a1, Col1a2, Col3a1, Col5a1, Col5a2, and Col6a3) and markers of terminal KC differentiation including the components of cornified cell envelope (Lce1l, Prr9, Tchh, and Tchhl1) were downregulated in mouse epidermis after DMBA exposure (Figure 4C; Table S10).

In contrast to mice, IPA analysis of DMBA-induced transcripts in the NMR epidermis revealed significant enrichment of the cell death pathways including ferroptosis (Ctsb, Sat1, and Steap3) and p53 signaling (Figure 4B; Table S15). Group of genes involved in regulation of apoptosis and upregulated in DMBA-treated NMR epidermis included Aen, Bak1, and Bax, stimulating cell death, as well as an apoptosis inhibitor Birc3 (Figure 4D; Table S11). These data corroborated with immunofluorescence detection of active Caspase-3 showing a significant increase of apoptotic cells in the epidermis of DMBA-treated NMRs versus controls, and lack of such changes in mice (Figures 2K and S2E).

Because ferroptosis-associated genes (Ctsb, Sat1, and Steap3) were upregulated in DMBA-treated NMR epidermis, we performed immunofluorescence analyses of the established ferroptosis marker 4-hydroxynonenal (4-HNE) and compared its expression in NMRs and mice.⁴³ Interestingly, a predominance of 4-HNE+ cells was seen in homeostatic NMR epidermis versus normal mouse epidermis (Figure S4A), while a further increase of 4-HNE+ cells was observed in DMBA/TPA-treated NMR epidermis; in contrast, no changes in their number were seen in mouse epidermis after DMBA/TPA treatment (Figure S4B). Another ferroptosis marker, Cathepsin B, was also increased in pH2AX + KCs in DMBA-treated NMR epidermis (Figure S4C). Furthermore, treatment of mice with Erastin, a ferroptosis stimulator,⁴⁴ significantly increased number of 4-HNE+ cells and Cathepsin B+/pH2AX + cells in mouse epidermis treated with DMBA/TPA (Figures S4B and S4C). Erastin treatment also reduced a number of tumors in DMBA/TPA-treated FVB mice 6 weeks after treatment, thus demonstrating that activation of ferroptosis indeed retards tumor formation in mice (Figure S4D).

In contrast to mouse epidermis, DMBA also increased the expression of several groups of TSGs in NMR epidermis including extracellular matrix regulators (Clu, Tgfbi, and Thbs1), growth factor receptors, ligands, and inhibitors (Bmp2, Hcar2, Igfbp5, Nrg2, and Socs3), and ubiquitin-protein ligase Nedd4l (Figure 4D; Table S11). Several genes regulating cell cycle (Ccnd2, Ccne1, and Ccng1) and DNA damage/repair (Ddb2, H2ax, and Tk1) were also induced in the NMR epidermis relative to controls (Figure 4D; Table S11). Interestingly, increased expression of several cytokine genes (Cxcl1, Cxcl14, Il8, and Il33) was counterbalanced with elevated expression of inhibitors of Jak/Stat and NFkB signaling pathways (Socs3, Nfkbia, and Nfkbiz), hinting a tight control over inflammation in DMBA-treated NMR epidermis (Figure 4D; Table S11).

Conversely, the most downregulated cohort comprised genes controlling antigen presentation (Cd83, Hla-Dpa1, Hla-Dpb1, Hla-Dqa1, Hla-Dqb1, Hla-Dra, and Hla-Drb), macrophage activation and cytokine signaling (Csf1r, Il18bp, Il34, Il36g, and Tnfsf18) (Figure 4D; Tables S12 and S16). Together, these data demonstrate that unlike mouse epidermis, DMBA-exposed NMR epidermis activates robust cell death and tumor suppressor programs, while repressing the immune response genes controlling antigen presentation and macrophage activation.

NMR epidermis resists tumor formation by sustaining high TSG expression while repressing oncogenes

To further explore mechanisms underlying a resistance of the NMR epidermis to chemically induced skin cancer, we repeated the two-stage chemical carcinogenesis protocol⁶ as described earlier. However, epidermis was harvested at 6.5–7.5 weeks after the DMBA/TPA treatment, just prior visible papillomas arise in mice (the pre-tumor stage).⁴²

Bulk RNA-seq analyses of mouse pre-tumor epidermis revealed that DMBA/TPA treatment altered 1,922 genes by > 2-fold, with 793 upregulated and 1,129 downregulated relative to acetone-treated control (Figure 5A; Tables S17 and S18). In NMR epidermis, DMBA/TPA treatment yielded 2063 DEGs (537 upregulated and 1,526 downregulated) compared with controls (Figure 5A; Tables S19 and S20).

Figure 5.

RNA-Seq analysis reveals maintenance of high expression levels of TSGs and downregulation of oncogenes in NMR epidermis after DMBA/TPA treatment

(A) Differentially expressed genes (DEGs), including both upregulated and downregulated genes, in mouse and NMR epidermis after 6.5–7.5 weeks of DMBA/TPA treatment compared with respective untreated controls.

(B) Comparison of the changes in mouse and NMR epidermal transcriptomes between the DMBA treatment alone (day 10) and after DMBA/TPA treatment (days 46–52). Number of genes showing distinct patterns of changes in expression is indicated. Reciprocal differences in the number of genes up- and downregulated in DMBA/TPA-treated epidermis versus DMBA-exposed epidermis between NMRs and mice are highlighted by red.

