Tumorigenic Activity of Merkel Cell Polyomavirus T Antigens Expressed in the Stratified Epithelium of Mice

Merkel cell carcinoma (MCC) is a highly aggressive and frequently lethal skin cancer. Risk factors for developing MCC include excessive sun exposure, advanced age, and immunosuppression induced by HIV infection, malignancy, or medical conditions such as organ transplantation or autoimmune disease. Given the immune status risk, a search for etiologic pathogens in MCC led to the discovery of Merkel cell polyomavirus (MCPyV; ref. 1).

MCPyV is one of the 13 known human polyomaviruses, each containing a small circular, dsDNA genome organized into two transcriptional regions, early and late, separated by a noncoding regulatory region (2). The MCPyV early region encodes for large T antigen (LT), 57kT, small T antigen (ST), and ALTO (alternate LT open reading frame; refs. 3, 4). The late region encodes for the viral coat proteins VP1 and VP2. LT and ST share an N-terminal DNA J domain followed by unique domains due to alternate splicing. LT contains an RB1-binding or LXCXE motif, nuclear localization signal, a viral origin DNA-binding domain, and a helicase domain. Full-length LT is required for viral replication. ST contains a unique domain that binds to protein phosphatase PP2A. Notably, in at least 80% of MCC, clonal integration of the MCPyV viral genome results in expression of an intact ST and a truncated form of LT, but not the viral coat proteins or progeny virions (4). The truncated LTs isolated from numerous MCC tumor specimens typically retain the J domain and LXCXE motif, but not the DNA-binding and helicase domains.

The prevalence of MCPyV infection in children and adults is high although there are no reports of symptoms resulting from acute infection (5–7). MCPyV infection appears to be cutaneotropic in healthy individuals because virions can be readily isolated from skin swabs (8). It is not known what cells support MCPyV replication although skin keratinocytes, skin progenitor cells, or Merkel cells are likely candidates. Normal Merkel cells function as the gentle touch receptor and are present in the basal layer of the epidermis (9). Because normal Merkel cells are terminally differentiated and do not undergo cell division, ongoing studies seek to identify the Merkel cell progenitor as well as the MCC cell-of-origin (10, 11).

Mouse models for cancer have been useful to determine the oncogenic activity of many oncogenes and tumor suppressors. In particular, mice that express the human papillomavirus (HPV16) E6 and E7 viral oncogenes in keratinocytes have demonstrated their oncogenic potential. Here, we demonstrate that MCC tumor-derived MCPyV T antigens have tumorigenic activity in vivo.

The MCPyV early region, isolated from MCC tumor specimen MCCw168 (MCPyV168; GenBank: KC426954.1), was PCR amplified and cloned into the pBigT plasmid vector between XhoI and NotI restriction sites (12, 13). A 6-kb fragment containing the neomycin resistance gene and the triple SV40 polyadenylation (tpA) that function as transcription terminators (stop) flanked by two loxP sites (loxP-stop-loxP or LSL), MCPyV168, and bovine growth hormone polyadenylation terminator was cut from the pBigT vector and cloned into the pROSA26PA plasmid between PacI and AscI restriction sites (12). Linearized ROSA26-LSL-MCPyV168 plasmid was electroporated into 129SvEV/C57BL/6 (v6.5) hybrid embryonic stem (ES) cells. ES clones were selected with neomycin and screened by PCR for integration into the ROSA26 locus. Sequencing of the short and long homology arms, MCPyV168 gene, and loxP sites was performed to confirm correct targeting. Two appropriately targeted ES cells with normal karyotype were injected into blastocysts derived from C57BL/6 mice and implanted into pseudo-pregnant females. High-grade chimeras were generated and crossed with C57BL/6 mice to obtain germline transmission. The mouse strain has been designated Gt(ROSA)26Sor<tm1(MCPyV168)Jdec> (MGI:5576261) by Mouse Genome Informatics. The conditional-ready allele is designated ROSA26-LSL-MCPyV168. Initially, the ROSA26-LSL-MCPyV168 mice were crossed with B6N.Cg-Tg(K14Cre)1Amc/J mouse strain (Jackson Laboratory).

To perform studies in a defined mouse genetic background, the ROSA26-LSL-MCPyV168 mice were backcrossed to FVB/N for 10 generations. The N10 generation was intercrossed to produce N10F1 and N10F2 mice homozygous for the ROSA26-LSL-MCPyV168 allele. One male and two female N10F2 mice on the FVB/N background were transferred to the American Association of Laboratory Animal Care-approved McArdle Laboratory Animal Care Unit of the University of Wisconsin School of Medicine and Public Health (Madison, WI). The ROSA26-LSL-MCPyV168 mice were crossed with transgenic mice expressing Cre recombinase driven by the human keratin 14 (Krt14 or K14) promoter (K14Cre) on the FVB/N murine genetic background that have been described previously (14).

