KLF4 and PCNA identify stages of tumor initiation in a conditional model of cutaneous squamous epithelial neoplasia

Abstract

KLF4 is induced upon growth-arrest in vitro and during epithelial maturation in vivo, and is essential for proper cell fate specification of post-mitotic cells. In spite of a normal role in post-mitotic cells, expression is upregulated and constitutive in certain tumor types. KLF4 functions as an oncogene in vitro, and enforced expression in basal cells of mouse skin rapidly induces lesions similar to hyperplasia, dysplasia and squamous cell carcinoma (SCC). Here we used conditional expression to characterize early steps in KLF4-mediated tumor initiation. In contrast to SCC-like lesions that result when using a conditional, keratin 14 promoter-dependent strategy, lower conditional expression achieved using a MMTV promoter induced only epidermal cycling within morphologically normal skin, a process we termed occult cell turnover. Surprisingly, KLF4-induced hyperplastic lesions showed increased transgene-derived mRNA and protein in maturing, PCNA-negative cells, a property of endogenous KLF4. In contrast, hyperplastic lesions induced by GLI1, a control, showed uniform transgene expression. In KLF4-induced dysplasia and SCC the complementarity of KLF4 and PCNA was replaced by concordance of the two proteins. These studies show that KLF4 transcripts are normally suppressed in cycling cells in a promoter-independent fashion, consistent with a post-transcriptional control, and reveal loss of this control in the transition from hyperplasia to dysplasia. Like the mouse tumors, human cutaneous SCCs and adjacent dysplasias frequently showed maturation-independence of KLF4, with co-expression of KLF4 and PCNA. A smaller subset of human SCCs showed complementarity of KLF4 and PCNA, similar to hyperplastic mouse skin. The results identify parallels between a mouse model and human primary tumors, and show that successive increases of KLF4 in the nuclei of basal keratinocytes leads to occult cell turnover followed by hyperplasia, dysplasia, and invasive SCC.

Fig. 1

Characterization of anti-KLF4. A, HEK293 cells were transfected with the indicated plasmid and cell extracts were analyzed by immunoblot. Ponceau S stain of the filter demonstrated equal loading (not shown). B, Total RNA was isolated from the indicated cell lines and analyzed by Northern blot (left panel). A β-tubulin probe (TUBB1) served as a control for loading. To detect the endogenous protein, anti-KLF4 (upper panels) or normal rabbit immunoglobulin (lower panels) was applied to MCF10A cells. Bound antibody was detected using a red-fluorescent secondary antibody. Signal corresponding to KLF4 is shown either alone (TRITC) or else superimposed on the cell nuclei (TRITC+DAPI). Scale bars, 50μ. C–D, Tissue sections were stained with anti-KLF4. Bound antibody is indicated as a brown precipitate, and sections were counterstained blue using hematoxylin. C, Normal skin was obtained by reduction mammoplasty (left panel) or by excision of uninvolved skin adjacent to a SCC (right panel). D, Uninvolved adnexal epithelium in a patient with SCC (left panel) or stratified squamous epithelium and tumor cells in a patient with BCC (right panel). Brackets indicate cell layers with more prominent staining, solid arrows denote the dermo-epidermal junction (DEJ), and open arrows indicate the tumor-stromal border. Scale bars, 100μ.

Fig. 2

Low-level expression of KLF4 induces epithelial turnover without overt hyperplasia. Transgenic animals carrying the reverse tetracycline-responsive transcriptional activator (rtTA) under control of the MMTV promoter (i.e., MMTV-rtTA) were crossed with tetracycline response element (TRE)-KLF4 transgenic mice, and bitransgenic offspring were induced with dox (upper panels). Paraffin sections of skin were analyzed by immunostaining. Control animals were administered sucrose-water without dox. As an additional control, animals expressing KLF4 under control of a K14-rtTA transgene were induced and analyzed in parallel (lower panels). Closed arrows indicate the DEJ, and open arrows indicate follicle epithelium. Scale bars, 50μ.

Fig. 3

KLF4 and PCNA identify stages of tumor initiation. K14-rtTA;TRE-KLF4 bitransgenic mice were induced with dox for 14 days. The skin was sampled at multiple sites, fixed in paraffin, and tissue samples showing representative histologies were analyzed by immunostaining. After staining of serial sections with one of three different antibodies, images were recorded at a similar position on each sample. A, Lesion-adjacent skin exhibiting little or no morphologic abnormality. On normal skin analyzed in parallel, K14 staining was limited to basal keratinocytes, as expected (not shown). B, Hyperplasia. C, Dysplasia. D, SCC-like lesion. Panels A and C show adjacent areas of one tissue section, while B and D were sections taken from another mouse. Insets show the cell-type specific staining at increased magnification. Arrows indicate the DEJ, and open arrowheads indicate examples of KLF4-positive, basal keratinocytes.

Fig. 4

TRE-KLF4 and TRE-GLI1 transgene expression in hyperplastic skin. K14-rtTA;TRE-KLF4 (A) or K14-rtTA;TRE-GLI1 (B) bitransgenic mice were induced with dox, and cryosections of hyperplastic skin were analyzed. This 21 day induction was performed independently of the experiment shown in Fig. 3. A, mRNA in situ hybridization analysis using the indicated anti-sense probe. Positive staining is indicated as a red precipitate. Orientation lines correspond to the skin surface (solid lines) or to the DEJ (dashed lines). An H&E-stained cryosection is shown at bottom left. B, Immunostain of GLI1-induced hyperplastic lesions with GLI1 antibody. Normal rabbit immunoglobulin served as a control (Ctl). Staining by anti-GLI1 was low or negative on normal, interfollicular mouse skin (not shown). Arrows indicate the DEJ. Scale bars, 100μ.

Fig. 5

KLF4 expression in human cutaneous SCC. A, Tumors showing different KLF4 staining patterns (see Table 1). B, Comparison of staining in tumors with that of adjacent skin. Arrows indicate the DEJ in sections showing epidermis (left), or else the tumor-stromal interface for sections showing tumor cells (right). Arrowheads indicate maturing cells with expression of KLF4. Scale bars, 100μ.

Fig. 6

Co-localization of KLF4 and PCNA in human tissues. Serial sections of paraffin-embedded samples were analyzed by immunostaining. Identical orientation lines were placed on the matched images. Arrows indicate the DEJ or tumor-stromal borders, and arrowheads indicate examples of PCNA-positive cells. Scale bars, 50μ.