Myc proteins as therapeutic targets

Abstract

Myc proteins (c-myc, Mycn and Mycl) target proliferative and apoptotic pathways vital for progression in cancer. Amplification of the MYCN gene has emerged as one of the clearest indicators of aggressive and chemotherapy-refractory disease in children with neuroblastoma, the most common extracranial solid tumor of childhood. Phosphorylation and ubiquitin-mediated modulation of Myc protein influence stability and represent potential targets for therapeutic intervention. Phosphorylation of Myc proteins is controlled in-part by the receptor tyrosine kinase/phosphatidylinositol 3-kinase/Akt/mTOR signaling, with additional contributions from Aurora A kinase. Myc proteins regulate apoptosis in part through interactions with the p53/Mdm2/Arf signaling pathway. Mutation in p53 is commonly observed in patients with relapsed neuroblastoma, contributing to both biology and therapeutic resistance. This review examines Myc function and regulation in neuroblastoma, and discusses emerging therapies that target Mycn.

Introduction

Neuroblastoma, a neoplasm of peripheral neural crest origin, is the most common malignant extracranial solid tumor of childhood and accounts for 15% of cancer deaths in children (Park et al., 2008). Approximately 650 new cases are diagnosed in the United States annually with peak incidence in early childhood (ages 0–4 years). The most common site of origin is in the adrenal medulla; however, tumors can occur anywhere along the sympathetic chain.

Patients with new diagnoses are typically stratified into risk groups based on age, stage, histopathology, DNA index and genetic/genomic factors. Amplification of the proto-oncogene MYCN occurs in ~25% of tumors and is the best characterized genetic-risk factor for high-risk chemotherapy-refractory disease (Brodeur et al., 1984; Seeger et al., 1985; Riley et al., 2004). Deletion or suppression of caspase 8, loss of chromosomes 1p and 11q, and gain of 17q also correlate with aggressive disease (Bown et al., 1999; Guo et al., 1999; Riley et al., 2004; Attiyeh et al., 2005; Stupack et al., 2006; Maris et al., 2008b).

In contrast with most other treatment-refractory cancers, neuroblastomas, irrespective of risk group, generally respond to initial therapy, which typically includes high doses of chemotherapy (Park et al., 2008). Low- and intermediate-risk patients are subsequently cured and rarely progress. Patients with high-risk disease typically relapse with treatment-refractory tumors. It is conceivable that low- and high-risk neuroblastoma are entirely separate disease entities. Overall, prognosis in the high-risk group is quite poor, with long-term survival of 30–40%.

Familial neuroblastoma

Neuroblastoma may be associated with Hirschsprung disease and neurofibromatosis type 1 (Clausen et al., 1989; Rohrer et al., 2002). Mutations in PHOX2B, a homeodomain-containing transcription factor important for the development in the autonomic nervous system, underlie the congenital central hypoventilation syndrome and confer a heritable predisposition to neuroblastoma; however, these mutations are not found in spontaneous tumors (Mosse et al., 2004; Trochet et al., 2005; Raabe et al., 2008). In fact, familial neuroblastoma is quite rare, and is most often because of dominant gain-of-function mutations in the orphan receptor tyrosine kinase (RTK) anaplastic lymphoma kinase (ALK). ALK, which is linked to MYCN on chromosome 2p23, also shows sporadic gain-of-function mutation in 8% of spontaneous tumors (Chen et al., 2008; George et al., 2008; Janoueix-Lerosey et al., 2008; Mossé et al., 2008). Although direct connections between MYCN and ALK have yet to be elucidated, activation of Alk and other RTKs may contribute to stabilization of Mycn protein (detailed below).