(C) Comparison of the number of oncogenes up- or downregulated in DMBA/TPA-treated mouse and NMR epidermis demonstrate predominance of genes downregulated in the NMR epidermis over other gene categories.

(D) Selected oncogenes up- or downregulated in mouse and NMR epidermis after DMBA/TPA treatment (expression fold changes versus controls).

(E) Comparison of the number of TSGs up- or downregulated in DMBA/TPA-treated mouse and NMR epidermis.

(F) RNA-seq data showing expression of selected TSGs in homeostatic untreated epidermis, DMBA/TPA-treated and control epidermis of mice and NMRs (Fragments Per Kilobase of transcript per Million reads, FPKM). High expression levels of vast majority of TSGs seen in homeostatic epidermal KCs of NMRs is maintained after DMBA/TPA treatment.

Comparing epidermal transcriptomes after DMBA alone (day 10) with DMBA/TPA treatment (weeks 6.5–7.5) revealed that 129 of the 356 DEGs DMBA-responsive genes in mice (36%) remained dysregulated at the pre-tumor stage, whereas 142 out of 284 such genes in NMRs (50%) still differed from controls (Figure 5B; Tables S21 and S22).

Strikingly, 61 genes that were induced by DMBA in NMRs became repressed following TPA treatment at day 46, while only 5 genes with the same pattern were detected in mice, including regulator of cell proliferation cyclin E1 (Ccne1) and several oncogenes (Ets1, Jun, and Rab4b) (Figure 5B; Tables S21 and S22). The opposite shift (DMBA-down and DMBA/TPA-up) was rare in NMRs (1 gene), but common in mice (41 genes), including 17 genes encoding non-epidermal keratins and keratin-associated proteins, 7 collagens, and 3 components of the epidermal barrier (Lce1l, Tchh, and Tchhl1) (Figure 5B; Table S21). These data suggest that DMBA/TPA alters epidermal differentiation program and extracellular matrix organization in mouse epidermis at the pre-tumor stage, while NMR epidermis, instead, suppresses several oncogenes without activating KC differentiation and ECM pathways.

Cross-referencing DMBA/TPA-responsive DEGs treatment with the TSG database (TSG 2.0, University of Texas, Houston, TX)³⁷ revealed modulation of a large number of anti-cancer genes (Figure 5E; Tables S23 and S24). Interestingly, DMBA/TPA induced a higher number of TSGs in mouse epidermis compared with NMR epidermis (55 genes versus 23 genes, respectively), whereas more repressed TSGs were detected in NMRs versus mice (157 and 81 genes, respectively) (Figure 5E; Tables S23 and S24).

Because homeostatic epidermal KCs in NMRs already express 324 TSGs at higher levels than normal mouse epidermal KCs (Figure 1A), we compared the DMBA/TPA transcriptional shift with baseline TSG expression in untreated epidermis of both species. We found that most of the TSGs expressed in homeostatic NMR epidermis also maintained their elevated expression after DMBA/TPA treatment (Figure 5F), and only 25 of 157 downregulated TSGs correspond to genes that are normally higher in untreated NMR epidermis versus mouse epidermis (Tables S1 and S20). Strikingly, among DNA damage/repair TSGs, only 2 genes were altered: Ddb2 (upregulated) and Msh6 (downregulated), suggesting that the vast majority of the DNA repair genes elevated in steady-state NMR epidermis are unchanged after carcinogen exposure (Tables S1 and S20).

Cross-referencing the DMBA/TPA DEGs with the CGC 2.0 database³⁸ revealed the divergent oncogenic programs in the two species. In mouse epidermis, 17 cancer-promoting genes were induced including few oncogenes (Ddr2, Ehf, Epas1, Etv4, and Mafb) and 3 components of Wnt signaling pathway (Lef1, Wnt5a, and Wnt11) (Figures 5C and 5D; Tables S23, S25, S26, and S27). By contrast, only 3 oncogenes were upregulated in the NMR epidermis after DMBA/TPA treatment (Lmo2, Plag1, and Tmsb4x), while 80 cancer-promoting genes were downregulated, among them several proto-oncogenes (Cbl, Jun, Mafb, Mn1, Myc, Mycn, Pim1, Rel, Ski, and Skil) (Figures 5C and 5D; Tables S24, S25, S28, and S29).

Collectively, our data demonstrates that resistance of the NMR epidermis to DMBA/TPA-induced tumor formation is associated with high TSG expression and repression of numerous oncogenes, whereas the transcriptional response of murine epidermis to DMBA/TPA at the pre-tumor stage is accompanied by upregulation of both oncogenes and TSGs.

