Chronic expression of p16INK4a in the epidermis induces Wnt-mediated hyperplasia and promotes tumor initiation
Results .Epidermal p16 induction causes partial senescence features . To study the effects of p16-expressing cells on the adult skin we crossed mice carrying a doxycycline-activated human p16 gene (tet-p16)21with K5-rtTA mice31, allowing its inducible activation in the basal epidermis. Transgenic p16 protein was detected in ~40% of basal keratinocytes in the interfollicular epidermis (IFE) after 2 days of doxycycline (dox) treatment at 3 weeks of age (Fig.? 1a–c ). Tissues showed reduced phosphorylated Rb levels, consistent with expected Rb activation (Fig.? 1d ). Staining for Ki67 revealed a dramatic reduction in the number of proliferating IFE basal cells after 2 days of p16 activation (Fig.? 1e, f ). However, after 2 weeks of p16 activation, overall proliferation rates returned to those of control animals.Fig. 1: p16 induces cell-cycle arrest and cell hypertrophy, and disrupts hair-follicle growth.a Immunostaining of human p16 (brown) in back skin sections of K5-rtTA/tet-p16 and control tet-p16 mice treated with doxycycline (dox) for 2 days (2d) starting at 3 weeks of age. b Skin sections from the same mice stained for p16 (red) and K14 (green), which labels the basal epidermis. c Percentage of p16 + keratinocytes in the interfollicular epidermis (IFE) in mice treated with dox for 2 days (2d), 2 weeks (2w), or 6 months (6?m). Dots indicate individual mice, n ?=?5, 5, 4 in respective groups. d Skin sections from mice treated with dox for 2 days, stained for Rb phosphorylated on Thr-821/826 (p-Rb) (red) and K14 (green). e Immunostaining of the proliferation marker Ki67 (brown) in sections from mice treated with dox for 2 days (2d) or 2 weeks (2w). f Percentage of IFE keratinocytes expressing Ki67 in the same mice (dots). n ?=?4, 6, 4, 4. g SA-βGal staining (blue) of skin sections from?indicated mice, treated with dox for 2 weeks. h Percentage of SA-βGal positive IFE cells from indicated mice (dots) treated with dox for 2 weeks. n ?=?9, 10. i Skin sections from indicated mice stained for E-Cadherin (brown) to indicate cell circumference. j Area of epidermal keratinocytes from indicated mice (dots). Shown are values for control mice (tet-p16), and for p16 ? and p16 + keratinocytes in K5-rtTA/tet-p16 mice. n ?=?3, 6, 6 mice, >60 cells were scored per mouse. k H&E-stained skin sections from the indicated mice treated with dox for 2 weeks, showing hair-follicle morphology. l Number of hair follicles per field in indicated mice (dots), following dox treatment for 6 months. n ?=?9 per group. m FACS analysis of epidermal cells from indicated mice treated with dox for 6 months, stained for CD34, CD49f, and Sca1. Charts show Sca1 ? (follicular) cells only. Gate indicates percentage of CD34 + /CD49f high hair-follicle stem cells. Cells were pooled from three mice per group. All graphs indicate mean across mice?±?S.E.M, all scores were conducted visually from images. Blue labels DNA in all fluorescence images. P ?0.0001, t test. ns non-significant. Scale bars—20?μm, except k —100?μm. Full size image Increased numbers of cells expressing the senescence markers SA-βGal and SenTraGor32were detected in mice that have undergone 2 weeks of p16 activation (Fig.? 1g, h and Supplementary Fig.? 1a, b ), yet those were lower than the numbers of p16-expressing cells, suggesting that not all transgene-expressing cells enter a fully senescent state. The average size of keratinocytes was increased in p16-expressing mice, specifically cells expressing the transgene, resulting in a thickening of the epidermis, with no increase in cell numbers (Fig.? 1i, j and Supplementary Fig. 1c–e). Changes in other known markers of senescence such as p21, Lamin B1, DNA damage, or secreted factors were not detected. Together these findings indicate that p16 expression in epidermal keratinocytes induces cell-cycle arrest, cell hypertrophy, and some features of senescence.p16 induction led to delayed growth of hair follicles (HF), reduced hair-follicle stem cell (HFSC) proliferation, and the formation of deformed HFs during the anagen stage initiated at 3 weeks of age (Fig.? 1k and Supplementary