Myc family of proto-oncogenes

Broadly implicated in oncogenesis, the human MYC family of proto-oncogenes is among the most studied genes in cancer (Meyer and Penn, 2008). Early insertional mutagenesis studies in mouse identified c-MYC (homologous to the v-myc gene that drives avian myelocytosis) as capable of transformation by retroviral promoter insertion (Payne et al., 1982). MYCN was subsequently identified as an MYC homolog amplified in neuroblastoma tumors (reviewed in Meyer and Penn, 2008). Amplification of MYCN has emerged as among the clearest genetic indicators of high-risk, aggressive disease (Brodeur et al., 1984; Seeger et al., 1985). MYC family members (c-MYC, MYC and MYCL) show differential expression in normal tissues. Expression of murine N-myc in particular is elevated in normal mammalian developing retina, forebrain, hindbrain, intestine, kidney and has functions in neuronal progenitor cells, developing lung tissues, hematopoetic stem cells and programmed cell death in the developing limb (Zimmerman et al., 1986; Mugrauer et al., 1988; Downs et al., 1989; Hirvonen et al., 1990; Hirning et al., 1991; Knoepfler et al., 2002; Bettess et al., 2005; Okubo et al., 2005; Ota et al., 2007; Martins et al., 2008; Xu et al., 2009).

Myc proteins are basic helix-loop-helix leucine zipper transcription factors. Mycn and c-Myc proteins share several regions of homology and share similar cellular functions. Myc proteins localize to the nucleus and form heterodimers with the basic helix-loop-helix molecule, Max (Blackwood and Eisenman, 1991; Prendergast et al., 1991; Berberich and Cole, 1992; Blackwood et al., 1992; Kato et al., 1992). Myc/Max heterodimers bind to DNA at specific CAC(G/A)TG ‘E-box’ sequences to drive transcription of targets important for proliferation, apoptosis and differentiation (Blackwell et al., 1990, 1993; Amati et al., 1992; Kretzner et al., 1992). Max also heterodimerizes with Mxd or Mnt proteins to influence the transcription of other downstream genes and often to antagonize the proliferative effects of Myc proteins (Ayer et al., 1993; Zervos et al., 1993; Hurlin et al., 1995, 1997; Walkley et al., 2005). In many systems, Mxd/Max or Mnt/Max heterodimers oppose the actions of Myc/Max to repress transcription; however, very little is known about these interactions in neuroblastoma (Grandori et al., 2000; Patel et al., 2004). The complex competition among Myc, Mxd and Mnt proteins for binding to Max further modulates the effects of both c-Myc, and probably Mycn, on gene expression.

Downstream of Mycn

E-boxes are common (~25% of known promoters) with >10 000 sites per cell (Fernandez et al., 2003; Li et al., 2003; Zeller et al., 2006). That there are more E-box sequences than Myc molecules in cells represents a conundrum apparently common to many transcription factors (Farnham, 2009). Adding further complexity to this system, Myc proteins also regulate downstream targets through cap-dependent methylation, altering both global translation as well as the translation of specific proteins (Barna et al., 2008; Cole and Cowling, 2008).

Chromatin immunoprecipitation experiments show that c-Myc and Mycn proteins bind to promoters with variable specificity determined by DNA ultrastructure and cellular context (Fernandez et al., 2003; Li et al., 2003; Mao et al., 2003; Chen et al., 2004; Guccione et al., 2006; Zeller et al., 2006; Kim et al., 2008; Martinato et al., 2008; Westermann et al., 2008). Myc/Max heterodimers bind to E-boxes and interact with a variety of histone modifiers, increasing histone acetylation (Bouchard et al., 2001). These alterations modify chromatin at promoters, affecting gene expression (McMahon et al., 2000; Frank et al., 2001; Vervoorts et al., 2003; Guccione et al., 2006; Martinato et al., 2008; Liu et al., 2009). In fact, histone modification by Mycn effectors is being exploited pharmacologically using histone deacetylase inhibitors. These drugs may alter acetylation in neuroblastomas and other MYC dependent processes thereby modulating transcription (reviewed in Witt et al., 2009).

Several studies have attempted to elucidate specific transcriptional targets for Mycn in neuroblastoma (Alaminos et al., 2003). A large number of targets important for cell cycle control and differentiation have been characterized, including: downregulation of SKP2 and TP53INP1 with resultant decrease in p21^WAF1 (Bell et al., 2007), downregulation of DKK1 upstream of the wnt/β-catenin pathway (Koppen et al., 2007), upregulation of NLRR1 both important in neural cell proliferation (Hossain et al., 2008), downregulation of Fyn kinase important in differentiation (Berwanger et al., 2002), regulation of multiple genes responsible for pluripotency (Cotterman and Knoepfler, 2009) and modulation of apoptosis by upregulation of p53 and Mdm2 (Slack et al., 2005b). The multidrug resistance gene MRP1 is regulated by Mycn, driving chemotherapy resistance (Manohar et al., 2004). Importantly, Mycn upregulates oncogenic microRNAs, which have wide ranging effects on cancer (reviewed in Schulte et al., 2009). Mycn also controls several proteins important in ribosome biogenesis (Boon et al., 2001) affecting protein synthesis (reviewed in Ruggero, 2009).