K14Cre-MCPyV168 mice were evaluated weekly starting at 3 weeks of age to monitor for acute phenotypes and the presence of papillomas. The K14E6E7 transgenic mice included in this study express the human papillomavirus type 16 (HPV16) E6 and E7 oncogenes driven by the K14 promoter and have been described previously (15, 16). All procedures were performed according to protocols approved by the Dana-Farber Cancer Institute Animal Care and Use Committee (Boston, MA) and the University of Wisconsin Institutional Animal Care and Use Committee.

Genomic DNA was isolated from tail snips and resuspended in Tris/EDTA buffer. Equivalent DNA concentrations from each sample were analyzed by separate PCR reactions to identify the wild-type or recombined ROSA26 allele, presence of the Cre recombinase gene, and the loxP-stop-loxP or excised cassette. See Table 1 for primer sequences. The PCR products were evaluated using agarose gel electrophoresis. All primers were synthesized by Integrated DNA Technologies.

Tissues were harvested, fixed in 4% paraformaldehyde, and embedded in paraffin. Serial sections (5 μm thick) were cut and hematoxylin and eosin (H&E)-stained sections were evaluated microscopically for histopathological features. Images were captured using a Zeiss AxioImager M2 microscope and AxioVision software version 4.8.2.

Where indicated, mice were treated with 10 μL per gram body weight of a 12.5 mg/mL solution of bromodeoxyuridine (BrdUrd) by intraperitoneal injection one hour before sacrifice. For BrdUrd quantification, representative tissue sections of the ear epithelium were processed for BrdUrd-specific IHC. One slide from at least three individual mice was analyzed by microscopy and 10 images (×20) were captured. The total number of cells and BrdUrd-positive cells were quantified with an automated cell counting program [developed by David Ornelles (Wake Forest University School of Medicine, Winston-Salem, NC); unpublished data] using ImageJ software version 1.47 (NIH, Bethesda, MD). The percentage of total BrdUrd-positive cells was calculated for each sample using the average of 10 fields from each mouse. These values were then averaged among the mice in each group. The SD reflects variation between individual mice. A two-sided Wilcoxon rank-sum test was used to compare the average percentage of BrdUrd-positive cells between groups. Statistical analysis was performed using MSTAT statistical software version 6.1.2 (http://www.mcardle.wisc.edu/mstat; last accessed May 21, 2014).

Protein concentrations of whole tissue lysates were determined using Bio-Rad Protein Assay reagent (Bio-Rad). Equivalent concentrations of protein (35 μg skin, 20 μg footpads) were resolved using precast polyacrylamide gels (Mini-Protean TGX AnyKD or Mini-Protean TGX 4%–20% gradient gels; Bio-Rad) and then transferred to 0.45 μm nitrocellulose membrane (Whatman Protran BA85; GE Healthcare). Following transfer, membranes were stained with 0.1% Ponceau S in 1% acetic acid to verify equal loading and successful transfer of proteins and then blocked with 5% nonfat dry milk in TBS-BGT (TBS containing BSA and glycine supplemented with 0.1% Tween-20). Separated and immobilized proteins were analyzed using the following primary antibodies at the indicated dilutions: β-actin (1:5000; Sigma), MCPyV LT (Ab3; 1:5000; refs. 13, 17), MCPyV ST/LT (Ab5; 1:5000; ref. 13), and survivin (1:1000; Cell Signaling Technology). To visualize immune complexes, horseradish peroxidase-conjugated secondary antibodies (1:10,000; Jackson Immunoresearch) and chemiluminescent substrates (Amersham ECL Plus; GE Healthcare) were applied to membranes, followed by exposure to film (Kodak Biomax XAR; Sigma). All washes were performed with TBS-BGT, and blots were stripped with 0.2 mol/L sodium hydroxide where applicable.

Tissue sections were deparaffinized and rehydrated with xylenes and graded ethanol, respectively. For IHC staining, endogenous peroxidase activity was quenched with 3% H₂O₂ in methanol. Heat-induced antigen retrieval was performed in 0.01 mol/L citrate buffer, pH 6.0, or 0.01 mol/L Tris/0.001 mol/L EDTA, pH 9.0, solutions. Antibodies against the following proteins were used: ATM, phospho-serine 1981 (pS1981; Millipore), BrdUrd (Calbiochem), cytokeratin 8 (CK8; Developmental Studies Hybridoma Bank), cytokeratin 20 (CK20; Dako), γ-H2AX, phospho-serine 139 (Millipore), keratin 14 (K14; Covance), mini-chromosome maintenance protein 7 (MCM7; ThermoScientific), biotinylated horse anti-mouse/rabbit IgGs (Vector Laboratories), Alexa-Fluor488, Alexa-Fluor594, and Alex-Fluor647 (Molecular Probes). For IHC staining, proteins were visualized with 3,3′-diaminobenzidine (Vector Laboratories), and tissues were counterstained with hematoxylin. For immunofluorescence analysis, tissues were mounted with Vectashield mounting media containing DAPI (Vector Labs). All images were taken with a Zeiss AxioImager M2 microscope using the AxioVision software version 4.8.2.

