|Home | About | Journals | Submit | Contact Us | Français|
Chemoresistance in neuroblastoma is a significant issue complicating treatment of this common pediatric solid tumor. MYCN-amplified neuroblastomas are infrequently mutated at p53 and are chemosensitive at diagnosis but acquire p53 mutations and chemoresistance with relapse. Paradoxically, Myc-driven transformation is thought to require apoptotic blockade. We used the TH-MYCN transgenic murine model to examine the role of p53-driven apoptosis on neuroblastoma tumorigenesis and the response to chemotherapy. Tumors formed with high penetrance and low latency in p53-haploinsufficient TH-MYCN mice. Cyclophosphamide (CPM) induced a complete remission in p53 wild type TH-MYCN tumors, mirroring the sensitivity of childhood neuroblastoma to this agent. Treated tumors showed a prominent proliferation block, induction of p53 protein, and massive apoptosis proceeding through induction of the Bcl-2 homology domain-3-only proteins PUMA and Bim, leading to the activation of Bax and cleavage of caspase-3 and -9. Apoptosis induced by CPM was reduced in p53-haploinsufficient tumors. Treatment of MYCN-expressing human neuroblastoma cell lines with CPM induced apoptosis that was suppressible by siRNA to p53. Taken together, the results indicate that the p53 pathway plays a significant role in opposing MYCN-driven oncogenesis in a mouse model of neuroblastoma and that basal inactivation of the pathway is achieved in progressing tumors. This, in part, explains the striking sensitivity of such tumors to chemotoxic agents that induce p53-dependent apoptosis and is consistent with clinical observations that therapy-associated mutations in p53 are a likely contributor to the biology of tumors at relapse and secondarily mediate resistance to therapy.
Neuroblastoma is the most common extracranial solid tumor of childhood . The proto-oncogene MYCN is amplified in one third of neuroblastomas and is associated with high-risk disease and poor survival [1–5]. The ability of Myc proteins to induce malignant transformation is limited by coordinate induction of both proliferation and apoptosis [6–8]. Consequently, mutations of p53 are common in several Myc-driven malignancies at diagnosis and enhance oncogenesis.
In neuroblastoma, genetic defects in the p53 pathway function (mutation in p53 or p14ARF, amplification of MDM2) are rare at diagnosis but are present in cell lines established from patients treated with prolonged therapy [9–12]. In the absence of such mutations, amplification of MYCN in newly diagnosed neuroblastomas is typically associated with epigenetic abnormalities that impair apoptosis: silencing of caspase-8 or overexpression of the antiapoptotic proteins Bcl-2 and survivin [13–15]. These events are independently associated with poor outcome in neuroblastoma and represent potential mechanisms contributing to malignant progression [13,15,16].
Despite such defects in apoptotic signaling, newly diagnosed neuroblastoma tumors typically respond to chemotherapeutics such as cyclophosphamide (CPM) irrespective of risk group, suggesting that epigenetic regulation of apoptotic pathways is either insufficient to block apoptosis in response to cytotoxic chemotherapy or is not initially present at diagnosis [17,18]. Because virtually all patients respond to initial therapy, outcome in this disease is mainly determined by the likelihood of relapse, with high-risk patients (including tumors that show amplification of MYCN) frequently relapsing with drug-resistant tumors [10,19].
The TH-MYCN model shows that coordinate mutation of p53 is not required for MYCN-driven oncogenesis in neuroblastoma, mirroring the situation in human high-risk tumors. It follows that apoptosis is functionally suppressed by other means. However, this presents a paradox in that MYCN-driven murine neuroblastomas retain exquisite sensitivity to chemotherapeutics, which exert their effects through induction of apoptosis. To examine this question, we sought to determine whether the p53 pathway was active or inactive in TH-MYCN and whether induction of the pathway occurred in response to chemotherapy.
TH-MYCN is a native neuroblastoma model in which tumorigenesis is driven by targeted expression of a TH-MYCN transgene in the neural crest of transgenic mice . These animals develop an aggressive malignancy that is morphologically, genetically, and clinically similar to human high-risk MYCN-amplified neuroblastoma [21,22]. We show that p53 haploinsufficiency enhances the effect of MYCN-driven transformation in neuroblastoma, that basal p53 pathway function and apoptosis are suppressed in such tumors (in contrast to in vitro observations), that haploinsufficiency amplifies the effect of MYCN on transformation in neuroblastoma, and that the chemoresponse of effective agents such as CPM requires intact p53-driven apoptosis. Mechanistically, CPM induces p53 and activates a variety of p53 targets critical to apoptosis, including PUMA, Bim, Bax, caspase-3, caspase-9, and poly(ADP-ribose) polymerase (PARP). The in vivo response of TH-MYCN-driven tumors to CPM is similar to that observed in cultured human cell lines treated with CPM. In three cell lines with equivalent levels of p53 protein but increasing levels of Mycn, the apoptotic response to CPM correlated with the induction of p53 and could be blocked by siRNA to p53.