Different groups seeking to stratify neuroblastoma risk using gene expression micoarrays have generated gene lists that are largely non-overlapping (Berwanger et al., 2002; Ohira et al., 2005; Schramm et al., 2005, 2009; Oberthuer et al., 2006). Similar strategies using real-time PCR transcript analysis, are also being performed (Vermeulen et al., 2009). Direct comparison of targets from chromatin immunoprecipitation shows that Mycn and c-Myc have many overlapping targets (Westermann et al., 2008). Further refinement of microarray and chromatin immunoprecipitation techniques and identification of critical transcriptional targets specific to Mycn may provide insights into both biology and therapy for neuroblastoma.

Mycn in cell cycle control

In normal cells, levels of Myc proteins are tightly regulated, with increased levels driven in part through activation of phosphatidylinositol 3-kinase (PI3K), which stabilizes Mycn and c-Myc proteins (Figure 1), enabling entry into the cell cycle (Marqués et al., 2008). In addition to the clearly defined function of Myc family members in control of the cell cycle, c-Myc contributes non-transcriptionally to the initation of DNA replication (Dominguez-Sola et al., 2007). Myc/Max dimers also bind and inhibit Miz-1, a helix-loop-helix transcription factor (reviewed in Herold et al., 2009). Free Miz-1 promotes transcription of p15INK4b and p21Cip1 proteins associated with cell cycle arrest. Inhibition of Miz-1 in response to Myc/Max binding contributes to immortalization, transformation and oncogenesis (Seoane et al., 2001; Staller et al., 2001; Herold et al., 2008). Interactions between Mycn and Miz-1 are incompletely characterized. Expression of Miz-1 has been associated with favorable outcome in neuroblastoma, however, consistent with an interaction among Miz-1, Mycn and Max (Ikegaki et al., 2007).

Figure 1.

Schematic model of regulatory pathways involved in Mycn stabilization. In proliferating cells, Mycn is initially phosphorylated at serine 62 by cyclin-dependent kinase 1/CyclinB. The S62-phosphorylated protein is stabilized and competent to enter the nucleus. Nuclear Mycn binds to its partner, the helix-loop-helix transcription factor Max and stimulates the transcription of a target genes important in cell cycle, proliferation, differentiation and apoptosis. This priming phosphorylation at S62 allows the binding of Gsk3β, as well as Pin1 and PP2A in a complex also containing Axin. Active Gsk3β phosphorylates Mycn at threonine 58, producing doubly phosphorylated, stabilized and transcriptionally active Mycn. In a Pin1-mediated process, PP2A dephosphorylates the S62 phosphate, enabling binding of ubiquitin ligase Fbw7. Fbw7 subsequently drives poly-ubiquitination and proteasomal degradation of Mycn. Aurora A kinase can bind to and stabilize polyubiquitinated/phosphorylated Mycn, potentially leaving it competent to bind to Max and activate transcription. In this model, upstream signaling starts at the membrane with receptor tyrosine kinases (RTKs), which are either activated by ligand or by constitutively activating mutations (in Alk, for example). RTKs activate phosphatidylinositol 3-kinase (PI3K), which catalyzes the conversion of phosphatidylinositol-3,4-bisphosphate (PIP2) to phosphatidylinositol-3,4,5-triphosphate (PIP3). PIP3 then binds to Akt, localizing it to the membrane and allowing the phosphorylation and activation of Akt by membrane bound Pdk1 at threonine 308. Active Akt then phosphorylates Glycogen synthase kinase 3β (Gsk3β) and inactivates it, blocking Gsk3β-mediated phosphorylation of Mycn T58 and thereby stabilizing Mycn. Akt also activates mammalian target of rapamycin (mTOR) through several indirect signaling mechanisms including through Tsc2/Tsc1 and Rheb. mTOR exists in two distinct complexes. mTORC1 complex, the rapamycin sensitive form, consists of mTOR bound to multiple effector proteins, including Raptor and Lst8, and is important for the promotion of global translation through S6K and 4EBP. mTORC1 also directly phosphorylates and inhibits PP2A, enabling the accumulation of doubly phosphorylated, active Mycn and contributing to the general proliferation-promoting downstream effects of the PI3K/Akt/Mycn pathway (Cheng et al., 2007). The mTORC2 complex, also known as Pdk2, contains mTOR, rictor, Lst8 and mSin1. mTORC2 phosphorylates Akt at serine 473 further activating this kinase. Inhibitors of this pathway could potentially destabilize Mycn. These inhibitors are currently in clinical trials or in clinical development and are denoted by red octagons. The Aurora A Kinase inhibitor is denoted by an orange octagon, which represents targeted kinase inhibition, but no anticipated activity against Mycn itself. Relevant references are contained within the corresponding text.