MCPyV is routinely shed from the skin, suggesting that the virus infects one or more epidermal cell types in the cutaneous stratified epithelium (8). In addition, several laboratories have reported that normal Merkel cells are derived from keratin 14 (K14)-positive epidermal progenitor cells (18, 19). To test whether expression of MCC-derived MCPyV T antigens in K14-positive epithelial cells could induce tumors, we generated a targeted knockin mouse strain capable of tissue-specific expression (Fig. 1A). The MCPyV early region DNA, isolated from an MCC tumor (MCCw168), expresses an N-terminal fragment (275 residues) of LT as well as wild-type ST and is referred to as MCPyV168 (13). Although the ROSA26 promoter used in the transgene is constitutively active and ubiquitously expressed in mice (20), the presence of the loxP-stop-loxP (LSL) cassette immediately upstream prevents transcription of the MCPyV168 gene until excised by Cre recombinase (21, 22). Because the K14 promoter is active in basal keratinocytes of the epidermis (23), we chose to induce deletion of the LSL cassette with Cre recombinase driven by the K14 promoter (K14Cre; Fig. 1A). The ROSA26-LSL-MCPyV168 and K14Cre mice were crossed to yield K14Cre/⁺; ROSA26-LSL-MCPyV168/⁺ mice, resulting in mice heterozygous for both alleles and hereafter referred to as K14Cre-MCPyV168 mice. PCR genotyping with DNA isolated from mice on a mixed genetic (Fig. 1B) and FVB/N backgrounds (Fig. 1C), along with immunoblotting for MCPyV T antigen protein expression in lysates prepared from the skin and footpads of adult mice (Fig. 1D), shows that expression of Cre recombinase and subsequent excision of the LSL sequence correlated with robust expression of the ST and truncated LT viral proteins. Therefore, expression of the tumor-derived MCPyV T antigens in the stratified epithelium was dependent on conditional deletion of the LSL cassette by K14-driven Cre recombinase expression.

Expression of the MCPyV ST and truncated LT was associated with several overt phenotypes (Fig. 2). Within 8 to 10 days of birth, the K14Cre-MCPyV168 pups on both mixed (Fig. 2A1) and FVB/N (Fig. 2B1) genetic backgrounds had a markedly smaller body size and ruffled, disorganized fur coat compared with their Cre-negative ROSA26-LSL-MCPyV168 littermates. By postnatal day 18, the ruffled pelage of the K14Cre-MCPyV168 mice on mixed (Fig. 2A2 and 2A3) and FVB/N (Fig. 2B2) backgrounds had become more pronounced, and signs of hyperplasia and skin thickening, or acanthosis, were evident. Interestingly, some K14Cre-MCPyV168 mice on the mixed background developed blisters on their footpads (Fig. 2A3 and 2A4). This was the only phenotype unique to the mixed background and was not observed in K14Cre-MCPyV168 mice on the FVB/N background. After weaning, the K14Cre-MCPyV168 pups on the mixed genetic background lost their epithelial phenotypes by around 2 months of age and were eventually indistinguishable from their Cre-negative littermates (not shown).

The K14Cre-MCPyV168 mice on the FVB/N genetic background exhibited more gross epithelial phenotypes than those on the mixed background. Notably, expression of the MCPyV T antigens had a fairly toxic effect in the FVB/N background. The breeding crosses between ROSA26-LSL-MCPyV168 and K14Cre mice on the FVB/N genetic background yielded approximately 51% (n = 50/98) Cre-positive, K14Cre-MCPyV168 mice. Of these mice, nearly one-third (30%, n = 15/50) died or was euthanized before weaning at 21 days of age due to stunted development or health issues consistent with malnourishment. The surviving mice continued to exhibit ruffled fur and flaky, thickened skin that persisted throughout their lifespan and led to mild to moderate alopecia in older mice. Progressive whisker shortening and loss were also observed in K14Cre-MCPyV168 mice (Fig. 2B4) in comparison with the control ROSA26-LSL-MCPyV168 mice (Fig. 2B3). Consistent with the skin phenotypes, the ears of K14Cre-MCPyV168 mice were thickened, reddened, and highly keratinized (Fig. 2B5 and 2B6) in comparison with those of the ROSA26-LSL-MCPyV168 control mice. In addition, K14Cre-MCPyV168 mice typically developed small eyes and cataracts (clouding of lens) by as early as 10 weeks of age (Fig. 2B5 and 2B6). Finally, expression of the MCPyV T antigens in the stratified epithelium was associated with the spontaneous development of benign tumors of epithelial origin or papillomas (Fig. 2B7 and 2B8). Taken together, these observations indicate that expression of the MCC tumor-derived MCPyV T antigens in the stratified epithelium facilitates the development of multiple gross phenotypes in the skin and several additional anatomical locations. Because of the enhanced development of skin-associated phenotypes observed with K14Cre-MCPyV168 mice on the FVB/N genetic background, these mice were used for further characterization of the effects of MCPyV T antigen expression in the cutaneous stratified epithelium.

