To improve preclinical studies and their translation, patient-derived xenografts (PDXs) are increasingly used. They have human-specific tumor characteristics and reflect intra and inter-tumor heterogeneity. However, the endocrine milieu differs between humans and host mice. In light of sex-specific cancer biology and a rise in endocrine-related cancers there is an urgent need to correctly reflect the hormonal milieu in PDX models. We show that female mice of NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) strain widely used for PDXs has 17-β-estradiol (E2) and testosterone (T) levels comparable to C57Bl6 females but higher progesterone (P4) levels. E2 levels are comparable, T levels are lower and P4 levels higher than those observed in postmenopausal women. Ovariectomy increases T to levels observed in postmenopausal women. Subcutaneous E2 and combined E2/P4 silicon pellets provide NSG females with premenopausal ovarian hormone levels. These procedures humanize the endocrine environment of experimental animals, improving PDX relevance in women’s health-related research.

To improve on the quality of preclinical studies and their clinical translatability, patient-derived xenograft (PDX) models are increasingly used because they reflect inter- and intra-patient heterogeneity as well as human-specific tumor cell characteristics. However, the endocrine milieu of human patients, which affects grafted tumor cells may differ from mice. This is a growing concern as evidence of sex-specific biology in cancer has accumulated and an increase in the incidence of endocrine-related cancers has been observed highlighting the need to correctly reflect the hormonal milieu in PDX models. Here, we address the need to better model different female endocrine milieus in xenograft studies. Using an improved Liquid Chromatography-Mass Spectrometry (LC-MS) protocol for concomitant analysis of four different ovarian steroids in low volume plasma samples, we show that female mice of NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) strain frequently used for xenografts have 17-β-estradiol (E2) and testosterone (T) levels comparable to widely used C57Bl6 strain but higher progesterone (P4) levels. While NSG E2 levels are comparable, T levels are lower and P4 levels higher compared to those observed in menopausal women. Following ovariectomy, T levels increase to those found in postmenopausal women. Subcutaneous implantation of E2 and combined E2 and P4 silicon pellets mimic ovarian hormone levels of premenopausal women in follicular and luteal phase of the menstrual cycle. Thus, straightforward procedures can effectively humanize the endocrine environment of experimental animals and improve physiologic relevance in women’s health-related research.

Estrogen and progesterone receptor (ER, PR) signaling control breast development and impinge on breast carcinogenesis. ER is an established driver of ER + disease but the role of the PR, itself an ER target gene, is debated. We assess the issue in clinically relevant settings by a genetic approach and inject ER + breast cancer cell lines and patient-derived tumor cells to the milk ducts of immunocompromised mice. Such ER + xenografts were exposed to physiologically relevant levels of 17-β-estradiol (E2) and progesterone (P4). We find that independently both premenopausal E2 and P4 levels increase tumor growth and combined treatment enhances metastatic spread. The proliferative responses are patient-specific with MYC and androgen receptor (AR) signatures determining P4 response. PR is required for tumor growth in patient samples and sufficient to drive tumor growth and metastasis in ER signaling ablated tumor cells. Our findings suggest that endocrine therapy may need to be personalized, and that abrogating PR expression can be a therapeutic option.

Hormonal contraception exposes women to synthetic progesterone receptor (PR) agonists, progestins, and transiently increases breast cancer risk. How progesterone and progestins affect the breast epithelium is poorly understood because we lack adequate models to study this. We hypothesized that individual progestins differentially affect breast epithelial cell proliferation and hence breast cancer risk. Using mouse mammary tissue ex vivo, we show that testosterone-related progestins induce the PR target and mediator of PR signaling-induced cell proliferation receptor activator of NF-κB ligand (Rankl), whereas progestins with anti-androgenic properties in reporter assays do not. We develop intraductal xenografts of human breast epithelial cells from 36 women, show they remain hormone-responsive and that progesterone and the androgenic progestins, desogestrel, gestodene, and levonorgestrel, promote proliferation but the anti-androgenic, chlormadinone, and cyproterone acetate, do not. Prolonged exposure to androgenic progestins elicits hyperproliferation with cytologic changes. Androgen receptor inhibition interferes with PR agonist- and levonorgestrel-induced RANKL expression and reduces levonorgestrel-driven cell proliferation. Thus, different progestins have distinct biological activities in the breast epithelium to be considered for more informed choices in hormonal contraception.

Estrogens and progesterone control breast development and carcinogenesis via their cognate receptors expressed in a subset of luminal cells in the mammary epithelium. How they control the extracellular matrix, important to breast physiology and tumorigenesis, remains unclear. Here we report that both hormones induce the secreted protease Adamts18 in myoepithelial cells by controlling Wnt4 expression with consequent paracrine canonical Wnt signaling activation. Adamts18 is required for stem cell activation, has multiple binding partners in the basement membrane and interacts genetically with the basal membrane-specific proteoglycan, Col18a1, pointing to the basement membrane as part of the stem cell niche. In vitro, ADAMTS18 cleaves fibronectin; in vivo, Adamts18 deletion causes increased collagen deposition during puberty, which results in impaired Hippo signaling and reduced Fgfr2 expression both of which control stem cell function. Thus, Adamts18 links luminal hormone receptor signaling to basement membrane remodeling and stem cell activation.

