Mesenchymal-epithelial interactions involving epiregulin in tuberous sclerosis complex hamartomas

Abstract

Patients with tuberous sclerosis complex (TSC) develop hamartomas containing biallelic inactivating mutations in either TSC1 or TSC2, resulting in mammalian target of rapamycin (mTOR) activation. Hamartomas overgrow epithelial and mesenchymal cells in TSC skin. The pathogenetic mechanisms for these changes had not been investigated, and the existence or location of cells with biallelic mutations ("two-hit" cells) was unclear. We compared TSC skin hamartomas (angiofibromas and periungual fibromas) with normal-appearing skin of the same patient, and we observed more proliferation and mTOR activation in hamartoma epidermis. Two-hit cells were not detected in the epidermis. Fibroblast-like cells in the dermis, however, exhibited allelic deletion of TSC2, in both touch preparations of fresh tumor samples and cells grown from TSC skin tumors, suggesting that increased epidermal proliferation and mTOR activation were not caused by second-hit mutations in the keratinocytes but by mesenchymal-epithelial interactions. Gene expression arrays, used to identify potential paracrine factors released by mesenchymal cells, revealed more epiregulin mRNA in fibroblast-like angiofibroma and periungual fibroma cells than in fibroblasts from normal-appearing skin of the same patient. Elevation of epiregulin mRNA was confirmed with real-time PCR, and increased amounts of epiregulin protein were demonstrated with immunoprecipitation. Epiregulin stimulated keratinocyte proliferation and phosphorylation of ribosomal protein S6 in vitro. These results suggest that hamartomatous TSC skin tumors are induced by paracrine factors released by two-hit cells in the dermis and that proliferation with mTOR activation of the overlying epidermis is an effect of epiregulin.

Fig. 1.

The epidermis of TSC skin tumors shows increased staining for Ki-67 compared with normal-appearing skin. Sections of patient's normal skin (A and D), angiofibromas (B and C), and periungual fibromas (E and F) from two TSC patients were stained with anti-Ki-67 antibody, with normal mouse IgG1 (C and F) as negative controls for A and B, and D and E, respectively. (Scale bars, 35 μm.) Results were similar with three angiofibromas and five periungual fibromas from five patients.

Fig. 2.

Immunoreactivity for phosphorylated ribosomal protein S6 in the epidermis and dermis of TSC skin tumors but not normal-appearing skin. Sections of normal-appearing skin (A and B) or periungual fibromas (C and D) were incubated with rabbit anti-phospho-S6 (1:200) (A and C) or control rabbit IgG) (B and D) and stained with Vectastain ABC-alkaline phosphatase with Vector red substrate). Patterns of staining were similar in three periungual fibromas and four angiofibromas from four patients.

Fig. 3.

Immunoprecipitation of epiregulin from medium of TSC skin tumor cells. PF cells or fibroblasts from normal-appearing skin (NL) from the same TSC patient were grown in 10% FBS/DMEM without or with 100 ng/ml EGF and labeled with [³⁵S]methionine for 24 h. Proteins precipitated from conditioned medium with anti-epiregulin antibody or normal rabbit IgG were separated by SDS/PAGE to identify a band corresponding to epiregulin secreted by TSC skin tumor cells. Added recombinant epiregulin competed with radiolabeled endogenous epiregulin; absence of the band indicates antibody specificity.

Fig. 4.

Effect of epiregulin on proliferation of human epidermal keratinocytes. Cells were grown and treated with human recombinant epiregulin as described in Materials and Methods. Cell proliferation was quantified by the amount of BrdU incorporation using chemiluminescence ELISA. Data are means ± SD of triplicate experiments. *, P < 0.01 versus control.

Fig. 5.

Effect of epiregulin on phosphorylation of S6 ribosomal protein in keratinocytes. After incubation for 48 h in keratinocyte-SFM (without added EGF and bovine pituitary extract), cells were incubated without or with 100 ng/ml epiregulin for the indicated times (A and C) or without or with epiregulin at the indicated concentrations for 1 h (B and D) before separation of cell proteins by SDS/PAGE and reaction of Western blots with antibodies against S6 and phosphorylated S6. Shown are representative blots (A and B) and corresponding graphs from triplicate experiments (C and D), in which values are expressed as the means ± SD of the ratios of phospho- to total S6 for treated (at each time or concentration) relative to that of control (time = 0 or concentration = 0). *, P < 0.05; **, P < 0.01 vs. control.