Medulloblastoma (MB) is the most common pediatric tumor of the CNS, representing ∼20% of all childhood CNS tumors. Although in recent years many molecular mechanisms that control MB development have been clarified, the effects of biological factors such as sex on this tumor remain to be explained. Epidemiological data, in fact, indicate a significant difference in the incidence of MB between the 2 sexes, with considerably higher susceptibility of males than females. Besides this different susceptibility, female sex is also a significant favorable prognostic factor in MB, with girls having a much better outcome. Despite these literature data, there has been little investigation into estrogen influence on MB development. In our study, we evaluated how hormone deficiency resulting from ovariectomy and hormone replacement influences the development of early and advanced MB stages in Patched1 heterozygous mice, a well-characterized mouse model of radiation-induced MB. Susceptibility to MB development was significantly increased in ovariectomized Ptch1+/− females and restored to levels observed in control mice after estrogen replacement. We next investigated the molecular mechanisms by which estrogen might influence tumor progression and show that ERβ, but not ERα, is involved in modulation of MB development by estrogens. Finally, our study shows that a functional interaction between estrogen- and IGF-I-mediated pathways may be responsible for the effects observed.
Brain tumors are the leading cause of cancer-related mortality in children, with medulloblastoma (MB), a highly malignant primitive neuroectodermal tumor of the cerebellum, representing about 20% of all childhood primary CNS tumors. MB peak incidence is around 7 years of age, whereas in adults the disease is much less common, comprising only 1% of primary brain tumors.1, 2 This cerebellar neoplasm typically arises in the midline vermis and often invades and obliterates the fourth ventricle; an insidious feature of MB is its propensity to metastasize and disseminate through the subarachnoid space, with ∼30% of children demonstrating cerebrospinal fluid (CSF) metastasis at diagnosis.3 MB patients are currently stratified into 2 groups, standard and high risk, on the basis of clinical parameters, such as extent of tumor at the time of diagnosis and completeness of surgical resection. High-risk patients are those <3 years of age or having a residual tumor mass after surgery ≥1.5 cm2 or metastatic disease at diagnosis.4 Currently, multimodality treatment, including surgery, radiotherapy and chemotherapy, is considered the most effective strategy against MB, and this combined approach has indeed significantly improved the outcome: 5-year disease-free survival rates of 80% or more are now being reported by multiple groups for patients with standard-risk MB, whereas the 5-year event-free survival rates for high-risk disease are 60%.5, 6 However, these remarkable improvements in survival are achieved at a high cost to quality of life, with many survivors experiencing significant long-term neurocognitive and neuroendocrine effects.7 Thus, new treatment options, more effective and less toxic, are urgently needed. In this regard, a rapidly developing field, which offers hope for the future, is represented by the possibility to incorporate into chemotherapy strategies agents targeting specific oncogenic signaling pathways involved in MB formation, such as Sonic Hedgehog (SHH), WNT and ERBB.4 Another possible actor, and thus a possible target in this complex scenario, may be represented by the estrogen-receptor signaling pathway, which appears to play an important role in MB. In fact, as shown by epidemiological studies, male sex is a risk factor for MB,8 with ∼65% of patients being males. Notably, the predominance of males over females persists among white and black children9 and in the adult population.10 Besides this different susceptibility, other authors found that female sex is also a significant favorable prognostic factor in MB, with girls having a much better outcome.11–13 Curran et al.13 actually reported that females with MB have a survival advantage only in older children (>3 years) and in adults, suggesting the existence of an interaction between sex and age in regards to outcome, possibly because of hormonal differences, among other factors. Indeed, circulating estrogen levels progressively increase in girls from infancy to late prepuberty,14 and thus significant differences between females and males only exist in children of 4 years or older, whereas under 3 years of age, mean serum estradiol levels are similar between the sexes.15
Germline heterozygous inactivation of PTCH predisposes to MB, in line with the PTCH protein role as a receptor and negative regulator for the secreted protein ligand SHH, which is involved in many signaling processes that control cell fate and growth in the developing cerebellum. Ptch1+/− mice carry a germline mutation of Ptch1 that leads to development of MB in a low percentage of mice on CD1 genetic background by 20 weeks of age, a process that can be strongly promoted by neonatal irradiation.16 In addition, preneoplastic MB stages consisting of hyperplastic cerebellar lesions of increasing degree of altered cellular morphology and size are present at high incidence in the cerebellum of asymptomatic mice.17
To gain further insights into the role of estrogen signaling in MB tumorigenesis, we have studied how hormone deficiency resulting from ovariectomy and hormone replacement influences the development of early and advanced MB stages in Ptch1+/− mice. Our results suggest that estrogens protect against Ptch1-associated MB formation, and that a functional interaction between estrogen- and IGF-I-mediated pathways is likely to play a role in the effects observed.