Taken together, these data document the activity in murine tumors of a frontline chemotherapeutic agent used in children with newly diagnosed neuroblastoma and add to evidence validating mice transgenic for TH-MYCN as a preclinical platform for developmental therapeutics. Furthermore, these observations highlight the utility of mice transgenic for TH-MYCN as a preclinical model in which to delineate the kinetics and details of apoptotic signaling in response to CPM. The induction and functional activation of p53 in these tumors reinforce the importance of p53 signaling in the therapeutic response to cytotoxic chemotherapy. These observations are also consistent with clinical data that cytotoxic chemotherapy selects for loss of p53 function, that recurrence in neuroblastoma is subsequently driven by p53 mutation, and that p53 mutation in relapsed tumors contributes secondarily to therapeutic resistance [10,23,24].
SHEP, SK-N-SH, SH-SY5Y, and Kelly neuroblastoma cell lines were obtained from the University of California at San Francisco. Cells were grown in Dulbecco's modified Eagle's medium with 10% FBS at 37°C. 4-Hydroxyperocyclophosphamide (4OH-CPM) was obtained from Susan Ludeman (University of North Carolina) and was dissolved at 20 mg/ml in water. For in vivo therapy, CPM from the University of California at San Francisco Clinical Pharmacy was dissolved at 1 mg/ml.
Cell growth and proliferation were assessed by soluble tetrazolium (WST-1; Roche, Indianapolis, IN). Necrosis and apoptosis were assessed using a Cell Death Detection ELISA (Roche) that detects intracellular or released nucleosomal H2b DNA. Values were corrected for vehicle (DMSO). Cells were plated and assayed at 6 to 24 hours using a Synergy HT-1 spectrophotometer (Biotek, Winooski, VT). Percent toxicity was standardized using Triton X-100 as a positive control (0–100 mM).
Cells were suspended in nondenaturing lysis buffer (Cell Signaling Technology, Danvers, MA) with 0.5% SDS. Lysates were sonicated and cleared at 14,000g at 4°C. Protein content was assayed by the BCA method (Pierce, Rockford, IL), and 20 to 40 µg of protein was analyzed on 4% to 12% gradient denaturing gels (Invitrogen, Carlsbad, CA). Membranes were incubated overnight at 4°C with primary antibodies [cleaved caspase-3 (9664; Cell Signaling Technology), caspase-9 (9508; Cell Signaling Technology), p53 (9282; Cell Signaling Technology), Bim (4582; Cell Signaling Technology), Bax (2772; Cell Signaling Technology), PARP (9542; Cell Signaling Technology), PUMA (4976; Cell Signaling Technology), and β-tubulin (05-661; Upstate, Billerica, MA)] and visualized using HRP-conjugated secondary antibodies (Calbiochem, Gibbstown, NJ) and ECL-Plus reagents.
For frozen sections, mice were killed after therapy, and tumors were embedded in Tissue-Tex O.C.T. compound (Sakura, Torrance, CA), cooled at -80°C (24 hours), sectioned at 4-µm thickness, mounted on Superfrost slides (Fisher Scientific, Pittsburgh, PA), and placed in 70% ethanol (1 hour at 4°C). Slides were washed three times with wash buffer (Dako, Carpinteria, CA), treated with 3% hydrogen peroxide (for 10 minutes), rewashed, blocked with casein (for 10 minutes), and incubated with caspase-3 antibody (CST 9664) at a dilution of 1:100 at room temperature in a humidified chamber (for 2 hours). Slides were rewashed three times and incubated with rabbit Envision+ secondary antibody (Dako) at room temperature in a humidified chamber (for 30 minutes) treated with Dako DAB+ complex (for 4 minutes), rinsed with water, counterstained with Gills modified hematoxylin (American Master Tech, Lodi, CA), dehydrated, and mounted for analysis. Paraffin-embedded tissues were cut and mounted, deparaffinized with xylene through graded alcohol to water, and then placed in PBS. Heat-induced epitope retrieval was performed at 120°C in citrate buffer at pH 6.0, followed by treatment in 3% peroxide (for 20 minutes). Washing and antibody incubations were as previously mentioned. Ki-67 staining in paraffinized sections was performed using a Ki-67 rat primary antibody (M7249; Dako) in a 3-hour incubation followed by processing as previously mentioned. Development was with streptavidin HRP (Amersham, Piscataway, NJ) at a dilution of 1:100 for 1 hour at room temperature.