Fig.? 1f, g ), reminiscent of the effect of p14 ARF activation33. Upon prolonged p16 expression, mice displayed a decrease in the number of HFs and hair loss (Fig.? 1l and Supplementary Fig.? 1 f, h), and the numbers of CD34 + /CD49f high HFSCs were dramatically reduced (Fig.? 1m ). p16 expression thus blocks the proliferation of HFSCs and leads to their gradual depletion.Prolonged p16 expression induces hyperplasia and dysplasia . To examine the effects of prolonged p16 expression on the epidermis, we activated the transgene for 6 months. At this timepoint we detected p16 protein in ~20% of IFE cells (Fig.? 1c ). Strikingly, p16-expressing mice exhibited a thickened epidermis, in which both cell number and average cell size were increased (Fig.? 2a–c ). We observed an increased proportion of suprabasal differentiated cells in p16-expressing skins, while basal layer marker expression (K14, K5) was maintained and expanded in some regions (Fig.? 2a, d, e and Supplementary Fig.? 2a, b ). These changes contrast with the normal young and aged mouse epidermis, which contains only a thin suprabasal layer (Supplementary Fig.? 2c ). p16-induced mice also often showed an expansion of the dermis with dense collagen deposition regions, and a reduction of fat layer width (Supplementary Fig.? 2d ). Although we did not observe a prominent inflammatory reaction, numbers of epidermal and dermal T-cells, as well as of macrophages, were increased (Supplementary Fig.? 2e, f ). Control mice in which GFP was activated instead of p16 for the same period did not show these changes (Supplementary Fig.? 2g ).Fig. 2: Chronic p16 expression causes epidermal hyperplasia and dysplasia.a H&E-stained skin sections of K5-rtTA/tet-p16 and control tet-p16 mice treated with dox for 6 months (6m). Typical (middle) and pronounced (right) phenotype regions are shown for p16-induced mice. b IFE keratinocyte cell number per field (40×) in indicated mice (dots). n ?=?7, 12 in respective groups. c Average area of IFE cells from the same mice, visually measured from E-cadherin-stained sections. n ?=?7, 12. d Sections from the same mice stained for K10 (red) labeling suprabasal cells and K14 (green) labeling basal cells. e Percentage of K10 + epidermal cells isolated from dissociated mouse ear epidermis of indicated mice (dots) after 6 months of dox treatment, scored by FACS. n ?=?3, 4. f Skin sections from the indicated mice stained for Ki67 (brown). Right image shows a region with dysplasia. g Percentage of Ki67 + basal IFE cells in indicated mice (dots) after 6 months of dox treatment. n ?=?3, 8. h Skin sections from same mice co-stained for human p16 (red), the proliferation marker Mcm7 (white) and K14 (green). i Percentage of Mcm7 + basal IFE cells in control mice (tet-p16), and among p16 ? and p16 + cells in the epidermis of K5-rtTA/tet-p16 mice treated with dox for 6 months. n ?=?3, 7, 7. j Percentage of BrdU + basal epidermal cells in same mice, scored from skin sections co-stained for BrdU and p16. n ?=?9 per group. k Serial sections of a human actinic keratosis lesion stained for p16 (middle) and Rb (right) (both brown). Staining for p16 in normal human skin is shown in left panel. Dotted lines label the basement membrane. l Percentage of p16 + cells within individual human actinic keratosis (AK) lesions (dots), among lesions showing p16 expression. n ?=?47. m AK lesion co-stained for p16 (red), Ki67 (white) and K14 (green). n Percentage of Ki67 + cells among p16 ? and p16 + cells in human AK lesions (dots). n ?=?47. All graphs indicate mean across mice?±?S.E.M, except panels l , n which indicate mean across lesions. All scores were conducted visually on images, except e . Blue labels DNA in all images. P ?0.0001, t test. ns non-significant. Scale bars—20?