Material and Methods
Mice lacking one Ptch1 allele (Ptch1neo67/+, named Ptch1+/− throughout the text), derived by gene targeting of 129/Sv ES cells and maintained on outbred CD1 background,18 were housed in the animal facility at ENEA-Casaccia (Rome, Italy) and genotyped as described.19 All experimental protocols were reviewed by our Institutional Animal Care and Use Committee.
All Ptch1+/− females involved in our study (n = 174) were whole-body irradiated at postnatal day 2 (P2) with a single 3-Gy dose of X-rays. Irradiation was performed using a Gilardoni CHF 320G X-ray generator (Gilardoni S.p.A., Mandello del Lario, Italy; HVL = 1.6 mmCu) operated at 250 kVp, 15 mA, with filters of 2.0 mm Al and 0.5 mm Cu. At 4 weeks of age, a subgroup of mice (n = 100) were anesthetized (65 mg/kg sodium pentobarbital i.p.) and bilaterally ovariectomized (OVX); the success of ovariectomy was checked at necropsy by marked atrophy of the uterine horns. One week after ovariectomy, a group of OVX mice (OVX + E2; n = 37) were implanted subcutaneously with 90-day release, 0.36 mg 17β-estradiol pellets (Innovative Research of America, Sarasota, FL); these pellets are designed to produce 100–125 pg/ml of serum estradiol (as indicated by the supplier). The remaining OVX females were left untreated (n = 63). Control mice (CN, n = 74) were sham operated.
To evaluate full MB development, animals (CN, n = 36; OVX, n = 34; OVX + E2, n = 25) were observed daily for their whole life span; upon health decline (i.e., severe weight loss, paralysis, ruffling of fur and inactivity), they were euthanized and autopsied. Normally appearing and tumor-bearing brains were fixed in 4% buffered formalin for histology or stored at −80°C for immunoblot analysis.
The remaining mice (CN, n = 38; OVX, n = 29; OVX + E2, n = 12) were euthanized at different times, i.e., 4 (baseline), 6 and 8 weeks of age for evaluation of preneoplastic lesions. Six sections, recovered with intervals of 70 μm to ensure representative sampling, were examined for each cerebellum.
The number of animals allocated to each experimental group at the beginning of the study is summarized in Table 1.
Table 1. Experimental groups
Immunohistochemical analysis was carried out on 3-μm-thick paraffin sections as described.20 Antibodies used include: anti-ERα (polyclonal, 1:500; Santa Cruz Biotechnology, Santa Cruz, CA), anti-ERβ (polyclonal, 1:50; Santa Cruz Biotechnology), anti-ERβ (polyclonal, 1:500; Upstate, Biotechnology, Lake Placid, NY), anti-ERβ (polyclonal, 1:50; Abcam, Cambridge, UK), anti-ERβ (polyclonal, 1:50; Affinity BioReagents, Golden, CO), anti-phospho-IGF-I Receptor β (Tyr1131) (polyclonal, 1:50; Cell Signaling Technology, Danvers, MA), anti-phospho-p44/42 MAPK (polyclonal, 1:50; Cell Signaling Technology) and anti-Ki67 (polyclonal, 1:800; Novocastra, Novocastra Laboratories, Newcastle, UK). Immunohistochemical scoring was carried out by investigators blinded to treatment groups. The number of positive (brown stained) cells was determined as the percentage of the total number of cells counted in each different preneoplastic cerebellar lesion per study group (i.e., CN, OVX and OVX + E2).
Samples (30-μg aliquots as determined by the Bradford assay, Bio-Rad Laboratories, Hercules, CA) were separated on SDS-PAGE and transferred to nitrocellulose membranes Hybond P+ (Amersham Biosciences, Bucks, UK). Blots were developed using horseradish peroxidase-conjugated secondary antibodies (Bio-Rad Laboratories). Proteins were visualized by chemiluminescence detection (SuperSignal West Pico Chemiluminescent Substrate, Pierce Biotechnology, Rockford, IL). Protein levels were quantified by densitometric analysis using the Scion Image Beta 4.02 software package (Scion Corporation, Frederick, MD). Filters were stripped and reprobed with anti-β-Actin antibody (monoclonal, 1:5,000; Sigma Chemical, St. Louis, MO) for normalization. Antibodies used include anti-ERα (polyclonal, 1:500; Santa Cruz Biotechnology), anti-ERβ (polyclonal, 1:500; Santa Cruz Biotechnology) and anti N-Myc (monoclonal, 1:1,000; Calbiochem, San Diego, CA).