Transcription of MYCN is downregulated by the neuroblastoma differentiating agent retinoic acid and upregulated by several known transcription factors including E2F and Sp1/Sp3 (Thiele et al., 1985; Inge et al., 2002; Kramps et al., 2004; Kanemaru et al., 2008). Sonic hedgehog indirectly regulates transcription of MYCN in developing neurons (Kenney et al., 2003, 2004; Oliver et al., 2003; Mill et al., 2005). Levels of Myc mRNA are also by alternate internal ribosomal entry sites (Barna et al., 2008; Cobbold et al., 2008). Although natural antisense transcripts are frequently co-amplified with MYCN in neuroblastoma, the importance of these co-amplified sequences remains unclear (Krystal et al., 1990; Armstrong and Krystal, 1992; Jacobs et al., 2009).

Mycn and apoptosis

In addition to promoting proliferation, the expression of c-Myc and Mycn actually drives apoptosis (Askew et al., 1991; Evan et al., 1992; Shi et al., 1992; Fulda et al., 1999; Paffhausen et al., 2007; Ushmorov et al., 2008). Transformation by Myc proteins, therefore, requires concomitant inhibition of apoptosis. Tissue-specific expression of a switchable allele of c-Myc also induced apoptosis in vivo. In contrast, proliferation and tumorigenesis required lower level, continuous and deregulated expression of c-Myc (Murphy et al., 2008).

Although the association between MYCN amplification and poor outcome has been reproduced in numerous studies over decades, a similar association between expression of Mycn and outcome remains controversial (Chan et al., 1997; Cohn et al., 2000; Tang et al., 2006). Data from preclinical models of switchable myc above suggest that low levels of myc proteins may drive proliferation, with higher levels required to induce apoptosis. If it is true in the context of Mycn and neuroblastoma, then MYCN amplification may serve primarily to dysregulate Mycn during the cell cycle, rather than simply driving high-level expression. This hypothesis is consistent with the claims that amplification and not overexpression is predictive of aggressive disease, although inconsistent with observations that MYCN is often amplified and overexpressed to extreme levels in neuroblastoma.

Mechanisms through which apoptosis is inhibited as a contributer to Mycn-driven transformation are complex and not yet fully elucidated. Silencing of the apoptotic initiator Casp8 is observed frequently in neuroblastoma (Stupack et al., 2006). Crosstalk with the p53 pathway has been implicated in apoptosis mediated by both Mycn and c-Myc (reviewed in Hoffman and Liebermann, 2008; Van Maerken et al., 2009b). Mutations in p53 are rare in primary neuroblastoma (<2%) irrespective of MYCN amplification (Vogan et al., 1993). However, inactivating mutations in p53 and in p53 pathway members are common at relapse (Keshelava et al., 1997, 2000; Tweddle et al., 2001; Carr et al., 2006). Myc proteins indirectly regulate the p53 pathway and p53-dependent apoptosis through the p14^ARF-MDM2-p53 axis (reviewed in Van Maerken et al., 2009b).