We sought to determine whether there was histologic or molecular evidence of neoplastic progression that accompanied these gross phenotypes. We examined tissue from ROSA26-LSL-MCPyV168 and K14Cre-MCPyV168 mice, as well as tissue harvested from K14E6E7 mice as a positive control. K14-directed expression of the HPV16 oncogenes E6 and E7 has been shown to promote cell-cycle progression, induce epithelial hyperplasia, and perturb differentiation (15, 16). The epithelium of K14Cre-MCPyV168 mice was characterized by hyperplasia and thickening of the stratum corneum, also known as hyperkeratosis, compared with the normal epithelium of ROSA26-LSL-MCPyV168 mice (Fig. 3A). Notably, the degree of hyperplasia and hyperkeratosis in K14Cre-MCPyV168 mice was comparable with that seen in the stratified epithelium of K14E6E7 mice. In addition, the stratified epithelium of the whisker pads, footpads, and dorsal skin of neonatal and adult K14Cre-MCPyV168 mice also showed extensive hyperplasia and hyperkeratosis in comparison with age-matched control mice (Supplementary Fig. S1). These histological traits indicate that targeted expression of the MCPyV tumor-derived T antigens in the stratified epithelium induced epithelial hyperplasia to a degree comparable with that induced by E6 and E7 oncogenes from the well-validated oncogenic virus, HPV16.

Immunoblotting of tissue lysates of K14Cre-MCPyV168 mice demonstrated that MCPyV ST and truncated LT were expressed in the stratified epithelia (Fig. 1D). To assess the impact of MCPyV T antigen expression on RB1 function, we performed immunostaining for the mini-chromosome maintenance protein 7 (MCM7) in the same ear tissue samples. Mcm7 is an E2F-responsive gene and its expression is normally restricted to the highly proliferative basal layer of the stratified epithelium (24, 25). Previously, we demonstrated that expression of K14-HPV16 E7 protein redistributes MCM7 expression to the suprabasal layers and throughout the entire thickness of the stratified epithelium through its ability to bind and inactivate RB1 (26, 27). These findings were recapitulated in the analysis presented here (Fig. 3A). Although the ear epithelium of ROSA26-LSL-MCPyV168 mice contained MCM7-positive cells only in the basal layer, the K14E6E7 mice showed MCM7 positivity throughout the entire stratified epithelium. Similarly, tissue from K14Cre-MCPyV168 mice had robust MCM7 expression throughout all layers of the stratified epithelium in the ear epidermis as well as in the whisker pads, footpads, and dorsal skin (Fig. 3A and Supplementary Fig. S1). The intensity and pattern of MCM7 expression in K14Cre-MCPyV168 epithelium were comparable with that in K14E6E7 mice. These data suggest that the MCPyV T antigens can induce expression and spatial redistribution of the E2F-responsive gene Mcm7 in the stratified epithelium of mice.

To evaluate the effect of MCPyV T antigen expression on differentiation of the stratified epithelium, we performed immunofluorescence for K14 and keratin 10 (Krt10 or K10) on the same ear samples (Fig. 3A). The epithelium of the ROSA26-LSL-MCPyV168 mice revealed a normal pattern of differentiation with K10 expression present in the differentiated, suprabasal layers of the epithelium and K14 expression restricted to the basal layer. In contrast, the epithelium of K14Cre-MCPyV168 mice contained an expanded layer of K10-positive cells consistent with the hyperplastic histopathology observed in the epithelium of these mice. Concomitantly, there was a drastic expansion of the K14-positive population, such that every cell in the stratified epithelium appeared to express K14. This aberrant keratin expression pattern was again similar to that observed in K14E6E7 mice, indicating that the tumor-derived MCPyV T antigens can disrupt the normal differentiation program of the stratified epithelium in vivo to the same extent as HPV16 E6 and E7 oncoproteins.

To assess the impact of MCPyV T antigen expression on cellular proliferation in vivo, we measured DNA replication by evaluating BrdUrd incorporation within a 1-hour window before sacrifice. IHC was performed to identify BrdUrd-positive cells in the epithelium and the percentage of BrdUrd-positive cells was calculated (Fig. 3B). The control ROSA26-LSL-MCPyV168 ear epithelium contained approximately 3% BrdUrd-positive cells and, as anticipated, these cells were only present in the basal layer. Strikingly, there was a 5-fold increase in the total percentage of BrdUrd-positive cells in K14Cre-MCPyV168 ear epithelium compared with control mice (15.5%; P = 0.036). Moreover, expression of the MCPyV T antigens changed the spatial distribution of BrdUrd-incorporating cells from a strictly basal staining pattern to one where BrdUrd-positive cells were observed throughout the parabasal and suprabasal layers of the stratified epithelium (Fig. 3B). Expression of the HPV16 E6 and E7 oncogenes also induced a significant increase in the total percentage of BrdUrd-positive cells compared with ROSA26-LSL-MCPyV168 mice (15.4%; P = 0.036). There was no difference between the percentage of BrdUrd-positive cells in the epithelium of K14Cre-MCPyV168 and K14E6E7 mice (P = 0.69). Therefore, in vivo expression of the MCPyV T antigens in the stratified epithelium leads to a significant increase in cellular proliferation comparable with that induced by the HPV16 E6 and E7 proteins. Collectively, these results indicate that epithelial expression of the MCPyV T antigens can induce phenotypes consistent with neoplastic progression at a level similar to the HPV16 oncogenes.