Oncogenes repressed in the NMR epidermis upon carcinogen treatment encode the components of NF-kB, Wnt, and tyrosine kinase receptor signaling pathways

To uncover functional relationships within cancer-promoting genes downregulated in the DMBA/TPA-treated NMR epidermis, we performed STRING protein-protein interaction network analysis, as described previously.³² In total, the clustering analysis revealed 246 predicted physical protein-protein interactions (Figures 6A and 6B; Tables S30 and S31). The network enriched for regulators of NF-kB pathway (Nfkb2, Rel, Birc3, and Bcl3), Wnt signaling (Bcl9, Bcl9l, and Tcf7l2), and tyrosine kinase receptor signaling (Erbb2, Fgfr3, Map3k4, Map3k8, and Pdgfb) (Figures 6A and 6B; Tables S30 and S31). A second cluster comprised cell cycle drivers Ccnd1 and Ccne1 with their upstream regulators, such as proto-oncogenes Myc, Mycn, and p53 inhibitor Mdm2 (Figure 6A; Table S30).

Figure 6.

STRING protein-protein interaction network analysis of oncogenes downregulated in the NMR epidermis after DMBA/TPA treatment

(A) Functionally related gene clusters significantly enriched among the oncogenes.

(B) STRING protein-protein interaction network showing physical associations between proteins among oncogenes downregulated in DMBA/TPA-treated NMR epidermis. Color nodes represent the proteins associated with corresponding GO term or KEGG pathways, line thickness indicate the strength of data support, active interaction sources – experimental data and datasets, disconnected nodes in the network are hidden.

STRING protein-protein interaction network analysis of oncogenes downregulated in the NMR epidermis after DMBA/TPA treatment.

Interestingly, a large cluster representing oncogenes downregulated in the DMBA/TPA-treated NMR epidermis contained epigenetic transcriptional activators, such as H3K4 histone methyltransferases (Kdm6a, Kmt2c, and Kmt2d), histone acetyltransferases (Kat6a, Kat6b, and Crebbp), regulators of enhancer-promoter interactions (Brd4), and RNA polymerase II activity (Taf15 and Mllt1) (Figure 6A; Table S30). Notably, the Polycomb H3K27 histone methyltransferase Ezh2, which silences many anti-cancer genes,⁴⁵ was also downregulated in the NMR epidermis after DMBA/TPA treatment (Figure 6A; Tables S30 and S31).

Similar analysis of cancer-promoting genes altered in mouse epidermis after DMBA/TPA treatment (19 downregulated and 17 upregulated, Figures 5C and 5D) showed no coherent clustering. Together, these data suggest that NMR epidermis employs unique species-specific anti-tumor program that coordinately suppresses functionally cooperating oncogenes and the epigenetic machinery regulators promoting active gene transcription or repressing anti-cancer genes.

Discussion

Data obtained during last decade reveal that remarkable resistance of NMRs to spontaneous and experimentally induced carcinogenesis is mediated by multiple mechanisms including unique genome organization of the tumor suppressor Ink4 locus, efficient DNA repair system, markedly increased stability of p53 protein, production of large amounts of higher molecular weight hyaluronic acid with unusual properties, decreased activity of the Ras effector ERK and AKT signaling pathways, and dampened inflammatory response.^{21,25,27,29,34,46,47,48} Recent data demonstrate that although NMR cells are capable of neoplastic transformation in vitro, experimental induction of epithelial lung cancer in vivo in NMRs requires simultaneous introduction of oncogenic fusion gene and inactivation of p53 and Rb1 TSGs.^49,50

In this manuscript, we demonstrate for the first time that cancer resistance of the NMR skin is not associated with unique endocrine or metabolic features of these animals¹⁸ and that NMR skin possesses intrinsic anti-tumorigenic properties in a tissue-autonomous manner. We show that (1) homeostatic epidermal KCs in NMRs exhibit highly elevated levels of numerous anti-cancer genes, compared with mouse KCs; (2) NMR skin grafts transplanted onto nude mice do not develop papillomas after DMBA/TPA treatment; (3) in contrast to mice, DNA mutations in the NMR epidermis are not increased and mutated cells are effectively eliminated during time-course of DMBA/TPA treatment; and (4) elevated expression of TSGs in the NMR epidermis is maintained during chemical carcinogenesis protocol, while expression of numerous oncogenes is decreased.

TSGs of different classes protect cells from neoplastic transformation and are frequently mutated in cancers.^3,51,52 SCC development is accompanied by inactivating mutations of several TSGs, including p53, Cdkn2a, Cdkn2b, Rb, Notch1, and Pten.^13,53 In NMR skin, selected TSGs are upregulated during aging, possibly contributing to the lack of spontaneous aging-associated skin tumorigenesis.³² Our data demonstrate that in contrast to mice, homeostatic epidermal KCs of NMRs show higher expression of 324 TSGs encoding extracellular or cell membrane-associated inhibitors of pro-oncogenic Hedgehog, IGF, and Wnt signaling pathways, intracellular stimulators of apoptosis, inhibitors of cell proliferation, transcription factors and epigenetic regulators (Figure 1). Importantly, most of these TSGs also maintained their elevated expression after DMBA/TPA treatment (Figure 5F). These data suggest existence of multi-level anti-cancer protection program in homeostatic epidermal KCs contributing to resistance of the NMR epidermis to DMBA/TPA treatment.