At baseline, p53 is tightly regulated by Mdm2, which binds to and inhibits the transactivation domain of p53 (Oliner et al., 1993; Thut et al., 1997). Mdm2 also ubiquitinates and targets p53 protein for degradation (reviewed in Coutts et al., 2009). Further regulation is conferred by the tumor suppressor protein p14^ARF, which binds to and inhibits Mdm2, allowing activation and stabilization of p53 (Kamijo et al., 1998; Zindy et al., 1998; Weber et al., 1999; Midgley et al., 2000; Lin and Lowe, 2001). The balance between apoptosis and survival remains in equilibrium through multiple feedback and feed-forward loops affected by other signaling pathways including external apoptotic stimuli.

In neuroblastoma cells, Mycn directly stimulates transcription of MDM2 (Slack et al., 2005a; Barbieri et al., 2006; Chen et al., 2009). The resulting inhibition of p53 may in part allow cells to escape Mycn-primed apoptosis. Myc and Mycn also indirectly inhibit p14^ARF, through directly driving transcription factors including TWIST1, resulting in activation of MDM2 and escape from apoptosis (Maestro et al., 1999; Valsesia-Wittmann et al., 2004). The p14^ARF protein can in-turn feed back and bind to Myc and Mycn proteins, abrogating their ability to activate downstream targets (Qi et al., 2004; Amente et al., 2007). The complex interactions among Myc/Mycn, Mdm2 and p14^ARF provide mechanisms through which Mycn both indirectly activates or inactivates p53, altering the sensitivity of cells to apoptotic stimuli (reviewed in Li and Hann, 2009; Van Maerken et al., 2009b).

Neuroblastoma tumors at diagnosis are generally wild type for p53 and respond to chemotherapy. Children with high-risk tumors generally relapse, whereas those with low and intermediate disease are typically cured. As stated above, Mycn-induced downregulation of the p53 axis could potentially underlie the minimal residual disease that drives subsequent relapse in newly diagnosed MYCN-amplified neuroblastomas. This hypothesis is supported by the identification of Mdm2 as an Mycn target, and in studies showing that Mdm2 haploinsufficiency inhibits tumorigenesis in MYCN-driven models for neuroblastoma (Slack et al., 2005a; Chen et al., 2009). In fact, small molecule inhibitors of the p53/Mdm2 interaction (‘Nutlins’) are currently in development and have shown some promise for the preclinical treatment of cancers including neuroblastoma (Barbieri et al., 2006; Chen et al., 2009; Van Maerken et al., 2009a).

Modeling MYCN-amplified neuroblastoma

Mice carrying an MYCN transgene under control of the rat tyrosine hydroxylase promoter develop neuroblastoma tumors several months after birth (Norris et al., 2000; Weiss et al., 2000). Tumors from mice transgenic for TH-MYCN develop in adrenal and mesenteric ganglia and in paraspinous locations. Histolology and genetics show similarities with high-risk neuroblastoma (Weiss et al., 2000; Hackett et al., 2003; Moore et al., 2008). In relatively resistant strains or subspecies (for example C57BL/6, BALBc or Mus musculus castaneus), a range of differentiation was observed in murine tumors. However, when crossed into highly penetrant strains, such as 129 SvJ, the histology was uniformly undifferentiated small round blue cells.

Cell lines derived from TH-MYCN tumors retain the ability to form tumors in nude mice (Cheng et al., 2007). Unlike xenograft models of neuroblastoma, TH-MYCN tumors show native host interactions with the tumor microenvironment including vascular cells (Chesler et al., 2007). When treated with chemotherapy, tumors in this model, similar to their human p53 wild-type counterparts, undergo apoptosis in a p53-dependent manner. Furthermore, tumors from TH-MYCN;p53−/+ mice were refractory to cytoxic chemotherapy (Chesler et al., 2008). These data and corresponding data from human tumors (Keshelava et al., 1998, 2000; Xue et al., 2007) collectively suggest a model in which newly diagnosed neuroblastoma tumors impair the p14^ARF-MDM2-p53 axis in the absence of p53 mutation, enabling tumors to arise. Subsequent chemotherapy subsequently provides strong selective pressure for inactivating mutations in p53 or in components of the p53 pathway, resulting in chemotherapy-refractory tumors seen clinically in relapsed patients. Thus, although mutation at p53 rarely contributes to the biology of primary neuroblastoma, therapy-selected mutations in p53 drive a genetically distinct tumor at relapse. This therapy-associated alteration in the biology of neuroblastoma presents a challenge in identifying effective treatments for relapsed tumors.