K14-directed expression of the MCPyV T antigens resulted in smaller eye size and cataract development (Fig. 2B5 and 2B6). The K14 promoter has been previously shown to drive transgene expression in the postnatal mouse anterior lens epithelium (28). To determine whether expression of the MCPyV T antigen affects differentiation of the lens epithelium, eyes from ROSA26-LSL-MCPyV168 and K14Cre-MCPyV168 mice were collected and representative longitudinal sections used for histopathological analysis (Supplementary Fig. S2). Abnormalities in the eyes of K14Cre-MCPyV168 mice were primarily restricted to the lens epithelium. Low magnification images indicate that the ocular lens in K14Cre-MCPyV168 mice was overall less organized and irregularly shaped compared with the lens from ROSA26-LSL-MCPyV168 control mice. The normally tightly regulated process of terminal differentiation in the transition zone of the differentiating lens epithelium observed in the ROSA26-LSL-MCPyV168 mice was significantly disrupted in the K14Cre-MCPyV168 mice. The posterior lens is normally composed of terminally differentiated fiber cells and devoid of epithelial cells, as observed in the ROSA26-LSL-MCPyV168 control mice. However, in the eyes of K14Cre-MCPyV168 mice, the posterior lens was abnormally occupied by a layer of disorganized epithelial cells. The fiber cell compartment in the lens of some K14Cre-MCPyV168 mice also contained cellular aggregations of an indeterminate origin. These data therefore indicate that targeted MCPyV T antigen expression in the stratified epithelium can disrupt the normal pattern of cellular differentiation in multiple anatomical sites including the skin and ocular lens.

The spontaneous development of papillomas was the most striking epithelial phenotype observed in K14Cre-MCPyV168 mice on the FVB/N genetic background (Fig. 2B7 and 2B8). Of the K14Cre-MCPyV168 mice that survived weaning (n = 35/50), nearly half (n = 16/35 mice, 45.7%) spontaneously developed papillomas (Fig. 4A; black portion of bar graph). These papillomas developed within 2 to 3 months of age with some as early as 6 weeks of age. Moreover, 44% of the K14Cre-MCPyV168 mice that developed papillomas developed multiple papillomas (n = 7/16 mice, 43.75%). Papilloma incidence was irrespective of sex, and the tumors arose on both singly and multiply caged animals. Papillomas developed on the hairy skin of K14Cre-MCPyV168 mice at locations including the dorsal skin, hind legs, and at the area surrounding the base of tail (Fig. 4B). Some of the papillomas exhibited a sessile or smooth macroscopic appearance, whereas others were more exophytic in nature. Interestingly, tumor regression occurred spontaneously in approximately one quarter of the K14Cre-MCPyV168 mice that developed papillomas (n = 4/16; data not shown). We excluded the possibility that the papillomas observed on K14Cre-MCPyV168 mice were a result of infection with the murine papillomavirus (MmuPV1) that was recently reported to infect laboratory mice and induce papillomas (29). Although MmuPV1-induced papillomas are positive for L1 immunofluorescent staining, staining for MmuPV1 L1 protein was negative in the MCPyV-associated papillomas (data not shown).

The MCPyV-induced papillomas were evaluated for the presence of MCPyV T antigens, and the histopathology, differentiation, and extent of MCM7 expression were analyzed. The skin and papillomas were harvested from two individual K14Cre-MCPyV168 mice and compared by immunoblotting (Fig. 4C). No MCPyV T antigens were detected in the skin of ROSA26-LSL-MCPyV168 control mice, yet expression of the truncated MCPyV LT and ST was detected in lysates from the skin and papillomas harvested from the K14Cre-MCPyV168 mice. There was some variability in the levels of ST among the various papillomas.

The MCPyV-associated papillomas displayed conventional histopathological hallmarks associated with papillomatosis (Fig. 4D). Epithelial invaginations were evident in low-magnification images of a representative H&E-stained papilloma and higher magnification images revealed highly keratinized regions within the epithelial folds. Immunofluorescence analysis for K10 confirms the advanced differentiation of these keratinized regions. Consistent with our analysis of cellular differentiation in K14Cre-MCPyV168–stratified epithelium (Fig. 3A), the K14-positive fraction was significantly expanded in the papilloma compared with age and location-matched skin from a ROSA26-LSL-MCPyV168 mouse (Fig. 4D). Moreover, the MCM7 staining pattern within the papilloma was intensely positive throughout the full thickness of epithelium (Fig. 4D).

