Advertisement
Journal of Investigative Dermatology Home

HDAC1/2 Control Proliferation and Survival in Adult Epidermis and Pre‒Basal Cell Carcinoma through p16 and p53

  • Xuming Zhu
    Affiliations
    Department of Dermatology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA

    Black Family Stem Cell Institute, Icahn School of Medicine at Mount Sinai, New York, New York, USA

    Department of Cell, Developmental and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, New York, USA
    Search for articles by this author
  • Matthew Leboeuf
    Affiliations
    Department of Dermatology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA

    Cell & Molecular Biology Graduate Group, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA

    Department of Surgery, Geisel School of Medicine at Dartmouth, Lebanon, New Hampshire, USA
    Search for articles by this author
  • Fang Liu
    Affiliations
    Department of Dermatology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for articles by this author
  • Marina Grachtchouk
    Affiliations
    Department of Dermatology, Michigan Medicine, University of Michigan, Ann Arbor, Michigan, USA
    Search for articles by this author
  • John T. Seykora
    Affiliations
    Department of Dermatology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for articles by this author
  • Edward E. Morrisey
    Affiliations
    Department of Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for articles by this author
  • Andrzej A. Dlugosz
    Affiliations
    Department of Dermatology, Michigan Medicine, University of Michigan, Ann Arbor, Michigan, USA
    Search for articles by this author
  • Sarah E. Millar
    Correspondence
    Correspondence: Sarah E. Millar, Black Family Stem Cell Institute, Icahn School of Medicine at Mount Sinai, Icahn Building, Floor 13 Room 020C, 1425 Madison Ave, New York, New York 10029, USA.
    Affiliations
    Department of Dermatology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA

    Black Family Stem Cell Institute, Icahn School of Medicine at Mount Sinai, New York, New York, USA

    Department of Cell, Developmental and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, New York, USA

    Department of Dermatology, Icahn School of Medicine at Mount Sinai, New York, New York, USA
    Search for articles by this author
Open AccessPublished:July 17, 2021DOI:https://doi.org/10.1016/j.jid.2021.05.026
      HDAC inhibitors show therapeutic promise for skin malignancies; however, the roles of specific HDACs in adult epidermal homeostasis and in disease are poorly understood. We find that homozygous epidermal codeletion of Hdac1 and Hdac2 in adult mouse epidermis causes reduced basal cell proliferation, apoptosis, inappropriate differentiation, and eventual loss of Hdac1/2-null keratinocytes. Hdac1/2-deficient epidermis displays elevated acetylated p53 and increased expression of the senescence gene p16. Loss of p53 partially restores basal proliferation, whereas p16 deletion promotes long-term survival of Hdac1/2-null keratinocytes. In activated GLI2-driven pre–basal cell carcinoma, Hdac1/2 deletion dramatically reduces proliferation and increases apoptosis, and knockout of either p53 or p16 partially rescues both proliferation and basal cell viability. Topical application of the HDAC inhibitor romidepsin to the normal epidermis or to GLI2ΔN-driven lesions produces similar defects to those caused by genetic Hdac1/2 deletion, and these are partially rescued by loss of p16. These data reveal essential roles for HDAC1/2 in maintaining proliferation and survival of adult epidermal and basal cell carcinoma progenitors and suggest that the efficacy of therapeutic HDAC1/2 inhibition will depend in part on the mutational status of p53 and p16.

      Abbreviations:

      BCC (basal cell carcinoma), HDAC (histone deacetylase), HH (hedgehog), K (keratin), P (postnatal day), rtTA (reverse tetracycline transactivator)

      Introduction

      Adult epidermal homeostasis involves both self-renewal and differentiation of stem and progenitor cells. The delicate balance between these two processes is perturbed in common skin conditions, including epidermal malignancies and hyperproliferative diseases. Chromatin regulators such as HDACs play critical roles in ensuring faithful inheritance of gene expression programs from mother to daughter cells and in permitting broad-scale, rapid responses to differentiation signals. HDACs function by removing acetyl groups from histone tails, generating compacted chromatin and a repressive transcriptional environment (
      • Kouzarides T.
      Acetylation: a regulatory modification to rival phosphorylation?.
      ). They also deacetylate key transcription factors, altering their activity and/or stability (
      • Higashitsuji H.
      • Higashitsuji H.
      • Masuda T.
      • Liu Y.
      • Itoh K.
      • Fujita J.
      Enhanced deacetylation of p53 by the anti-apoptotic protein HSCO in association with histone deacetylase 1 [published correction appears in J Biol Chem 2020;295:9264].
      ;
      • Tang Y.
      • Zhao W.
      • Chen Y.
      • Zhao Y.
      • Gu W.
      Acetylation is indispensable for p53 activation [published correction appears in Cell 2008;133:1290].
      ). HDAC inhibitors show promise as a potential treatment for cutaneous diseases, including skin cancers. However, the genetic requirements for individual HDAC proteins in adult skin are not fully understood. Delineating these requirements and unraveling the underlying mechanisms will be important in designing more specific and effective therapeutics.
      In line with decreased proliferation generally caused by HDAC inhibitors (
      • Duvic M.
      Histone deacetylase inhibitors for cutaneous T-cell lymphoma.
      ), constitutive codeletion of Hdac1 or Hdac2 in embryonic mouse epidermis causes complete failure of epidermal stratification, reduced proliferation, and increased apoptosis, associated with upregulation of the senescence factor p16 and hyperacetylation of the apoptotic regulator p53 (
      • LeBoeuf M.
      • Terrell A.
      • Trivedi S.
      • Sinha S.
      • Epstein J.A.
      • Olson E.N.
      • et al.
      Hdac1 and Hdac2 act redundantly to control p63 and p53 functions in epidermal progenitor cells.
      ). By contrast, homozygous epidermal deletion of Hdac1 combined with heterozygous deletion of Hdac2 using a constitutive Keratin (K) 5 (K5) promoter to drive expression of Cre recombinase causes hyperproliferation, accelerated differentiation, hair follicle degeneration, and accumulation of c-MYC protein in adult mouse epidermis, and accelerated tumor development in a mouse squamous cell carcinoma model (
      • Hughes M.W.
      • Jiang T.X.
      • Lin S.J.
      • Leung Y.
      • Kobielak K.
      • Widelitz R.B.
      • et al.
      Disrupted ectodermal organ morphogenesis in mice with a conditional histone deacetylase 1, 2 deletion in the epidermis.
      ;
      • Winter M.
      • Moser M.A.
      • Meunier D.
      • Fischer C.
      • Machat G.
      • Mattes K.
      • et al.
      Divergent roles of HDAC1 and HDAC2 in the regulation of epidermal development and tumorigenesis.
      ), suggesting that the proproliferative functions of elevated c-MYC may override the effects of partial loss of HDAC1/2 in this context. The consequences of double homozygous genetic deletion of Hdac1/2 in vivo have not yet been evaluated either in the normal adult epidermis or in skin tumor models.
      Basal cell carcinoma (BCC) is the most common skin tumor in Caucasian populations and is linked to sporadic or familial mutations, most commonly in the Hedgehog (HH) receptor PTCH1 or its coreceptor SMO, that cause constitutive signaling through the HH pathway. Although BCC lesions rarely metastasize, they are locally invasive and can cause massive local tissue damage if not treated promptly. Furthermore, familial BCC causing multiple lesions is hard to treat by excision (
      • Atwood S.X.
      • Whitson R.J.
      • Oro A.E.
      Advanced treatment for basal cell carcinomas.
      ). The Food and Drug Administration‒approved SMO antagonists vismodegib and sonidegib are effective in the initial treatment of BCC (
      • Sekulic A.
      • Von Hoff D.
      Hedgehog pathway inhibition.
      ), but lesions recur owing to acquisition of secondary downstream mutations in or downstream of SMO (
      • Atwood S.X.
      • Li M.
      • Lee A.
      • Tang J.Y.
      • Oro A.E.
      GLI activation by atypical protein kinase C ι/λ regulates the growth of basal cell carcinomas.
      ;
      • Brinkhuizen T.
      • Reinders M.G.
      • van Geel M.
      • Hendriksen A.J.
      • Paulussen A.D.
      • Winnepenninckx V.J.
      • et al.
      Acquired resistance to the Hedgehog pathway inhibitor vismodegib due to smoothened mutations in treatment of locally advanced basal cell carcinoma.
      ;
      • Pricl S.
      • Cortelazzi B.
      • Dal Col V.
      • Marson D.
      • Laurini E.
      • Fermeglia M.
      • et al.
      Smoothened (SMO) receptor mutations dictate resistance to vismodegib in basal cell carcinoma.
      ;
      • Sharpe H.J.
      • Pau G.
      • Dijkgraaf G.J.
      • Basset-Seguin N.
      • Modrusan Z.
      • Januario T.
      • et al.
      Genomic analysis of smoothened inhibitor resistance in basal cell carcinoma.
      ;
      • Urman N.M.
      • Mirza A.
      • Atwood S.X.
      • Whitson R.J.
      • Sarin K.Y.
      • Tang J.Y.
      • et al.
      Tumor-derived suppressor of fused mutations reveal hedgehog pathway interactions.
      ;
      • Wang C.
      • Wu H.
      • Evron T.
      • Vardy E.
      • Han G.W.
      • Huang X.P.
      • et al.
      Structural basis for smoothened receptor modulation and chemoresistance to anticancer drugs.
      ), resulting in a need for additional therapeutics that function downstream of SMO or target parallel pathways important for tumor development.
      HH signaling can upregulate HDAC1, which deacetylates GLI1 to enhance its activity in a positive feedback loop, suggesting HDAC inhibitors as potential treatments for HH-mediated malignancies (
      • Canettieri G.
      • Di Marcotullio L.
      • Greco A.
      • Coni S.
      • Antonucci L.
      • Infante P.
      • et al.
      Histone deacetylase and Cullin3-REN(KCTD11) ubiquitin ligase interplay regulates Hedgehog signalling through Gli acetylation.
      ). The HDAC inhibitor vorinostat was found to be mildly effective in reducing the growth of mouse BCC cells in vitro and in xenografts, and its effects were enhanced by cotreatment with a small-molecule inhibitor of an atypical protein kinase C family member that recruits HDAC1 to GLI1 (
      • Mirza A.N.
      • Fry M.A.
      • Urman N.M.
      • Atwood S.X.
      • Roffey J.
      • Ott G.R.
      • et al.
      Combined inhibition of atypical PKC and histone deacetylase 1 is cooperative in basal cell carcinoma treatment.
      ). The latter study utilized cells that were deficient in p53 as well as PTCH1; however, approximately 50% of human BCC do not display P53 mutations (
      • Rady P.
      • Scinicariello F.
      • Wagner Jr., R.F.
      • Tyring S.K.
      p53 mutations in basal cell carcinomas.
      ;
      • Soufir N.
      • Molès J.P.
      • Vilmer C.
      • Moch C.
      • Verola O.
      • Rivet J.
      • et al.
      P16 UV mutations in human skin epithelial tumors.
      ;
      • Zhang H.
      • Ping X.L.
      • Lee P.K.
      • Wu X.L.
      • Yao Y.J.
      • Zhang M.J.
      • et al.
      Role of PTCH and p53 genes in early-onset basal cell carcinoma.
      ). Thus, the consequences of homozygous loss of both HDAC1 and HDAC2 in postnatal epidermis in vivo and the extent to which targeting these factors could be therapeutically useful in BCC and other hyperproliferative skin conditions are not yet entirely clear.
      To address these questions, we carried out inducible in vivo epidermal-specific homozygous codeletion of Hdac1 and Hdac2 in postnatal mouse epidermis and in a mouse model of pre-BCC expressing a mutant, constitutively active form of the SHH pathway downstream transcription factor GLI2 (GLI2ΔN) (
      • Grachtchouk M.
      • Pero J.
      • Yang S.H.
      • Ermilov A.N.
      • Michael L.E.
      • Wang A.
      • et al.
      Basal cell carcinomas in mice arise from hair follicle stem cells and multiple epithelial progenitor populations.
      ). Our findings reveal crucial roles for HDAC1/2 in maintaining the proliferation and survival of adult epidermal progenitor cells.
      Our data further suggest HDAC inhibitors as promising therapeutics for epidermal hyperproliferative conditions and malignancies but indicate that their efficacy would be reduced in carcinomas that carry concomitant loss of function mutations in p53 and/or p16.