Mycn as a target for therapy

In light of both the frequency and importance of MYCN amplification in pathogenesis of high-risk neuroblastoma, blockade of Mycn signaling represents an important approach for the developmental therapeutics. The resistance of high-risk relapsed neuroblastoma to conventional chemotherapy and the high morbidity and mortality in these patients present a formidable challenge for clinicians. The prominence of MYCN amplification in the pathogenesis of this disease points to Mycn as a potential therapeutic target.

Neuroblastoma presents both common and unique challenges for therapy. The initial response to chemotherapy even in high-risk disease is somewhat unusual. However, the acquisition of p53 mutant, therapy-refractory disease is common to many cancers (McDermott et al., 2008). Deregulated expression of Mycn may contribute to genomic instability and, in combination with the strong selective pressures of chemotherapy and radiation, select for mutation at p53 through interactions with Mdm2 and p14Arf detailed above, and by driving the cell cycle and activating p53-dependent check points. Alleviation of check-point activation by blocking Mycn itself could conceivably impair the acquisition of p53 mutant, refractory disease.

Transcription factors have long been considered as targets for cancer therapy; however, clinical approaches to block this class of molecules remain elusive. Clearly, the most direct means of silencing these molecules would be by disrupting Myc synthesis directly through RNAi methods and there are numerous examples of siRNA used successfully on cultured cells. However, although these approaches are extremely useful tools in the laboratory, they have yet to achieve regular use in the clinic largely because of difficulty with delivery (Whitehead et al., 2009).

It is difficult to develop drugs with activities sufficient to block protein–protein interactions or binding of such factors to target sequences. Molecules developed against c-Myc have been primarily directed against the Myc/Max interaction domain and so far have fairly low potency. Progress is being made in this area, however, and the development of inhibitors, which dissociate Myc/Max heterodimers is important (Yin et al., 2003; Wang et al., 2006; Berg, 2008; Brooks and Hurley, 2009). Chemists are good at developing kinase inhibitors, however, and the stability of Myc molecules is regulated at a number of levels by kinases and critical phospho-residues.

Post-transcriptional modification and stabilization of Myc proteins

In neuroblastoma and in neural progenitor cells, Mycn protein stability is regulated by a complex signaling network involving both feedback and feed-forward loops (Figure 1) (Kenney et al., 2004; Sjostrom et al., 2005; Chesler et al., 2006; Kang et al., 2008). Although both c-Myc and Mycn are widely phosphorylated, sites critical to stabilization of Myc proteins are located in the N-terminal Myc Box I, which is commonly mutated in c-Myc in Burkitt lymphoma (Bhatia et al., 1993; Smith-Sørensen et al., 1996; Bahram et al., 2000). Comparable mutations in c-Myc and Mycn are rare in solid tumors including neuroblastoma. The important residues in the Myc Box I region of c-Myc and Mycn are threonine 58 and serine 62 (Henriksson et al., 1993; Pulverer et al., 1994; Lutterbach and Hann, 1999; Sjostrom et al., 2005).

Phosphorylation at S62 stabilizes Mycn protein and primes it for phosphorylation at T58. Candidate kinases proposed to phosphorylate c-Myc at S62 include extracellular signal-regulated kinase, the canonical downstream effector of the Ras/Raf/MAPK; c-Jun N-terminal kinase and cyclin-dependent kinase 1 pathway (Henriksson et al., 1993; Lutterbach and Hann, 1999; Sears et al., 2000; Benassi et al., 2006). As shown in cerebellar neural progenitor cells and synchronized mitotic SJ8 neuroblastoma cells, cyclin-dependent kinase 1 is thus far the only candidate priming kinase for Mycn at S62 (Sjostrom et al., 2005).