μm. Full size image Strikingly, proliferation rates in the epidermis of mice expressing p16 for 6 months were dramatically increased, with 81% of basal keratinocytes expressing Ki67 relative to 33% in control mice (Fig.? 2f, g ). Furthermore, some epidermal regions exhibited dysplasia (atypia), exhibiting changes in nuclear morphology and basal layer organization (Fig.? 2f , right panel). Co-staining for p16 and the proliferation marker Mcm7 revealed that 93% of basal IFE keratinocytes in which p16 was not activated (p16 – ) were proliferating, as opposed to 26% of basal IFE cells in control mice (Fig.? 2h, i ). A corresponding increase in BrdU incorporation, marking cells in S-phase, was observed (Fig.? 2j ). These findings indicate that prolonged p16 activity in a subset of epidermal cells is sufficient to induce the proliferation of the majority of surrounding keratinocytes. Surprisingly, 30% of p16-expressing (p16 + ) cells also stained positive for Mcm7, and 5% stained positive for BrdU, suggesting that some of the transgene-expressing cells possessed the ability to proliferate (Fig.? 2i, j ). Chronic epidermal p16 expression thus leads to hyperplasia and regional dysplasia, consistent with paracrine stimulation emanating from the p16 + cells.We tested whether p16-induced hyperplasia influenced the regenerative capacity of the skin following wounding. Healing was delayed in mice in which p16 was activated for 2 weeks prior to wounding, and a similar delay was observed in mice in which p16 was activated for 6 months (Supplementary Fig.? 3 ), indicating that the p16-induced hyperplasia does not lead to enhanced regenerative capacity.A subset of cells in human actinic keratosis express p16 . The appearance of dysplastic regions in p16-expressing mice was reminiscent of early-stage human premalignant epidermal lesions. To establish whether p16 + cells are detected within such lesions, we stained sections of human actinic keratosis (AK) lesions, the common precursors to cutaneous squamous cell carcinoma (cuSCC)34. Previous studies reported varying percentages and staining patterns of p16 in these lesions26 , 27 , 28. We detected p16 protein in 47 of 72 lesions (65%), collected in three medical centers, with a mean of 11.3% p16 + cells per lesion in those that contained such cells (0.8–41.8%, n ?=?47) (Fig.? 2k, l ). p16 + cells were often interspersed among or adjacent to dividing, p16 – , cells (Fig.? 2m, n and Supplementary Fig.? 4a ). An average of 6% of the cells in AK lesions positively stained for SenTraGor, suggesting the presence of senescent cells in them (Supplementary Fig.? 4b, c ). Rb protein was also detected in these lesions, excluding Rb deletion or silencing, which would render p16 ineffective (Fig.? 2k , right panel). Furthermore, as observed in the p16-induced mice, some p16 + cells were proliferating, based on Ki67 expression, yet in lower numbers than the surrounding cells (Fig.? 2m, n ). These findings are consistent with potential crosstalk between p16 + and p16 – subpopulations in human AK lesions, as suggested by the mouse model.p16 promotes formation of carcinogen-induced papillomas . The finding that p16 expression was sufficient to induce epidermal hyperplasia and dysplasia raised the possibility that cells carrying oncogenic mutations would be influenced by a p16-activated environment. We, therefore, tested whether chronic p16 expression would affect the rates of papilloma formation after carcinogen treatment, using the DMBA/TPA model35. We treated K5-rtTA/tet-p16 mice, as well as sibling control tet-p16 mice, with a single topical dose of the mutagen DMBA at 3 weeks of age, and 1 week later initiated p16 activation by dox treatment, and simultaneously began twice-weekly topical treatments with TPA. We tracked papilloma formation for 5 months. Strikingly, p16-expressing mice developed approximately double the number of papillomas than control mice (Fig.? 3a, b ). This indicates that p16 generates an environment that supports the ability of keratinocytes bearing an oncogenic mutation to initiate papillomas.Fig. 3: p16 promotes carcinogen-induced papilloma formation but slows lesion?growth.a Average number of papillomas per mouse formed over time (age in weeks) in K5-rtTA/tet-p16 and sibling control tet-p16 mice treated once with DMBA at 3 weeks, followed by continued TPA treatment and p16 activation by dox starting at 4 weeks. n ?=?14,?11 mice?per group combined from four independent experiments. b Images of indicated mice at 27 weeks of age, showing papillomas formed on back skins. c H&E staining of papillomas from control and p16-induced mice. d Mean diameter of papillomas from the same mice at 27 weeks. n ?