      Results

       Short-term inducible deletion of Hdac1/2 causes defective plantar epidermis

      Because epidermal homeostasis is influenced by secreted signals from cycling hair follicles (
      • Oh H.S.
      • Smart R.C.
      An estrogen receptor pathway regulates the telogen-anagen hair follicle transition and influences epidermal cell proliferation.
      ) and Hdac1 deletion affects postnatal hair follicle differentiation and cycling (
      • Hughes M.W.
      • Jiang T.X.
      • Lin S.J.
      • Leung Y.
      • Kobielak K.
      • Widelitz R.B.
      • et al.
      Disrupted ectodermal organ morphogenesis in mice with a conditional histone deacetylase 1, 2 deletion in the epidermis.
      ;
      • Winter M.
      • Moser M.A.
      • Meunier D.
      • Fischer C.
      • Machat G.
      • Mattes K.
      • et al.
      Divergent roles of HDAC1 and HDAC2 in the regulation of epidermal development and tumorigenesis.
      ), we examined the effects of Hdac1/2 deletion in the epidermis by analyzing hairless plantar mouse skin (Figure 1a), which exhibits ubiquitous expression of HDAC1 and HDAC2 (Figure 1b). We used a bitransgenic K5-reverse tetracycline transactivator (K5-rtTA) tetO-Cre system for doxycycline-inducible deletion of Hdac1 and Hdac2 in K5 promoter‒active basal cells (
      • Ramírez A.
      • Bravo A.
      • Jorcano J.L.
      • Vidal M.
      Sequences 5' of the bovine keratin 5 gene direct tissue- and cell-type-specific expression of a lacZ gene in the adult and during development.
      ;
      • Teta M.
      • Choi Y.S.
      • Okegbe T.
      • Wong G.
      • Tam O.H.
      • Chong M.M.
      • et al.
      Inducible deletion of epidermal Dicer and Drosha reveals multiple functions for miRNAs in postnatal skin.
      ). K5 promoter‒active basal cells serve as stem cells for the epidermis: they both self-renew and give rise to suprabasal progeny (
      • Blanpain C.
      • Fuchs E.
      Epidermal stem cells of the skin.
      ). Thus, as the epidermis turns over, which is completed within approximately 8–10 days in adult mice (
      • Potten C.S.
      • Saffhill R.
      • Maibach H.I.
      Measurement of the transit time for cells through the epidermis and stratum corneum of the mouse and guinea-pig.
      ), the Hdac1/2-deleted basal cells give rise to Hdac1/2–deleted suprabasal and terminally differentiated cells. Hdac1/2 deletion was initiated in K5-rtTA tetO-Cre mice carrying Hdac1fl/fl Hdac2fl/+ (Hdac1cKO Hdac2cHet), Hdac1 fl/+ Hdac2fl/fl (Hdac1cHet Hdac2cKO), or Hdac1fl/fl Hdac2fl/fl (Hdac1/2cKO) alleles from postnatal day (P) 20, and plantar skin samples were analyzed on P27. At this time point, HDAC1 and HDAC2 were efficiently deleted in basal and many suprabasal cells in the plantar epidermis of Hdac1cKO Hdac2cHet and Hdac1cHet Hdac2cKO mice, respectively (Supplementary Figure S1e–h); however, epidermal thickness and proliferation were not significantly altered compared with those in littermate controls lacking K5-rtTA or tetO-Cre (Supplementary Figure S1a–d and i–l and q and r), and apoptotic basal cells were absent (Supplementary Figure S1m–p).
      Figure thumbnail gr1
      Figure 1Hdac1/2 deletion causes apoptosis and reduced the proliferation of plantar epidermis that is partially rescued by codeletion of p53. (a) Schematic of plantar foot and timing of doxycycline treatment in panels c–z; samples in c–z were analyzed on P27. (b) IF for HDAC1, HDAC2, and K14 in wild-type P30 plantar skin. (c, d, p, q) Plantar skin histology for the genotypes indicated. (e–j, m, n, r–w, m, n) IF for the indicated proteins and genotypes. (m’–n’’) RNAscope with the indicated probes and genotypes. (k, l, x, y) TUNEL staining for the genotypes indicated. Arrows in l and y indicate apoptotic basal cells. White dashed lines indicate epidermal–dermal borders. (o) Quantitation of HDAC1/2-positive plantar epidermal basal cells; n = 5 Hdac1/2cKO and control mice. (z) Quantitation of Ki-67‒positive plantar epidermal basal cells in the genotypes indicated. Each data point represents the average percentage from ≥5 ×20 fields from each of four mice of each genotype. P-values were calculated with two-tailed Student’s t-test; error bars indicate SEM. Bar = 50 μm for b, e–n”, and r–y and 25 μm for c, d, p, and q. +ve indicates positive responses. IF, immunofluorescence; K, keratin; P, postnatal day.
      By contrast, in Hdac1/2cKO double knockout plantar epidermis, which displayed efficient removal of both HDAC1 and HDAC2 in the basal cell layer, with less than 10% of basal cells remaining positive for HDAC1/2 by immunofluorescence after 7 days of induction (Figure 1 e, f, and o), epidermal thickness (Figure 1c and d) and proliferation (Figure 1i, j, and z) were significantly reduced, and sporadic TUNEL-positive basal cells were present that were never observed in littermate controls lacking K5-rtTA or tetO-Cre (Figure 1k and l). Furthermore, the levels of H3K9Ac, which is normally deacetylated by HDACs (
      • Seto E.
      • Yoshida M.
      Erasers of histone acetylation: the histone deacetylase enzymes.
      ), were dramatically elevated (Figure 1g and h).

       Decreased proliferation of Hdac1/2CKO basal cells is partially rescued by loss of p53

      In line with data from the embryonic epidermis, we found that acetylated p53 was elevated in postnatal Hdac1/2-deficient plantar epidermis (Figure 1m and n), and p16 mRNA and protein were respectively increased in the plantar epidermis and hairy skin epithelial cells lacking HDAC1/2 (Supplementary Figure S2a and b) after 7 days of induction. Furthermore, mRNA levels for the p53-target genes p21 and MDM2 were elevated in Hdac1/2-deficient plantar epidermis (Figure 1m’ and n”). To test whether the effects of short-term Hdac1/2 deletion were dependent on p53 or p16, we generated Hdac1/2cKO mice that were also null for either p53 (p53KO) (
      • Jacks T.
      • Remington L.
      • Williams B.O.
      • Schmitt E.M.
      • Halachmi S.
      • Bronson R.T.
      • et al.
      Tumor spectrum analysis in p53-mutant mice.
      ) or p16 (p16KO) (
      • Sharpless N.E.
      • Bardeesy N.
      • Lee K.H.
      • Carrasco D.
      • Castrillon D.H.
      • Aguirre A.J.
      • et al.
      Loss of p16Ink4a with retention of p19Arf predisposes mice to tumorigenesis.
      ). Loss of p53 alone had no effect on epidermal thickness relative to that in the controls (Figure 1c and p). The absence of p53 in Hdac1/2cKO mice did not restore epidermal thickness to control levels (Figure 1q, compared with Figure 1c and d) and did not rescue hyperacetylation of H3K9Ac (Figure 1g, h, t, and u). Loss of p53 partially restored proliferation (Figure 1i, j, v, w, and z) but did not prevent apoptosis (Figure 1k, l, x, and y) of Hdac1/2-deleted basal cells. Interestingly, loss of p16 did not significantly rescue epidermal thickness (Supplementary Figure S2c–f), H3K9 hyperacetylation (Supplementary Figure S2k–n), proliferation (Supplementary Figure S2o–r and w), or apoptosis (Supplementary Figure S2s–v) of Hdac1/2-deleted epidermis. Together, these data indicate that the immediate effects of HDAC1/2 deletion on proliferation are mediated in part through increased p53 activity.