Since the discovery of MCPyV in 2008, several studies have revealed an ability of the virus to perturb a variety of cellular processes and pathways in vitro (13, 30–35). We analyzed whether expression of MCC-derived MCPyV T antigen perturbed some of these same molecular pathways in vivo. MCPyV LT activates transcription and subsequent expression of the BIRC5 gene encoding the antiapoptotic protein survivin, an activity dependent on the LXCXE RB-binding motif of LT (30). In lysates prepared from ROSA26-LSL-MCPyV168 control mouse skin, survivin protein was undetectable. In contrast, there was a substantial increase in levels of survivin in both the skin and papillomas harvested from K14Cre-MCPyV168 mice (Fig. 5A). The induction of survivin and the increased level of MCM7 expression provide two lines of indirect evidence for RB1 inactivation by the truncated LT expressed in K14Cre-MCPyV168–stratified epithelium of mice.

Expression of MCPyV LT in vitro has also been recently reported to activate the host DNA damage response (33, 36). Therefore, we evaluated whether an active DNA damage response was present in the stratified epithelium of ROSA26-LSL-MCPyV168 and K14Cre-MCPyV168 mice. The presence of phosphorylated histone 2 variant X (γ-H2AX) and phosphorylated Ataxia telangiectasia mutated (ATM) kinase was assessed by immunostaining. There were no γ-H2AX or pS1981-ATM–positive cells in the skin of ROSA26-LSL-MCPyV168 mice. In contrast, high levels of γ-H2AX and pS1981-ATM were observed in the nuclei of cells within papillomas arising in K14Cre-MCPyV168 mice, but not in the surrounding skin (Fig. 5B). The γ-H2AX and pS1981-ATM–positive cells in the papillomas from K14Cre-MCPyV168 were primarily concentrated in the relatively differentiated portions of the tumor. Indeed, immunofluorescence analysis verified that those regions positive for K10 expression in the more keratinized and differentiated portions of the papilloma correlated with the areas where γ-H2AX and pS1981-ATM–positive cells were most frequently observed.

The human malignancy with the strongest causal link to MCPyV is MCC (reviewed in ref. 37). Because normal Merkel cells are K14 positive and arise from epidermal progenitors in mice (18, 19), we hypothesized that the MCPyV168 transgene was expressed in Merkel cells. Unfortunately, we have been unable to visually detect the MCPyV T antigens by IHC in tissues harvested from K14Cre-MCPyV168 mice. Nonetheless, we evaluated the status of Merkel cells in K14Cre-MCPyV168 mice using coimmunofluorescence for two well-established markers of Merkel cells, cytokeratin 8 (CK8) and cytokeratin 20 (CK20; ref. 38; Fig. 6). The colocalization of these two Merkel cell markers was most evident in the murine whisker vibrissae, where the CK8⁺ and CK20⁺ Merkel cells lined the vibrissae follicle (Fig. 6A). Merkel cells were also visualized in the footpads and, to a lesser extent, in the dorsal skin (data not shown). On the basis of our observation that MCPyV T antigen expression in vivo induces strong expression of the E2F-responsive gene Mcm7 in the stratified epithelium (Figs. 3 and Fig. 4), we used MCM7 immunoreactivity as an indirect readout for MCPyV T antigen expression in Merkel cells. Coimmunofluorescence for MCM7 and CK8 was performed on representative tissue sections of whisker pads from adult ROSA26-LSL-MCPyV168 and K14Cre-MCPyV168 mice (Fig. 6B) to visualize MCM7 expression in Merkel cells. CK8-positive Merkel cells were observed in the vibrissae follicles of all mice. The vibrissae follicles of ROSA26-LSL-MCPyV168 control mice showed very few MCM7-positive cells, and, as expected, none of the CK8-positive Merkel cells were positive for MCM7 immunoreactivity given their terminally differentiated status (Fig. 6B, top; ref. 10). In contrast, the CK8-positive Merkel cells were MCM7 positive in the vibrissae of K14Cre-MCPyV168 mice. Qualitatively, there were no differences in Merkel cell number or morphology in the whisker vibrissae of K14Cre-MCPyV168 mice compared with ROSA26-LSL-MCPyV168 control mice (data not shown). These observations suggest that MCPyV T antigen expression results in increased levels of the E2F-responsive gene Mcm7 in Merkel cells, presumably through the inactivation of RB1.

Evidence supporting a causal link between MCPyV and MCC continues to grow, yet the precise molecular mechanisms involved in virus-induced transformation and carcinogenesis are unknown. To evaluate how the MCPyV T antigens found in MCC tumors interact with and affect the cellular environment in vivo, we developed a mouse model that expresses truncated LT and wild-type ST in the skin. Our studies reveal that K14-directed expression of MCC tumor-derived MCPyV T antigens profoundly disrupts the normal properties of the stratified epithelium and promotes tumorigenesis. Several gross phenotypes developed in K14Cre-MCPyV168 mice and were restricted to anatomical locations that contain stratified epithelium including the skin, footpads, whisker pads, and eyes (Fig. 2). Histopathologically, MCPyV T antigen expression in the stratified epithelium promoted hyperplasia, hyperkeratosis, aberrant differentiation, and papilloma formation (Figs. 3 and 4 and Supplementary Figs. S1 and S2). The epithelium of K14Cre-MCPyV168 mice showed increased intensity and frequency of MCM7 expression (Fig. 3) and increased levels of the antiapoptotic protein survivin (Fig. 5), indicating that the MCPyV T antigens deregulate E2F-responsive gene expression. Cellular proliferation was also significantly increased in the epithelium of K14Cre-MCPyV168 mice (Fig. 3).