=?84,?136 papillomas?per group. e Immunostaining for p16 (brown) in papillomas from same mice. Enlarged inset is shown on right. f Staining for Ki67 (brown) in papillomas from same mice. g Percentage of Ki67-stained area of the epithelial component of papillomas (dots) from indicated mice, scored by image analysis. n ?=?13,8 in respective groups. h Papillomas from control and p16-induced mice stained for K14 (green) and K10 (red), marking the basal and suprabasal layers, respectively. i Relative mRNA levels of epidermal differentiation-associated genes in whole papillomas from control and p16-expressing mice, measured by mRNA-seq. n ?=?4, 3. P ?0.05 for all genes. Graphs indicate mean values across mice ( a ) or across lesions ( d , g , i )?±?S.E.M. P ?0.01, P ?0.001, P ?0.0001, t test. Scale bars—200?μm, except f —100?μm. Full size image Interestingly, the papillomas that developed in p16-expressing mice were smaller than those in control mice, and contained fewer proliferating cells, which were more restricted to the basal epidermal regions of the lesions (Fig.? 3c–g and Supplementary Fig.? 5a ). p16 + cells were readily detected within these papillomas (Fig.? 3e ). SenTraGor staining indicated higher numbers of senescent cells within the p16-induced papillomas, relative to the adjacent non-transformed epidermis (Supplementary Fig.? 5b,c ), consistent with previous reports indicating that p16 exerts a stronger pro-senescent effect when activated in the context of a strong pro-mitotic signal36. Expression profiling of whole dissected papillomas revealed a dramatic elevation in markers of differentiated epidermal layers and reduced expression of stem cell-associated markers, in addition to reduced expression of cell cycle, signaling, and cytoskeletal genes (Fig.? 3h, i and Supplementary Fig.? 5d, e ).Together these findings indicate that p16-expressing cells in the non-transformed tissue stimulate neighboring mutated oncogene-expressing cells to form growing premalignant tumors. However, p16 expression within formed oncogene-expressing lesions restricts proliferation, increasing senescence and differentiation, and thereby reduces tumor growth, consistent with its known tumor-suppressive roles.p16 expression activates Wnt in epidermal keratinocytes . We next set out to uncover the mechanisms by which p16-expressing cells stimulate proliferation and induce hyperplasia. We generated triple-transgenic K5-rtTA/tet-p16/tet-GFP mice, in which p16 and GFP were co-activated, and treated the mice with dox for 6 months (Fig.? 4a ). Co-expression of GFP did not reduce the numbers of detected p16 + cells (Supplementary Fig.? 6a, b ). We then isolated GFP + keratinocytes from these mice and from K5-rtTA/tet-GFP control mice expressing only GFP (Fig.? 4b and Supplementary Fig.? 6c ). The transcriptome of GFP + cells from p16-expressing mice revealed changes in multiple genes relative to GFP + cells from control mice (624 upregulated and 750 genes downregulated, Supplementary Data? 1 ), representing pathways known to be involved in senescence, including cell-cycle regulation, cytoskeletal structure, cell adhesion, protein turnover and metabolism (Fig.? 4c ). Several cytokine-encoding genes were upregulated, including Ccl20 , Cxcl1, and Cxcl9 , as well as Tgfβ ligands, yet these did not constitute a full SASP (Fig.? 4d ). Interestingly, we found that the levels of several genes encoding Wnt ligands, as well as known Wnt transcriptional target genes, were elevated, while Fzd receptor-encoding genes showed reduced mRNA levels (Fig.? 4e ). Tcf7 , which encodes the Wnt-target-activating Tcf1 transcription factor (and a Wnt target itself)37 , 38was upregulated; conversely, Tcf7l1 and Tcf7l2 , which respectively encode the Tcf3 and Tcf4 factors and are known to often act as Wnt target repressors37 , 38, were downregulated, consistent with transcriptional target activation (Fig.? 4e ). The proliferation-associated Ccnd genes (encoding Cyclin D proteins) which also respond to Wnt, were among the upregulated genes, as was often previously observed in senescent cells39. Consistent with Wnt-pathway activation, we detected by