       HDAC1/2 are required for long-term maintenance of postnatal epidermal progenitor cells

      A total of 5–10% of basal cells remained undeleted 7 days after initiating doxycycline treatment in Hdac1/2cKO mice (Figure 1o). To determine whether undeleted cells could out compete deleted cells in the longer term, we analyzed plantar epidermis from mice that were induced with doxycycline food for 14 days (from P20 to P34) or for 22 days (P20‒P42). On P34, the majority of epidermal basal stem cells still lacked HDAC1/2 and had populated the suprabasal layers with Hdac1/2-deleted progeny (Figure 2c and d). Basal cell proliferation remained reduced at this stage (Figure 2e and f). We noted the appearance of suprabasal cells with enlarged nuclei and premature cornification (Figure 2a and b, arrow), indicating that HDAC1/2 play critical roles in suprabasal as well as in basal cells. In line with this, K5 and K14 expressions were expanded, and cells double positive for K14 and the suprabasal marker K10 or the granular layer marker loricrin were present (Figure 2e–j, arrows). Consistent with aberrant differentiation, we found that acetylated c-Myc was dramatically elevated in Hdac1/2cKO plantar epidermis (Figure 2k and l).
      Figure thumbnail gr2
      Figure 2Accelerated differentiation of Hdac1/2-mutant epidermis is associated with increased levels of acetylated c-MYC. Schematic shows the timing of doxycycline treatment; plantar skin samples were analyzed on P34. (a, b) Histology and (c–l) IF for the indicated proteins in control and in Hdac1/2-mutant skin as indicated; n ≥ 4 mice for each genotype. Bar = 25 μm for a and b and 50 μm for c–l. IF, immunofluorescence; K, keratin; P, postnatal day.
      By contrast with the persistence of HDAC1/2-deleted cells on P34, by P42, the epidermis was largely repopulated by undeleted cells (Figure 3a’, b’, and y). The repopulated epidermis was thickened (Figure 3a and b) and hyperproliferative (Figure 3c, d, and z) relative to controls. Expression of K14, K5, and K10 was expanded (Figure 3a’, b’, and c–f and Supplementary Figure S3a–d), and cells double positive for K14 and loricrin were present (Supplementary Figure S3c and d, arrow); however, premature cornification was less obvious than on P34 (Figure 3b, compared with Figure 2b), and the levels of acetylated c-MYC were similar to those of the controls (Supplementary Figure S3e and f). Basal cells no longer showed evidence of increased H3K9Ac (Figure 3e and f). Consistent with loss of Hdac1/2-deleted cells, basal cell apoptosis was absent (Figure 3g and h), and the levels of p53Ac in Hdac1/2cKO-mutant epidermis were similar to those of the controls (Supplementary Figure S3g and h). CD3+ T cells were absent in mutant plantar skin on P42 (Supplementary Figure S4a and b); however, the repopulated skin displayed signs of a persistent inflammatory response, including elevated expression of the inflammatory marker CD26 (
      • Arwert E.N.
      • Mentink R.A.
      • Driskell R.R.
      • Hoste E.
      • Goldie S.J.
      • Quist S.
      • et al.
      Upregulation of CD26 expression in epithelial cells and stromal cells during wound-induced skin tumour formation.
      ;
      • Novelli M.
      • Savoia P.
      • Fierro M.T.
      • Verrone A.
      • Quaglino P.
      • Bernengo M.G.
      Keratinocytes express dipeptidyl-peptidase IV (CD26) in benign and malignant skin diseases.
      ) in the epidermis and underlying dermis (Supplementary Figure S4c and d) and increased numbers of dermal mast cells compared with controls (Supplementary Figure S4e–g). These observations suggested that hyperplasia and disorganized differentiation were secondary to a damage response caused by loss of HDAC1/2 at earlier stages.
      Figure thumbnail gr3
      Figure 3Long-term maintenance of Hdac1/2-deleted cells requires p16. The schematic shows the timing of doxycycline treatment; samples were analyzed on P42. (a, b, i, j, q, r) Histology of plantar skin for the genotypes indicated. Arrow in r indicates premature cornification. (a’–f, i’–n, q’–v) IF for the indicated proteins and genotypes. (g, h, o, p, w, x) TUNEL staining for the genotypes indicated. Arrow in x indicates apoptotic cell. (y) Quantitation of HDAC1/2-positive plantar epidermal basal cells in the genotypes indicated. n ≥ 4 mice for each genotype were analyzed. (z) Left graph: quantitation of Ki-67‒positive basal cells in control and in Hdac1/2-mutant plantar epidermis; n = 5 mice for each genotype were analyzed. Right graph: quantitation of HDAC1/2-positive and HDAC1/2-negative (deleted) basal cells staining positive for Ki-67 in Hdac1/2cKO p16KO plantar epidermis; n = 3 mice analyzed. In y and z, each data point represents the average percentage from ≥5 ×20 fields from a single mouse. P-values were calculated with two-tailed Student’s t-test; error bars indicate SEM. Bar = 25 μm for a, b, i, j, q, and r and 50 μm for a’‒h, i’‒p, and q’‒x. +ve indicates positive responses. K, keratin; P, postnatal day.

       HDAC1/2 maintain epidermal progenitor cell fitness by suppressing p16

      To determine whether HDAC1/2-deleted cells were lost through apoptosis, decreased proliferation, accelerated differentiation, and/or senescence, we examined whether their ability to compete with undeleted cells was altered by codeletion of p53 or p16. p53 codeletion failed to rescue long-term survival of Hdac1/2-deleted cells in Hdac1/2cKO p53KO triple mutant plantar epidermis (Figure 3a–p and y); however, p16 codeletion resulted in statistically significantly increased retention of HDAC1/2-deleted basal cells (Figure 3a’, b’, q’, r’, and y). Epidermal thickness in p16KO mice was similar to that in controls (Figure 3a and q), but in Hdac1/2cKO p16KO mice, it was reduced compared with the thickness of Hdac1/2cKO epidermis on P42 (Figure 3b and r). Triple mutant epidermis had an abnormal, disorganized structure characterized by appearance of enlarged cells, loss of cell‒cell adhesion in more superficial epidermal layers, and premature cornification (Figure 3r, arrow). Cells double positive for K14/K10 (Supplementary Figure S3a, b, i, and j, arrow) and K14/loricrin (Supplementary Figure S3c, d, k, and l, arrows) were more prominent in Hdac1/2cKO p16KO than in Hdac1/2cKO epidermis, and acetylated c-MYC was elevated (Supplementary Figure S3e, f, m, and n). Furthermore, Hdac1/2cKO p16KO triple-mutant epidermis displayed higher numbers of nuclei showing elevated H3K9Ac than were observed in Hdac1/2cKO skin at the same stage (Figure 3e, f, u, and v).
      Interestingly, the persistence of HDAC1/2-deleted cells in the triple-mutant epidermis was not due to the rescue of their apoptosis by p16 deletion because Hdac1/2cKO p16KO triple-mutant plantar epidermis on P42 displayed increased apoptosis (Figure 3g, h, w, and x, arrow) and elevated acetylated p53 (Supplementary Figure S3g, h, o, and p) compared with the control, Hdac1/2cKO or p16KO epidermis at the same stage. Furthermore, analysis of the proliferative status of HDAC1/2-negative cells compared with that of undeleted cells in the same triple-mutant Hdac1/2cKO p16KO samples showed that cells lacking HDAC1/2 were less proliferative than the undeleted cells (Figure 3z). These results suggest that p16 deletion enables the survival of Hdac1/2-deleted keratinocytes by preventing their senescence rather than by rescuing decreased proliferation, accelerated differentiation, or apoptosis.
      Taken together, our data indicate that HDAC1/2 play at least three major and distinct roles in postnatal epidermis: (i) deacetylation of p53 to permit progenitor cell proliferation, (ii) deacetylation of c-MYC to prevent premature and disordered differentiation, and (iii) repression of p16 to suppress senescence and allow for long-term maintenance of progenitor cells.

       Forced GLI2ΔN expression in epidermis initiates BCC-like tumorigenesis

      To investigate whether homozygous codeletion of Hdac1/2 could combat initial stages of epidermal tumorigenesis, we utilized doxycycline-inducible epithelial-specific expression of a mutant, constitutively active form of GLI2, GLI2ΔN in K5-rtTA tetO-GLI2ΔN mice. GLI2ΔN stimulates HH signaling and can promote the formation of BCC-like growths (
      • Grachtchouk M.
      • Pero J.
      • Yang S.H.
      • Ermilov A.N.
      • Michael L.E.
      • Wang A.
      • et al.
      Basal cell carcinomas in mice arise from hair follicle stem cells and multiple epithelial progenitor populations.
      ). Adult K5-rtTA tetO-GLI2ΔN mice placed on high dose (200 μg/ml) doxycycline drinking water displayed invagination of the plantar epidermis (Supplementary Figure S5a and b), efficient expression of MYC-tagged GLI2ΔN (Supplementary Figure S5c and d), increased proliferation (Supplementary Figure S5c–e), and expression of the BCC markers K17 and SOX9 (
      • Anderson-Dockter H.
      • Clark T.
      • Iwamoto S.
      • Lu M.
      • Fiore D.
      • Falanga J.K.
      • et al.
      Diagnostic utility of cytokeratin 17 immunostaining in morpheaform basal cell carcinoma and for facilitating the detection of tumor cells at the surgical margins.
      ;
      • Vidal V.P.
      • Ortonne N.
      • Schedl A.
      SOX9 expression is a general marker of basal cell carcinoma and adnexal-related neoplasms.
      ) (Supplementary Figure S5f–i) within 7 days of initiating induction, providing a rapid and consistent system for analyzing molecular mechanisms underlying proliferative responses to GLI2ΔN. Forced expression of GLI2ΔN did not affect HDAC1/2 expression (Supplementary Figure S5j and k), although interestingly, acetylated p53 was slightly increased (Supplementary Figure S5l and m), and p16 mRNA levels were elevated (Supplementary Figure S5n and o) in GLI2ΔN-expressing epidermis.
      GLI2ΔN-expressing mice died after approximately 10 days of induction with high dose doxycycline water; however, mice treated with a lower dose of doxycycline water (10 μg/ml) survived for at least 1 month and developed tumors in plantar skin resembling human BCC (Supplementary Figure S5p and q), including the appearance of highly proliferative cells at the leading edges of tumors (Supplementary Figure S5r and s) and expression of K17 and SOX9 (Supplementary Figure S5t–w). Thus, forced expression of Gli2ΔN promotes BCC-like tumorigenesis in the plantar skin of K5-rtTA tetO-GLI2ΔN mice.

       Hdac1/2 deletion in pre-BCC–like lesions significantly reduces proliferation and induces apoptosis