Cancer-associated fibroblasts (CAFs) are important at all tumor stages. CSL/RBPJκ suppresses the gene expression program leading to CAF activation and associated metabolic reprogramming, as well as autophagy. Little is known about CSL protein turnover, especially in the tumor microenvironment. We report that, in human dermal fibroblasts (HDFs), conditions inducing autophagy—often found in tumor stroma—down-regulate CSL protein levels but do not affect its mRNA levels. Genetic or pharmacologic targeting of the autophagic machinery blocks CSL down-modulation. Mechanistically, endogenous CSL associates with the autophagy and signaling adaptor p62/SQSTM1, which is required for CSL down-modulation by autophagy. This is functionally significant, because both CSL and p62 levels are lower in skin cancer-derived CAFs, in which autophagy is increased. Increasing cellular CSL levels stabilizes p62 and down-modulates the autophagic process. We reveal here an autophagy-initiated mechanism for CSL down-modulation, which could be targeted for stroma-focused cancer prevention and treatment.

Crosstalk between the epithelial and stromal compartment is essential to maintain tissue homeostasis. Normal tissue environment can suppress the ability of incipient neoplastic cells to survive and proliferate; as a consequence, tumor cells often acquire the ability to create a permissive microenvironment. So far, genetic and epigenetic changes of the epithelium have been implicated as likely primary determinants of carcinogenesis, while alterations of the underlying mesenchyme have been assumed to be only a reactive consequence of a stromal-epithelial co-evolution process. Increasing amount of evidence points to the ability of the microenvironment to acquire a pro-carcinogenic state independently of the presence of tumor cells. Alteration of the stromal microenvironment, such as the appearance of cancer-associated fibroblasts (CAFs), is emerging as an important event in tumor initiation as well as progression. We previously showed that mesenchymal deletion of CSL, the key effector of “canonical” Notch signaling, endowed with an intrinsic repressive function, results in spontaneous multifocal keratinocyte tumor formation, preceded by CAF phenotype acquisition in both mouse and human stromal fibroblast. Recently we discovered, in the same mouse and human cell system, that CAF activation is accompanied by stromal cell senescence and that the 2 events are genetically linked. More specifically, deletion or silencing of the CSL gene in dermal fibroblasts induces cellular senescence and increased expression of senescence determinants, such as p15, p16, p21 and miR-34a. In parallel, CSL silencing also results in induction of the senescence messaging secretome (SMS) constituents, such as IL6, IL8, Pai-1 and MMPs. We also identified key senescence-effector genes, specifically p16, p21, miR-34a and IL6, together with a broad spectrum of CAF effector genes as direct targets of CSL transcriptional repression. Two CSL binding sites on p21 promoter have adjacent p53 recognition sequences, raising the possibility of direct interaction between the 2 proteins. Indeed, we subsequently uncovered that CSL and p53 can physically interact and that CSL suppresses p53 activity, blocking its association to p300, a critical chromatin modifier and transcriptional activator. Previous work has shown that stromal cell senescence can be associated with induction of SMS and CAF effector genes, yet whether or not the 2 events can be genetically dissociated remained unclear. We therefore set out to test the impact of p53, a well-known determinant of cellular senescence, on CSL dependent senescence and CAF activation. We found that loss of p53 overcame the senescence driven by CSL knockdown. Surprisingly, concomitant CSL and p53 silencing also enhanced the expression of CAF effector genes, like Cox-2 and periostin, known to be involved in inflammation and macrophage recruitment. To assess the in vivo significance of these findings, we developed a mouse ear intradermal injection assay for cancer and stromal cell expansion by in vivo imaging of fluorescently labeled cells. Skin or head and neck squamous cell carcinoma (SCC) cells admixed with fibroblasts deficient for both CSL and p53 showed enhanced outgrowth, leukocyte/ macrophage infiltration and angiogenesis in comparison with CSL alone. In most cases, in situ epithelial lesions do not progress into an invasive cancer and additional genetic changes are needed for progression into a malignancy. The events governing this process are not fully understood, but most likely a tumor-permissive microenvironment is required. Stromal changes associated with cell senescence are thought to play an important role in the increased incidence of age-related cancers. However, this process may be more relevant in the initial phase of cancer development rather than at an advanced stage, the latter being characterized by increased fibroblast density and proliferation. Concurring with this possibility, we found senescent stromal cells in human premalignant skin lesions, actinic keratosis (AK) and in situ SCC, while little or none could be detected in invasive SCCs. Stromal dermal fibroblasts underlying AK, in situ and invasive SCC showed CSL down-modulation, while p53 gene expression was down-modulated only in SCCs. Furthermore, relative to normal HDFs, SCC-derived CAFs showed lower levels of CSL. Importantly, CSL overexpression in these cells resulted in suppression of CAF markers and their ability to promote SCC cell proliferation

Nat. Cell Biol. 17, 1193–1204 (2015); published online 24 August 2015; corrected after print 28 August 2015 An error in the print version of this Article meant that Witold W. Kilarski's name was incorrect. This has been corrected in all online versions.