TH-MYCN mice were maintained in hemizygotic matings as previously described . Hemizygous TH-MYCN were serially crossed to p53+/- mice (FVBN)  and E2F1-luciferase mice (FVBN)  for more than four generations to re-establish tumor penetrance in the 129X1/SvJ strain. A similar approach was used in luciferase crosses. For the analysis of the in vivo effects of CPMon tumor growth in mice transgenic for TH-MYCN (in strain 129X1/SvJ), a tumor intervention design was implemented. The assay end point was tumor mass. In parallel experiments, tumor response and progression were confirmed using luciferase bioimaging of TH-MYCN/E2F1-Luc doubly transgenic animals, in which a luciferase signal is emitted from proliferating tumor tissue and is a surrogate indicator of Mycn activity. Animals (seven per arm) with palpable tumors (approximately 60 days of life) were treated three times per week with intraperitoneal injections of CPM (150 mg/kg in 100 µl of saline) or vehicle (saline) and were then killed. In parallel, animals (three per arm) were followed by noninvasive imaging, using the Xenogen IVIS Lumina system (Caliper Life Sciences, Hopkinton, MA) with a background bioluminescence setting of 38,000 photons/sec per square centimeters. At sacrifice, tumors were imaged, excised, measured, weighed, and snap-frozen for further analysis. Lysates were prepared for immunoblot analysis as previously mentioned, except that tumors were initially homogenized in TBS with protease inhibitors (Compleat; Roche, Inc.) before lysis. Significance analysis was performed using the Student's t test. All animals were handled in accordance with institutional guidelines for safe and ethical treatment of mice.
Given the critical role of p53 in regulating myc-driven oncogenesis, we sought to evaluate induction of apoptosis by p53 in spontaneously growing neuroblastoma tumors originating within a native murine background. We therefore used the well-characterized TH-MYCN neuroblastoma model in which tyrosine hydroxylase (TH)-mediated overexpression of MYCN is targeted to the peripheral neural crest . This model faithfully recapitulates high-risk human neuroblastoma tumorigenesis by all major clinicopathologic criteria. To evaluate the effect of p53 dysfunction on TH-MYCN-driven neuroblastoma tumor formation, we introduced TH-MYCN into a p53-deficient background by serial crosses against animals with an inactivated germ line p53 allele . Tumors arose in doubly transgenic animals with higher penetrance and reduced latency (Figure 1A). The initial anatomic appearance of these tumors was similar to that of p53 wild type TH-MYCN tumors (Figure 1B), but end-stage tumors were larger and were characterized by a higher density of proliferating neuroblasts and lower levels of apoptosis and necrosis on histologic analysis (Figure 1, C and D). Taken together, the data imply that, in neuroblastoma tumor progression, inactivation of p53 could enhance MYCN oncogenicity, at least in part by decoupling the growth-stimulatory activity of MYCN from its ability to concurrently drive apoptosis.
Because intact p53 function inhibits growth and progression of TH-MYCN neuroblastoma tumors, we used CPM to assess whether established murine TH-MYCN tumors are sensitive to chemotherapeutics that exert their activity at least in part through p53 up-regulation. Cyclophosphamide induced complete regression of established tumors, with a dramatic effect on survival intervals (cumulative tumor free days of life; Figure 2A) and time to death from overwhelming tumor burden (Figure 2B). Tumors in the control cohort were readily apparent on necropsy and by luciferase bioimaging of parallel animals doubly transgenic for TH-MYCN and E2F1-luciferase (representative animal shown in Figure 2C). Seven CPM-treated animals were killed at 100 days of life, with no evidence of tumor on necropsy and negative luciferase signals (representative animal shown in Figure 2D). These data show that tumors in TH-MYCN transgenic mice are sensitive to CPM, mirroring similar observations in MYCN-amplified human neuroblastoma.