      Immunofluorescence for HDAC1 or HDAC2 revealed the expression of both these proteins in pre-BCC lesions of K5-rtTA tetO-GLI2ΔN mice induced with high dose doxycycline water for 7 days (Figure 4a and b). To test whether the deletion of HDAC1 and HDAC2 affected the initiation of pre-BCC, we generated K5-rtTA tetO-Gli2ΔN tetO-Cre Hdac1fl/+ Hdac2fl/fl (Gli2ΔNK5-rtTA Hdac1cHet/Hdac2cKO), K5-rtTA tetO-Gli2ΔN tetO-Cre Hdac1fl/fl Hdac2fl/+ (Gli2ΔNK5-rtTA Hdac1cKO/Hdac2cHet), and K5-rtTA tetO-Gli2ΔN tetO-Cre Hdac1fl/fl Hdac2fl/fl (Gli2ΔNK5-rtTA Hdac1/2cKO) mice as well as controls lacking tetO-Cre and placed them on high dose doxycycline water. GLI2ΔN expression was apparent by 4 days after initiating doxycycline treatment (Figure 4c and d). HDAC1/2 were mosaically deleted in basal cells by 4 days (Figure 4c and d); by 7 days after initiating treatment, HDAC1/2 were efficiently deleted (Figure 4e and f), p53 was hyperacetylated (Figure 4g and h), and p16 mRNA levels were significantly increased (Figure 4i). Subsequent analyses were therefore conducted on tissue from mice at the 7-day time point.
      Figure thumbnail gr4
      Figure 4Hdac1/2 deletion increases p53 acetylation and elevates p16 mRNA levels in GLI2ΔN-expressing plantar epidermis. Schematics show the timing of high-dose doxycycline treatment; samples were analyzed on (a, b, e-i) P55 or (c, d) P52. (a, b) IF for Myc-tagged GLI2ΔN (a) and HDAC1 or (b) HDAC2 in induced K5-rtTA tetO-GLI2ΔN plantar epidermis. (c–f) IF for GLI2ΔN and HDAC1/2 in the plantar epidermis of the indicated genotypes induced for (c, d) 4 or (e, f) 7 days. (g, h) IF for GLI2ΔN and acetylated p53 in the plantar epidermis of the indicated genotypes induced for 7 days. (i) qPCR for p16 mRNA in induced K5-rtTA tetO-GLI2ΔN Hdac1/2cKO plantar epidermis compared with that in K5-rtTA tetO-GLI2ΔN Hdac1/2control. P-values were calculated with two-tailed Student’s t-test; error bars indicate SEM. n ≥ 5 pairs of mutant and control mice in a–h; n = 4 pairs of mutant and control mice in i. Bar = 25 μm in a–f and 50 μm in g and h. IF, immunofluorescence; P, postnatal day; rtTA, reverse tetracycline transactivator.
      Homozygous loss of Hdac1 combined with heterozygous loss of Hdac2 in Gli2ΔN-expressing plantar epidermis (Supplementary Figure S6e and f) produced a slight decrease in epidermal thickening and invagination (Supplementary Figure S6a and b) and a mild but significant decrease in proliferation (Supplementary Figure S6i, j, and q) but no increase in apoptosis (Supplementary Figure S6m and n), whereas heterozygous loss of Hdac1 combined with homozygous loss of Hdac2 (Supplementary Figure S6g and h) had no obvious effect on epidermal thickening and invagination (Supplementary Figure S6c and d), proliferation (Supplementary Figure S6k, l, and r), or apoptosis (Supplementary Figure S6o and p), indicating that the functions of HDAC2 in this context are partially redundant with those of HDAC1.
      Double homozygous deletion of Hdac1/2 in GLI2ΔN-expressing epidermis (Figure 5g and h) significantly reduced thickening and invagination (Figure 5a and b) and proliferation (Figure 5m, n, y), increased basal cell apoptosis (Figure 5s, t, z), and caused hyperacetylation of c-Myc (Supplementary Figure S7i and j). Expression of K10 (Supplementary Figure S7a and b) and loricrin (Supplementary Figure S7e and f) was slightly expanded in some regions compared with their expression in the controls lacking tetO-Cre.
      Figure thumbnail gr5
      Figure 5Hdac1/2 deletion reduces proliferation and increases apoptosis of GLI2ΔN-expressing plantar epidermis through p53- and p16-dependent mechanisms. Schematic shows the timing of doxycycline treatment; samples were analyzed on P55. (a–f) Histology of plantar epidermis for the indicated genotypes. Arrows in b, d, and f indicate prematurely cornified cells. (g‒r) IF for the indicated proteins and genotypes. (s–x) TUNEL staining for the indicated genotypes. Arrows in t indicate apoptotic cells. White dashed lines indicate epidermal–dermal boundaries. Quantitation of (y) Ki-67‒positive or (z) TUNEL-positive GLI2ΔN-expressing plantar epidermal basal cells for the indicated genotypes; n ≥ 3 mice of each genotype were analyzed. P-values were calculated with two-tailed Student’s t-test; error bars indicate SEM. +ve indicates positive responses. Bar = 25 μm for a–f and 50 μm for g–x. IF, immunofluorescence; P, postnatal day.

       Deletion of p53 or p16 partially rescues basal proliferation and apoptosis in Hdac1/2-deficient GLI2ΔN-expressing plantar epidermis

      To determine whether the effects of Hdac1/2 deletion in pre-BCC were mediated by p53 or p16, we generated Gli2ΔNK5-rtTA Hdac1/2control mice lacking tetO-Cre and Gli2ΔNK5-rtTA Hdac1/2cKO mice that either carried null mutations in p53 or p16 or were wild type for p53 and p16. In the absence of Hdac1/2 deletion, loss of p53 or p16 did not affect GLI2ΔN-mediated epidermal thickening and invagination (Figure 5a, c, and e), HDAC1/2 expression (Figure 5g, i, and k), proliferation (Figure 5m, o, q, and y), or apoptosis (Figure 5s, u, w, and z). However, concomitant loss of either p53 or p16 partially rescued the decreased thickness (Figure 5b, d, and f), the decreased proliferation (Figure 5n, p, r, and y), and the increased apoptosis (Figure 5 t, v, x, and z) caused by Hdac1/2 deletion in GLI2ΔN-expressing epidermis. Premature cornification was observed in GLI2ΔN-expressing plantar epidermis deleted for Hdac1/2, with or without concomitant loss of p53 or p16 (arrows in Figure 5b, d, and f). Interestingly, codeletion of p16 further exacerbated expanded expression of K10 (Supplementary Figure S7a–d, arrow) and loricrin (Supplementary Figure S7e–h, arrows), enlargement of suprabasal nuclei (Figure 5b and f), and elevated acetylated c-MYC levels (Supplementary Figure S7i–l) caused by Hdac1/2 deletion in GLI2ΔN-expressing epidermis. Thus, both p53 and p16 are required to maintain proliferation and prevent apoptosis in pre-BCC.

       Romidepsin reduces proliferation and induces apoptosis in normal and GLI2ΔN-expressing plantar epidermis

      Our data from genetic deletion experiments suggested that small-molecule HDAC1/2 inhibitors might be effective in treating BCC and other hyperproliferative skin conditions. Because mouse plantar epidermis is too thick to permit easy access of topically applied chemical inhibitors, we tested this by topically applying romidepsin or vehicle only to dorsal paw skin of P50 control or GLI2ΔN-expressing mice that had been induced with high-dose doxycycline water from P46. After 4 consecutive days of inhibitor application to dorsal paw skin, romidepsin caused histological changes (Figure 6a, b, a’, and b’), including epidermal damage, hair follicle defects, and the presence of an inflammatory infiltrate in the dermis (Supplementary Figure S8a and b). Romidepsin treatment resulted in increased H3K9 acetylation (Figure 6c, d, c’, and d’) and significantly reduced proliferation (Figure 6e, f, q, e’, f’, and q’) and basal cell apoptosis (Figure 6g, h, g’, and h’, arrows) of both normal and GLI2ΔN-expressing epidermis compared with that in vehicle-treated skin. Notably, the epidermal thickening and invaginations observed in GLI2ΔN-expressing dorsal paw epidermis were diminished after romidepsin treatment (Figure 6a’ and b’). In parallel, topical romidepsin altered the patterns of K10 and loricrin expression (Supplementary Figure S8c–f and k–n) and caused hyperacetylation of both c-MYC (Supplementary Figure S8g, h, o, and p) and p53 (Supplementary Figure S8i, j, q, and r). These phenotypes were in line with the effects of genetic deletion of Hdac1/2 in normal and GLI2ΔN-expressing interfollicular epidermis. By contrast with the short-term effects of genetic loss of Hdac1/2 and despite reduced proliferation, epidermal thickness was increased rather than decreased by topical romidepsin (Figure 6a and b). This was likely due to a more pronounced effect on c-MYC hyperacetylation in suprabasal cells caused by topical application of the inhibitor, which can be expected to penetrate suprabasal layers before affecting Hdac1/2 basal cells, compared with K5 promoter‒driven deletion, which first removes Hdac1/2 in the basal layer. In line with this, forced expression of c-Myc in suprabasal epidermis causes hyperplasia (
      • Waikel R.L.
      • Wang X.J.
      • Roop D.R.
      Targeted expression of c-Myc in the epidermis alters normal proliferation, differentiation and UV-B induced apoptosis.
      ). Together with our observation of differentiation defects in Hdac1/2-deleted suprabasal cells (Figure 2), these data indicate that HDAC1/2 play key roles in suprabasal as well as in basal cells. These results suggest topical romidepsin as a promising therapeutic for BCC and other hyperproliferative skin conditions in human patients.
      Figure thumbnail gr6
      Figure 6p16 deletion partially rescues reduced IFE and pre-BCC proliferation and prevents the increased apoptosis caused by topical romidepsin. Schematics show the timelines for topical application of romidepsin or vehicle to dorsal paw skin starting on P53 (both diagrams) and high-dose oral doxycycline treatment starting on P49 (right diagram). All samples were analyzed on P57. (a, b, i, j, a’, b’, i’, j’) Histology of dorsal paw skin of the indicated genotypes after treatment with vehicle (DMSO) or romidepsin as indicated. (c–f, k–n, c’–f’, k’–n’) IF for the indicated proteins in dorsal paw skin of the indicated genotypes after vehicle or romidepsin treatment. (g, h, o, p, g’, h’, o’, p’) TUNEL assay for the genotypes indicated after vehicle or romidepsin treatment. Arrows in h and h’ indicate apoptotic cells. (q) Quantitation of Ki-67‒positive IFE basal cells in control or in p16-null dorsal paw IFE treated with vehicle or romidepsin. (q’) Quantitation of Ki-67‒positive basal cells in GLI2ΔN-expressing dorsal paw IFE with or without p16 deletion after treatment with vehicle or romidepsin; n = 4 mice of each genotype analyzed. P-values were calculated with two-tailed Student’s t-test; error bars indicate SEM. Bar = 25 μm for a, b, i, j, a’, b’, i', and j’ and 50 μm for c–h, k–p, c’–h’, and k’–p’. +ve indicates positive responses. BCC, basal cell carcinoma; IF, immunofluorescence; IFE, interfollicular epidermis; P, postnatal day.

       The effects of topical romidepsin depend in part on intact p16 function

      To test whether the loss of p16 could impair the effectiveness of chemical HDAC inhibition in BCC, we compared the effects of topical romidepsin on normal dorsal paw skin, and on GLI2ΔN-induced pre-BCC lesions, in mice with or without homozygous p16 deletion. Loss of p16 had no effect on morphology (Figure 6a, i, a’, and i'), H3K9Ac levels (Figure 6c, k, c’, and k’), proliferation (Figure 6e, m, q, e’, m’, and q’), or apoptosis (Figure 6g, o, g’, and o’) of vehicle-treated normal or GLI2ΔN-expressing epidermis. By contrast, whereas p16 deletion in romidepsin-treated epidermis did not alter H3K9Ac levels caused by topical romidepsin (Figure 6d, l, d’, and l’), it restored proliferation from approximately 25% to approximately 60% of the levels seen in vehicle-treated normal epidermis, which was highly statistically significant (Figure 6e, f, m, n, and q), and also decreased basal cell apoptosis resulting from topical romidepsin treatment (Figure 6g, h, o, and p). Similarly, loss of p16 in romidepsin-treated pre-BCC lesions restored epithelial invaginations (Figure 6b’ and j’), increased proliferation from <10% to approximately 25% of the levels seen in vehicle-treated lesions (Figure 6e’, f’, m’, n’, and q’), and prevented basal cell apoptosis (Figure 6g’, h’, o’, and p’). Thus, the efficacy of romidepsin in blocking proliferation and inducing apoptosis in both normal skin and pre-BCC‒like lesions was significantly diminished in the absence of p16.