immunohistology increased numbers of cells expressing nuclear β-catenin in the epidermis of p16-expressing mice, as well as cells expressing Tcf1, Cyclin D1 and CD44 (Fig.? 4f, g ).Fig. 4: p16 expression activates Wnt-pathway-associated genes.a Diagram of transgenic mouse lines crossed for co-induction of GFP and p16. b FACS plots showing GFP levels and gates used for isolation of GFP + epidermal cells from K5-rtTA/tet-GFP/tet-p16 (right) and control K5-rtTA/tet-GFP (left) mice, after 6 months of induction. Plots show the CD31 ? /CD140a ? /CD45 ? cell fraction only. c Gene sets whose expression was preferentially upregulated (blue) or downregulated (grey) in GFP + cells from p16-expressing mice relative to GFP + cells from control mice. Values indicate –log 10 (adjusted P value) by hypergeometric test. d mRNA levels of genes encoding cytokines and genes associated with the TGFβ pathway upregulated in p16-expressing GFP + ?cells, measured by mRNA-seq. Values were normalized to the mean expression levels in GFP + ?cells from control mice, defined as 1. P ?0.05 for all genes, n ?=?4 samples per group, each pooled from 3 mice. e Relative mRNA levels of genes associated with the Wnt pathway in the same samples. P ?0.05 for all genes, n ?=?4 samples per group, each pooled from 3 mice. f Skin sections from K5-rtTA/tet-p16 and control tet-p16 mice treated with dox for 6 months stained for β-Catenin, Tcf1, Cyclin D1 or CD44. Arrowheads indicate representative positively stained cells. g Percentages of nuclear β-catenin + , Tcf1 + , and CyclinD1 + IFE cells in control and p16-expressing mice after 6 months of induction, as scored visually from images. P values calculated by t test. All graphs indicate mean across mice?±?S.E.M. Scale bars—20?μm. Full size image Isolation and transcriptome analysis of the GFP – cell fraction from p16-expressing mice revealed that, relative to the GFP + cells from the same mice, this fraction was enriched for proliferation-associated gene sets (as expected from their hyperproliferative state), and showed reduced expression of senescence-related sets, such as oxidative stress response, protein translation, and cytoskeletal reorganization (Supplementary Fig.? 6c–e ). Keratinocyte differentiation genes were somewhat elevated, reflecting the presence of diverse differentiation states in the expanded epidermis (Supplementary Fig.? 6f ). We found that, similar to the GFP + p16-expressing cells, the GFP – fraction showed elevation in Wnt-associated genes relative to cells from control mice, and that the Tcf7l1 gene was further silenced in these cells (Supplementary Fig.? 6g ). Together these findings indicate that p16 expression leads to activation of Wnt-pathway genes in both the p16-expressing cells and the surrounding IFE cells. The activation of the Wnt pathway by p16 appeared to occur gradually after prolonged p16 expression, as we did not observe elevation in Wnt-associated genes after 2 days or 2 weeks of p16 induction (Supplementary Fig.? 6h ).p16-induced Wnt secretion drives keratinocyte proliferation . To better dissect p16-induced Wnt activity, we overexpressed p16 in cultured primary mouse keratinocytes by lentiviral transduction. p16 overexpression led to rapid cell-cycle arrest, acquisition of a senescent morphology, and expression of senescence-associated markers, yet, interestingly, with no evidence of SA-βGal activity (Fig.? 5a, b ). Within 10 days of p16 expression, the levels of multiple Wnt ligands and targets were upregulated in the p16-expressing cells (Fig.? 5c ). ELISA measurement indicated that Wnt3a levels in conditioned medium (CM) of p16-expressing cells was elevated, reaching 3?ng/ml (Fig.? 5d ). Treatment of the cells with XAV-939, a Tankyrase1/2 inhibitor that causes β-catenin degradation40, suppressed the upregulation of Wnt target genes, including the genes encoding Cyclin D1 and D2, as well as of Wnt ligands (Fig.? 5c ). The activator Tcf7 gene was upregulated in p16-expressing cells, whereas the repressors Tcf7l1 and Tcf7l2 were both downregulated by p16 overexpression, as was observed in the transgenic mice, and XAV-939 treatment reversed this effect (Fig.? 5c ). These findings indicate that p16 is sufficient to activate the Wnt pathway in keratinocytes, driving both target and ligand activation in a β-catenin-dependent manner.Fig. 5: p16-overexpressing primary keratinocytes stimulate proliferation of naive cells through Wnt secretion.a Primary mouse keratinocytes infected with a p16-expressing or control lentivirus, co-stained for K14 (green), p16 (red) and BrdU (indicating S-phase, white), or for SA-βGal, 10 days after infection. b Relative mRNA levels of the indicated genes in the same keratinocytes, measured by qRT-PCR. n ?