Among TSGs upregulated in homeostatic epidermal KCs of NMRs, DNA repair genes represent the largest functional group that encode the components of all major DNA damage/repair pathways (interstrand cross-link DNA repair/Fanconi anemia pathway, DNA double-strand break repair by homologous recombination, DNA base excision repair, mismatch repair, and non-homologous end joining) (Figure 1). Recent data demonstrate that more efficient DNA repair in NMR cells is mediated, at least in part, by the prolonged binding to chromatin of the DNA sensor sGAS having unique molecular structure and facilitating recruitment of two important components of the homologous recombination pathway FANCI and RAD50 to the DNA damage sites.²⁸ Interestingly, Fanci shows over 20-fold higher expression levels in homeostatic NMR KCs versus mouse KCs, while its expression in the NMR epidermis further increases after DMBA treatment (Tables S1 and S11).

Strikingly, among other DNA damage/repair genes, only 2 genes were altered after DMBA/TPA treatment, indicating that the vast majority of the DNA repair genes elevated in steady-state NMR epidermis are unchanged after carcinogen exposure (Tables S1 and S20). This suggests that additional efforts are required to define the set of transcription factors and/or epigenetic regulators controlling high expression levels of Fanci and other DNA repair genes in homeostatic and DMBA-treated NMR KCs, as well as to assess the potential contribution of the DMBA-induced mutations in gene regulatory regions to the gene expression differences observed between mice and NMRs.

Interestingly, some DNA repair genes (Mbd4, Msh3, Neil2, and Tdg) increased in homeostatic NMR KCs are also upregulated in the NMR liver in comparison to mice,⁵⁴ while others (Ddb2) are also upregulated in the NMR fibroblasts during the UV-induced DNA damage.²¹ Because NMR fibroblasts show more efficient excision repair after UV irradiation compared with mouse fibroblasts,²¹ we speculate that more robust DNA repair machinery in NMR KCs contributes to lack of overall increase of the DNA mutations and to decline of the DNA insertions/deletions in the NMR epidermis after DMBA/TPA treatment (Figure 3).

Our data showing reciprocal dynamics in the number of pH2AX + cells in the epidermis between NMRs and mice 6.5–7.5 weeks after DMBA exposure suggest the effective elimination of DNA-damaged cells in NMRs, possibly via multiple cell death-associated mechanisms including apoptosis and ferroptosis. Recent observations show that in response to DNA-damaging agents, cell necrosis predominates over apoptosis in NMR dermal fibroblasts.⁵⁵ We demonstrate that in homeostatic epidermal KCs of NMRs, several genes encoding stimulators of apoptosis (Apaf1, Bcl2l11, Bclaf1, Bmf, and Dapk2) are upregulated compared with mouse KCs (Figure 1), while DMBA treatment results in increase of apoptosis-promoting genes (Aen, Bak1, and Bax) accompanied by increase of Caspase-3+ cells in the NMR epidermis (Figures 4D and S2E). In addition to apoptosis-related genes, increased expression of the ferroptosis marker 4-HNE and three ferroptosis-associated genes (Steap3, Sat1, and Ctsb) were seen in DMBA-treated NMR epidermis (Figures S4A–S4C; Table S15). Consistently with this data, stimulation of ferroptosis in DMBA/TPA-treated mice results in an increase of 4-HNE+ cells and retardation of tumor development (Figure S4D), demonstrating that ferroptosis indeed contributes to skin cancer resistance in NMRs. However, additional efforts are required to fully uncover mechanisms of cell death responsible for eliminating DNA-damaged KCs and, possibly, tumor-initiating epidermal cells,⁵⁶ which, similarly to other epithelial stem cells, could be eliminated via p53-dependent apoptosis.⁵⁷

Most likely, resistance to chemical carcinogenesis in NMR skin is mediated by unique complex microenvironment involving interactions between different cell lineages including epithelial and mesenchymal cells. For instance, epidermal cell death and skin inflammation are tightly linked: increased apoptosis and necroptosis triggers inflammatory cascades and, conversely, inflammation causes excessive KC death.⁵⁸ NMRs and mice also differ in their immune system including the lack of natural killer cells and the presence of circulating cytotoxic γδ T-cells in NMRs.^59,60 These differences, together with distinctions in skin morphology including the markedly reduced number of hair follicles in NMRs and presence of higher molecular weight hyaluronic acid in extracellular matrix of NMRs,^27,31,32 might also contribute to the resistance of NMRs to skin carcinogenesis. Recent data demonstrate that resistance of NMRs to 3-methylcholantrene-induced fibrosarcoma formation is associated with a dampened cutaneous inflammatory response due to loss-of-function mutations in Ripk3 and Mlkl genes that function as mediators of necroptosis and inflammation.³⁴ Our data showing a markedly decreased number of Cd3+ cells in NMR skin compared with mouse skin after DMBA/TPA treatment are concordant with findings of Oka et al. Most likely, multiple mechanisms contribute to the inhibition of inflammation in the NMR skin including elevated expressions of inhibitors of Jak/Stat and NFkB signaling pathways (Socs3, Nfkbia, and Nfkbiz) in DMBA/TPA-treated NMR epidermis (Figure 5).