      Discussion

      In this study, we show that HDAC1 and HDAC2 function in postnatal epidermis to maintain proliferation in part by deacetylating p53 and permit long-term cell survival by repressing p16. Interestingly, whereas we found that inducible homozygous deletion of Hdac1 together with heterozygous loss of Hdac2 did not alter proliferation in postnatal plantar epidermis in the short term, previous studies have described hair follicle defects and epidermal hyperproliferation in hairy skin of mice with constitutive homozygous loss of Hdac1, which were exacerbated by heterozygous Hdac2 deletion (
      • Hughes M.W.
      • Jiang T.X.
      • Lin S.J.
      • Leung Y.
      • Kobielak K.
      • Widelitz R.B.
      • et al.
      Disrupted ectodermal organ morphogenesis in mice with a conditional histone deacetylase 1, 2 deletion in the epidermis.
      ;
      • Winter M.
      • Moser M.A.
      • Meunier D.
      • Fischer C.
      • Machat G.
      • Mattes K.
      • et al.
      Divergent roles of HDAC1 and HDAC2 in the regulation of epidermal development and tumorigenesis.
      ). Thus, the influence of degenerating hair follicles and the effects of longer-term compound homozygous/heterozygous deletion of Hdac1 and Hdac2, reveal nonredundant functions of these regulators in the adult epidermis.
      Similar to its effects in normal plantar epidermis, homozygous Hdac1/2 deletion caused decreased proliferation, increased apoptosis, and premature differentiation in GLI2ΔN-expressing pre-BCC lesions. Codeletion of p53 partially rescued both proliferation and apoptosis, providing an explanation for the relatively mild effects of the HDAC inhibitor vorinostat observed in mice xenografted with Ptch1–/– P53–/– BCC cells (
      • Mirza A.N.
      • Fry M.A.
      • Urman N.M.
      • Atwood S.X.
      • Roffey J.
      • Ott G.R.
      • et al.
      Combined inhibition of atypical PKC and histone deacetylase 1 is cooperative in basal cell carcinoma treatment.
      ). Although not as common as P53 mutations, loss of heterozygosity for P16 occurs in approximately 4% of human BCC (
      • Soufir N.
      • Molès J.P.
      • Vilmer C.
      • Moch C.
      • Verola O.
      • Rivet J.
      • et al.
      P16 UV mutations in human skin epithelial tumors.
      ). Interestingly, we observed a partial rescue of proliferation and decreased apoptosis when p16 was codeleted with Hdac1/2. These data indicate that in hyperproliferative lesions, HDAC1/2-mediated suppression of the functions of both p16 and p53 is important to maintain proliferation and prevent apoptosis.
      To test directly whether HDAC inhibition could reduce proliferation and increase apoptosis of GLI2ΔN-expressing pre-BCC lesions, we used topical application of the class I HDAC inhibitor romidepsin. Topical romidepsin was effective at increasing H3K9 acetylation, increasing apoptosis, and inducing premature differentiation and decreased proliferation to <10% of the levels observed in vehicle-treated lesions. Homozygous deletion of p16 restored the proliferation of romidepsin-treated lesions to approximately 25% of the levels seen in vehicle-treated skin.
      Taken together, our data suggest that topical HDAC1/2 inhibition is likely to prove effective in reducing the proliferation of pre-BCC lesions that lack P53 or P16 mutations. Collectively, approximately 46% of human BCC lack such mutations, suggesting the potential susceptibility of a large fraction of BCC to romidepsin treatment and highlighting the importance of genotyping BCC for P53 and P16 before initiating therapy.
      The experiments described in this study focused on the effects of HDAC1/2 deletion or inhibition on early stages of BCC initiation rather than their effects in established tumors. Importantly, however, we observed similar effects of homozygous Hdac1/2 deletion in normal hairless skin, where HH signaling is generally not active, and in pre-BCC lesions caused by GLI2ΔN-induced constitutive HH signaling. These results show that removal of HDAC1/2 function has significant consequences unrelated to the HH pathway. This conclusion is supported by the partial rescue of Hdac1/2-mutant phenotypes on codeletion of p53 or p16, which are not canonical HH targets. Thus, our data reveal general effects of HDAC inhibition on epidermal proliferation, suggesting that this approach could be useful in treating established BCC lesions and hyperproliferative skin conditions unrelated to BCC. The effects of topical inhibition on normal skin also highlight the potential for side effects in normal tissues; indeed, systemic romidepsin is associated with thrombocytopenia (
      • Bishton M.J.
      • Harrison S.J.
      • Martin B.P.
      • McLaughlin N.
      • James C.
      • Josefsson E.C.
      • et al.
      Deciphering the molecular and biologic processes that mediate histone deacetylase inhibitor-induced thrombocytopenia [published correction appears in Blood 2015;125:3824–3825].
      ) and electrocardiographic changes (
      • Sager P.T.
      • Balser B.
      • Wolfson J.
      • Nichols J.
      • Pilot R.
      • Jones S.
      • et al.
      Electrocardiographic effects of class 1 selective histone deacetylase inhibitor Romidepsin.
      ). In light of this, topical application limited to lesional skin may be an attractive therapeutic option.

      Materials and Methods

       Mice

      The mouse lines used and methods for genotyping, induction, and topical romidepsin application are detailed in Supplementary Materials and Methods. All mouse experiments were performed with approved animal protocols according to the institutional guidelines established by the Institutional Animal Care and Use Committees committees of the University of Pennsylvania (Philadelphia, PA) and the Icahn School of Medicine at Mount Sinai (New York, NY).

       Histology, immunostaining, and TUNEL assays

      Tissues dissected from humanely killed mice were fixed in 4% paraformaldehyde/PBS (Affymetrix/USB, Santa Clara, CA) overnight at 4 °C and then paraffin embedded and sectioned at 5 μm. Detailed experimental procedures, antibodies, and statistical analysis methods are provided in Supplementary Materials and Methods.

       Real-time PCR

      Reverse transcription was performed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA), and cDNA was subjected to real-time PCR using the StepOnePlus system and SYBR Green Kit (Applied Biosystems). Detailed methods and primers are provided in Supplementary Materials and Methods.

       Data availability statement

      The authors declare that the main data supporting the findings of this study are available within the article and its supplementary information files. All correspondence and material requests should be addressed to SEM. This study includes no data deposited in external repositories.

      ORCIDs

      Conflict of Interest

      The authors state no conflict of interest.

      Acknowledgments

      We thank Adam Glick for K5-rtTA mice; Eric N. Olson for Hdac1fl/fl and Hdac2fl/fl mice; Steven Prouty for histology; and Mingang Xu, Katherine M. Szigety, and Deborah J. Moran for helpful discussions. This work was supported by grant RO1AR063146 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases/National Institutes of Health, USA to SEM and by grant P30AR069589 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases/National Institutes of Health, USA to support the Penn Skin Biology and Diseases Resource-based Center (Philadelphia, PA).

      Author Contributions

      Conceptualization: XZ, ML, MG, EEM, AAD, SEM; Formal Analysis: JTS; Funding Acquisition: SEM; Investigation: XZ, ML, FL; Methodology: XZ, ML; Project Administration, SEM; Resources: MG, EEM, AAD; Supervision: SEM; Validation: XZ, ML; Visualization: XZ; Writing - Original Draft Preparation: XZ; Writing - Review and Editing: XZ, MG, ML, SEM

      Supplementary Materials and Methods

       Mice

      All mice were maintained on a mixed FVB/N/C57BL/6J/SJL/J strain background. The following mouse lines were used: Hdac1fl (
      • Montgomery R.L.
      • Davis C.A.
      • Potthoff M.J.
      • Haberland M.
      • Fielitz J.
      • Qi X.
      • et al.
      Histone deacetylases 1 and 2 redundantly regulate cardiac morphogenesis, growth, and contractility.
      ), Hdac2fl (
      • Montgomery R.L.
      • Davis C.A.
      • Potthoff M.J.
      • Haberland M.
      • Fielitz J.
      • Qi X.
      • et al.
      Histone deacetylases 1 and 2 redundantly regulate cardiac morphogenesis, growth, and contractility.
      ), keratin (K) 5-reverse tetracycline transactivator (
      • Diamond I.
      • Owolabi T.
      • Marco M.
      • Lam C.
      • Glick A.
      Conditional gene expression in the epidermis of transgenic mice using the tetracycline-regulated transactivators tTA and rTA linked to the keratin 5 promoter.
      ), tetO-Cre (The Jackson Laboratories, Bar Harbor, ME, strain #006234) (
      • Perl A.K.
      • Wert S.E.
      • Nagy A.
      • Lobe C.G.
      • Whitsett J.A.
      Early restriction of peripheral and proximal cell lineages during formation of the lung.
      ), tetO-Gli2ΔN (
      • Grachtchouk M.
      • Pero J.
      • Yang S.H.
      • Ermilov A.N.
      • Michael L.E.
      • Wang A.
      • et al.
      Basal cell carcinomas in mice arise from hair follicle stem cells and multiple epithelial progenitor populations.
      ), p53KO (The Jackson Laboratories, strain #002101) (
      • Jacks T.
      • Remington L.
      • Williams B.O.
      • Schmitt E.M.
      • Halachmi S.
      • Bronson R.T.
      • et al.
      Tumor spectrum analysis in p53-mutant mice.
      ), and p16KO (NCI Mouse Repository line 01XE4) (
      • Sharpless N.E.
      • Bardeesy N.
      • Lee K.H.
      • Carrasco D.
      • Castrillon D.H.
      • Aguirre A.J.
      • et al.
      Loss of p16Ink4a with retention of p19Arf predisposes mice to tumorigenesis.
      ). To induce Cre-mediated recombination, mice were placed on doxycycline chow (6 g/kg, Bio-Serv, Flemington, NJ) or doxycycline (Sigma-Aldrich, St. Louis, MO) containing water at high (200 μg/ml in 5% sucrose) or low (10 μg/ml in 5% sucrose) doses. Mice were assigned to control or experimental groups on the basis of genotype; within each group, mice were assigned randomly to experimental treatments. For most experiments, we used n = 5 mice per group, which provides 80% power at a two-sided significance level of 0.05 to detect an effect size of 2.0s where s is the SD. Mice of both sexes were analyzed, and no sex-specific differences were observed in the data. All mouse experiments were performed with approved animal protocols according to institutional guidelines established by the Institutional Animal Care and Use Committee committees of the University of Pennsylvania (Philadelphia, PA) and the Icahn School of Medicine at Mount Sinai (New York, NY).