Myc and Mycn proteins monophosphorylated at S62 are substrates for a second phosphorylation at T58, controlled by glycogen synthase kinase 3β (Gsk3β). Gsk3β signals in the Wnt pathway (reviewed in MacDonald et al., 2009), and also downstream of the PI3K/Akt/mTOR pathway (reviewed in Engelman, 2009; Ma and Blenis, 2009; Memmott and Dennis, 2009). The S62 priming phosphorylation site in Myc/Mycn promotes interaction with a complex containing Gsk3β, Pin1, PP2A and Axin. This complex induces the phosphorylation at T58 by Gsk3β in both Myc and Mycn. For c-Myc, Pin1 has been shown to regulate dephosphorylation of S62 by PP2A, a phosphatase and tumor suppressor that itself is regulated by mTOR (Peterson et al., 1999; Gingras et al., 2001; Hartley and Cooper, 2002; Yeh et al., 2004; Arnold and Sears, 2006; Arnold et al., 2009). Regulation of Mycn dephosphorylation by Pin1 and PP2A has not been fully characterized.

Myc proteins monophosphorylated at T58, then bind to the E3 ligase Fbxw7 and are targeted for ubiquitination and degradation (Saksela et al., 1992; Lutterbach and Hann, 1994; Sears et al., 1999; Gregory and Hann, 2000; Oliver et al., 2003; Herbst et al., 2004; Kenney et al., 2004; Welcker et al., 2004; Yada et al., 2004; Otto et al., 2009). Aurora kinase A, which has critical functions in cell cycle regulation and spindle assembly, contributes at this step to the stabilization of phosphorylated and ubiquitinated Mycn, but not of equivalently modified c-myc (Otto et al., 2009). Consistent with these findings, expression of Aurora A kinase is a negative prognostic factor in neuroblastoma (Shang et al., 2009). Although the emerging function of inhibitors of Aurora kinase A in cancer therapy suggested that inhibitors of Aurora kinase A might have unique and multifaceted functions in the therapy of neuroblastoma, stabilization of Mycn apparently requires a scaffold function of Aurora kinase and was independent of Aurora A kinase activity (Maris, 2009; Otto et al., 2009).

Myc proteins as downstream targets of PI3K

The phosphorylation of c-Myc and Mycn is thus regulated directly by Gsk3β, and indirectly by upstream signaling through RTKs, PI3K, Akt and mTOR. RTKs are important to PI3K signaling in neuroblastoma. As mentioned above, activating mutations in the orphan RTK ALK are found in ~9% of neuroblastoma. Other RTKs that could potentially regulate stabilization of Mycn include the insulin-like growth factor receptor and the Trk family of neurotropin receptors (TrkA, TrkB and TrkC). Although mutation in Trk genes is not typically observed in neuroblastoma, expression of TrkA correlates with favorable prognosis, whereas expression of TrkB correlates with amplification of MYCN and poor outcome (reviewed in Brodeur et al., 2009). Although neither Trk nor Alk proteins have been directly linked to stabilization of Mycn, these kinases are all upstream activators of PI3K, implying a possible connection (Bai et al., 2000; Norris et al., 2000; Ho et al., 2002; Marzec et al., 2007).

RTKs activate PI3K, which catalyzes the conversion of phosphatidylinositol-3,4-bisphosphate (PIP2) to phosphatidylinositol-3,4,5-triphosphate (PIP3) (reviewed in Ma and Blenis, 2009; Memmott and Dennis, 2009). PIP3 binds to Akt and localizes it to the membrane, enabling phosphorylation at T308 by the kinase PDK1. Activation of Akt leads to phosphorylation and inactivation of Gsk3β, stabilizing Mycn by blocking phosphorylation at T58. Activated Akt also phosphorylates and inhibits the tuberous sclerosis 2 (Tsc2) tumor suppressor protein. Tsc2 binds to Tsc1, enabling the Ras-related GTPase Rheb to stimulate mTOR bound in the mTORC1 complex of proteins. As mentioned above, mTORC1 downregulates PP2A, which normally dephosphorylates Myc/Mycn at S62, targeting it for ubiquitination, and promoting stabilization of Myc and Mycn. Thus, activation of RTKs and PI3K converge on Akt to phosphorylate and inhibit Gsk3β, blocking phosphorylation at T58 and promoting stabilization. In addition, Akt activates mTORC1, inhibiting dephosphorylation of T58/S62-phosphorylated Myc/Mycn at S62, leading to further stabilization of Myc/Mycn.

