Prevention of KLF4-mediated tumor initiation and malignant transformation by UAB30 rexinoid

Abstract

The transcription factor KLF4 acts in post-mitotic epithelial cells to promote differentiation and functions in a context-dependent fashion as an oncogene. In the skin KLF4 is co-expressed with the nuclear receptors RARgamma and RXRalpha, and formation of the skin permeability barrier is a shared function of these three proteins. We utilized a KLF4-transgenic mouse model of skin cancer in combination with cultured epithelial cells to examine functional interactions between KLF4 and retinoic acid receptors. In cultured cells, activation of a conditional, KLF4-estrogen receptor fusion protein by 4-hydroxytamoxifen resulted in rapid upregulation of transcripts for nuclear receptors including RARgamma and RXRalpha. We tested retinoids in epithelial cell transformation assays, including an RAR-selective agonist (all-trans RA), an RXR-selective agonist (9-cis UAB30, rexinoid), and a pan agonist (9-cis RA). Unlike for several other genes, transformation by KLF4 was inhibited by each retinoid, implicating distinct nuclear receptor heterodimers as modulators of KLF4 transforming activity. When RXRalpha expression was suppressed by RNAi in cultured cells, transformation was promoted and the inhibitory effect of 9-cis UAB30 was attenuated. Similarly as shown for other mouse models of skin cancer, rexinoid prevented skin tumor initiation resulting from induction of KLF4 in basal keratinocytes. Rexinoid permitted KLF4 expression and KLF4-induced cell cycling, but attenuated the KLF4-induced misexpression of cytokeratin 1 in basal cells. Neoplastic lesions including hyperplasia, dysplasia and squamous cell carcinoma-like lesions were prevented for up to 30 days. Taken together, the results identify retinoid receptors including RXRalpha as ligand-dependent inhibitors of KLF4-mediated transformation or tumorigenesis.

Figure 1

The RXR-selective retinoid 9cUAB30 inhibits KLF4-mediated in vitro transformation. (A) Retroviral supernatants were generated using the expression vectors indicated on the left and applied to RK3E cells. 9cUAB30 or vehicle (DMSO) was added to the culture media every two days. For dishes transduced with KLF4-ER, 4OHT was added to the culture media. Transformed foci were quantified by staining and counting of foci. Columns, mean (n = 4); bars, SE; p value, unpaired t test (two-tailed). (B) Dose response analysis (upper) and analysis of functionally distinct retinoids (lower and see Suppl. Fig. 1). KLF4-transformed foci were generated and quantified as described in (A), with addition of retinoid or DMSO to the culture media. Columns, mean (n = 2); bars, range. (C) Receptor selectivity of retinoid agonists. HEK293 cells were transfected with expression vectors encoding the indicated Gal4 fusion proteins, a luciferase reporter plasmid containing Gal4 binding sites, and an internal control. Following treatment with the indicated agonist for 24 h, normalized firefly luciferase activity was determined. Columns, mean (n = 4); bars, SE. (D) Immunoblot analysis of KLF4-ER and KLF4-regulated genes. RK3E cells were stably transduced with KLF4-ER retrovirus, encoding a 4OHT-dependent allele. Cells were treated with 4OHT + DMSO or with 4OHT + 9cUAB30 for the indicated interval. β-actin served as a control for loading.

Figure 2

(A) Skin tumor initiation by wild type human KLF4 is inhibited by 9cUAB30. Dox-inducible KLF4 transgenic mice were fed chow admixed with 9cUAB30 or placebo. Following dox administration for 30 days, the skin phenotype was determined by gross examination and by histologic analysis of fixed tissues. Penetrance of the skin phenotype is defined as any observed abnormality (gross or microscopic). (B) Analysis of 9cUAB30 effects on skin tumor initiation/progression. Mice (N = 2 per treatment group) were treated for 7 days as indicated and skin samples were analyzed by microscopic examination and immunostaining. Asterisks indicate SCC-like cells within the dermis. Arrowheads indicate the DEJ. Insets show a portion of the section at a higher magnification. Scale bars, 100 μ.

Figure 3

Reprogramming of basal keratinocytes in TO-KLF4 transgenic mice is inhibited by 9cUAB30. (A) Immunostaining analysis of KLF4 transgenic mouse skin. Mice were treated with Dox for 48 h. Insets show a portion of the section at a higher magnification. (B) K1 antibody stained the outer root sheath cells following induction of KLF4, but did not stain these cells in samples from control mice (not shown). (C) Mice were treated as indicated for 7 days, and expression of K1 in basal keratinocytes of the interfollicular skin was examined by immunostaining. Arrows indicate the DEJ. Scale bars, 50 μ.

Figure 4

Regulation of NR gene expression by KLF4. (A) RK3E epithelial cells stably transduced with a KLF4-ER expression vector were treated with 4OHT for two hours. RNA was isolated and assayed by microarray. Fold changes were calculated using DMSO-treated KLF4-ER cells as the control. (B) KLF4-ER cells or Vector-transduced RK3E cells were treated with 4OHT as indicated, and RXRα was analyzed by immunoblot. (C) Cultures were treated as indicated, and mRNA expression was analyzed by sqRT-PCR. Gapd served as a loading control.

Figure 5

RXRα knockdown attenuates the rexinoid inhibitory effect on KLF4. (A) Lentiviral vectors encoding shRNAs to RXRα were transduced into RK3E cells, and stable populations were selected using puromycin. Cell extracts were isolated at 96 h post-transduction and RXRα was analyzed by immunoblot. β-actin served as a loading control. (B) KLF4 transforming activity was assessed in knockdown cells and control cells in the presence or absence of 9cUAB30 using similar methods as in Figure 1A. (C) Quantitation of KLF4 transforming activity in the presence or absence of 9cUAB30. Each lentiviral construct was introduced into RK3E cells in two independent transduction experiments. Columns, mean (n = 2); bars, range. (D) The data shown in (C) was used to calculate the relative KLF4 transformation efficiency (i.e., the ratio of transformed foci observed in dishes treated with either 9cUAB30 or DMSO). Columns, mean (n = 4); bars, SE; p value, one way ANOVA with Tukey's Multiple Comparison Test, comparing each shRNA-expressing cell line to Vector control cells).