=?3 replicates. P ?0.05 for all genes. c Relative mRNA levels of the indicated Wnt-associated genes measured by qRT-PCR in control and p16-expressing keratinocytes. Values are also shown for the same cells treated with the β-catenin inhibitor XAV-939 (5?μM) for 24?h. n ?=?3 replicates. P ?0.05 for all comparisons indicated by lines: p16 versus control (dark blue versus dark grey, short lines), and p16 versus p16+XAV-939 (light blue versus dark blue, long lines). d Secreted Wnt3a protein levels in medium of control and p16-expressing cells, measured by ELISA. n ?=?3 replicates. P ?0.0001 e Fold increase over 4 days in numbers of primary keratinocytes cultured with conditioned media from p16-expressing (CM-p16) or from control (CM-Cont) keratinocytes, or with 10?ng/ml recombinant Wnt3a, in the absence or presence of XAV-939 (darker and lighter columns, respectively). n ?=?6 replicates. P ?0.0001. f Relative mRNA levels of the indicated Wnt-associated genes measured by qRT-PCR in keratinocytes treated for 24?h with conditioned media from p16-expressing or control cells, or with Wnt3a, in the presence or absence of XAV-939. n ?=?3 replicates. P ?0.05 for all comparisons indicated by lines: CM-p16 versus CM-Cont (dark blue versus dark grey, short lines), and CM-p16 versus CM-p16+XAV-939 (light blue versus dark blue, long lines). All graphs indicate mean across replicates?±?S.E.M., t test. Scale bars—50?μm. Full size image To test whether Wnt activation and secretion by p16-expressing keratinocytes can influence naive cells, we treated proliferating primary keratinocytes with conditioned media from p16-expressing or control (empty vector-expressing) keratinocytes, or, as a positive control, with recombinant Wnt3a. While control-cell CM had no effect on the numbers of treated cells, the CM of p16-expressing cells led to enhanced proliferation of the treated cells, resulting in a 1.7-fold increase in cell numbers after 4 days of treatment, similar to the effect of Wnt3a (Fig.? 5e ). This stimulatory effect was almost completely lost when XAV-939 was administered together with CM or Wnt3a, indicating that proliferation is β-catenin mediated (Fig.? 5e ). Treated cells showed upregulation of Wnt target genes and downregulation of Tcf71,2 , as well as increased levels of genes encoding Wnt ligands (Fig.? 5f ), mirroring the changes observed in the p16-expressing mice. These changes were blocked by XAV-939 (Fig.? 5f ). Together, these findings indicate that Wnt secretion by p16-expressing cells can stimulate the proliferation of naive keratinocytes, and that Wnt signals act on both the p16-expressing cells themselves and on the non-expressing proliferating cells, activating transcriptional targets as well as Wnt ligands, in a β-catenin-dependent manner (Supplementary Fig.? 7 ).The Wnt-pathway mediates p16-induced hyperplasia . We next tested whether Wnt activation contributes to the hyperplasia observed in p16-expressing mice. We activated p16 for 6 months, and then treated the mice systemically with XAV-939 for 1 week. XAV-939 treatment led to a significant reduction in epidermal proliferation rates, as well as a reduction in the expression of Wnt ligand and target genes (Fig.? 6a–c ). Wnt inhibition was therefore sufficient for a partial reversal of the ongoing hyperplastic effects of p16 expression.Fig. 6: Wnt activity mediates p16-induced epidermal hyperplasia.a Skin sections from control tet-p16 mice (left) and K5-rtTA/tet-p16 mice treated with dox for 6 months, and subsequently treated for 1 week with XAV-939 (right) or with vehicle (middle), stained for Ki67 (brown). b Percentage of Ki67 + basal IFE cells in sections from same mice (dots). n ?