RNA extraction and reverse transcription-PCR
Total RNA was isolated from tumors of CN (n=3) and OVX (n = 3) Ptch1+/− mice using SV Total RNA isolation system (Promega Corporation, Madison, WI) and stored at −80°C until further processing. Total RNA (2 μg) was reverse-transcribed using RETROscript™ (Ambion, Austin, TX) according to the manufacturer's instructions and amplified for Gli1, Gli2 and β-Actin, as described.20 Scanning of the ethidium bromide gels was done with Electrophoresis Documentation and Analysis System (EDAS 290, Eastman Kodak, Rochester, NY), and data were analyzed for mass and molecular weight using the 1D automatic lane and band finder (Image Analysis software, Eastman Kodak). For each sample, 3 different PCR amplifications were performed, and means ± SE were calculated.
Analyses were performed using GraphPad Prism version 4.02 for Windows (GraphPad Software, San Diego, CA). We used Mann-Whitney test for comparison of tumor multiplicity and Fisher's exact test for analysis of tumor incidence. p values are for 2-sided tests; p values ≤ 0.05 were considered statistically significant.
Effect of estrogen modulation on early and advanced MB stages
To assess the role of estrogen deprivation on preneoplastic cerebellar lesions and MB development, we irradiated Ptch1+/− female mice at P2 and performed ovariectomy at 4 weeks of age. Treatment with 17β-estradiol started 1 week after ovariectomy in the relevant group (OVX + E2). Cerebella of OVX Ptch1+/− mice and intact littermates were examined for the presence of preneoplastic lesions or full tumors. At 4 weeks of age (baseline), mice show no symptoms of MB, but a large percentage of them (75%) shows cerebellar abnormalities that may represent a pool of partially transformed cells on their way to becoming tumors (Fig. 1a). At 6 weeks of age, 89% (8/9) of intact mice were found to have preneoplastic alterations compared to 82% (9/11) of OVX mice; at this time point, OVX + E2 mice were not sacrificed because of the short treatment period (Fig. 1a). At 8 weeks of age, the incidence of cerebellar abnormalities in OVX Ptch1+/− females remained essentially unchanged (83%; 15/18), in contrast with intact mice in which the incidence was 59% (10/17), showing that one-third of preneoplastic lesions undergo regression in mice with intact ovarian function in a 2-week period. In OVX + E2 mice, the incidence of preneoplastic lesions was 58% (7/12), 1.4-fold lower than untreated OVX mice and very similar to intact CN mice (Fig. 1a). Notably, the final incidence of MB in the lifetime study reflected the trend observed for preneoplastic lesions at 8 weeks of age, with an MB incidence of 85% (29/34) in OVX compared to 58% (21/36) in intact mice (p = 0.0173) and with 56% (14/25) in OVX + E2 mice (p = 0.0181 vs. OVX) (Fig. 1b). Overall, these data strongly support a role for estrogens in modulating Ptch1-associated MB.
To assess whether the lower growth potential of putative MB precursor lesions in mice with intact ovarian function was accompanied by a decrease in proliferation, we compared the proliferation rate of preneoplastic lesions at 6 weeks in OVX and CN mice (Fig. 1c). Immunohistochemistry was carried out on cerebellum sections using an antibody directed against the proliferation marker Ki67. The mean values of positive cells were 56.3% ± 2.7% and 42.8% ± 2.7% in the OVX and CN groups, respectively, showing a significantly higher proliferative potential of preneoplastic lesions from OVX mice (p = 0.0049). This suggests that estrogens may inhibit MB development by suppressing proliferation of early preneoplastic areas in Ptch1+/− cerebellum.
Effects of estrogen modulation on ERs expression
Previous studies suggested that ERs play a role in the maintenance of cerebellar function; specifically, ERβ has been found to be strongly expressed during cerebellum development, in particular in the external granule layer,21 where cells of origin of MB, the cerebellar granule cell progenitors, proliferate actively during the first postnatal weeks.
