A subset of patients with NSCLC, HNSCC, mCRC and pancreatic cancer are responding to therapy by several agents directed against the epidermal growth factor receptor (EGFR). Uniformly patients develop a papulopustular follicutis often accompanied by alopecia, xeroderma and changes in nails and eyelashes. To model this skin rash in a mouse model, EGFR was ablated in the epidermis in the litters derived from Keratin 5 driven Cre recombinase transgenic mice crossed with EGFR floxed mice. The skin of double transgenic mice (EGFR KO) reproduced the hallmarks of the skin lesions of patients undergoing chemotherapy with anti-EGFR agents: inflammation, pruritis, dry skin with neutrophilic pustules and infiltration of mast cells, macrophages and lymphocytes. In previous reports we detailed the temporal biochemical changes in chemokines and cytokines that are upregulated in KO skin starting 3 days after birth and the phenotypic consequences of this biochemistry starting 7 days after birth. We also showed similar changes in plasma cytokine/chemokine levels apparently emanating from the skin. A characteristic hematological change was identified with neutrophilia, eosinophilia and reduction in lymphocytes. Cutaneous inflammation preceded changes in the hair follicle phenotype and resulted in I aberrant expression of differentiation markers in the epidermis. The first infiltrating cells are macrophages and mast cells. In an attempt to prevent the skin inflammation, we crossed the KO mice with mice deficient in major pathways leading to an inflammatory reaction. Backcrosses of the epidermal ablated EGFR KO mice with mice genetically ablated for iNOS, TNFR1/R2, MyD88, CCR2 and Rag1 micemice retained the skin phenotype and the increased recruitment of macrophages in the skin. This genetic approach revealed a lack of a primary role for the IL-1/TLR family, CCL2 and T/B cells in the development of the lesions. In a second approach we successfully reduced macrophage infiltration in the skin by subcutaneous injection of clodronate/liposomes. Clodronate treatment partially repaired the architecture of epidermal EGFR KO skin with more parallel growing hair follicles, less dermal cellularity, fewer keratin plugs, and reduced levels of a subset of cytokines, chemokines and chemokine receptors in both FVB/N and C57Bl/6 background. Clodronate treatment also reduced infiltrating mast cells and aberrant expression of several epidermal differentiation markers in the KO skin. In collaboration with Elise Kohn and Seth Steinberg we examined the plasma collected from 10 patients participating in a Gefitinib trial for ovary cancer. Examining a series of changes in circulating cytokines and chemokines before and one month after treatment we determined that inflammatory mediators levels were related to rash grade and pruritus. Most circulating factors examined increased after treatment, but unexpectedly, patients whose pretreatment levels of these factors were the lowest tended to have greater rash and pruritus following treatment. Lower Eotaxin/CCL11 and IL-18 levels before and after treatment with Gefitinib wereas associated with stronger rash and pruritus. Lower levels of IL-1ra and MDC/CCL22 after the treatment were significantly associated with stronger rash while only lower levels of IL-1ra showed a trend to be associated with pruritus. In contrast, higher levels of RANTES/CCL5 before the treatment initiates were associated with pruritus. We measured the blood cell counts of eight patients and analyzed for trends or significant association with the rash grade or the pruritus experience. As in the mouse model, patients exhibited neutrophilia and lymphopenia on treatment. The difference (after treatment minus before treatment) between the percentage of granulocytes (higher) and lymphocytes (lower) in the blood cell counts was significantly associated with rash severity and pruritus. Moreover, larger differences in the absolute number of platelets are significantly associated with pruritus. The mouse studies suggest that macrophages or mast cells may be fundamentally causative in the rash phenotype and human adverse skin response to anti-EGFR drugs may be predicted based on circulating chemokine/cytokine levels before treatment begins. In previous studies we have demonstrated that ablation of MyD88 reduces skin tumor induction in mice. Using cultured keratinocytes transformed by oncogenic Ras, we have mapped the MyD88 dependent pathway and discovered that the activator is IL-1a that is upregulated downstream from activation of EGFR. This leads to NFkB activation, expression of multiple pro-inflammatory factors and both autocrine and paracrine consequences important to tumor induction. Ras transformed MyD88 deficient keratinocytes do not form tumors in orthografts but paradoxically IL-1R deficient keratinocytes do form tumors in orthografts. This suggested that IL-1R may respond differently to tumor promoters, such as the phorbol ester, TPA. Acute TPA response is attenuated in MyD88-/- mice (lower MPO and chemokine expression in the skin) and a similar phenotype is observed in mice with an epidermal specific deletion of MyD88. However IL-1R deficiency (genetic or pharmacological blockade) does not alter the acute response to TPA. This result suggests that MyD88 contribution to acute TPA mediated recruitment of PMNs is mostly independent of IL-1 signaling. Primary keratinocytes deficient for MyD88 failed to upregulate CXCL1 expression when treated with TPA. However, the induction of CXCL1 by TPA in IL-1R deficient keratinocytes is maintained. This provides a mechanism for the difference observed in tumor formation among the two strains in vivo. The epidermal hyperplasia caused by repeated TPA application (4x) is reduced in MyD88 totally deficient mice but is partially maintained in mice with keratinocyte specific deletion of MyD88. This result suggests that a non epidermal population contributes to the hyperproliferation associated with tumor promotion. Another growth factor of interest for skin tumor induction is Hepatocyte Growth Factor (HGF). We have been studying skin tumor formation in mice overexpressing HGF in collaboration with Glenn Merlino. These mice have been crossed to mice overexpressing PKCalpha in skin, a model that leads to induced acute neutrophilic inflammation when activated topically. These are double transgenic mice (DT). DT mice have a greatly increased incidence of skin tumor formation after DMBA and TPA. After 16 weeks of tumor promotion (in the absence of prior DMBA), squamous papillomas appear in 100% of DT mice. No ras mutations have been found in those DT papillomas. These results suggest that aberrant HGF expression can compensate for the lack of initiation, likely through the transactivation of EGFR. It provides a potential mechanism for the development of SCC in the absence of ras mutations. K5-PKCalpha and DT mice are more sensitive to TPA in vivo as evidenced by microabscess formation, chemokines expressions, Cox-2 induction and epidermal hyperplasia. Levels of activated ras (ras-GTP pull down assay) are higher in HGF and DT keratinocytes in the absence of activated ras when compared to WT or K5-PKCa keratinocytes. All EGFR ligands expressed by keratinocytes are upregulated in HGF and DT primary keratinocytes in the absence of ras transduction. Treatment with an EGFR inhibitor (PD168393) attenuates the baseline Ras-like phenotype of HGF and DT keratinocytes. Together these results suggest that high expression of HGF in the tumor environment could substitute for ras activation.
Showing the most recent 10 out of 21 publications