Although K14Cre-MCPyV168 mice developed benign epithelial tumors (Fig. 4), they did not develop MCC during the time course of our studies. The cell of origin for MCC is controversial, and it is not known whether Merkel cells give rise to MCC (39). Normal Merkel cells are located in the basal layer of the epidermis in glabrous and hairy skin and express a discrete set of established markers including CK20 (reviewed in ref. 38). Merkel cells are required for gentle touch (40) and function as neuromechanical receptors (9, 41, 42). Normal Merkel cells are terminally differentiated and descend from a population of K14-positive epidermal progenitor cells (18, 19). Therefore, there are several possible cellular origins of MCC in addition to Merkel cells, including epidermal or dermal progenitors, perhaps a unique Merkel cell precursor, or possibly nonepidermal cells derived from other tissues. In the K14Cre-MCPyV168 mouse model, Merkel cells appeared to be relatively unaffected by MCPyV T antigen expression apart from being immunoreactive for MCM7 (Fig. 6). There were no qualitative differences in Merkel cell distribution, morphology, or number in K14Cre-MCPyV168 mice compared with Rosa26-LSL-MCPyV168 control mice. These findings suggest that other factors in addition to MCPyV, such as cooperating oncogenes, loss of tumor suppressors, or other known risk factors (43) are required for MCC development. Alternatively, MCC may develop from a cell lineage independent of a K14-derived precursor cell, or mouse K14-derived cells are distinct from those in human.

Nonetheless, MCPyV T antigen expression had a marked effect on keratinocytes in the stratified epithelium. Prominent epithelial phenotypes developed in two different murine genetic backgrounds, although the scope and severity were more pronounced in K14Cre-MCPyV168 mice on the FVB/N background (Fig. 2). This observation is consistent with previous results that found that FVB/N mice are more susceptible to chemically induced skin carcinogenesis than other murine genetic backgrounds (44). Malignant progression induced by the K14-driven expression of the HPV16 genome was only observed in FVB/N mice (45). The FVB/N strain carries a polymorphic variant of the Ptch1 gene that contributes to spontaneous and oncogene-induced squamous carcinoma of the skin (46). This suggests the possibility that the HPV and MCPyV viral oncogenes can cooperate with perturbations in the Hedgehog signaling pathway to induce malignancy.

A remarkable and perhaps somewhat surprising observation in this study was that the skin phenotype induced by MCPyV T antigens was essentially identical to that induced by the high-risk HPV16 oncogenes E6 and E7 (Fig. 3). The stratified epithelium of K14Cre-MCPyV168 mice had an equivalent level of hyperplasia, disrupted cellular differentiation, and increased cellular proliferation as K14E6E7 transgenic mice. Moreover, the spatial pattern of MCM7 expression in the stratified epithelium of K14Cre-MCPyV168 and K14E6E7 mice was disrupted to a similar extent, while the intensity of MCM7 expression was stronger in the skin expressing the MCPyV T antigens. The ability of HPV16 E7 to increase MCM7 levels is due to direct binding and functional inactivation of the RB1 tumor suppressor protein (26). Because both HPV16 E7 and MCPyV LT contain an LXCXE RB1-binding motif, it is reasonable to speculate that the same function of MCPyV LT is responsible for the changes in MCM7 expression seen in the K14Cre-MCPyV168 epithelium. The similarities in epithelial phenotypes and tissue histopathology of K14Cre-MCPyV168 and K14E6E7 transgenic mice suggest that the MCPyV T antigens are at least as potent as the HPV16 E6 and E7 oncogenes in promoting hyperplasia, E2F-responsive gene expression, aberrant cellular differentiation, and cellular proliferation in the stratified epithelium of mice. These results also indicate that keratinocytes are sensitive to MCPyV T antigen expression, at least in the context of the K14Cre-MCPyV168 mouse model. Whether this finding has implications on the understanding of MCPyV tropism or the cell of origin for MCC has yet to be determined.