To analyze the effect of CPM on apoptosis, necrosis, and proliferation, we treated a cohort of transgenic animals with early, established tumors, sacrificing animals at several time points within 24 hours of CPM treatment. Tumors showed no changes in vascularity (data not shown) and exhibited no obvious cellular or structural changes (observable apoptosis or necrosis) in response to CPM treatment (Figure W1, A–C). In vehicle-treated tumors, immunoreactivity for the proliferation marker Ki-67 was positive in most nuclei (Figure 3, A and B). In response to CPM, Ki-67 labeling decreased dramatically even by 24 hours (Figure 3C). Whereas control tumors had negligible staining for the apoptosis marker cleaved caspase-3 (Figure 3D), high levels of cleaved caspase-3 were observed 3 and 6 hours after treatment with CPM (Figure 3, E and F). A wave of apoptosis throughout the tumor was demonstrated at low power (Figure 3, H–J), with maximal staining visible at 6 hours after treatment. At 24 hours, apoptotic staining was associated with foci of necrosis (Figure 3J, inset). Significantly, levels of apoptosis induced by similar treatment of p53-haplodeficient tumors with CPM were minimal, as assessed by immunostaining for cleaved caspase-3 (Figure W2). No significant differences in apoptosis were seen by immunohistologic analysis in vehicle-treated p53+/- versus p53+/+ tumors (data not shown).
To evaluate apoptotic signaling in response to CPM, we treated a cohort of early tumor-bearing transgenic animals and assessed levels of key apoptotic proteins by immunoblot (Figure 4A). In contrast to basal apoptotic activity detectable by immunoblot analysis of cultured neuroblastoma cell lines, activation of the p53 pathway and caspase-3 cleavage were minimal in vehicle-treated tumors, consistent with the low-level caspase-3 staining seen in Figure 4. Rapid induction of p53 was observed at 3 hours after treatment with CPM, with a peak at 6 hours after treatment (Figure 4A). Cleaved caspase-3 and -9 were maximal at 6 hours after treatment and were sustained during a 12-hour period. These data indicate that basal apoptotic activity in spontaneously progressing TH-MYCN neuroblastomas is minimal but that p53-driven apoptosis is intact and can be strongly induced in response to cytotoxic chemotherapy.
PUMA and Bim are Bcl-2 homology 3 (BH3)-only proteins, proapoptotic components of the mitochondrial cell death pathway, and are critical effectors of p53-induced apoptosis in Myc protein- driven tumors . In response to CPM treatment, PUMA was strongly induced in vivo, peaking at 3 hours and dissipating by 12 hours (Figure 4A). Bax, a downstream target of PUMA and a critical effector of myc-induced mitochondrial apoptosis, was strongly expressed (Figure 4B). Bim, a BH3-only protein necessary for apoptosis in myc-driven lymphoma, was induced prominently. Cleavage of caspases-3 and -9 (Figure 4A) and PARP (Figure 4B) occurred concurrently, indicating high levels of apoptosis. Taken together, these findings indicate that the proapoptotic effect of CPM proceeds through the intrinsic p53 regulatory pathway of BH3 proteins, implicating PUMA, Bax, and Bim as key effectors of p53 function in primary murine neuroblastoma tumors.
To determine whether levels of Mycn influence response to CPM in human tumors, we examined a panel of human neuroblastoma cell lines in which levels of Mycn were undetectable (SH-EP), intermediate (SH-SY5Y), or high (Kelly). All lines expressed roughly comparable levels of p53 at baseline (Figure 5A). Cells were treated with CPM, and proliferation and apoptosis were assessed using soluble MTT and histone DNA-cleavage assays, respectively (Figure 5, B and C). Treatment with 4OH-CPM (the active metabolite of CPM) resulted in effective growth inhibition in 10% serum (Figure 5B) or in low serum (data not shown). 4-Hydroxyperocyclophosphamide induced apoptosis in all cell lines, with increased apoptosis noted in the MYCN-amplified Kelly cell line (Figure 5C). These data are consistent with published findings that neuroblastoma cells amplified for MYCN are sensitized to the effects of cytotoxic chemotherapeutics . Coincident with the induction of apoptosis by 4OH-CPM, activation of p53 was still observed in such cells. We treated MYCN-amplified Kelly cells with 4OH-CPM and cisplatin [cis-diamminedichloridoplatinum(II), CDDP], a bifunctional alkylator also used in frontline therapy for human neuroblastoma. In contrast to a lack of apoptosis observed in vehicle-treated spontaneous in vivo tumors, the vehicle-treated cultured cell lines displayed detectable basal apoptosis by immunoblot analysis. Analogous to the response observed in TH-MYCN tumors growing in vivo, p53 was strongly induced and phosphorylated in response to treatment with 4OH-CPM, and maximal activity correlated with strong induction of caspase-3 cleavage at 6 hours (Figure 5D). Comparable activation was observed in response to CDDP, although with a slightly longer time course of p53 activation (Figure W3). Treatment of these cells with siRNA against p53 led to decreased levels of p53 total and phosphorylated proteins 6 hours after treatment with 4OH-CPM, with a concomitant decrease in levels of the apoptotic marker cleaved caspase-3 (Figure W4). These results are consistent with previous findings indicating that the induction of p53 observed in murine neuroblastoma is also evident in human tumors and that MYCN-amplified human neuroblastoma likely requires p53 for induction of apoptosis in response to cytotoxic chemotherapy.