       Histology, immunostaining, TUNEL assays, and RNAscope

      Tissues dissected from killed mice were fixed in 4% paraformaldehyde/PBS (Affymetrix/USB, Santa Clara, CA) overnight at 4 °C and then paraffin embedded and sectioned at 5 μm. Paraffin sections were dewaxed and rehydrated through xylene substitute (Sigma-Aldrich) and graded ethanol solutions. Sections were microwave treated with antigen unmasking solution (Vector Labs, Burlingame, CA). For immunostaining, sections were incubated with primary antibodies overnight, followed by incubation with Alexa Fluor‒labeled secondary antibodies (Thermo Fisher Scientific, Waltham, MA and Vector Labs) and application of DAPI containing mounting medium (Vector Labs). For TUNEL assays, sections were incubated with a TUNEL labeling mixture (Roche, Basel, Switzerland) at 37 °C for 1 hour and then washed with PBS with Tween 20 and mounted with DAPI containing mounting medium. RNAscope (Advanced Cell Diagnostics, Newark, CA) was carried out with probes for p21 (Advanced Cell Diagnostics, #408551), Mdm2 (Advanced Cell Diagnostics, #447641), and p16 (Advanced Cell Diagnostics, #411011), following the manufacturer’s instructions.

       Antibodies

      The following primary antibodies were used: rabbit anti-HDAC1 (raised against the C-terminal region of human HDAC1, using a keyhole limpet haemocyanin‒conjugated synthetic peptide) (Thermo Fisher Scientific, 49-1025, 1:1,000), rabbit anti-HDAC2 (raised against a synthetic 11 amino acid peptide derived from the C-terminus of the mouse HDAC2 protein) (Thermo Fisher Scientific, 51-5100, 1:1,000), mouse anti-K14 (raised against a synthetic peptide corresponding to human cytokeratin 14 [C terminal]) (Abcam, Cambridge, United Kingdom, ab7800-500, 1:200), rabbit anti-acetyl-H3K9 (raised against a synthetic peptide corresponding to human Histone H3 amino acids 1-100 [N terminal] [acetyl K9] conjugated to keyhole limpet haemocyanin) (Abcam, ab10812, 1:500), rat anti‒Ki-67 (eBioscience, 14-5698-82, 1:200), rabbit anti-K5 (raised against a peptide sequence derived from the C-terminus of the mouse K5 protein) (Lapcorp Drug Development, Princeton, NJ, PRB-160P, 1:1,000), rabbit anti‒acetyl-p53 (raised against a synthetic acetylated peptide derived from human p53 around the acetylation site of lysine 381 [Abcam, ab61241, 1:200]), mouse anti-p16 (raised against amino acids 1-168 representing full length p16 of mouse origin) (Santa Cruz Biotechnology, Dallas, TX, sc-1661, 1:200), rabbit anti-K10 (raised against a peptide sequence derived from the C-terminus of the mouse K10 protein) (Lapcorp Drug Development, PRB-159 P, 1:1,000), rabbit anti-loricrin (raised against a peptide sequence derived from the C-terminus of the mouse loricrin protein) (Lapcorp Drug Development, PRB-145 P, 1:1,000), rabbit anti‒acetyl-c-MYC (raised against a keyhole limpet haemocyanin‒conjugated linear peptide corresponding to human c-Myc acetylated at Lys157) (MilliporeSigma, Burlington, MA, ABE27, 1:200), rabbit anti-K17 (raised against a synthetic peptide within human Cytokeratin 17 amino acids 1-100 [N terminal]) (Abcam, ab109725, 1:500), rabbit anti-SOX9 (raised against a keyhole limpet haemocyanin‒conjugated linear peptide corresponding to the C-terminal sequence of human SOX9) (MilliporeSigma, AB5535, 1:200), mouse anti‒MYC-tag (raised against a synthetic peptide corresponding to residues 410-419 of human c-Myc [EQKLISEEDL]) (Cell Signaling Technology, Danvers, MA, #2276, 1:1,000); rabbit anti‒Ki-67 (raised against a synthetic peptide within human Ki-67 amino acids 1,200-1,300) (Abcam, ab16667, 1:200), rabbit anti-CD26 (raised against a synthetic peptide corresponding to Human DPP4; on the basis of the spacer region of Human DPP4, mid-molecule, before the cysteine-rich region) (Abcam, ab28340, 1:400), and rabbit anti-CD3 (raised against a synthetic peptide KAKAKPVTRGAGA corresponding to amino acids 156-168 of Human CD3 Epsilon chain) (Abcam, ab5690, 1:200).

       Real-time PCR

      Full-thickness skin samples were incubated in Dispase II (Roche) solution for 1 hour at 37 °C, and the epidermis was separated from the dermis with forceps. Total RNA was extracted from the epidermis using TRIzol (Thermo Fisher Scientific) and purified using the RNeasy kit (Qiagen, Hilden, Germany). Reverse transcription was performed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA), and cDNA was subjected to real-time PCR using the StepOnePlus system and SYBR Green Kit (Applied Biosystems). Gapdh was used as an internal control, and expression differences were determined using the 2−ΔΔCT method. Primers for p16 were 5'- GAGCATGGGTCGCAGGTTCT-3' for p16 forward and 5'-GGTAGTGGGGTCCTCGCAGT-3' for p16 reverse.

       Topical application of romidepsin

      Romidepsin (Selleckchem, Pittsburgh, PA) was diluted with DMSO to 18.5 mM. Before applying to mouse skin, romidepsin solution was diluted with corn oil to a final concentration of 0.37 mM. After vigorous vortexing, 20 μl romidepsin in DMSO/corn oil or DMSO/corn oil vehicle alone was immediately applied to dorsal paw skin twice per day for 4 consecutive days.