Furthermore, downregulation of genes involved in the control of macrophage activation, antigen presentation, and cytokine signaling (Figure 4) is also likely to play a role in attenuating inflammation in NMR skin after DMBA/TPA treatment. Cd11b-positive macrophages are present in control NMR skin grafts, as well as in the dermis of control NMRs and NMRs treated by DMBA or DMBA/TPA (Figures S2B and S2F), while our previous data also demonstrates their presence in the dermis of homeostatic young and old NMRs.³² In the context of recent data demonstrating that Cd11b agonists are capable of reprograming tumor-associated macrophages from an immune-suppressive toward an anti-tumor phenotype,⁶¹ additional analyses including single-cell RNA sequencing are needed to define potential contribution of the local microenvironment to differences in the phenotype of dermal Cd11b-positive cells between NMRs and mice.

We also demonstrate here that one of the unique features mediating resistance of the NMR epidermis to chemically induced carcinogenesis is the downregulation of 80 genes encoding proteins from different functional groups that stimulate the neoplastic process including apoptosis/cell cycle regulators, growth factors/receptors, and their downstream signaling components, proto-oncogenes, and epigenetic regulators of active transcription (Figures 5 and 6). Some of these proteins, such as apoptosis inhibitor Birc3, distinct components of the EGF, FGF, and PDGF signaling pathways, cyclins D1 and E1, and the proto-oncogenes Cbl, Jun, Myc, Rel, and Ski play important roles in SCC development and progression in many epithelial cells including epidermal KCs.^2,3,4,5 However, mechanisms regulating the expression of these genes in DMBA/TPA-treated NMR epidermis, as well as the markedly elevated expression of numerous TSGs in homeostatic epidermis of NMRs remain to be defined.

In summary, these data provide evidence that despite the relatively high proliferation rate in homeostatic conditions,³² NMR epidermis harbors a potent tissue-autonomous anti-cancer protective system that include elevated expression of TSGs, effective elimination of DNA-damaged cells after mutagen exposure, inhibition of the inflammatory response, and downregulation of oncogenes, preventing neoplastic transformation. We hope that this data will serve as a platform for further investigations into the mechanisms controlling anti-cancer protection in human epithelial cells and for designing novel approaches to cancer prevention in humans.

Limitations of the study

Although our data provides evidence for existence of intrinsic multi-level anti-cancer protective program in NMR epithelial cells, several limitations should be considered. Additional efforts are required to define a set of transcription factors and/or epigenetic regulators controlling high expression levels of tumor suppressor and DNA repair genes in homeostatic and DMBA-treated NMR KCs. Also, more analyses are required to fully uncover mechanisms of cell death responsible for eliminating DNA-damaged KCs in carcinogen-exposed NMR epidermis. Furthermore, contribution of tumor-initiating epithelial cells to resistance of the NMR skin to chemically applied carcinogens needs to be carefully explored. Finally, additional analyses including single-cell RNA sequencing are needed to define potential contribution of the distinct components of local microenvironment including immune cells to differences in the response to carcinogens between NMRs and mice.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Vladimir A. Botchkarev (vladbotc@bu.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The sequencing data are available from the NCBI Gene Expression Omnibus (accession codes GSE334395, GSE334586, and GSE334587), and from the NCBI Sequence Read Archive (SRA) under accession codes PRJNA1480147 and PRJNA1481185. Other data reported in this paper will be shared by the lead contact upon request.
• •This study does not report original code or analysis tool.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

Authors thank Dr. Igor Malashchuk for generation of preliminary data for this study; Drs. Michael Fessing and Ignaty Leshchiner for data discussion/advise; Drs. Shelley Buffenstein, Vera Gorbunova, Andrey Seluanov, and Ewan St. J. Smith for help with establishing NMR facilities; and Zoe Smith and John Bland for help with a part of the animal experiments. This study was supported in part via grant 4R33AR078093 from the NIAMS (to V.A.B. and A.A.S.). I.F. is a recipient of the Research Grant from the Dermatology Foundation.

Author contributions

I.F. performed the experiments, generated and analyzed data, and wrote manuscript. A.M. designed and performed the experiments and analyzed and discussed the results. E.R. performed the experiments and generated data. N.B. generated, analyzed and discussed the data. A.S. designed and performed the experiments, analyzed the data, wrote the manuscript. V.B. designed the experiments, analyzed the data, and wrote the manuscript.

Declaration of interests

The authors declare no competing interests.