The critical importance of Mycn phosphorylation and stability as a downstream target of PI3K/Akt/mTOR in neuroblastoma cells is apparent when neuroblastoma cells are treated with a broad spectrum PI3K inhibitor. Activation of Akt predicts poor outcome in neuroblastoma patients (Opel et al., 2007). Treated cells show a decreased proliferation, which is largely rescued when they are engineered to express T58/S62 phosphorylation site-deficient Mycn mutants (Sjostrom et al., 2005; Chesler et al., 2006). These data show that degradation of Mycn is a critical downstream factor in the efficacy of PI3K/mTOR pathway and suggest that clinical inhibitors of PI3K should show activity in Mycn-driven neuroblastoma (detailed review in Fulda, 2009).

Mycn stability as a therapeutic target in neuroblastoma

The post-translational modification and stabilization of Myc and Mycn proteins represents an area with promise for currently available and emerging-targeted therapies. Inhibitors of RTKs (Alk and Trk), PI3K and Akt should activate Gsk3β (which is negatively regulated by phosphorylation), whereas inhibitors of mTOR kinase activity should inhibit the de-activation of PP2A by mTORC1, enabling PP2A to dephosphorylate S62, collectively driving degradation of Mycn in neuroblastoma (Figure 1). Specific Alk and Trk inhibitors are currently in development and have shown promise preclinically and in phase I trials (reviewed in Chiarle et al., 2008; Li and Morris, 2008; Brodeur et al., 2009; Mossé et al., 2009).

Downstream of RTKs, inhibitors of PI3K, mTOR, dual inhibitors of PI3K/mTOR and inhibitors of Akt are all in clinical trials (reviewed in Garcia-Echeverria and Sellers, 2008; Engelman, 2009). Allosteric inhibitors of mTORC1 (so-called rapalogs) block mTOR independently of ATP binding, are currently being tested in neuroblastoma and have shown mixed results preclincally (Houghton et al., 2008; Johnsen et al., 2008; Maris et al., 2008a; Wagner and Danks, 2009). These agents only affect some outputs of mTORC1. In comparison, ATP-competitive inhibitors of mTOR more completely block mTORC1 and also block mTORC2 (Feldman et al., 2009; Thoreen et al., 2009; Yu et al., 2009; Zask et al., 2009). The availability of inhibitors targeting RTKs, PI3K, Akt and mTOR, and the use of these to destabilize Mycn protein, represent important areas of investigation. Further, because of the complex interrelation of pathway members, inhibition at one point often induces feedback activation in other signaling pathways, justifying the need to test these agents in combination using preclinical models.

The regulation of Mycn phosphorylation also involves a priming phosphorylation at S62. As efficient destabilization of Mycn would require activators of kinases responsible for these phosphorylation steps, the ability to finesse phosphorylation at S62 presents a therapeutic challenge. Aurora kinase A represents an additional therapeutic target, as Aurora kinase A stabilizes Mycn at later steps. Inhibitors of Aurora A kinase are currently in clinical trials in neuroblastoma (Shang et al., 2009). However, asMycn is stabilized by a kinase-independent activity of Aurora A, these inhibitors are unlikely to affect Mycn protein (Gautschi et al., 2008; Maris, 2009). An allosteric and ATP-competitive small molecule inhibitor of Aurora A–Mycn interactions, if it could be developed, should retain the ability to block kinase-dependent functions, whereas also twisting Aurora A kinase, thereby disrupting a scaffolding function and degrading Mycn protein. The interplay among RTKs, PI3K, Akt, mTORC1/2, Aurora A and Mycn is quite complex. However, the broad functions for these kinases in cancer biology, and the specific function in regulating the stabilization of Mycn proteins suggests functions in both Mycn-driven and Mycn-independent cancers including neuroblastoma. The current availability of clinical Alk, Trk, PI3K, mTOR and Aurora inhibitors presents an important translational opportunity to test these agents in children with high-risk neuroblastoma.