       Statistical information

      For quantification of Ki-67‒ or TUNEL-positive basal keratinocytes, basal cells were counted from five or more ×20 fields per mouse. The numbers of mice used in each experiment are detailed in the figure legends. For quantification of mast cells, toluidine blue‒positive dermal cells from five ×20 fields were counted. For real-time qPCR, at least three pairs of control and mutant samples were used, and three technical replicates were analyzed for each sample. Significance was calculated with a two-tailed Student’s t-test. P < 0.05 was considered to be significant.
      Figure thumbnail fx1
      Supplementary Figure S1A single copy of Hdac1 or Hdac2 is sufficient to maintain the short-term homeostasis of the plantar epidermis. The schematic shows the timing of oral doxycycline treatment for panels a–r; all samples were analyzed on P27. (a–d) Histology reveals that plantar epidermis in (b) Hdac1cKO Hdac2cHet and (d) Hdac1cHet Hdac2cKO mice is similar to those of the (a, c) controls. (e–h) IF staining for (e, f) HDAC1 (red) or (g, h) HDAC2 (red) shows that Hdac1 and Hdac2 are efficiently deleted in (f) Hdac1cKO Hdac2cHet and (h) Hdac1cHet Hdac2cKO plantar epidermis relative to their expression in the respective littermate controls. Basal cells are marked by IF for K14 (green). (i–l) IF for Ki-67 (green) shows that levels of epidermal proliferation in (j) Hdac1cKO Hdac2cHe and (l) Hdac1cHet Hdac2cKO plantar epidermis are similar to those in the respective littermate (i, k) controls. Basal cells are marked by IF for K5 (red). (m–p) TUNEL assay (green) shows that basal cell apoptosis is absent in (n) Hdac1cKO Hdac2cHet and (p) Hdac1cHet Hdac2cKO plantar epidermis and in their respective littermate (m, o) controls. Epidermal–dermal borders are marked by dashed white lines. (q, r) Quantitation of the percentage of Ki-67‒positive basal cells shows no statistical difference between (q) Hdac1cKO Hdac2cHet and (r) Hdac1cHet Hdac2cKO plantar epidermis and their respective littermate controls. Four mice of each genotype were analyzed; a two-tailed Student’s t-test was used to calculate P-value. P < 0.05 was considered to be significant; error bars indicate SEM. Bar = 25 μm for a‒d and 50 μm for e‒p. IF, immunofluorescence; K, keratin; P, postnatal day.
      Figure thumbnail fx2
      Supplementary Figure S2Loss of p16 does not rescue proliferation or prevent apoptosis of basal cells lacking Hdac1/2. (a, c–w) The schematic shows the timing of oral doxycycline treatment; samples were analyzed on P27. (b) Mice were placed on oral doxycycline on P19, and the skin was analyzed on P28. (a) qPCR shows that p16 mRNA expression is elevated in K5-rtTA tetO-Cre Hdac1fl/fl Hdac2fl/fl plantar epidermis relative to that in the littermate controls. Data were normalized to Gapdh. Three pairs of mice were analyzed; a two-tailed Student’s t-test was used to calculate the P-value. P < 0.05 was considered to be significant; error bars indicate SEM. (b) Dorsal skin of K5-rtTA tetO-Cre Hdac1fl/fl Hdac2fl/fl mice treated with doxycycline from P19 and analyzed on P28 shows mosaic deletion of HDAC1/2 (green) and elevated levels of p16 protein (red) in HDAC1/2-deficient cells in hair follicles (white arrow) and IFE (yellow arrow). (c–f) Histology reveals the thinning of the plantar epidermis in (d) Hdac1/2cKO and (f) Hdac1/2cKO p16KO mice relative to that in the (c) control and (e) p16KO samples. (g–j) IF for HDAC1/2 (red) reveals efficient deletion in K14-positive basal cells (green) in (h) Hdac1/2cKO and (j) Hdac1/2cKO p16KO plantar epidermis compared with that in the (g) control and (i) p16KO samples. (k–n) IF for H3K9Ac (red) reveals increased levels in (l) Hdac1/2cKO and (n) Hdac1/2cKO p16KO plantar epidermis compared with that in the (k) control and (m) p16KO samples. IF for K14 (green) marks the basal cells. (o–r) Ki-67 staining (green) reveals reduced proliferation of K5-positive basal cells (red) in (p) Hdac1/2cKO and (r) Hdac1/2cKO p16KO plantar epidermis compared with that in the (o) control and (q) p16KO samples. (s–v) Basal cell apoptosis (green) is observed in (t) Hdac1/2cKO and (v) Hdac1/2cKO p16KO plantar epidermis but not in (s) control or (u) p16KO epidermis. (w) Quantitation of the percentage of Ki-67‒positive basal cells shows that proliferation of plantar epidermis is unaltered by loss of p16 alone compared with that in the control; proliferation is significantly reduced in Hdac1/2cKO epidermis, and decreased proliferation is not rescued by concomitant deletion of p16. At least five mice of each genotype were analyzed. A two-tailed Student’s t-test was used to calculate the P-values. P < 0.05 was considered to be significant; error bars indicate SEM. White dashed lines in b and s–v indicate the boundary between epidermis and dermis. Bar = 25 μm in b–f and 50 μm in g–v. IF, immunofluorescence; IFE, interfollicular epidermis; K, keratin; P, postnatal day; rtTA, reverse tetracycline transactivator.
      Figure thumbnail fx3
      Supplementary Figure S3p16 deletion exacerbates differentiation defects in Hdac1/2cKO plantar epidermis after long-term doxycycline treatment. The schematic shows the timing of oral doxycycline treatment. All samples were analyzed on P42, 22 days after initiating treatment. (a, b) IF shows that the expression of both K14 (green) and K10 (red) is expanded in hyperplastic Hdac1/2cKO plantar epidermis relative to that in the control 22 days after initiating doxycycline treatment. (c, d) IF reveals the persistence of some cells double positive for LOR (red) and K14 (green) in hyperplastic (d) Hdac1/2cKO epidermis (white arrow) relative to in the control. (e, f) The levels of acetylated c-MYC (red) are similar in hyperplastic Hdac1/2cKO epidermis and in control. (g, h) The levels of p53Ac are similar in hyperplastic Hdac1/2cKO epidermis and control. (i, j) Differentiation of (i) p16-null epidermis indicated by expression of K10 (red) and K14 (green) is similar to that of the (a) control. (j) The absence of p16 in Hdac1/2cKO mice results in the appearance of cells strongly double positive for K10 and K14 (arrow; compared with b). (k, l) Differentiation of (k) p16-null epidermis indicated by expression of LOR (red) and K14 (green) is similar to that of the (c) control. (l) The absence of p16 in Hdac1/2cKO mice results in the appearance of cells strongly double positive for LOR and K14 (arrow; compare with d). (n) Cells with high levels of acetylated c-MYC (red) are retained in Hdac1/2cKO p16KO triple-mutant epidermis after 22 days of doxycycline treatment but are absent in (m) p16KO epidermis, (e) control epidermis, and (f) repopulated Hdac1/2cKO epidermis. (p) Cells with high levels of acetylated p53 are retained in Hdac1/2cKO p16KO triple-mutant epidermis after long-term doxycycline induction but are absent in (o) p16KO epidermis, (g) control epidermis, and (h) repopulated Hdac1/2cKO epidermis. At least four mice of each genotype were analyzed in each experiment. Bar = 50 μm. IF, immunofluorescence; K, keratin; LOR, loricrin; P, postnatal day.
      Figure thumbnail fx4
      Supplementary Figure S4Hdac1/2cKO skin displays inflammation 22 days after initiating deletion. (a–f) At 22 days after initiating doxycycline treatment, Hdac1/2cKO plantar skin displays thickening of the epidermis relative to the control. (a, b) CD3+ T cells are absent in control and mutant skin; (c, d) however, expression of the inflammatory marker CD26 is elevated in the mutant epidermis and dermal fibroblasts relative to that in the control (red signal, arrows), and (e, f) mutant dermis displays increased numbers of mast cells (purple signal, arrows). Dashed lines represent epidermal–dermal borders. Bar = 50 μm in a–d and 30 μm in e and f. (g) Quantitation of mast cell numbers shows that these are significantly increased in Hdac1/2cKO plantar skin relative to that in the controls 22 days after initiating doxycycline treatment. Five pairs of mutant and control mice were analyzed. A two-tailed Student’s t-test was used to calculate the P-value. P < 0.05 was considered to be significant; error bars indicate SEM. K, keratin.
      Figure thumbnail fx5
      Supplementary Figure S5Forced expression of GLI2ΔN in plantar epidermis initiates BCC-like tumorigenesis. (a–o) The left schematic indicates the timing of treatment with high-dose doxycycline water; samples were analyzed on P55, 7 days after initiating treatment. (pw) The right schematic indicates the timing of treatment with low-dose doxycycline water; samples were analyzed on P69, 21 days after initiating treatment. (a, b) Histology shows the thickening and invagination of plantar epidermis 7 days after initiating high-dose doxycycline treatment to induce the expression of GLI2ΔN compared with epidermis in the littermate control. (c, d) IF shows efficient induction of the expression of GLI2ΔN (green) and increased proliferation (red) in (d) K5-rtTA tetO-GLI2ΔN epidermis compared with that in the (c) control. (e) Quantitation of the percentage of Ki-67‒positive basal cells shows that proliferation is significantly increased in induced K5-rtTA tetO-GLI2ΔN epidermis compared with that in the control. Five pairs of control and mutant mice were analyzed. A two-tailed Student’s t-test was used to calculate the P-value. P < 0.05 was considered significant; error bars indicate SEM. (f–i) GLI2ΔN-expressing plantar epidermis shows ectopic expression of BCC markers (f, g) K17 (red) and (h, i) SOX9 (red). (j–m) HDAC1/2 protein levels are unchanged, and acetylated p53 is slightly elevated in GLI2ΔN-expressing plantar epidermis. (n, o) RNAscope shows slightly elevated levels of p16 mRNA in GLI2ΔN-expressing plantar epidermis. (p, q) Histology shows that long-term induction with low-dose doxycycline water results in tumor development resembling human superficial BCC. (r, s) Lesions resulting from long-term induction of K5-rtTA tetO-GLI2ΔN mice show robust proliferation at the leading edges of the downgrowths, as indicated by Ki-67 staining (red). (t–w) Lesions resulting from long-term induction of K5-rtTA tetO-GLI2ΔN mice express the BCC markers (t, u) K17 (red) and (v, w) SOX9 (red) (arrows in u and w). IF for myc-tagged GLI2ΔN is shown in green. (a–o) Five pairs of GLI2ΔN-expressing and control mice were analyzed; (p–w) two pairs of GLI2ΔN-expressing and control mice were analyzed. Control mice lacked K5-rtTA or tetO-GLI2ΔN transgenes. Bar = 25 μm in for a, b, f–i, p, and q; 50 μm in c, d, and j–o; and 30 μm in r–w. BCC, basal cell carcinoma; IF, immunofluorescence; K, keratin; rtTA, reverse tetracycline transactivator.
      Figure thumbnail fx6
      Supplementary Figure S6HDAC1 and HDAC2 act semiredundantly in regulating pre-BCC development. The schematic shows the timing of treatment with high-dose doxycycline water; all samples were analyzed on P55 after 7 days of treatment. (a–d) Histology shows that (b) homozygous loss of Hdac1 combined with heterozygous loss of Hdac2 or (d) heterozygous loss of Hdac1 combined with homozygous loss of Hdac2 has no obvious effect on the morphology of GLI2ΔN-driven pre-BCC in plantar epidermis compared with that in the (a, c) controls. (e–h) IF for GLI2ΔN (green) and (e, f) HDAC1 (red) or (g, h) HDAC2 (red) shows that Hdac1 or Hdac2 are efficiently deleted in GLI2ΔN-expressing basal cells in mice carrying homozygous Hdac1cKO or homozygous Hdac2cKO alleles, respectively. (i–l) IF for GLI2ΔN (green) and Ki-67 (red) shows that (j) homozygous loss of Hdac1 combined with heterozygous Hdac2 deletion reduces proliferation compared with that in the (i) littermate control, whereas (l) heterozygous loss of Hdac1 combined with homozygous Hdac2 deletion does not affect proliferation compared with (k) control. (m–p) TUNEL assay (green) shows that basal cell apoptosis is absent in GLI2ΔN-expressing epidermis with (n) homozygous loss of Hdac1 combined with heterozygous Hdac2 deletion or (p) heterozygous loss of Hdac1 combined with homozygous Hdac2 deletion as well as in (m, o) littermate controls. (q, r) Quantitation of the percentage of GLI2ΔN-expressing basal cells positive for Ki67 shows a significant decrease in proliferation in the (q) epidermis with homozygous loss of Hdac1 combined with heterozygous Hdac2 deletion but no significant change in proliferation in the (r) epidermis with heterozygous loss of Hdac1 combined with homozygous Hdac2 deletion. (q) Four pairs of control and mutant mice were analyzed; (r) three pairs of control and mutant mice were analyzed. A two-tailed Student’s t-test was used to calculate the P-values. P < 0.05 was considered significant; error bars indicate SEM. Bar = 25 μm in a–d and 50 μm in e–p. BCC, basal cell carcinoma; IF, immunofluorescence; P, postnatal day.
      Figure thumbnail fx7
      Supplementary Figure S7Loss of p16 exacerbates abnormal differentiation in Hdac1/2-deficient GLI2ΔN-expressing plantar epidermis. The schematic shows the timing of treatment with high-dose doxycycline water; plantar epidermis was analyzed on P55 after 7 days of treatment. (a–d) IF for GLI2ΔN (green) and K10 (red) shows that K10 expression is slightly expanded in (b) Hdac1/2-deleted GLI2ΔN-expressing epidermis compared with that in (a) GLI2ΔN-expressing control; (c) deletion of p16 alone does not alter K10 expression in GLI2ΔN-expressing epidermis (compare with that in a); (d) codeletion of p16 and Hdac1/2 enhances the expansion of K10 expression (arrow) compared with (b) Hdac1/2 deletion alone. (e–h) IF for GLI2ΔN (green) and LOR (red) shows that LOR expression is slightly expanded in (f) Hdac1/2-deleted GLI2ΔN-expressing epidermis (arrow) compared with that in (e) GLI2ΔN-expressing control; (g) deletion of p16 alone does not alter LOR expression in GLI2ΔN-expressing epidermis (compare with that in e); (h) codeletion of p16 and Hdac1/2 enhances the expansion of LOR expression (arrow) compared with that in (f) Hdac1/2 deletion alone. (i–l) IF for GLI2ΔN (green) and acetylated c-MYC (red) shows that acetylated c-MYC levels are higher in (j) Hdac1/2 deleted GLI2ΔN-expressing epidermis (arrow) than in (i) GLI2ΔN-expressing control; (k) deletion of p16 alone does not alter acetylated c-MYC levels in GLI2ΔN-expressing epidermis (compare with that in l); (l) codeletion of p16 and Hdac1/2 appears to further enhance the levels of acetylated c-MYC (arrow) compared with (j) Hdac1/2 deletion alone (arrow). Five mice of each genotype were analyzed. Bar = 50 μm. IF, immunofluorescence; K, keratin; LOR, loricrin; P, postnatal day.
      Figure thumbnail fx8
      Supplementary Figure S8Topical romidepsin causes abnormal differentiation of normal and GLI2ΔN-expressing IFE. The schematics indicate the time frames for treatment of dorsal paw skin with topical romidepsin or vehicle and oral high-dose doxycycline water; all samples were analyzed on P57. (a, b) Romidepsin-treated dorsal paw skin exhibits epidermal damage (yellow arrow), abnormal hair follicles, and the presence of an inflammatory infiltrate in the dermis (black arrows). (c–j) Romidepsin-treated dorsal paw IFE displays expanded (d) K10 (red) and (f) LOR (red) expression (yellow arrows) and increased levels of (h) acetylated c-MYC (red) and (j) acetylated P53 (red), compared with (c, e, g, i) vehicle-treated IFE. White dashed lines indicate epidermal–dermal borders. (k–r) Romidepsin-treated GLI2ΔN-expressing dorsal IFE displays slightly expanded (l) K10 (red) and (n) LOR (red) expression (yellow arrows) and increased levels of (p) acetylated c-MYC (red) and (r) acetylated P53 (red), compared with (k, m, o, q) vehicle-treated GLI2ΔN-expressing IFE. IF for myc-tagged GLI2ΔN is shown in green. n = 5 vehicle-treated and 5 romidepsin-treated mice in a–j; n = 4 vehicle-treated and 4 romidepsin-treated mice in k–r. Bar = 50 μm. IF, immunofluorescence; IFE, interfollicular epidermis; K, keratin; LOR, loricrin; WT, wild-type.