Interestingly, the similarities between the tumor-derived MCPyV T antigens and the HPV16 E6 and E7 oncogenes at the molecular level did not yield similar results with respect to spontaneous tumorigenesis. Close to half (45.7%) of all K14Cre-MCPyV168 mice on the FVB/N genetic background that survived past weaning spontaneously developed papillomas (Fig. 4). We have previously demonstrated that the HPV16 E6 and E7 oncogenes expressed either singly or combined in mice using the K14 promoter are capable of spontaneous skin tumorigenesis (15, 16, 47). However, by 6 months of age only approximately 5% of K14E6E7 mice had developed skin tumors and a peak incidence of 40% was not achieved until the mice were 12 months old. Approximately 70% of these were squamous tumors and the rest were sebaceous epitheliomas. In contrast, most papillomas on K14Cre-MCPyV168 mice developed between 2 months and 3 months of age and some within 6 weeks. Within the time frame of our study, the skin tumors observed on K14Cre-MCPyV168 mice remained benign. In contrast, most of the squamous tumors present on K14E6E7 mice were malignant carcinomas as evidenced by the level of tumor cell differentiation. This observation may again suggest that MCPyV-associated skin carcinogenesis requires other cofactors in addition to the MCPyV T antigens.

Although benign, the MCPyV-associated papillomas that developed on K14Cre-MCPyV168 mice appear to contain an environment favorable for malignant conversion. Like the stratified epithelium (Fig. 3), the papillomas on K14Cre-MCPyV168 mice exhibited significantly increased and reorganized spatial expression of MCM7 (Fig. 4) and increased expression of the antiapoptotic protein survivin (Fig. 5A). The MCPyV-associated papillomas also had markers of an active cellular DNA damage response (Fig. 5B). Interestingly, γ-H2AX and pS1981-ATM positive cells were only observed in the papillomas and not on neighboring skin (Fig. 5B), even though the epithelium in both the papilloma and the skin expresses the MCPyV T antigens (Fig. 4C). The observation of an active DNA damage response in K14Cre-MCPyV168 papillomas in vivo is somewhat contrary to published reports, which indicate that an active DNA damage response is detected in vitro as the result of MCPyV viral DNA replication facilitated by the C-terminus of MCPyV large T antigen (33, 36). This scenario is not possible in our mouse model (Fig. 1A), as the large T antigen species encoded by MCPyV168 produces a C-terminally truncated large T antigen incapable of driving viral DNA replication and does not contain a viral origin of replication. Consequently, the presence of an active DNA damage response implies that functions of the preserved domains present in the tumor-derived MCPyV truncated LT and ST are able to activate the DNA damage response independently of viral replication. Future studies are required to determine these functions and how the cellular microenvironment of MCPyV-associated papillomas may facilitate activation of the DNA damage response. Nonetheless, these observations suggest that expression of the tumor-derived MCPyV T antigens alone is not sufficient to induce phosphorylation of H2AX and ATM in vivo in the stratified epithelium, as this signature was only observed in papillomas and not in neighboring skin despite T antigen expression. The DNA damage response specific to MCPyV-associated tumors may represent an inherent genomic instability or the process of neoplastic progression, and together with deregulation of E2F-responsive genes may create an environment with heightened potential for malignancy.

Mouse models of cancer have contributed greatly to our understanding of cellular and molecular biology. Although there are certain limitations inherent to the species specificity of certain cellular processes, most mouse model studies provide valuable knowledge that informs and directs research on human disease. This study provides evidence that the MCPyV T antigens are tumorigenic when expressed in murine skin, the first example of the in vivo oncogenic potential of the MCPyV T antigens. The flexible design of the mouse model described here will enable study of the tropism of MCPyV and the cell of origin of MCC. Ultimately, future studies using this mouse model will provide insight into MCPyV and the T antigens, the host pathways that they perturb, and MCC oncogenesis and may ultimately provide opportunities to develop therapeutics specific for MCPyV-induced disease.

No potential conflicts of interest were disclosed.

Conception and design: M.E. Spurgeon, J. Cheng, P.F. Lambert, J.A. DeCaprio

Development of methodology: M.E. Spurgeon, J. Cheng, J.A. DeCaprio

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.E. Spurgeon, J. Cheng, R.T. Bronson, J.A. DeCaprio

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.E. Spurgeon, J. Cheng, R.T. Bronson, J.A. DeCaprio

Writing, review, and/or revision of the manuscript: M.E. Spurgeon, J. Cheng, P.F. Lambert, J.A. DeCaprio

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.E. Spurgeon, J. Cheng, J.A. DeCaprio

Study supervision: J. Cheng, P.F. Lambert, J.A. DeCaprio

Other (performed pathology): R.T. Bronson

The authors acknowledge the assistance of Anne Griep, Aayushi Uberoi, and Harlene Edwards (University of Wisconsin) for histopathology analysis; James W. Horner (current address: University of Texas M.D. Anderson Cancer Center, Houston, TX) and the Dana-Farber Cancer Institute Transgenic Core; and Frank Costantini (Columbia University Medical Center, New York, NY) for pBigT and pROSA26-PA plasmids.

This work was supported in part by U.S. Public Health Service grants R01 CA63113, R01 CA173023, and P01 CA050661 (J.A. DeCaprio) and P01 CA022443 (P.F. Lambert). M.E. Spurgeon was supported by the University of Wisconsin Investigative Dermatology Training Program T32AR055893 and the National Institute of Allergy and Infectious Diseases T32 Training Program AI078985.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.