=?7, 8, 8, combined from two independent experiments. c Relative mRNA levels of the indicated Wnt-associated genes in isolated epidermis from the same mice, measured by qRT-PCR. n ?=?8 mice per group. P ?0.05 for all comparisons between p16-expressing mice (blue) and XAV-939-treated mice (green). d Skin sections from control (tet-p16), K5-rtTA/tet-p16 and K5-rtTA/tet-p16/tet-Tcf3 mice treated with dox for 6 months, stained for p16 (red) and Tcf3 (white). e Skin sections from the same mice as in d stained for Ki67 (brown). f Percentage of Ki67 + basal IFE cells in mice expressing p16 (p16) or p16 together with Tcf3 (p16+Tcf3) for 6 months, and from controls (Cont). n ?=?7, 4, 5. g Relative mRNA levels of the indicated Wnt-associated genes in isolated epidermis from the same mice, measured by qRT-PCR. n ?=?7, 4, 5. Scores were done visually from images. All graphs indicate mean values across mice?±?S.E.M. P ?0.05 for all comparisons between p16-expressing mice (blue) and mice co-expressing p16 and Tcf3 (green). P ?0.0001. t test was used for all significance tests. Scale bars—20?μm. Full size image We next tested whether co-activation of the repressive Tcf3 (Tcf7l1) together with p16 would influence Wnt-pathway activation and hyperplasia. We crossed K5-rtTA/tet-p16 mice with mice carrying an inducible Tcf3 transgene (tet-Tcf3)41to generate triple-transgenic mice. Co-activation of Tcf3 together with p16 for 6 months resulted in significantly reduced epidermal hyperplasia, as well as in reduced levels of Wnt ligand and target genes (Fig.? 6d–g ). Tcf3 expression thus largely blocks the pro-proliferative effects of p16. Together these results indicate that activation of the Wnt pathway upon chronic p16 expression contributes to the induction of epidermal hyperplasia.Senolytic elimination of p16 + cells suppresses hyperplasia . To study whether the continued presence of p16-expressing cells in the epidermis is required for maintenance of hyperplasia, even after prolonged stimulation, we tested whether these cells can be eliminated by senolytic treatment. The Bcl-2 protein-family inhibitor ABT-737 and its analog ABT-263 (navitoclax) inhibit Bcl-2, Bcl-xl, and Bcl-w, and were shown to have senolytic activity in various settings16 , 42 , 43. We previously showed that ABT-737 eliminates epidermal cells in which the p53 activator, p14 ARF , was overexpressed42. However, it was unclear, in light of the partial induction of senescence by p16, whether ABT-737 would be effective in eliminating p16-expressing cells. Following six months of p16 induction, we treated mice with ABT-737 six times over nine days and then sacrificed the mice. ABT-737-treated mice showed a significant reduction in the percentage of p16 + cells in the IFE (an approximate 60% reduction), and a corresponding reduction in basal cell proliferation and epidermal thickness, indicating partial reversal of the hyperplasia (Fig.? 7a–d ). Furthermore, the expression of Wnt ligands and targets was significantly reduced in the ABT-737-treated mice (Fig.? 7e ). p16-expressing cells are thus sensitive to senolytic treatment, and their residence in the epidermis is necessary for the ongoing maintenance of Wnt stimulation and consequent hyperplasia.Fig. 7: Senolytic treatment inhibits p16-induced hyperplasia and Wnt activation.a Skin sections from K5-rtTA/tet-p16 and control tet-p16 mice treated with dox for 6 months and subsequently treated with ABT-737 (6 injections over 9 days), stained by H&E, for p16, or for Ki67. b Percentage of p16 + IFE cells in ABT-737-treated and untreated p16-induced and control?mice (dots). c Percentage of Ki67 + basal IFE cells in same mice. d Epidermal thickness in same mice. e Relative mRNA levels of the indicated Wnt-pathway-associated genes in whole skins isolated from the same mice, measured by qRT-PCR. P ?0.05 for all comparisons indicated by lines: p16 versus control (dark blue versus dark grey, long lines) and p16+ABT-737 versus p16 (light blue versus dark blue, short lines). n ?=?4, 4, 4, 5 in respective groups in all panels. All graphs indicate mean values across mice?±?S.E.M. Scores were conducted visually from images. P ?0.0001, t test. Scale bars—20?μm. Full size image .