We sought to characterize ERs immunoreactivity in MBs from the different experimental groups. We used Western blotting and immunohistochemistry in MB samples for evaluation of ERα expression. As shown in Figure 2, no detectable ERα expression was observed by immunoblotting (Fig. 2a) or immunostaining (Figs. 2b and 2c) in tumors from the different experimental groups. In fact, only Purkinje cells showed both cytoplasmatic and nuclear ERα immunoreactivity, with no difference in the expression level among CN, OVX and OVX + E2 mice. This suggests that this receptor is not involved in modulation of MB development by estrogen.
Next, we performed evaluation of ERβ expression in MB samples. We used 4 different antibodies subjected to several antigen retrievals for immunohistochemical detection of ERβ (see Material and Methods); however, none of them was able to function properly. Thus, we quantified ERβ protein levels by Western blotting in tumors from CN, OVX and OVX + E2 groups (Figs. 2e and 2f). We found that ERβ expression was 25% lower in OVX compared to CN tumors (p = 0.05; Fig. 2f). Importantly, in OVX mice receiving estrogen replacement, ERβ expression in MBs was comparable to the levels observed in tumors from intact mice. Thus, decrease of ERβ expression and tumor progression seems to correlate in MBs.
Effects of estrogen modulation on IGF-I signaling
Several points of convergence exist between estrogen and insulin-like growth factor-I (IGF-I) signaling in the brain,22 and increased IGF pathway activity has been observed in human MB.23 We analyzed the expression of phosphorylated (P) IGF-IR, ERK1/2 and N-Myc induction in MBs from the various groups. As shown in Figures 3a–3c, membrane immunostaining with anti-P-IGF-IR was found in all tumors examined; however, a more intense immunoreactivity was evident in tumors from OVX (Fig. 3b) compared to CN or OVX + E2 groups (Figs. 3a and 3c).
Next, we performed immunohistochemical analysis with an antibody directed against P-ERK-1/2 on preneoplastic lesions from 8-week old mice and full-blown tumors (Figs. 3d–3i). There was intense P-ERK-1/2 staining in preneoplastic areas from OVX mice (Fig. 3e), compared to very scarce positivity in analogous lesions from intact mice (Fig. 3d) or OVX mice receiving estrogen replacement (Fig. 3f). Finally, activation of the ERK-1/2 pathway in fully developed tumors was highly consistent with the pattern of precursor lesions, with marked nuclear signal in MBs from OVX (Fig. 3h) compared to weak and diffuse staining in those from CN and OVX + E2 mice (Figs. 3g and 3i).
Western blot analysis was next performed to evaluate N-Myc expression levels (Figs. 3j and 3k). For both N-Myc isoforms, we found significantly higher levels of expression in tumors from OVX compared to tumors from CN mice (67 KDa: p = 0.0138; 64 KDa: p = 0.0359). In mice receiving estrogen replacement, protein expression in tumors was very similar to tumors from CN mice, while both isoforms were statistically significantly different between tumors from the OVX and OVX + E2 groups (67 kDa: p = 0.0033; 64 kDa: p = 0.0235).
The N-Myc protein appears to be activated by several mitogenic signals, including Shh.24 Therefore, to assess whether the increased expression of N-Myc and the increase of MB incidence in OVX animals were the result of differential deregulation of the Shh pathway, we quantified by reverse transcription (RT)-PCR mRNA expression levels of downstream Shh-responsive genes, Gli1 and Gli2, in tumors from OVX and CN mice (n = 3 per group). In all MBs examined, a higher expression of Gli1 was observed compared to Gli2, but no significant difference in Gli1 or Gli2 levels was detected in OVX vs. CN mice, with a constant Gli1:Gli2 ratio of 2.1, indicating that estrogen deprivation does not interfere with Shh signaling in cerebellar tumors (Fig. 4).