Declaration of generative AI and AI-assisted technologies in the writing process

The authors did not use AI for help in preparation of this paper.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Cytokeratin 14-FITC	Santa Cruz	Cat#SC-53253; RRID:AB_2134820
Cytokeratin 10	Biolegend	Cat#905401; RRID:AB_2565049
Ki67	Abcam	Cat#AB15580; RRID:AB_443209
Caspase-3	Abcam	Cat#AB13847; RRID:AB_443014
pH2AX	Cell Signaling	Cat#9718T; RRID:AB_2118009
CD3	Bio-Rad	Cat#MCA1477; RRID:AB_321245
CD8	Bio-Rad	Cat#MCA752S; RRID:AB_324551
CD8a	BD Pharmingen™	Cat#550281; RRID:AB_2275792
CD11b	ThermoFisher	Cat#MA110080;RRID:AB_11153179
MHCII	DRFZ	Fatima et al.32
4HNE Michael Adducts	Millipore	Cat#393207; RRID:AB_10679522
Cathepsin B (CTSB)	Proteintech	Cat#12216-1-AP; RRID:AB_2086929
Alpha-6 integrin (Human, NMR)-PE	Abcam	Cat#AB95703; RRID:AB_10681089
Alpha-6 integrin (Human, NMR)-PE	Invitrogen	Cat#12-0495-82; RRID:AB_891480
Chemicals, peptides, and recombinant proteins
Phorbol 12-myristate 13-acetate	Sigma-Aldrich	P8139-1 MG; CAS:16561-29-8
7,12-Dimethylbenz[a]anthracene	Sigma-Aldrich	D3254; CAS: 57-97-6
Dispase II	Roche	Cat#04942078001
Accutase	Gibco	Cat#A1110501
Liberase DH	Roche	Cat#05401054001
TRI Reagent	Zymo Research	Cat#R2050-1-200
Critical commercial assays
LIVE/DEAD™ Fixable Violet Dead Cell Stain Kit	Invitrogen	Cat# L34955
Direct-Zol Miniprep Kit	Zymo Research	Cat#R2052
Direct-Zol Microprep Kit	Zymo Research	Cat#R2060
NEBNext® rRNA Depletion Kit v2	New England Biolabs	Cat# E7400
NEBNext® Ultra™ II Directional RNA Library Prep Kit for Illumina	New England Biolabs	Cat# E7760S
QIAamp DNA Mini Kit	Qiagen	Cat#56304
Swift 2S® Sonic DNA Library kit	Swift Biosciences	Cat#42024
Deposited data
RNA-seq raw and analyzed data	This paper	GEO: GSE334395, GSE334586, GSE334587
WGS raw data	This paper	PRJNA1480147, PRJNA1481185
Experimental models: Organisms/strains
Naked mole-rat	Boston University ASC	Fatima et al.32
Mouse: NU/J: Foxn1nu	The Jackson Laboratory	RRID:IMSR_JAX:002019
Mouse: FVB/NJ	The Jackson Laboratory	RRID:IMSR_JAX:001800
Software and algorithms
Cutadapt	N/A	https://cutadapt.readthedocs.org/
Bowtie2	N/A	https://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Tophat2	N/A	https://ccb.jhu.edu/software/tophat/index.shtml
Ballgown	N/A	https://www.bioconductor.org/packages/release/bioc/html/ballgown.html
QIAGEN IPA	N/A	https://digitalinsights.qiagen.com/products-overview/discovery-insights-portfolio/analysis-and-visualization/qiagen-ipa/
TSG 2.0	N/A	http://bioinfo.mc.vanderbilt.edu/TSGene/
CGC	N/A	https://cancer.sanger.ac.uk/cosmic/login
ClusterProfiler	N/A	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html
ImageJ	N/A	https://imagej.net/ij/download.html
GraphPad Prism	N/A	https://www.graphpad.com/

Experimental model and study participant details

Animals

All experiments were conducted in accordance with the protocols approved by Boston University Institutional Animal Care and Use Committee (IACUC; protocol numbers PROTO202100000008, PROTO201900096) and United Kingdom Animal (Scientific Procedures) Act 1986 Amendment Regulations 2012 under a Home Office Project License. Female FVB (FVB/NJ) and nude mice (NU/J: Foxn1^nu) were purchased from Jackson Laboratories and maintained under specific pathogen-free conditions. Mice were housed under a 12 h light/12 h dark cycles at a temperature of ∼22 °C and humidity of ∼50%. Animals were maintained under standard husbandry conditions with ad libitum access to food and water. NMRs (Heterocephalus glaber) were maintained in a custom-made caging system with conventional mouse/rat cages connected by different lengths of tunnel. The room harboring NMR colonies was warmed to 28 °C, 50–60% humidity levels were maintained, and bedding, nesting material and water-enriched food were provided according to recommendations published previously (Buffenstein, 2005). Nude mice were approximately 7–8 weeks old at the time of grafting. At the initiation of skin carcinogenesis experiments, nude mice were 12 weeks old, FVB mice were approximately 7–8 weeks old, and naked mole-rats (NMRs; males and females) were approximately 6 months old.

Naked mole-rat skin transplantation onto nude mice

Skin grafting experiments were performed in Animal Science Center, Boston University according to the IACUC guidelines. Immunodeficient nude mice (NU/J) were acclimatized for two weeks before the transplantation experiment (n = 12). 6-7 month-old NMRs (n = 2) were euthanized, nude mice were anesthetized, and their back skin was sterilized with Povidone-Iodine Swab-sticks, then cleaned with alcohol pads and marked for excision. Mice were injected with pre-surgical analgesics for pain relief and 6 mm full-thickness excision was made on the back skin. To keep graft-bed moist, a small amount of sterile PBS was applied, and pre-marked NMR skin was excised and laid flatly within the graft bed, fully covering the area. Edges of the grafts were then glued to the ends of mice skin using skinstitch glue. After glue was dried, graft was covered by sterile gauze and secured with bandage with the help of sutures until skin was fully healed. Few weeks after grafting animals were fully recovered, and two-stage chemically induced carcinogenesis was applied.