To our knowledge, this study is the first in vivo demonstration of the importance of intact p53 function to neuroblastoma tumor progression and chemoresponsiveness. In this MYCN-driven malignancy, the proliferative and apoptotic functions of MYCN are effectively uncoupled by functional suppression of p53-induced apoptosis (in the absence of p53 mutation). Furthermore, any coexistent apoptotic defects required for transformation by MYCN are insufficient to block the apoptotic effect induced by cytotoxic chemotherapy. Mutation at p53 is not generally observed in neuroblastoma tumors at diagnosis; however, epigenetic alterations (including silencing of caspase-8, increased levels of survivin and Bcl2, and cytoplasmic sequestration of p53) likely contribute to malignant progression [13,15,29]. Although these epigenetic abnormalities may underlie some relative chemoresistance, our data are consistent with observations in other tumor types suggesting that these abnormalities spare intrinsic mitochondrial apoptotic signaling, which can drive apoptosis in response to cytotoxic chemotherapy.
Cyclophosphamide, a first-line agent in children with newly diagnosed neuroblastoma, is active in murine neuroblastoma. Baseline apoptosis in spontaneously progressing TH-MYCN tumors is suppressed, and activation of such proteins is not detectable by immunoblot analysis, in contrast to observations in cultured neuroblastoma tumor cell lines in vitro. Cyclophosphamide treatment induces massive apoptosis proceeding through activation of p53 and the downstream Bcl-2-homology domain-3 (BH3) proteins, PUMA and Bim, indicating that apoptosis proceeds through mitochondrial-dependent pathways. Bax, a known effector of apoptosis in murine lymphoma, is also strongly induced . PUMA, Bim, and Bax are all key effectors of p53-induced apoptosis, and alterations in PUMA expression modulate the phenotype of c-myc-driven tumors in genetically engineered mouse models of hematopoietic malignances . PUMA complexes directly with p53, linking nuclear and cytosolic p53 fractions, and is required for induction of apoptosis by nuclear p53 . Our observations argue that p53-driven apoptosis is functional but suppressed in spontaneous TH-MYCN tumors growing in vivo, that p53 and its effector PUMA are potential therapeutic targets in neuroblastoma, and that agents that enhance the activity of p53 or PUMA may show efficacy in this disease.
Impaired p53 signaling likely enhances the oncogenic activity of MYCN in neuroblastoma [10–12,33,34] and probably contributes to relapse as additional (mutational) defects in this pathway are acquired with genotoxic therapy. These studies help to clarify observations in cell lines from human neuroblastoma, in which acquisition of mutations in p53 occurred in cell lines established from relapsed tumors [11,24,33]. Collectively, our murine data are in agreement with human cell line data generated here and published by others, suggesting that cytotoxic drugs drive relapse by selecting for mutations in p53 or in the p53 pathway. We are currently generating mice transgenic for TH-MYCN that are inducibly dysfunctional at p53 and are backcrossing them into the penetrant 129/SvJ strain to determine the utility of these animals as a model for drug-resistant high-risk neuroblastoma at relapse.
The authors thank Scott Lowe, Gerard Evan, Jason Shohet, and members of the Weiss laboratory for useful discussions. The authors also thank Larry Donehower for supplying p53+/- mice, Eric Holland for supplying E2F1-luciferase mice, and Susan Ludeman for supplying the 4OH-CPM.
1Grant numbers and sources of support: National Institutes of Health (grants R01CA102321 (W.A.W.), K08NS053530 (L.C.), the Thrasher Research Fund (W.A.W.) and the Samuel Waxman Cancer Research Foundation (W.A.W.), The Katie Dougherty Foundation (W.A.W.), the Children's Neuroblastoma Cancer Foundation (L.C.), the Campini Family Foundation (L.C.), and the UK Neuroblastoma Society (L.C.).