In our study, we explored the effects of estrogen deficiency and replacement on the progression of early and advanced MB stages in a well-characterized mouse model of MB. We show that tumor formation is significantly accelerated in female Ptch1+/− mice after ovarian hormone withdrawal, and that a higher level of cell proliferation is present in preneoplastic cerebellar lesions from OVX mice, suggesting a relationship between increased tumorigenesis and higher proliferative potential. Importantly, we show that both tumorigenesis and cell proliferation are inhibited by 17β-estradiol, a finding that strongly supports a protective role of female sex hormones against MB. Overall, data from our study reflect the clear sex differences in MB incidence observed in both children and adults.8
The mechanisms for a potential influence of sex on the biology of MB are not known, and to shed light on this specific matter, we examined ERs protein levels in MBs from our different treatment groups. Immunoanalysis revealed undetectable levels of ERα in all tumors from Ptch1+/− females, regardless of hormone status. In contrast, ERβ was expressed in all tumors examined, in keeping with studies showing that ERβ is a dominant ER subtype in the adult rodent cerebellum.25–27 In addition, we show here that ERβ protein levels decreased significantly in tumors from OVX mice, an effect reverted by estrogen replacement that restored ERβ expression to CN levels. Notably, Urbanska et al.28 recently reported that well-differentiated desmoplastic and neuroblastic human MBs tend to have higher levels of nuclear ERβ when compared to poorly differentiated MBs. Moreover, compared to normal tissues, a decreased ERβ expression has also been observed in other cancers, such as breast,29 prostate,30 ovary,31 colon and rectum32; in squamous skin papillomas, a lower expression was noticed in tumors with a higher proliferative rate,33 and in cutaneous melanoma lower ERβ protein levels were observed in thicker, more invasive tumors.34 However, what mechanisms account for the decreased expression of ERβ in tumors and whether a mechanistic connection exists between ERβ dysfunction—especially loss of expression—and tumor progression remain open questions. Indeed, ERα has been commonly associated with stimulation of growth, whereas ERβ is usually linked with growth suppression or inhibition of cell proliferation.35
It is known that multiple signaling pathways are associated with MB formation and growth.36, 37 Among these, the IGFs and their cognate receptors (IGFRs) have been shown to cooperate with Shh in MB formation, with studies showing that the combined activation of the Shh/Ptch1 and IGF-I signaling pathways represents an important mechanism in MB pathogenesis.38 In fact, IGF-IR, its phosphorylated active form, and the substrate for the active receptor insulin receptor substrate-1 were found in a majority of MB surgical specimens, although their expression levels did not correlate with a specific tumor subtype.39 Moreover, various components of the IGF system were found to be elevated in tumor tissue, CSF and peripheral blood of pediatric MB patients.40 Remarkably, evidence has also accumulated over the past years indicating a close interdependence between the actions of IGF-I and estradiol in the brain.41 Indeed, there is indication that the expression of ERs and IGF-IR is crossregulated in the brain, and, interestingly, this regulation in the cerebellum appears different from other areas of the brain. Specifically, it has been shown that ERβ may downregulate IGF-IR expression in the cerebellum.21 In keeping with these hypotheses, we show that the withdrawal of estrogen, with the concomitant reduction in ERβ level observed in OVX-MBs, is associated with P-IGF-IR upregulation. This latter event, in turn, triggers the IGF-I pathway, with activation of key signal transduction pathways downstream of IGF-IR, such as the cascade of ERK/MAPKs and the inhibition of phosphorylation and degradation of N-Myc through the PI3K/Akt signaling pathway. The consequent increase of both cell proliferation and survival may explain the higher proliferation rate in preneoplastic lesions and the increased MB incidence observed in OVX mice in our study. In addition, our results show that Gli1 and Gli2 expression is similar in tumors from CN and OVX groups, suggesting that estrogen signaling does not interfere with Shh signaling in cerebellar tumors. Thus, ovariectomy may promote MB development with a mechanism that converges on activation of IGF-I-mediated pathways.
A discrepancy exists between our data and results from a recent study by Belcher et al.42 reporting that estrogens have a positive effect on the development of MB, because 17β-estradiol stimulated tumor growth in OVX mice xenotransplanted with the ERβ-positive MB D283Med cells. At present, no data are available that could help explaining this divergence, although a possible interpretation resides in the use of different experimental protocols/mouse models. To the best of our knowledge, there are no other experimental studies addressing this specific issue; thus, further research is clearly needed. Notably, a protective role of estrogen has also been underlined for other brain tumors. Indeed, Plunkett et al.43 reported a survival advantage for females over males in a rat model of glioblastoma, with ovariectomy negating this gain and estrogen replacement restoring survival advantage. Moreover, in a subsequent study, they also demonstrated a therapeutic effect of an estradiol-based therapy for glioblastoma.44
In summary, our study has shown for the first time a protective role of endogenous estrogen against MB in an accurate mouse model of the human disease. Our study also suggests that ERβ, but not ERα, is involved in modulation of MB development by estrogens. Finally, the study shows that a functional interaction between estrogen- and IGF-I-mediated pathways may be responsible of the observed effects. Our results are in keeping with the epidemiological data showing clear sex differences in MB incidence and outcome and strongly support the need to investigate new therapeutic approaches in the treatment of this highly malignant disease by manipulation of estrogen signaling.
The authors thank Prof. Maria Marino for useful comment on the manuscript and Mr. Orsio Allegrucci and Mrs. Angela Sandri for assistance in handling and treating the mice.