Two-stage chemically induced skin carcinogenesis protocol

Two-stage chemically induced carcinogenesis was performed as described previously.⁶ 7,12-dimethylbenz[a]-anthracene (DMBA) (1 mg/mL; 0.1%, w/v or 390 nmol per 100-μL) and 12-O-tetra-decanoylphorbol-13-acetate (TPA) (0.1 mg/mL; 0.01%, w/v or 16.2 nmol per 100-μL) were dissolved in acetone. 6-7-month-old NMRs (n = 12) and FVB mice (7–9 weeks old, n = 15) of both sexes were used for experiments. FVB/N mice develop an unusually high incidence of squamous cell carcinomas after DMBA/TPA treatment compared to other mouse strains.⁶² In FVB mice, back skin hairs in resting (telogen) phase of the hair cycle were shaved using surgical clipper and shaver. FVB mice and NMRs were divided into 3 groups: Acetone-treated (Control, n = 6), DMBA-treated (n = 7), DMBA/TPA-treated (n = 8). Acetone or a single dose of DMBA (100 μL) was applied on the back skin of animals. At day 10, animals were euthanized, and skin was harvested from DMBA only group and half of the control group. In remaining animals, 1 week after DMBA treatment, acetone or TPA was applied for up to 6.5–12 weeks (FVB mice), 13–20 weeks (nude mice) and 25 weeks (NMRs).

For stimulation of ferroptosis in FVB mice, Erastin was dissolved in 70% ethanol and applied topically at a final dose of 10 mg/kg according to recommendations, published previously.⁴⁴ Application began one day after DMBA initiation and continued daily for 8 days, after which Erastin was applied three times per week following TPA promotion. FVB mice were assigned to six groups: acetone-treated controls (n = 6), DMBA + vehicle (n = 4), DMBA + Erastin (n = 5), DMBA/TPA + vehicle (n = 5), DMBA/TPA + Erastin (n = 8), and Erastin-only controls (n = 3). Skin samples were collected from DMBA-only groups on day 10 and from DMBA/TPA groups at week 11, then processed using established protocols. At experimental endpoints, animals were euthanized, and skin was harvested and processed according to established protocols (see below).

Method details

Histology and immunofluorescent analysis

Skin samples were collected from the dorsum of animals, covered in Tissue-Tek medium (VWR, UK), snap-frozen in liquid nitrogen, and stored in −80 °C freezer.^63,64 For histology and immunofluorescent analysis, 6–8μm sections were cut using Cryostat and stored at −80 until use. For hematoxylin/eosin staining, sections were fixed in 10% formalin, and staining was performed using Vectorlabs Kit (Cat# H-3502) using manufacturer’s protocol. For immunofluorescence, 8um cryosections were fixed in 10% formalin or Acetone. Sections were then blocked in 10% serum and incubated with primary antibodies overnight at +4 C with or without prior antigen retrieval, followed by 3 washes with 1× PBS and incubation with the secondary antibodies (1:500) at 37°C for 45min, as described previously.^65,66,67 For double staining of pH2AX/CTSB, sections were first incubated with primary and secondary antibody for pH2AX, followed by sequential incubation with primary and secondary antibody for CTSB. Immunomorphometric analysis was performed by quantifying positive cells or immunofluorescence signal intensity—such as corrected total cell fluorescence (CTCF) using ImageJ software, as described previously.⁶⁸

Quantification of immunofluorescence staining was carried out using two complementary metrics: the proportion of positive nuclei relative to total nuclei, and evaluation of an equivalent number of microscopic fields per sample acquired under uniform imaging conditions. Microscopic fields were selected consistently across samples to ensure assessment of comparable tissue regions, thereby reducing sampling variability and enabling reliable quantitative comparisons. Statistical analysis was performed using unpaired Student’s t test; differences were deemed significant if p < 0.05.

Fluorescence-activated cell sorting (FACS)

Whole back skin was isolated from mice and NMRs and incubated with 0.5% of Dispase II (Roche, #04942078001) at 4 °C overnight in 1xHBSS (Sigma, #H2387) without Ca²⁺ and Mg²⁺. Epidermis was separated from the dermis and was cut into small pieces and digested with Accutase (Gbico, # A1110501) containing 0.1% of Liberase DH (Roche, #05401054001) at 37 °C and centrifuged at 800 rpm for 20 min. Cells were filtered with 70 μm Nylon Mesh, and stained with a6-integrin (Cd49f)-PE (Invitrogen, #2196653, 1:500) antibodies for 30 min on ice, washed by using 1% FBS, followed by staining with LEAVE/DEAD™ Fixable Violet Dead Cell Stain Kit (Life Technologies, #L34955, 1:1000) for 30 min on ice intercepted by 1% FBS washing between each staining. Cd49f-expressing epidermal keratinocytes were separated from other cell populations, as described previously (Figures S1A and S1B).⁶⁹

RNA-sequencing analysis

For RNA-Seq analysis of epidermis, skin was collected and cleaned thoroughly to remove any fat or muscle tissue and washed in PBS. Skin was then digested as described above, epidermis was carefully separated from the dermis and was quick-frozen in liquid nitrogen and stored at −80 for further use. Frozen tissue was then minced and homogenized in TRI Reagent (Zymo Research# R2050-1-200) and RNA was isolated using Direct-Zol Miniprep Kit (Zymo Research# R2052) following manufacturer’s protocol. For RNA-Seq analysis of epidermal keratinocytes, NMR and mouse skin was digested overnight with Dispase II (5 mg/mL) and separated epidermis was then digested with TrypLE Express. Detached keratinocytes were washed, labeled with anti-CD49f antibody and FACS-sorted, as described above. Keratinocytes were lysed in TRI Reagent and RNA was isolated using Direct-Zol Microprep Kit (Zymo Research# R2060).