      References

        • Anderson-Dockter H.
        • Clark T.
        • Iwamoto S.
        • Lu M.
        • Fiore D.
        • Falanga J.K.
        • et al.
        Diagnostic utility of cytokeratin 17 immunostaining in morpheaform basal cell carcinoma and for facilitating the detection of tumor cells at the surgical margins.
        Dermatol Surg. 2012; 38: 1357-1366
        • Arwert E.N.
        • Mentink R.A.
        • Driskell R.R.
        • Hoste E.
        • Goldie S.J.
        • Quist S.
        • et al.
        Upregulation of CD26 expression in epithelial cells and stromal cells during wound-induced skin tumour formation.
        Oncogene. 2012; 31: 992-1000
        • Atwood S.X.
        • Li M.
        • Lee A.
        • Tang J.Y.
        • Oro A.E.
        GLI activation by atypical protein kinase C ι/λ regulates the growth of basal cell carcinomas.
        Nature. 2013; 494: 484-488
        • Atwood S.X.
        • Whitson R.J.
        • Oro A.E.
        Advanced treatment for basal cell carcinomas.
        Cold Spring Harb Perspect Med. 2014; 4: a013581
        • Bishton M.J.
        • Harrison S.J.
        • Martin B.P.
        • McLaughlin N.
        • James C.
        • Josefsson E.C.
        • et al.
        Deciphering the molecular and biologic processes that mediate histone deacetylase inhibitor-induced thrombocytopenia [published correction appears in Blood 2015;125:3824–3825].
        Blood. 2011; 117: 3658-3668
        • Blanpain C.
        • Fuchs E.
        Epidermal stem cells of the skin.
        Annu Rev Cell Dev Biol. 2006; 22: 339-373
        • Brinkhuizen T.
        • Reinders M.G.
        • van Geel M.
        • Hendriksen A.J.
        • Paulussen A.D.
        • Winnepenninckx V.J.
        • et al.
        Acquired resistance to the Hedgehog pathway inhibitor vismodegib due to smoothened mutations in treatment of locally advanced basal cell carcinoma.
        J Am Acad Dermatol. 2014; 71: 1005-1008
        • Canettieri G.
        • Di Marcotullio L.
        • Greco A.
        • Coni S.
        • Antonucci L.
        • Infante P.
        • et al.
        Histone deacetylase and Cullin3-REN(KCTD11) ubiquitin ligase interplay regulates Hedgehog signalling through Gli acetylation.
        Nat Cell Biol. 2010; 12: 132-142
        • Duvic M.
        Histone deacetylase inhibitors for cutaneous T-cell lymphoma.
        Dermatol Clin. 2015; 33: 757-764
        • Grachtchouk M.
        • Pero J.
        • Yang S.H.
        • Ermilov A.N.
        • Michael L.E.
        • Wang A.
        • et al.
        Basal cell carcinomas in mice arise from hair follicle stem cells and multiple epithelial progenitor populations.
        J Clin Invest. 2011; 121: 1768-1781
        • Higashitsuji H.
        • Higashitsuji H.
        • Masuda T.
        • Liu Y.
        • Itoh K.
        • Fujita J.
        Enhanced deacetylation of p53 by the anti-apoptotic protein HSCO in association with histone deacetylase 1 [published correction appears in J Biol Chem 2020;295:9264].
        J Biol Chem. 2007; 282: 13716-13725
        • Hughes M.W.
        • Jiang T.X.
        • Lin S.J.
        • Leung Y.
        • Kobielak K.
        • Widelitz R.B.
        • et al.
        Disrupted ectodermal organ morphogenesis in mice with a conditional histone deacetylase 1, 2 deletion in the epidermis.
        J Invest Dermatol. 2014; 134: 24-32
        • Jacks T.
        • Remington L.
        • Williams B.O.
        • Schmitt E.M.
        • Halachmi S.
        • Bronson R.T.
        • et al.
        Tumor spectrum analysis in p53-mutant mice.
        Curr Biol. 1994; 4: 1-7
        • Kouzarides T.
        Acetylation: a regulatory modification to rival phosphorylation?.
        EMBO J. 2000; 19: 1176-1179
        • LeBoeuf M.
        • Terrell A.
        • Trivedi S.
        • Sinha S.
        • Epstein J.A.
        • Olson E.N.
        • et al.
        Hdac1 and Hdac2 act redundantly to control p63 and p53 functions in epidermal progenitor cells.
        Dev Cell. 2010; 19: 807-818
        • Mirza A.N.
        • Fry M.A.
        • Urman N.M.
        • Atwood S.X.
        • Roffey J.
        • Ott G.R.
        • et al.
        Combined inhibition of atypical PKC and histone deacetylase 1 is cooperative in basal cell carcinoma treatment.
        JCI Insight. 2017; 2: e97071
        • Novelli M.
        • Savoia P.
        • Fierro M.T.
        • Verrone A.
        • Quaglino P.
        • Bernengo M.G.
        Keratinocytes express dipeptidyl-peptidase IV (CD26) in benign and malignant skin diseases.
        Br J Dermatol. 1996; 134: 1052-1056
        • Oh H.S.
        • Smart R.C.
        An estrogen receptor pathway regulates the telogen-anagen hair follicle transition and influences epidermal cell proliferation.
        Proc Natl Acad Sci USA. 1996; 93: 12525-12530
        • Potten C.S.
        • Saffhill R.
        • Maibach H.I.
        Measurement of the transit time for cells through the epidermis and stratum corneum of the mouse and guinea-pig.
        Cell Tissue Kinet. 1987; 20: 461-472
        • Pricl S.
        • Cortelazzi B.
        • Dal Col V.
        • Marson D.
        • Laurini E.
        • Fermeglia M.
        • et al.
        Smoothened (SMO) receptor mutations dictate resistance to vismodegib in basal cell carcinoma.
        Mol Oncol. 2015; 9: 389-397
        • Rady P.
        • Scinicariello F.
        • Wagner Jr., R.F.
        • Tyring S.K.
        p53 mutations in basal cell carcinomas.
        Cancer Res. 1992; 52: 3804-3806
        • Ramírez A.
        • Bravo A.
        • Jorcano J.L.
        • Vidal M.
        Sequences 5' of the bovine keratin 5 gene direct tissue- and cell-type-specific expression of a lacZ gene in the adult and during development.
        Differentiation. 1994; 58: 53-64
        • Sager P.T.
        • Balser B.
        • Wolfson J.
        • Nichols J.
        • Pilot R.
        • Jones S.
        • et al.
        Electrocardiographic effects of class 1 selective histone deacetylase inhibitor Romidepsin.
        Cancer Med. 2015; 4: 1178-1185
        • Sekulic A.
        • Von Hoff D.
        Hedgehog pathway inhibition.
        Cell. 2016; 164: 831
        • Seto E.
        • Yoshida M.
        Erasers of histone acetylation: the histone deacetylase enzymes.
        Cold Spring Harb Perspect Biol. 2014; 6: a018713
        • Sharpe H.J.
        • Pau G.
        • Dijkgraaf G.J.
        • Basset-Seguin N.
        • Modrusan Z.
        • Januario T.
        • et al.
        Genomic analysis of smoothened inhibitor resistance in basal cell carcinoma.
        Cancer Cell. 2015; 27: 327-341
        • Sharpless N.E.
        • Bardeesy N.
        • Lee K.H.
        • Carrasco D.
        • Castrillon D.H.
        • Aguirre A.J.
        • et al.
        Loss of p16Ink4a with retention of p19Arf predisposes mice to tumorigenesis.
        Nature. 2001; 413: 86-91
        • Soufir N.
        • Molès J.P.
        • Vilmer C.
        • Moch C.
        • Verola O.
        • Rivet J.
        • et al.
        P16 UV mutations in human skin epithelial tumors.
        Oncogene. 1999; 18: 5477-5481
        • Tang Y.
        • Zhao W.
        • Chen Y.
        • Zhao Y.
        • Gu W.
        Acetylation is indispensable for p53 activation [published correction appears in Cell 2008;133:1290].
        Cell. 2008; 133: 612-626
        • Teta M.
        • Choi Y.S.
        • Okegbe T.
        • Wong G.
        • Tam O.H.
        • Chong M.M.
        • et al.
        Inducible deletion of epidermal Dicer and Drosha reveals multiple functions for miRNAs in postnatal skin.
        Development. 2012; 139: 1405-1416
        • Urman N.M.
        • Mirza A.
        • Atwood S.X.
        • Whitson R.J.
        • Sarin K.Y.
        • Tang J.Y.
        • et al.
        Tumor-derived suppressor of fused mutations reveal hedgehog pathway interactions.
        PLoS One. 2016; 11e0168031
        • Vidal V.P.
        • Ortonne N.
        • Schedl A.
        SOX9 expression is a general marker of basal cell carcinoma and adnexal-related neoplasms.
        J Cutan Pathol. 2008; 35: 373-379
        • Waikel R.L.
        • Wang X.J.
        • Roop D.R.
        Targeted expression of c-Myc in the epidermis alters normal proliferation, differentiation and UV-B induced apoptosis.
        Oncogene. 1999; 18: 4870-4878
        • Wang C.
        • Wu H.
        • Evron T.
        • Vardy E.
        • Han G.W.
        • Huang X.P.
        • et al.
        Structural basis for smoothened receptor modulation and chemoresistance to anticancer drugs.
        Nat Commun. 2014; 5: 4355
        • Winter M.
        • Moser M.A.
        • Meunier D.
        • Fischer C.
        • Machat G.
        • Mattes K.
        • et al.
        Divergent roles of HDAC1 and HDAC2 in the regulation of epidermal development and tumorigenesis.
        EMBO J. 2013; 32: 3176-3191
        • Zhang H.
        • Ping X.L.
        • Lee P.K.
        • Wu X.L.
        • Yao Y.J.
        • Zhang M.J.
        • et al.
        Role of PTCH and p53 genes in early-onset basal cell carcinoma.
        Am J Pathol. 2001; 158: 381-385