For RNA-sequencing library preparation, concentration, quality and integrity of RNA was checked using Qubit and 5200 Fragment Analyzer and RNA with RIN>7 was used for the libraries. Libraries were prepared using Next Generation Sequencing Library Preparation kit from NEBNext according to manufacturer protocol. Briefly, ribosomal RNAs (rRNAs) were depleted using NEBNext rRNA Depletion Kit v2 (cat# E7400L) and libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (cat# E7760S). After purifications, quality of the libraries was assessed on 5200 Fragment Analyzer with DNA Chip. NMR libraries were sequenced by LC Sciences (Houston, TX), while mouse libraries were sequenced by Med Genomics (Foster City, CA). The following libraries were included in the analysis: Acetone (n = 2) and DMBA (n = 2) at day 10, Acetone (n = 2) and DMBA/TPA (n = 2) at day 46–52.

Whole genome sequencing

For whole genome sequencing (WGS), skin was processed and digested as described previously. Frozen epidermal samples were then minced and homogenized in lysis buffer and DNA was isolated using QIAamp DNA Mini Kit (Qiagen; Cat#56304) according to manufacturer protocol. For WGS library preparation, DNA concentration, quality and integrity were checked using Qubit and 5200 Fragment Analyzer. DNA was fragmented using Covaris and libraries were prepared using Swift 2S Sonic DNA Library kit (Swift Biosciences; Cat# 42024) kit according to manufacturer protocol. After purifications, quality of the libraries was assessed on 5200 Fragment Analyzer with DNA Chip. The following libraries were included in the analysis: Acetone-treated (n = 2) and DMBA-treated (n = 2) collected at day 10, Acetone-treated (n = 2) and DMBA/TPA-treated (n = 2) collected at day 46.

WGS libraries were sequenced and analyzed by Med Genomics (Foster City, CA). The paired-end reads were aligned to the reference Naked mole rat (Heterocephalus_glaber) genome or Mouse genome (GRCm38) downloaded from Ensembl database. Somatic variant calling was performed using Strelka program (v2.9.10). Default settings were used for variant calling. Reads after duplicate removal were used for variant calling. The identified somatic variants were further filtered and only passed variants were considered for downstream analysis. Default parameters provided by Strelka were used to filter passed somatic variants.

Bioinformatics and STRING protein-protein interaction network analyses

Bioinformatics analysis of the RNAseq data was performed by the LC Sciences (Houston, TX). Firstly, Cutadapt⁷⁰ and Perl scripts were used to remove the reads that contained adaptor contamination, low quality bases and undetermined bases. Then, sequence quality was verified using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Bowtie2⁷¹ and Tophat2⁷² were used to map reads to the genome of Heterocephalus glaber (https://useast.ensembl.org/Heterocephalus_glaber_female/Info/Index). The differentially expressed mRNAs were selected with log2 (fold change) > 1 or log2 (fold change) <-1 and with parametric F-test comparing nested linear models (p value <0.05) by R package Ballgown. Ballgown was used to generate a list and heatmaps of differentially expressed genes, and in-house generated Perl scripts (LC Science, Houston, TX) were used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses. In addition, to cluster differentially expressed genes into different functional groups, the QIAGEN Ingenuity Pathway Analysis (QIAGEN IPA) was used (https://digitalinsights.qiagen.com/products-overview/discovery-insights-portfolio/analysis-and-visualization/qiagen-ipa/) as a general platform, and skin-relevant functional gene sub-categories were manually organized accordingly to previously published data.^73,74 Also, the significantly differentially expressed genes were input into the Tumor Suppressor Genes database (TSG 2.0, University of Texas, Houston, TX)³⁷ or CGC (https://cancer.sanger.ac.uk/census) databases. In addition, ClusterProfiler R Package was used to identify KEGG Pathway enrichment and GO functional term enrichment for each of the gene sets.

To predict functional interaction of proteins encoded by differentially expressed genes in DMBA/TPA-treated epidermis of NMRs and mice, the search tool for retrieval of interacting genes (STRING) (https://string-db.org, vision 11.0) database was employed. Active interaction sources, including text mining, experiments, databases, and co-expression as well as species limited to “Heterocephalus glaber” and an interaction score >0.4 were applied to construct the PPI networks. GO and KEGG pathways were selected with the threshold of adjusted p-value <0.05.

Quantification and statistical analysis

Statistical analysis

All statistical analysis was done using GraphPad Prism (versions 10–11; GraphPad Software). Statistical comparisons were performed using Student’s t test between the control and treated groups within each specie unless otherwise stated. In all experiments, data are presented as mean ± SEM, mean measuring central tendency and SEM for standard error of the mean. Unless otherwise stated, n represents number of individual animals in each group, as specified for each species. Statistical significance was defined as p < 0.05. Exact sample sizes and additional statistical or other relevant details are provided in the figure legends. Levels of statistical significance are indicated as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Supplemental information

Table S31. Putative physical STRING protein-protein interactions of oncogenes downregulated in NMR epidermis after DMBA/TPA treatment, related to

Figure 6